Handbook of Pronunciation Many Languages - Luciano Caneparia

From: L.
Canepari (2005) A Handbook of Pronunciation Lincom Europa
Contents
p.
Foreword
A Handbook of pronunciation
English, Italian, French, German,
Spanish, Portuguese, Russian, Arabic,
Hindi, Chinese, Japanese, Esperanto
1
1
16
21
23
28
34
36
37
39
41
46
1. Prelude
A synthesis of Natural Phonetics Tonetics
Transcriptions
e contents of the HPr ( of the HPh)
Observations on phonetic terminology
Guide to the gures
Guide to dierent types of transcriptions
Transcribing by hand
Generic symbols
e ocial IPA chart
canIPA correspondent oIPA symbols
2. English American British (neutral other accents)
123
3. Italian
152
4. French (neutral other accents)
186
5. German (neutral other accents)
231
6. Spanish Iberian American
257
7. Portuguese Brazilian Lusitanian
282
8. Russian
314
9. Arabic
330
10. Hindi
345
11. Chinese Mandarin
370
12. Japanese
392
13. Esperanto
413
424
427
Utilizable bibliography
Index
Language index
From: L. Canepari (2005) A Handbook of Pronunciation Lincom Europa
1. Prelude
A synthesis of Phonetics Tonetics
1.1.1. th the aid of some indispensable diagrams (mostly taken from the
HPh, but with simplications and other modications), we will provide the key
to adequately use the rich potentialities of Phonetics, with regard to the 12 languages dealt with in this book: English, Italian, French, German, Spanish, Portuguese, Russian, Arabic, Hindi, Chinese, Japanese, and Esperanto.
Above all we take into consideration voicing, vowels, consonants, and intonation
(including tones). For in-depth analyses, the reader is referred to the companion
volume HPh (A Handbook of Phonetics).
1.1.2. 1.1 shows the glottis (which is the space between the vocal folds) and
the positions adopted during the phonation types we are interested in. In addition
to normal breathing, we have a complete stoppage ( the stop (), which technically can be neither voiced nor voiceless, but has more anities with the latter
type, and will schematically be represented with ), besides voicelessness ((f), ) and
voicing ((v), ).
Furthermore, we also nd lenis (or lenited] voicelessness and voicing (respectively (, v), , ), and mixed phonation (or half-voicing, (), with three schematic
1.1. Phonation types and positions of the glottis.
. breathing
. occlusion
(){}
(, P)
. voicelessness
(f, s, )
(, s, h, ), A)
+
+
. lenis voicing
(v, z, H, ")
. mixed phonation
. creaky voice
(Ca)
. lenis voicelessness
. voicing
(v, z, , m, a)
(, , , ) {1 (pbX ), 2
(bpX b), 3 (bpX )}: 1 (|'a, ap'a)
2 (&aa'a) 3 (a'pa, 'a|)
. falsetto
(ma)
a handbook of pronunciation
icons, which we will shortly see), where half of the interested phone is voiced,
while the other half is voiceless.
Generally, the choice between the three of them depends on context: after
a pause (or silence) or a voiceless C, the rst half is still voiceless; whereas the second half, which is in contact with voiced phones, is voiced: ( ), as in German: Bett
/'bEt/ ('Et). Instead, before a pause or a voiceless C, the two halves are exchanged:
( ), as in English: Bed /'bEd/ ('b;). On the other hand, within words or phrases/
sentences, between voiced phones, the central part is the voiceless one, whereas the
two margins (each one for a fourth of the total duration) are voiced: ( ), as in Neapolitan-Italian pronunciation: dato /'dato/ ('dA;&). We have added several particulars in 1.1, where there are two peculiar phonation types, too: creak ( ) and
falsetto: ( ), which are useful for Chinese and Hindi respectively.
In 4.1.7-12 of the HPh, we have explained how to verify and check if voice is
1.2. Dierent phonation types exemplied by some languages: American British English
(with mediatic British variants); Italian (with two regional variants: Naples Rome); French;
(Lusitanian) Portuguese; German; (Mandarin) Chinese; Hindi; Japanese.
('pha;)
('spa;)
('ba;)
('b;)
('phl;I)
('h;&Sp)
pie
spy
buy
bib
play
headship ({Amer./Brit.} Eng.)
/ /
('kT, 'T) ('wks, 'ws)

actor
('Eka) ('wks)
(Brit.)
wicks
actor
wicks (mediatic Brit.)

('fa:va)
fava
(u'gwan:to)
(&ukan'tan:te)
un guanto
('sud:H, -d:) ('Es:th, -t)
un cantante
sud
(sa'pe:te)
sapete
est
(Ital.)

(&ilkap'pOt:to)
(&ilkp'pOt:o)
il cappotto
il cappotto (~)
(&ugn'dan:de)
un cantante (~)
(s'e;Ie)
sapete (~)
(sa'be:de)
sapete () (reg.)
('pe) ('p#) ('pi) ('p) ('ppa) ('pis)) ('tX)
('Atu)
pied
alto (Lus. Port.)
poids
puis
plat
peuple
prisme
quatre (French)

('phaen) ('aen)
('li:pI)
('a:b)
('ap&Ilt)
('Ek&f)
Pein
lieblich
aber
Abbild
Bergdorf
Bein
(Germ.)
(5pa)
(pha2U)
(5t,2bu2tUU)
('1pa)
(a63qa)
(63pa3b)
pai
paishu
tingbudng
bi
bici
bba (Chinese)
('kaan) ('khaan)
('gaan)
('gHaan)
('m21)
(3i'i)
kaan
gaan
ghaan
mq? (Hindi)
chichi (Japanese)
khaan
1. prelude
present or not, depending on vocal-fold vibrations during the productions of phones.

1.2 shows the various phonation types applied to the 12 HPr languages, only.
1.1.3. As far as vowels are concerned, let us recall that from a phonetic point of
view it is more convenient to use the term vocoids, while reserving the more traditional terms for phonemes and graphemes, in addition to more general expositions,
as at the beginning of the HPh.
1.3-5 will help to reconstruct the typical modalities for the production and
identication of vocoids, or vowel phones, which have three fundamental components that concisely are: the fronting and raising of the dorsum (or central part
of the tongue), with dierent degrees of jaw opening, and dierent lip positions,
since lip rounding doubles the number of possible vocoids.
1.3. Orograms with the extreme points for vocoid articulations.
1.4. Vocogram with the extreme points for vocoid articulations (and corresponding labiograms).
u
i
nt
ce -cen
nt tr
ra al
ba l
ck
ba -cen
ck tra
l
fro
nt
fro rou
nt nd
e
ce cen d
nt tra
r
l
ba al ro rou
ck u
n nd
ba cen ded .
ck tra
ro l ro
un u
de nd
d .
M [*]
[]
X []
x []

A
3
Y y
Y
[]
[] #
[]
[]
5
%
T
+
u
U
o
high ()

lower-high ()
higher-mid ()
lower-mid ()
Y{%}
higher-low ()
~
low ()
y{%}u
= +
fro
i
I
e
fro
nt
1.5. Vocoid classication (with labiograms of high vocoids).
Furthermore, 1.6 will complete our general view, by providing all possible
vocoids, in orograms which contain a miniature vocogram, in order to help to see
1.6. Vocoid orograms.
unrounded
{{*}}
{{}}
{{}}
{{}}
rounded
{{}}
{{}}
{{}}
{{}}
1. prelude
nuances and dierences better.

For vocoids, voicing is the normal type, so much so because voicelessness is considered to be the marked characteristic. Also the addition of possible nasalization
is considered marked ( 1.7; in 11.17 of the HPh we explain how to verify and
check exactly how nasalized vocoids are produced).
For the eight vocoids which appear without any grey background, and with
symbols in double brackets, no language has been found that uses them systematically. 1.8 shows the various lip positions, by means of labiograms.
1.7. Oral (or normal) vocoids and nasalized vocoids.
1.8. Labiograms for the dierent vocoids.

spread
neutral
(or normal)
half-round(ed)
vertically
rounded
round(ed)
M {{*}}
{ W}
Y {% }
y {% } u
{{}}
{ : w}
{T }
Y {T } U
X {{}}
{ , }
{{}} {+ P}
{+ P} o
x {{}}
{ }
{{}} { }
# { }
{ ,}
{{}} {@ }
{@ } O
aA
{ a}
{{}} { }
{ }
1.1.4. Also for consonants we follow a more scientic terminology, which denes the consonantal phones as contoids, while reserving the more traditional
terms for phonemes and graphemes, in addition to more general expositions.
e three fundamental components for the production and identication of
contoids are: manners and places of articulation, and phonation types. e latter
are extremely useful for contoids, since depending on whether voicing is present
or not they double their number, for distinctive or phonemic purposes, as of-
ten happens with (t, d c, G f, v), which then form diphonic pairs, which are distinguished only by the dierent type of phonation.
ere are seven fundamental manners of articulation: nasal, stop, stop-strictive,
constrictive, approximant, trill, lateral. Traditionally (in spite of an actual and objective diculty, instead of the clearer articulatory terms that we use), some terms
of auditory origin, such as aricate for stop-strictive or fricative for constrictive,
are still widespread.
Depending on languages, some internal subdivisions are possible, as that between grooved (constrictive and stopstrictive) contoids, which are marked in comparison with the simple slit ones, from which they can be distinguished by the
addition of a groove along the longitudinal part of the tongue. For instance, (s, z
S, Z q, Q c, G) are grooved articulations.
Another important subdivision applies to trills, such as (r, K), which consist in
two rapid tappings (respectively of the tip of the tongue or of the uvula), and taps,
such as (R, ), with just one tapping. Also aps are possible, such as ([, ), which
produce a rapid contact in passing to a fronter position, as can be seen from the
illustration ( 1.14.3).
1.9-15 present, according to the seven manners (with further internal subdivisions), the contoid articulations found in dealing with the 12 languages described
in 2-13, including the variants indicated. is is why we call them HPr con 1.9.1. Nasal contoids.
)/m
/[
/m
/M
/M
/"
/M
/n
/n
/N
/n
/n
/N
/n
/~
/N
/\
/m
/,
1. prelude
toids, in order to recall both the completeness of the HPr (in itself), and that these
consonantal articulations are not the only existing ones.
Rather, since there are only 200 in comparison with the 462 given in the HPh,
they are less than half (again, of course, considering them as mere articulations,
since potentially each one may produce a diphonic pair, with two distinct elements because of the addition of voicing. As a matter of fact, they almost double
the actual number of phones; there are 283 in the HPr, but 772 in the HPh).
/M
1.9.2. Semi-nasal contoids.

/n
p/b
(/{
Q/
p/b
]/7
t/d
t/d
4/7
T/D
B/
T/D
t/d
+/_
/8
k/g
$/
1.10. Stop contoids.
p/
./
k/
%/
w/
k/
1.11.1..
(at) stop-strictive contoids.
1.11.1.. Grooved stop-strictive contoids.
q/Q
q/Q
/"
C/
c/G
&/1
C/G
c/g
v/
1.11.2. (Grooved and at) stop-semi-strictive contoids.
/6
f/v
5/
f/v
w/W
/D
Q/z
/J
x/
/)
X/
1.12.1.. (at) constrictive contoids.
h/
s/z
s/z
S/Z
X/5
x/
./
//\
S/q
/=
/B
1.12.1.. Grooved constrictive contoids.
1. prelude
1.12.2. (Grooved and at) semi-constrictive contoids.
/d
/,
/y
/m
1.13.1. Approximant contoids.
F/B
/V
/j
/V
W/w
x/
/H
1.13.2. Semi-approximant contoids.
/b
/d
/%
/F
/=
/j
1.13.3. Lateralized approximant contoids.
/S
/>
+
/
/<
, semilateral approximants
semiapproximant
+
/<
10
/5
/r
/5
{/D
1.14.1. Trill contoids.
'/K
/R
/R
/e
/R
1.14.2. Tap contoids.

[/
3/r
1.14.3. ap contoids.
+
/[
r/m
{V}/e
1.15.1. (Bi)lateral contoids (and two semilateral (, ))).
/R
/l
/$
/L
+
/
/l
/L
+
/
/)
1.15.2. Unilateral contoids.
1.15.3. Lateral tap contoids.
+
/
+
+
/]
a/l
+
/
In 1.9-15, some orograms are marked with an eight-point asterisk; they indicate the few articulations which actually correspond to the canonical or ocial
ones. Near those marked with another slightly dierent articulation appears,
which however is represented with the same symbol. In these synopses, very precise symbols are given, in order to satisfactorily couple articulations and symbols,
1. prelude
11
although for some of them more common symbols can then be used (as can be
seen in 10.2-8 of the HPh]
In addition, 1.16-7, with their labiograms, dorsograms and palatograms, are
the necessary help to distinguish and observe well all the nuances, which contribute to dierentiate the contoids given there.
=+
1.16.1. (Two perspectives of) consonant labiograms.
mpb
Mfv
SZcG
[({
+
+
wu
&1
ji
1.16.2. Further consonant labiograms (only front view).

szqQ
td
kgx
1.16.3. Consonant orograms with dierent labial positions.
/B
1.17. Consonant dorsograms and palatograms.

grooved constr.
at consrtict.
approximant
(appr.) lateral
s, z , S, Z
, , J x,
, j, w
l, {L}
s, z
S, Z
T, D
(appr.) unilateral constr. unilat.
!,
!,
q, Q
c, G
12
@
F
B
b
> fv 5
M
>)
m[ mMM
p ( p
]
F >
7
@ b { b
&
S
" n
t 4
>d 7
>
q q
QQ
f w
v W
>
> sz sz
> d
>
d
l R
> n N n
t T
B
d D
> "
>
>
[n] N
>
Q
D z
R
V
e
l $
]
[5] [5] r
[R] [R] R

[
r
m
a
[] l
n
[T] [t]
[D] [d]
Cc & C cv
G 1Gg
~
+
_
> {D
> [
>
S X x . / S
> Z 5 \ q = B

>
n
N
\m ,
k
>
g
k%
w
k
>
>

>J
,
> j 3
% F
<
x
> )
ym
W
V w
=
j
&
> x H hH
<
'
K

3
r
1.18. Table of the contoids dealt with in the HPr. (@ indicates an intermediate articulation manner not present here, 10 of the HPh.]
bilabial
bilabial rounded
palatalized bilabial
uvularized bilabial
labiodentalized bilabial
labiodental
labiodental rounded
postalveolariz. labiodental r.
palatalized labiodental
uvularized labiodental
pro-dental
dental
denti-alveolar
velarized dental
uvularized dental
alveolar
alveobilabial
labiodentalized alveolar
alveolar semi-rounded
semi-paltaliz. alveolar semi-r.
velarized alveolar
velarized alveolar rounded
uvularized alveolar
postalveolar
postalveolar rounded
velarized postalveolar
postalveo-palatal (lowered tip)
postalveo-palatal protruded
postalveo-palatal over-rounded
postalveo-palatal (raised tip)
postalveo-palatal over-rounded
postalveo-prevelar
postalveo-prevelar protruded
postalveo-velar protruded
prepalatal
bilabialized prepalatal
labiodentalized prepalatal
palatal
postpalatal
postpalatal rounded
prevelar
postalveolariz. prevelar round.
provelar
provelar rounded
velar
velar rounded
labiodentalized velar
velarbilabial
velaralveolar
uvulo-postalveolariz. velar r.
uvular
pharyngealized uvular
pharyngeal
laryngeal
laryngeal rounded
1. prelude
13
Obviously, the total possible places of articulation are far more numerous than
those needed for our 12 languages. For each of them, in the respective chapter, a
table is given, which contains all the necessary contoids (and denitions) for the
neutral pronunciation of every single language.
However, the table in 1.18 provides these contoids (except 7 complex coarticulations, which appear among the orogram synopses, though), for a useful overall
comparison.
1.19 1.20 show the mechanisms of three particular types of explosion:
lateral, nasal, and inaudible, respectively for a stop followed by a homorganic lateral or nasal contoid, or else by another dierent contoid, with an intermediate
phase, made up of an'articulation with two simultaneous occlusions.
1.19.
Lateral () nasal
() explosions.
() tl, t, t
((Tl, T, T))
1.20.
Inaudible explosion.
pt ((pT))
() tn, t
((Tn, T))
kt ((kT))
1.1.5. Intonation is the most evasive aspect of languages and (perhaps because
of this) it is generally very badly and supercially dealt with, that is when it is not
completely neglected (which would, sometimes, be preferable rather than causing damage, or simply confusing readers).
In this synthesis, 1.21-3 will help us to present intonation, without useless
absurd and harmful complications. In fact, without scaring (but also without
baing or disorienting anyone), 1.21 very intuitively shows the normal pitch
movements, in neutral British pronunciation, of the four intonemes of English
and the two most frequent preintonemes (of its four).
An intonation group includes a given number of syllables (formed by phones/phonemes so they are phono-syllables), which are combined in order to constitute
some rhythm groups, made up of the words occurring in a particular utterance.
1
See you on
Satur
day.
se e you on
d a y?
[Wi ll I]
Sat ur
[y won't they]
1.21.
Iconic
tono-graphic
examples.
see you on Sat

urday?
[If I
don't] see you

total di
on Sat d ay [it'll be a
sa s
te r.]
ur
[If I
don't] see you

worry a
on Sa t urd ay [don't
bout
it.]
14
e most normal and frequent intonation group is composed of two parts: a preintoneme and an intoneme. Obviously, the preintoneme precedes the intoneme,
which concerns the last strong stress of the intonation group. ere are three marked
intonemes: conclusive /./, interrogative /?/, and suspensive //. A fourth unmarked
continuative /,/ intoneme completes the inventory. e rst two, as shown by the
rst two examples in 1.21, are respectively used to state or to ask something. is
kind of question is dened a total question, because the answer which regards the
whole question, in its totality must be Yes or No (or Perhaps, I don't know, ]
1.1.6. Our third example is a partial question, because it inquires about why (by
taking the rest for granted, or known) and its answer cannot be simply Yes or No.
us, there is only one part of the question, which is the one marked by the interrogative word (such as: when, why, where, what, who, how, ]
As can be seen from the movements in the third example, the intoneme that has
to be used, in partial questions, is not the interrogative one at all (as, however,
schools lead us to believe, with their grammars, and even the recordings of language-teaching courses!). On the contrary, the appropriate intoneme is a conclusive one, whereas its preintoneme is actually interrogative indeed.
As a matter of fact, a statement like en you come back again (as an answer to
a question like en will you tell us about it?) is intonationally dierent, from the
very start, from en'll you come back again? In fact, even before hearing // () 'll,
which is the only syntactical dierence, we can perceive that /'wEn/ (5wn:) is
already dierent in the two examples, because in the question it has a slightly rising pitch movement (while in the statement it is level), as can be seen from 1.23.
e interrogative preintoneme begins with / / ( ), whereas the normal one has
no particular symbol. (Here the space after the isolated symbol helps to identify
the preintoneme; but in actual examples it does not appear at all.)
e statement can dier also because it may have a slightly weaker stress and
dierent pitch: (&wn) up to (wn), and to /wn/ (wn) as well (with a clear segmental change). Here are the transcriptions of both sentences (including assimilatory coarticulation for /n/ (~) + heterosyllabic /j/, while /-n j-/ = (-n j-), as a compromise): en you come back again (5w~: j&khm'bk 'gn:3 3), (&w~5j;u, w~5j;u, w~5j;u), and en'll you come back again? (5wn j&khm'bk 'gn:3 3).
(As some readers might have observed, in our inter(dia)phonemic transcription
of English, we use both /l/ and //, although they are not in opposition, (l, );
whereas, an intraphonemic kind of transcription would give only /l/, with (l, ).
Instead, we also have // (, ), the latter occurring before // or /j/. Our transcription is diaphonemic too, since it also shows dierences between American and
British English, as for instance in no /'nOU/ ('n;)a ('n;)b.)
e last two examples illustrate the suspensive intoneme, //, which is used to draw
attention to what one is going to say (or not to say) in a kind of suspense, and the
continuative one, /,/, which instead does not produce this eect, as it simply divides
the utterance (just in order to continue). is is done either to avoid strings which
are too long, or to subdivide them into parts which present semantic cohesion between the elements of each group, in comparison with those of another group.
1. prelude
15
1.22. e four neutral intonemes of British English.
conclusive
/./ (2 ' 3 3) >6
interrogative
/?/ (2 ' 2 1) >
suspensive
// (2 ' 3 2) >
1.23. Two neutral preintonemes of British English.

normal
//()@
continuative
/,/ (2 ' 2) '@
interrogative
/ / ( ) @
1.1.7. We can now carefully see (or see again) the tonograms of the intonemes
and preintonemes given in 1.22-3. is is also useful to run trials in order to
check how much we approach or not what is shown there, depending on our
own spontaneous pronunciation, which may be more or less regional. It is likely
that the major dierences exist for the suspensive intoneme, which is the most varied and imaginative one.
In 2-13 we always also give the imperative, / /, and emphatic, / /, preintonemes. ese are respectively used, above all, to give orders, to curse or to exclaim, declare
Lastly, let us observe that a continuative intoneme can substitute a conclusive
one, when it is needed to attenuate the impact of the latter, in order to be less
abrupt, or nicer, as also happens in partial questions addressed to strangers, such
as at's the time? or at's your name?, or How much does it cost?
For many other things, including paraphonics (which concerns states of mind,
attitudes), it is necessary to refer the readers to the HPh or MaPI. We simply add
1.24, that shows pitch modications in (low or mid) parentheses and in quotations, that occur in the text of e North Wind and the Sun, which we have used
for the transcriptions given at the end of each chapter (before possible appendixes, as for English and French).
1.24. Tonograms of parentheses and quotations.
low parenthesis: ( ) / / mid parenthesis: ( ) / / quotation: (^ ) /^ /
1.1.8. 1.25 helps us understand the relationship tone languages have with intonation. As a matter of fact, the dierent tones (which are the realizations of particular tonemes, such as those of Chinese, for instance) obviously have their own
peculiar characteristics, as can be seen from 1.26.
erefore, intonation is added to the existing tones, by modifying them in preintonemes, according to what is shown in 1.25. In the tonograms there, the grey
parts indicate how the preintonemes slightly deform lexical pitches (by delimiting
the available extents). In comparison with an emphatic preintoneme, / /, the normal one, / /, is more compressed towards the middle part; an interrogative, / /, is
raised, whereas an imperative one, / /, is falling.
16
Obviously, also in intonemes there are some (even greater, 13.9 of the HPh)
modications, which are indicated by the corresponding tonograms, for Chinese
and Japanese ( 11-2; or in the phonosyntheses of the HPh, for the various tone
languages given there).
Naturally, what has been presented here about intonation holds for the languages treated, with all due dierences at a tonetic level which are indicated by
possible observations and respective tonograms. ese sections might seem to be
too short, but in actual fact, they provide all that is needed, whenever one adequately knows the tonetic method, which is an integral and necessary part of the
phonetic method (within natural phonetics).
e tonemic symbols, / . ? ,/, can be applied to all languages, since intonation functions are shared, although their actual tonetic realizations are extremely
dierent, as can be seen in the individual tonograms given.
1.25. e four preintonemes for tone languages.
/ / ( ) (( ))
/ / ( ) (( ))
/ / ( ) (( ))
/ /( ) (( ))
1.26. e four basic ton(em)es of (Mandarin) Chinese.
1 /5/ (5) >1
2 // ('1) >Q
3 // (2) >5
4 /6/ (63) >Z
Transcriptions
1.2.1. en it comes to analyzing the pronunciation of a language, for learning or teaching purposes, it is necessary to use two fundamental types of transcription: phonetic and phonemic.
Both of them must start from real interlinguistic strategies, rather than from intralinguistic considerations. In fact, an intralinguistic transcription has the sole
aim to be used for just one language and above all for native speakers. erefore, it is limited to what is merely phonemic, without concerning comparisons
with other languages.
All this is quite legitimate, though, when no connection with other languages
is felt necessary. In this case, even very generic symbols may be used, provided each
phoneme has a dierent symbol from those of all other phonemes.
In theory, it could be sucient to transcribe the English diphthongs (Ii, I, a,
, a, /, u/Uu) (here American variants appear after a slash), simply as /i:,
e:, ai, oi, au, o: u:/ (if not even as /ij, ej, aj, oj, aw, ow, uw/). According to this
way of thinking, German (ae, ao, OY) are rightly indicated as /aj, aw, oj/, or
even Spanish or Italian (ai, au) as /aj, aw/.
1. prelude
17
1.2.2. However, it is immediately clear that dealing with descriptions and teaching indications such as /aj, aw/ are rarely faithful and highly misleading, for such
dierent realities as English (a, a) (with centralized and non-high second elements), German (ae, ao) (with peripheral and non-high second elements), and
Spanish or Italian (ai, au) (with peripheral and decidedly high second elements).
Besides, transcribing German (OY) as /oj/ would mean completely ignoring
(or concealing) the fact that in neutral pronunciation also the second element
is rounded (and neither fully high nor front). For English it would be desirable
once and for all not to continue to conceal the evident fact that not only are (I,
/) real diphthongs, but that (Ii, u/Uu) are diphthongs too.
Instead, interphonemic transcriptions seek to take into proper consideration
the characteristics of each language, even though within a less rich and less precise
symbol inventory, which is however capable of using the similarities and dierences among the various languages in a better way.
erefore, we have: English /aE, aO/, German /ae, ao/, Spanish or Italian /ai, au/.
A careful analysis of the vocograms in 3 5 will surely explain the notational
dierence between English and German. To complete the series given above, we
have English /Ii, EI, OE, OU, Uu/, and German /OY/.
1.2.3. e most eective phonetic transcriptions are the taxophonic ones. ey
resort to every useful symbol (among the available ones for the dierent phones:
vocoids and contoids) and all the most precise prosodic elements, in order to manage to show the necessary nuances. Only in this way is it possible to make real and
valid comparisons between dierent pronunciations, either regarding dierent languages or dierent areas for the same language. Otherwise, everything becomes
approximate and denitely less useful, since people may think they are working
well, while generally they are simply deceiving themselves.
Even excessive simplications, to help learners, are not the best solution in order to teachlearn pronunciation really well. For instance, when English phoneticians continue to use () even in more accurate transcriptions (perhaps with diacritics), for all occurrences of // (and perhaps /U/ too), they miss the opportunity to show reality, as when to go the man further are rendered as (t'gU, 'mn,
'f:) (or ('f:)) in British pronunciation, instead of actual (T'g;, 'm;n,
'f;) (as happens to the quite often reported cases of (i:, u:) for (Ii, u/Uu)).
1.2.4. erefore, for those who use several languages, but also for those who utilize one foreign language only, the most recommendable phonemic transcriptions
are the interphonemic ones, because they use symbols in a less arbitrary way. In
fact, they do not atten reality, by only showing what is functional, but compare
dierent languages better, by considering similarities and dierences as well.
In Japanese, it is certainly better to use /M/ rather than /u/, even if there is no
possibility of confusing them, since this language has no /u/. However, the use of
/M/ highlights the dierences from other languages which do have /u/.
e kind of diaphonemic transcription is important, as well. It generally rests
on an interphonemic basis, although an intraphonemic basis is possible, too (but
18
with all the imperfections that this inevitably implies).

diaphonemes are fundamental to transcribe partially dierent accents of the
same language simultaneously. In fact, by using a sole kind of transcription and
occupying less space, it calls to mind that there are dierences and indicates systematic variations (with no need to repeat the same words but only change the symbols of the parts that really dier).
In this way, for instance, /Uu, OU/ represent (and trace back to) (u, ) (Br.) and
(Uu, ) (Am.): two /'tUu/ ('Th;u, 'ThU;u), go /'gOU/ ('g;, 'g;).
Equally we have: /;/ for (A:) (Br.) and for () (Am.): last /'l;st/ ('lA;sT, 'lsT); or
/;/ for () (Br.) and for (O:) (Am.): lost /'l;st/ ('lsT, 'lO;sT); and /, / in /'lE/ for
('lT) (Br.) or for ('lm) (Am.).
1.2.5. atever the phonemic transcription used, to go on to the phonetic one
(and then to an adequate pronunciation) each distinctive element each phoneme must be considered as one of a given number of xed points in the phonemic space, either vocalic or consonantal, of a certain language.
Each of these points, or elements, is necessarily dierent from all the others because of its function, and is also dierent from a zero occurrence. For instance,
// is dierent from /I/, or from //, , but it is dierent from /`/ too. As a matter of fact, beat /'bIit/ diers from heat /'hIit/, but they are also dierent from eat
/'Iit/ (or //'Ìit//, to show the relationship more clearly).
Even the opposite process the decoding of an oral message is carried out
according to these principles. Each phone of a given utterance depending on the
global meaning and the speaker's phonic system as well must be assigned to an
agreed phoneme (as a realization of it). en this has to be placed into a specic
phonemic space, as already said.
1.2.6. It is important to keep in mind the dierence between phonemes, phones,
and sounds. phonemes have a distinctive value within a given language, since it
can change the meaning of words, as for instance: (I can) read /'<Iid/ and (I have)
read /'<Ed/, or hit /'hIt/, hat /'ht/, hut /'ht/, hot /'ht/. As already said, a phoneme
has the function of distinguishing itself from all other phonemes of a given language, that is of not being what the others are: it is form.
phones, instead, have an identifying value, in one or more languages, since they
contribute to characterize pronunciation, by using more or less typical and recognizable segments, beyond simple phonemic representations. In a language, or in
an accent, the phoneme /s/ can be articulated as dental, (s) (as in Latin American
Spanish, or in neutral Italian), or as alveolar, () (as in Castilian Spanish, which is
used in central-northern Spain, or in Greek, Danish, nnish, or very often in
northern Italian).
In English and French, /s/ is more often realized as denti-alveolar, intermediate between (s) and (), which can be represented by (s); but it can be appropriate
to use a (more) specic symbol, (s), to show the dierence at least in the early
stages of learning in order to demonstrate its actual realization, even though it
can hardly be perceived (all the more so because quite a few native speakers of
1. prelude
19
these languages use the dental type instead). A phone has the function to maintain
consistency between the elements of a given pronunciation: it is substance.
1.2.7. On the other hand, sounds have their value exclusively in transmission,
their function being to enable human communication through sound waves.
erefore, a sound is a single emission, in practice unrepeatable even by the same
person. It can oscillate quite a bit, often producing quite dierent realizations: it
is matter.
As can be seen in 2.4 of the HPh, one way to allude to the fact that actual
sounds are always a bit dierent would be to represent them with dierent fonts:
o, o, o, o
In conclusion, many dierent yet similar sounds constitute a single phone.
en, in the context of a particular language, several phones, not wholly identical
(but with the alternation governed by xed and systematic rules, which can and
should be discovered and then explained simply and completely), constitute a phoneme. e phones referred to by a given phoneme are called taxophones (or combinatory phones, or allophones {a more ambiguous and less advisable term, since
it implies modications not necessarily due to combination, but simply any sort
of dierence, for any reason, whether general or random}).
Sometimes, in the course of listening to recorded materials, certain sounds still
need to be disregarded in the process of constructing the phonetic inventory of
that language. is is because it is possible for single speakers to occasionally produce sounds presenting abnormal deviations, whereas it is essential to consider
mainly what is more typical and frequent.
1.2.8. erefore, actual sounds are practically innite, and phonetics and phonology would be decidedly complex if it were not possible to rely on the systematicity of phones. In fact, the set of phones is the result of a past classication and
structuring of sounds (abstracting from the unrepeatability of the same sound).
is set allows us to reach recognizable types, which can in turn be represented
through precise phonetic symbols, regardless of any particular language.
us, what is similar in dierent languages and dialects is realized by phones
and their symbols, which make it possible to compare dierent languages (an essential point in order to be able to learn and describe those languages).
erefore, the phonemic representation of single languages necessarily uses a selection of (phonemic) symbols, with distinctive functions and purposes, even if
of course some symbols may be the same but with rather dierent phonetic values.
us plain phonemic symbols (generally chosen among the most common, as
the ocial IPA ones, International Phonetic Alphabet, or oIPA), do not represent pronunciation exactly, but rather the relationships between the phonemes of
a given language.
ey are useful for the specic purpose of keeping the current writing system
distinct from the phonic level. erefore, phonemic transcriptions make it possible to avoid interference deriving from not knowing orthographic rules, or from
20
the inconsistency of these rules.

However the exact pronunciation, for purposes of learning and teaching, can
only be shown with the careful, consistent, and systematic (because normalized]
transcriptions of a phonetic alphabet such as canIPA.
1.2.9. In the books which do not completely ignore it, intonation is usually
treated after vowels, consonants, and stress (and other prosodic characteristics,
such as length) this is due to the greater diculties involved in describing intonation. However, intonation should not be ignored, or relegated to the end in teaching since it is inseparable in actual language from the other elements.
erefore, phonetic transcriptions, in the strict sense of the word, simplify reality a bit by indicating it only partially; in fact, it is as if they indicated in any
case a continuative intoneme, as in Italian ('tan:to2), ('tEm:po2), (a'ko:Ra2). (e
dot at medium height indicates the presence of intonation, even though of an unmarked type. e symbol (:) denotes length.)
However, in tonetic and phonotonetic transcriptions, the notation (') represents
not only stress but also a mid-range pitch (both in the context of tones and in that
of intonation) it is thus in contrast with other signs such as (5), (), (6), (), .
e examples given also in the chapters on the 12 languages are of a phonetic type, with possible durations, but without intonation. For instance, to a French
transcription like (p'i) /pa'Ki/ Paris we will have to add an adequate pitch, as if
it were say in a conclusive intoneme. us, we ought to have at least (pi)
(since another slight dierence exists, 4.3.5).
However, the low pitch on (i) depends on its occurring in the stressed syllable of a conclusive intoneme; thus, in actual fact, we ought to transcribe it as (pi3 3), that gives us intonational information, which has to be added when necessary (as for a conclusive intoneme), provided we already fully know the intonation
characteristics of the language.
1.2.10. erefore, it is certainly useful to have right at the very beginning of
the study of the pronunciation of a language at least an idea of the pitch movements of that language. One ought to manage to (re)produce them adequately and
be able to nd their tonograms again easily, whenever they are useful (regardless
of laziness and the illusion one does not need them).
However, by indicating pitch in transcriptions of isolated words, for non-tone
languages (which simply means without any ton(em)es, certainly not without
intonation!), people could be led to think that all stressed syllables of French are
low-pitched whereas the other French intonemes and preintonemes are not so.
erefore, in conclusion, a realistic solution to the problem states that, for all
non-tone languages, intonation is marked only in sentences, which are given as examples of actual utterances. On the contrary, for tone languages (although, again,
intonation is not generally given), it is necessary, instead, to show all the ton(em)es, since they are a part of the phonemic system itself, which is distinctive,
not of intonation (although both ton{em}es and intonation rely on pitch).
As a matter of fact, tonemes are real phonemes (not at the segmental but su-
1. prelude
21
prasegmental level), as can be seen from the examples associated with 1.26:
(5ma) /5ma/ ma mother, ('1ma) /ma/ m hemp, (2maa) /ma/ m horse, (63ma)
/6ma/ m scold, curse (In 11, we will see that Mandarin Chinese also has a zero toneme and very important taxotones, too.)
e contents of the HPr ( of the HPh)
1.3.1. e HPr is about applied phonetics, since (as has been said in 1.1.1) it
gives accurate and fairly extended descriptions of 12 languages ( English, Italian,
French, German, Spanish, Portuguese, Russian, Arabic, Hindi, {Mandarin} Chinese, Japanese, and Esperanto) by putting into practice what is explained in the
HPh. We thus include the crucial phonemic component constituted by functional phonetics which is realized with phonemic transcriptions given together with
phonetic transcriptions, so as to show the relationship between phonemes and
phones
Together with these two types of notation, the ocial spelling of each language
is provided (or a transliteration in the case of the four Asian languages). Spelling
is given last, so that it will interfere as little as possible with the ecacy of the phonetic method and the associated transcriptions thus the reader will come upon
the spelling only after having xed in mind the phonic structures. As will be seen,
pronunciation variants are also given, and these will be worth studying.
Natural phonetics is an artistic science, and the best way to extend its rich potential concretely (after the necessary fundamental premises) is through applying its
techniques and knowledge to languages which are in great demand for teachinglearning. In fact, these languages can be taught to others, or they can be self-taught (while remembering that, in order to teach adequately, it is necessary to
have learned well rst).
erefore, it is best to begin the course of phonetic analysis with one's own language. e language one speaks is frequently a regional variant, more or less strongly marked. Consequently, it is important to understand how neutral pronunciation operates and how it diers from one's own pronunciation. is method helps
to build consciousness of one's own speech patterns. ese patterns can then be
compared with those of the neutral pronunciation, which exists in every language
of culture, even though speakers and society rarely have a consistent and precise
idea of what it is.
1.3.2. In every society, at least when there is a written language and literary production, a neutral pronunciation exists, which is the pronunciation used by professionals in public speaking (particularly actors, presenters, and announcers of high
quality). e Danish linguist Otto Jespersen declared that the best language is
spoken when the social and regional origin of the speaker are least obvious. Neutral pronunciation brings this miracle to pass, sometimes even to the point of
making the identication of social and regional origin completely impossible.
e supporters of regional pronunciations, who encourage individuals to
22
maintain their genuine pronunciation (even when they carry out public professions), deceive themselves as well because of their lack of success in bettering
their own pronunciation (a task which would have required commitment and persistence).
In other cases, they fail to understand the importance of being able to free themselves from a heavy burden (often extremely distasteful to the individual in question, at least on an unconscious level), all the while without renouncing their origins or identity. In fact, regional origins can be exhibited more eectively by alternating between one's dialect and the national language (competently, and above
all by choice). e least preferable approach is to remain prisoners of an uncontrollable hybrid, which is neither the ocial language nor a local dialect. In Italy and
the German-speaking countries, dialects dier largely not only in terms of pronunciation but also of grammar and vocabulary too; in fact dialect rigorously speaking refers (or should refer) to these other dierences, while accent predominantly involves dierences in pronunciation.
In the other 12 chapters of the HPr, therefore, we will apply the phonetic method, which consists of a careful and rigorous comparison of the phonemes phones
and intonation of the languages described, listening to good audio recordings, using the two types of transcription (etic and emic), analyzing the phonetic gures,
and showing commitment. e last quality is clearly indispensable: it is not enough
simply to want to know a sustained eort to learn according to an eective method is also required.
1.3.3. It is very important to attain greater precision in identifying and representing the true realizations of each phoneme. is task therefore implies phonetic transcription with the aid of a symbol inventory endowed with a sucient number of taxophones, so as to be able to represent the actual phonetic reality, and not
merely what this reality is assumed to be.
In this way, the articulatory basis, the collective phono-tonetic habits inherent in each language, is implicitly made evident, without the need for specic and
complicated eorts (more targeted and burdensome, but normally leading to
less satisfactory results). us it will not be necessary to transcribe the partial nasalization each time in cases such as thanking ('k) ((('k))), unless it should
become more noteworthy, in which case it could be transcribed as ('k); the
same holds for words like none ('nn:, 'nn:) {(('nvn:, 'nn:))}, if pronounced dierently from ('nn:, 'nAn:). It will be more than sucient to observe, once and for all,
that a light nasalization in such cases is practically inevitable, in contact with following () segments.
In the case of intonation, mechanical instruments are capable of extracting and
measuring single characteristics (through a series of separate processes), but they
end up confusedly mixing up the various components. Expert human perception,
on the other hand, can sift out superuous elements and concentrate on what is
truly important.
For these reasons, it is impossible to accept without reserves the raw melodic
curves produced by acoustic methods. It is necessary instead to lter them in imi-
1. prelude
23
tation of the process employed by the human ear, which manages to compensate
in an extremely eective way for the many irregularities, whether objective or incidental, which can occur. e results are then standardized in tonograms, by taking a sort of average of large numbers of utterances.
Observations on phonetic terminology
1.3.4. In the eld of terminology, as well, scientic rigor is a great help, while
the vagueness and lack of clarity of certain obsolete traditions with little scientic
foundation is decidedly negative. On every page of the HPr ( HPh, as well) this
requirement is addressed continually indeed, everything becomes simpler and
easier to understand when the technical terms are clear and intuitive (even more
so than in normal speech). For example, tonic should refer only to tone pitch
and not to stress.
A diphthong should contain only vocalic elements, such as in ('ai), not vocalic
and consonantal elements together, as in ('ja). Were it otherwise, it would follow
that ('la, 'ma, 'sa) would be diphthongs as well in fact, while (a, i, u) are vocalic
elements, (j, l, m, s, t, r, h) are consonants. us, contrary to the opinion broadly promulgated by grammarians, Italian ('ia) for instance is a true diphthong as
well ( 5.2-3 of the HPh).
e concept of a phonetic syllable also remains overly subject to the inuence
of writing systems and of grammatical and metrical traditions. It is natural that
books on linguistics and dialectology should speak about the phonic side of things
as well, using transcriptions. However, a minimum of rigor would be of great benet, since otherwise there is the risk of spreading and reinforcing unmitigated errors, which compromise and discourage eective learning.
For example, with the third millennium already here, books are still published
which give the provisory IPA table (of 1993), naturally with the mistakes included as well (corrected in 1996), when it would have been a simple matter to download, or refer to, the (currently) denitive table shown on the ocial website. Obviously, this would only bring us to the level of the ocial position, which is hardly satisfactory; but at least major errors and embarrassingly naive missteps could
be avoided (within the limits of the incompleteness and ingenuousness of the ocial table) ere are also those who succeed in producing brilliant achievements
X ( the monogram with a sort of umbrella on top, in the place of the
like /c/
normal /c/, or the possible /tXS/).
As we have said in 8.5 of the HPh, we prefer not to avoid the perfectly grammatical use of suxal comparative and superlative for front and back, contrary to
mostly non-phonetician common usage.
1.3.5. It must be quite clear that doing phonetics means giving a symbol to a
sound. is task is not so trivial as the man on the street and perhaps acoustic phoneticians and theoretical phonologists might imagine. In fact, giving a symbol
to a sound necessarily implies several successive and linked phases, in which hear-
24
ing, mimicry, kinesthesia, comparison, adjustment, and mnemonic storage are all
activated.
rst of all, it is necessary to be capable of suciently perceiving the sound, so
that it can be identied with a precise phone which can adequately represent it.
e next step is to be capable of reproducing the sound using the appropriate
phone, which is carried out especially through imitation which is immediate, ,
occurring promptly after hearing the sound.
e third step, which is indispensable, is to produce the phone using kinesthesia ( consciousness of the necessary articulatory and phonatory movements), even when the auditory stimulus is not close at hand. However, auditory memory
can still be an important guide, whether it is particular, referring to a specic
sound in a given language, or general, involving comparison with similar phones
with the help of experience in listening to and producing the phones of many languages.
In this manner, it is possible to produce a particular phone even days, months,
or years after having heard it (and to be reasonably sure of producing the correct
phone, even in the case of languages which one has never heard). e secret of
good notation is for it to be realistic, and consequently, truly useful.
As a matter of fact, the fourth point, which is fundamental and decisive, is as
our denition suggests nding a way to symbolize the specic phone by choosing the most appropriate symbol out of several hundred (not just a few dozen) elements. If, after careful consideration, none of the available symbols is capable of
decently representing a particular phone, it becomes necessary to nd its position
with regard to all other known phones. In this way, it will become possible to decide whether it is truly a new phone in need of a new symbol. If a new symbol is
necessary, it can then be designed by following the general criteria of necessity, distinctness, and availability (as can be seen in the HPh).
1.3.6. erefore, doing phonetics means managing to truly enter the phonic
system of one or more languages, thanks to a rich array of symbols as well. e
symbols of the ocial IPA (oIPA = an o alphabet! 7.4 of the HPh) are not
at all sucient, and they create the illusion of successfully doing phonetics, while
all that has been achieved is at most a bit of phonology. Far too often, these studies proceed without the minimal understanding of what the phonetic structure in
the case in question actually is.
Clearly, it is phonology which is a part of phonetics ( 1.9 and above all
3.1.3 of the HPh), and not the other way round as some believe. In fact, within
the rubric of phonetic analysis and description, there is a functional component
as well. erefore, functional phonetics (or phonemics) is an indispensable part
of phonetics, but only a part. ere is little which could be done working with
phonology alone, just as there is little which could be done with just acoustics. Instead, what is needed is a global vision: articulatory, auditory, functional, descriptive, and contrastive (while verifying various characteristics acoustically).
On this subject, it is interesting to note that the phonotonetic data of the HPh,
and of the HPr, have been compared with a sizable collection of acoustic data from
1. prelude
25
various sources, or at times, with data from a single source furnished by dierent
authors. In practice, the correspondence is dramatic, not only with data from a single source, but also with data from multiple audio recordings analyzed accurately
and subsequently normalized. en we speak of normalized data, we refer to the
practice of averaging the results of multiple speakers and many utterances in various contexts, while keeping in mind phonological considerations and excluding
inappropriate samples.
e discoveries of sociolinguistics also require normalization; otherwise there
is a signicant risk of irremediably creating confusion, even when scientic data are used. is fact has been responsibly demonstrated in several recent works.
We will not cite these sources (as examples to follow), simply because this ought
to be the normal way of doing things, not the alarmist or scoopist manner of
far too many publications.
1.3.7. According to this global framework, every phonic system is an autonomous and complete organism. It contains phonemes, with their taxophones, as well
as prosodemes, with their particular realizations (concerning duration, stress, tones,
and intonation). To give a simple example, a vocalic element of a language, no matter how similar to an element of another language, must be considered in relationship only with the other vocalic (and consonantal and prosodic) elements of its
own phonic system, within its own phonemic space ( 1.5 of the HPh]
erefore, if it is necessary to codify (pronounce, or transcribe), and also to decodify (listen, or trans-read), it is essential to constantly refer to only elements from
the system of the language in question. e term trans-read should be taken quite
seriously, since it means to read a transcription appropriately, using the actual
phones (as well as tones and intonation) of the language transcribed. It does not
mean at all to take a shot at reading a transcription, using merely the phones of
one's own personal accent.
1.3.8. erefore, in the pronunciation of a particular language, it is essential to
use only the phones and prosodic elements of that language. Instead, it is almost
a rule that people use the phones of their mother tongue, adding a few phones
from the other language because these are not present in their own language. Of
course, the rough idea is to complete the inventory when things are missing, but
ideally all work should take place within a single system, even with parts containing elements which are similar in the two languages.
In fact, no matter how similar they may seem, the elements of one language are
never exactly like those of another due to the relationships with other elements,
if nothing else. For example, the Italian /i/ is similar to the Spanish, or Portuguese
(Brazilian or Lusitanian), or French ones. However, the Spanish /i/ is in opposition with only four vowel phonemes (/e, a, o, u/), while the Italian or Brazilian /i/
is in opposition with six (/e, E, a, O, o, u/), and the Lusitanian /i/ is in opposition
with eight (/e, E, a, A, O, o, u, e/, which in an exclusively Lusitanian not diaphonemic system, would be represented by /e, E, a, , O, o, u, /, together with /i/, of
course).
26
At rst sight, the Brazilian system could seem exactly the same as the Italian
one, as both have /i, e, E, a, O, o, u/. Instead, the two systems are dierent, particularly since in Brazilian Portuguese (and also in Lusitanian, but with additional
dierences), even nasalized taxophones are expected ((i, e/, A, 9/, u), when followed by (), a nasal consonant element); without these nasalized phones the
pronunciation would not be genuine (while in Italian it would be a regional one
if with nasalized vowels).
As for French, /i/ is in opposition with fourteen other vowel phonemes (/e, E, a,
O, o, u, y, , , , , ^/ and //, , the traditional //), including the four nasalized vowels (/, , , ^/), which are true phonemes, in French, and not mere
taxophones.
Apart from these considerations, the actual phonetic realizations are not exactly the same, even though we use the same phone (i). e dierences can be seen
by comparing the vocograms (or vowel quadrilaterals) of these languages, in 213. e same is true for the other elements which correspond.
1.3.9. If we now consider German and English, even without going into too
much detail (the details are available, of course, in 2 5), it is clear that we
will have to take into account phonemic vowel duration. Instead, in the Romance
languages we have considered, vowel length is practically only phonetic (although it varies considerably from language to language; consult the specic chapters, or to see the dierences more rapidly, consult the transcriptions at the end of
those chapters).
In German and in English, the /i/ (short i) is considerably more open than
in the Romance languages, giving (I) in German and () in English, respectively.
However, even more pertinent to the present discussion of phonic systems is the
fact that in the Germanic languages, the opposition of duration is relevant (,
phonemic, distinctive) as well: German /I, i:/ (Schi schief /'SIf, 'Si:f/ ('SIf, 'Si:f)); English /I, Ii/ (bit beat /'bIt, 'bIit/ ('bT, 'bIiT) as we have observed repeatedly, we prefer a less abstract type of notation than the predominating form, which less usefully continues to give /i:/, even for English, ).
at emerges quite clearly from this discussion is the sorry state of those grammar texts and language courses which describe the German /I/, or worse still, the
English one, as a short i, as in French vite, or in Spanish listo, or in Italian tto.
en one considers that many Italian regional pronunciations have a vowel in tto which is not at all short, the absurdity is evident! Of course, the same is true
when one tries to teach the close and short French (or Spanish, or Italian) /i/ (i)
by citing the same old misleading example: machine. Often, silence is golden
1.3.10. Foreign and regional accents result from pronouncing a national language according to the phono-tonetic system of a particular and recognizable area
( of a particular and recognizable social group), especially in Italy, and in German, Spanish, or Portuguese speaking countries. erefore, it is important to begin to carefully examine one's own pronunciation, so as to be in a position to work
towards the goal desired. is task involves learning to analyze one's own sounds,
1. prelude
27
classifying them into precise phones (and transcribing them with adequate symbols), and then seeing how these phones are in turn part of particular phonemes.
At this point, it is indispensable to have a reliable and careful description available for the system of the language one is seeking to learn. e necessary comparisons can then begin they should be carried out objectively and systematically.
Obviously, it is imperative to follow an extensive series of specically directed exercises, followed by careful checks. ese exercises consist of listening to good recordings and recording oneself to perform merciless examinations of how much progress has been made without cheating, otherwise all eort is wasted.
For the pronunciation of Italian, Italians can rely on the seven chapters of the
MaPI treating regional pronunciations for the initial diagnosis of how regional
their pronunciation is. To (attempt to) achieve a neutral Italian pronunciation,
Italians (and foreigners as well) can use the rest of the MaPI (including the two audiotapes which come with it), together with the DiPI. ese two formulae refer
to two works by the present author (Manuale di pronuncia italiana {A Handbook
of Italian Pronunciation} and Dizionario di pronuncia italiana {A Dictionary of
Italian Pronunciation}, in the bibliography).
1.3.11. In this handbook, we suggest the use of Esperanto ( 13: a planned
and auxiliary international language, which means it is a suitable second language
for all {certainly not the only language, which would be absurd}), in order to apply the phonetic method. e aim is to manage to produce the sounds and intonation of Esperanto, which are denitely simpler than those of other languages.
(Of course, the comparison holds even more for morpho-syntactic and lexeme-semantic aspects, as well. In fact, the structural comparison at all levels, between Esperanto and some studied languages, including one's own national language, is very revealing. Above all, it is a good preparation for the serious problems of interference, by helping one to see them concretely and systematically, not
only as occasional inconvenient troubles. Language teaching ought to acknowledge this opportunity and use it accordingly.)
It is a doubly useful drill, since it requires just a limited commitment, which
should, however, be taken seriously. Besides, it is very unlikely for anyone to have
heard neutral Esperanto before. In fact, even if one happened to hear somebody
speak this language, certainly the pronunciations used must have been individual,
regional, ethnic, or national. erefore, one has to approach neutral Esperanto
pronunciation relying exclusively on the descriptions, gures, and explanations
provided here, in 13.
en the dilemma of self-evaluation can arise: how can people be sure they have
really attained their goal, which means neutral Esperanto pronunciation? is is
also a part of the phonetic method, where mere imitation (with all its problems of
delity and authenticity) is not the main aim at all although, of course, for natural languages, it is always fundamental to have a reliable sound-model to follow. On
the other hand, we know fully well that transcriptions are reliable models too, as
much as recordings (and sometimes even more), provided they are faithful and serious, mainly if they are supplied with intonation, objectively and realistically
28
marked.
However, the real strength of the phonetic method resides exactly in being aware of what to do in order to attain the (hopefully neutral) pronunciation of a
certain language, or of several languages. is is so because we are not in the position of a child any longer, when complete phonotonetic (and phonotonemic, and
also paraphonic) learning was right, including certain regional peculiarities, which
could be happily renounced but that is the biological method!
nally, awareness has this further advantage as well it allows one to choose
what is thought to be most advisable, between various possibilities.
Guide to the gures
1.4.1. e orograms of the HPr (and HPh) use symbols according to certain conventions, which make the orograms easier to understand (and tell apart). It is
therefore important to become familiar with these conventions, so as to be able to
make the best use possible of the rich iconic framework they provide. It is dicult
to understand why some phonetics texts contain only a small number of illustrations, or even none at all. It is certainly true, however, that it is better to give no
illustrations at all rather than to provide imprecise (or erroneous) ones. It would
be better still if certain books were not produced at all
In the vowel orograms ( 1.7), it is of great importance to pay careful attention to the location of the marker indicating the center of the back of the tongue.
It is even more important to take note of the precise location within the white (or
transparent) miniature vocogram, placed at the center of the oral cavity (with respect to the still more precise location seen in the normal, larger vocograms), and
the shape assumed by the entire back of the tongue. In this way, the various vowel orograms can be compared (or just a subset of them, such as those relevant for
a particular language).
All this eort should lead to a real understanding of the vocoid articulations and
of the dierent movements of which they are composed, so that an active panorama is produced in the mind, and not simply a passive vision of things. In phonetics, merely passive and memorized knowledge has little use except to confuse
and discourage people!
Obviously, the true analysis and description of the vocoids of a given language
occur through the use of the large vocograms (vocograms in the strict sense), given that these diagrams succeed in showing nuances very precisely (as can be seen
in 2-13, or in the phonosyntheses of 16-23 of the HPh, as well as in the MaPI
and the various Italian regional pronunciations given therein).
1.4.2. erefore, we will now consider what can be found in vocograms, which
should be observed, analyzed, and scrutinized calmly in all of their particulars. Vocograms are full of details, without which it is impossible to come close to the
spirit of a language, manifested especially through vocoids, then through pitch,
and nally through contoids. Even a single millimeter makes a notable dierence
1. prelude
29
on a vocogram (as well as on an orogram or on a tonogram).

is is the magic of phonetics. In fact, those who fail to perceive it accuse phonetics of being cold, dry, incomprehensible, dicult, and useless to boot. Instead,
it is extremely useful, which is fundamental, and fun as well!
Vocograms are subdivided into 30 boxes, where the appropriate markers are
placed according to the shape imparted to the lips. Round markers denote lip
rounding (as in (u, o, O)), and square ones denote normal lips (or spread lips; in
any case unrounded, for example (i, e, E, a)), 1.4 1.7-8. It is also true that
the vowel orograms (in the miniature vocogram part) also contain circular or
square markers according to the lip position. However, it is clearly much easier to
see the markers in the (large) vocograms, where it is essential to use them properly ( 1.27).
It is possible to use the diamond markers as well (squares rotated by 45: $) to
show lip positions which are half-rounded, halfway between round and normal (as
in (), 8.10 of the HPh), or for perfectly coinciding rounded and unrounded
V (as (, %), 1.27).
Besides the shape of the markers, their content and shading are also important.
ite markers (where the normal shape is not lled in by any shading) represent
unstressed vocoids, such as English (), or for example those represented by >o in
the following Italian examples: poich grido (pi'ke, 'gri:d) (or possibly, in certain languages, half-stressed vocoids, but not fully stressed ones). Solid black markers represent vocoids which are always stressed, as in yes ('js), or in Italian no ('nO).
Naturally, there are also vocoids which can occur stressed or unstressed; for
these, the symbols are black-and-white ( black with a white center), as in singing
('s) or in neutral British English lover ('lv), while we have lovers ('lv, -),
or neutral American English lover ('lv) and lovers ('lv, -).
1.27. Dierent markers for vocoids.
half-rounded (), or coincident (, %)
unrounded:
rounded:
unstressed /u/ ()
unstressed /i, / (i, )
stressed or unstressed /U/ ()
stressed or unstressed /I, / (, )
stressed /E/ ()
1.4.3. Markers can also be lled with grey in order to indicate variants (contextual ones the fundamental taxophones, pronounced using special phones; possible ones, such as those used in regional accents geophones; or those related to various social groupings sociophones). In neutral British English, for instance, there
are quite a few taxophones, especially including those produced when there is a
following () for example in hut hull /'ht, 'h/ ('hT, 'h:), or feet feel ('fIiT, 'fi;I)
(while feeling has ('fIil)). It is impossible to continue to ignore such variations in
vocograms and phonetic transcriptions.
A representative example of Italian geophones is given with the varied realiza-
30
tions of the phoneme /a/ (a), which include: (, A, , , , , ). Some of these realizations can constitute sociophones, since they are more commonly found in
broader (instead of less broad) regional accents (these details can be found in the
MaPI in both cases). Grey markers can also have white centers in cases where they
refer to vocoids which can occur unstressed as well.
At times, it can be necessary to improvise a dierence in the marker or in its
shading in order to represent important realizations which depend upon the position in the word with respect to word boundaries, stress, syllable structure, less
common use, or simple occasional variation. In this manner, it becomes possible
to avoid the use of supplementary vocograms. e purpose of these special markers will be explained clearly, whether in a text placed close to the vocogram, or in
the main treatment. e most common convention is the use of dashed lines, particularly for unstressed vowels which are represented by ordinary unshaded
(white) markers.
For examples of these rules in practice, the reader is invited to consult the vocogram of the international French accent ( 4.4.1), southern French (
4.4.3), or German (with its various accents: 5), or Brazilian Portuguese, Russian,
Arabic ( 7-10).
1.28. Markers for variants.
unstressed // ()
stressed or unstressed /I, / (, )
stressed /E/ (e, E)
unstressed /u/ (%)

stressed or unstressed /U/ (P)
1.4.4. But let us now proceed to the conventions concerning diphthongs

(formed by two tautosyllabic vocoids , two vocoids contained in a single syllable, 5.2-3 of the HPh). Diphthongs are denoted with the appropriate marker
for the starting element, together with an unbroken black line which moves all the
way to the precise location of the second element ( 1.29-31).
If the endpoint is an unrounded vowel, nothing is added; instead, with rounded vowels, a minuscule black dot is placed at the end of the line. If the endpoint is
half-rounded, the marker at the end is a diamond this marker is thus analogous
to the normal-sized marker that would be used for a half-rounded rst element ($).
Meanwhile, the phonemic and phonetic transcriptions completing the information provided are placed around the vocogram.
1.29. Diphthongs (occurring stressed and unstressed).
wide (ai)
narrow (a)
wide, with rounded

2nd element (au)
narrow, with rounded
2nd element (a)
monotimbric (aa)
1. prelude
31
1.4.5. Diphthongs are considered wide, when their gure contains a fairly long
line, and narrow, when the line is rather short. Besides these ditimbric (two-timbre) diphthongs, beginning and ending with distinct vocoids, monotimbric (one-timbre) diphthongs also exist, with the second element in the same box as the rst
element, but at a slightly dierent point of that box.
is last group consists of quite narrow diphthongs (which therefore are almost
like long vowel phonemes), and often the line is extremely short. Because the distance is so short, a dashed line can be reduced especially in such cases to a single
short segment, or to the black dot alone for diphthongs with rounded second elements.
Completing the survey, we come to vowel gemination, or vocoid doubling. ese
cases involve vocoids which are neither short nor monotimbric diphthongs. A given vocoid is repeated within the same phonetic syllable, but without any movement in the vocogram at all: (aa) ( 1.30).
en the rst element of a diphthong is the same as that of a monophthong
given together in the vocogram, it is possible to show the monophthong and the
diphthong together by showing the diphthong with a dashed line, instead of an
unbroken one (which would indicate the diphthong alone).
en variant diphthongs occur, they are denoted by grey markers together
with unbroken lines (or, in the case of an unstressed variant, the marker will be an
unshaded gure with the edge and line both dashed).
1.30. Short and long monophthongs and diphthongs with the same starting point (here, all
stressed).
(short or long) monophthong
and (wide) diphthong (a, a:, ai)
and (wide) diphthong with
rounded 2nd element (a, a:, au)
and (narrow) diphthong with
and (wide) diphthong (a, a:, a)
rounded 2nd element (a, a:, a)
(short or long) monophthong and monotimbric diphthong (a, a:, aa)
1.4.6. Moreover, according to the direction of their movement, diphthongs can

be classied into three categories: opening (when the second element is lower), closing (when the second element is higher), and centering (when the endpoint is (),
or ()). In 1.31 (where all the symbols given are unrounded, for the sake of simplicity), the diphthongs given with thinner lines in the rst and third vocograms
((a)), or in the second and third (()), could be considered closing, opening, or
centering, depending upon the phonological interpretation and upon what sorts
of similar diphthongs are present (or absent) in that language in other areas of the
vocogram.
For example, if (a) comes together with (, ), or () with (ie, uo), then
(a) will be considered closing, and () opening. Analogously, diphthongs whose
second element is not exactly central, including cases with (, , X, x , #, P, )
(and (, , , , Y, T, , @, )), could be protably considered centering,
if structural conditions so indicate.
32
1.31. Dierent diphthongs.
closing
opening
centering
For example, in British English, beers /'bIz/ ('b;) and bear(s) /'bE{z}/
('b;{}) are in the same group as beer /'bI/ ('b;), even in accents which present
(, a) for //. erefore, the best decisions are normally made considering
both phonemic and phonetic transcriptions.
1.4.7. In the vowel orograms (and in the vocogram) supplied in 1.3-4, we give
the most extreme vocogram positions, with the express purpose of showing the
limits of the region of oral space used for vocoids. Instead, the orograms in 1.6
give positions which are more commonly found in the languages of the world;
these are slightly less peripheral.
In fact, certain trapezoids (or even vowel triangles) show all of the symbols
perfectly aligned along the edges, threaded along the lines (rather like pearls,
since they are all circular as well), so that they extend outside of the margins. We
nd such gures decidedly odd, given that the objective reality of vocograms is
quite dierent.
1.4.8. th orograms, which are fundamental for consonants, we follow several conventions, some of which are more intuitive than others. For example, it is
sucient to indicate nasals by showing the velum lowered, as in the cases of
1.9.1-2. Nasalized articulations are indicated in the same way, including vocoids
( 1.7, on the bottom). ere is also nasal explosion ( 1.19, on the right).
Stops are shown with a raised velum, and as with nasals, there is contact between
two or more articulators ( 1.10).
Constrictive orograms show the articulators close to one another ( 1.12.1-2
1.32; 1.1, , for ()), together with a useful convention (albeit a bit less obvious and objective) consisting of a horizontal black line, immediately above the
base of the orograms. is line is intended to allude to the constriction in this
case, the noise of friction produced by the ow of air through the narrowest point
of the articulation. If the line is not continuous, but divided into three segments
(as in the case of (,)), the contoid is semi-constrictive (intermediate between a constrictive and an approximant).
In the case of grooved constrictives ( 1.32 1.12.1. {and 9.13 of the
HPh}), there is a curved line placed on the tongue blade. is curve is meant to be
a reminder of the longitudinal groove that characterizes these articulations.
e curve (like the horizontal line) is segmented in the case of semi-constrictives. e same marks naturally occur in the diagrams for stopstrictives as well.
1. prelude
33
1.32. Slit and grooved constrictives: () and (s).
slit
grooved
In the case of approximants, there is visibly more space between the articulators,
and the horizontal line (used with constrictives) is omitted. However, there can
be a black arrow, slightly smaller than the one used for laterals, and this arrow signies lateral contraction lateralization a trait accompanying and characterizing
some of the approximants ( 1.13.3). Semi-approximants have a dotted horizontal line ( 1.13.2).
Trills, taps, and aps ( 1.14.1-3) are marked with a black dot placed upon the
articulator which is in motion (whether this is the tongue tip, the uvula, or the
lips). Moreover, a dashed outline is added to trill orograms, and two dashed outlines to ap orograms ( 1.33). In the same gure, we can see sequential articulatory compositions as well, which show two rapid tappings for (r), followed by
two open positions, with a white background.
For the tap, (R), there is only one contact, for a rapid tapping; whereas, the ap,
([), consists of three quite dierent (and quick) phases: rstly, the tip is brought
behind the alveolar ridge; then, while moving forwards, it touches the ridge (and
this is the pertinent articulation); nally, the third phase corresponds to the detachment at every tapping, such as the two white ones in (r).
1.33.
Trills, (r),
taps, (R),
and aps, ([).
Laterals are identiable with an arrow placed on the most fundamental part of
the articulation. If the arrow is black, the contoid is bilateral ( 1.15.1); if white,
the contoid is unilateral ( 1.15.2). If the contoid is instead a lateral tap, a white
dot is shown ( 1.15.3). (Much more is to be found in the HPh, on other types of
laterals, with dierent additions.)
1.34.
(Bi)lateral, (l), and
unilateral, (), articulations.
+
l
1.4.9. Stopstrictive orograms have a small black part, which refers to the stop
phase of these contoids. e stop phase (as can be seen in 1.11.1-2) is homorganic to the place of articulation of the constrictive phase, which immediately follows
34
and constitutes the second part of these phones. e result is a unitary (though
compound) phone, since its full duration corresponds to the length of other stops
or constrictives, not to the length of two phones combined in sequence.
Stopstrictives naturally have a horizontal line at the base of the gure. Grooved
stopstrictives have a curved line as well, representing the groove ( 1.11.1.). Stop-semi-constrictives have a black horizontal line divided into three parts. (Here we
indicate only the stop-strictives which occur in the 12 languages dealt with in this
book. Other languages present further types too, which are dealt with in the HPh.]
1.4.10. Palatograms ( 1.17, two lower rows) are used to show contact with the
palate throughout the course of an articulation. Palatograms are mainly useful for
certain contoids. It would be possible to employ them to add detail to the description of vocoids, particularly those which are not back. However, vocograms
and orograms are more useful in this case for purposes of description and teaching. en parts of the palatogram are shaded black, this convention signies full
contact of the articulators (in the stop phase of stopstrictive contoids). Grey shading indicates instead, naturally, constrictive contact (as in the characteristic constrictive phase of stopstrictives). If the palatograms of the constrictives, (, s, S) are
compared to those of the corresponding stopstrictives, (, q, c), this dierence
becomes clear immediately. It is useful to observe carefully (in the last part of every
set of diagrams, in 1.17) the palatograms and dorsograms which show the
dierence for the voiced alveolar articulation between stop, (D), and (bi)lateral,
(l), in addition to unilateral, (), and constrictive, (), as well (although these do
not occur in the 12 languages of the HPr).
1.4.11. Dorsograms (as in 1.17, rst row) give a new perspective, which is not
longitudinal but instead transversal. ese gures are particularly helpful to illustrate the dierence between slit tongue position (the unmarked position, since it
requires fewer phonic features) and grooved tongue position or lateral contraction
(or lateralization) these are the two marked possibilities, in comparison with the
unmarked position.
en arrows are used in labiograms given in prole (as in 1.5 1.16), they
indicate the direction of the characteristic movements, as produced by particular
facial muscles. Frontal labiograms require no particular explanations ( 1.8 and
those just mentioned) the vertical gap, which is steadily larger as the jaw opens,
is fairly intuitive. By carefully observing 1.16, let us consider attentively the labial dierences for (, S, ) (keeping in mind what has been said in 4.2.4 of the HPh,
about lip-position detection). However, we think it useful to call to mind that (S)
is protruded, not simply rounded ( 1.16.1).
Another useful type of diagram is the laryngogram (as in 1.1), which should
be analyzed attentively. ese laryngograms are naturally optical (as well as being
schematic and frozen in a particular instant), representing what can be seen with
a laryngoscope, or a throat-specialist's mirror. ey are not acoustic laryngograms,
which measure vocal fold vibration.
1. prelude
35
1.4.12. We will now move on to tonograms, which are divided into three juxtaposed bands, of high, mid, and low pitch (where the levels are not absolute, but
relative to the voice of each particular speaker). Both in preintonemes and in intonemes ( 1.1.5-8 1.21-3 {and, in the HPh 6.4.5.1-4 13.8-34}), as in
tones as well ( 1.26 {and, in the HPh again 6.4.4 12.17-18}), lines (or
dashes) placed at dierent heights and with various slopes represent stressed phono-syllables. Dots, on the other hand, represent unstressed syllables, while shorter
lines/dashes represent half-stressed syllables (with secondary stress, but with pitch
indicated by the position in the tonogram).
In phonotonetic transcriptions, secondary stress is indicated by two dots placed
close together (smaller than a single dot), variously oriented according to the tonetic necessities. Secondary stress on a medium level pitch is denoted with (&) in
order not to create confusion with the hyphen we use to show syllable boundaries.
is use is consistent with marking primary stress with ('). Unstressed phono-syllables (or better, weakly-stressed , weaker than half-stressed syllables) with mid
level pitch are not marked in any particular way. On the other hand, in tone languages, syllables with mid pitch and weak stress are preceded by a dot placed at
medium height, (2).
Guide to dierent types of transcriptions
1.5.1. In the HPr, there is no doubt that a great number of symbols is used.
However, these symbols are not superuous for those who desire to do phonetics
thoroughly (and not merely easily, and inevitably supercially). A small number of symbols ineluctably leads to mediocrity, whereas a large number of symbols
opens the way towards the true understanding and savoring of pronunciation.
Even if study has been methodical, concentrated, and accompanied by careful
exercises, it will remain occasionally necessary to check the value, the nature, and
the connections of certain symbols (and concepts as well) which are less frequent.
e best way to do so is to look for explanations in the right place, or places. In
fact, the necessary answers, verications, and connections, as well as new perspectives, can be found in the general index (contents) and in the analytical index (index), by ipping through the chapters and the sections, and by paying attention
to the tables, the lists of symbols, and the groups of gures.
e major categories are, of course, vowels, consonants, intonation, other prosodic traits (stress, pitch, duration/length), and paraphonics.
It would be complicated and probably useless to try to re-explain these matters in a general synthesis (probably too compressed and complex). erefore we
will merely suggest following the directions given here, emphasizing only the
meaning of the dierent parentheses used to enclose the symbols.
Slashes / / always denote phonemes, on a theoretical and abstract level; instead, brackets ( ) are used exclusively for phones (and taxophones) , the
practical and concrete side of things, which nonetheless naturally comprise essential generalizations and normalizations, without which it would be necessary to
36
speak only of single, unrepeatable realizations of particular individuals. For example, play /'plaE/ ('phla;).
1.5.2. Doubling the parentheses indicates a further accentuation of their
primitive values. us, double slashes // // refer to a still more abstract or theoretical level of phonological characteristics, such as what we have with German
wiederhaben //'vi:dKha:bn//, with respect to the ordinary phonemic or phonetic
transcription /'vi:dha:b/ ('vi:d&ha;b). Double brackets (( )) indicate instead
more exact symbols, in cases where showing extremely precise nuances is desired.
An example of the use of double brackets could occur when showing the articulation with a high tongue tip, ((s, x)) (where these symbols need to be drawn carefully so they can be clearly distinguished from (, S)), instead of the articulation with
a low tongue tip, (s, S), which is considered more normal.
Another case is given by partial (or even substantial) nasalization, as indicated
above ( 1.3.3): thanking ('k) ((('k))). e ocial IPA has no way to
show light automatic nasalization, and in fact, it arbitrarily and misleadingly employs our symbol () for a particular type of phonation, creaky voice (for which
we use ()).
nally, when angle brackets > enclose symbols or diacritics (such as @), these
refer to paraphonic elements. Instead, when they enclose orthographic symbols
(such as >a), they refer to graphemic elements.
Symbols not enclosed in brackets or slashes represent phones in treatments of
general phonetics, as in the orograms shown up to now.
Instead, in the consonant tables in 2-13 (and in the phonosyntheses, in 1623 of the HPh] the symbols not enclosed in slashes represent phonemes, even
though they are represented by fairly specic symbols (this is to avoid sacricing
precision, but without making the visual eect too intimidating in fact, here the
symbols represent phones as well]
In the phonemic transcriptions accompanying the phonetic transcriptions, it is
also possible to use more generic symbols to indicate phonemes. We have followed
this approach in 2-13, where we have also indicated the corresponding results
and presented them in easy reference form in the consonant tables.
Transcribing by hand
1.5.3. nally, several fundamental observations should be made on the topic
of transcriptions carried out by hand. It is important to draw every symbol (and
every diacritical marking) exactly as it is printed in this manual (where we employ
the elegant Simoncini Garamond character set, in our personal version called
imonani). It is necessary to be very careful not to confuse one symbol with another one which is similar to it, but obviously not exactly the same and with a
dierent value.
To start out, transcriptions should not be carried out in cursive (or in italics). In
fact, contrary to habits in normal handwriting, the symbols should not be connected
1. prelude
37
to each other, but should instead be kept separate, as in print. Moreover, symbols
should not be simplied, changing n into u, or m into ; or l into or a. Dots
should not be omitted, and so should not be written in place of i, nor for j.
Stylistic additions and modications should be avoided. us, d must remain
dierent from and from , just as is dierent from the grapheme g, and h from
. e same goes for cases like z, which should be kept distinct from both Z and .
Moreover, (L, r, R, ) are, by the same token, quite distinct from (/H, , [, y), .
Naturally, small capitals should be avoided as well, since the symbols (, , , I, G,
, ,, , K) are dierent from (a/A, b, e/E, i/, g, l, n, p, r).
In conclusion, it is necessary to set aside every normal writing habit which could
lead to confusion between the symbols. e best strategy to reach this essential
goal is to begin observing every symbol with great care. A typographic eye should
be developed which pays attention to every detail, from the size and orientation
of a stroke ((t, T, , ), (B, 6)), or of a symbol ((e, , ), (K, , , ), (A, a, ), (,
E, ), (X, x v, , O), (r, <, ), (h, H, ), (f, j, 3 J, , , )), to the presence or absence of a sort of serif, or the type of serif present: (i, I, u, , U o, , ).
In working towards this objective, the observations in 8-9 of the HPh will
help the reader to look at, not just see, the symbols found throughout the text. In
this way, it will prove easier to accept the fact that (g) always has the value present in the word get /'gEt/, and not the value in gem /'GEm/ (while gif or .gif
can be pronounced either /'gIf/ or /'GIf/), .
Generic symbols (for phonetic categories)
1.5.4. It might prove useful, sooner or later, to have symbols available which
do not directly represent particular segments, but rather whole phonic categories.
For this task, phonetic and phonological formulae can be used, and the resulting
symbols can be employed, for example, on the edges of vocograms or tables. We
therefore provide a list of appropriate symbols of this type.
1.35 gives a schematic presentation of the seven fundamental manners of articulation, for contoids. Also given are useful groupings and subdivisions, including the distinction between obstruents and sonants (however, the mixed manners
of articulation, typical of approximants, and even more of trills and laterals, are
not included).
e category of obstruent contoids includes stops (but not nasals, even though
these could technically be considered stops with added nasalization), stopstrictives,
constrictives (including constrictive trills and constrictive laterals), and approximants (only the peripheral ones). e sonants comprise, on the other hand, central or lateralized approximants, besides nasals and trills (together with taps and
aps), and laterals (including unilaterals and tapped laterals).
In various languages, for any manner of articulation except trills (and taps and
aps), phonetically semi- articulations are possible ( less tense with no full
contact, also for nasals, stops, stopstrictives and laterals).
38
1.35. Groupings of the fundamental manners of articulation.

0
F !

` _
vocoid/vowel
0
0
C
=
reduced V (in duration; = )

shortened V
nasalized V
devoiced V
voiced lenis V
voiceless lenis V
half-nasalized V
rounded V
unrounded V
advanced V
retracted V
lowered V
raised V
normal V or under other conventions
creaky V (or laryngealized)
contoid/consonant
reduced C (in duration; = )
shortened C
glottalized voiceless C, with simultaneous
()
intense (syllabic) C
devoiced C
voiceless lenis C or under other conventions, especially diaphonemic
voiced lenis C
voiceless C
voiced C
rounded C
unrounded C
palatalized C
velarized/uvularized C
advanced C
retracted C
tenser/closer C
less tense/close C
voiced creaky/laryngealized C
lateral C
constrictive lateral C
voiceless lateral C
lateral tap C
unilateral C
semilateral C
lateral trill/tap C
intense (syllabic) lateral trill/tap C
@ voiceless lateral trill/tap C
nasal C
M intense (syllabic) nasal C
voiceless nasal C
sonant (or sonorant) C
% intense (syllabic) sonant C
# voiceless sonant C
trill (or trill tap) C
voiceless trill (or trill tap) C
constrictive trill C
tap C
lateralized tap C
ap C
lateralized ap C
median approximant C
approximant C
semi-approximant C
lateralized approximant C
J lateralized semiapproximant C
peripheral approximant C
obstruent C (F, !, ), in diphonic pairs
laryngeal approximant C
laryngeal constrictive C
F stop C
K semi-stop C
(slit) constrictive C
(slit) semi-constrictive C
_ grooved constrictive C
` grooved semi-constrictive C
(generic) constrictive C
(generic) semi-constrictive C
` (slit) stop-strictive C
@ (slit) stop-semi-strictive C
(slit) semi-stop-strictive C
grooved stop-strictive C
& grooved stop-semi-strictive C
I grooved semi-stop-strictive C
! (generic) stop-strictive C
(generic) stop-semi-constrictive C
1. prelude
39
(generic) semi-stop-strictive C
| pause
lateral stop-strictive C
\ potential pause
trill(ed)/tap(ped) stop-strictive C
|| longer pause
* lexeme
(low) parenthesis
$ grammeme
(mid) parenthesis
O rhythm group
^ quotation
o reduced rhythm group
. emic conclusive intoneme
phono-syllable
? emic interrogative intoneme
i reduced phono-syllable
emic suspensive intoneme
light syllable
, emic continuative intoneme
9 heavy syllable
normal preintoneme (no sign)
` zero phone/phoneme
interrogative preintoneme
0 indicates proximity to 00, 000, 00
imperative preintoneme
indicates proximity to 0 , ,
emphatic preintoneme
0$ C with audible explosion
& supplementary interrogative preintoneme
0 C with inaudible explosion
(in French)
05 = /=/ /=h/, (=) (=h)
5 ( ' ) tone with strong stress, 12.17 of
0 = // /h/, () (H)
the HPh
0 ejective C
(& ) tone with mid stress, 12.17 of the
0 injective/implosive C
HPh
60 dejective/click C
1 (2 3) tone with weak stress, 12.17 of the
0 prenalized dejective C
HPh
0 prenasalized C
8 ( " ) tone with extrastrong stress,
' stressed V (with strong/primary stress)
12.17 of the HPh
& half-stressed V (with mid/medium/secon- 9 (6 0 ) falling tone, 12.18 of the HPh
dary/half-strong stress)
( ` ) rising tone, 12.18 of the HPh
unstressed V (with weak stress)
Japanese akusento (distinctive pitch lower destressed V (with reduced stress, up to
ing, 12.3.2.1-4)
weak; starting from ')
# (@ /) shift diacritic, 8.11, 9.5 of
" over-stressed V (with extrastrong stress)
the HPh
: long V
@ paraphonic element ( 14.3-5 of the
; half-long V
HPh) or grapheme
less than long V
( ) phonetic transcrition
less than half-long V
/ / phonemic transcrition
| utterance-nal V
(( )) hyperphonetic transcrition
| after a pausa or silence V
// // hyperphonemic transcrition
word-nal V
( ) symbol/phon(em)e which can fall (or be
word-initial V
lacking)
syllable-nal V
(( )) potential symbol/phon(em)e, which can
-- within a word, word-internal V
be used, as in 1.7-8.
e ocial IPA chart

1.6. For informative purposes we also show the ocial IPA chart ( 1.36). It
highlights mostly its defects and global deciencies, however.
e rst section gives the ocial pulmonic consonants, which are 58 (with a
minor addition of ten other symbols). e second small section shows ve dejec-
40
1.36. Ocial IPA Chart (1996).
~~~ ~
(ocial, revised 1993 and corrected in 1996)
~~~ (~)
Plosive
Bilabial Labiodent. Dental
Alveolar Postalveol. Retroex
Palatal
Velar
Uvular
k g
G
,
K
Nasal
Trill
Tap or ap
Fricative
Lateral fric.
()
b
m
B f
d
n
r
R
s z
!
<
l
S Z
Pharyng. Glottal
h H
V
>
j
Approxim.
$
L
Lateral app.
ere symbols appear in pairs, the one to the right is voiced. Shaded areas denote articulations judged impossible.
Clicks
Voiced implosives Ejectives
Bilabial
| Dental
! (Post)alveolar
Palatoalveolar
~ Alveol. lateral
Bilabial
Dental/alveol.
Palatal
Velar
Uvular
as in:
p Bilabial
t Dental/alveol.
k Velar
s Alveol. fricat.
Front
Close-mid
Open-mid
Open
Voiceless labial-velar fric.

Voiced labial-velar app.
Voiced labial-palatal app.
Voiceless epiglottal fric.
Voiced epiglottal fric.
Epiglottal plosive
Voiceless alveolo-palatal fric.

Voiced alveolo-palatal fric.
Alveolar lateral ap
Simultaneous S and x
Aricates and double articulat.

can be represented by two symbols joined by a tie bar if necess.
~ ~
Close
'
w
~~~ (~~-~)
i y
I Y
e
Central
Back
M u
U
o
O
A
ere symbols appear in pairs, the one

to the right (and U) is rounded.
0
`
or
>1
1
1
1
<1
Extra-high
High
Mid
Low
Extra-low
Downstep (relative)
Upstep (relative)
(Diacritics can be placed above a symbol with a descender, )

= Voiceless
) Breathy voiced b "
Y Dental

U P
Voiced
Apical

Creaky voiced
y Linguolabial u p
Laminal

t5 d5
5 Aspirated
g More rounded o
t d
e
Labialized
Nasalized
G Less rounded O
tJ dJ
d
J Palatalized
Nasal release
Advanced
%
t d
Velarized
Lateral release d
! Retaracted
m
/ Pharyngealized t/ d/
No audibile rel. d
Velarized
or
pharyngealized

t
7 Centralized
F Raised
l ( = voiced labial-velar fricative)
Z Mid-centralized
c Syllabic
f Lowered

L ( = voiceless velar approximant)
W Non-syllabic
Advanced Tongue Root
k

Retracted Tongue Root
& A&

& Rhotacized
8
3
1
1
21
"1
4[1
or
~
Rising
Falling
High rising
Low falling
Rising-falling
Global rise
Global fall
'
&
:
;
z
.
Primary stress
Secondary stress:
&foUn'tISn
Long
a:
Half-long a;
Extra-short q
Syllable break:
<i.kt
Minor (foot) group

Major (intonation) gr.
Linking (absence of a
break)
1. prelude
41
tives (or clicks), ve injectives (or implosives, including the theoretical uvular
one), and four examples of ejectives.
en the ocial quadrilateral follows, which tries to explain the vowels, which
are 28, in twelve pairs (for lip-position spread or rounded), plus four isolated elements. e collocation of (, , ) is highly unhappy.
e paltry section on tone and stress is misleading and leads one to think it is
not limited to some examples. en suprasegmental signs are shown.
Lastly, we nd the section on diacritics, which are thus necessary to avoid the
risk of being utterly vague and ambiguous. In fact, since only very few segmental
symbols are available, one must use some ocial diacritics (even two or three for
one symbol), which can be of some use when trying to hint at some important articulatory dierences.
Unfortunately, this partial solution also betrays one of the fundamental principles, which used to dierentiate IPA from all other phonetic alphabets, precisely
for its outright refusal to use diacritical signs, simply to indicate articulatory characteristics. Instead, now several diacritics are put above or under a generic symbol, or after it as a superscript tiny character.
e chart provides nothing at all to show intonation. Even the ToBI system
(which someone seems to consider the intonation component of ocial IPA] is
very far from being really useful, since it fails to separate intonation both from randomness and paraphonics.
e CANIPA in the HPr their correspondent OIPA symbols
1.7. For all the articulations given in the HPr (with their unitary symbols), we
thought it useful to add the corresponding oIPA transcriptions (given within
), in order to show their composition (almost as in chemical or algebraic formulae).
is will be useful both to understand the combinations of the few basic symbols with so many diacritics (even if we did not use all those which could have
been necessary for absolute precision, although oIPA followers usually omit them
in their approximate and vague transcriptions), and to highlight that it is unthinkable to do diacritical transcriptions, as all other phonetic alphabets do. One of
the fundamental criteria of the original IPA, in fact, was to avoid articulatory diacritics.
e cases where these two IPA systems coincide are indicated with (=); whereas
a partial correspondence is marked with (=); the dierences, without diacritics,
with ().
Let us recall that it is paramount to carefully observe the orograms, to constantly compare them, and to nd similarities also between symbols, by starting from
the ocial ones, from which the others have been derived (although with useful
modications and some necessary substitutions). As we have already said, our rm
intention is to avoid articulatory diacritics and prefer unitary symbols, instead.
is is also done to avoid that there may be second- or third-class symbols (and
42
thus second- or third-class phones, as well), because they are accompanied by diacritics (which are not easy to combine accurately).
Vocoids
Spread/unrounded
(i)
(I)
(e)
()
(E)
()
(i) (=)
(I) (=)
(e) (=)
(ef ) or (EF )
(E) (=)
(a) ()
()
()
()
()
()
()
(i&) or (i]) or ()
(I&) or (I7)
(e& ) or ( ) or (e7 )
(EZ ) or (eZ )
(aZ ) or (E7 ) or (E& )
(a7 ) or (a& )
()
()
()
()
()
(a)
() (=)
(f) or (HF )
() or (F )
() (=)
() (=)
(a7& ) or (f )
(M)
()
(X)
(x)
()
(A)
(M
7 ) or (M
)
(M
Z ) or (M
f ) or (M
7f ) or (7F )
() or (7)
(Z ) or (7F ) or (Z) or (7f )
() or (7 )
(A ) or (A7 )
(*)
()
()
()
()
()
(M) ()
(M
f ) or (f)
() ()
(f) or (F )
() ()
(A) ()
Rounded
(Y)
()
(y) ()
(Y) ()
()
()
()
()
(y)
(Y)
()
(#)
()
()
() ()
(f ) or (F )
() ()
() ()
(y7 ) or (y. )
(Y7 ) or (Y& )
(7 ) or (& )
(Z ) or (7F ) or (Z ) or (7f )
(7 ) or (& )
(7 ) or (& )
(%)
(T)
(+)
()
(@)
()
(%) (=)
(%f ) or (+F )
(+) (=)
() (=)
(f ) ()
(7& ) ()
()
()
(P)
()
()
()
(u7 ) or (u )
(U7 ) or (U )
(o7 ) or (o )
(OZ ) or (OF ) or (oZ ) or (of )
(O7 ) or (O )
(7 ) or ( )
(u)
(U)
(o)
()
(O)
()
(u) (=)
(U) (=)
(o) (=)
(OF ) or (of )
(O) (=)
() (=)
Semi-rounded
()
()
(W)
(ig ) or (yG)
(g ) or (%G )
(M
g ) or (uG )
()
(:)
(w)
(Ig ) or (YG )
(I.g ) or (Y.G )
(M
hg ) or (UG )
()
(eg ) or (G )
1. prelude
(,)
()
(e.g ) or (g ) or (e7g )
(g) or (oG )
()
()
()
(ehg ) or (EHg )
(g ) or (G )
(Zg ), (7g@ ), (OZG ) or (O7@G )
()
()
(,)
(Eg ) or (G )
(aZg ) or (
.G )
(g ) or (g7 ) or (OG ) or (O7G )
()
()
(a)
(aHg ) or (HG )
(a:7g ) or (hg ) or (7:G )
(Ag ) or (A7g ) or (G ) or (7G )
43
(/, \)
((/, ))
(/, m)
(/, ,)
(Xm)
()
(Xn )
(,) (=)
Semi-nasal
((/, M))
((/, ))
((/, ))
((/, n))
((/, ))
(m
f )
(nh )
(Nh )
(Nh )
(
h )
Stop
Contoids
Nasal
({)}, m)
(/, [)
(/, m)
(/, M)
((/, ))
(/, M)
(/, )
(/, ")
((/, ))
((/, n))
(/, N)
((/, ))
((/, ))
({}, n)
((/, ))
(/, n)
(/, n)
(/, N)
((/, n))
(/, ~)
((/, ))
(/, N)
((/, ))
((/, ))
(/, )
(m= , m) (, =)
(m)
(mJ)
(m)
(M)
(M) (=)
(MJ)
(M)
(nY ) or (n )
(n& )
(n ) or (nXm)
(n )
(n fJ )
(n , n ) (, =)
(n )
(n )
(n &)
(N ) (=)
(n. )
(n J)
(n J)
(N) (=)
(N. )
( )
() (=)
(p, b)
((, {)
(p, b)
((], 7))
(t, d)
(t, d)
((4, 7))
(T, D)
(, )
((T, D))
((t, d))
(+, _)
(, )
((, 8))
((, ))
(k, g)
(, /)
()
(p, b) (=)
(p, b)
(pJ, bJ)
(p, b)
(tY , dY )
(t!, d!)
(t:Y , d:Y )
(t, d )
(, ) (=)
(t: J, d: J)
(t: J, d: J)
(tJ, d J)
(, ) (=)
(& , .)
(k , g )
(k, g) (=)
() (=)
() (=)
Stop-strictive
(p, /)
(, /)
(, )
(., )
(k, )
(%, /)
((, ))
(w, )
(, W)
(pXF)
(pXf)
(tx, dx)
(t=H#x< , dx<#H )
(X, X)
(X: )
(kxx , gX)
(kxx, gX)
(XX, Gx)
44
(k, K)
(HH=XK , GxKH#)
(q, Q)
((q, Q))
(, )
(, ")
(C, )
(c, G)
(&, 1)
((C, G))
((c, g))
((v, ))
(, )
(, )
(tXsY , dXzY )
(tXs:Y , dXzY:)
(tXs , dXz )
(X, X)
(tXSJ, d XZJ)
(tXS, d XZ)
(tXS g, d XZg)
(:tXSJ, d: XZJ)
(:tXS, d: XZ )
(:tXSg, d: XZg)
(tXs J, dXz J)
(tXs J, dXz J)
Stop-semi-constrictive
(, )
(, )
(, )
(Xh , hX )
(kXxf , ghX)
(ftX , dfX)
Constrictive
(, 6)
(f, v)
(5, )
(f, v)
((w, W))
(, )
(/, D)
(Q, z)
(P, p)
(, J)
((, /))
((, ))
(x, )
(X, )
(h)
()
(F, B) ()
(f, v) (=)
(f, v)
(fJ, vJ)
(, )
(, ) (=)
(& )
(H<F, <FH )
(H=># , >#H )
(, ) (=, )
(. )
(x , )
(x, ) (=, =)
(X, ) (=)
() ()
(h) ()
(s, z)
((s, z))
(, )
(, )
(, )
()
(sY , zY )
(s& , z& )
(s& , z&)
(s , z )
(, ) (=)
()
(, )
(S, Z)
(, )
((X, 5))
((x, ))
((, ))
(., )
(/, \)
(S, q)
(, =)
(, B)
(SJ, ZJ)
(S, Z)
(Sg , Zg )
(S:J , Z:J)
(S:, Z:)
(S:g , Z:g )
(S: , Z: )
(S:, Z:)
(S:, Z: )
(s J, z J)
(s J, z J)
Semi-constrictive
(, )
(/, , )
(/, y)
(/, m)
(ff , vf )
( )
(h)
(h)
(, d)
(, )
(shY , zYh )
(ShJ, ZhJ)
Approximant
(F, B)
(/, V)
(/, )
(/, )
(, )
(/, )
(, j)
(/, )
(, j)
(/, )
(/, )
(, V)
(W, w)
(, )
(x, )
(/, H)
(F , Bf)
(V) (=)
(V)
(V& )
(:f , :f )
(h )
(h) o (), (j) (, =)
(. )
(:], :]j) or (
]+ ,
7 )
(
)
(w
)
(
= , ) ()
(', w) (, =)
(Xh , f )
(Xh /, f /)
(Hf )
Semi-approximant
(/, )
(/, %)
(/, F)
(j)
(f. )
(j:]f) or (
]f )
1. prelude
(/, =)
(/, j)
(
h )
(w
f )
Laryngeal approximant
(h, H)
(/, )
(+)
()
()
(=)
()
(hf , Hf )
(Hf )
(h J)
(h )
('
)
(h )
(h )
45
ap
({}, [)
({V}, e)*
(/, )
(, R )
(, R )
(# )
Constrictive trill
(, )
(K#=H , K#H )
Lateralized ap
(r, m)
( , )
(R: &, R:&)

((R: &, R:&)
Lateralized approximant (+ 2 semi-)

(/, S)
(/, )
(/, )
(/, )
(/, >)
(/, )
(/, <)
(/, <)
(/, )
(VX< &)
( &)
(< &)
(<< &)
(<< &)
(<< &)
(X < &)
(<X &< )
(X
h < &)
Trill
(/, r)
(/, 5)*
(/, 5)
({{}, D)
({'}, K)
(r )
(r )
(r &)
(r+ J, r J)
(K= , K) (, =)
Tap
({5}, R)
(/, R)*
(/, R)
(/, e)
({[}, )
({3}, r)
(R+ , R )
(R )
(R < )
(#) ()
(R+ J, R J)
(KW+ , KW )
(Bi)lateral
((/, l))
((/, ))
((/, R))
({a}, l)
({}, )
({}, ])
(/, )
((/, l))
(/, $)
((L))
({}, )
(/, L)
(/, )
(lY)
(lY) or (Y)
(lY:)
(l , l )
(l), (l ) or (), ()
(lf), (l) or (), (f)
(l ) or ( )
(l <) or (: )
($) (=)
(l: J)
(l J, l J)
(L) (=)
() ()
Unilateral
(/, )
(/, |)
(/, )
(l@ )
(l@ )
(l@ )
Tap(ped) lateral
(/, )
(/, )
() ()
(:)
Semilateral
(/, )
(/, )
(f)
(f )
2. English
2.0. In this chapter we will deal with the American and British neutral accents
(or standard accents). We will also make a teaching proposal for an international accent of English that could usefully be employed in pronunciation books and
pronouncing dictionaries (and in common dictionaries, too), as well as in everyday teaching. e kind of transcription we use is dia#onemic expressly devised
for this kind of description, together with its corresponding #onetic and #onotonetic transcriptions.
Furthermore, we will also consider the American and British mediatic accents
(from non-local radio), which are now as frequently heard as the neutral ones,
in the news, in the movies, and in songs.
To conclude, we will describe (without going into great detail, mainly by using
our usual accurate symbols and many diagrams) some other accents (with internal variations): those of Canada, Australia, New Zealand, and England (for the
latter we will give the traditional, aected, and Cockney accents). In a book in
progress Eli Pronunciation* we will deal with all the native accents of English all over the world (including many non-native accents), by working directly
on a substantial number of recordings, as well.
2.1. It will be useful to list the correspondences between our diaphonemic symbols and the phonemic symbols used in recent dictionaries, and especially in the
three current English pronouncing dictionaries (Longman, Cambridge, Oxford).
ese do not always agree for certain aspects, but we show them (between ) after the diaphonemes, with examples.
Vowels
/i/ /i, I/ lady /'lEIdi/ ('lIDi)
/I/ /I/ bit /'bIt/ ('bT)
/E/ /e, E/ let /'lEt/ ('lT)
// /, a/ hat /'ht/ ('hT)
// /, / hut /'ht/ ('hT)a ('hT)b
// /O, A:, A/ hot /'ht/ ('hAT)a ('hT)b
/U/ /U/ book /'bUk/ ('bk)
/u/ /u, U/ inuenza /Influ'Enz/ (&Mfl'nz)a (-)b
// // anoèr /'n/ ('n)a ('n)b contain /kn'tEIn/ (khn'Th;In)
/A:/ /A:, A/ spa /'spA:/ ('spA:)
/O:/ /O:, O/ law /'lO:/ ('lO:)a ('l:)b
/Ii/ /i:, i/ see /'sIi/ ('sI;i)
/EI/ /eI/ dZ /'dEI/ ('D;I)

/aE/ /aI, I/ y /'flaE/ ('fla;)
/OE/ /OI/ boy /'bOE/ ('b;)
/aO/ /aU/ cow /'kaO/ ('kha;)
/OU/ /U, oU/ go /'gOU/ ('g;)a ('g;)b
/Uu/ /u:, u/ who /'hUu/ ('hU;u)a ('h;u)b
// /I, , , / wies /'wISz/ ('wS)a
('wS)b
/;/ /, A:/ last /'l;st/ ('lsT)a ('lA;sT)b
/A;/ /A:, / pasta /'pA;st/ ('phA;sT)a
('phsT)b
/;/ /, O:/ so /'s;/ ('sO:)a ('s;)b
/O;/ /O:, / false /'fO;s/ ('fO;s, 'fAs)a
('f;s, 'fs)b
/;/ /r, !:, :r, r, / hurry /'h;<i/
('h;i)a ('h>i)b
2. english
/:</ /:r, :r, r, !:/ furry /'f:<i/ ('f;i)a
('f;>i)b
/:/ /:, :, :r, r, !:/ fur /'f:/ ('f:)a
('f:)b
/A:/ /A:, A:r/ car /'kA:/ ('khA:<)a ('khA:)b
/O:/ /O:, O:r/ door /'dO:/ ('D:<)a ('D:)b
// /r, , r, r, / wonder /'wnd/
('wnD)a ('wnD)b
/i/ /i, I, ir, Ir, I, i, ir, i/
happier /'hpi/ ('hpi)a ('hpi)b
/I/ /I, Ir, Ir, I, Ir, I/ here /'hI/
('hI;)a ('h;)b
/E/ /e, E:, er, Er, e, er, E/ ère
/'E/ (';)a (';)b
/U/ /U, Ur, Ur, U, Ur, U/ moor
/'mU/ ('mU;)a ('m;)b (with further
pronunciations)
/u/ /u, U, ur, Ur, u, u, U,
ur, u:/ rescuer /'<Eskju/ ('<skj)a
('<skj)b
/Ii/ /i:, i:r, i:, i:, i:r, i:/ seer
/'sIi/ ('sIi)a ('sIi)b
/EI/ /eI, eIr, eI, eI, eIr, eI/
plZer /'plEI/ ('phlI)a ('phlI)b
/aE/ /aI, I, aIr, aI, aI, aIr,
aI/ liar /'laE/ ('la)a ('la)b
/OE/ /OI, OIr, OI, OI, OIr, OI/
employer /Im'plOE/ (m'phl)a (m'phl)b
/aO/ /aU, aUr, aU, aU, aUr, aU/
tower /'taO/ ('Tha)a ('Tha)b
/OU/ /U, oUr, oU, oU, oUr/
slower /'slOU/ ('sl)a ('sl)b
/Uu/ /u:, u:r, u:, u:, u:r/ doer
/'dUu/ ('DUu)a ('Du)b
/</ /r, r, r, / wonderi /'wnd<I/ ('wnD<, -D<)a ('wnD>,
-N>)b
/I</ /Ir, Ir, Ir/ heari /'hI<I/ ('h<)a ('h>)b
/E</ /er, er, Er, E:r/ cari /'kE<I/
('kh<)a ('kh>)b
/U</ /Ur, Ur, Ur/ curi /'kjU<I/
('khj<)a ('khj>)b (with f. pron.)
// /, e, E/ dictionary /'dIkSn<i/
('DkS&<i)a ('DkS>i)b
47
/K:/ /, O:/ repertory /'<EptK:<i/ ('<p&T<i)a ('>pT>i, ->i)b
// /U, , T, / regular /'<Egjl/ ('<gjl)a ('>gjl, -gj-)b
Consonants
/m/ /m/ some /'sm/ ('sm:)a ('sm:)b
/n/ /n/ sun /'sn/ ('sn:)a ('sn:)b
// // su /'s/ ('s:)a ('s:)b
// /m, m, m/ rhy`m /'<I/ ('<)a ('>-)b
// /n, n, n/ cotton /'kt/ ('khAT)a
('kh-)b
/p/ /p/ pa$ /'pk/ ('phk)
/b/ /b/ ba$ /'bk/ ('bk)
/t/ /t/ o /'tUu/ ('ThU;u)a ('Th;u)b
/d/ /d/ do /'dUu/ ('DU;u)a ('D;u)b
/k/ /k/ came /'kEIm/ ('kh;Im)
/g/ /g/ game /'gEIm/ ('g;Im)
/c/ /tS/ >ain /'cEIn/ ('ch;In)
/G/ /dZ/ Jane /'GEIn/ ('G;In)
/f/ /f/ f[ /'fjUu/ ('fj;u)
/v/ /v/ vi[ /'vjUu/ ('vj;u)
// // wreath /'<Ii/ ('<Ii)a ('>Ii)b
// // wreathe /'<Ii/ ('<I;i)a ('>I;i)b
/s/ /s/ ice /'aEs/ ('as)
/z/ /z/ \es /'aEz/ ('a;)
/S/ /S/ dilution /d'lUuSn/ (D'lUuSn)a
(D'luS)b
/Z/ /Z/ delusion /d'lUuZn/ (D'lUuZn)a
(D'luZ)b
/</ /r/ rate /'<EIt/ ('<IT)a ('>IT)b
/j/ /j/ yate /'jEIt/ ('jIT)
/w/ /w/ wait /'wEIt/ ('wIT)
/h/ /h/ hate /'hEIt/ ('hIT)
/l/ /l/ late /'lEIt/ ('lIT)
// /t, / ci /'sIi/ ('s[i)a ('sTi)b
// /, r/ car /'kA:/ ('khA:<)a ('khA:)b
// /j, / n[ /'nUu/ ('nU;u)a ('nj;u)b
// /h, / when /'wEn/ ('wn: 'hwn:
'Wn:)
// /l/ lull /'l/ ('l,:)a ('l:)b
// /l/ little /'lI/ ('l[)a ('lT)b
48
Vowels
2.1.1. English has a high number of vowel phonemes, so it may be advisable to
subdivide them into groups, rather than keeping them all together. is is also useful to make easier comparisons with other languages, and to avoid possible confusions. e essential English vowel phonemes are: (short and long) mono#òs
/I, E, , A:, , , O:, U, :, / and di#òs /Ii, EI, aE, OE, aO, OU, Uu/. Furthermore,
there are some dia#onemes\ /;, A;, ;, O;, ;/, and unstressed /i, u, / (plus some
other possible devices). But, in the volume English Pronunciation instead of /Ii,
Uu ;, A;, ;, O;, ; /, we use /ii, uu X, [:, ], r:, : /, which are more convenient.
We denitely prefer to deal with English pronunciation in a diaphonemic way.
It is important to show especially what the American and British accents have in
common, so that their structural dierences are made clearer and more natural.
American monophthongs
2.1.2.1. 2.1 shows the American monophthongs (the British ones are given
in 2.2). Let us start from the eight bla$ markers, which indicate the realizations
of the following eight vowel phonemes (in stressed or unstressed syllables): () /I/,
() /E/, () //, (A:) /A:/, (A) //, () //, (O:) /O:/, () /U/. Although in the mediatic
American accent ( 2.4.2.2) /A:/ and // are often neutralized because they can
both be realized as (A[:]) in the neutral American accent we keep them apart, for
three good reasons. rstly, they are in actual fact dierent, even if chiey only in
terms of length. Furthermore, in this way we can keep a diaphonemic relation
with the British neutral accent. And nally, this will help us in highlighting the
characteristics of other accents, starting from comparable bases, although they are
actually shared only by a minority of speakers. Let us stress, in fact, that a neutral
pronunciation is always learned voluntarily.
Let us now illustrate the phonemes in 2.1. It must be recalled that the transcriptions, including phonemic ones, bear a stress mark, even for monosyllabic
words, unless they are usually unstressed in sentences, like the preposition in /In/
(n), compared to the adverb in /'In/ ('n:), or to the noun inn /'In/ ('n:)): ('hT)
/'hIt/ hit, ('js) /'jEs/ yes, ('m;n) /'mn/ man, ('fA;) /'fA:/ faèr, ('lAT) /'lt/ lot,
('<n:) /'<n/ run, ('lO:n) /'lO:n/ lYn, ('phT) /'pUt/ put.
For the sake of descriptive precision, although generally variations do not require dierent symbols, it is worthwhile to notice the taxophones of /I, E, , ,
U/; please note the two that change their symbols as well, in comparison with (,
): ('bE:, 'h,:) /'bE, 'h/ bell hull.
2.1.2.2. e gr\ markers in the vocogram also show three variants of /I, U, O:/
(I, U, :) (compared to normal (, , O:)). ey occur in (stressed or unstressed) syllables checked by //: (I) /I/, ([j]U) /[j]U/, (:<) /O:/. ere are no dierences in
terms of the nuclear element (or, perhaps, nucleal element) in () /E/, (A:<)
/A:/, because they coincide with the black markers for /E, A:/ (, A:). Instead, the
2. english
49
typical American realization of /:/ is through an intense contoid (:), rather than
a vocoid as in British pronunciation ((:)).
Examples: ('hI;) /'hI/ here, ('phjU;) /'pjU/ pure, ('D:<) /'dO:/ door, (';)
/'E/ ère, ('fA:<) /'fA:/ far, ('f:) /'f:/ fur A possible pronunciation with (, )
(instead of (I, U), for /I, U/} does not sound strange, although it is not the
most frequently heard.
For /O:</ (that is, before vowels: /O:</), the realization is (O:<) (dierent from
/O:/, marked by a grey marker): ('bO;<)a (->-)b /'bO:<I/ bori In American English the diaphonemic transcriptions /I<, E<, U</ correspond to /I<, E<, U</, with
their typical articulations shown by the black markers. By the way, the dierence
between /I<, E<, U</ and /I, E, U/ (taken into consideration above) should
be carefully noted. Here are a few examples: ('h<)a ('h>)b /'hI<I/ heari,
('b<)a ('b>)b /'bE<I/ beari, ('D<)a ('Dj>)b /'dU<I/ duri
It must be also noticed that, in American pronunciation, the phonemic sequence
/jU/ has the variant /j:/ (more typically mediatic): ('phjU; 'phj:) pure (the variant /[j]:</ occurs for /jU<, U</, too): ('khj<is 'khj;is) curious, (n'D<ns -'D;ns) endurance, ('D< 'D;) duri is is also true of other consonants with a
palatal component: ('SU; 'S:) /'SU 'S:/ sure but ('phU;, -:<) /'pU, -O:/ poor.
e dierence between ('mO;<n) /'mO:nI/ morni and ('mon, 'mo;<nI)
mourni (which, by now, belongs only to traditional American pronunciation)
is no longer neutral, nor is it modern, but simply regional. As a matter of fact, they
are now both pronounced ('m;<n) (('m;n)b) /'mO:nI/ (and the traditional
pronunciation of mourni could be shown diaphonemically as /'mOnI/ or
/'mo:nI/).
2.1. American monophthongs.
/[]/ (, ), /</ (<), /;/ ([:])
/u/ ()
/U[]/ (, ), /U/ (U), /U</ (<)
/:, :<, ;</ (:), // (), /</ (<, )
/[]/ (, ,)
/O:/ (O:), /O:</ (O:<), /O:/ (:<), /O;/ (O: A)
// (A), /;/ (O: A), /;</ (O:< A<)
/A:/ (A:), /A;/ (A:)
F
/i, i/ (i)
/I[]/ (, ), /I/ (I), /I</ (<)
// = //, // (, *, |)
/E[]/ (, E), /E/ (), /E</ (<)
2.1.2.3. e four white markers show the unstressed realizations of the remaining three phonemes: (i) /i/, () /u/, (, ) //. ey occur in the following contexts
/i/ at the end of lexemes: ('lIDi[]) /'lEIdi[z]/ lady ladies, ('ni&) /'EniI/ aJì, and before vowels: (<i'kT) /<i'kt/ react /u/ chiey before vowels: (&sc'ISn) /sIcu'EISn/ situation (Th'Gi) /tu'Gi/ to Algy A few words ending in -ue
[value atue virtue iue tiue] may have a reduced variant, especially in a preintoneme, which might usefully be represented with the diaphoneme /u/: ('vlj[u],
'sTc[u]) /'vlju, 'stcu/ value atue
nally, // is the most frequent unstressed vowel phoneme of the English language (in particular American, Oceanian, and South African): (f'ThAg<f) /f'tg<f/ #otogra#er. In words like ('sf)a ('sf)b /'sOUf/ sofa we can see that a
50
nal // before a pause (even a short one), is realized as if it were the (unstressed)
phoneme //; however, if a pause is not there, this does not occur.
2.1.2.4. In certain types of diaphonemic transcription, it would certainly be
better to use precisely this notation. Let us make this fact clear at once, with suitable examples in phonetic and (dia)phonemic transcriptions, in order to avoid any
unintentional misunderstandings. In the plural, we have: ('sf)a ('sf)b /'sOUfz/ sofas even when not in an intoneme: (h'f wz';, -f w-)a (h'sf
wz';, -f w-)b /h'sOUf wz'OUd/ her sofa was old ( 2.1.3.4).
On the other hand, if we introduce even a simple continuative intoneme, with
a short pause, we have: (pwz[h]'f2\ pwz')a (pwz[h]'sf2\ pwz';)b /Itwz[h]'sOUf,\ twz'OUd/ it was her sofa àt was old.
us, this is what happens to /./, /?/, // /,/, ( with intonemes and pauses
of any length). e same is true, but only in the neutral British accent, of /./,
/?/, //, /,/: ('D;)b /'dI/ dear, ('snD)b /'sEnd/ sender Instead, for /E/, the
modern neutral pronunciation has (): ('kh;)b /'kE/ care One generation ago,
it was ('kh;)b; while, earlier than that, it was ('khE;)b; still earlier, it was ('khE;)b;
and before that, ('kh;)b but let us close this micro-diachronic digression.
e second taxophone of //, (), occurs in contact with /k, g, w, / (as they
are velar, or at least have a velar component): ('bk 'gn: -;In) /'bk 'gE[I]n/
ba$ again, (khn'Th;In) /kn'tEIn/ contain, (Th'g;)a (-;)b /t'gOU/ to go, ('s
'sO:, 'sA;)a (-;)b /'s 's;/ si a so, ('wn:) /'wInd/ è wind, ('I~G,
-G) /'EInG, -G/ ael
As can be seen from the vocograms, () substantially is () with no lip rounding (and the symbol itself makes this quite clear). However, in these cases () could
be used, without great problems, as is done by some native speakers. On the other hand, the correct articulation may be produced, spontaneously, even by foreigners, when they are able to adequately reproduce all the other phones that realize
the English phonemes.
British monophthongs (and centering diphthongs)
2.1.3.1. Let us now consider the corresponding vocogram for the Briti accent
( 2.2). Here we have nine bla$ markers (for nine either stressed or unstressed
phonemes): () /I/, () /E/, () //, (A:) /A:/, () //, () //, (:) /O:/, () /U/, (:)
/:/. We nd: ('hT) /'hIt/ hit, ('js) /'jEs/ yes, ('m;n) /'mn/ man, ('fA;) /'fA:/
faèr, ('fA:) /'fA:/ far, ('lT) /'lt/ lot, ('>n:) /'<n/ run, ('l:n) /'lO:n/ lYn, ('D:)
/'dO:/ door, ('phT) /'pUt/ put and ('w:) /'w:d/ word (for American English, of
course, this last word was in the group of //, as we have seen above). If we consider farèr /'fA:/, we can see that it is pronounced exactly as faèr ('fA;) (with
no // pronounced), while in American English both r // are realized: ('fA;<). For
an American pronunciation like ('fA;), for farèr, and other words, see below:
diimilation ( 2.3.3.5).
Let us quickly add that even for (:) /O:/, as in ('w:[]) /'wO:[z]/ war(s) by now,
2. english
51
pronunciations like ('w;, 'w;) are old-fashioned (or regional).

For /I, E, , , U/ we have some modications, with an actual change only
for two of them: (';, 'h:) /', 'h/ Al hull (in comparison with (, )).
2.2. British monophthongs.
/i, i/ (i)
/I/ (), /I/ ()
// = /I/ ()
f
/E[]/ (, )
/[]/ (, )
/[]/ (, ), /;/ (A:)
/u/ ()
/U[]/ (, )
/[]/ (, *, |)
/O:/ (:), /O;/ (:, )
/:</ (:>), /:/ (:), /;</ (:>)
// (), /;/ ()
/A:/ (A:), /A;/ ()
2.1.3.2. For British English, /I, E, U/ are given in 2.3: ('h;) /'hI/ here,
('phj;) /'pjU/ pure, (';) /'E/ ère We have already seen that in British English a simple vocoid, with no contoid, occurs in cases like ('f:) /'f:/ fur It is to
be noticed that, in this type of pronunciation, the phonemic sequence /[j]U/, by
this time, is almost exclusively substituted with its variant /[j]O:/: ('phj: 'phj;)
pure, ('ph: 'ph;) poor, ('S: 'S;) sure
It is curious to note that, in the British pronunciation of the sixties, a similar
trend was very strong, but it soon stopped: ('phj:, 'khj;>is, n'Dj;>ns, 'Dj;>, 'S:)
pure curious endurance during sure (at that time, of course, the neutral pronunciation had () /I/ for modern (i) /i/: ('khj;>s), curious). Another curiosity
is that, in the rst half of the twentieth century, the same pronunciation (j:) /j:/
(which today no longer occurs, except as an old-fashioned variant) was a variant of
/I, I</: ('D; 'Dj:) dear ('h; 'hj:) here (sn's; -'sj:) sincere ('khl>ns 'khlj;>ns) clearance however, it still remains as a secondary variant in ('j; 'j:) year
2.1.3.3. Also in the British accent, the four white markers show the realizations
of the remaining phonemes, always in unstressed positions: (i) /i/, () /u/, (, )
//. ey occur in the same contexts /i/ at the end of a lexeme: ('lIDi[]) /'lEIdi[z]/
lady ladies, ('ni&) /'EniI/ aJì, and before a vowel: (>i'kT) /<i'kt/ react
/u/ occurs chiey before vowels: (&sc'IS) /sIcu'EISn/ situation (Th'Gi) /tu'Gi/ to Algy e possibility of nding /'vlju/ value ( 2.1.2.3) is less common
in British English.
nally, // is the most frequent unstressed vowel phoneme in British English,
too, also because of the vocalization of /, :/): (f'Thg>f) /f'tg<fz/ #otogra#ers ((f'ThAg<f)a) ('>c) /'<Icd/ Ri>ard (('<c)a), (Th'g;) /t'gOU/
to go ((Th'g;)a). Of course, we also nd (|) /, / (before pauses): ('sf)
/'sOUf/ sofa {('sf)a}, ('A;fT) /';ft/ after {('[;]fT)a}.
2.1.3.4. However, in British pronunciation, // has another rather important
taxophone, (). Phonetically, it is the short version of /:/ (:), and occurs for //
followed by the grammemes /z, d/, and for non-prepausal // (while, if // is
nal before a pause, it becomes ()).
52
So we nd: ('fA;) {('fA;)a} faèrs and faèr's ('A;ns) {('[;]ns)a} an~ered and ('phlZz&man ->- 'ma;n) {(-- -z- 'm-)a} è pleasure is mine
('n 'g;) {(- 'g:, 'g;)a} anoèr girl
In addition to //, this happens to non-prepausal // as well (also in American
English, generally only up to (), with no need to use ()): ('phk 'khT) {(-
'khT)a} alpaca coat so, as can be seen, () occurs, even near a /k/ , instead of
(), (a'phk 'khT) {('khT)a} I'll pa$ a coat] but: ('phk|) {('phk|)a} alpaca
However, even this taxophone can be represented by the usual realization of //:
(), as many native speakers do. As a matter of fact, compared to (), this other
taxophone may be less important and almost exclusively limited to the British accent.
Both in American and in Briti English, but with a traditional or regional connotation, /i, u/ may be realized as if they were /I, U/: ('lID[]), ('n&), (<'kT)a
(>'kT)b, (&sc'ISn)a (-S)b (Th'G)a (--)b. A possible compromise is (I) for
/i/: (-DI[], 'nI-, <I-a/>I-b, -GI). On the other hand, in American English /u/ =
/w/ is quite frequent, too: (&sc'ISn, &sc'wISn), ('v, 'vjw) (('vj)b) /'vljuI/ valui, and /OU/ = /w/ as well: ('fAl, 'fAlw) (('fl)b) /'flOUI/ followi, ('SDi, -Dwi) ((-Di)b) /'SdOUi/ adowy
2.1.3.5. It is useful to recall here (although this is also true of the preceding cases of //) that, in neutral British pronunciation (as well as Oceanian and South
African), the normal realization of // is (): (ph'f;mns) /p'fO:mns/ performance (&s'Th;In) /s'tEIn/ ascertain (h'hzbn) /h'hzbnd/ her husband Of
course, this holds good unless in absolute nal position before a pause, (|), or nal
in a rhythm group, or with the grammemes /z, d/, (, , ): ('h>) /'hI<I/
heari ('h; w'j;u) /'hI wI'jUu/ here wi` you, ('b;) /'bIz/ beers, ('h;> n';) /'hI n{d}'E/ here and ère
e modern neutral British pronunciation of /E/ does not change any longer
according to context, but it is always () (with a strong tendency to a long monophthong, through (), up to (:) /E:/, as has, for a few generations, already happened to the previous /O/): ('b;[]) /'bE[z]/ bear(s) For /A:, :, O:/, as we
have seen, // is lost and realized through a zero phone, (`): ('khA:[]) /'kA:[z]/
car(s), ('f:[]) /'f:[z]/ fur(s), ('D:[]) /'dO:[z]/ door(s).
In absolute nal position and before pauses, we nd: ('b;|) /'bI|/ beer,
('khj;|) (and (-j:|)) /'kjU|/ cure and (f'Thg>f|) /f'tg<f|/ #otogra#er
But, if nal in a rhythm group, we have: ('b; f'mI;i) {('bI; f'mI;i)a} /'bI f 2.3. British // diphthongs.
/I, I, I</ (>, z+, D+, +|)
/E, E</ ()
/aE, aE, aE</ (a, a+|)
+ "/a/& (a>, az+, aD+, a+|)
/j/U, -U, -U</ (j>, jz+,

jD+, j+|) (+ /jO:, -</}
/aO, aO</ (a, a+|)
+ "/a/& (a>, az+, aD+, a+|)
{+ "/A/& (A>, Az+, AD+, A+|)}
2. english
53
'mIi/ a beer for me, (&k'khj;> 'z 'g;, -j:> 'z-) {(&k'khjU; 'z 'g;, -j: 'z-)a}
/t'kjU 'Izt 'gUd/ àt cure isn't good, (f'Thg>f 'DDT) ((f'ThAg<f 'DDT)a) /f'tg<f 'dIdIt/ è #otogra#er did it.
Diphthongs
2.1.4.1. Let us now consider the seven phonemic diphthongs of American English. ey have ten realizations, which are necessary for a good pronunciation (
2.4). e bla$ markers stand for the seven phonemes, while the three gr\ ones
show taxophones, or contextual variants. Meanwhile, we will see: (Ii) /Ii/, (I) /EI/,
(a) /aE/, () /OE/, (a) /aO/, () /OU/, (Uu) /Uu/: ('ThI;i) /'tIi/ tea, ('D;I) /'dEI/ dZ,
('ha;) /'haE/ high, ('b;) /'bOE/ boy, ('na;) /'naO/ now, ('g;) /'gOU/ go, ('hU;u)
/'hUu/ who Besides, we need: (iI) /Ii/, (ju) /jUu/, (Uu, u) /Uu/, ([j]uU) /[j]Uu/:
('fi;I, 'fiI) /'fIi/ feel, ('ju) /'jUu/ you`, ('nU;u, 'n;u) /'nUu/ n[, ('ju;U, 'juU)
/'jUu/ yule, ('khu;U, 'khuU) /'kUu/ cool
e other diphthongs, even if followed by //, do not change much their components, apart from those with front second elements; besides (except for /OU/,
which has only //, in neutral pronunciation), they freely uctuate between // and
// (with a possible realization of // as // ()): ('s;, 's) /'sEI/ sail, ('fa;,
'fa) /'faE/ le, ('b;, 'b) /'bOE/ boil, ('fa;, 'fa) /'faO/ fowl but only
('s;) /'sOU/ soul.
2.4. American diphthongs.
/Ii/ (iI)
/Ii/ (Ii)
/jUu/ (ju, jUu)

/[]Uu/ (Uu, u), /[j/]Uu/ ([j/`]uU)
/EI[]/ (I, )
/OU[]/ ([])
/OE[]/ (, )
/aO[]/ (a[])
/aE[]/ (a, a)
2.1.4.2. For the corresponding Briti diphthongs, we nd seven fundamental

types, plus six taxophones. e modern pronunciation diers only slightly from
the more traditional one (and so there is little dierence from the American one)
for /Ii, EI, aE, OE, aO/ and for /Ii/, too, as can be seen better through a careful comparison between the British ( 2.5) and the American ( 2.4) vocograms. But
there is a bigger dierence for /Uu, OU/ and /Uu, OU/. Indeed, we have: ('ThI;i) /'tIi/
tea, ('D;I) /'dEI/ dZ, ('ha;) /'haE/ high, ('b;) /'bOE/ boy, ('na;) /'naO/ now, and also: ('fi;I, 'fiI) /'fIi/ feel, ('s;, 's) /'sEI/ sail, ('fa;, 'fa) /'faE/ le, ('b;, 'b)
/'bOE/ boil, but only: ('fa;) /'faO/ fowl, ('ju;U) /'jUu/ yule, ('khu;U) /'kUu/ cool
We nd then: ('s;) /'sOU/ soul and above all ('s;) /'sOU/ so, as well as: ('ju)
/'jUu/ you`, ('nj;u) /'nUu/ n[, ('h;u) /'hUu/ who (// can always be realized
as // ()).
Of course, the most peculiar diphthong is /OU/ (), not followed by (): ('n;3 3
54
'G;2 5w 'g;3 3) /'nOU. 'GOU, 'wOUnt 'gOU./ No, Joe won't go. e rst element of
/OU/ is central and unrounded, (), while in American pronunciation it is back
and rounded, (): ('n;23 'G;2 'w 'g;23)a. In the British accent, at the beginning of the twentieth century, (o) was widespread; until the fties it was (P),
always with lip rounding, while (, ), at that time, sounded rather aected.
2.5. British diphthongs.
/Ii/ (iI)
/Ii/ (Ii)
/EI[]/ (I, )
/aE[]/ (a, a)
/[j/]Uu/ ([j]u)
/[j/]Uu/ ([j]uU)
/OE[]/ (, )
/OU[]/ (, )
/aO[]/ (a[])
2.1.4.3. In the vocogram showing /I, E, [j]U/ ( 2.3), there are also /aE,
aO/. As a matter of fact, in a typical British pronunciation (besides remaining stable, as in American English), both can frequently reduce to /a/ (a;) (and
(a;|)): ('fa, 'fa|) and ('fa;, 'fa;|) /'faE[z]/ re(s). Otherwise, /aO/ can become /A:/ (A;) (and (A;|)), up to coincide with /A:[]/: ('Tha, 'Tha|),
('Tha;, 'Tha;|) (also ('ThA;, 'ThA;|), besides ('ThA:[])) /'taO[z]/ tower(s).
In this kind of pronunciation (sometimes dened smoothing), even the rarer
/EI, OE, OU/ can always be lessened, respectively, to /E:, O:, :/. us
/E:/ (up to coincide with /E/): ('lI, 'lI|), ('l;, 'l;|), ('l;[]) (and also
('l;[], 'l:[])) /'lEI[z]/ lZer(s); /O:/ (up to coincide with the old /O/): (m'phl, -|), (-;[], -;|), /Im'plOE[z]/ employer(s); /:/ (up to coincide
with /:/): ('m, 'm|), ('m;, 'm;|), ('m:[]) /'mOU[z]/ mower(s) (
('fa[], 'Tha[], 'lI[], m'phl[], 'm[])a).
2.1.4.4. In neutral (both American and Briti) pronunciation, /</ remains:
('m<i)a ('m>i)b /'m<i/ marry dierent from ('m<i)a ('m>i)b /'mE<i/ merry Besides, we have: ('sTA;<i)a ('sTA;>i)b /'stA:<i/ arry ('sAi)b /'s<i/ sorry
('sTO;<i)a ('sT;>i)b /'stO:<i/ or(e)y
e dierence only by now in traditional American pronunciation between
('O;<) /'O:</ aural and ('o;<) /'O</ (sometimes, represented as /'o:</ or, even,
as in the misleading transcription /'OU</) oral is neither neutral, nor any longer
modern: it is ('O;<) for both. At most, sometimes (in both accents), one can resort
to /'</ for oral in order to avoid ambiguities. But, for this very reason, not infrequently, people even say ('a<) /'aO</ aural]
Vowel diaphonemes
2.1.5.1. ere is a dierence, especially in British English, between ('f;i)a
('f;>i)b /'f:<i/ furry and ('h;i)a ('h>i)b /'h;<i/ hurry, occurring in just a few words.
2. english
55
In a diaphonemic transcription, it is conveniently represented by /:</ /;</, as

we have just seen.
We must now also introduce the diaphoneme /</, occurring in the context
/</, especially in the set /I<, E<, U</ (corresponding to /I, E, U/);
more rarely it occurs in /aE<, aO<, OE</, as well.
Typically, in American English, // is dropped and realized as zero; so we have:
('m<i)a ('m>i)b /'mE<i/ Mary (consequently, in American pronunciation, it is
the same as merry, and, in current and widespread pronunciation, which however is not neutral but mediatic, 2.4.2.2 the same goes for marry, too), ('khl<)a ('khl>)b /'klI<I/ cleari, ('khl<T)a ('khl>T)b /'klI<It/ clear it, ('D<)a
('Dj>)b /'dU<I/ duri. Often, we have: ('D;)a /'d:<I/ and ('Dj;>)b /'djO:<I/ (this example shows the diaphoneme //, too).
erefore, in American English, ('sp<T, -T) holds good both of /'spI<It/ spirit
{('sp<T, -T) ('sp>T)b), and of /'spI<It/ spear it {('sp<T, -T) ('sp>T)b). On the other hand, some speakers may distinguish, saying: ('sp<T, -T) /'spI<It/ and ('spIT, -T)
/'spI<It/. Furthermore: ('hI; n';)a ('h;> n';)b /'hI< n[d]'E/ here and
ère
2.1.5.2. en // is preceded by a consonant, it may be dropped (more frequently so in the British accent): (&l'mn[[]<i, -nT<i)a (&l'mN>i)b /El'mEnt<i/
elementary th this kind of sux, in American English, a secondary stress is
kept, when it is preceded by an unstressed syllable: ('DkS&<i, -&n<i)a ('DkS>i,
-ni)b /'dIkS<i/ dictionary ('khsT&m<i)a ('khsTm>i, -mi)b /'kstm<i/ cuomary (A;<'Thkjl&T<i)a (A;'Thkjl>i)b /A:'tIkjltK:<i/ articulatory It may thus be
convenient to use the diaphonemes /, K:/, as well. is is very economical, because
it helps save some of the space given to transcriptions, especially in dictionaries,
without renouncing precious information. For the last example given, there is a
variant (chiey British): (A;&Tkj'lI>i)b /A:tIkj'lEIt<i/).
Especially for Briti English, it may be useful to use the diaphoneme //, to
show the oscillation between (unstressed) /j/ and /jU/ (j, j): ('>gjl, -gj-) /'<Egjl/ regular /A:'tIkjltK:<i A:tIkj'lEIt<i/ (-kj-, -kj-), ('khmj&nz, -mj-); in
American English, /j/ denitely prevails.
Besides, the handy diaphonemes /ju, u/ may be useful, to show the uctuation
between (jUu'naT, j-)a (ju-, j-)b /ju'naEt/ unite, ('sTcUu, -)a (-u, -)b /'stcu/
atue
It is convenient to use the diaphoneme // in other contexts, too: (v'lAs[i, -sTi)a
(v'lsTi, -sTi)b /v'lsi/ veloci, ('nAvlsT, -vsT)a ('n-)b /'nvlIst/ noveli, ('gvn, -vn, -vn)a ('gv, -vn)b /'gvn/ governor .
2.1.5.3. Two other diaphonemes, /;, ;/, are more important in distinguishing
between the American and British modern neutral accents. e rst, /;/, shows
the dierence between //a (but it often behaves like a long monophthong /:/)
and /A:/b (chiey before /f, , s/ and before /0/), as in: ('g<[;]sp)a ('g>A;sp)b
/'g<;sp/ grasp ('l[;]sT)a ('lA;sT)b /'l;st/ la ('ph[;]s)a ('phA;s)b /'p;s/ pa ('h[;]f)a
('hA;f)b /'h;f/ half ('[;]fT)a ('A;fT)b /';ft/ after ('ph[;])a ('phA;)b /'p;/ pa`
56
More examples are: ('D[;]ns)a ('DA;ns)b /'d;ns/ dance ('phl[;]nT)a ('phlA;nT)b

/'pl;nt/ plant ('kh[;]nT)a ('khA;nT)b /'k;nt/ can't (g'z[;]mp)a (g'zA;mp)b /Ig'z;mp/ example (kh'm:n, -;-)a (-A:n)b /k'm;nd/ command ('[;]nT)a ('A;nT)b
/';nt/ aunt ('<[;])a ('>A;)b /'<;/ raèr (m'<;)a (-'>A:)b /m'<;/ morale,
(b'n[;]n)a (b'nA;n)b /b'n;n/ banana.
Even in British English, there are forms with //: ('sp) /'sp/ asp ('phnT)
/'pnt/ pant , of course, besides: ('b;nD) /'bnd/ band ('m;n) /'mn/ man
('ms) /'ms/ ma`s .
2.1.5.4. e second of these diaphonemes, /;/, shows the dierence between
/O:/a (but //a occurs, too) and //b (in particular before /f, , s, , g/ and /</), as
in: ('O;f, 'Af)a ('f)b /';f/ o ('khlO;, -A)a ('khl)b /'kl;/ clo` ('lO;s, 'lAs)a ('ls)b
/'l;s/ lo ('lO;sT, 'lAsT)a ('lsT)b /'l;st/ lo ('sO:, 'sA;)a ('s;)b /'s;/ so ('DO:,
'DA;)a ('D;)b /'d;g/ dog ('hO;<b, 'hA-)a ('h>b)b /'h;<b/ horrible ('O;<Gn,
'A-)a ('>Gn)b /';<Gn/ origin ('flO;<D, 'flA-)a ('fl>D)b /'fl;<d/ orida
('khwO;<, 'khwA-)a ('khw>)b /'kw;</ quarrel ('gO:n, 'gA;n)a ('g;n)b /'g;n/ gone
But, for /;/, there are even cases like: //a (but /O:/a is to be preferred) //b, as
in: ('sAi)b /'s<i/ sorry (Th'mA< -'mO;-)a (Th'm>)b /t'm;<OU/ tomorrow ('wAc 'wO;c)a ('wc)b /'w;c/ wab ('f<A; 'f<O:)a ('f>;)b /'f<;g/ frog
('chAklT 'chO;-)a ('chklT)b /'c;klt/ >ocolate ('DA; 'DO:)a and ('DAli 'DO;li)a
('D; 'Dli)b /'d;, 'd;li/ doll(y) ('wAnT, 'wO;nT)a ('wnT)b /'w;nt/ want
It is interesting to notice the peculiarity of ('wO;m, 'wAm)a ('w;T)b /'wO;/ water In England, ('wT) is no neutral pronunciation; it can be found most commonly in an area including Oxford and Reading. In a few words with /;f, ;, ;s,
;ft, ;st/, even in British English, a minority pronunciation with /O:/ is possible,
besides the preferred one with //; they are: o cough trough bro` fro` cro
lo to soft croft co fro lo oft often soften
2.1.5.5. ere are another couple of diaphonemes, /A;, O;/, which are relatively
less signicant, because /A;/ is used especially in words of foreign origin, written
with an a\ ('phA;sT)a ('phsT)b /'pA;st/ paa, (&vIiT'nA:m, -'-)a (-'n;m)b /vIiEt'nA;m/ Vietnam
ile /O;/ occurs in particular in words written with ausC aunC alC\ (O;'sTI;,
A-)a (;'sT;, -)b /O;'stI/ auere, ('hO;nT, 'hAnT)a ('h;nT)b /'hO;nt/ haunt ('sO;T,
'sAT)a ('s;T 'sT)b /'sO;t/ salt As the examples show, /O;/ concerns, above all,
American English. On the other hand, /A;/ may present twofold possibilities, in
both accents, according to words and to speakers.
e (socio)diaphoneme //
2.1.6.1. e last vowel diaphoneme we must consider is // = /, I/. is refers
to the alternation in the realizations of //: between // and /I/. Clearly, // () prevails in the American accent, while /I/ () prevails in the Briti one, even if things
are a little more complicated. Indeed, in American English, too, there are cases of
2. english
57
// = /I/, chiey in a more rend and more conservative way of speaking. By the
same token, in British English, there are cases of // = //, chiey in a less rened
and more innovative way of speaking. So, we nd a greater convergence at a more
up-to-date and modern level.
Here are a few examples: (s'vI;)a (s'v;)b /s'vI/ severe, ('ksGnT, 'gz-)a
(-G-)b /'EksGnt, 'Egz-/ exigent, (&khAmp<'h~Sn)a (&khmp>'h~S)b /kmp<'hEnSn/.
2.1.6.2. It may be a good idea to take stock of the situation about some (real or
seeming) suxes and prexes, because we still nd old-fashioned and outdated transcriptions, especially in bilingual dictionaries. e modern neutral pronunciation,
British too, by now, has // (while /I/ sounds quite pompous) in: -ace ('phls)
/'pls/ palace -ate ('chAklT)a ('chklT)b /'cklt/ >ocolate -iI ('hpli) /'hpli/ happiI -i ('khwAn[[][i)a ('khwnTTi)b /'kwni/ quanti
en /-i/ is preceded by /s/, it can often become /-sti/. So, a notation like /-si/
includes both possibilities, while excluding that the two diaphonemes /, / may
work together, because of contextual incompatibilities. at means that, if // falls,
then // automatically becomes /t/, because it is preceded by /s/): necei universi veloci /n'sEsi, jUun'v:si, v'lsi/.
Besides, we have: -le ('hpls)a ('hpls)b /'hOUpls/ hopele -ne ('gDns)
/'gUdns/ goodne For -e /s/ prevails, chiey in American English, while in British English /Is/ is also possible (actre waitre); in some cases, /Es/ too (du>e);
for prince we have: ('ph<nss, -s)a (p>n'ss, 'ph>nss)b. To end with, -let ('b<IslT)a ('b>-)b /'b<EIslt/ bracelet -ret ('skA;<lT)a ('skA;lT)b /'skA:lt/ scarlet
2.1.6.3. We have // (which means, mainly // in American, but /I/ in Briti
English] for: -ed ('wI[)a ('wIT)b /'wEId/ waited -es ('h;<)a ('h;s)b /'hO:sz/ horses -e ('bgsT)a ('bgsT)b /'bIgIst/ bigge -et ('ThkT)a (-T)b /'tIkt/ ti$et
('vEvT)a ('vvT)b /'vEvt/ velvet (but ('nT<sT, 'n[[]&<sT)a ('n>sT, 'nT&>sT)b
/'Int<st, 'In<Est/ intere] -ite ('ApzT, -s-)a ('pzT, -s-)b /'pzt, -s-/ opposite
Besides: -ice ('Afs, 'O;-)a ('fs)b /';fs/ oce -ine (g'zmn)a (-n)b /Ig'zmn/
examine -i ('v<&fa)a ('v>&fa)b /'vE<faE/ veri
Furthermore: be- (b'khm:)a (b'khm:)b /b'km/ become de- (D'm;n,
-:n)a (D'mA:n)b /d'm;nd/ demand (but de /dIi-/: (DIi'nIc)a (-c)b /dIi'nEIc/ denature] pre- (ph<'Thn:)a (ph>-)b /p<'tEnd/ pretend re- (<'Tha)a (>'Tha)b /<'taE/ retire (but re- /<Ii-/: (<Ii'g;In)a (>Ii-)b /<Ii'gEIn/ regain]
In words like become demand pretend retire eleven we could add that // has
a possible variant /i/ (or even /Ii/). is, generally, belongs to a formal American
pronunciation; while, the British one is at the opposite side. erefore, it is safer
to stick to what we have just said. Of course, everyone will decide for themselves,
especially through the regular consultation of a reliable pronunciation dictionary
(but it is much better to look up words, regularly, in more than one dictionary).
To end with, most internal -e- -i- (in unchecked syllables), generally, have //
(namely, as a trend, // in American and /I/ in British pronunciation): ('lmnT)a
('lmnT)b /'Elmnt/ element (&nsT'ThUuSn)a (&nsT'ThjuS)b /Inst'tUuSn/ initution
58
2.1.6.4. On the contrary, regularly we have /I/ in: -ic(s) (f'n[k[s])a (-Tk[s])b /f'nEIk[s]/ #onetic(s) -ical (s'Th<k)a (->-)b /s'tI<Ik/ sarical -i ('sTnD)
/'stndI/ andi -ip ('f<~Sp)a ('f>-)b /'f<EndSIp/ friendip -ive (n'Thns)
/In'tEnsIv/ intensive
is group includes: -age ('vl) /'vIlIG/ village -i ('glS) /'IglIS/ Engli -i
('lgwsT) /'lIgwIst/ liui e- and i- (at the beginning of words, in unchecked
syllables): ('lvn) /I'lEvn/ eleven ('Thjn) /I'tljn/ Italian im- and in- (in
checked syllables): (m'ph;<TT, -<T)a (-;T- -T)b /Im'pO:tt/ important (n'sT;) /In'stEd/ inead In some American pronunciations, all these words may
have either /I/ or //.
As far as // is concerned, Oceanian and South African English are more like
American English.
Usually, // is unstressed, being the weak vowel par excellence. ere are two
forms, however, that are very often heard even with stressed // (except in formal
pronunciation): ('chD<n, 'ch-, 'ch-, 'ch-)a (->-)b /'cId<n, 'c-, 'cU-/ >ildren ('DzT, 'DT, -) /'dzt, 'd-/ doesn't (this is given in an example in 2.6.4,
too). Let us consider also the possibility of restressing for emphasis, as in: Oh, I
didn't buy it: it wasn't the ('I;i, ') dre, it was ju a (';I, ') dre (example
adapted from a phonetics newsgroup).
Consonants
2.2.0. At the beginning of this section, we will show the table of the consonantal articulations of (American and British) neutral English: 2.6. It is useful to
make regular reference to it, in order to thoroughly understand the English consonantal system.
1.9-15 show the orograms of all the contoids needed to describe English (and
the other languages dealt with in HPr), including secondary, occasional, or regional variants, arranged according to their manner of articulation.
F pb
(M)
f v
(n)
n
(t d) T D
( )b
k g
cG
laryngeal
velar rounded
velar
palatal
{postalveolariz.}
prevelar round.
(~)

s z ( )a
([|m)a
l
prepalatal
postalveopalatal
protruded
postalveolar prot.
postalveolar
velarized alveol.
alveolar
dental
labiodental
bilabial
3.6. Table of neutral English consonants.
()
S Z
j
(>)b
<a
()
// ([a, ma, Tb), //l// /l, / (l, -lj, -j, 0, ), /</ (<a, >b)
h (H)
2. english
59
Nasals
2.2.1.1. English has three nasal phonemes\ /m, n, /. e velar phoneme does
not occur at the beginning of English words, but it is normal in internal and nal
positions: ('n;Im) /'nEIm/ name, ('n:) /'In/ ìn, (':) /'I/ ì, ('b;) /'b/
ba, (b'khm)a (b'khm)b /b'kmI/ becomi, ('s) /'sII/ sii,
('lO: 'g;, 'lA;)a ('l; 'g;)b /'l; 'gOU/ lo ago, ('sT<[k])a ('s>-)b
/'st<EI/ reèni
In unstressed syllables, after /t, d , s, z/, there is a typical intense (syllabic)
realization of /n/ {() //; after /, /, we nd ((2))}. It is denitely worthwhile to
use it in (dia)phonemic transcriptions, as well, although clearly its origin is /n/:
('khAT, -)a ('khT, -)b /'kt/ cotton, ('wD) /'wUd/ wooden, (';)a (';)b
((-2)) /':/ earèn, ('m:< 'T)a ('m:)b ((2-)) /'mO: n't/ more àn àt,
('f[;]s)a ('fA;s)b /'f;s/ faen, ('chz)a ('chz)b /'cOUz/ >osen After /0t,
0d/, it is best to use /n/: ('wnsTn) /'wInstn/ Winon ('hDn)a ('h-)b
/'hOUdn/ Holden
Furthermore, we have /n/ (= (n)a ()b) after /S, Z/: (ph<&nnsi'ISn)a (--, -S)b
/p<nnsi'EISn/ pronunciation (ph'<Zn)a (->Z)b /p'<IZn/ Parisian In a preintoneme, (n) can easily become (), chiey after /S, Z c, G/.
After other consonants, /n/ (n, n) is more usual: ('l~cn)a ('l-)b /'lncn/
lun>eon (<'lGn)a (>-)b /<'lIGn/ religion ('<bn)a ('>bn)b /'<Ibn/ ribbon,
('~jn)a ('~jn)b /'njn/ onion, ('ThIkn) /'tEIkn/ taken
en speed is higher, especially in a preintoneme, we can go as far as (c, G
p, b k, g): ('l~cn, -c)a ('l-)b (<'lGn, -G)a (>-)b ('pn, -p)a
('-)b /'OUpn/ open ('<bn, -b)a ('>-)b ('ThIkn, -k), (';<gn, -g)a (';g-)b
/'O:gn/ organ On the other hand, when speed is lower, or for emphasis, // can
easily become (n): ('ls, -sn) /'lIs/ lien
Regularly, we have: (n[D]n) /n[d]n/: ('lnn) /'lEnn/ Lennon, ('lnDn)a ('ln-)b
/'lndn/ London but, ('wAnT, -n, -nTn, 'wO;n-)a ('wnTn -n)b /'w;ntn/
wanton
2.2.1.2. Aimilation is very important, and it must not be neglected either in
the description of languages, or in teaching and learning. Let us notice that, of the
three English nasal phonemes, the two marked ones, /m, /, resist well; while, the
unmarked one, /n/, undergoes several changes, contrary to what phonemic transcriptions generally seem to indicate.
However, proceeding in an organized manner, we have: ('D<I;im)a ('>-)b
/'d<Iimd/ dreamed, ('D<mT)a ('>-)b /'d<Emt/ dreamt, ('sm&Tham)a ('sm-)b /'smtaEmz/ sometimes, ('s;Im 'kha;n) /'sEIm 'kaEnd/ same kind, (sm'Gli) /sm'GEli/
some jelI ('s;IM 'vs, -m '6s) /'sEIm 'vOEs/ same voice, ('s;IM 'fkT, -M] 'f-, -m
'kT) /'sEIm 'fkt/ same fact, ('khMfT, -M]f-, -m-)a ('khMfT, -M]f-, -m-)b
/'kmft/ comfort As can be seen, only with labiodentals, /m/ shows a slight let-up, becoming labiodental, (Mf M]f), but this happens just in trivial cases from a
lexical-semantic point of view. Otherwise, /f, v/ may become bilabial (constrictives): (m, m6).
60
Our examples also show that at present a homorganic (labiodental) stop may
very often be inserted into the sequence (mf) ( only with voiceless /f/); but a too
frequent use is better avoided. is homorganic insertion can happen with other
sequences too: ('w;<m, -mp)a ('w;m-)b /'wO:m/ warm`, ('mTi, 'mpTi) /'Emti/ emp, ('sms, 'smps) /'sIms/ Sim(p)son.
Even // resists well: ('<O:, '<A;)a ('>;)b /'<;d/ wroed, ('w&bk) /'wIbk/ wiba$ Only for the sux -i () /I/ (but at a non-neutral level, not to
be followed), can we have /In, n, /: ('sl -n -n) /'sElI/ selli. We can also
nd: ('l, -k) /'lE/ le` ('khsTn, -ks-, -z-, -T) /'kInstn/ Kion ('za[i, g-)a (-Ti)b /'zaEi/ anxie, ('Ss, 'k-) /'Ss/ anxious.
2.2.1.3. On the contrary, except in a very accurate way of speaking (even too
accurate!), /n/ assimilates to a following contoid: (m'blk) /Im'blk/ in bla$ (M'vaT) /In'vaEt/ invite (M'fkT M]-) /In'fEkt/ infect ('bAks, -)a (-ks)b /In'bks/ in è box ('Th -t) /'tEn/ ten` ('Thns -ts) /'tEns/ tense ('phns,
-ts)a (-)b /'pIns/ pincer (for (()) see below).
For some time, /nz/ has been simplied (but /ndz/ can always be restored, according to current spelling): ('wnz, -ndz)a (-)b /'wInz/ Windsor ('lnzi, -dzi)
/'lInzi/ Linds\ Even in /nz/, a /d/ can be inserted: ('khln:, -n:d) /'klEnz/ cleanse
('khlnz, -ndz) /'klEnzI/ cleansi. But this is less and less recommendable,
specially with grammemes: ('Thjn -nd -) /I'tljnz, -/ Italians ('GA;n
-nd)a ('G;-)b /'Gnz/ John's
Furthermore, we have (~) (postalveopalatal, ((n))): ('~c) /'Inc/ in> (~'chan)a
(-)b /In'caEn/ in ina ('I~G) /'EInG/ ael ('Th~Sn -~cn -~+Sn)a (-~S
-~c -~+S)b attention But we nd (~) (prepalatal) before /j/ (if this is heterosyllabic): (khm'ph~jn) /km'pnjn/ companion ('~jn)a ('-)b /'njn/ onion
Besides: ('hn<i)a ('hN>i)b /'hEn<i/ Henry (n'Th<nsk)a (N'h>-)b /In't<InsIk/ intrinsic ('Th 'g:, 'g;)a ('g:)b /'tEn 'g:z/ ten girls . Also (m'wn[m])a
(-nT)b /In'wIn/ in winter should be noticed.
A better transcription for (~c, ~G, ~+S) would be ((nc, nG, ntS)), with ((n)) and
even ((t)) (for a postalveopalatal stop). It was stated above that dental (()) could
be represented with (n), as well, chiey in (n), because the simple fact that /n/
has (dental) () allows us to infer that we automatically have (()), by assimilation.
On the contrary, a special symbol would be more important in (()), to show
that it is not (nn) (alveolar, but dental, coming from (()) /n/), as, for instance,
in: (n'n;&Tha;m)a (-;-)b /In'nOUtaEm/ in no time. On the other hand, for some
speakers, a less completely assimilated realization produces exactly (nn). is is
not very dierent auditorily, so more precise symbols could even be avoided, with
no real drawbacks.
In lexical composition, as well as for the negative prex un-, in a slow careful
way of speaking, people try to keep (n), while currently assimilation to the place
of articulation of a following consonant is quite regular.
As an actual compromise, here we will show that it is possible to maintain an
apical contact while adding a secondary coarticulation (with no full contact) bilabial, (); labiodental, (); velar, (n): (&mb'lIivb, &-)a (&mb-, &-)b /nb-
2. english
61
'lIivb/ unbelievable (M'f<nDli, -)a (M'f>-, -)b /n'f<Endli/ unfriendI ('g<ITf, n-, -f)a ('g>-, n-, -f)b /n'g<EItf/ urateful.
In all other cases, with less dierent coarticulations, assimilation (which is often considered less recommendable, on mere written and grammatical bases) is
more elusive. It is therefore used spontaneously, though unconsciously.
To end, simplication is also possible, though less frequently than in the past,
in cases like: ('snts -ns) /'sEnts/ cen_ ('pha;nd -n) /'paOndz/ pounds ('l~c
-~S)a ('l-)b /'lnc/ lun> ('l~cn -~Sn)a ('l-)b /'lncn/ lun>eon (<'v~: -~:)a
(>-)b /<'vEnG/ revee ('I~G -~Z -G, -Z) /'EInG -G/ ael
Stops
2.2.2.1. ere are three diphonic pairs of ops ( pairs of both a voiceless and
a voiced articulation): /p, b t, d k, g/. Of course, /t, d/ are alveolar (T, D): (Th'D;I)
/t'dEI/ todZ, ('DsT&TUuT)a ('DsT&TjuT)b /'dEsttUut/ deitute However, before /,
s, z/, /t, d/ become dental (t, d): ('It) /'EIt/ eigh`, ('w) /'wId/ wid`,
('hts) /'hts/ ha_, ('h;d) /'hEdz/ heads
On the contrary, before /</ (>)b, in British English, /t, d/ become postalveolar,
always due to assimilation: ('Th<a;)a ('h>a;)b /'t<aE/ try, ('khnT<i)a ('khN>i)b
/'knt<i/ country, ('D<k)a ('>k)b /'d<Ik/ drink
For /t<, d</, however, several pronunciations are possible, mostly with /t, d/ realized as stop-strictives (or aricates): in American English ([h]<, < C[h]<, <
c[h]<, G<)a; in British English ([h]>, > [h]>, "> C[h]>, > c[h]>, G>)b.
Furthermore, a (homorganic) constrictive realization is possible for /</ (which is
then, respectively, alveolar or postalveolar rounded, (, )): (T[h], D)a ([h], )b.
So: ('Th<a;, 'h<-, 'Ch<-, 'ch<-, 'Th-)a ('h>a;, 'h>-, 'h>-, 'Ch>-, 'ch>-, 'h-)b /'t<aE/
try, ('khnT<i, -<i, -C<i, -c<i, -Ti)a ('khN>i, ->i, ->i, -C>i, -c>i, -i)b /'knt<i/
country, ('D<k, '<-, '<-, 'G<-, 'D-)a ('>k, '>-, '">-, '>-, 'G>-, '-)b /'d<Ik/
drink
All these pronunciations are possible as neutral ones too, although opinions regarding their correctness may be inuenced by spelling. On the other hand, from
a structural point of view, /c<, G</ could represent a tting parallelism with /S</,
as in ('S<k)a ('S>-)b /'S<Ik/ rink
2.2.2.2. Some other transformations of /t, d/ are much more signicant; indeed,
although to foreigners (T, D) might seem more marked, actually, in the natives'
phonological system, /t, d/ are an unmarked diphonic pair of stops (as happens to
/n/ in comparison with /m, /). From an articulatory point of view, /t, d/ are liable
to assimilation; not to complicate things, but rather to make them easier.
So, /t, d/, before /p, b, m, w/, generally, become (p, b): (&p'm;n) /t'mn/
àt man (&p'b;) /t'bOE/ àt boy ('nAp 'wn:)a ('np 'wn:)b /'nt 'wn/ not
one; likewise, before /k, g/, they become (k, g): ('wAk k~j'DU;u, 'w-)a ('wk k~j'D;u)b /'wt knj'dUu/ what can you do? (&k'g:, -;)a (-:)b /t'g:/ àt
girl
62
Even the rare sequences /pf, bv/ present some kind of assimilation (in one direction or in the other): ('khp, -]f)a ('kh-)b /'kpfU/ cupful, ('Ab6is, 'A7v-)a
('-)b /'bvis/ obvious
Prevelar articulations, which are automatic by coarticulation, need not be expressly written down: ('k, 'gT) {((', 'T))} /'IkI, 'gEt/ ìnki get
2.2.2.3. One fundamental thing, already seen in previous examples, which must
not be neglected in learning and teaching, is that, in stressed syllables, initial /p,
t, k/ are aspirated (unless they are preceded by /s/ in the same syllable and in a
same lexeme), also after silence (even in an unstressed syllable) (0h) /0/: (Th'D;I) /t'dEI/ todZ, ('phIik) /'pIik/ peak (but: ('spIik) /'spIik/ speak] ('ThIk) /'tEIk/
take (but we have: ('sTIk) /'stEIk/ ake] ('kh;)a ('kh;)b /'kE/ care (but: ('sk;)a
('sk;)b /'skE/ scare]. Nevertheless, one should notice: (ms'ph<n[[])a (ms'ph>nT)b /mIs'p<Ind/ misprinted (with dierent phono-syllables and dierent
morphemes).
American t // ([, m)
2.2.3.1. An important characteristic of the neutral American accent (which is,
however, not neutral in the British accent, although it is fairly widespread) regards
/t/ which, in given contexts, is realized as a voiced alveolar ap, ([) (which before
() is lateralized, as well, laterally contracted: (m)).
But, let us see, rst, when it remains a voiceless alveolar stop (though, in certain
cases, it may become a laryngeal or glottal stop, ()).
In stressed (even unaspirated, in /'st/) or in half-stressed syllable: ('Thn:) /'tEn/
ten, ('sTm:) /'stEm/ em, (A;<'Thkjl&T<i)a (A;'Thkjl>i)b /A:'tIkjltK:<i/ articulatory (+ (A;&Tkj'lI>i)b /A:tIkj'lEIt<i/}, ('<Uum&Tz)a ('>u-)b /'<UumtIz/ rheumatism
After a pause or after consonants (dierent from /n, , /): (Th'ThIk) /t'tEIk/ to
take, ('kT)a (-)b /'kt/ actor, ('[;]fT)a ('A;fT)b /';ft/ after, ('m[p]Ti) /'Em[p]ti/ emp
Before consonants\ ('chTni, -ni)a ('chTni, -ni)b /'ctni/ utn\, ('phT<i)a
('ph>i)b /'pOUt<i/ poetry, ('lITli, -li) /'lEItli/ lateI, (T'ln[[]k, 'l-)a (T'lnTk,
'l-)b /t'lnIk/ Atlantic.
It remains (T) even in words in -Vtic (with no secondary stress, too): ('lUunTk)a
('lu-)b /'lUuntIk/ lunatic, ('phAlTk)a ('ph-)b /'pltIk/ politic, ('<mTk)a ('>-)b /'<ImtIk/ ari`metic It is the same even between /, / and //: ('n;<T,
-<)a ('n;T)b /'nO:t/ Norton, ('hT, -)a ('hT)b /'hIt/ Hilton Besides, we
nd: ('khlnTn, -nT, -n)a ('khlnTn -n)b /'klIntn, -nt/ Clinton
2.2.3.2. Let us now turn to the contexts where /t/ (T) becomes // ([, m), in normal (not slow, nor particularly careful) speaking.
Between a stressed (or unstressed) vowel and another vowel, or (, ): ('b[i)a
('bTi)b /'bEi/ Bet, ('vIi[)a ('vIiT)b /'vIiOU/ veto, (&vz'bl[i)a (-Ti)b /vIz'bIl-
2. english
63
i/ visibili, (&['msTk)a (&T-)b /'mIstIk/ atomiic, ('lm)a ('lT)b /'lE/ letter, ('l[)a ('lT)b /'lI/ little
Between /n, , / and a vowel, or (, ) (remembering that, as our examples show,
([, m) may often be dropped after /n/, (n[[], n[m])): ('bn[[]m)a ('bnTm)b /'bnm/ bantam, ('wn[m])a ('wnT)b /'wIn/ winter ('mn[[])a ('mnT)b /'mn/
mantel, (';[i)a (';Ti)b /':i/ ìr, ('phA;<[i)a ('phA;Ti)b /'pA:i/ par, ('SEm, -T)a
('ST)b /'SE/ elter
Even before a stressed vowel (provided it is heterosyllabic): (ph'ThI[, ph['I[)a (ph'ThIT)b /p'tEIOU p'EIOU/ potato, (['O:)a ('Th:)b /'O:, 'tO:/ at
all (it should be noted that there is a dierence, between the two accents, for at all]
e sequences /n, , / may even have fused realizations, (, m, e): ('phI)a
/'pEInI/ painti, ('phA;mi)a /'pA:i/ par, ('fO;ei)a /'fO:i/ faul.
However, in American English, when speed is reduced, or when more attention
is paid to the way of speaking, // ([) becomes /t/ (T): ('lm, 'lT) /'lE/ letter
('wn[m], -nT) /'wIn/ winter e same goes, even in a normal manner of speaking, for //: ('SEm, -T) /'SE/ elter, ('O;m, -T, 'A-) /'O;/ alter is happens
even to less common words, such as: ('vIi[, -T) /'vIiOU/ veto, ('phlI[, -T)
/'plEIOU/ Plato (d'fIi[&z, -&Tz) /dI'fIitIz/ defeatism.
Also in Australia, New Zealand, South Africa (and in towns in Wales and Ireland) // is ([, m); while, Scotland is well-known for // = (), even between vowels
and before () // (and even before its typical pronunciation of (R) //). e same
change, // = (), occurs even in broad accents in England, in particular, in London, Birmingham, .
Unexplosion
2.2.4.1. Notably, in English, stops (both voiced and voiceless) are unreleased,
chiey after /[m, n, , , ]/, before pauses or consonants. is means that their
third phase ( their o-glide) is inaudible, incomplete. It is useful to put the diacritic () after a proper symbol, to show unreleased contoids, especially at rst and,
of course, when it is the subject in question, as here.
erefore, (voiced or voiceless) stops are unreleased after (stressed or unstressed)
vowels, even followed by homorganic N (/mp, mb nt, nd k, g/), or by /, /.
is holds good except in very slow or careful pronunciation.
Here are some examples, although it is to be remembered that, when single
words are said in isolation, before a pause, it is more usual to produce (and let
hear) the o-glide, indicated by ($): ('bA;)a ('b;)b /'bb/ Bob ('khp)a ('khp)b
/'kp/ cup ('khmp) /'kmp/ camp ('hT) /'ht/ hat ('h;n) /'hnd/ hand
('<;)a ('>;)b /'<Ed/ red ('S;<T)a ('S;T)b /'SO:t/ ort ('fO;T, 'fAT)a ('f;T, 'fT)b
/'fO;t/ fault ('blk) /'blk/ bla$ ('bk) /'bk/ bank ('D;) /'dIg/ dig. en
said in isolation, as plain examples, before a pause, they are released\ ('bA;$)a
('b;$)b ('h;n$) ('S;<T$)a ('S;T$)b ('bk$), .
Before a consonant, they are unreleased: ('kT) /'kt/ act ('kT)a (-)b /'kt/
actor ('<;b)a ('>;b)b /'<bd/ rubbed, ('Tkns, 'kk-) /'tkIns/ Atkinson,
64
('b;g) /'bEgd/ begged ('sTApT)a ('sTpT)b /'stpt/ opped, ('hAT&dA, -O;, --)a ('hTD, --)b /'htd;g/ hotdog ('sTAp 'DUu)a ('sTp 'Du)b /'stp 'dUuI/ op doi
More examples: ('Gk 'phi;I, -iI) /'Gk 'pIi/ Ja$ Peel, ('bA;b 'g;)a ('b;b
'g;)b /'bb 'gOUz/ Bob goes ('skATln)a ('skT-)b /'sktlnd/ Scotland (p'bats
T'b-) /It'baEts/ it bites (T'ThIks, 'Th-) /It'tEIks/ it takes (k'kh-)b
/It'k<iz/ it carries
Intermediate articulations are possible between the two extreme ones, above all
when people pay special attention to their speech (although this must not lead us
to think that these pronunciations are necessarily better). In fact, for /t, d/ (+ /p,
b k, g/, in addition to (pp, pb p, bb kk, kg k, gg)), the alveolar contact can
be maintained, by adding either a bilabial, (B, ), or a velar, (, ), coarticulation.
A bilabial or velar articulation is also possible, to which an alveolar coarticulation
can be added: respectively (Q, ) or ($, ).
2.2.4.2. e so-called nasal and lateral explosions are included in this group.
It is essential that transitions from /t, d/ to /n, l, / are direct, with no o-glide
similar to aspiration, and even with no insertion of vocoids.
erefore, we have: ('phTni) /'pItni/ Pitn\, ('wD) /'wUd/ wooden, (khm'phlIiTli, -li) /km'plIitli/ completeI, ('l[)a ('lT)b /'lI/ little, ('swnD) /'swInd/
~indle.
Laryngeal stop ()
2.2.5.1. It is a good thing to include the laryngeal (or glottal) stop () in the
symbols inventory of the English phonological system, even if, strictly speaking,
there are no (classical) minimal pairs, in order to declare its phonological status.
e fact is that it is important, too, to have () from a descriptive and teaching
point of view.
In the (American and British) neutral pronunciation () is used, when there is
some emphasis, before vowels, especially stressed ones: (ts'O;f)a (-';-)b /Its'[]O:f/
it's Yful! In British pronunciation, () may be used even to avoid the insertion of
a non-etymological /</ (intrusive, at the end of 2.2.9.4): ('l: n';D) /'lO: n[d]'O:d/ lY and order instead of the frequent ('l:> n';D) (('lO: n';<D)a).
2.2.5.2. Furthermore, even in neutral pronunciation, before consonants, we often have /t/ = (): ('skAlnd)a ('sk-)b ('bats) ('ThIks) ('kh<i)a (->-)b
(adjusting some examples just seen).
In a more systematic way, proceeding by groups, this occurs with vowels (chiefly, but not necessarily, stressed; and even with interspersed sonants, /m, n, , , /),
before /m, n, l, < j, w/: ('naT&m;, -p-, --)a (-;)b /'naEtmE/ nightmare, ('nAT
'na;, -)a ('n-)b /'nt 'naO/ not now, ('nAT 'jT, 'nA)a ('n-)b /'nt 'jEt/ not yet (also
(&nAc'T, n-, -)a (&nc'T, n-, -)b), ('lITli, -li) /'lEItli/ lateI
Before /t, d c, G/: ('gT 'Da;n, -) /'gEt 'daOn/ get down, ('g<IT 'Gk, -)a ('g>-,
-k)b /'g<EIt 'GOUk/ great joke Also before /p, b k, g f, v , s, z S, Z/: ('fp&bO;,
2. english
65
--, -T-)a (-:)b /'fUtbO:/ football, ('khk&gT, --, -T-)a (--)b /'ktgt/ catgut ('at ';,
-)a (-;)b /'aOt 'E/ out ère, ('nT&SE, --)a ('nT&S, --)b /'ntSE/ nutell
Let us also notice: (a'kh[;]nT {'DUuT}, -n, -n {--, -})a (-A;n- {-uT, -})b /aE'k;nt {'dUuIt}/ I can't [do it] (wi'wnT {'DUuT}, -n, -n {--, -})a (-n- {-uT, -})b
/wi'wOUnt {'dUuIt}/ we won't [do it] In absolute nal position, () is not used, except for /It, t/, and this only in informal speech: (wi'wAnT, -O;nT)a (-nT)b /wi'w;nt/ we want (wi'wAn[[]T, -T, -)a (-nTT, -)b /wi'w;nIt/ we want it ('ThkT,
-)a (-T, -)b /'tIkt/ ti$et.
However, the change /t/ = () is less frequent before /h/: ('IT 'hts -) /'EIt 'hts/
eight ha_ It sometimes occurs in /t/: ('khAT, -)a ('khT -)b /'kt/ cotton
In /nt/, it is more frequent (but it is only possible with /, /): ('sk<nT, -,
-[[]n)a ('sk>nTn -)b /'sk<ntn/ Scranton ('khlnT, -, -Tn)a ('khlnTn -)b
/'klIntn/ Clinton ('ET, -, -[n)a ('T, -Tn -)b /'klIntn/ Elton
/t/ = () is no neutral pronunciation, before /, , /: ('b[i)a ('bTi)b /'bEi/ Bet,
('bm)a ('bT)b /'bE/ better, ('l[)a ('lT)b /'lI/ little (('bi), ('b)a, ('b)b,
('l)). Nevertheless, it can occur in /t'/: (&nAT'Iivn, &nA['-, &nA'-)a ('n-)b /nt'Iivn/ not even It is accepted even for it before a pause: ('ThIkT, -) /'tEIkIt/ take it
ereas, it is possible, for /p/, (only) before /p, b/, and for /k/, (only) before /k,
g/ (otherwise, it is not neutral): ('sp 'phaD, -)a ('s-, -D)b /'sOUp 'paOd/ soap
powder, ('bk&khIs, -&kh-) /'bUkkEIs/ bookcase, ('bk 'gA;<D, -)a ('gA;D)b /'bk
'gA:d/ ba$ garden
British glottalization
2.2.6.1. As far as Briti pronunciation is concerned we must report the glottalization of /p, t, k/, before a pause or a consonant. By the end of the nineteenth
century, it was only occasional, but it is now very widespread, often, even among
ne neutral speakers.
Naturally, there are gradations, both in intensity and in frequency. It is therefore not really necessary to introduce glottalization into pronunciation. However,
its complete avoidance may sound too accurate or even pretentious.
In any case, it might be more advisable to restrict it to the rst level, to simultaneous glottalization, or real glottalization (or synglottalization). Consequently,
while a stop (0) is being articulated, at the same time, a laryngeal (or glottal)
stop () is produced. is is not added before the contoid (0) giving two
phones (or two segments), but is simply coarticulated with that () so that this
additional closure is not too intrusive: (P, T, ).
Here are some examples, showing the absence or presence of synglottalization:
('phIip, 'phIiP) /'pIip/ peep, ('hT, 'hT) /'ht/ hot, ('bk, 'b) /'bk/ ba$.
Furthermore, before contoids: ('kT, 'T) /'kt/ act ('kT, 'T) /'kt/ actor, ('Tkns, 'Tk-, 'kk-, 'k-) /'tkIns/ Atkinson, ('hT&D, --, 'hT-) /'htd;g/ hotdog ('sTp 'Du, 'sTP) /'stp 'dUuI/ op doi
More examples: ('Gk 'phi;I, 'G, -iI) /'Gk 'pIi/ Ja$ Peel (p'bats, P-, -
T- T-, -ts) /It'baEts/ it bites (T'ThIks, T-, -, -s) /It'tEIks/ it takes (k'kh>i, -
66
T- T-) /It'k<iz/ it carries ('skTln, 'skT-) /'sktlnd/ Scotland ('Thp&msT,

-P&m-) /'tpmOUst/ topmo, ('phTni, -Tni) /'ptni/ Putn\.
Still more examples: ('ph>, ->-) /'pEt</ petrol ('Iikw, -w-) /'Iikw/ equal
('phpjl, -Pj-) /'ppjl/ popular (k'saT, -) /Ik'saEd/ excited ('bts, -ts)
/'bEts/ be_ ('bts, -ts) /'bEts/ bel_ ('bnts, -ts) /'bEnts/ ben_ ('wks, -s) /'wIks/
wi$s ('wks, -s) /'wIks/ Wilkes ('wks, -s) /'wIks/ winks
For the substitution of /t/ with () 2.2.5.2.
2.2.6.2. A stronger degree of glottalization is the glottal reinforcement, producing () just before /p, t, k/: (p, T, k) (therefore, preglottalization). is is
more evident and cumbersome since we have two segments, two phones, even if
the laryngeal stop is unreleased, while, in these cases, /p, t, k/ are actually released:
((p$, T$, k$)) before pauses (but not before contoids, in a sentence).
Let us see, now, our examples (in progression): ('phIip, 'phIiP, 'phIip) /'pIip/
peep, ('hT, 'hT, 'hT) /'ht/ hot, ('bk, 'b, 'bk) /'bk/ ba$.
Before contoids: ('kT, 'T, 'kT) /'kt/ act ('kT, 'T, 'kT) /'kt/ actor,
('Tkns, 'kk-, 'Tk-, 'k-, 'Tk-, 'kk-) /'tkIns/ Atkinson, ('hT&D, --,
'hT-, 'hT-) /'htd;g/ hotdog ('sTp 'Du, 'sTP, 'sTp) /'stp 'dUuI/ op doi
nally: ('Gk 'phi;I, 'G, 'Gk, -iI) /'Gk 'pIi/ Ja$ Peel (p'bats, P-, p-
T- T, T-, -ts -ts) /It'baEts/ it bites (T'ThIks, T-, -, -s -ks) /It'tEIks/ it takes
(k'kh>i, -, k- T- T-T-) /It'k<iz/ it carries ('skTln, 'skT-, 'skT-) /'sktlnd/ Scotland ('Thp&msT, -P&m-, -p&m-) /'tpmOUst/ topmo, ('phTni, -Tni,
-Tni) /'ptni/ Putn\.
And: ('ph>, ->-, ->-) /'pEt</ petrol ('Iikw, -w-, -kw-) /'Iikw/ equal ('phpjl, -Pj-, -pj-) /'ppjl/ popular (k'saT, -, k-) /Ik'saEd/ excited ('bts, -ts,
-ts) /'bEts/ be_ ('bts, -ts, -ts) /'bEts/ bel_ ('bnts, -ts, -ts) /'bEnts/ ben_ ('wks,
-s, -ks) /'wIks/ wi$s ('wks, -s, -ks) /'wIks/ Wilkes ('wks, -s, -ks) /'wIks/
winks
e replacement of /t/ by () has been dealt with above ( 2.2.5.2).
Lenitions
2.2.6.3. In quick informal speech, in British pronunciation, simple /p, b t, d
k, g/, before unstressed vowels may be weakened, and transformed into constrictive phones (more or less tense, while the two apical ones are slit constrictives,
dierent from the more usual grooved constrictives, (s, z)), (, 6 Q, z x, ): ('phIp, 'phI) /'pEIpz/ papers, ('>b, '>6) /'<b/ rubber, ('lT, 'lQ) /'lE/ letter, ('fID, 'fIz) /'fEIdI/ fadi, ('bIk, 'bIx) /'bEIk/ baker, ('Dg, 'D)
/'dIgI/ diggi.
In American pronunciation, one possible lenition is just a partial voicing of /p,
k/, (p, k, ): ('phIp, -p, -) /'pEIpz/ papers, ('bIk, -k, -) /'bEIk/ baker A further variation of /t/ = // = ([, m), may be (A, b, r) ({partially} devoiced or
{totally} voiceless): ('bTi, -[i, -Ai, -i) /'bEi/ Bet ('lT, -m, -b, -r) /'lE/ letter
In American English, the change /d/ = ([, m) is also frequent: ('lD 'lm) /'ld-
2. english
67
/ ladder ( ('lm) /'l/ latter which may have a slight dierence in duration,
() shorter than a semi-chrone, (;)), ('<aD '<am) /'<aEd/ rider ( ('<am) /'<aE/
writer] ('bDi 'b[i) /'bEdi/ beddy ( ('b[i) /'bEi/ Bet] However, such a pronunciation may not be considered neutral, though very widespread, but only mediatic.
Another (and neutral) way to keep a dierence, partially recovers the voicelessness of //: ('lb '<ab 'bAi), or better still totally: ('lr '<ar 'bi) (as mentioned and illustrated above), with no lengthening of the vocoid before /d/ ([, m).
In informal Briti pronunciation, chiey in monosyllables of low semantic value, /t/ may be realized as ([): (Th&gT'ph>zT, -&g[-) /tgE'p<Ezt/ to get a
present (&gT'mc, &g[-) /g'mc/ got a mab (&khwaT&bT'h>fk,
&khwa[&b[-) /kwaEbIv't<fIk/ quite a bit of trac (T'kh;Tmi, [-) /It'k:tmi/ it occurred to me (&s;T*'G;, &s;[*-) /sO:v'Gb/ sort of job ('wT 'phTi, 'w[ -) /w'pIi/ what a pi ('nT nli's, 'n[ n-) /nOUnli'Is/ not
onI ìs (bT';s 'T, b['-) /b'O:sOU 't/ but also àt (bTa'D;u, b[a-)
/baE'dUu/ but I do.
Stop-strictives (or aricates)
2.2.7.1. English has just one (diphonic) pair of stopstrictives, (c, G) /c, G/. For
segments, or phones, articulatory terms are preferred over auditory ones (and, of
course, acoustic ones), because they are much more adequate and clear, generally
self-explanatory.
For this reason, we are happy to avoid aricate, in favor of a more descriptive
and tangible (even checkable) term, such as preopped conrictive, which we will
presently reduce to op-rictive, after explaining that they are unitary phones, or
sounds, in that they have a total duration comparable to that of any other single phone, like (p, t, T, k) or (f, s, S, x), not like the sum of two of them (as in (ts,
kx)).
In addition, they must be homorganic ( produced at the same place of articulation). So, the rst half of a stop-strictive consonantal phone is a short stop, while
its second half is a short fricative one (or, better, a constrictive one). e place
of articulation is determined by the second component, to which the rst one is
just a mere closure, correctly at the same place (even if no actual stop phone exists
at that place, in any real language).
e simpler and more convenient way to symbolize stop-strictive phones is by
means of two monographed symbols. Of course, the second one is the more specic, so the rst can be a looser one, because its only function is to show a closure,
which may be generically labial, pre-lingual or post-lingual. For this reason, the
stop phases of the various possible stop-strictive phones, are suciently shown by
using simply (p, b t, d k, g).
As we said, the only (diphonic) pair of stop-strictives of the English language is
(c, G) /c, G/. In stressed syllables (or after pauses, even in the rare cases of unstressed syllable, as in aucerian), /c/ is aspirated, as /p, t, k/ are (although most
68
native phoneticians do not say that, in the least): ('chmni) /'cImni/ >imn\.
Usually, /c, G/ have (a slight) labial protrusion, and, most often, they are articulated with the tongue tip in a high position (but we need not really use special
symbols, such as ((c, g))).
Besides, /G/ (as any other voiced phoneme in diphonic pairs) is partially devoiced before a pause or before a voiceless consonant: ('G;)a ('G;)b /'GG/
judge.
ile English stops are very often inaudibly released, English /c, G/ always
show an audible plosion, even when they occur before themselves (notice that we
prefer to mark this plosion only here, by means of (0$)): ('wAc$ 'khfi 'wO;c$)a
('wc$ 'khfi)b /'w;c 'kEfli/ wab carefulI ('wc$ 'chI;i 'Wc$ 'hw-) /'wIc
'cIiz/ whi> >eese ('lA:<$ 'ch;c)a ('lA:$ 'ch;c)b /'lA:G 'c:c/ a large >ur>.
e only possible reduction may be in changing the rst stop-strictive with the
corresponding constrictive (fricative) one, (S, Z): ('wS 'chI;i 'WS 'hw-), ('lA:<
'ch;c)a ('lA: 'ch;c)b is can also occur with /nG, nc/ before /c, G/ (seen that
/nG, nc/ have a less frequent variant /nZ, nS/): ('O;<~G 'GUus, -~Z, 'A<-)a ('>~G
'Gus, -~Z)b /';<nG 'GUus/ orae juice
For British English, we must add that /c/, as well as /p, t, k/, can show the two
kind of glottalization we saw ( 2.2.6.1-2), with the same frequence and degree of
advisability. But, for /c/, it occurs even before vowels: ('fc, 'f, 'fc) /'fEc/ feb,
('b~c, -~, -~c) /'bEnc/ ben>, ('s;cT, -T, -cT) /'s:ct/ sear>ed, ('s;cmi, -mi,
-cmi) /'s:cmi/ sear> me, ('fcT, -T, -cT) /'fEcIt/ feb it, ('Thic, -, -c)
/'tIic/ tea>er.
nally, chiey in British English, /c/ may become (S), before a pause or a C\
('fS, 'b~S, 's;ST, 's;Smi) (examples we have already seen) and ('khc 'khS)
/'kc/ cab (5wc 'bk3 3 5wS 'bk3 3) /'wIc 'bUk/ whi> book?
Constrictives (or fricatives)
2.2.8.1. Also for this manner of articulation (as for the stop-strictive one, rather
than aricate), we prefer to use an articulatory term, because of its greater clarity.
ere are four (diphonic) pairs, /f, v , s, z S, Z/ (f, v , s, z S, Z). For /f,
v/ readers are referred to what has been said about /pf, bv/ ( 2.2.2.2). We now add
some examples that show the frequent reduction or dropping of /v/: (a'kh[;]m
b'lI;iv&T, -iV&-, -i&-)a (-A;m b-)b /aE'k;n b'lIivt/ I can't believe àt ('gvmi
'fa;, 'gVmi, 'gmi) /'gIvmi 'faEv/ give me ve! (a'S;<TDT, aV-, a-)a (-;T-)b
/aEv'SO:tdt/ I've ortened it (in spite of an information loss in comparison with
I ortened it), ('la;vz 'Th<b, -Vz, -z, -)a (-, 'Th>-)b /E'laEvz 'tE<b/ èir lives are terrible (in spite of the ambiguity with èir lies are terrible),
('f:2 'fa;2 'sks2 'sv2 'IT2 'fa;V2 'sk2 'sb2 'fa;2 'sk2 's2)a ('f:2)b /'fO: 'faEv 'sIks
'sEv 'EIt/ four, ve, six, seven, eight And, let us note: ('G 'phT, 'fa;6 'bts)
/'GEf 'pIt, 'faEv 'bIts/ Je Pit ve bits.
/, / are slit dental (whereas in American English a prodental or interdental
2. english
69
articulation is possible, perhaps more often indeed, which may be transcribed with
((w, W)); however, since their auditory impression is not very dierent, the ocial symbols, (, ), may be sucient. It is important for foreigners to acquire this slit articulation well: ('k) /'Ik/ ìnk, ('s) /'Is/ ìs, ('w;) /'wI/ wi`. en a sentence begins with a more or less pleonastic I ìnk, generally, a semi-constrictive //
is used: (&k)a (-)b; the same can occur for the //, (), of grammemes: è ìs
Besides, in quick informal speech, // in forms such as è àt ìs bo` is
normally assimilated: ('wts z'Tha;m, 'wAts)a ('wts)b /'wts 'taEm/ what's è
time?, ('wts 'zT, 'wAts)a ('wts)b /'wts 't/ what's àt?, ('bs 'sa;d)a ('bs)b
/'bOU 'saEdz/ bo` sides (hiz'wIm, hiz-, hizz-)a (-T)b /hiz'wIt/ he's è
waiter ('m;<n, -, -)a (-;n-)b /In'mO:nI/ in è morni ('O: 'Tha;m, 'O:-, 'O:l-)a (':)b /'O: 'taEm/ all è time
2.2.8.2. /s, z/ are grooved dental constrictives, usually pronounced with the tip
of the tongue raised, denti-alveolar, so that they could be transcribed with (s,
z), especially for comparative purposes, in order to emphasize the dierence between (s, z), pronounced with the tip of the tongue lowered. On the other hand,
native speakers themselves may indierently have one articulation or another, even
vacillating, so the plain symbols can safely be used: ('s;Im) /'sEIm/ same, ('<z)a
('>z)b /'<OUzz/ roses, ('sz)a ('sz)b /'sIzz/ sciors
For word-initial sm- sn- sl- (as well as for the non-autochthonous sr-), /s/ (s) is
normal (contrary to some other languages): ('smO:)a ('sm:)b /'smO:/ small
('snIk) /'snEIk/ snake, ('slIip) /'slIip/ sleep ((&s<Ii'lA;k, s<-, &S<Ii-)a (-k)b /s<Ii'lA;k, s<I-, S<Ii-/ Sri Lanka).
For dis- followed by a voiced stop, there are several possibilities, both phonetic
and phonemic: (&Dsb'lI;i, -s-) /dIsb'lIiv/ disbelieve (Ds'D;In, -s'-, -'D-, -z'D-, -'sT-)
/dIs'dEIn, -z'd-, -s't-/ disdain (Ds'gsT, -s'-, -'g-, -z'g-, -'sk-)a (-sT)b /dIs'gst, -z'g-, -'sk-/
disgu (Ds'ga;, -s'-, -'g-, -z'g-, -'sk-) /dIs'gaEz, -z'g-, -'sk-/ disguise
Let us now observe (but only here) that /s, z/ preceded by one or more consonants, are usually articulated as (dental/denti-alveolar) approximants, (, d), rather
than as constrictives: (('SAp))a (('Sp))b /'Sps/ ops (('k)) /'ks/ ànks
(('bE:D))a (('b:D))b /'bEz/ bells (('hn:dD)) /'hndz/ hands. But it is sucient to transcribe: ('SAps)a ('Sps)b, ('ks, 'bE:, 'h;nd)a ('b:)b.
In American English, the sequences /s, z/ are realized as (<, <) (with alveolar
constrictives): ('f;T) /'f:st/ r ('ph;) /'p:s/ purse (h'k;T) /h'sk:t/ her skirt
(f'mO: 'phIip) /f'smO: 'pIip/ for small people ('D:<) /'dO:z/ doors ('f:) /'f:z/
furs In British English, /s<, z</ generally become (>, >): ('nju&>iI) /'nUuz<Ii/
n[sreel.
More often, /S Z/ are produced with the tip of the tongue raised, as apico-postalveo-palatal rounded contoids, so that more suitable symbols, contrastively, could be ((x, )). But, as several natives pronounce them with the tip of the
tongue lowered, (S, Z) will be sucient. e most important thing to keep in mind
is that they have a certain degree of lip protrusion (indeed, they must be labeled
as protruded]: ('Sp) /'SIp/ ip, ('phlZ)a (-)b /'plEZ/ pleasure
For /s, z/, too, assimilation is rather important. As a matter of fact, /s, z/ = /S, Z/
70
(S, Z) before /S c, G j/: (S'SAp)a (-p)b /Is'Sp/ ìs op, (S'GA:<)a (-A:)b /Is'GA:/ ìs jar, (S'jI;)a (-;)b /Is'jI/ ìs year, (Ii'SU;u)a (-;u)b /Iiz'SUuz/ èse
oes, ('hSi, 'hSSi) /'hzSi/ has e? even /sc/ = (Sc) is possible: ('khwscn, -Sc-)
/'kwEscn/ queion
Generally, with you your there is a complete fusion between elements: (a'mS,
-)a (a'mS -)b /aE'mIsju/ I mi you, (hi'nIiD Z'hEp, -G - -U-)a (Z'hp, -G -
-;-)b /hi'nIidz jK:'hEp/ he needs your help, (Z's;I, - &-, --) /zju'sEI/ as you sZ.
Let us also consider: ('ZT -) /z'jEt/ as yet. (Making use of diaphonemes like /,
K:/ may help in reducing the space of a phonemic transcription, chiey in a dictionary.)
Approximants
2.2.9.0. In order to present the important components of this particular manner of articulation in a simple way, we will proceed by specic categories.
English r /</ (<)a (>)b ( //)
2.2.9.1. e English /r/ phoneme is completely dierent from that of most
languages, which have alveolar trills or taps (r, R). It is thus extremely important
to use a dierent symbol for English r, even at a phonemic level: /</. Furthermore,
American and British English have two quite dierent articulations, although
from an auditory point of view the impression is quite similar. However, there
are some perceptible dierences: suce to say that the American type has a relatively higher intrinsic timbre than the British.
Once and for all, it is of paramount importance to establish the exact articulation of both kinds of approximants. Unfortunately, except in very few cases, even
among native English phoneticians, there exists odd and perhaps too-traditional
ideas about the precise nature and articulation of /</, which are not based on real
analyses of sounds and accurate kinesthesia as well. It is true that the American r
is articulated in a backer position than the British one, but its retraction refers to
the dorsum not to the tip of the tongue.
2.2.9.2. It is proved that the American /</ is a prevelar approximant, with a very
slight and (almost) uninuential raising of the tip of the tongue towards the
postalveolar region. Instead, the British sound is decidedly postalveolar, (>), in the
specic meaning of an area after the alveolar one, approached by the tip of the
tongue (not by the lamina, as in the IPA ocial point of view). It is actually an
apico-postalveolar articulation.
It will be very important to observe the orograms of these two approximants
very carefully ( 1.13.3). Both of them are laterally contracted, just as real lateral
phones, but there is no contact with the roof of the mouth (as, instead, with real
laterals). e absence of such a lateral contraction would simply deprive these ar-
2. english
71
ticulations of their typical timbre, which is so similar (in these two appoximant
phones), even though their actual articulations are relatively very dierent.
In addition, both (<) and (>) show a certain amount of lip rounding (more evident in stressed syllables), which changing both towards a duller timbre contributes in making them less dierent auditorily, while remaining articulatorily rather dierent.
2.2.9.3. Once the exact articulations are clear, it is easy to understand why, for
/t<, d</, the British pronunciation regularly undergoes assimilation, giving (>, >).
On the other hand, the fact that the auditory impression is so similar for these two
types of phones, may explain why, even in the American pronunciation, (>, >)
can be used, besides the more usual ones, (T<, D<).
Certainly, it is very strange that the majority of phoneticians (even native ones)
keep on using the symbol (>) to hint at the American type, which is far from being postaveolar. By the way, the term poalveolar corresponds to the ocial one
retroex, which picturesquely tries hard to pass itself o as a real point of articulation, while, in fact, it is at most just a very peculiar articulatory modication.
But, as is well known, good kinesthetic, auditory (and even acoustic) skills are not
the same for all people
Up to now, we have seen several examples of /</ (<)a (>)b, and several others will
follow. Let us remember only that our diaphonemic transcription rigorously distinguishes between /</, which is always pronounced in the two accents, and //,
which is pronounced, as such, only in American English. As a matter of fact, in
British English, // corresponds to zero, as r is pronounced only before vowels:
('<I;)a ('>;)b /'<I/ rear ('<;)a ('>;)b /'<E/ rare ('<:<)a ('>:)b /'<O:/ roar.
In American pronunciation, //, preceded by vowels or consonants, is realized
as (). It is the same even for /</ (and, by and large, for /</); /:, :<, ;</ are realized as (:): ('m;D)a ('m;D)b /'m:d/ murder ('m;D<, -D)a ('m;D>, ->)b
/'m:d<, -d</ murderer Also (<, :<)a, for (, :)a are acceptable, even if less
frequent.
2.2.9.4. However, // is pronounced, even in British English, when it occurs
nal in a rhythm group before a following rhythm-group initial vowel (and there
is no intervening pause, not even a short one). In this way, the two words are
bound together, and // becomes /</ (>): ('khA:< '<a;v)a ('khA:> '>a;v)b
/'kA:< '<aEvd/ è car arrived, ('ThIk 'kh; vj'Ef, -jU-)a ('ThIk 'kh;> vj;'sf, -j-)b /'tEIk 'kE< vjK:'sEf/ take care of yourself
On the other hand, on the analogy of word-nal /, I, E, U, O:, A:/, very
frequently, also nal /, I, U, O:, A:/ are realized as the previous ones, even if no
etymological r is present in their spelling: (ia'D[]vT ->vT) /iaE'dIvIt/ è
idea of it ('laz 'ls -z> 'ls) /'laEz 'ElIs/ Eliza Ellis ('GI;i&bIi 'S: 'A;s[k]T -:>)
/'GIibIi 'SO: 'A:skt/ G. B. ShY asked.
is use is very widespread, chiey for //, although good speakers try to avoid
it, but many others use it airily, even teaching it to foreigners (who should avoid
it, unless they are very uent and have a very good command of British English).
72
In a broad New Zealand accent, /</ is (), instead of (>), usual also in neutral
New Zealand English. As in American English (except in typical Southern, Eastern, and Black accents) and Canadian English, also Irish English (in the whole island) has // = /</. e same goes both for an area in the South Island in New Zealand and for the West Country in the southwest of England (as well as for some
more limited areas in the North of England). A typical Scottish accent, usually, has
/<, / = (R).
As a speech defect, /</ is realized as a labiodental (V). is is so widespread, especially in Great Britain, that someone considers it to be normal (all the more so because it is frequent in the mediatic British accent).
e other approximants
2.2.9.5. e voiced palatal approximant, /j/ (j), has no particular characteristics. It is therefore more interesting to talk about the diaphoneme //, that restricting ourselves to the two neutral accents distinguishes American English from
British English, because, between /n, t, d/ and /Uu, U/, in non-weak syllables (
those with primary or secondary stress), in American pronunciation // becomes
zero.
is means that, in American English, /nU[u], tU[u], dU[u]/ correspond to
/tU[u], dU[u], nU[u]/, while, in British English, they are /njU[u], tjU[u], djU[u]/:
('nU;u)a ('nj;u)b n[, ('ThU;u)a ('Thj;u)b /'tUub/ tube, ('DUuk)a ('Djuk)b /'dUuk/
duke.
It is true that, in American pronunciation, one can even nd ('nj;u, 'Thj;u,
'Djuk) and even a compromise realization, ('n;u, 'Th;u, 'Duk). However, the
more usual pronunciation has (Uu), even if ('DU;u) may happen to correspond to
both /'dUu/ do and /'dUu/ due .
In you and your, due to the assimilation of /j/ to preceding /t, d/, there are noteworthy expressions such as: ('D~c, -)a ('D~c -)b /'dOUncu, -/ don't you?,
('wDc, -)a (- -)b /'wUdcu, -/ wouldn't you?, ('khG, -)a (- -)b /'kUGu,
-/ could you?, ('DG 'b< 'g;, -GU)a ('DG; 'b> 'g; -G)b /'dIGU 'b<
'gOU/ did your broèr go? For the assimilation to preceding /s, z/, see above ( 2.2.4).
Although rarely, English phonotactics presents sequences such as: ('jp) /'jIp/ yip
('jIisT) /'jIist/ yea. Before /Ii/, /j/ may be realized as a semi-constrictive contoid, (,),
which is stronger than (j): ((',IisT)). On the other hand, in unstressed syllables, /j/
(j) may lessen, up to a semi-approximant, (): (('nb-ls)) /'nEbjls/ nebulous.
2.2.9.6. e velar rounded approximant, (w) /w/, has no particular characteristics, apart from such rare sequences as in: ('w;) /'wUd/ wood ('wUunD)a ('wunD)b /'wUundd/ wounded. Before /Uu/, /w/ may be realized as a semi-constrictive
contoid, (m): (('mUu-))a (('mu-)b; while, in unstressed syllables, it may lessen, up to
a semi-approximant, (j): ((j'aT)) /wI'aOt/ wiòut. It is important to remember
that /w/ has a strong bilabial component, which causes changes in the realization
of preceding /n, t, d/: ('khp&wO;k, --, -T-)a (-w;k)b /'ktwO:k/ caalk ('kh;<m-
2. english
73
wO;, -n-)a ('kh;mw;, -n-)b /'kO:nwO:/ Cornwall.

By now, the sequence /w/ (w, W, hw) has only a secondary role, which (at least
in theory) allows us to distinguish words beginning with wh- from those with a
simple w-. It is to be said that such a distinction is no longer neutral; it can still be
found either as a voluntary eort, or chiey in some non-urban American pronunciations or in some northern British ones: ('wc) /'wIc/ wib ('wc 'Wc
'hwc) /'wIc/ whi> ('wT) /'wEt/ wet ('wT WT 'hwT) /'wEt/ whet
is distinction is more stable in Scotland, Ireland (including Ulster) and in
New Zealand. For /hj/, we can have (j, ): ('hj;u, 'j-, '-) /'hjUuG/ huge; whereas (j) is acceptable only in American pronunciation.
2.2.9.7. e last English approximant (although too many phoneticians even
native speakers insist in classifying it as a constrictive) is (h) /h/. It occurs before
vowels and between voiced sounds it can become voiced: ('hT) /'ht/ hat
('hU;u)a ('h;u)b /'hUu/ who (ph'hps, ph'Hps)a (ph-)b /p'hps/ perhaps (b'ha;n, -H-)a (b-)b /b'haEnd/ behind.
In British (and Welsh and Australian) uneducated pronunciation, /h/ often becomes zero, /`/: /'`t, 'Ùu/, hat who, . However, it must be clearly stated
that in non-emphatic grammemes (not occurring after a pause) the change /h/ =
/`/ is quite normal: ('Thlm) /'tElIm/ tell him, ('ThIk)a ('ThIk)b /'tEIk/ take her
(very dierent from ('ThE: "hm:)a ('Th: "hm:)b /'tE "hIm/ tell !, ('ThIk "h:)a
('ThIk "h:)b /'tEIk "h:/ take !]
On the other hand, in comparison with Romance languages, we must emphasize the importance of aspiration for /p, t, k, c/, at the very beginning of stressed
syllables (and even of unstressed syllables after pauses), including second elements
of compounds: (Th'ThIk) /t'tEIk/ to take ('ThE&Th)a ('Th-)b /'tEtEI/ telltale
(but ('sTIk) /'stEIk/ ake).
Laterals
2.2.10.1. e only lateral English phoneme, from a strict intraphonemic point
of view, is \l\ (l, ), with two very important taxophones, or contextual allophones
(together with others, by coarticulation, as we will see). In actual fact, given their
considerable importance, from a descriptive and teaching point of view, our diaphonemic transcription makes use of /l, / (instead of a more abstract /&/ which
is decidedly less interphonemic for //). ere is one caveat: although we have
decided to include // among our diaphonemic symbols, this does not imply that
we consider it a real phoneme, as no opposition exists in English between // and
/l/. It is simply a very useful guide for foreigners to make a safe and straightforward
choice between them. On the other hand, in an almost neutral pronunciation, certain speakers may present cases such as: ('khADl)a ('kh-)b /'kdlI/ codli and
('khAD])a ('kh-)b, besides (-D, -Dl, -D) /'kdlI/ coddli, including
(-Dl), which unies dierent pronunciations into one.
Traditional transcriptions excessively hide many characteristics, including the
74
dierence between (, ), and the actual fact that, before (a heterosyllabic) /j/, not
only does () not occur (as, instead, it does before any other consonants) but it is
prepalatal, (), ('m-jn). In other contexts, it is alveolar, (l), as in: ('lUusaT)a ('lu-,
'lju-)b /'lUusaEt, 'l-/ leucite Even between a stressed short vowel and another vowel, which makes one syllable with /l/, in the neutral (American or British) pronunciation, (l) (not ()) is used: ('fAl-i)a ('fl-i)b /'fli/ folI ('fAl-)a ('fl-)b /'flOU/
follow ('Thl-m) /'tElIm/ tell him ('fIilT) /'fIilIt/ feel it, ('fO:l 'aT)a ('f:l)b /'fO:l 'aOt/
fall out ('bl 'vn) /'bIl 'Evnz/ Bill Evans.
2.2.10.2. eir distribution is in any case rather simple, once it has been clearly explained. As we have said, in diaphonemic transcriptions, it is convenient to
use the two primary symbols, /l, /; although (to be precise) it would be useful to
use at least four symbols, (l, , ], ) (besides () in British {and British-like} pronunciation, before /<, t<, d</), in addition to three intense ones, (, , ). Furthermore,
at least in specic descriptions, the taxophones with dental contact, ((l, )), should
be indicated, and also those with lip-rounding after V or C which have a labial
component): ((t, , 3, , , )). Especially the coarticulation of labialization is automatic; therefore, it is necessary to mark it only where it is being explained, in order to draw attention to the phenomenon.
Before V we regularly have /l/ (l): ('lI;i) /'lIiv/ leave, ('l[;]sT)a ('lA;sT)b /'l;st/ la,
('lk) /'lUk/ look, ('l;n) /'lOEn/ loin. To be rigorous, lip-rounding also occurs before rounded V, by coarticulation: (('tk, 't;n)); however, a special symbol like
((t)) is not needed, since it is absolutely inevitable to prepare the lips for the rounded vocoids that follow, within the syllable.
In fact, an articulation of /l/ without lip-rounding, ((l)), would somehow be perceived as something strange, exactly as for /k, g/ followed either by front V or
by /j/, or else by rounded V or by /w/: the articulations ((, )) and ((k, g)), respectively, are natural and automatic: (('hT)) ('khT) /'kIt/ kit, (('T)) ('gT) /'gEt/ get,
(('j;u)) ('khj;u) /'kjUub/ cube, (('kW:<))a ('kh:<) (-:)b /'kO:/ core, (('gUus))a
(('gus))b ('g-) /'gUus/ goose, (('kWaT)) ('kwaT) /'kwaEt/ quite.
erefore, in particular for //, an objective pronunciation, obtained by juxtaposing /p/ and //, for instance, would produce an eect that may perplex native
speakers. Strictly speaking, in fact, (p) would have something less in comparison
with the genuine ((p)), as in (('phIip)) ('phIip) /'pIip/ people.
2.2.10.3. It is important to notice that, for postvocalic and tautosyllabic l, after
rounded V in the various languages, labial coarticulation regularly occurs; therefore, it need not be marked, as instead we are doing here. Before heterosyllabic /j/,
we nd () (and (())): ('mjn) /'mIljn/ million, (('bjn)) ('b-) /'bUljn/ bullion,
(('O: j'nI;i))a ((':))b (-) /'O: j'nIid/ all you need before /, ts, dz/, we have ((,
)), (): (('fi)) ('fi) /'fIi/ l`y (('O: 'Tha;m))a ((':))b ('O: 'Tha;m)a (':)b
/'O: 'taEm/ all è time, (('bEts))a (('bts))b ('bEts)a ('bts)b /'bEts/ bel_, (('fO;ts))a
(('f;ts))b (-ts) /'fO:ts/ faul_. th /, / + /l/, we have ((l)) (dental, but not velarized): (('m 'lIT))a (('m-))b (-n 'lIT) /'mn 'lEIt/ a mon` late, ((w'l;))a
(-;)b (w'l-) /wI'lv/ wi` love. In British English, before /<, t<, d</, it is realized
2. english
75
as postalveolar, ((, )): ('chD<n)a ('ch>n)b ('ch-)b /'cId<n/ >ildren, ((O;'<Di))a ((;'>Di))b (O;-)a (;-)b /O:'<Edi/ already.
Before pauses, or another C, we have ((, )), (): ('b:) /'bI/ bill, ('bT) /'bIt/
built, (('h,:))a ('h:)b /'h/ hull, (('fO:))a (('f:))b (-) /'fO:/ fall, (('fO;k))a (('f;k))b
(-k) /'fO:k/ Falk. In the case of words with //, we generally nd nasalization:
(('kh:n)) ('kh:n) /'kIn/ kiln (('E:m))a ((':m))b (-:m) /'Em/ elm.
en // and a word-initial V meet with no pause between, we have ((l, t)): ('fIilT) /'fIilIt/ feel it, (('O:t 'v)a (':l 'v)b ('O:)a (':)b /'O: 'OUv/ all over.
As far as intense l is concerned, ////, we nd lip-rounding after C with a labial
component, /m, p, b f, v c, G S, Z <, w/. In addition, before V within words, we
have ((, K)): ('khAD)a ('khD-)b /'kdI/ coddli ( the beginning of 2.2.10.1),
(('Th<mbK))a ('h>-)b (-b) /'t<mbI/ trumbli. However, before V belonging
to a following word, a semi-velarized alveolar articulation is found, ((, )) (even before front V]\ (('l[ '[li))a (('lT 'Tli))b ('l[)a ('lT)b /'lI 'Ili/ Little ItaI, (('phIip
'Ii[))a ((-T))b ('phIip) /'pIip 'IiI/ people eati. Lastly, before a pause or a C we
have ((, )): ('l[)a ('lT)b /'lI/ little, ('k)a ('-)b /'k/ uncle, (('ThIb)) (-b) /'tEIb/ table, (('O;f))a (';-)b (-f) /'O:f/ Yful. After /, /, the contact is dental: (('b%))
('bn) /'bEn/ benàl.
Often, many transcriptions present sequences of /l/, because they refer to
slow or careful speech: ('nAvlsT)a ('n-)b /'nvlIst/ noveli (b'lI;i) /b'lIiv/ believe (ph'lIis) /p'lIis/ police (kh'lZn)a (-Z)b /k'lIZn/ collision currently,
though, we nd: (('nAvKsT))a ('n-)b, ((bK'I;i)), ((phK'Iis)), ((kh'Zn))a (-Z)b. It is interesting to compare the following forms, which generally maintain a slight dierence of syllabic structure, in comparison with the cases previously seen: ('blIic)
/'blIic/ blea> ('phlI;i) /'plIiz/ please.
2.2.10.4. We will now report, though not recommend, the frequent insertion
of a homorganic stop before /, s, S/ (not /f/) preceded by //: ('fi -ti) ((-ti))
/'fIi/ l`y ('fO;s -ts, 'fA-)a ('f;s -ts, 'f-)b ((-ts)) /'fO;s/ false ('wES -TS, -c)a
('wS -TS, -c)b /'wES/ Wel is occurs with /z/, as well, but more rarely: ('bE:
-:d)a ('b: -:d)b /'bEz/ bells ('fi;I -d, 'fiI-) /'fIiz/ feels. On the other hand,
the reverse simplication may be heard, too, as in: ('fi;Id -, 'fiI-) ((-)) /'fIidz/
elds ('bEts, -s)a ('bts, -s)b ((-s)) /'bEts/ bel_. Especially when there are possible ambiguities (and grammemes >-s -'s), spelling is better respected.
In non-neutral American pronunciation, /l/ can be realized as (]) (sometimes
even ()) before V or /j/. is can also occur in New Zealand, in northern Wales,
but most of all in Scotland. On the other hand, in Ireland, in South Africa, and
in southern Wales, usually, (l) may be heard, even before C or pauses.
Other less systematic dierences
2.2.11.1. nally, there are some more or less isolated words, which are pronounced dierently in the two neutral accents. Others often show both pronunciations. In American pronunciation, word-initial syllables are commonly full,
76
not pronounced with //: /vEI'kEISn/a /v-/b vacation /nOU'vEmb/a /n-/b november Even the article a often has a full timbre (which could sound too formal in
British pronunciation): /z'gUd '<Iiz, -zEI-/a /-z-/b ère's a good reason
Furthermore, bisyllabic verbs in -ate more frequently in American English, bear
initial stress: /'dIktEIt, -'tEIt/a /-'tEIt/b dictate, /'vaEb<EIt/a /-'b<EIt/b vibrate. e sux
-ile is /- -aE/a /-aE/b: /'hstA/ ('hAsT, -a)a ('hsTa)b hoile (and mobile as
well), /'f:A/ ('f;[ -a)a ('f;Ta)b fertile /'mIsA/ ('ms)a ('msa)b miile
(and reptile erile] But we have: /'ds -aE/a /'dOUsaE, 'dsaE/b docile and /'GUuvn, -naE/a /-naE, -n/b juvenile. We also nd: /'p<-/a /'p<OU-/b pro-: /'p<sEsI,
--/a /'p<OU-, 'p<-/b procei, /'p<g<Es, -s/a /'p<OU-, 'p<-/b progre (but: /'p<OUg<m/ programa, -mmeb).
To end with, let us see: /sIvl'zEISn -laE-/a /-laE-, -l-/b civilization, /'kpl<i/a
/k'pIl<i/b capillary, /'kO:<l<i, 'k-/a /k'<l<i/b corollary, /'lb<tK:<i/a /l'b<t<i/b laboratory, /'nOUbdi, -b-, -b-/a ('n&bDi, -&bADi, -bDi)a /--, --/b ('n&bDi, -bDi)b nobody (and aJ- every- some-] Let us also notice this (seemingly
curious) word: /'aEn/a ('an, 'a;<n)a /'aEn/b ('an)b iron
2.2.11.2. Here is a collection of over 100 interesting cases, which we present exceptionally in alphabetic order, giving current spelling rst. However, we must be
aware that, not infrequently, speakers of one of the two accents (a and b) use pronunciations shown for the other accent. For single words, this is quite obvious:
addre (an) /'d<Es, 'd<Es/a /'d-/b
advertisement /dv'taEzmnt, '-,
d'v:smnt, -z-/a /d'v:-/b
agave /'gA:vi/a /'gEIvi, -A:-, 'gEIv/b
albino /'baEnOU/a /-'bIi-/b
American ('m<kn, 'm;-)a ('m>kn)b
anti- /&naE-, -i-/a /-i-/b
apparatus /p'<s/a /-t-, -EIt-/b
artisan /'A:z/a /A:'zn, 'A:zn/b
ate /'EIt/a /'Et, 'EIt/b
ballet /b'lEI/a /'blEI/b
because /b'kz, -;z/a /-z/b
Berkire /'b:kSI, -S/a /'bA:kS,
-I/b
beta /'bEI/a /'bIi/b
cabaret /kb'<EI/a /'kb<EI/b
cadre /'kd<i, 'kA:-, -EI/a /'kA:d, -d<,
'kEI-/b
centenary /sn'tEn<i, 'sEnn<i/a /-'tIin<i, -'tE-/b
>arade /S'<EId/a /-A:d/b
>ais /'Ssi, 'c-, -ss/a /'Ssi/b
>irrup /'c:<p, 'cI<-/a /'cI<-/b

cigarette /'sIg<Et, -'<Et/a /-'<Et/b
circumstances /'s:kmstnsz, -t-/a
/-t-, -t-, -tA:-/b
clerk /'kl:k/a /'klA:k/b
comrade /'km<d, -d/a /-EId, -d,
'km-/b
consomm /kns'mEI, 'knsmEI/a
/kn'smEI, 'kns-/b
controver /'knt<v:si/a /'kn-,
kn't<vsi/b
cordial /'kO:G/a /-dj-/b
coup /kUu'pEI/a /'kUupEI/b
creek /'k<Iik, 'k<Ik/a /'k<Iik/b
cu$oo /'kUukUu, 'kU-/a /'kUkUu/b
data /'dEI, --, -A:-/a /-EI-, -A:-/b
dmod /dEImOU'dEI/a /dEI'mOUdEI/b
derby /'d:bi/a /'dA:bi/b
deterrent /d't:<nt, -E<-/a /-E<-/b
doctrinal /'dkt<n/a /dk't<aE-, 'dk-/b
drama /'d<A:m, 'd<-/a /'d<A:-/b
dynas /'daEnsti/a /'dI-, 'daE-/b
eièr /'Ii 'aE-/a /'aE-, 'Ii-/b
2. english
erase /I'<EIs/a /-z/b

gure /'fIgj/a /'fIg/b
nancier /faEnn'sI, -nn-, f'nnsi, faE-/a /faE'nn-, f-/b (just one
stress pattern)
fracas /'f<ks, 'f<EI-/a /'f<kA:/b
from /'f<m, 'f<m/a /'f<m/b
garage /g'<A:Z, -G/a /'g<A:Z, -G, -IG/b
hara /h'<s, 'h<s/a /'h-, h-/b
herb /'h:b ':b/a /'h-/b
hero /'hI<OU, 'hIi<-/a /'hI<-/b
humble /'[h]mb/a /'h-/b
humo[u]r /'hjUum 'j-/a /'hj-/b
idea /aE'dIi/a /aE'dI/b
idyll /'aEd/a /'aE-, 'I-/b
inquiry /'Ikw<i, -'kwaE-/a /-'kwaE-/b
isolate /'aEslEIt 'I-/a /'aE-/b
khaki /'kA:ki, 'k-/a /'kA:-/b
lao /'lsOU, -Uu, l'sUu/a /l'sUu, l-,
'lsOU/b
leisure /'lIiZ, 'lE-/a /'lE-/b
lever /'lEv, 'lIi-/a /'lIi-/b
lieutenent /lUu'tEnnt/a /lEf't-, lf't-/b
margarine /'mA:G<n, -<Iin/a
/mA:G'<Iin, 'mA:G<In, -g-/b
maage /m'sA:Z, -G/a /'msA:Z, -G/b
melan>oI /'mElkli/a /-li, -li/b
migraine /'maEg<EIn/a /'maE-, 'mI-,
'mIi-/b
miscellaJ /mIs'lEIni/a /m'sEl-, mIs'lEI-/b
multi- /&mlaE-, -i-/a /-i-/b
musta>e [mou-] /'mstS, m'st-/a
/m'stA:S, mU-/b
neièr /'nIi 'naE-/a /'naE-, 'nIi-/b
nougat /'nUugt/a /'nUugA:, 'ngt/b
of /'v, 'v/a /'v/b
omega /OU'mEg, -EI-, -Ii-/a /'OUmIg/b
omicron /'mIk<n, 'OU-/a /OU'maEk<n, -n, 'mIk<n/b
on /n, O:n/a /n/b
patriot /'pEIt<it/a /'pEI-, 'p-/b
plateau /pl'tOU/a /'pltOU/b
premier /p<'mI, 'p<Ii-, 'p<E-/a /'p<E-,
'p<Ii-/b
77
prince /'p<InsEs, -s/a /p<In'sEs, 'p<InsEs/b

privacy /'p<aEvsi/a /'p<I-, 'p<aE-/b
puma /'pUum, 'pjUu-/a /'pjUu-/b
quinine /'kwaEnaEn, -nIn/a /kwI'nIin,
'kwInIin/b
raèr /'<; '<-/a ('<[;], '<-)a
/'<;-/b ('<A;)b
ration /'<Sn, -EI-/a /'<-/b
record (a) /'<Ekd/a /'<EkO;d/b
reveille /'<Evli/a /<'v-, -'vE-/b
(edule /'skEG/a /'SEGUu, 'sk-/b
semi- /&sEmaE-, -i-/a /-i-/b
one /'SOUn/a /'Sn/b
simultaneous /saEm'tEInis/a /'sIm-/b
sojourn /'sOUGn/a /'sG[:]n, 's-/b
solder /'sd, 'sO:-/a /'sOU-, 's-/b
spina> /'spInIc/a /-G, -c/b
squirrel /'skw:< -I<-/a /-I<-/b
st[arde /'stUuds/a /'st-, -'dEs/b
stirrup /'st:A;Tm, -I-)b
subaltern /s'bO;n/a /'sb-/b
suggest /s'GEst, sg'G-/a /s'G-/b
rup /'s:<p, 'sI<-/a /'sI<-/b
tomato /t'mEIOU/a /-A:-/b
tourniquet /'tUnkt, -:-/a /'tUnkEI, -O:-/b
tra>ea /'t<EIki/a /'t<EI-, t<'kIi/b
trait /'t<EIt/a /'t<EI, -t/b
upon /'pn, -n, -O:n/a /-n/b
vase /'vEIs, -z/a /'vA:z/b
water /'wO:, 'w-/a /'wO:-/b
what /'wt, 'wt 'hw-/a /'wt 'hw-/b
wigwam /'wIgwm, -O:m)a (-m)b
wra` /'</a /'<, '<O:/b
xerox /'zI<ks, 'zIi<-/a /'zI<-, 'zE<-/b
yogurt, -ourt -gh- /'jOUgt/a /'j-, 'jOU-,
-Ut/b
Z /'zIi/a /'zEd/b
zebra /'zIib</a /'zE-, 'zIi-/b
zeni` /'zIin 'zE-/a /'zE-/b
zero /'zI<OU, 'zIi<-/a /'zI<-/b.
78
ere is an additional short list of words that in American English may often
have also /O:/ (O:)a, besides the more usual // (A)a ()b: >ocolate doll dolI dol#in god golf gone mo$ on resolve revolver rolf solvate solve omp ~amp
Structures
2.3.0. In this section, we will deal with various macro-segmental aspects, including intonation.
Unstressed syllables and reduced forms
2.3.1.1. As can be seen from many of the previous transcriptions, English unstressed syllables almost always have /, , I/: (m'nA[ns)a (-'nT-)b /m'nns/
monotonous, (ph<'vn[[])a (ph>'vnT)b /p<'vnd/ prevented, ('Dm)a (-T)b /'Edz/ editors, ('mnsT)a (-sT)b /'mInst/ minier On the contrary, in
foreign accents of English, unstressed syllables too often show full vowels.
Indeed, many vowels (and consonants) may disappear entirely in comparison
with spelling, as in: ('khMf[b, -fT-, -f- -mT-)a (-MfT-, -fT-, -f- -mT-)b /'knfb, -mf-/ comfortable, ('mA;<g<T, -T)a ('mA;g>T, -T)b /'mA:g<t/ Margaret,
(&jun'v;[i, -Ti)a (-'v;sTi, -sTi)b /jUun'v:si/ universi
However, not all unstressed syllables have vowel reduction or fall: ('khAmnT)a
('kh-)b /'kmEnt/ comment, ('sfO;T, -A;T)a /'sfO:t, -A:t/ (-T, -;T)b /'sft,
-O:t/ as#alt, ('khUupAn 'khj-)a ('khupn)b /'kUupn/ coupon
Only regular consultation of a pronunciation dictionary (or, better, dictionaries) can give the exact structure of words and sentences, in English as in any other language.
In English sentences, respect of the reduction, or weakening, of many grammemes (or functional words) is vital: articles, prepositions, conjunctions, auxiliary and modal verbs, some pronouns and some other forms. ere are not many
reduced forms (using a clearer term than the traditional one weak forms) about
eighty but they are the most frequent ones. ey are listed below (with examples), in alphabetical order, for easy consultation.
2.3.1.2. e examples given illustrate various elements simultaneously. It is
worthwhile to observe them very carefully and to consider all the variants given
(but only in phonetic transcription, for the sake of space, in a very economical
way, while full transcriptions would have been more monotonous, and would not
be able to show the same things with precision):
a\ ('lm)a (-T)b a letter ('m;n) a man ('ga;) a guy (n'wa;) in a while
('junT)a (-T)b a unit ('n;Im) a name
am\ (am'wE:)a (-:)b I'm well (aM'fa;n) I'm ne (am'gl;, aM-, a-) I'm
glad (ma'<O:, -A;, -)a (-'>;)b am I wro?
an\ (n'p) an apple ('gA[ n'as&kh<Iim, -T'-)a ('g-, ->-)b got an ice cream? (n-
2. english
79
'A:<m)a (n'A:m)b an arm (n';Im) an aim

and\ (n'n:, -) and èn ('b<D 'bm)a ('b>D 'bT)b bread and butter
('bA;b 'khIT)a ('b;b)b Bob and Kate ('j;u n'a;, nD'-) you and I
aJ\ (z'ni 'b<;, --)a (->-, 'b>-, --)b is ère aJ bread? ('hv~c 'gA[
ni'm:<, ni-, -T i, -T ni, - i, - ni, -c)a ('gT, -)b haven't you got aJ
more?
are\ ('I;iz 'jU;, 'j:<)a ('j: 'j;)b èse are yours (A<j'wE:, j-, j-
&A;<j-)a (Aj'w:, j-, j- &A;j-)b are you well? but notice how are you? (&ha'A;<j) (-'A;j) (since ('ha; 'j;u) ('-) would be contrastive)
as\ (z'mm 'fkT, -, *'f-, 'f-)a (-'mT>-, ->-)b as a matter of fact (Si's;,
SSi-) as e said ('GsT z'g;, GsT-)a ('G-)b ju as good ('nAt s'laT z'lks, s-, -at s-, -T'l-)a ('nt s-)b not so light as it looks
at\ (t'ThIb, -) at è table ('lIisT, T'l-) at lea (k'khl;I, 'k-) at
ClZ (p'w;k, 'w-, T'w-)a (-;k)b at work
be\ (bi'g;) be good! ('l[p bi'mI;i, - -T)a ('lT-)b let it be me (a'wm[p] bi'lO:, -, -A;)a (--, -, -;)b I won't be lo
been\ (abbn'w;I, -bn- -bn-, aD- a- ah-)a (-bn-, -bIin-)b I had been
YZ [I'd]
but\ (bt'n:, b-) but èn (bk'g;, b-)a (-;)b but go (bp'bli, b-) but
BilI
by\ (ba'O:&mIin, b-, b-)a (-:, b-, b-)b by all means ('s;D ba'pha;n,
b-, b-)a ('s;-, b-, b-)b sold by è pound
can\ (khna'hvT) can I have it? (wik'plh;I, -km-) we can plZ (jk'g;, -k-, j-)a (-;)b you can go
could\ ('k kD'DU;u, -, -T)a (-;u)b àt could do (wikb'mIkT, -D'm-) we
could make it
did\ (DDi'sT;I 'lO;g, DDi-, -A-)a (-g)b did he Z loer? ('wn DDi'khm:,
DDi-, Di-)a (-m:)b when did he come? ('ha; D'SI;i&n, D-, S-)a (-)b how
did e know? ('ha; DDk'g;, DD-, D-, D-, -'g-, -T'g-)a (-;)b how did it
go? ('ha; DdI'lakT, D-, dd-, d-) how did `\ like it? ('w; DG'g;,
--, -Dj-, DG-, G-)a ('w;, -;)b where did you go?
do\ (G'nT, G-, Dj-)a (--)b do you know it? [d'you] ('w; G'khIipT, G-,
Dj- Dj-)a ('w;)b where do you keep it? [d'you] ('s; D'wI;i, D-)a ('s;)b so
do we ('w; DI'l;, D-)a ('w;)b where do `\ live? ('wT Dj'wAnT, -j-,
Dj-, Dj-, Dij-, 'wA-, -, -O;nT)a ('wnT)b what do you want? (&DA;<'chD<
'g;)a (&DA;'ch> 'g;)b do our >ildren go?
does\ (Dzp'w;k, -'w-, -T'w-)a (-'w;k)b does it work? ('wTDzi 'mI;in, -tdi, -tsi-,
--, 'wA-)a ('w-)b what does he mean? [what's] ('wn DSi'DUuT, 'w~ i-)a
(-'DuT)b when does e do it? [when's] ('ha; Dz'lk, dz-, z, -T'lk) how
does it look? [how's]
for\ ('lkfT, -f<T)a (-f>T, -fT -f>T)b look for it ('sT;I f'wIik, -f<-)a (-f>-,
-f-, -f>-)b Z for a week (tsf'j;u)a (tsf'j;u, -f'j-, -f'j-)b it's for you
from\ (f<m'sku;U)a (f>m-)b from (ool (f<';)a (f>';)b from ère ('w;
'A;<j&f<m, -Am -j-, &w<)a ('w> 'A;j&f>m, -j-, &w>-)b where are you from?
had\ (hDa'sIinT, h-) had I seen it (ID'Da;, ID- -IhD-) `\ had died
80
[`\'d] ([bbn'Dn:, -Db-, -n- -n- Th-)a (T-, -n-, -Iin-, -n:)b it had been
done [it'd] ('m;n g'gO:n, D-, -A;n)a (-;n)b è man had gone (Si'h T'sT;I, -T) /Si'hd t'stEI/ e had to Z (+ (-[ -)a (-T -)b /- -/);
has\ (hzi'gO:n, h-, -A;n)a (-;n)b has he gone? (Siz'Dn:, Siz- Sihz-)a (-n:)b e
has done [e's] (ts'bn:, [z- thz-)a (ts-, -I;in, -n:, Tz-)b it has been [it's]
('G:<G 'khm:, s- - h-)a ('G:G, -m:)b George has come (hi'h T'sT;I, -s) he has to Z
have\ (hvj'sIinT, h-, -j-) have you seen it? (Iv'gO:n, -A;n I- -h-)a
(-;n)b `\ have gone [`\'ve] (wi'khm:, -f'k- wi- wih-)a (-m:)b we have
come [we've] (wi'h T'sT;I, -f) /wi'hv t'stEI/ we have to Z (j'SD,
j-, -, -&h) you ould have (j'SD 'Dn:, -*, -v, j-)a (-n:)b you ould
have done [ould 've] + (-'kh-, -'w-, -'msT-a/--b, -'mI-, -'ma[-a/-T-b) could
would mu mZ might
he\ (hi'wnT) he went ('wni 'sO;T, w-, w-)a ('s;T)b when he sY it ('hzi) has
he?
her\ (h'a;)a (h>'-)b her \es ('Thl)a (-)b tell her ('gv 'hT)a (->-, - h-)b
give her her hat (Th'm, Thh-, Thw-)a (Thh'm, Th-, Th-, Thw-)b
to her moèr
him\ (a'sO;m -m)a (-;m)b I sY him ('l[m 'n: -m -)a ('lT-)b let him in
his\ (h'phn:) his pen (hi'Thk z'bk) he took his book (hZ'ju) his you`
I\ (a'sI;i, -)a (a-, -)b I see (a'ThIk, ,- -- -- -w-)a (a-, - -- --
-w-)b I will take [I'll] ('ha; kDa'sT;I, -- --)a (-a-, -- --)b how could I
Z?
if\ (fa'm;I, f-, -) if I mZ (fj's;I&s, f-, -, --)a (-)b if you sZ so (f'nA
f'j;u, f-, -, -T)a (-n f'-)b if not for you
in\ (n'lnDn)a (--)b in London (m'ph<s)a (->s)b in Paris ('khnD)a
(-)b in Canada (hi's[ n'ch;, -, nn -T - -T n-)a (-T, -;)b he
sat in è >air ('b<Ik[ n'ThU;u -T -)a ('b>-, -T, -;u)b break it in o (amn'h;i, - -mn- am-)a (->i, -)b I am in a hurry [I'm]
is\ (z'j;u, -T'j- -) is it you? (ts'mI;i [z-)a (ts- Tz-)b it is me [it's] ('khm
'hI; -m -)a ('h;)b Kim is here [Kim's] ('s z'fa;n 'D;I, 's '- 's s'-) ìs
is a ne dZ, ('<;z z'nas 'wmn, '<;z '-, '<;z z'-)a ('>;z)b Rose is a
nice woman, (s'DS z'waT, '-, s'-, s-) ìs di is white
it\ (z'Th<U;u, z-, , -T'T-)a (-'h>;u, -'-)b is it true? (ts'O: '<aT, ts'-, s'- [z-)a
(-: '>- Tz-)b it is all right [it's] ('js ['; j'sT;)a (T'; j'sT;)b yes it is
('ThIkT, -T)a (-T)b take it ([bi'g; ':, T- pw- Tw-)a (T-)b it will
be a good ì [it'll] ([bbi'nas, Tw-, -Db- pw- Tw-)a (T-, )b it would
be nice [it'd]
i_\ (ts'Th;) i_ tail (T'h;D ts'ThIi 'b<kn, ts-)a ('b>-)b it had i_ tee`
broken
ju\ (avGs[T]'sIinm, -Gs-) (-G-, -G-)b I have ju seen him [I've] (Siz'GsgO:n 'w;I, -s-, -A;n)a (--, --, -;n)b e has ju gone YZ [e's] (Gs'lak
'sn:, Gs-, -T'l-)a (Gs-, -n:)b ju like his son
maJ\ ('mni 'b;) maJ boys ('ha; mni'm:<, mni-, mni-)a (-:)b how
maJ more?
2. english
81
mZ\ (jmI'[;]sk, j- -mi-)a (-A;sk)b you mZ ask her (wimI'g; -m-)a

(-;)b we mZ go (SimI'sT;I -m-) e mZ Z
me\ (hi'khO;bmi, -Dmi)a (-;-)b he called me ('Smi 'w;I)a ('Smi)b ow
me è wZ
mu\ (ams'g;, -m#-, -)a (-, -;)b I mu go (pmsbi'Dn:, -)a (-n:)b it
mu be done (Sims'ph;I) e mu pZ
my\ (ma'h;, m-, m-)a (m-, m-)b my head ('nAT Tma'nAl, --, --,
-- -- -i-, -)a ('n-, --, --)b not to my knowledge
no\ (n'm:< D'a; n-)a (n'm: n-)b no more do I (n' '<Iiz n-)a
(n' '>- n-)b no oèr reason (tsn'g; -n-)a (-- -n-)b it's no good
nor\ (&nIi'fS n;<'flS, n- &na-)a (&na-, n;-, n- &nIi-)b neièr nor e
(&nIi'hI;i n'a;, nO;<- &na-)a (&na-, n>-, n;>- &nIi-)b neièr he nor I
not, n't\ (['zT)a (T'-)b it isn't (a'wnT, -)a (-, -nT)b I won't (hi'DzT,
-'D-)a (--, -'D-)b he doesn't (wi'Dm 'phl;I)a (--)b we don't plZ (['z
'g;)a (T'-)b it isn't good ('zi, -Ti) isn't he? ('z, -T) isn't it? (hi'jus&T) he used not to
of\ ('ff v'm;I, *'m-) è f` of MZ ('khp 'ThI;i, -'T-, -'T-)a (-p)b a cup
of tea ('f;sT v'O:, *'-)a ('f;-, -:)b r of all
on (the most reduced form only occurs when no ambiguity with in is possible):
(&pO;n'bAks, -An- -n-, -nn-)a (&p-, -ks)b up on è box (&tsO;mma'sa;, -Am-, -m- -m- [z-)a (-m-, -m- Tz-)b it is on my side [it's] ('w[n ';, 'wA-, -O;n, -An)a ('wTn ';, -n)b what on ear`!
once\ (wns'm:<, wn-)a (wns'm:, wn-)b once more (= again] (&wns'm:<)a (&wns'm:)b once more (= one more time]
one(s) (the form without /w/ may be judged as dialectal or regional): ('bg&wn,
-wn -n)a (-w-)b a big one (&ts'g;b&wn, -wn, -D- -Dn -D &[z-)a (-w- &Tz-)b àt is a good one [àt's] (wn'O;w 'hps, w-, -,
-I, -O;w-)a (wn';w 'hps, -, I-, -;w-)b one alwZs hopes (&z'nas&wn, -wn -sn)a (&z-, -w- -sn)b òse are nice ones
or\ (Th'D;I T'mO;<, ;<T- -mA<-)a (T-, ;T-, ->)b todZ or tomorrow ('ThU;u
'<I;i 'pha;nd ;<-)a ('Th;u ;-, '-, ->-)b o or `ree pounds ('Es)a (>'s)b or else
our\ (A;<'sku;U)a (A;-)b our (ool (&IizA;<'phn:)a (-z>A;-)b èse are our pens
per\ ('ffTi p'snT)a (p-)b f per cent ('fa; p'smp p'nm, -m, -nT)a (p's-,
p>'-)b ve per cent per annum
all (in American English it is a stylistic choice): (&Sla'ThIkT, --)a (Sla'ThIkT, Sa-, --, --)b all I take it? (aS'DU;u, a-, ,-)a (-;u, -)b I all
do (Swi'g;, Swi-, Swi-)a (-;)b all we go?
e\ (Si'wnT) e went (Si'hT, -sT) e has to ('hzSi, -TSi) hasn't e?
ould\ (SDi'khm:)a (-m:)b ould he come (jSD'DU;uT, j-)a (-'DuT)b you
ould do it (aSg'g;&na, -Sg-, -D'g-, -)a (-;-, -)b I ould go now (aS'k&s, -St'-, -)a (-, -)b I ould ìnk so ([iS'khm 'aT, -'-, -iS'kh-)a (-Ti-, --)b àt he ould come out
sir\ ('jss, -s-)a (-)b yes, sir ('ns)a ('ns)b no, sir (s'GA;n)a (s'G;n)b Sir
John (s'f<)a (s>'f>)b Sir Alfred (s'chA:<, -A;<)a (s'chA:, s-,
82
#-)b Sir arles

so\ ('nAt s'g;D zp'w;, s'-, -A;, -p-, -T'-)a ('nt s-, -;, s'-)b not so good
as it was (ts'nAt s'fa;n T'd;I, s-, ts-, s- [z-)a (-t s- Tz-)b it is not so
ne todZ [it's] ('nAt s'; s-)a ('nt s'; s-)b not so old ('v s'mni, s-)a (- s-)b /'Ev sOU'mEni/ ever so maJ
some (determ.): (wG'lak s'ThI;i, sm-) would you like some tea? (Dj'wAn
s'm:<, sm-, Dj-, -'wO;-)a (-w-, -:)b do you want some more? (a'h&sm)a
(-m)b I have some
St, Saint\ (sIm'phIim, -mp'ph-)a (s'phIiT, sm-, sIm-)b St. Peter (sI'khl;,
-k'kh-)a (s'khl;, s-, sI-)b St Clair (sIn[[]'nni, -T-)a (sT'nTni,
sn-, sIn-, --)b St AnòJ
su>\ (&sc'ph;s s-)a (&sc'ph;s s-)b su> a person (&sc': s-)a (&s-
s-)b su> a ì
àn\ ('m:< 'T)a ('m:)b more àn àt (Si'fan 'm<i, Sis- Si-)a (-n,
->i)b e is ner àn Mary [e's] (ts'ls nn'~c, n, ts- [z-)a ( Tz-)b
it is le àn an in> [it's] (&ts'm:< na'h; n- &[z-)a (-: -:>
n- &Tz-)b àt is more àn I have [àt's]
àt (conj. rel. pron.): ('nAt [p'mm, 'nA, -'m-, -T'm-)a ('n-, T-, -T)b
not àt it matters ('sI;in c'n&T, TS, Tj-, j-, --) seen àt you
know àt ('D;I {p}wi'mT) è dZ (àt) we met
è\ ('bk) è book ('m;n) è man ('jE:, 'j-, i'j-)a (-:)b è yell (i'n:) è end (i' 'D;I)a (i')b è oèr dZ ('hIiT -) è heat ('wts
'Tha;m, z-, 'wA- -[z-)a ('w- -Tz-)b what is è time? [what's], (&z'kht
';, &zz- &z-, -)a (-;)b is è cat ère?, (n'has, -, nn-) in è
house ('w;I f<m's[i, -'s-)a (f>-, -Ti)b /'wEI f<m'sIi/ YZ from è
ci
èir\ (I'bO;t 'bT -)a (-;t 'bT, - -)b `\ bought èir boat ('has v<';n --, *-, -)a (->';n ->'-)b a house of èir own
èm\ (wi'sO;m, -, -O;m)a (-'s;-, -;m)b we sY èm ('gvm, -, -vm,
-v) give èm
èn\ ('sUup 'f;sT ~'chkn, -)a ('sup 'f;-, -kn)b soup r èn >i$en (a'mak&g\ bn'gn:\ a'ma 'nAT, -n-, -, -T&g-, -T 'n-)a (-\, -T)b I
might go but èn again I might not (n'[;]fT 'Tha;m, n-)a (-'A;ft> -,
-> 'Th-)b èn after a time
ère (exist.): (<'mni, <-, , <-)a (>-, >-, >-, >-)b ère are
maJ (z'lAT, -)a (z'lT, -, -)b ère is a lot ('hv, -)a (-)b
haven't ère?
`\\ ([]';)a (';)b `\ are ère (&wnI'wnT - -) when `\
went ([I]'s;I []- -w-) `\ will sZ [`\'ll]
ìs\ (s'Iivn, s'-, 's-) ìs eveni (s'phn: s-) ìs pen (ns'w;I, nn- n- -s-) in ìs wZ (&O;ns'ThIb, -nn-, &A- -s-)a (&-)b on ìs table
('wts 's, -s, -s, 'zs, 'sIs, 'wA- -[z)a ('w- -Tz)b what is ìs? [what's]
till\ ('wIT Tli'khm:, Ti-)a (-m:)b wait till he comes (Th'ThUuzDI, Th-, -i)a
(-'Thju-)b till tuesdZ
time(s)\ ('f;s&Tham a'wnt&, -sTm, -sTm, -sTm, -n&-, -n&-)a (-;s-, -am,
2. english
83
-m, -m, -m)b è r time I went ère ('<I;i&Tham 'f:< 'ThwE:, -T-,
-T-, -T-)a ('>-, 'f:> -, --, --, --, -:)b `ree times four are elve
to\ (Th'lnDn)a (--)b to London (Th'sku;U) to (ool (Th'gln, 'Thw-) to
Eland (Th'j;u, Th'j-, Thi'j-) to you (Th';n) to Ann ('khmTT, -TwT)a
('khm-)b come to it (j'hT, -fT, j-) you have to (Th'g;) to give (Th'wn:, Th-) to win (Th'IiT Th-)a ( Th'-)b to eat (Th'O;f, -'A- Th-)a ( Th'f)b to oer for to before consonants (, ) can be very short; and, before voiceless consonants, they are often devoiced: ('Thn T'fa;) ten to ve;
up\ ('mIkp j'ma;n, jU- -p-)a (-pj-, -j;- -p-)b make up your mind (z'wn p'; p- z-)a (z'wn p'; p- >z-)b ère is one up ère
[ère's]
upon\ ('wn &pO;n'n, &pA-, &p-, p-)a ('wn &pn'n, p-)b one upon
anoèr ('la;n pO;n'la;n, -pA-, -p-, -p-)a (-p-, -p-)b line upon line
us\ ('Thls) tell us (lts'g;, ls-, ls-)a (-;)b let's go! but: ('l[s 'g;)a ('lTs 'g;)b let us go
was\ (awz'<O:, -A;)a (-'>;)b I was wro (hiw'f<n:)a (->-)b he was a
friend
we\ (wi'm;I) we mZ ('A;<nTwi, -mpwi, -mwi)a ('A;n-, 'A;m-)b aren't we?
were\ (Iw':)a (-w>-)b `\ were ill (wj';, -j-)a (w-, -;)b were you
ère?
what\ ('sI;i wTjv'Dn:, wA-, w-, -TS-, -c-, -v- hw- W- -j- jh-)a (w-, w-,
)b see what you have done! [you've] (hi'nU;u w[i'wAn[[], wA-, w-, -'wO;n-
hw- W-)a (-'nj;u wTi'wnT, w-, )b he kn[ what he wanted (&wmj'sI, -j, &wA-, w- hw- W- -;)a (&wT-, )b what are you sZi?
[what're] (&wTDj 's;I, -TG, -TS, -c, -G, -, &wA-, w- hw- W-)a (&w-, )b
what do you sZ? [d'you] (&wTDj'DU;u, -TG, -TS, -c, -G, &wDj-, &wDj,
-, &wA-, w- hw- W-)a (&w-, )b what do you do? [d'you]
when (not interr.): (mwna'sO;T, w-, -n'-)a (-;-)b and when I sY it
(&sw~j'g ';, -w-, -T-)a (&s-, -;)b so when you get ère
where (not interr.): ('phlIs wiw'fa;n, w-)a (w>i-, w>i-)b è place
where he was found ('khnTi w-, w-)b a
country where people si
who\ ('m;n 'DDT, h-) è man who did it
will\ ('[ 'DU;u 'pw -Tw)a (-T 'D;u, )b àt will do [àt'll] (j'sI;i,
j-, ju- j jw) you will see [you'll] (wlp'w;k, -'w-, -T'w- wl-)a
(-;k)b will it work? ([bi'mI;i pw- Tw-)a (T-)b it will be me [it'll]
('GA;m wbi'hI;, -n bi-)a ('G;-, -;)b John will be here [ John'll] ('ch;c
[w]bi'f:, w-)a (-;c)b è >ur> will be full
would\ (wDpbi'g;, -b-, -Tb-) would it be good? (ID'DU;u ID- -wD-)a
(-;u)b `\ would do [`\'d] (hi'khm:, -'k- hi hi- hiw-)a (-m:)b
he would come [he'd] ([bbi'nas, -Db- pw- Tw-)a (T-)b it would be nice
[it'd], ('GI;im wd's;I&s, -n d-)a (-)b Jean would sZ so
you\ (fj'DU;u, -j-)a (-;u)b if you do (a'Thj, -j, ,- a- aw-)a (-j,
-)b I will tell you [I'll] ('kj, -j)a (-j)b ànk you ('A;<~c, -)a ('A;~c)b aren't you? ('DDc 'sI;i, -c, -TS-) didn't you see? ('DDc 'khwT, -,
84
-TS-, - j-) didn't you quit?

your\ (j'phlIs, jU-, j;<-)a (j-, j;-)b your place (&wTS'n;Im, -U-, -;<-, -tsS-,
-tsj-, &wA- -[Z- -[zj-)a (&w-, -'n-, -;'n- -TZ- -Tzj-)b what is your name?
[what's]
2.3.1.3. en prepositions become postpositions, they have full vowels:
('hU;u j'wI[&f;<, -j-)a (j-, -T&f;)b who are you waiti for? ('w; j'khm&f<m, -Am, -j-)a ('w;> j'khm&f>m)b where are you comi from? ('wt
si'lk&T, 'w[ zi-, 'wA-)a ('wt si-, -T zi-)b what is he looki at?
And when a preposition is followed by an unstressed pronoun, there are two
possibilities, according to rhythm and speaking rate (or tempo): (hi'wI[fj, -j,
-&f;<j)a (hi'wITfj, -&f;j)b he waited for you (wI'lkm, -&-)a (w'lkT, -&T)b we're looki at her.
Of course, with emphasis, things change: (wI'lk T"h:| 'nA[ T"hm:)a (w'lk T"h:| 'nT T"hm:)b we're looki at her not at him, (a"DU;u&n)a (a"D;u&n)b I do know
e forms beginning by h- after a pause, never lose initial /h/: (hi'n;)a
(-;)b he knows (hvj'sI;in, h-, -j-)a (-)b have you seen her? (h'A;<j, -j)a (-'A;j)b who are you? .
2.3.1.4. Here we will make some examples of compounds with reduced second
elements, especially in British English: ('sT<O:&b<i)a ('s>;bi, -b>i)b rYberry
/-b<i/ (in particular with monosyllabic roots, 2.3.5), ('khb)a ('khb)b cupboard, ('wEkm)a ('w-)b welcome, ('snDI, -Di)a ('sn-)b sunday (often /-dEI/ in
an intoneme, but /-di/ in a preintoneme, /-dE/);
('phIim&b)a (-Tb, -b>)b Peterborough /-b<K/, ('Dm&b)a (-mb>, -b)b
Edinburgh /-b</, ('nU;u&b<i)a ('njubi, -b>i)b N[bury /-b<i/ (in particular
with monosyllabic roots, 2.3.5), ('lsT)a ('lsT)b Leicester, ('wsT)a ('wsT)b
Worcester, ('n;<fk)a ('n;fk)b Norfolk, ('Aksf)a ('ksf)b Oxford;
('khsm) Casham, ('D;m)a ('D>m)b Durham /;</ and ('khn&hm)a ('khnm)b Cunniham /-m/, ('glnD) Eland, ('phsmn)a ('phs-)b poman, ('Gn[[]mn)a (-nT-)b gentlemen, ('s<&mni)a ('s>mni)b ceremony /-mKni/, ('phlm) PImou$;
('sO;s&phn)a ('s;spn)b saucepan /-pn/, ('nAnsns, -sns)a ('nnsns)b nonsense
/-sns/, ('j;<kSI, -S)a ('j;kS, -S)b Yorkshire /-SE/, ('hnsm) handsome, ('hnDsn)a (-D-)b Henderson, ('fksTn, -&sTn)a ('fksTn)b Folkestone /-stKn/.
Furthermore: ('la&b<<i, -b<<i, -bi, -b<i)a ('lab>>i, -b>i, bi, -b>i)b library
('khsT&m<i)a ('khsTm>i, -mi, -m>i)b cuomary (D'<kT<i, -Ti, -T<i, Da-)a
(D'>kT>i, -i, ->i, Da-)b directory ('D;<m&T<i)a ('D;m>i, -T>i)b dormitory.
Taxophonics
2.3.2.1. From the examples given thus far, the use of #onetic duration for the
various English phones will be suciently clear. However, we will summarize its
2. english
85
characteristics. In stressed syllables, the long vowels (/A:, O:, :/, and the possible
long ones from the diaphonemes /A;, ;, ;, O;, ;/) as well as the diphthongs (/Ii,
EI, aE, OE, aO, OU, Uu/) undergo a little shortening half-orteni, indeed: from
(:, ;) to (;, ) when they are followed, within the same word or rhythm
group, by at least one of the following three elements: (1) a voicele consonant, or
(2) an unreed vowel (and that changes them into diphthongs or triphthongs,
(, )), or (3) a whole unreed llable. e second elements of compounds
have secondary-stressed syllables; so they have no inuence on length.
erefore, we have: ('phl;I) /'plEI/ plZ, ('phl;I) /'plEIz/ plZs, ('phl;I) /'plEId/
plZed, ('phl;I&bk) /'plEIbk/ plZba$, but ('phlI) (or, possibly, ('phl;))
/'plEII/ plZi, ('phlI)a (-)b /'plEI/ plZer, ('phlIb) /'plEIb/ plZable, and
('phlIT) /'plEIt/ plate, ('phlIts) /'plEIts/ plates, ('phlI[)a (-T)b /'plEId/ plated,
('phlI[)a (-T)b /'plEII/ plati, ('phInT) /'pEInt/ paint
Equally: ('khA:<)a (-A:)b /'kA:/ car, ('khA:<)a (-A:)b /'kA:z/ cars, ('khA:<)a (-A:)b
/'kA:d/ card, ('khA:<D&b;<, -[b]&b-)a ('khA:D&b;, -b&b-)b /'kA:dbO:d/ cardboard, but:
('khA;<T)a (-A;T)b /'kA:t/ cart, ('khA;<bn)a (-A;b-)b /'kA:bn/ carbon, ('khA;<D)a
(-A;D-)b /'kA:dI/ cardi, ('kh[;]nT)a (-A;nT)b /'k;nt/ can't
Besides, also unstressed or half-stressed syllables shorten, as seen in cardboard (a
true compound, as to cupboard /'kbd/, that is crystallized, by now, so that a
more suitable spelling for the latter could certainly be cubbard]\ (phA;<'ThspnT)a (also (ph-)a) (-A;'Ths-)b /pA:'tIspnt/ participant (phA;<'Ths&pIT)a (also
(ph-)a) (-A;'Ths-)b /pA:'tIspEIt/ participate ('m;I&DI) /'mEIdEI/ MZdZ Besides:
('wIspIp&b[;]skT)a (-&bA;skT)b /'wEIstpEIp-b;skt/ waepaper basket
ese degrees of length hold good both in intonemes and preintonemes.
2.3.2.2. For the English stressed vowels (/I, E, , , , U/), there is another interesting fact about phonetic length. In syllables checked by nal voiced consonants,
occurring in diphonic pairs ( /b, d, g, G v, , z, Z/), short stressed vowels undergo a little lengthening half-leèni, indeed: from () to (;) ('l;) /'lId/ lid
(but ('lT) /'lIt/ lit], ('m;) /'md/ mad (but ('mT) /'mt/ mat] ('b;)a ('b;)b
/'bz/ buzz (but ('bs)a ('bs)b /'bs/ bus] ese degrees of length hold good even
in both intonemes and preintonemes.
On the other hand, when nal stressed syllables are checked by an isolated
voiced consonant ( not forming a diphonic pair that is /m, n, /), instead of
the vocoid, the contoid is a little lengthened (but only in intonemes, before pauses): ('Thn:) /'tEn/ ten, ('j:)a ('j:)b /'j/ you, ('b:) /'bU/ bull (but ('b&h;<n)a
(-;n)b /'bUhO:n/ bullhorn ('b '<;n)a ('>-)b /'bU '<n/ è bull ran).
In both accents, though, there is an exception to the exception, insofar as /, /
are half-lengthened (in a preintoneme, too), instead of a following contoid:
('GA;n)a ('G;n)b /'Gn/ John, ('b;) /'b/ ba, ('h;n) /'hnd/ hand
In the sequences /I, E, U/, the rst element is half-lengthened, both in intonemes and preintonemes; this occurs before vowels as well, if nal in rhythm groups:
('hI;)a ('h;)b /'hI/ here, (';)a (';)b /'E/ ère, ('phjU;, 'phj:)a ('phj;,
'phj:)b /'pjU/ pure ('hI; n';)a ('h;> n';)b /'hI< n[d]'E/ here and ère
We should notice that, in American English, /I, E, U/ followed by vowels,
86
within words or rhythm groups, become /I<, E<, U</: ('sp<T) both for /'spI<It/
('sp>T)b spear it and for /'spI<It/ ('sp>T)b spirit ('h<)a ('h>)b /'hI<I/ heari
2.3.2.3. Even as far as (partial) devoici is concerned, the examples thus far
will have already been a clear general survey. A short summary is, however, useful,
because there are also some particular remarks to be made, only here, even if we
need not mark them all in our transcriptions.
e devoicing of voiced diphonic consonants (/b, d, g G v, , z, Z/), before
pauses or before voiceless consonants (podevoicing), is very important: ('bA;)a
('b;)b /'bb/ Bob ('D;) /'dd/ dad ('G;)a ('G;)b /'GG/ judge ('v:)a
('v:)b /'v:v/ verve ('bA; 's;)a ('b;-)b /'bb 'sEz/ Bob sZs ('D; 'ThkT) /'dd
'tUkIt/ dad took it.
Of less importance is their devoicing after pauses or after voiceless consonants
(predevoicing), which is slighter, too. It is true that for some speakers it is as
strong as postdevoicing, but it is usually less evident, and we need not mark it in
our transcriptions (although they could be shown by means of a dot under a symbol: (b, a) (or above: (g, Z)): ('bA;)a ('b;)b /'bb/ Bob, .
A dot could be used even after aspirated /p, t, k, c/, but we will do that only
here, because (h) is sucient: (('ph&;I)) /'plEI/ plZ (('khwaT)) /'kwaEt/ quite It is
the same also for the other voiceless consonants (although their devoicing is only
slight, and therefore usually it need not be marked): (('f;u)) /'fjUu/ f[ (('T))a
(('>-))b /'<Et/ `reat (('s2:<))a (('s2:))b /'snO:/ snore
Everyday-speech simplications
2.3.3.1. In normal non-slow speech certain articulatory simplications are
quite normal. In particular, /t, d/, between C are easily dropped: ('msli)a
('m-)b /'mOUstli/ moI ('hnsm) /'hndsm/ handsome ('phsmn)a (--)b
/'pOUstmn/ poman ('ph;fkli)a (-;f-)b /'p:fIktli/ perfectI ('nks 'D;I) /'nEkst
'dEI/ next dZ ('f; ':)a ('f;s)b /'f:st 'I/ r ì ('mS p'ThI[)a (-T)b
/'mSt p'tEIOUz/ maed potatoes
is simplication occurs for /sts/, as well: ('phsts, -ss, -ss)a ('ph-)b /'pOUsts/
pos ('Thsts, -ss, -ss) /'tEsts/ tes ('Thkst s'lkSn, -ks s-, s'-)a (-S)b /'tEkst s'lEkSn/ text selection.
Besides: ('mU;uv 'bk)a ('m;uv)b /'mUuvd 'bk/ moved ba$ ('lk 'lak) /'lUkt
'laEk/ looked like ('<Iicmi)a ('>-)b /'<Iictmi/ rea>ed me ('Th; 'bA;)a ('Th;
'b;)b /'tOUd 'bb/ told Bob ('khp 'khwaT) /'kEpt 'kwaEt/ kept quiet
In addition to simplications, there are assimilations: ('h;m 'm;I, -n 'm-)
/'hnd 'mEId/ handmade ('khl; 'SAp, -)a (-; 'Sp)b /'klOUzd 'Sp/ closed op
('kh[;] 'g;, -n)a (-A; 'g;)b /'k;nt 'gOU/ can't go
Even for vowels, simplications are frequent: ('g[ 'O:, -A;)a ('gT ';)b /'gE
'l;/ get alo ('[;]fT <'wa;)a ('A;f >-)b /';ft 'waE/ after a while (T'<fk)a
(T'>-)b /t'<IfIk/ terric (ab'lI;i) /aEb'lIiv/ I believe (p'lIis) /p'lIis/ è police
2. english
87
And there are combinations, too: ('lT<li, -[<-)a ('l>li, -T>-)b /'lI<li/ literalI (ph'Thkjlli, -kjli, -kli)a (ph-, -kjl- -kjl-)b /p'tIkjlli/ particularI
('ph<Abli, -bb-, -bb-, &ph<Ali)a ('ph>-)b /'p<bbli/ probabI
2.3.3.2. In vowel combinations, within words or sentences, several simplications are possible: ('s;, 'sI) /'sEII/ sZi ('S;, 'S)a (-;, -)b /'SOUI/
owi ('si;, 'sIi) /'sIiI/ seei ('n;, -) /'nOEI/ annoyi
Besides: ('d<a;T, -aT)a ('>-)b /'d<aEIt/ dry it ('IiT, I'IiT) /EI'Iit/ `\ eat
('s;T, 'sIT) /'sEIIt/ sZ it ('na: n'n:, 'na;) /'naO n[d]'En/ now and èn ('b;
'g:, -;)a (-:)b /'bOE n[d]'g:/ boy and girl ('g; 'w;I, 'g:)a ('g;, 'g:)b
/'gOU 'wEI/ go YZ ('g; 'A;n, 'g:, 'O:n)a ('g; ';n, 'g:)b /'gOU ';n/ go on
Here are some other frequent cases that it is good to know: ('kcli, 'kSli, -Sli
&kSi) /'kcli/ actualI ('s;li, -T-, &si)a ('s;-)b /'s:tli/ certainI (D'<kTli,
'D<kli)a (D'>-, '>-)b /d'<Ektli/ directI ('Iizi, 'Iizli) /'Iizli/ easiI (g'zkTli, 'gzkli)
/Ig'zktli/ exactI (v<i'g;, vi-)a (v>i-, vi-)b /'vE ('kj, -j, 'h-, 'kj, 'kj) {(('kj))} /'kju/ ànk you
Others: (b&k, p&k-, &kh-, -'kh; b'kh; -O: -A;)a (-'kh; -;)b /b'k;z/ because ('Th:, n-)a (n-)b /n'tI/ until ('ls, n-)a (n-)b /n'lEs/ unle ('nf,
'-)a (-f, '-)b /'nf/ enough (&js'DI;i) /'jEs n'dIid/ yes indeed ('kh;<, f'- f&-,
&k-)a (-;s)b /v'kO:s/ of course (Z'uZ, '-, -Zw, Z'j-, 'j-) /z'jUuZw/ as usual (ph'hps, -&H-, &ph<-, ph-)a (ph-, &ph>-)b /p'hps/ perhaps
2.3.3.3. Some other cases: (w, w, w {,'Thj&wT, -})a ({-, --})b /wE,
w, w/ well, [I'll tell you what] (&nli'kh;, -ni-)a (--, -'-)b /IfOUnliaE'kUd/ if
onI I could ('skjuzmi, k'-) /Ik'skjUuzmi/ excuse me! (ba'ba;, b-, b-, &ba'ba;)a
(b-)b /baE'baE/ bye-bye (gb'ba;, gb-, g'-, g'- -D'b-, 'ba;) /gUd'baE/ goodbye
(gb'm;<n, gb-, g'-, g'- -D'm-, 'm;<n)a (-;n-, 'm;n-)b /gUd'mO:nI/ good
morni
Besides: ('js, 'jhs, 'jh, 'js, 'j, 'j, j, js), and ('j, 'j, 'ja:, 'jA:)a (-)b,
('j, 'jp, 'jp, 'j, j)a (-p, -, -)b /'jEs 'jE 'jEp 'jp/ yes! ('n;, 'n,
'np)a ('n-)b /'nOU/ no! (&O;'<aT, O;'-, -)a (&;-, ;-)b /O:'<aEt/ all right (khm'A;n,
-'O:n)a (-';n)b /km';n/ come on! (khm'n:) /km'In/ come in! (khm'I;, -'H-)a
(-;)b /km'hI/ come here!
More examples: (s'Da;n, s-, sT-) /sIt'daOn/ sit down (D';, aD'n;)a
(D';, aD'n;)b /aEdOUn[t]'nOU/ I don't know (D';, a&DD';)a (D';, a&DD';)b /aEdId[t]'nOU/ I didn't know ('spk&s, ak-)a ('spk&s)b
/aEIk'spEktsOU/ I expect so (a'sp, s-)a (-, s-)b /aEs'pOUz/ I suppose ('k
j'<aT, -)a (-, j'>-)b /aE'Ik j'<aEt/ I ìnk you're right
nally: (agn'DUuT, - -&gn-, --, -O;-, -A-, -T)a (-, -uT -&gn-, --,
-;-, --)b /aEmgn'dUuIt -gOUn- -'gOUIt-/ I'm goi to do it [gonna do it] (hizgn'DT -&gn-, --, -O;-, -A-, -T)a (-&gn-, --, -;-, --)b /hizgnu'dIt -OU- -'gOUIt-/ he's goi to add it (wi&wAn's;I, -wO;-, -nT-, -n-)a (-w-)b /wiw;n'sEI/ we want
to sZ [wanna] (wi&wAn'IiT, -wO;-, -nT-)a (-nT-)b /wiw;nu'Iit/ we want to eat ('sm 'Df<nT, 'smp)a ('sm-, ->nT, 'smp)b /'smI 'dIf<nt/ someì dierent
88
(f'f:, f'-, f'-, f'-) /fU'fI/ fulll (M'va<nmnT, -mm-, -m-, -a<m-, -,
-)a (->nm-, ->mm-, ->m-, -am-, -am-, -, -)b /In'vaE<nmnt, E-, -/ environment ('gvnmnT, -vm-, -vm-, -v-, -vm- -b- -mm-)a ('gvnm-, -vmm-,
-vm-, -v-, -vm- -b- -mm-)b /'gvnmnt/ government.
American English has, too: ('khnD&DIT, 'khn-, -DT) /'knddEIt, -dt/ candidate (nT'A;<kTk, n['-, n'-, -'A;<Tk) /n'A:ktIk/ antarctic
2.3.3.4. It is important to know that, in English, phonic syllabication generally follows morphemic divisions. is allows some slight dierences to be maintained that Romance languages, instead, usually lose: ('n;Im) /'nEIm/ a name
(n';Im) /n'EIm/ an aim ('n< 'sa;n)a (->)b /'n<OU 'saEn/ a narrow sign
(n'< 'sa;n)a (->)b /n'<OU 'saEn/ an arrow sign
In addition, let us note: ('wa; 'chU;u)a (-;u)b /'waE 'cUuz/ why >oose ('waT
'SU;u, -)a (-;u)b /'waEt 'SUuz/ white oes (ma'Th<;In)a (-'h>-)b /maE't<EIn/ my
train ('maT '<;In, -)a ('>-)b /'maEt '<EIn) might rain ('blk 'Tha;) /'blk 'taE/ a
bla$ tie ('blkT 'a;) /'blkt 'aE/ a bla$ed \e
Forms like (ms'spE:)a (-:)b /mIs'spE/ mipell and (ms'sm) /mIs'smI/ Mi
Smi` may seem a bit strange. Indeed, as consonants often fall between others, so,
in less slow manners of speaking, even (m'spE:, m'sm)a (-:)b occur.
For British English, the following examples are usual, too: (s'b;g 'm;n, s-)b
/sOU'bIg 'mn/ so big a man ('bg 'm;n)b /'bIg 'mn/ a bigger man ('laT 'fa)b /'laEt 'faE/ light a re ('laT 'fa)b /'laEt 'faE/ a lighter re (tsz'w
T'wIT)b /Itsz'wE t'wEIt/ it's as well to wait (hiz'wT&wIT)b /hiz'wEtwEIt/
he's a welterweight
Let us add an important remark about the syllabic structure regarding /'0/,
which has /I, E, , , , U/, even preceded by /j, w, <, l/, or with nal /, /. Although we will not mark it systematically, but only here (as it would be an almost
useless increase in weight of our transcriptions), it is worthwhile knowing that a
single consonant and the preceding short stressed vowel belong to the same syllable: ('lm-)a ('lT-)b /'lI/ litter ('bg-)a ('bg-)b /'bEg/ letter ('phk-)a ('phk-)b /'pk/ pa$er ('khm-)a ('khT-)b /'k/ cutter ('hAm-)a ('hT-)b /'h/ hotter ('lk-)a ('lk-)b /'lUk/ looker ('sl-)a ('sl-)b /'sEl/ seller ('kh<-i)a ('kh>-i)b /'kh<i/ carrier ('skn-)a ('skn-)b /'skn/ scanner
More examples: ('<n-)a ('>n-)b /'<nI/ runni ('lv-)a ('lv-)b /'lvI/
lovi ('mT-, 'm-)a ('mT- 'm-)b /'mt/ mutton ('khAT-, 'khA-)a ('khT- 'kh-)b /'kt/ cotton ('ls-) /'lIs/ lien ('mD-) /'mId/ middle ('m[-)a
('mT-)b /'mE/ metal ('Dv-) /'dEv/ devil ('mT-<k)a ('m->k)b /'mEt<Ik/ metric
('k-j<T)a ('k-j>T)b /'kj<t/ accurate ('<Ak-w)a ('>k-w)b /'<kw/ Ro$well
('<Ak-li)a ('>k-li)b /'<kli/ Ro$l\ ('<p-li)a ('>p-li)b /'<Ipli/ Ripl\
On the contrary, stressed long vowels and diphthongs belong to dierent syllables as to following single consonants: ('nO;-[i)a ('n;-Ti)b /'nO:i/ naugh ('nUu-T,
'nUu-)a ('nju-T 'nju-)b /'nUut/ N[ton ('lI-m)a ('lI-T)b /'lEI/ later
('bI-k) /'bEIkI/ baki ('Ii-kw) /'Iikw/ equal ('nUu-T<)a ('nju->)b /'nUut</ neutral ('<a-pli)a ('>a-pli)b /'<aEpli/ ripeI
2. english
89
American dissimilation of r
2.3.3.5. To simplify the articulation of words and rhythm groups with two //'s,
American pronunciation can have variants with /`/ for the rst //, although not
very frequently, even in stressed syllables: ('fA;<, 'fA;) /'fA:/ farèr ('f;,
'f;) /'f:/ furèr (';<D, ';D) /'O:d/ order ('m;D, 'm;D) /'m;d/ murder ('kh;<n, 'kh;n) /'kO:n/ corner ('mAmm, '-) /'mm/ èrmometer ('f;<wD, 'f;wD) /'fO:wd/ forward ('f;<&w;D, 'f;&w;D) /'fO:w:d/ for[ord
(s'ph<a;, s'-) /s'p<aEz/ surprise ('gvn, -vn, -vn) /'gvn/ governor (ph'Thkjl, ph'-) /p'tIkjl/ particular
Also: ('khn[m]&b<i, -&b-) {('khnTbi, -b>i, -&b>i)b} /'knb<i/ Canterbury
('wO;m&b<i, -[-, 'wA-) {('w;Tbi, -b>i, -&b>i)b} /'wO;b<i/ Waterbury ('Am&b;n, 'A[-) /'b:n/ Otterburn ('b;n, 'b;-, b'nA:<, b-) /'b:nd, b'nA:d/ Bernard ('khm&pl, -[-) /'kpIl/ caterpillar ('ED'b<i, -D&-) {('D&b>i, -bi, -b>i)b} /'Edb<i/ elderberry ('<z&vwA;<, -z&- -&vw;<) /'<EzvwA:/ reservoir ('n;<<p, 'n;-) /'nO:<p/ Nor`rup ('sn, 'sn) /'sn/ Souèrner ('m;D<, -D, 'm;D-) /'m:d</ murderer ('n;<n, 'n;<n, 'n;n) /'nO:n/
Norèrner
Less systematically, dissimilation is possible even in rhythm groups: (h'hA;<T,
h'-) /h'hA:t/ her heart (j'm;i, j'-) /j'm:si/ your mercy (A;<'ph;<c, A;'-) /A:'pO:c/ our por> ('f:m, '-, '-, '-) /E'f:mz/ èir rms ('D;[i, '-,
'-, '-) /E'd:i/ `\'re dirty (f'w:, f'-) /f'w:d/ for word (;<'hI;, ;'-, '-,
'-, -'H-) /O:'hI, '-/ or here
In addition to //, the following examples will show dissimilation for /</ (which
is used in British pronunciation too, due to a kind of simplication, even by analogy): ('la&b<<i, -b<<i, -bi, -b<i, -&b-, -b-)a ('lab>>i, -b>i, bi, -b>i, -b>i)b /'laEb<<i/ library ('fb<&<i, -b<<i, -bj-, -bj-)a ('fb>>i, ->>i, -b>i, bi, -b>i, -bj>i,
-bj-)b /'fEb<u<i, -bju-/ February ('sk<&T<i, -k-)a ('sk>>i, -&T>i, -k-)b /'sEk<t<i/ secretary.
In addition: (sT'nAg<f -g-)a (-g>- -g-)b /st'ng<f/ enogra#er (f'ThAg<f -g-)a (-g>f -g-)b /f'tg<f/ #otogra#er (ph<'fs ph- ph-)a
(ph>'fs ph-)b /p<'fEs/ profeor (ph<&nnsi'ISn ph- ph-)a (ph>&nnsi'IS ph-)b /p<nnsi'EISn/ pronunciation (ph<'ph<&T'ph>>i, ph>- ph-)b /p<'p<tK<i 'p<Ep<tK:<i/ preparatory
All in all, we can see that dissimilation mostly occurs with: (';<) /'O:/, usually,
in stressed syllables (where, even if // is not pronounced, words do not become
ambiguous, because the vowel timbre alone is distinctive; see the examples above),
and with: () // in unstressed syllables: surprise particular caterpillar governor
èrmometer
Morphonological remark
2.3.3.6. Now, thanks to transcriptions (which do not hide reality, as spelling
does) we will resolve a widespread problem for foreigners knowing which pronunciation to use for the grammemes >-ed -(e)s -'s, -s'.
90
Simply, we have:
/d/ (D) after voiced phonemes ( vowels, diphthongs, and voiced consonants,
except the very /d/): /'plEId/ plZed, /'h;<id/ hurried, /'bA:d/ barred, /'<bd/ rubbed,
/'GGd/ judged, /'plnd/ planned
/t/ (T) after voiceless consonants (except the very /t/): /'stpt/ opped, /'w;St/
waed, /'swIct/ ~ibed, /'l;ft/ laughed
/d/ (D)a (D)b after /t, d/, in order to be able to pronounce them: /'wEId/ waited, /'nIidd/ needed, /'stA:d/ arted
nally, we have:
/z/ (z) after voiced phonemes ( vowels, diphthongs, and voiced consonants,
except the grooved ones, /z, Z, G/): /'gOUz/ goes, /'flaEz/ ies, /'dEIz/ dZs, /'lEIdiz/
ladies, /'<bz/ rubs, /'wEIvz/ waves, /'plnz/ plans, /'Gnz/ John's, /'<Icdz/ Ri>ard's
/s/ (s) after voiceless consonants (except the grooved ones, /s, S, c/): /'tps/ tops,
/'<aEts/ writes, /'b<Es/ brea`s, /'GEfs/ Je's, /maE'pE<nts/ my paren_'
/z/ (z)a (z)b after /s, z S, Z c, G/, in order to be able to pronounce them: /'kIsz/ kies, /'<OUzz/ roses, /'dISz/ dies, /'swIcz/ ~ibes, /'ksz/ axes, /'cA:zz/
arles's
Stress
2.3.4.1. We know that (the position of) stress may be distinctive, in English:
('mp;<T)a (-;T)b import (noun, adj.), (m'ph;<T)a (-;T)b import (verb); ('ph<zT)a (->-)b present (noun, adj.), (p<'znT)a (p>-)b present (verb).
English sentences usually keep the stresses of their words well, even in lexical
monosyllables, while grammatical monosyllables lack any stress (as, in general, do
polysyllabic unstressed syllables): ('smz 'bO;t '<I;i 'nU;u 'smO: 'blk 'khts)a ('b;t
'>I;i 'nj;u 'sm:)b Sam has bought `ree n[ small bla$ ca_ but we have: (w'lA:< 'kh<a;D *'phIip)a (-w'lA:)b and ère was a large crowd of people,
(b[&ts*'g<I[sT m'ph;<Ts, -s)a (bT&-, -'g>IT-, -;-)b but it's of è greate importance
However, in long words (and in sentences as well), many syllables with full vowels generally receive secondary stresses (especially when they occur near unstressed
and reduced syllables): (&v'sT&mIT)a (&v>'sT-)b overeimate, (&vz'bl[i)a
(-Ti)b visibili, (&ph;pn&Dkj'l<[i)a (&ph;-, ->Ti)b perpendiculari, ('<kg&na)a
('>-)b recognize, ('mks&k)a (-)b Mexico.
In compounds, the more frequent structure is ('&) (more rarely (&'): (&ff'ThI;in)
fteen). Sometimes, even ('') occurs, as in collocations (or occasional or free
compounds, which are, then, modiable): ('blk&b;)a (-;)b bla$bird (but ('blk
'b:)a ('b:)b bla$ bird), ('b&DA, -O;)a (-)b bulldog
Of course, there are also many instances like: ('glS&ThIic)a (-)b Eli tea>er
a teacher of English and ('glS 'ThIic)a (-)b Eli tea>er a teacher who is
English.
2. english
91
2.3.4.2. Let us now consider compounds such as r cla (noun and adverb)
and r-cla (adjective), and the collocation r cla in a sentence like àt was
è r cla to be considered From a phonetic point of view, they are alike: ('f;[T]
'khl[;]s)a ('f;s[T] 'khlA;s)b; however, from a phonemic point of view, and for teaching and lexicographical purposes as well, it could be very useful to distinguish
them as: /'f:st'kl;s/ (compounds: /''/) and /'f:st 'kl;s/ (collocation: /' '/).
Besides, patterns are exibly structured. As a matter of fact, we have: (&ff'ThI;in)
fteen and ('ph;I ff'ThI;in) page fteen but ('ff&ThIim 'ph;IG)a (-)b fteen
pages ('b<;n[D] 'nU;u)a ('nj;u)b brandn[ but ('b<;n[D]&nUu km'phjum)a
(-&nju, -T)b a brandn[ computer
Moreover: ('skn[D] 'h;n) secondhand but ('skn&hn 'khl;[])a
(-;[])b secondhand cloès and ([]'O: &skn'h;n)a (>':)b `\'re all secondhand also (&fT'nU;un)a (&A;fT'n;un)b afternoon and (&gDfT'nU;un, gD&-,
gD&-)a (&gDA;fT'n;un, gD&-, gD&-)b good afternoon but ('fT&nU;un 'ThI;i)a ('A;fT&n;un 'ThI;i)b afternoon tea
A few cases can vary according to speech rate, but also whether they occur in
intonemes or preintonemes, as well as according to personal choices. Here, we will
make use of dierent degrees of intermediate stress, too, which (without an emphatic one, (")) are, in descending order: ('), (), (&), (), ( ). It is worthwhile observing nuances carefully: (&Dmn'sT<ISn 'ks&saz, 'Dmn&s-, -'sT<ISnks&saz)a
(-'sT>IS 'ks&saz)b demonration exercises ('l&vIm 'Ap&Im, 'l&vImAp&Im, 'lvIm&ApIm)a (-T> 'p&>IT)b elevator operator ('laT&has 'khIip, 'laT&haskhIip, 'laThas&khIip)a (-)b lighòuse keeper let us notice: ('laT 'has&khIip)a (-)b light housekeeper
2.3.4.3. To feel certain about the stress patterns of compounds, it is necessary
to look them up in reliable dictionaries. But pronunciation dictionaries are not always the best choice, for this aspect, although, of course, they have to be consulted. We willingly recommend the Random House dictionaries which, for secondary stress, are almost perfect; of course, the stress patterns shown are American
ones, but, in general, they may hold good even for British English, which, in the
meanwhile, may have added kept some other possible variants (mainly collocation-like, rather than compound-like, so less useful ones: weekend icecream N[
York N[ Zealand N[ Hampire).
In addition, the Oxford Advanced Learner's dictionaries show the marked
cases of primary stress in several lexical collocations (which are quite unpredictable, above all for foreigners).
In (dia)phonemic transcriptions such as ours, the most typical and numerous
compounds are shown with a single primary stress /'/; the secondary one is easily recoverable, because the second lexeme necessarily bears a secondary stress.
cce versa, most dictionaries printed in the include secondary stress, /'&/;
but usually the non-IPA symbols they use put stresses after stressed syllables, unfortunately, not before, and simply through a dierence in thickness (which, sometimes, is not evident enough, even with both of them in prsentia); as a matter of
fact, we happen to nd, in scrib instead of /In'sk<aEb/ inscribe and viz bil
92
te for /vIz'bIli/ (&vz'bl[i)a (-Ti)b visibility But some American dictionaries

are misleading, because they mark secondary stress for most unstressed syllables
bearing full vowels.
Regrettably, mainly dictionaries published in the (even pronouncing dictionaries) do not use secondary stress wisely enough. As a matter of fact, a collocation
like ('snT< 'hIi[)a (->-, -T)b /'sEnt< 'hIiI/ central heati is, usually, represented as */&sentrl 'hi:tI/, exactly like (&snT<'lsTk)a (->-)b /sEnt<'lIstIk/ centraliic
(their */&sentr'lIstIk/).
However, the more they mark the better, even when things are predictable, provided they do so in an exact and accurate way. Indeed, teaching transcriptions, especially for beginners, should show several characteristics, with no absurd and
groundless fear that they may confuse. In reality, too simple a transcription is less
useful and, sometimes, misleading.
2.3.4.4. As regards diaphonemic transcriptions in compounds with suxes, it
is sucient to know which of them are always non(half)reable (//) and
which are prosodically (half)reable (/%/). As a matter of fact, the others, that
have full vowels, are always (half)reable (/&/). In addition to those with /, ,
/, the following are always unstressed: /-Ik, -Iks, -I, -IS, -Ist, -Iv, -fI/ -ic -ics -i
-i -i -ive -#il\ /'<<Ik, 'pltIks, 'lIg<I, 'jElOUIS, 'nvlIst, d'sk<IptIv, 'glfI/ rhetoric politics lieri yellowi noveli descriptive alo#il (for
-#ile we have /-&faE, -fI/).
Instead, the following are half-reed (if preceded by an unstressed syllable), but
unreed (if preceded by a stressed syllable): /-hUd, -Izm, -aEt, -aEz, -SIp, -jUu/
-hood -ism -ite -ize [-ise] -ip -ule\ /'wUmn(&)hUd/ womanhood ( /'caE[d]hUd/
>ildhood] /'tE<(&)<Izm/ terrorism ( /'bUdIzm, 'bUu-/ Buddhism] /'t<tski(&)aEt/
Tro_kyite ( /'sfaEt/ sulte] /'k<I(&)saEz/ criticize (but /'bptaEz/ baptize in addition to /bp'taEz/] /'skl(&)SIp/ (olarip ( /'f<En[d]SIp/ friendip] /'ml(&)kjUu/ molecule ( /'glbjUu/ globule]
Intonation
2.3.5. As far as intonation is concerned, close observation of the tonograms for
preintonemes and intonemes ( 2.7-8) of both accents is sucient. Technically,
we talk about intonation groups (or tone groups, for short), which are generally
composed of a rst part, the preintoneme, and a second, or intoneme. e latter is
the most important for conveying pragmatic meanings, such as atement,
queion, . It is common knowledge that these structures depend on oròlogy (
expressive speech) and semantics. ese will produce particular eects, but always
within usual primary intonation patterns, which are exible, though systematic.
An added complication is para#onics (which marks attitudes, moods, feelings
and social roles). All this is typical of any common messages, even in every-day simple conversation:
/./: (av'Gs[p] 'bO;[ 'nU;u 'DkS&<i23)a (av5Gs[p] 'b;T 'nj;u 'DkS>i3 3)b /aEv-
2. english
93
'Gst 'bO:t 'nUu 'dIkS<i./ I've ju bought a n[ dictionary

/?/: (Dj'spIik 'glS wE:21, G-)a (Dj5spIik 'glS 'w:21, G-)b /dj'spIik 'IglIS 'wE?/ Do you speak Eli well?
//: (jkn'h; 6sT<O:&b<iz2 2 'blU;u&b<i23)a (jkn5h; 's>;biz32 ;'blubi3 3)b
/jukn'hv 'st<O:b<iz K:'blUub<iz./ You can have rYberries or blueberries
2.3.6. ere are several and quite varied queion tags, or tag queions in English, while, other languages generally have xed formulas. In the English language, they are morphologically determined (by modifying auxiliary and modal
verbs and changing their positive/negative polarity). ey have two dierent functions: conrmations of somebody's suppositions (by means of conclusive intonemes), or actual queions, to really ask something, for lack of any certainty.
Let us see a few examples: It's cold todZ, isn't it? or You're American, aren't you?
or \ aren't well, are `\? For conrmation, we will have: /Its'kOUd t'dEI. 'Is{}t./, /j<'mE<kn. 'A:ntjUu./, /EI'A:nt 'wE. 'A:EI./; while, for information: /Its'kOUd t'dEI. 'Is{t}t?/, /j<'mE<kn. 'A:ntjUu?/, /EI'A:nt 'wE. 'A:EI?/.
2.7. American intonation.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 6 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
/ / (2 2 5 1 1 ' 2 2 ' 2)
/./ (2 ' 3 3)
/ / ( 2 2 5 1 1 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 5 1 1 ' 2 2 ' 2)
// (2 ' 3 2)
/ / ( 2 2 5 1 1 ' 2 2 ' 2)
/,/ (2 ' 2)
2.8. British intonation.
Other accents
2.4.0. is section is likely to be very important for descriptive and communicative purposes, as it deals with pronunciations going beyond the neutral accents.
94
International neutral accent

2.4.1.1. As a teaching application for foreigners, the proposal of an international accent seems to be welcome and useful. It will certainly have more real advantages than learning without a method, in a wild and uncontrolled way, and acquiring elements of both neutral accents, together with many personal (and regional) peculiarities and interferences due to spelling inconsistencies.
ings will become simpler and more straightforward, when, at long last, a dictionary with diaphonemic transcriptions is available. e international accent
is mainly based on the ~~ pronunciation, which covers the whole globe. Although it is slightly more American-like, which is the more widespread accent and
also the less diverging if possible from current spelling, still it is not too American. Of course, we are referring to newsreaders, not to local correspondents, who
in certain cases may even be speaking English as a second language.
Indeed, it simplies actual complexities of real accents, above all the British
one, eliminating unnecessary and unwanted distinctions (not shown by spelling,
among other things), to recover a more organic and general situation. All this is
done, starting from actual pronunciations, which, moreover, lack any connotations that can be easily localized. Many singers and actors use it.
So, roughly, this international pronunciation is intermediate between the
American and British neutral pronunciations. Besides, it does not sound strange
to any native speaker. It simply is more organized, but with no undue or far-fetched exaggerations.
2.4.1.2. Starting from our diaphonemic transcription, the international accent
is obtained mainly by bringing the diaphonemes /;, A; ;, O; ;, / to their most
natural matrices (/, A: , O: :, / (, A: , : :, )), more traditional and
widespread, as well as less apart from spelling, as we have already said. In this way,
we can simplify the hard task of foreigners, who unfortunately are forced to
learn from spelling. Actually, our modern-language teaching is still in a pitiful
condition as far as pronunciation is concerned, as it is often completely neglected.
us, we have: ('lsT)i ('l[;]sT)a ('lA;sT)b /'l;st/ la, ('phA;sT)i ('phA;sT)a ('phsT)b /'pA;st/ paa, ('s<i)i ('sAi)b /'s<i/ sorry, ('lsT)i ('lO;sT, 'lA-)a ('lsT)b
/'l;st/ lo, ('f;s)i (see below for (), which is a semi-lateral contoid, with no real
contact) ('fO;s, 'fA-)a ('f;s, 'f-)b /'fO;s/ false, ('DS)i ('DS)a ('DS)b /'dISz/ dies,
('wnT)i ('wAn[[], 'wO;-)a ('wnT)b /'w;nd/ wanted, ('h;<i)i ('h;i)a ('h>i)b
/'h;<i/ hurry 2.9-10 give the vocalic articulations of the international English
accent.
2.4.1.3. As for unstressed syllables with possible full timbres, the international
pronunciation, instead, has the less prominent ones, even if not extreme (() is a
semi-approximant, 1.13.3): ('<gjl)i /'<Egjl/ regular ('DkSn<i)i /'dIkSn<i/
dictionary ('D;mc<i)i /'dO:mtK:<i/ dormitory. us, for /t, d/ + /{}</, /c<, G</
are preferable (as many native speakers do themselves, all over the world).
2.9 shows that the international phonemes /, , O:, u/ are (, , :, u):
2. english
95
2.9. International monophthongs.

/i+/ (i) /i/ (i)
/I/ ()
// = //, // (, +|)
/E/ ()
// ()
// (), /;/ ()
/u/ (u)
/U/ ()
/:<, ;</ (:<), /</ (<), // (q), /:/ (:q)
/O:, O;/ (:)
/, ;/ ()
/A:, A;/ (A:)
2.10. International diphthongs.
/EI/ ()
/OU/ (), /OE/ ()
/aE/ (a)
/aO/ (a)
/j/Uu/ (juu)
f
/Ii/ (ii)
('hT)i ('hT)a ('hT)b /'ht/ hut, ('hT)i ('hAT)a ('hT)b /'ht/ hot, ('s:)i ('sO:)a ('s:)b
/'sO:/ sY, (&scu'Sn)i (&sc'ISn)a (&sc'IS)b /sIcu'EISn/ situation
For /:/, we have (:): ('w:)i ('w:)a ('w:)b /'w:d/ word for /:<, ;</ (:<):
('h;<i)i ('h;i)a ('h>i)b /'h;<i/ hurry for //, in unstressed syllables, we nd ():
('b<[])i ('b<[])i (-[])a (-, -)b /'b<[z]/ broèr[s]
For /I, E, U/ we have (;): ('h;)i ('hI;)a ('h;)b /'hI/ hear, (';)i (';)a
(';)b /'E/ ère, ('ph;)i ('phU;)a ('ph;)b /'pU/ poor and for /I<, E<, U</ we
have (<, <, <): ('h<)i ('h<)a ('h>)b /'hI<I/ heari, ('m<i)i ('m<i)a ('m>i)b /'mE<i/ Mary, ('Dj<)i ('D<, 'D;-)a ('Dj>, 'Dj;>-)b /'dU<I/ duri
2.4.1.4. As to diphthongs, it is sucient to notice: /Ii, Uu, OU/ (ii, uu, ) (notice that (ii, uu) are diphthongs, although very narrow): ('Thi;i)i ('ThI;i)a:b /'tIi/ tea,
('fju;u)i ('fj;u)a:b /'fjUu/ f[, ('Thu;u)i ('ThU;u)a ('Th;u)b /'tUu/ o, ('n;)i ('n;)a
('n;)b /'nOU/ no. Besides, for /aE, OE/, a second element like () is enough (
2.10, to avoid introducing/learning another vocoid; so much so that a pronunciation with () is much better than foreign *(ai, Oi), and it is near the native
one): ('ga;)i ('ga;)a:b /'gaE/ guy, ('b;)i ('b;)a:b /'bOE/ boy Little has to be said
about /EI, aO/: ('D;)i ('D;I)a:b /'dEI/ dZ, ('na;)i (na;)a:b /'naO/ now
Neither the () taxophone of // near velars, nor other taxophones of /I, E, ,
U , Ii, Uu EI, aE, OE/ + /, / are necessary, but their use makes one's pronunciation
more authentic, although this international model is already denitely more
realistic than those oered in so many textbooks, even by English-speaking phoneticians (where, among other thing, /i:, u:/ are still indicated). us, taxophones
like ( #, #, #, # @, iI, uU a, ) (but (I, )) can really change one's pronunciation from simple international (for foreigners) into native international
pronunciation. Instead, the use of (i:, u:) clearly indicates either a non native-like
pronunciation, or one which is regional or socially inappropriate.
96
For /, /, it is better to use (, ) (semilateral, 1.15.1), which, articulatorily,

are decidedly simpler than (, ) (in case, even velar can be used velar semilateral (), ) which, auditorily, are decidedly better than (l, )).
As for consonants, suce to say that () is acceptable for /t/ ( 2.2.2.4) and that
([) may be good for //, mainly after vowels, while, (T) is more recommendable, after /n, , /: ('bTi, -[i)i ('b[i)a ('bTi)b /'bEi/ Bet, ('ThwnTi)i ('Thwn[[]i)a ('ThwnTi)b /'twEni/ en
us, for r we have: /</ (<), // (). Besides, /w/ simply corresponds to /w/:
('wn:)i /'wEn/ when, and // to /j/: ('Thju;un)i /'tUun/ tune, ('Dju;un)i /'dUun/ dune,
('nju;u)i /'nUu/ n[
e intonation of international English has a restrained and more general
movement, as can be seen from 2.11 (although the rst stressed syllable in a preintoneme could be half-high, (), instead of just raised mid).
2.4.1.5. Both the and the American models have some problems of social
acceptability. , though still very widely used, both in the news and in
some kinds of British sitcom and movies, has always had a strong connotation of
articial aectation, which makes it disagreeable to many native speakers. is
generally associated with a high social position ( members of the aristocracy,
of the higher clergy or military ranks, Tory MP's, prestigious university professors,
), and a certain age group (over 50 years of age). Clothing, too, should be suciently formal, to be suitable for the accent. If these conditions are lacking
for common natives the British neutral accent could prove to be denitely inappropriate.
It is to be said that, paradoxically, even an impersonation of an speaker (even
only partially successful, especially if belonging to certain particular varieties) may
give rise to negative feelings from British listeners belonging to the middle or
working class. Indeed, these people might nd a few snobbish phonetic traits, easily recognizable as marked ones, even if mixed with foreign traits, especially if the
speakers are young(er).
Against these (empirically checked) diculties, it might be better to choose the
American neutral accent. is is certainly recommendable to learners in North
America, but not in the British Isles, where most people would consider it inappropriate.
2.11. International intonation.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 ' 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
2. english
97
We reckon that this dilemma may be faced, in a practical and diplomatic way,
aiming at acquiring an international accent, which although it might seem to
be nobody's accent would have none of the possible negative connotations of
the two mediatic accents (we will be dealing with below), or even of the two national neutral ones (which some people seem to consider nobody's accent, as
well, since in both nations only about 3% of native speakers actually use them).
Mediatic accents
2.4.2.0. In addition to the two neutral accents, the American and the British
ones, and to the international accent, we believe it is important to show the two
kinds of accent actually used by American and British native speakers. Naturally,
mediatic pronunciations are often used by North-American people in the midwestern states, and by English people in the southeastern counties, respectively, as
well as by most spoken-word mass media ( radio and television) that do not use
the neutral accents nor more local ones.
erefore, 2.12-18 (which speak for themselves) must be carefully analyzed,
comparing them with the neutral ones, in order to capture the dierences, which
are sometimes not slight! In ordinary people's opinion, mainly if their own pronunciation is directly concerned, these mediatic accents are thought to be less peculiar than the neutral ones.
On the other hand, as everyone knows, neutral pronunciation in percentage
terms is used the least by native speakers; but, it is the one generally aimed at by
advanced foreign learners, except for more or less frequent interferences, especially from their mother tongue, and individual peculiarities.
For this reason, we do not hesitate to show these actual realities, although we
will not get to the point of recommending an active usage by foreigners. On the
contrary, a passive usage is certainly welcome, to really understand native speakers, when they talk as they can. Actually, this happens every day, all over the
world, because school and society usually ignore (good) pronunciation.
However, as these mediatic pronunciations are really very widespread, and often considered as almost neutral (or, at least, less aected and less articial than
the neutral accent), many people would be willing to declare them to be neutral.
ey would do so, on the one hand, in opposition to pronunciations which are
more recognizable as local ones, and, on the other hand, to an unsubstantial neutral kind of pronunciation It is no rare fact that some mediatic speakers uctuate towards the neutral type (or away from it), for some words.
Here we wish to briey draw attention to some details, with reference only to the
respective (American or British) variant, for a direct (internal) comparison. In the
light of what has been seen so far, it will not be dicult to nd the corresponding
neutral forms of the other accent. On the contrary, it will be a very useful exercise.
98
Mediatic American English pronunciation

2.4.2.1. Traditionally it was called General American accent, since it is dierent from the stereotyped accents of New York City, and Eastern New England, or
of the Great Lakes ( Ontario, Erie, Huron, Michigan, Superior), or the South
(either the Deep South and its variants, or its mountains, Appalachia, Ozarks),
or of the American Blacks.
2.4.2.2. As can be seen from 2.12, /I, U, E/ are (, , E): ('hT)m;a ('hT)a /'hIt/ hit,
('phT)m;a ('phT)a /'pUt/ put (']E;)m;a ('l;)a /'lEg/ leg (let us also note ('f:)m;a ('f:)a
/'fI/ ll, ('fP:)m;a ('f:)a /'fU/ full] whereas, /I, U, O:/ are (>, >, :): ('h;>)m;a
('hI;)a /'hI/ here ('ph;>)m;a ('phU;)a /'pU/ poor ('w:)m;a ('wO:)a /'wO:/ wall
('s:)m;a ('sO:)a /'sO:/ sY (instead, /O:, O:</ are slightly raised). Equally, in /I/ it has,
(more) often, (I) (sometimes (i) too, 2.14): ('sII 'sii)m;a ('s)a /'sII/
sii
In addition, /u/ is fronter, (%): (&sc%'IS'n)m;a (&sc'ISn)a /sIcu'EISn/ situation // is fronter and higher, (x) (and in // it is rounded, ()): ('hxT)m;a ('hT)a
/'ht/ hut ('h:)m;a ('h,:)a /'h/ hull ere are some further slight modications
which however do not change the phonetic symbols.
Let us rather talk about /;/, which is diphthongized, (): ('phsT)m;a
('ph[;]sT)a /'p;st/ pa, as well as // () (which is considerably raised and
nasalized, too): ('m;n)m;a ('m;n)a /'mn/ man ('n'm)m;a ('nm)a /'nm/
animal (fn'ThsTk)m;a (fn'ThsTk)a /fn'tstIk/ fantaic ('kj%, -jx)m;a
('kj, -j)a /'kj, -j/ ànk you
Some phonemes neutralize, which leads to the merging into ('m<i) for /'mE<i,
'mE<i, 'm<i/ merry Mary marry In addition, /;/ merges into // or /O:/, generally as in American neutral pronunciation: ('sA;<i)m;a ('sA<i)a /'s<i/ sorry ('s:)m;a
('sO:)a /'s;/ so
We also nd the neutralization of /A:/ and // into (A[:]), including /A:<, A:/,
with an oscillating phonetic length: (A:, A;, A). Generally, in monosyllables (or in
nal-stressed words), we have (A:), when absolutely nal or followed by voiced C:
('bA:, 'n:, 'spA:, 'b<A:, 'fA:<, 'khA:<)m;a ('bA;, 'nA; 'spA:, 'b<A:, 'fA:<, 'khA:<)a
/'bb, 'nd 'spA:, 'b<A:, 'fA:, 'kA:d/ Bob nod spa bra far card. However, we have
(A), when followed by voiceless C: ('sTAp, 'hAT, 'SAk)m;a ('sTAp, 'hAT, 'SAk)a /'stp,
'ht, 'Sk/ op, hot, o$
In bisyllables (or penultimate-stressed words) we nd (A;) in intonemes, but (A)
in preintonemes: ('fA;<>, 'fA;>, 'bA;>, 'hA;m> 'mA;]i, 'khA;, 'phA;pi, 'DA;gm)m;a
('fA;<, 'fA; 'bA, 'hAm, 'mAli, 'khA, 'phApi, 'DAgm)a /'fA:, 'fA: 'b,
'h, 'mli, 'ppi, 'dgm/ farèr faèr boèr hotter MolI cotton poppy
dogma; ('hAm> n'hA;m>)m;a ('hAm n'hAm)a /'h< n[d]'h/ hotter and hotter
('hA<D> n'hA;<D>)m;a ('hA<D n'hA<D)a /'hA:d< n[d]'hA:d/ harder and harder
On the contrary, in plurisyllables (or in prepenultimate-stressed words) we have
(A): ('DAkj'm'nT, 'phA]Tks, 'DAgm'TsT, 'khA<Dgn)m;a ('DAkjmnT, 'phAlTks,
'DAgmTsT 'khA;<Dgn)a /'dkjmnt, 'pltIks, 'dgmtIst 'kA:dIgn/ document
politics dogmati cardigan
As can be seen from 2.12, the articulation of (:) and (A:) is mainly distin-
2. english
99
guished by labialization (which is slight since the two vocoids are low). us many
speakers can unify them (but, productively perceptively, things are rather complex and oscillating) by also obtaining ('sA;<i, 's:)m;a (('sA<i)a /'s<i/ sorry ('sO:)a
/'s;/ so), all the more so because, for /O:/, some intermediate articulations between /O:/ and /A:/ are decidedly common: (A), as in ('s:, 's;)m;a /'s;/ so
and ('s:, 's;A)m;a /'sO:/ sY
Besides, we have // (x)m;a (//), not only when nal before a pause, but even
after a pause, even if near velar(ized) C, where in the neutral accent we nd
()a: (x'ph];) /'plaE/ appI (x's;) /'saEd/ aside (x'ThmpT) /'tEmpt/ attempt
(x'Th;In) /b'tEIn/ obtain (x'];sn) /'lOUn/ alone (x'g;) /'gOU/ ago (x'w;I) /'wEI/ YZ (x'<;n) /'<aOnd/ around (x'kh>:) /'k:/ occur
2.12. Mediatic American monophthongs.
/i, i/ (i), /[']I/ (I)
/I[]/ (, ), /I/ (>), /I</ (<)
/[]/ ()
/E[]/ (E, E), /E/ (E>) /E<, </ (E<)
/[]/ (, ), /;/ ()
/u/ (%)
/U[]/ (, P), /U/ (>), /U</ (P<)
/, / (, *, x|, |x)
/:, :</ ('>:), // (>), /</ (<, >)
/[]/ (x, )
/O:/ (:, A:), /O:/ (:<), /O:</ (:<, :<)
/A:, / (A:), /;/ (:, A:),
/;</ (A:<, :<, :<)
2.4.2.3. 2.13 gives the diphthongs. Let us observe the dierences for /Ii, aE,
aO, OU, [j]Uu, Uu/ (i, , , , [j]%, )m;a (Ii, a, a, , Uu, ju, Uu)a:
('b;i)m;a ('bI;i)a /'bIi/ bee ('g;)m;a ('ga;)a /'gaE/ guy ('D;n)m;a ('Da;n)a /'daOn/
down ('n;s)m;a ('n;)a /'nOU/ no ('Th%;)m;a ('ThU;u)a /'tUu/ o ('n;*)m;a ('nU;u)a
/'nUu/ n[ ('fj;)m;a ('fj;u)a /'fjUu/ fuel
2.13. Mediatic American diphthongs.
/Ii[]/ (i, i)
( /I/}
/EI[]/ (I, )
/aO[]/ ([])
/jUu/ (j%), /[]Uu/ (%)

/[j/]Uu/ ([j/`]) ( /U/}
/OU[]/ ([]), /OE[]/ (, )
/aE[]/ (, )
2.4.2.4. As far as V are concerned, then, we have to pay attention to frequent

neutralizations (+ //), which however can present oscillations depending on words
or speakers. In extreme cases, which are not at all rare indeed, we can nd: ('f;)m;a
both for ('f:)a /'fI/ ll and ('fi;I, 'fiI)a /'fIi/ feel ('w;)m;a both for ('wE:)a /'wE/
well and ('w;, 'w)a /'wEI/ wale ('v;)m;a both for ('v;)a /'v/ Val and ('va;,
'va)a /'vaO/ vowel also ('fP;)m;a both for ('f:)a /'fU/ full and ('fu;U, 'fuU)a
/'fUu/ fool and ('g;)m;a both for ('g,:)a /'g/ gull and ('g;)a /'gOU/ goal. Two
100
further variants are fairly frequent, which produce: /I/ (i), /A:=, A:0/ (A<):
('khi:, 'sii)m;a for ('kh:, 's)a /'kI, 'sII/ ki sii and ('phA<k, 'khA<m>,
'DA<]i)m;a for ('phA;<k, 'khA;<m, 'DA;<l)a /'pA:k, 'kA:/ park carter, darli.
2.14. Mediatic American neutralizations (and two further possible variants).
/[']I/ (i)
/I, ii/ (;)
/U, uu/ (;)
/E, EI/ (;)
/, OU/ (;)
/, aO/ (;)
/A:/+/0, =/ = // (A<)
2.4.2.5. As far as C are concerned, keeping in mind that these observations are
general (and not absolutely obligatory for all speakers words) and that our
transcriptions are normalized, let us say that for /l[j]/ we have (][j]): (']T)m;a
('lT)a /'lIt/ lilt (']:)m;a ('l,:)a /'l/ lull ('mK]j'n)m;a ('mjn)a /'mIljn/ million
('nvsT)m;a ('nAvsT)a /'nvst/ noveli
Besides, /</ has a uvularized velar rounded articulation, which is darker (or
harder): ('<;>)m;a ('<I;)a /'<I/ rear ('<:<)m;a ('<:<)a /'<O:</ roar ('wn)m;a ('wn)a
/'wIn/ winner Please note that generally (<) exerts on // the same retracting and
raising eect of /k, g, , w/: (Th'<;In)m;a (Th'<;In)a /t'<EIn/ to rain.
As we have seen, a N nasalizes the following V It often also nasalizes the preceding V (as we indicate). Even intense C in contact are nasalized. en we nd syllables with /{}/ + /mp, nt, k/, we very frequently have ({}) + (Mp, Mp)
(T, T, ) (k, k) (which are too often described simply as (0) la franaise): ('khMp)m;a ('khmp)a /'kmp/ camp ('sTMp)m;a ('sTmp)a /'stmp/ ump
('kh, -T)m;a ('kh[;]nT)a /'k;nt/ can't ('Ds, -T)m;a ('DnT)a /'dOUnt/
don't ('Ik)m;a ('k)a /'Ik/ ìnk ('D<k)m;a ('D<k)a /'d<k/ drank
// behaves as in neutral American pronunciation, but often this use spreads to
/d/ as well, so that (']m>)m;a can represent either ('lm)a /'l/ latter or ('lD)a
/'ld/ ladder (&n'T;n, &nm>-)m;a (&nD'T;n)a /nd'stnd/ underand
e same goes for // = (`), as seen in n[ and it is also possible to have /w/:
2.15. Mediatic American intonation.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
2. english
101
('wn:, 'W-, 'hw-)m;a ('wn: 'W- 'hw-)a /'wEn/ when

ere is a tendency to slightly shorten stressed nal vowels and diphthongs,
which we will only mark here, because they oscillate (by indicating /O:, A:/ as doublings, ()m;a, for (;)a): ('si) /'sIi/ see ('Th%) /'tUu/ o ('DI) /'dEI/ dZ ('g)
/'gOU/ go ('h) /'haE/ high ('n) /'naO/ now ('b) /'bOE/ boy ('spAA) /'spA:/ spa
('s) /'sO:/ sY (also ('sA, 'sAA)).
Mediatic British English pronunciation
2.4.3.1. Journalists love to call it Estuary English, in reference to the ames
estuary, but it is not limited to this area, since from the south-eastern coasts of
England it surely stretches to Cambridge, Oxford, and Southampton, too. Of
course, the direct inuence of London is real, especially on the new towns (such
as Milton Keynes /'mIt 'kIinz/, in northern Buckinghamshire, 1967), which have
been built since 1946, each one planned as an autonomous whole (with factories,
houses, shops, ), in order to decentralize masses of populations, particularly
from London.
However, in general, the South-East has always shared to a lesser or greater extent the London-type pronunciation characteristics. us, this accent rather than
spreading has been emerging more and more, with the actual recognition of
its existence.
Many speakers nd that this accent is more genuine and authentic, in comparison with traditional rp (Received Pronunciation), which used to be the symbol of the prestigious and expensive and denitely private Public Schools,
such as Eton, Harrow, and nchester. /'A: 'pIi/ is also known as bbc English, because it was used by the bbc right from the beginning (1927 {and television, 1932}). But today it can be mainly heard only on international transmissions
of the bbc World Service radio broadcastings and the bbc World television
broadcastings, since most English people who do not use it nd it to be too
aected and litist.
2.16. Mediatic British monophthongs.
/i/ = /Ii/ (I)
/i/ (i), /I[]/ (, P[])
// (, *, a|)
// = //, /:/ ()
/E[]/ (, P[])
/[]/ (E, P[])
/u/ (%), /U[]/ (, UP[])

/[j]U, [j]U</ = /[j]O:, -</
([j]oo0, -oo<, [j]o[/])
/O:/ (oo0, o[/])
// (O)
/, |/ (a), // (P)
/A:/ (A)
erefore, foreigners must be familiar with the mediatic British accent too, but
with no real need to actually use it. However, often, the native speakers' conception of this accent is subjective and contrasting. In fact, if on the one hand they
may consider it to be neutral, or almost neutral, with respect to common peo-
102
ple, when public people are concerned as politicians are the same pronunciation can be dened as Cockney, that is very dialectal (and hardly appropriate).
But we do know that linguistic opinions are very colorful and personal.
2.4.3.2. As can be seen from 2.16, /, , O:/ are higher, (E, O, oo). e last one
is doubled (since it moves upwards slightly), or diphthongized (downwards) when
lexeme nal, or followed by the /z, d/ (o[/]) grammemes. Also /E/ is higher,
though remaining within its own box, () (to be better distinguished from // (E)):
('[h]E)m;b ('hT)b /'ht/ hat ('<Ok)m;b ('>k)b /'<k/ ro$ ('lo;on)m;b ('l:n)b /'lO:n/
lYn ('lo;)m;b ('l:)b /'lO:/ lY ('mo;)m;b ('m:)b /'mO:/ more ('lo;)m;b ('l:)b
/'lO:z/ lYs ('pho;)m;b ('ph:)b /'pO:d/ pored In the vocogram, the diphthongs due
to the vocalization of // are marked in grey: /I, E, , , U/. Note the dierent
symbols: ('b;P[])m;b ('b:)b /'bI/ bill ('b;P[])m;b ('b:)b /'bE/ bell (';P[])m;b (';)b
/'/ Al ('khP[])m;b ('khT)b /'kt/ cult, ('bU;P[])m;b ('b:)b /'bU/ bull.
ose who systematically realize // as () are inuenced by a sort of strategy
in reaction to the mediatic (and Cockney proper) closer articulation, which is realized as (E), but it does not belong to neutral pronunciation. In addition, it can
cause confusion with //, not neutralization (as some think), since these realizations are uttered by dierent speakers, not the same, although in the same places.
Besides, /u/ is fronter, (%): (&sc%'ISn)m;b (&sc'IS)b /sIcu'EISn/ situation;
/, []|/ are lower (and fronter), (a): ('baa)m;b ('bT)b /'b/ butter ('b;a)m;b
('b;)b /'bI/ beer; whereas, /:, A:{}/ are diphthongized, (, A): ('f;[])m;b
('f:[])b /'f:[z]/ fur[s] ('khA;[])m;b ('khA:[])b /'kA:[z]/ car[s]
2.4.3.3. 2.17 gives us the positions of /I, I< E, E</. We have just seen
/I|/ (beer] instead, for /I</ (and /I<, I/ not before pauses) more often we have
(:): ('[h];<n, --)m;b ('h>)b /'hI<I/ heari ('[h]:< n':, --, -;)m;b ('h;> n';)b /'hI< n[d]'E/ here and ère ('[h]: 'khamz 'san:, '[h];, &[h];'-, &[h]'-)m;b
('h; 'khmz 'sn:, &h'-)b /'hI 'kmz 'sn/ here comes è sun. It is to be noted
that even /E<, E/ are more often long monophthongs than narrow diphthongs:
('m;<I, 'm-)m;b ('m>i)b /'mE<i/ Mary (beside ère, just seen); /U/ practically
becomes /O:/: ('pho;)m;b ('ph:, 'ph;)b /'pU/ poor
Generally, in this accent, for // sequences we have (, a|): ('sIa)m;b
('sIi)b /'sIi/ seer (and also (A'DIa)m;b (a'D;)b /aE'dI/ idea), ('DT%a)m;b ('Du)b
2.17. Mediatic British // diphthongs.
/I, I</ (, :[], a|, :<)
/I[z/d]/ = /Ii/ (i, ia|)
/E, E</ (:[<], [<])
/[j]U, -U</
= /[j]O:, -</ ([j]oo0, -o[/])
/aE/ (A, Aa|), /aO/ (P, Pa|), /EI/ (I, Ia|), /OU/ (x, xa|), /OE/ (o, oa|),
/Ii/ (I, Ia|, I<), /Uu/ (T%, T%a|, T%<), /A:/ (A) + /A:/ (A, Aa|), /</ ([]<)
2. english
103
/'dUu/ doer In addition: ('phlIa)m;b ('phlI)b /'plEI/ plZer ('fAa)m;b ('fa)b

/'faE/ re ('hPa)m;b ('Tha)b /'taO/ tower and even: ('gxa)m;b ('g)b
/'gOU/ goer (m'phloa)m;b (m'phl)b /Im'plOE/ employer (all with /-z/ (-)).
For ('fA;)m;b ('fA:)b /'fA:/ far ('khA;)m;b ('khA:)b /'kA:z/ cars it is also possible to have /'fA:, 'kA:z/ ('fAa, 'khA)m;b.
2.4.3.4. 2.18 shows that there is a more retracted rst element for /Ii, Ii, aE/,
(I, I, A): ('b;I)m;b ('bI;i)b /'bIi/ bee ('hA;m)m;b ('Tha;m)b /'taEm/ time. e
rst element is lower and backer for /EI, OU/ (this last one is unrounded too, when
not followed by //, as in neutral British pronunciation), (I, x): ('D;I)m;b ('D;I)b
/'dEI/ dZ ('gx;)m;b ('g;)b /'gOU/ go (('gO;U[])m;b ('g;)b /'gOU/ goal]. e rst element is fronter for /aO, Uu/, (P, T%) (besides, /Uu/ is (u[])): ('h;Pn)m;b
('Tha;n)b /'taOn/ town ('hT;%)m;b ('Th;u)b /'tUu/ o ('chT;%)m;b ('Thj;u)b
/'tUub/ tube ('fj;u[])m;b ('fju;U)b /'fjUu/ fuel whereas, the rst element is higher for /OE/ (o): ('bo;)m;b ('b;)b /'bOE/ boy
For the diphthongs with front second elements, there are some remarkable
triphthongs which derive from the vocalization of // (and realized with or without ()); /Ii, EI, aE, OE/ (IP[], P[], AP[], oP[]): ('[h]IP[])m;b ('hi;I, 'hiI)b
/'hIi/ heel ('<IP[])m;b ('>;, '>)b /'<EI/ rail ('fAP[])m;b ('fa;, 'fa)b /'faE/ le
('oP[])m;b (';, ')b /'OE/ oil.
In [n]eièr the American-like pronunciation prevails: ('[n]Ia -A-)m;b ('[n]a -Ii-)b. Generally, /i/ is realized as if it were /Ii/ (and in the prex /0I-/ as well):
(<I'Ek)m;b (>i'kT)b /<i'kt/ react ('sI)m;b ('sTi)b /'sIi/ ci (<I'gA;)m;b (>'gA:)b /<'gA:d/ regard For /[0]/ () is more frequent (thus, we have an extension of what happens in neutral pronunciation for /[z/d]/): ('lp)m;b ('lp)b
/'lEpd/ leopard
Seeing that native phoneticians continue to overlook phonetic particulars, we
add some sociolinguistic variants for several phonemes (drawing from our archives
for the description of regional accents) for now without adding gures, and within the limits of the seven most typical diphthongs.
As regards what we have said above, we also provide the transcription of both
the most marked variants (broader, ()), and the least marked ones (more
rened, that is more controlled, ()): /Ii/ (I, i, i), /EI/ (I, , EI), /aE/ (A,
, I), /aO/ (P, P), /OE/ (o, ), /OU/ (x, , T), /Uu/ (T%, %, ).
ose who are really interested in them will quite easily manage to draw useful
comparisons, even without specic vocograms (however they can be found in
English Pronunciation*]
2.18. Mediatic British diphthongs.
/Ii[]/ (I, IP[])
/[j]Uu[]/ ([j]T%, [j]u[])
/EI[]/ (I, P[])
/OE[]/ (o, oP[])

/OU[]/ (x, OU[])
/aO[]/ (P[])
/aE[]/ (A, AP[])
104
2.19. Mediatic British neutralizations.
/I[i]/ (P[])
/-, -/ (P[])
/, EI, aO, / (aP[])
/[j]U[u]/ ([j]U[]) { /[j]O:/ ([j]oU[])}

/O:/ (oU[])
/, OU/ (OU[]), /OUl/ (xl)
{in derivatives, also (OUl)}
2.4.3.5. Even for mediatic British English, neutralizations are frequent before // (although less frequent than in Cockney, the typical and popular and less
educated dialect and accent of the East End of London). In fact, 2.19, we often nd ('f;P[])m;b both for ('f:)b /'fI/ ll and ('fi;I)b /'fIi/ feel in addition,
('[h]a;P[])m;b both for ('h;)b /'h/ Hal and ('h;)b /'hEI/ hail and both for
('ha;)b /'haO/ howl and ('h:)b /'h/ Hull, as well. Besides, we nd ('DO;U[])m;b
both for ('D;)b /'d/ doll and ('D;)b /'dOU/ dole and ('f;U[])m;b both for
('f:)b /'fU/ full and ('fu;U)b /'fUu/ fool (the last one is possible for ('fo;U[])m;b
('f:)b /'fO:/ fall too] (We will deal with () shortly.)
2.4.3.6. For the consonants, the strong preglottalization of /p, k, c/ must not
be forgotten as it practically occurs in all the cases indicated in 2.2.6.1-2
2.2.7.1; in addition, it is important to mention the massive substitution of /t/ with
(), in all the cases indicated in 2.2.5.2; whereas we can say that, generally, ()
replaces (T) /t, / of neutral pronunciation. erefore: ('D<Op)m;b ('>p)b /'d<p/
drop ('D<Op)m;b ('>pT)b /'d<pt/ dropped ('phEk)m;b ('phk)b /'pk/ pa$
('phEk)m;b ('phkT)b /'pkt/ pa$ed ('<Ic, -S)m;b ('>Iic, -S)b /'<Iic/ rea>
('<Ic, -S)m;b ('>IicT, -ST)b /'<Iict/ rea>ed ('<Aa)m;b ('>aT)b /'<aE/ writer
('f&boU[])m;b ('fp&b;, --, -T-)b /'fUtbO:/ football ('g 'D;Pn)m;b ('gT 'Da;n, -)b
/'gEt 'daOn/ get down ('lIlI)m;b ('lITli, -li)b /'lEItli/ lateI ('hA)m;b ('ThA;T)b
/'tA:t/ tart ('hsa, -a)m;b ('ThsT)b /'tEst/ tester
e preglottalization of /p, t, k/ (and of /c/, as in neutral British English) also
occurs between V both within words and in sentences: ('phpa)m;b ('php)b
/'pEp/ pepper ('ba)m;b ('bT)b /'bE/ better ('bIka)m;b ('bIk)b /'bEIk/
baker ('l, -)m;b ('lTT, -)b /'lEIt/ let it In the /st, sc/ sequences, there is the
possible variant /s/ () (by further assimilating to the following contoid, which is
typical of broader accents): ('sOp, '-)m;b ('sTp)b /'stp/ op ('s<;I~, '<-,
'>-)m;b ('s>;I~)b /'st<EInG/ rae ('khwscn, -c-, -cn, -Scn)m;b ('khwsc,
-Sc)b /'kwEscn, -Sc-/ queion
For //, ([) can occur, too: ('ph<I, 'ph<[I 'hwnI, -n[I)m;b ('ph>Ti,
'ThwnTi)b /'p<Ii, 'twEni/ pret en Even () (in addition to (n, -)):
('khln, 'P[], 'b)m;b ('khlnT, 'T, 'b;T)b /'klIntn, 'Etn, 'b:tn/ Clinton Elton Burton
Besides, we nd (n) for the grammeme >-i: ('<xln, '<OU-)m;b ('>l)b /'<OUlI/ rolli this example also shows the oscillation for /OU/ before /l/, under the in-
2. english
105
uence of /OU/ (OU[])m;b ()b, whereas in non-derivatives the normal realization

occurs: ('phxla)m;b ('phl)b /'pOUl/ polar It is possible to have /k/ in: ('samk, 'nI&k)m;b ('sm, 'ni&)b /'smI, 'EniI/ someì aJì.
From the examples, one can notice that the most frequent realization of /</ is not
postalveolar rounded ((>), as in neutral pronunciation), but (postalveolarized) prevelar rounded ((<), corresponding to the neutral American articulation). But there
are also four further quite frequent variants: the labiodental, (V), and its combinations with other articulations, labiodental rounded, (), velarized labiodental, (),
and postalveolarized labiodental, (S): (<I'<;I~, VI'V-, I'-, I'-, SI'S-)m;b
(>i'>;I~)b /<i'<EInG/ rearrae. In the case of /t<, d</, a realization which corresponds to the neutral articulation is possible: ('h<;In, -V-, --, --, -S-, 'h>-)m;b
('h>;In)b /'t<EIn/ train ('D<k, -V-, --, --, -S-, '">-)m;b ('>k)b /'d<Ik/ drink
Very frequently, a non-etymological // is inserted: (A'so<)m;b (a's;T)b
/aE'sO:It/ I sY it, even in cases like: (A&xv<';, -'h;)m;b (a&v'h:)b /aEOUv'h:d/ I overheard.
2.4.3.7. Our examples have already shown that one of the most evident characteristics, which is socially stigmatized, is the zero realization of /h/ (= (`)):
('[h]A;)m;b ('ha;)b /'haE/ high. Consequently, even hypercorrections are frequent,
as happens with the name of the letter h (not without a certain internal logic):
('[h]Ic)m;b ('Ic)b /'EIc/. On the other hand, the reduced form of him is less frequent: ('hP[]hm)m;b ('Thlm)b /'tElIm/ tell him equally for the reduced form
of a\ (I'mE;n)m;b ('m;n)b /'mn/ a man Also for /0n/, a less reduced form is
more frequent: (ph<&nansi'ISn, 'GEksn)m;b (ph>&nnsi'IS, 'Gks)b /p<nnsi'EISn, 'Gksn/ pronunciation Ja$son
e sequences /t, d, n/ have the peculiarity of typically corresponding to /c,
G/ and (~): ('chT;%n)m;b ('Thj;un)b /'tUun/ tune ('GT;%n)m;b ('Dj;un)b /'dUun/
dune ('~T;%)m;b ('nj;u)b /'nUu/ n[ (in Cockney we actually nd ('n;%) = /'nUu/).
Occasionally, /, / can become /f/ and /d, v/ respectively (which is a typical
Cockney pronunciation, and can be heard even on the borders of the Estuary
area {and in further mostly metropolitan areas, which have been inuenced by
this accent}): ('<;I, 'f<-)m;b ('>I;i)b /'<Ii/ `ree ('s, 'D-)m;b ('s)b /'Is/ ìs ('maa,
-va)m;b ('m)b /'m/ moèr.
2.4.3.8. ere is nothing to say about /l/, while for // there is a typical vocalization of (, ) which become (P) ( 2.19), after V or C as many examples have
already shown (particularly those of the neutralization before //). e lateralized
velar rounded approximant () (in brackets in the gure) indicates a less marked
pronunciation, where the lateralized contoid is added to the vocalization, (P), in
order to try to attenuate this characteristic, which is often socially stigmatized. So,
we can nd (), without lip-rounding, as a kind of halfway compromise.
Here are some examples, to complete the survey: ('lP[])m;b ('lT)b /'lIt/ little ('phIpP[])m;b ('phIip)b /'pIip/ people ('DvP[])m;b ('Dv)b /'dEvz/ devils
('mP[]k)m;b ('mk)b /'mIk/ milk ('woP[]a)m;b ('w;T)b /'wO:/ Walter
('GnP[]mn)m;b ('GnTmn)b /'GEnmn/ gentleman. nally, (&akPl'-
106
b, -P'h-)m;b (&k'h;bT)b /'k 'h:bt/ uncle Herbert where we can see the
normal kind of prevocalic l ((l)), and the intense one, again prevocalic: ('nOvs,
-)m;b ('nvsT)b /'nvIst/ noveli
2.4.3.9. In cases like the following, we often nd secondary stresses: ('DkS&n<I, 'sm&<I, 'Doom&o<I)m;b ('DkS>i, 'sm>i, 'D;m>i)b /'dIkSn<i, 'sEmt<i, 'dO:mtO:<i/ dictionary cemetery dormitory.
Regarding intonation, we have to say that the pitch of the rst stressed syllable
and of the following internal unstressed syllables in a preintoneme is less high than
in the neutral pronunciation, as can be seen from 2.20. e suspensive intoneme is more similar to the American one; and, often, the interrogative intoneme,
besides being as in neutral British English, can be rising-falling (again in 2.20,
where we show only the dierent type): (GspIk 'glS 'w;P[]12)m;b (Dj5spIik
'glS 'w:21, G-)b Do you speak Eli well?
2.20. Mediatic British intonation.
/ / (2 2 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 2 2 ' 2 2 ' 2)
/?/ (2 ' 1 2)
/ / ( 2 2 2 2 ' 2 2 ' 2)
// (2 2 2)
/ / ( 2 2 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
Text
2.5.0. e story e North Wind and the Sun (by Aesop) follows. It is given in
ve dierent (normalized) versions. In fact, they systematically and coherently
present the most typical characteristics, which are acknowledged as peculiar. We
start with the American and British versions in (neutral) English, which is the rst
step of the phonetic method. e international version follows together with the
two mediatic versions of American and British English.
In the other chapters of the book, for each language dealt with, at least two kinds
of foreign pronunciations are given: rst the foreign pronunciation of English, and
lastly the British English pronunciation of the foreign language in question, according to the same principles. e speakers are supposed to be neutral speakers of their
own language, uent in English (after prolonged contact with native speakers, but
with no help from the phonetic method), who have adequately learned the relative prominences, but who substantially use segments (vowels consonants) and
intonation elements, which are typical of neutral English (although, of course, a
neutral accent is not so common). Obviously, the same principle is valid for the
foreign pronunciations of English, given rst. Sometimes further accents have been
added as can be seen in the correspondent chapters.
2. english
107
Graphemic text
2.5.2.0. e Nor` Wind and è Sun were disputi whi> was è roer,
when a travel(l)er came alo wrapped in a warm cloak. \ agreed àt è one
who r succeeded in maki è travel(l)er take his cloak o ould be considered
roer àn è oèr.
en è Nor` Wind bl[ as hard as he could, but è more he bl[ è more
closeI did è travel(l)er fold his cloak around him; and at la è Nor` Wind gave
up è attempt. en è Sun one out warmI, and immediateI è travel(l)er took
o his cloak. And so è Nor` Wind was obliged to confe àt è Sun was è roer of è o.
Did you like è ory? Do you want to hear it again?
Neutral American pronunciation
2.5.2.1. ('n;< 'wn:D2 n'sn:2 wD'spju[ 'wc wz'sT<O;g23| 'wn 'Th<vl2 'kh;Im 'lO;2 '<pT n'w:<m 'khlk23| I'g<I;iD2\ t'wn 'f;s sk'sIiDD2 m'mIk 'Th<vl2 'ThIk 'khlk 6O;f2 2| &Sbbikn'sD "sT<O;g ni'23||
'n:2 'n;< 'wm:2 'blU;u2 'hA:<D2 zi'kh;23| bt'm:< i6blU;u2 2| 'm:<
'khlsli2 &Dd'Th<vl2\ 'f;D 'khlk '<anDm23| n[D]'l;sT2\ 'n;< 'w:2
'g;Iv 'p i'ThmpT23|| 'n:2 'sn 'S;n 'aT23 'w;<mli23| n'mIiDiTli2\ 'Th<vl
'Thk 'O;f23 'khlk23|| 's;2 'n;< 'wn:D2\ wz'bla; TkM'fs2| t'sn:23
w'sT<O;g23 v'ThU;u23||
&DGlak21 'sT;<i2| Dj'wAn [[]'h<[ gn:21|||)
Neutral British pronunciation
2.5.2.2. (5n; 'wn:D2 n'sn:2 wD5spjuT 'wc wz's>g3 3| 5wn 'h>vl2 5kh;Im 'l2 5>pT n'w:m 'khlk3 3| I'g>I;iD2\ t5wn 'f;s
sk'sIiDD2 m5mIk 'h>vl2 5ThIk 'khlk 'f32| &Sbbikn5sD "s>g
ni'3 3||
'n:2 5n; 'wm:2 5bl;u2 'hA:D2 zi'kh;3 3| bt5m:> i'bl;u32| 5m:
'khlsli2 &Dd'h>vl2\ 5f;D 'khlk '>anDm3 3| n[D]'lA;sT2\ 5n; 'w:2
5g;Iv 'p i'ThmpT3 3|| 'n:2 5sn 'Sn 'aT3 3 'w;mli3 3| n'mIiDiTli2\ 5h>vl
'Thk 'f3 3 'khlk3 3|| 's;2 5n; 'wn:D2 wz5bla; TkM'fs2| t'sn:3 3
w's>g>3 3 v'Th;u3 3||
&DG'lak21 'sT;>i2| Dj5wn T'h>T 'gn:21|||)
108
International English pronunciation

2.5.2.3. ('n; 'wn:D2 n'sn:2 wD'spjuuT 'wc wz'sc<g23| 'wn 'ch<vl2 'kh;m 'l;2 '<pT n'w:m 'khlk23| 'g<i;iD2\ t'wn u'f;s
sk'siiDD2 m'mk 'ch<vl2 'Thk 'khlk 'f2 2| &Sbbikn'sD "sc<g
ni'23||
'n:2 'n; 'wm:2 'blu;u2 'hA:D2 zi'kh;23| bt'm:< i'blu;u22| 'm: 'khlsli2 &Dd'ch<vl2\ 'f;D 'khlk '<anDm23| n[D]'lsT2\ 'n; 'w:2 'g;v
'p i'ThmpT23|| 'n:2 'sn 'S;n 'aT23 'w;mli23| n'miiDiTli2\ 'ch<vl 'Thk
'f23 'khlk23|| 's;2 'n; 'wn:D2 wz'bla; TkM'fs2| t'sn:23 w'sc<g<23 v'Thu;u23||
&DG'lak21 'sT;<i2| Dju'wn T'h<T 'gn:21|||)
Mediatic American pronunciation
2.5.2.4. (''n9;< 'wn:D2 'n'sn:2 w>D'spj%[I 'wc wz'sT<;g>23| 'wn ''Th<v]>2 'kh;Im '];2 '<pT n'w:<m 'kh]k23| I'g<;iD2\ t'wn %'f>;s
sk'si[D2 m'mIkI 'Th<v]>2 'ThIk 'kh]k ;f2 2| &SPbbikn'sm> "sT<;g>
'ni'x>23||
'n:2 ''n9;< 'wm:2 'b]%; 'hA:<D2 zi'kh;23| bt''m9:< ib]%;2 2| ''m9:<
'kh]s]i2 &Dd'Th<v]>2\ 'f;D 'kh]k '<nDm23| 'n[D]']sT2\ ''n9;<
'w:2 'g;Iv 'xp i'ThmpT23|| 'n:2 'sn 'S;sn 'T23 'w;<m]i23| 'n'mi[iT]i2\
'Th<v]> 'Thk ';f23 'kh]k23|| 's;2 ''n9;< 'wn:D2 wz'b]; TkM'fEs2|
t'sn:23 w'sT<;g>23 v'Th%;23||
&DG]k21 'sT;<i2| Dj'wn ''h<[ gn:21|||)
Mediatic British pronunciation
2.5.2.5. (noo 'wn:D2 n'san:2 wDspjT% 'wc wz'<Oga3 3|
wn 'h<Evl2 kh;Im 'lO2 <Ep n'wo;om 'khlxk3 3| I'g<;ID2\ wan %'fs sk'sIDD2 mmIk 'h<Evl2 hIk 'khlxk 'Of32| &SbbiknsD "<Og ni'aa3 3||
'n:2 noo 'wm:2 blT;%2 '[h]A;D2 zi'kh;3 3| bmo;< i'blT;%32| mo;
'khlxslI2 &Dd'h<Evla2\ fO;UD 'khlxk '<PnDm3 3| n[D]'lA2\ noo
'w:2 g;Iv 'ap i'hmp3 3|| 'n:2 san 'SOn 'P3 3 'woomlI3 3| n'mIGlI2\
h<Evl 'hk 'Of3 3 h'khlxk3 3|| 'sx;2 noo 'wn:D2 wzblA; kM'fs2| 'san:3 3 w'<Og<3 3 v'hT;%3 3||
&DG'lAk12 'oo<I2| G%wOn 'h;< 'gn:12|||)
2. english
109
Appendix: further accents

2.6.0. Concisely, we now provide the phonosyntheses of six accents. e practice had with the ve preceding accents ( the two neutral and two mediatic
American and British, and the international one) will surely allow us to identify
their peculiarities, starting from the diaphonemic transcriptions to see how they
are actually realized in the indicated areas. Obviously, if the readers cannot immediately nd some appropriate examples, they can use those given in the whole
chapter, depending on its sections. Clearly, the readers who are already familiar
with these other accents will more easily and spontaneously nd both the examples and the phonic values (together with connections and analogies). A more systematic treatment will be done in Eli Pronunciation* (in the bibliography); although a careful examination of the vocograms given here will certainly provide
more detailed information.
Canadian English pronunciation
2.6.1. ere are many peculiar unications of vowels, with the merging of /O:,
O;, , ;/ into (:) and of /O:, O:<, <, ;</ into (:<) (('s;<i) /'s<i/ sorry). In addition,
there are peculiar diphthongs with narrow taxophones of /EI, OU/ (eI, oU), and of
those of /aE, aO/ followed by a voiceless C within the word, (, ): ('nT, 'T)
/u/ (), /U/ (U)
/U</ (U<), /U[]/ (, U)
// = //, // (, *, |)
// (), /</ (<, ), /:, :</ (':)
/[]/ (, )
/O:, O:<, <, ;</ (:<)
/O:, O;, , ;/ (:)
/jUu/ (ju), /Uu/ (Uu)
/Uu/ (ju, Uu)
/[j/]Uu/ ([j/`]uU)
/OU/ (oU)
/OE[]/ (, )
/i, i/ (i), /I/ (I) >-ing

/I/ (I), /I</ (I<), /I[]/ (, )
/E[]/ (, ),
/E/ ()
/E</ (<),
/[]/ (, ), /</ (<), /;/ ()
/A:/ (A:), /A:<, A:/ (A:<)
/Ii[]/ (Ii, iI)
/EI[]/ (eI, e)
/aE[]/ (a, a, =, =)
/aO/ (a, =)
/ / (2 2 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / (2 2 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / (2 2 2 2 ' 2 2 ' 2)
// (2 ' 2 2)
/ / (2 2 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
110
Current and mediatic variants

/i|/ ([I]i), /I[]/ ([])
/I/ (In) >-ing
/E<, E</ (e<), /E/ (e;)
/[]/ (), /E[]/ (E, )
// (E) { (;))
/[]/ (, a)
/U[]/ (P[])
/O:/ (o:<, O:<)
/O:<, <, ;</ (o;<, O;<)
// (, , x)
/O:, O;, , ;/ (A:, A) { (:)}
/Uu/ (%, u) { (Uu)}
/Uu/ (u) { (Uu)}
/aE=/ (i)
/EI[]/ (e, )
/aO/ (, ), /aO=/ (x)
/OE[]/ (o, o)
/OU/ (P)
/Ii, EI, aE, aO, OE, OU, Uu/ + // = //
/'naEt, 'aOt/ night out. ere are neutral and non-neutral taxophones for many vowels and diphthongs + //. e neutral ones which change phones are: /, , U
Ii, EI, aE, OE, Uu/; the others are /I, E/ and // with other variants; in addition to a possible insertion of () before (). As far as phoneme distribution is concerned, certain words are pronounced with British vowel elements, others with
American ones. For the consonants, the use is similar to the American one.
In current and mediatic pronunciations (given in the second set of vocograms),
we nd that hypercorrection can produce /jUu/ for /Uu/ [noon too do), due to
the fact that, for /Uu/, careful speakers prefer /jUu/, after /n, t, d/. In addition, we
nd the nasalization of /, / (, ), even in // (above all for //);
lastly, for /l/, we have (]), in these kinds of pronunciations.
Australian English pronunciation

2.6.2. We present four dierent accents separately: neutral (cultivated Australian, in the rst three vocograms, which is used by a limited number of speakers,
who have learned it intentionally, as happens for all neutral accents).
e mediatic accent (general Australian, in the second series of three vocograms, typical of mass media and many speakers), the broad accent (broad Australian in the third series of three vocograms, typical of uneducated people, which
is heavily nasalized, too), and also the aected accent (modied Australian, in
the fourth series of three vocograms, used by a very small group of lite speakers,
who aim at imitating traditional or aected British pronunciation, which is considered to be too mannered and unacceptable).
For each accent, we will rst see the monophthongs (given in the rst vocogram), moving then to the diphthongs, and lastly to centering diphthongs (in the
third vocogram; in this case, the peculiarities of broad accents are quite evident).
For //, we regularly have //, except in aected pronunciations.
2. english
111
Neutral accent/Cultivated Australian E.

/i/ (i)
/u/ ()
/U/ (U)
/:/ (:), // (, *, |)
/O:/ (o:) {/O:|/ (o), /O:0/ (oo)}
/I[]/ (I, )
/E[]/ (e, )
/[]/ (E, )
/[]/ (, ), /A:/ (a:)
/Ii[]/
(i, i)
/EI[]/
(I, )
/aE[]/
(, )
/[]/ (, )
/Uu[]/
(, Uu)
/OE[]/
(o, o)
/OU[/l]/ (T,
OU, OUl)
/aO/ ()
/I[],
I</
(I[|])
/[j/]U[],
-U</
([j]U[|])
{ /[j]O:/}
/E,
E</
([|])
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / (2 2 ' 2 2 ' 2 2 ' 2)
// (2 ' 3 2)
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
Mediatic accent/General Australian E.

/i/ ()
/I[]/ (I, )
/U/ (U)
/:/ (:) // (, *, |)
/O:/ (o:)
/E[]/ (e, )
/[]/ (E, )
/[]/ (O, )
/[]/ (, ) /A:/ (a:)

/i/ (, )
/Ii[]/ (, )
/EI[]/ (, )
/aO/ ()
/Uu[]/ (T%,
, T, u)
/OE[]/
(o, o)
/OU[/l]/
(T, OU, OUl)
/aE[]/ (A, A)
/I[],
I</
(I[|])
/E,
E</
(e[|])
/[j/]U[],
-U</
([j]U[|])
{ /[j/]O:/}
e most typical characteristic (similar to mediatic British and Cockney pronunciations) consists in a wider pronunciation of the diphthongs /Ii, EI, aE, OE, aO,
OU, Uu/, as can be seen from the respective vocograms (in addition to (I, e) for /I,
E/, since, only in aected pronunciations, can we nd () for /E/). e diapho-
112
neme // follows British use (although some young Australians, especially females,
who live abroad, present a uctuating and non-neutral use of (>)). Let us notice
(and very well too) the various taxophones + //.
We systematically nd /OUl/ (whereas in Cockney and in mediatic British
English minimal pairs occur such as /Ul/ /OUl/, as polar] Occasionally, we
can have /O:/, for /O:/ (either /Or/, or /or/). Except in neutral pronunciation,
for the /Ii, EI, aE, OE/ diphthongs, we have (, ); besides, /{}/ ({},
{}), without vocalizations (with (, ), after phones with labial component).
e diaphoneme // has ([), as in American English. For /0, 0/, neutral
pronunciation has (0, 0); thus, it has (l), without vocalizations (which are
present, though, in broad pronunciations). It is possible to hear a non-neutral pronunciation with /l/ (]), for l + -y -ie -i -er (grammemes {or even pseudo-grammemes} added to //).
Broad accent/Broad Australian E.
/i/ ()
/I[]/ (I, )
/E[]/ (e, )
/u/ ()
/U/ (U)
/O:/ (o:)
/:/ (:), // (, *, |)
/[]/ (E, )
/[]/ (O, )
/[]/ (, ), /A:/ (:)

/I[], I</
/Uu[]/
=
/Ii/ ()+
(%, )
(>, 0, |)
/OE[]/
/E, E</
(o, o)
(e:[|])
/OU[/l]/
(T, OU, OUl)
/aE[]/ (, )
/i/ (, )
/Ii[]/ (, )
/aO/ (E)
/EI[]/ (, )
/U[, -<]/
=/Uu/ (%)
+(>, 0, |)
{ /O:[]/}
Aected accent/Modied Australian E.

/u/ (u), /U/ (U)
/U/ (U), /:/ (:), // (, *, |)
/i/ (i)
/I[]/ (I, )
f :0, o0)
/O:/ (: :,
/[]/ (, )
/A:/ (a:)
/Ii[]/
(Ii, iI)
/EI[]/
(EI, E)
/aO/ (ax)
/Uu[]/
(u, Uu)
/OU[/l]/ (T,
U, Ul)
/OE[]/
(, )
/aE[]/ (a, a)
/[]/ (, )
/[]/ (, )
f
/E[]/ (, )
/I[], I</
(I>, I0, I|)
/E, E</
(>,
0, |)
/[j]U[]/
([j]P>,
P0, P|)
{ /[j]O:/}
2. english
113
New Zealand English pronunciation

2.6.3. In addition to the neutral accent, with three vocograms (the rst three),
we present the mediatic accent (whose centering diphthongs {fourth and fth vocograms} correspond to the neutral ones) and the broad accent (whose monophthongs {sixth and seventh vocograms} correspond to the mediatic ones, whereas
the centering diphthongs are peculiar, including the unication of /E, E</ with
/I, I</, and the possible unication of /U, U</ with /O:, O:</). us, for /E</
(e>, e;>, e>) (even in less neutral speech, the timbre is always (e): (e>, e)). For /I</
we have (> >, ); whereas, for /I</, (I>, iI); for idea\ /aE'dIi/.
/</ (E> >) is always distinct (in all accents). Occasionally, we can nd /O:/,
for /O:/ (both /Or/, and /or/). e triphthongs are not attenuated into diphthongs. Even in neutral pronunciation, we have /I/ () (not (, )). In New
Zealand, /;, A;/ = /A:/, /;, O;/ = //, // = /, I/ ( /I/ is only an intentional choice);
ni /'fInIS/ ('fnS 'fnS), visit /'vIzt/ ('vzT, 'vzT).
e most typical characteristic consists in (, e) for /I, E/, in neutral pronunciation (but (, ), in the other accents), in addition to the pronunciation of the diphthongs /Ii, EI, aE, OE, aO, OU, Uu/ (which is similar to the Australian, mediatic British, and Cockney accents), as can be seen from the respective vocograms. Besides,
/:<, :/ is rounded, (:) (or only partially so, (+, +|), in the other pronunciation
kinds).
/i/ (i)
/u/ ()
/U[]/ (, U)
/O:/ (o:)
/I[], I/ (, , )
/E[]/ (e, )
/:/ (:), // (, *, |)
/[]/ (, )
/[]/ (O, )
/[]/ (E, )
/A:/ (a:)
/Ii[]/
(Ii, iI)
/EI[]/
(EI, E)
/aO/ (aP)
/Uu[]/
(, uu)
/OE[]/ (I, )
/OU[/l]/ (,
U, Ul)
/aE[]/ (a, a)
/I, I</
(I, I|)
/U, U</
(U, U|)
/O:, O:</
(o, o|)
/E, E</
(e, e|)
/ / (2 2 2 2 ' 2 2 ' 2)
/./ (2 ' 3 3)
/ / (2 2 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / (2 2 2 2 ' 2 2 ' 2)
// (2 2 2)
/ / (2 2 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
114
Mediatic-accent variants
/i/ (Ii) { /Ii/ (i, i)}
/I[]/ (, )
/E[]/ (, []0, )
/:/ (+, +|)
/[]/ (, []0, )
/A:/ (a)
/Ii[]/ (i, i)
/u/ (%)
/U[]/ (w, U)
// (, *, | )
/O:[]/ (o, oU)
/[]/ (O,
O)
#
/[]/ (, , )
/Uu[]/ (T%, Uu)

/OE[]/ (o, o)
/EI[]/ (I, )
/aO/ ()
/OU[/l]/ (@T, OU, OUl)

/aE[]/ (A, A)
Broad-accent variants
/Ii[]/ (i, i)
/Uu[]/ (%, u)
/OE[]/ (PI, P)
/OU[]/ (x%, U, Ul)
/aO/ (Ex)
/EI[]/ (a, a)
/I, I</
/E, E</ (iI, iI|)
{ /I, I</
/E, E</}
/aE[]/ (, )
/U, U</ (uU, uU|)
{ /U, U</
/O:, O:</ /O/
(o;, o;|)}
e diaphoneme // is distributed like in British English (although in the southern and rural part of the South Island, // is not silent, as in American pronunciation). On the other hand, even in non-neutral pronunciations of other New-Zealand areas, // can be pronounced in words mostly in monosyllables ending in r.
Neutral /</ is (>) (while non-neutral accents can even have ()); as in British English, /t<, d</ can be realized as ([h]>, ">), in addition to normal ([h]>, >) (whereas the other accents can also have (c[h]>, G>) and (c[h], G), including /st</ (Sc>,
Sc)). In non-neutral pronunciation, often /t, d/ = (c, G).
e neutral accent can have /hw/ (W), still rather extensively used, even in current pronunciation; besides, it has (l); whereas, non-neutral pronunciation has
(]) and also (, , ). Except in neutral pronunciation, for the diphthongs /Ii, EI, aE, OE/, we nd /, /; unless they are vocalized as (). For
/0, 0/, the neutral accent has ((0, , 0)); the other accents have (0,
0, 0 0P): milk /'mIk/ ('mk 'mk 'mk 'mk).
2. english
115
We systematically nd /OUl/ (whereas in Cockney and in mediatic British

English there are minimal pairs with /Ul/ /OUl/, as in polar). e reader is
invited to carefully observe the numerous taxophones + //, both neutral and non-neutral.
In non-neutral pronunciation, there are more or less regular vowel neutralizations mergers (realized as in the given vocograms; those appearing in round
brackets occur less often): normal /Ii, I/ = /I/ {/Ii/} /Ii, I/ = /iI/ {/I/}
/OU{0/} {0/}/ = /OU/ {//}; frequent /I, E/ = /Ii/ {/I, E/} /I<, E</ =
/Ii</ {/I<, E</} /E<, E</ = /E</ /El, l E, / = /E /; /, aO/ = /, aO/ /, /
= // {/OU/} /U, Uu/ = /U/ {/Uu/} /O:, U/ = /O:/.
e following are occasional\ /{j}U, :/ = /:/, /I, U/ = /U/ /I, / = //
/aO, A:/ = /A:/. In addition, non-neutral pronunciation can have, for own /OUn/
and the diphthonging of /E, /, particularly frequent for /{0}/. Currently, women is pronounced like its singular form.
/t, / (and /p, k, c/) behave as in American English (without glottalization; but,
between V they can behave as in British English, with continuous non-occlusive
realizations of /p, t, k/). Neutral pronunciation regularly has linking //, but avoids
linking when no etymological // occurs.
Generally, reduced forms are less frequent and less systematic. Tendentially,
there is no systematic reduction of you her more often (even unstressed) been has
its full form: /bIin/. In cases such as aect eect and allusion illusion above all in
less neutral pronunciations, for a we have // (); for i e /I, , /, we have (/).
Frequently (/) occurs for the article a, too.
For -ary -ery , the American stressing is frequent, but neutral pronunciation
prefers the British one.
Traditional British English pronunciation
2.6.4. is is the classic pronunciation known as (Received Pronunciation),
which was the only one to be admitted by the up to one or two generations
ago, with narrow /Ii, EI, OU, Uu/ (Ii, eI, P, Uu), and /aE, aO, OE/ with well distinguished rst elements, (, AP, O) (and diametrically opposed to the mediatic realizations, (A, P, o)). It has no glottalization of /p, t, k, c/ (not even (0), nor
(), with the only possible occurrence of () for /t/, before sonants: ('skTln,
'skln) Scotland). Let us notice carefully the taxophones of the short vowels and
diphthongs with front rst elements before //.
In addition, it has /</ only when it is etymological and spelt as r (in which case,
it rather has (), except in informal or colloquial speech, at times). Besides, /tj,
dj, sj, zj, j/ are highly frequent, as in ('khwsTjn, -scn 'sjUup, 'sUu-), queion
super (with (S Z, Sj Zj) only in colloquial or informal pronunciation, for /'kwEscn,
'sUup/ ('khwscn, 'sup)); /;f, -, -s, -ft, -st/ = /O:/ (O:) (as in ('O;f, 'lO;sT) o lo
for /';f, 'l;st/ ('f, 'lsT)); /</ (>, 'R, R, R, pR, bR, kR, gR).
It invariably presents (-) /-I/ it has (h, `) for /h/ hotel, and (`) in reduced
forms with h and in -ham; /w/ (w), ('wn:) when; in addition, unstressed my is
116
/m, mI, m/ besides, /EI, OU/ are (, P) before ('): vacation november; /-OU/ is
(P) in compounds when it is at the end of the rst lexeme (even if separated): window sill. For ar, /A/ (A;) is possible; even /O/ for /O:/. nally, it has (>)
for /<, </, and () for /i, i/: ('khA:, 'khA; 'mO; 'mE>, -R) /'kA:, 'mO:,
'mE<i/ car more Mary.
Triphthong attenuation is very frequent (but less than in aected pronunciation or in Cockney), even between word, also for /I/: ('f;, 'f: ~'GOT, ~-)
/'faE, aEIn'GOEIt/ re I enjoy it. Substantially, its intonation patterns correspond
to the neutral ones; with the conclusive and suspensive intonemes, a creaky phonation type is very frequent.
/I[], i, i/ (, )
/E[]/ (, )
/[]/ (, )
/Ii[]/ (Ii, iI)
/U[], u/ (, U)
// () {desinential as well)
// (, *, |), /:/ (:)
/[]/ (, ), /O:/ (O:)
/A:/ (A:), // ()
/[j]Uu[]/ ([j]Uu, [j]uU)
/EI[]/ (eI, e)
/OU[]/ (P, oU)
/aE[]/ (, )
/OE[]/ (O, O)
/aO/ (AP)
/I[], I</ (, |)
/E, E</ (E, E|)
/[j]U[], -U</ ([j], [j]|)

{ ([j]P, [j]P|) /[j]O:/
/[j]O/)
/O[], O</ (O, O|)
( /O:/ (O:, O:|)}
/Ii/ {(Ii, Ii|) ) (I:, I:|) ( /I/}
/Uu/ {(u, u|) ) (U:, U:|)

( /U/, not /O:, O/!}
/EI/ {(eI, eI|) ) (e, e|) ( /E/}
/OU/ {(P, P|) ) (P, P|)
/aE/ {(, |) ) (, |)
/OE/ {(O, O|) ) (O, O|)

/aO/ (A, A|) ( (AP, P|)}
( /A:/ (A:, A:|)}
Aected British English pronunciation

2.6.5. It can be aunted by aristocrats and people of high social, religious, and
cultural standings. But it is generally thought of as too aected. It is characterized
2. english
117
by more peripheral /I, U/, (I, U) (instead of (, )); but they are more centralized
in /I, I< E, E< U, U</ (x, | x, | Px, P|); it presents /</ (x>),
and () for /i|/, as well: ('mx>) Mary.
en the phoneme /I/ is fully unstressed, it is (), as /i/ too (instead of (i)),
and, before pauses, it even becomes (): (&vz'blT, -T); it has /-I/ (-n, ) also
// is (), as many // are, as well.
Besides, /E, / are closer (e, E); when /I, E, / are nal, in stressed checked syllables ending in voiced C before a pause, they diphthongize as (I, e, E): ('bI,
'we, 'bE) big web bad.
/:, :< A:/ are backer, (x:, , :) (and the last one occurs more frequently
than in neutral pronunciation); for ar, /A/ (;) is possible. e diphthongs are
considerably narrow and have particular timbres, manly /Ii, Uu/ (ii, uu) and /aO,
OE, OU/ (Ax, @, +) (and (P), for (Ii, u, a, , ) and ()). For short vowels and for diphthongs with front rst elements, there are taxophones requiring
dierent symbols.
/I[], i/ (I[|], , I), /I/ ()
/i/ (, |)
/E[]/ (e[|], )
/[]/ (E[|], )
/EI[]/
(, )
/aE[]/
(, )
/Ii/
(i:x, i:|)
( /I/
(Ix, I|)}
/EI/
(, |)
( /E/}
/aE/
(:, )
/[j]Uu[]/
/I[], I</
([j]uu,
(x, |)
[j]uu)
( (jx:)
/OU+/ (P)
= /j:/}
/OU[]/ (+, P)
/E, E</
/OE[]/
(, |)
(@, @)
/aO/ (Ax)
f
F
f
/Ii[]/
(ii, ii)
f
/U[], u/ (U, U)
/O:/ (o:)
// (, *, |), /:/ (x:)
/zd/ (, x|), // (, |), // ()
/[]/ (, )
/A:/ (:), // ()
/Uu/ (u:x,
u:|) ( /U/,
not /O:/!
(Ux, U|)}
/aO/
(A:, A)
/U/ (,
|)+/:/
/[j]U[], -U</
([j]Px, [j]P|)
( /[j]O:/
(jx:) = /j:/}
/O:zd|/
= /O:[]/
(o:|, o:|,
o:[x]zD|)
/OE/
(@x, @|)
It has no glottalization of /p, t, k, c/ ( neither (0), nor (); with the only possible occurrence of () for /t/, before sonants: ('skTln, 'skln) Scotland] /tj,
dj, sj, zj, j/ are fairly frequent, as in ('khwesTjn, -scn 'sjuup, 'suu-), queion
super (with (S Z, Sj Zj) only in colloquial or informal pronunciation, for /'kwEscn,
'sUup/ ('khwscn, 'sup)); /;f, -, -s, -ft, -st/ = /O:/ (o:), in addition to // () in
some words, today, but o maintains /O:/.
/</ (>, 'R, R, R, /</ (>, 'R, R, R, pR, bR, kR, gR) (also ('V), and even ('e,
118
'=) in some frequent words: very terrible sorry tomorrow); non-written and
non-etymological /</ is frequent. Often, the aspiration of /p, t, k, c/ is very
weak (contrary to Cockney usage). In addition, it has (`) for /h/ hotel /w/ (w),
('wen) when unstressed my is /mI, m/ /-OU/ is (P) in compounds when it is at
the end of the rst lexeme (even if separated): window sill.
Triphthong attenuation is extremely frequent (as in Cockney), even between
words, also for /I/: ('f;, 'f: ~'G@T, ~-) /'faE, aEIn'GOEIt/ re I enjoy it.
Substantially, its intonation patterns correspond to the neutral ones; with the conclusive and suspensive intonemes, the creaky phonation type is very frequent; paraphonic pitch expansion is typical; in intonemes, syllables are lengthened.
Cockney pronunciation (London)
2.6.6. Most typically, it is the speech of the working-class of è Ea End of
London which includes the harbor. e main vocalic characteristics reside in its
diphthongs, which we present in the second vocogram, while in the third vocogram we add both the le broad variants (, seven) and the broade ones (*, two
{with grey edges}). Frequently, speakers can uctuate between these three types:
/Ii/ (, i), /EI/ (, ), /aE/ (, ), /aO/ (EX, X, *) (the last variant, which
is generally described as (:, a:), is the most narrow diphthong of all), /OE/ (o,
o), /OU/ (P, @+, X*), /Uu/ (%, %). Usually, the diaphoneme // is /I/ ().
For the monophthongs, the most evident characteristics in addition to some
timbres are contextual diphthongizations. In fact, in the most typical and broad
accent, /E, , / occurring in stressed monosyllables in (bi)checked syllables
with /0, 00/ are pronounced (, E, O). For the rst two phonemes, this fact
is particularly clear with /n, nd, t, d , k, ks, g/ (although /t/ = ()) and with other voiced C (but also with voiceless ones), as in: ('E) /'dd/ dad (for ('D;).
Something similar happens to /O:{}/, which most typically is (oU) (although in
a less broad pronunciation it is (o:)), as in ('lo;Un, 'woU, 'oU>) /'lO:n, 'wO;,
'stO:<i/ lYn water ory (for ('l:n, 'w;T, 'sT;>i)). In an intermediate accent as in
the less broad one, in all positions, we always nd (o:, oU), respectively; instead,
in the most typical and broadest, we nd (oU), when in word-nal position before pauses.
However, in nal position, within sentences, or with the grammemes /z, d/,
we have (oU): ('phoU) pY pore pour poor (for ('ph:) /'pO:, 'pO:/, and ('ph:,
-;) /'pU, -O:</ for the last one {following the most international phonemic order}); ('phoU) pYs pores pours poor's (for ('ph:) /'pO:z, 'pO:z/, and ('ph:,
-;) /'pUz, -O:<z/).
/O:/ occurs more frequently (and the same is true of traditional and aected pronunciations) than in the neutral accent, especially for /;/: ('oUf, 'whloUf, 'wh>oUs)
/';f, 'kl;, 'k<;s/ o clo` cro. Even /:/ can be diphthongized (:, ;); and also the timbres of /A:, / are quite remarkable (for /{}|/, too): ('wh:, -:* 'f;v,
'f-) /'kA:, 'f:/ car furèr e vowels which are followed by nasal consonants
(and often those which are preceded by nasals, too) are nasalized (as is the diph-
2. english
119
/u/ (%), /U[]/ (, UP, )
// (, *, |)
/O:/ (oU) {(o:)} {(oU +
z/D, oU|, oU>) = /U/}
/i/ (i), /i|/ = /Ii/ ()

/I[]/ (, P, ), // ()
/E[]/ ([0], P) {()}
/:/ (:) ()
/[]/ (E[0], P) {(E)}
/[]/ (, aP)
// (O[0]) {(O)}
/A:/ (:) {(:)*]
/[j]Uu[]/
([j]%, [j]P) /Ii/ (i)
/OE[]/
/EI/ ()
(o, oP)
/OU/ (@+)
/OU/ (P)
{(X)*}
/aE[]/
/aO/ (X)
(, P)
{()*]
/[j]U/
([j]U, -U:, -U|)
/[j]U</
([j]U:>) /[j]O:/
/Ii[]/
(, P)
/aO/ (EX)
/EI[]/
(, P)
/I/
(I, I:, I|)
/I</ (I:>)
/E/
(e, e:, e|)
/E</ (e:>)
/aO/ (E, E|)
/I/ (I, I,
I|, I>)*
/E/ (e, e,
e|, e:>)*
/[j]Uu/
([j]%,
[j]T%)
/OE/ (o)
/aE/ ()
/aE/
(, |)
(, |)
/aE</ (>)
/[j]U/ ([j]oU,
-oU, -oU|,
-oU>)* = /[j]O:/
/aO/ (:)*
(, |)*
/aO/
(X, X|)
(, |)
/aE/
(, |)*
/aE/
(, |)
(, |)
/I/
(, :, |, :>)
/E/
(, :, |, :>)
/[j]U/
([j]o, [j]o:, [j]o|,
[j]o:>) = /[j]O:/
thong /aO/, quite often independently from context). For the grammeme /I/ we
have (In, n, ); and, for -ì (-fk) is frequent: ('m3n, 'samfk, 'n'I&fk)
120
/'maEn, 'smI, 'EniI/ mine someì aJì.

e fourth and fth vocograms show the realizations of /I, I< E, E< {j}U,
-< aE, aE< aO, aO</ (often /{j}U-/ becomes /{j}O:-/); instead, the sixth and seventh vocograms show the broadest variants, whereas the eighth to the tenth vocograms give the least broad variants.
In addition, we have: /EI, -</ ( , |)* (, | A, |), /OU, -</
(, | P, P|)* (@+, @+|), /OE, -</ (o, o|)* (o, o|) (o, o|), /Ii,
-</ (, | , |)* (i, i|), /Uu, -</ (, | %, %|), /I, -</ (, :, |,
;>), /E, -</ (, :, |, ;>), /U, -</ (o, o:, o|, o;>) (thus, as /{j}O:, -</).
We will now consider, in the last vocogram, the many (and typical) neutralizations of /{}/ + //, which is vocalized into (P) (in broader pronunciations, we
nd (U), while in less broad ones, (PU), which we do not mark): /I, I, Ii, Ii/
(P), /E, E, :/ (P), /EI, , aO, aO/ (P), // (aP, P*) (in the broadest accent, the diphthong may coincide with /OUl/ (Pl), when it is lexeme-internal),
/A:, aE, aE/ (P), /, OU/ (OP) (in derivatives of /OU/ we nd (OPl), instead of
(Pl), exactly as in mediatic pronunciation), /{j}U, -U, -Uu, -Uu, -O:/ ({j}ou)
(for /{j}U, -U, -Uu, -Uu/, we also nd a less broad realization, ({j}Uu)).
For /OE, OE/, we have (oP) (even in less broad accents); for /-, -, -/ we have
(P) (and (U)*, (PU)); -el -al and 'll after vowels, can be slightly lengthened (P;, PU)
(and the same is true of /-, -/). For /-, -, -/, we nd (Pl, Ul*), uncle
Herbert ['Erbert] (&kPl'b{}).
In the rst two vocograms, we have marked in grey also ve V and ve VV,
which before // may not undergo the typical neutralization shown in the last vocogram.
/I[], Ii[]/ (P)
/E[], :/ (P)
/, EI[], aO[]/ (P)
/[j]U[], -Uu[], -O:/ ([j]ou)

{/[j]U[], -Uu[]/ ([j]Uu)}
/-[], -/ (P) ( (PU), (U)*}
/OU/ (OP) /OUl/ (Pl)
{in deriv. + (OPl)}
// (aP) {(OP)*}, // (OP)
/A:, aE[]/ (P)
As far as C are concerned, the most typical characteristic refers to /p, t, k, c/,
which are typically preglottalized, (=), even (), in all cases where in the British
accent synglottalization is possible ( 2.2.6.1-2, 2.2.7.1), or where in mediatic
British English preglottalization occurs ( 2.4.3.6). Even for the phonetic realizations we nd some dierences. In fact, in the most typical and broadest pronunciations, /p, t, k/ are realized as the corresponding stopstrictives: (p, , w), even
aspirated (in the normal contexts expected for neutral pronunciation, too): (ph,
h, wh) (which can give the impression of stronger aspiration). However, the
most typical and broad element is the substitution of /t, / with () in all the cases
seen in 2.2.5.2, but with further typifying contexts ( except before a tautosyllabic stressed nucleus, or after pauses, or after /s/, ('h, |h, 's, s)).
Examples: ('b) /'b</ butter ('woU) /'wO;/ water ('whO) /'kt/ cotton ('s) /'sIt/ sitti ('bO) /'bm/ bottom ('b) /'b:t/ Burton ('P)
2. english
121
/'Et/ Elton ('whln) /'klIntn/ Clinton ('lP) /'lI/ little ('ph;na, -;-)
/'pA:tn/ partner ('sp>l) /'sEp<tli/ separateI ('lOs) /'lts/ lo_
More examples: ('phan, -n,, -n, -n[, -n) /'pEIn/ painter ('phan, -n,, -n, -n[, -n) /'pEInt/ paint it ('woU) /'wO:/ Walter ('oU,
-,, -[, -) /'hO:t/ halt it (';, -,, -[, -) /'stA:t/ art it (', ',,
'[, ') /'hIt/ hit it ('lP 'b 'b) /'lI 'bI v'b/ a little bit of butter
('ph 'p) /'pU 'p/ put up ('ph 'p) /'pU 'p/ put it up
In less broad pronunciations, an incomplete, attenuated stop is possible: (,),
which is less invasive; the vocoid preceding () can even be laryngealized, whereas () can become zero, especially before another vocoid (adding, however, the
creaky phonation type), ( = , = , = ): ('lP 'b 'b). Generally, forms
such as lill ('lP) /'lI/ and little ('lP, 'lP;, 'lPU) /'lI/ maintain some dierences
even if the latter is actually pronounced in this way; as a matter of fact, in addition to the creaky phonation type, // is often lengthened (at least in an intoneme).
Before V (even if derived from //, and even between words), also a less broad
variant, ([), is possible (or even () in elegant speech, which we do not indicate). It is also possible for (n[) to become (n); here we will report the relevant examples, without spelling, following the order in which they are given above (including water and Walter): ('b[, 'woU[, 's[n, 'b[m, 'l[P), ('phan[, -n
'phan[, -n), ('woU[, 'oU[, ';[, '[, 'l[P 'b[ 'b[, 'ph[ 'p, 'ph[[ 'p).
Other consonants can become (), especially /p, k/: ('O) /'stpI/ oppi
('fl 'l) /'fIlIp 'laEktIt/ Philip liked it ('whOn) /'kkni/ Co$n\ In a previous example, we have seen that typically /d/ becomes stopstrictive, (); besides,
commonly, /d/ is realized as (), when it is word-nal and followed by C or V
and in the grammeme sequence >-dn't /-dt/, as well: ('b> 'b) /'b<Ed 'b/
bread and butter ('g 'bo;) /'gUd 'bOE/ good boy ('{}) /aE'dIdt/ I didn't
For /st, st<, sc/, broad pronunciations have (, >, Sc): (';) /'stEI/ Z
('>O;) /'st<;/ ro ('wwhScn, -c) /'kwEscn/ queion In broad pronunciations, /, / become /f, v/; however, there are many intermediate nuances, including the realizations of normal pronunciation: (f, v , , , , ). More often, // can be realized as (`, , d, D): (s'EXs z'm3n, s-, ds-, Ds-) /Is'haOs Iz'maEn/ ìs house is mine. As we have seen, the typical realization of /h/ is (`),
which is a stigmatized pronunciation, and therefore can lead many speakers to hypercorrecting: ('h{}) /'Iit/ eat. For /n, t, d/, the typical Cockney pronunciation has no //, but, in less broad pronunciations, mediatic-like types are possible,
as well: ('n;%, 'n;%, '~;%, '~T;%) /'nUu/ n[ ('h';n, 'hK;n, 'hK;n, 'h;n,
'chK;n, 'ch;n) /'tUun/ tune ('%w, '%w, '%k, 'T%k, 'G%k, 'GT%k)
/'dUuk/ duke.
Triphthong attenuation is extremely frequent, even between words, also for
/I/: ('f;, 'f; ~'GoT, ~-) /'faE, aEIn'GOEIt/ re I enjoy it. Substantially,
its intonation patterns correspond to the neutral ones. For [n]eièr we generally
nd /Ii/.
3. Italian
3.0.1. is chapter will analyze modern neutral Italian pronunciation, keeping
it separate from traditional pronunciation, which will be looked at later, in order
to demonstrate the accepted and shared dierence between them. However, this
dierence is not evident in dictionaries and grammar books, which is hardly surprising given the inertia and lack of dynamism shown by schools and obviously
dictionaries and grammar books. ey continue to perpetuate conventional convictions, a legacy from grammar books and dictionaries from the 1800's, without
even doubting that things may have changed in the meantime Endless plagiarizing, with very few innovations as far as orthoepy ( pronunciation) is concerned.
is can be seen in the way csa for casa, /'kasa/ is still given. is is the traditional pronunciation, while the modern one has /'kaza/. Evidently, this shows a
lack of attention for the phonic aspect of the language. It is also true that this kind
of notation can, almost systematically, escape non-attentive dictionary-users, who
can thus presume that when a dictionary reads >-s- it means /z/. is case is not so
serious as it coincides exactly with modern pronunciation; but what would happen for -asi in qualsiasi?
In the Zingarelli dictionary (carrying the year 1997), the present writer has
given two variants, indicated as c{a (= /'kaza, -sa/), and many other words, increasing the double possibilities for /e, E o, O q, Q/ as well, though not reaching the
variety of pronunciation in DiPI, with its modern, traditional, acceptable, tolerated, slovenly, intentional and lofty variants (explained in MaPI, as well).
However, before we even begin to deal with these subtleties, we will have to
discredit a good few beliefs which are part of common culture, especially upheld
by teachers (even language teachers, abroad too), perpetuated by society and
schools worthy of the rst millennium of our era
Earlier we mentioned the pronunciation shown in dictionaries and grammar
books, which, instead of depicting the actual reality, as can be perceived by merely paying attention with open ears, they are easily satised with using and reusing what has always been printed, without checking to verify if it is still in current
use.
Unfortunately, the same happens in various brief guides to diction and pronunciation, that continue to proliferate, regardless of objective change (that one
should be able to perceive or, above all, want to perceive). Even in books about linguistics, glottology, dialectology and in historical grammar books or text books or
even descriptive grammars, one can nd a continual perpetration, as if it were
real, of what is unproposable as far as both vowels and consonants, or word-stress
and co-gemination are concerned.
3. italian
123
3.0.2. at brings us to the sore points. Writing is (erroneously) considered a

faithful indication of pronunciation; if pronunciation is looked at at all in teaching. More often than not it is spelling that is dealt with, relegating pronunciation
to quite a secondary position or else, it is even considered bothersome. Indeed,
very few people have a real perception of what exactly Italian pronunciation is (the
same goes for any other language). e most common belief amongst language
workers, language teachers, is that dealing with pronunciation is not part of
their duties, as if it were an unrelated aspect, or just an added complication to their
work.
Nevertheless, the rst manifestation of language is exactly through sound, therefore through pronunciation. Only telepathy could possibly make pronunciation
superuous. Despite this, for the whole of the third millennium we will have to
face this problem whether we like it or not is being the case we should nally begin to look at it properly; no more pushing the problem aside, nding all
kinds of excuses.
at is actually lacking here, is a friendly approach, one which is not misleading. As pronunciation is inevitable, at least for us common mortals, we have to
learn to accept it for what it is: the objective and perceptible manifestation of language.
It is neither, therefore, a wicked invention, nor a task comparable to Sisyphus'
eorts; it is merely a part of the teachinglearning process of any language. In
some languages spelling is (still) quite close to pronunciation, in others the gap is
wider, to greater or lesser extents, due to natural linguistic evolution, which is unstoppable, whereas the written word is always left behind, like an eternal defeat,
it never comes rst.
However, schools and society consider it instead to be the real indication of
pronunciation, when it is only a rough and poor way to render pronunciation.
Indeed, its aim is not to indicate pronunciation, but more modestly, to permit the
conservation of written documents independently from its pronunciation, so as
to allow people, who know the language, to nd the contents put in writing.
3.0.3. e very act of confusing spelling with pronunciation, inevitably leads
to interpretations and deductions which often have little to do with pronunciation. Let us now look at the concrete facts. en a person with no adequate
reading into pronunciation is asked how many vowel phonemes there are in Italian, the answer is, inevitably (from teachers themselves onwards), ve: a e i o
u. is erroneous answer, results from the examination of the wrong object. For
the Italian language the following procedure was undertaken: starting with the alphabet and excluding all consonants, obviously leaving only the ve letters: a e
i o u.
Clearly, the result is erroneous, because the calculation was reckoned on entities which have very little to do with pronunciation. letters are not sounds! In
neutral Italian, the letters: e and o, each indicate two phonemes, that are respectively /e, E/ and /o, O/, which are closed and open e and o, like in (se) corresse
(pi velocemente) if she ran faster with /e/ (), which is dierent from (lei) cor-
124
resse (i compiti) she corrected the homework, with /E/ (), or in (se) fosse (vero)
if it were true with /o/ (), which diers from (le) fosse (&avate) the dug
graves, with /O/ (). erefore, there are seven Italian vowel phonemes: /i, e, E,
a, O, o, u/.
Let us continue our search for the number of phonemes, looking at consonant
phonemes. e automatic answer (from a person with normal schooling, without
appropriate reading or consideration) is sixteen: b c d f g h l m n p q r s, t
v z. e very order of letters inevitably betrays the fact that the incredible gure
of sixteen was reached by means of an arithmetical operation, using the alphabet,
instead of actual sounds.
Hence, if you take the Latin alphabet, take away the ve Italian vowels and
the ve non-Italian consonants [j k w x y), it seems logical to get to the sad
list given above.
e number of phonemes in a language, however, cannot be obtained through
spelling, but through the distinctive sounds of that language. Just to begin with,
the list contains two letters which have nothing to do with Italian sounds: q is
practically useless; and h is only a diacritic, it is necessary to distinguish ha from
a, ho from o, hanno from anno, and also /ki*/ from ci /ci/, ro /'giro/ from
giro /'Giro/; there is even a vowel dierence between ho /O*/ and o /o*/. Incidentally, we can but disapprove of the method used by those teachers who, worried
only and exclusively about spelling (and not really understanding anything about
the pronunciation of the language!) dictate things such as */has'kritto, hannostu'djato/ to students, for ha &ritto hanno studiato he/she has written, they have
studied. e poor (selsh and dishonest) result of managing to obtain correct
written tests to show them o shamelessly produces, instead, a much worse (and
absurd) error in making pupils believe that above all, in formal situations, you really should say such obscenities as */hOs'kritto/ ho &ritto, instead of the only possible and admissible version: /Os'kritto/.
3.0.4. For the moment, let us say that there are 23 consonant phonemes in Italian and they are certainly not in alphabetical order, but in a phonic order, according to places and manners of articulation and phonation type, after having discovered them, not by graphic deduction, but by looking for opposites in minimal pairs, that contain two similar words, but with one dierent phoneme, which
makes the meaning of the two words change, as seen with /ki*, ci/ and /'giro, 'Giro/ (respectively and ci, ro and giro).
Furthermore, in the number of phonemes, we must bear in mind that the two
graphemes s and z each have (as do e o) two dierent phonemic values, as in presento [una persona I introduce a person, /pre'zEnto/, from presentare] and presento [un avvenimento I have a presentiment of an event, /pre'sEnto/, from presentire pre-sentire) in the same way we have razza (ray, spoke: /'raQQa/) and
razza (race, breed: /'raqqa/)
We must not forget other spelling problems that derive from the fact that Italian represents a transformation and evolution of Latin, which had a certain number of phonemes (obviously dierent from Italian, both phonically and numerical-
3. italian
125
ly speaking, as can be seen in the phonosyntheses of NPT/HPh, 22.1-4 or from

18 of MaPI {from 1999 onwards}).
As a matter of fact, classical Latin did not have /c, G/, which are still expressed
by ci gi in Italian (because Latin /ki, gi/ were transformed in time using the palatal
(i, i) of imperial Latin, to (ci, Gi) of medieval Latin {as well as clerical and academic Latin in Italy}). erefore, Italian has to then render /ki, gi/ with .
Similarly, Latin did not have /S, N, L/ or even /q, Q/, so in Latin, amicitia was /ami:'kitia/ (&mi'kItI) not /ami'ciqqja/ as in Italian, and in Italian clerical and academic Latin (whereas in {clerical and academic} Latin in other countries, the
phonic result obviously depends on the pronunciation of their national languages); hence, Italian spelling resorts to &(i) (i), and z for /q, Q/, as they
are new phonemes, foreign to classical Latin.
Furthermore, one must not believe everything that is unfortunately published; as a matter of fact, neutral Italian does not have vowels reduced to schwa,
not even in unstressed syllables, () (or (, ), 11.19 of NPT/HPh; nor to (, ,
, , , P, ), that are all regional pronunciations), nor assimilations such as
/ss/ actually, (s) for /st/ (like in questo), and /00/ for /r0/ [giorno] that are
very regional; nor /0wj/ like (0j) [continuiamo] which is not even Italian
(although some Italian {!} authors tried to pass it o as Italian). Obviously, in spontaneous speech, one can stammer or mue some sounds, but we certainly do not
get (, ), the only real possibilities for a neutral reduction of vocoids are explained in 10.11 of NPT/HPh (since in neutral pronunciation, one never has a
full (), not even in cases such as /'sum, 'frak/ sum (Lat.), frac (Fr.), which are
('frak:$, 'sum:$), where ($) at the most, stands for (); while, most coarticulation for
/wj/ can consist of ((3)) (provelar rounded + postpalatal approximants, instead of
canonical velar rounded + palatal approximants, (wj)).
Vowels
3.1.1. As examined in 3.0.3., for the ve graphemes a e i o u neutral Italian
has seven vowel phonemes, /i, e, E, a, O, o, u/, realized by nine taxophones, (i, e, ,
E, a, O, , o, u), as can be seen in 3.1. Before presenting the examples, we will
analyze the two taxophones which are seemingly foreign to the harmony of the
seven phonemes, : (, ).
e most intriguing aspect is that it deals with the pronunciation on the basis
of dierent principles of both /E, O/ and /e, o/. In the rst case we have the manifestation of half-closing, in fact, starting with /E, O/ we get to (, ), when there is no
longer a primary or strong stress, as is the case for the rst elements of compound
words from independent lexemes: (&pRndi'so:le) /prEndi'sole/ prendisole, (b'ke)
/bEn'ke*/ ben>, (&kpRi'lEt:to) /kOpri'lEtto/ copriletto, (pi'ke) /pOi'ke*/ poi>.
e other case regards the manifestation of half-opening, because starting with
/e, o/ we again get to (, ). is happens in /e, o/ endings after stressed syllables, when the stressed vowel is /i, u/, as in: ('vi:v, 'vi:v) /'vive, 'vivo/ vive vivo,
('ku:c, 'ku:c) /'kuce, 'kuco/ cuce cucio. Moreover, in an intoneme, the same can
126
also occur for /'e-o/: ('ve:do, -d) /'vedo/ vedo (but not for /'o-e/: ('do:ve) /'dove/
dove). at remains to be said, as can be guessed (and conrmation is always precious), is that this happens with a nal C too: ('ri:dR, 'ri:dn) /'rider/ rider(e) ridon(o), ('ip:siln, 'su:tR) /'ipsilon, 'sutor/ ipsilon, sutor (Lat.). ere are however,
other less regular cases which are dealt with in 3.3 of MaPI.
Let us take a look at some examples of the seven vowel phonemes: ('vi:ni) /'vini/
vini ('se:te) /'sete/ sete ('sEt:te) /'sEtte/ sette ('ra:na) /'rana/ rana ('Ot:to) /'Otto/ otto
('sot:to) /'sotto/ sotto (kul'tu:Ra) /kul'tura/ cultura. Clearly, /j, w/ are not vowels, but
(approximant) consonants, as in: ('pju) /'pju*/ pi ('kjE:do) /'kjEdo/ edo ('kwa)
/'kwa*/ qua ('bwO:no) /'bwOno/ buono. It is equally true that /ju, jE, wa, wO/ are
not diphthongs at all ( 3.1.2) but simply sequences of CV like /su, tE, va, nO/,
.
3.1. Italian monophthongs.
/i/ (i)
/e/ (e), ({'i/'u))
/E/ (E), ({'})
/u/ (u)
/o/ (o), ({'i/'u))
/O/ (O), ({'})
/a/ (a)
Diphthongs
3.1.2. Italian grammar books put a lot of eort into complicating what is, in
fact, quite simple. Indeed, instead of the three very common structures, the real diphthong ((', &, )), the hiatus ((', &)) and the heterophonic sequence
((0)), , (j), (w), and the like), they continue to consider only two of them:
diphthong (with fusion: syneresis) and hiatus (with separation: dieresis),
but with strained interpretations of medieval origin, of a graphic-grammatical and
graphic-metric nature. In fact, semi-vowels or semi-consonants do not exist:
they are merely an incredibly successful magic trick!
As a matter of fact (unless one expects to do magic in phonetics using graphic-grammatical categories), it is phonetically absurd to speak about a diphthong
for (j, w) (('pjE:no) /'pjEno/ pieno ('gwan:to) /'gwanto/ guanto] as only (i,
u) (('fai) /'fai/ fai ('pa;uza) /'pauza/ pausa] are real diphthongs, as any sequence
of (', &, ) (('a;uto) /'auto/ auto (au'tEn:tiko) /au'tEntiko/ autentico]
It is equally absurd to speak about hiatus for ('i, 'u), as only (i', u') are real hiatuses, as any other sequence like (', &) ((pa'u:Ra) /pa'ura/ paura] compared
to (pau'ro:zo) /pau'rozo/ pauroso a real diphthong. Hence, one cannot believe literature that only uses two categories ( diphthong and hiatus) and, what is more,
they dangerously mix them up to a point where they include in diphthongs
the heterophonic sequences (/j, w/), and in hiatuses the real diphthongs (as
in ('ma;i, 'ca;o, 'bO;a) /'mai, 'cao, 'bOa/ mai ciao boa 1.4.4).
3. italian
127
3.1.3. us, in languages like Italian, diphthongs are vowel sequences, which
are more or less common, formed by a combination of the seven vowel phonemes
and nine taxophones (/i, e, E, a, O, o, u/ (i, e, , E, a, O, , o, u)), that writing tidily reproposes without the problems presented for example by Germanic languages. ese really do have monophonemic diphthongs, because, in dierent accents they vary as to their realizations, independently from monophthongs (and
more than monophthongs do), and also because they have varied historical spellings.
erefore for Italian, it seems pointless in this chapter to make a list of diphthongs, of very dierent frequency; it would be appropriate however, to look at
8.26 of NPT/HPh (as well as 5.1.2-3 of MaPI]
By forcing (and violating) reality, grammar and metrics continue to call, above
all, /jE, wO/ ascendant diphthongs, because they derive from the Latin /E, O/ (,
), and they nd they need to have to invent (as the best lawyer for the worst
criminals) ctitious realities, like semi-consonants and semi-vowels.
ese seem to be nearly cinematographic special eects, that make one believe
one is facing something real, but which is completely invented! If phantaphonetics or virtual phonetics is not our aim, then sequences like (j, w) are naturally part of /{0}0/ groups, as in /'fjanko, 'franko 'gwado, 'grado 'twOno, 'trOno
'gjande, 'grande, 'glande/ anco franco guado grado tuono trono ande grande ande. Paradigmatically, /j, w/ are in opposition with /0/, clearly not with //
and can under no circumstance belong to vowel groups.
Even semi-vowels are a truly distorted reality and have been invented to try to
explain (but they are only deceived into trying to explain) what has no need of explanations. Indeed, what need do normal diphthongs, like /ai, au/, have for alibis
to defend themselves from grammatical and metric fanaticism, so as to demonstrate that they constitute one syllable and not two: ('ma;i, 'ka;u{to}) /'mai, 'kau{to}/
mai cau(to)?
Even /ia/ is a monosyllabic diphthong, as in ('mi;a) /'mia/ mia. It is evident,
in the same way as the Earth is round and rotates around the Sun; and yet, it has
not been at all easy for it to be accepted! e diphthong, hiatus, and /CV/-sequence matter, is much simpler: one does not have to be a scientist; all it takes is
observation and thought (using both ears)! Yet there is no change! Let us trustingly refer to 5.1.2-3 of MaPI as well.
Consonants
3.2.0. 3.2 shows the table of the neutral Italian consonant articulations, including their taxophones ((M, ~, R )), which are necessary for satisfactory pronunciation.
Instead, 1.9-15 gives orograms, grouped by manners of articulation of all contoids given in the chapters of this volume, even as secondary, occasional or regional variants of the 12 languages dealt with.
128
N
c G
velar rounded
(~)
velar
palatal
dental
(n)
t d
qQ
postalveo-palatal
(M)
alveolar
m
F pb
labiodental
bilabial
3.2. Table of Italian consonants.
()
k g
f v
s z
(l)
S {Z}
r|(R)
l
j
()
Nasals
3.2.1. ere are three nasal phonemes, /m, n, N/ (m, n, N), and four taxophones,
for /n/, that can be rendered by three supplementary symbols (M, ~, ) (a fourth
symbol, (()), in front of /t, d q, Q s/, can be useful to bring attention to the assimilation, from alveolar to dental, even if the dierence, though real, is less important than in other cases): ('mam:ma) /'mamma/ mamma ('ma:ma) /'mama/ m'ama ('nOn:no) /'nOnno/ nonno ('nO:no) /'nOno/ nono ('baN:No) /'baNNo/ bao (peR'NOk:ki) /per'NOkki/ per oc ('toM:fo) /'tonfo/ tonfo ('dEn:te) /'dEnte/ dente
('fra~:Ga) /'franGa/ frangia ('fa:go) /'fango/ fango.
In neutral Italian, /N/ is self-geminant ( 3.3.1.5.): ('soN:No, loN'NO:mo) /'soNNo,
loN'NOmo/ soo lo omo. We can observe that the self-geminant /N/ is in opposition to geminates, as in ('soN:No) /'soNNo/ soo and ('son:no) /'sonno/ sonno, but
not simple C (('so:no) /'sono/ sono); the minimal pair is between the rst two examples, not the third.
In order to closely examine the manifestation of nasal assimilation in neutral
Italian, a whole series of nouns could be shown (beginning with all possible C,
3.4 of MaPI] preceded by (kon) /kon/ con with, but we will limit ourselves to
(kom'mar:ko) /kom'marko/ con Marco and to the improbable (koN'NE;o) /koN'NEo/
con Gneo (a rare male name, even in ancient Rome).
Various problems (not only with consonants, but with vowels and intonation,
too) deriving from various regional pronunciations are examined in some chapters of MaPI, as well. We can now briey mention the fact that, very often, in the
north of Italy, the sequences of /n0/ are not homorganic by assimilation, but have
a velar articulation, (), or semi-prevelar (without full contact), (), as in (&uba'bi:n, &uba'bi:n) (concentrating, here, only on the nasals, because the actual regional dierences, doubtlessly, also regard other C, many V the type of phonation,
the syllabic structure and the intonation), for neutral (&umbam'bi:n) /umbam'bino/ un bambino.
In the same way, in the north, /N/ is very often not self-geminant: ('sO:No, 'so:No
3. italian
129
lo'NO:mo); but, generally, /N, nj/ are confused with something intermediate, which
we show quite generally here: (ka'pa;~ja, ka-) which stands both for /kam'panja/
Campania and for /kam'paNNa/ campaa. In central areas, /nj/ becomes (N) (but
in this case not self-geminant): (an'tO:No) /an'tOnjo/ Antonio (aN'NEn:to) /an'njEnto/
anniento
Stops
3.2.2. Italian has three diphonic pairs of stops, /p, b t, d k, g/ (p, b t, d k, g):
(peR'bE:ne) /per'bEne/ per bene (ti'dO) /ti'dO*/ ti do (ko'gwaL:Lo) /kon'gwaLLo/
conguao. Although before a front V and in front of /j/ /k, g/ are realized as prevelar, it is not necessary to systematically use the special symbols ((, )): (&kjakkje'ra:Re) /kjakkje'rare/ acerare ('kik:k) /'kikko/ cco ('gE:ga) /'gEnga/ ghenga.
e greatest regional problem concerning stops is provided in the gorgia toscana ( Tuscan throat) and in the central-southern voicing, which will be examined in general terms (also because there are dierences between areas; but for
more precise details, various chapters of MaPI can be consulted). erefore, the
following gives just a general idea: (&ihaFi'a:ni) /ikapi'tani/ i capitani and ('ciwe
im'un:) /'cinkwe im'punto/ cinque in punto
Stopstrictives
3.2.3. ere are two diphonic pairs, dental and postalveopalatal protruded, /q,
Q c, G/ (q, Q c, G): ('dan:qa, 'pjaq:qa) /'danqa, 'pjaqqa/ danza piazza ('QO:na,
aQ'QO:to) /'QOna, aQ'QOto/ zona azoto ('ce:na, 'fac:co) /'cena, 'facco/ cena, faccio ('GEn:te, 'OG:Gi) /'GEnte, 'OGGi/ gente oggi.
e main regional problems regarding the north concern the sequential realizations of /q, Q/, as: ('datsa, -a, -a, 'da- 'pja{t}tsa, -a, -a d'zO:na, d'-, d'- ad'zO:to, ad'-, ad'-) (or, in broader accents, simply as constrictive: ('dasa, -a, -a, 'da-
'pjassa -a, -a 'zO:na, '-, '- a'zO:to, a'-, a'-)), instead of real stopstrictive
phones, (q, Q). For almost the whole center and many parts of the south, without going into too much detail, the most obvious characteristic is given by the
change of /c/ to its corresponding constrictive, (S): ('pe:Se) /'pece/ pece (&diSe'Si:lja) /dice'cilja/ di Cecilia ('ka:So) /'kaco/ cacio. In Tuscany the same goes for /G/ = (Z): ('a:Zile, la'Zak:ka) /'aGile, la'Gakka/ agile la giacca.
It is important to focus on the fact that we are dealing with single postvocalic
/c, G/ (and hence, in this case, intervocalic too), thus avoiding ridiculous imitations which have, unfortunately ended up amongst examples shown in books
about linguistics, dialectology, language history, language teaching and sometimes
even in phonetics and phonology!
erefore, it is legitimate to give (la'Zi:a) /la'Gita/ la gita, as an example (although the ocial symbol () is hardly recommendable), but not simply *('Zi:a)
130
/'Gita/ gita, because after silence or after C, it stays for (G), ('Gi:a), as does, obviously, (aG'Gi:na) /aG'Gina/ a Gina (because of co-gemination, 3.3.2.1-3; certainly not *(aZ'Zi:na)). Naturally, the same is true of (la'Se:na) /la'cena/ la cena
while we can only have: (peR'ce:na, ac'ce:na) /per'cena, ac'cena/ per cena a cena
(absolutely not *(peR'Se:na, aS'Se:na))!
Apart from Tuscany (and some northern parts of Umbria and the Marches), in
the center and south (still generalizing the area and the precision in transcription,
too), we have /G/ = (GG): ('aGGile, laG'Gi:a)
Constrictives
3.2.4. Neutral Italian has two diphonic pairs of constrictives, /f, v s, z/ (f, v s,
z) as well as a voiceless postalveopalatal protruded self-geminant /S/ (S) (
3.3.1.5.), with the voiced correspondent, but non-geminant ( the beginning of
the same paragraph), in foreign words, above all French: ('fa:va) /'fava/ fava ('sO:zja) /'sOzja/ sosia ('peS:Se, loS'Sa:me, {laS}'SE:na) /'peSSe, loS'Same, {laS}'SEna/ pe&e, lo
ame, (la) &ena (&aba'Zu:r, &aZi'go) /aba'Zur, aZi'go/ abat-jour gigot
In front of any voiced C there is always /z/ in Italian: (biz'bE:tiko, zve'ni:R, zle'ga:Re, zden'ta:to) /biz'bEtiko, zve'nire, zle'gare, zden'tato/ bisbetico svenire slegare
sdentato ( (sten'ta:to) /sten'tato/ stentato]
As far as VsV is concerned, modern neutral pronunciation resolves the problem
of the traditional one ( 3.3.5.2.); actually, every postvocalic intralexemic -s- (
in simple words, not compound] is voiced, /z/ (z): ('ka:za, 'ri:z, 'pre:za, ci'ne:ze, Ge'lo:zo) /'kaza, 'rizo, 'preza, ci'neze, Ge'lozo/ casa riso presa cinese geloso Only in
compounding, is the initial -s- of lexeme or grammeme /s/ (s): (&pRndi'so:le, &sotto&segRe'ta:Rjo, &pResa'la:Rjo, kwal'si;asi, Ri'sal:to) /prEndi'sole, sottosegre'tarjo, presa'larjo, kwal'siasi, ri'salto/ prendisole sottosegretario presalario qualsiasi risalto e
last example means I jump again; whereas, the noun means to stand out and
is (Ri'zal:to) /ri'zalto/ Forms such as prosieguo, whose composition nowadays
is heard in a less denite way, vary: (pRo'sjE:gwo, -z-) /pro'sjEgwo, -z-/.
For /s, z/, in the north, an alveolar instead of dental articulation is frequent; ('O:ja) /'sOzja/ sosia (be warned that, too often, neutral Italian articulation, which is
dental {with lowered tip}, is dened alveolar, even in phonetic books!).
In central-southern regional pronunciations (except in Tuscany), for VsV (even
if with an approximant inserted after s), we do not have /z/, but /s/: ('rO:se, 'a:sja)
/'rOze, 'azja/ rose Asia. However, the prestige of voiced pronunciation, with /z/, is
such that, many speakers try to go from (s) to (z); but from a regional situation
without /z/ (and with no help from the written word), the substitution often generalizes (z) (or ()), even in words like: *(&laza'lu:t, &laa-) /lasa'lute/ la salute, that
is (&lasa'lu:t).
In the center (except orence and Prato) we nd the typical change /s/ = (q) after /n, r, l/: /'pEnso, 'pErso, 'falso/ ('pEn:qo, 'pEr:qo, 'fal:qo) penso, perso, falso. The
same happens in Tessin and in northern and easter Lombardy. It is very frequent
also in some areas of southern Italy, including its partial or complete voicing: (,
3. italian
131
Q) ( MaPI). However, speakers who have a kind of defective r produce ('pEV:so,

'pE:so, 'pEK:so) (not a stopstricive).
In the center and south of Italy, above all, there are areas in which preconsonantal /s, z/ are seen to be prepalatal, (, =): ('vi:ta, ='bat:to) /'vista, z'batto/ vista sbatto ( in the south-eastern parts of the Marches, Umbria, and Latium; and in western Sicily, and southern Calabria and Salento; as well as in northern Lombardy).
In Campania, /s0, z0/ are typically realized as postalveopalatal protruded constrictives, (S, Z), but only in front of non-apical C; thus, for instance: (S'pA;R, Z'gar:R)
/s'paro, z'garro/ sparo sgarro. We cannot but reveal that too many third-rate imitators (even in books!) show absurd pronunciations, for example in Neapolitan,
: *(S'ta:Re) /s'tare/ stare for (s'tA;R).
In the north, /S/ is often without lip protrusion, but, above all, it is not self-geminant: ('pe:e, lo'a:me, {la}'e:na) pece lo ame la scena; so, the rst and third examples are closer to the central-southern pronunciation of ('pe:Se, la'Se:na) /'pece,
la'cena/ pece la cena!
We can see how self-geminant /S/ opposes geminates, as in ('faS:Se) /'faSSe/ fa&e
and ('fac:ce) /'facce/ facce, but not single, as in ('fa:ce) /'face/ face (a literary
word), which in central Italian pronunciation (and very often in southern ones)
is ('fa:Se); however, the minimal pair is between the rst two examples, not the
third.
Approximants
3.2.5. e two Italian approximants are /j, w/ (j, w): ('jE:Ri) /'jEri/ ieri (gjac'ca:jo) /gjac'cajo/ acciaio ('wO:vo) /'wOvo/ uovo (kwa'lu:kw) /kwa'lunkwe/ qualunque
In central-southern pronunciation, we sometimes nd /j/ = (jj): ('paj:jo, 'pa;jjo)
for ('pa:jo) /'pajo/ paio In the south we often have even (i'e:Ri, u'o:v) for /'jEri, 'wOvo/ ieri uovo (simplifying a little).
Trills
3.2.6. ere is only one trill phoneme, /r/, with two important taxophones, (r,
R), which depend on the strength of the syllables; in stressed syllables, before or after the syllabic nucleus we nd a trill, (r) (lengthened, (r:), if in a checked syllable
of an intoneme); whereas in unstressed (or half-stressed) syllables we have a tap:
('ra:Ro) /'raro/ raro ('pOr:ta) /'pOrta/ porta (Ri'prE:ndeRe) /ri'prEndere/ riprendere
(&pRepa'rar:si) /prepa'rarsi/ prepararsi ('kar:Ro) /'karro/ carro (aR'ri:v) /ar'rivo/ arrivo (&aRRi'va:Re) /arri'vare/ arrivare As one can see, even for /rr/, the choice between
(r, R), depends on the position of the stress.
ere are various types of accents with a French r, in Italian, that range from
uvular, like (K, , , r), to labiodental (with or without uvularization), (V, ).
For regional pronunciations (as shown in MaPI), we could have a generalized
132
(R) or (r), independently of the stress; we could also have uvularization, (5, R); or
in Venice, (, , ).
Let us take a detailed look at the Sicilian type (in Sicily, of course, and in
southern Calabria and the Salentina Peninsula as well) for /r, rr, tr, dr, str, sdr/.
Indeed, simplifying a little (even the transcriptions) for /r, rr/, we have the voiced
slit alveolar constrictive, (z), self-geminant if postvocalic: ({laz}'za:djo) /{la}'radjo/
(la) radio ('kaz:z) /'karro/ carro It is clearly not the postalveolar trill phone ((#),
cacuminal, ), as it has been described for generations, without proof.
Relative to this, we also have the bizarre pair =, which is persistently presented as (#, #), without listening carefully. In fact the most normal articulation is given, simply, by the stopstrictives which correspond to (z), (., ) (slit
alveolar): ('.a:v) /'trave/ trave ('a:g) /'drago/ drago
Lastly, the /str/ sequence (lexically quite common), is more often rendered by
the voiceless postalveo-prevelar constrictive (self-geminant, if postvocalic) (.): ('na.:.) /'nastro/ nastro ({la.}'.a:da) /las'trada/ (la) strada; instead, the (less common)
sequence /zdr/ is more often rendered by a voiced (grooved + slit) alveolar sequence, (): ({la}'a:j) /{la}z'drajo/ (la) sdraio (much less commonly by (la{}'a:j)).
We will conclude by indicating the assimilation of /r0/, which is more typical
of eastern Sicily: ('pat:t) /'parto/ parto.
Laterals
3.2.7. Italian has two lateral phonemes, /l, L/, alveolar and palatal (this last is
quite rare in languages, and in neutral Italian it is self-geminant, as well): ('la:la)
/'lala/ l'ala ('lal:la) /'lalla/ Lalla ('dir:Li) /'dirLi/ dir ('fOL:La) /'fOLLa/ foa.
We can observe how the self-geminant /L/ opposes to geminates, as in ('paL:La)
/'paLLa/ paa and ('pal:la) /'palla/ palla, not to (('pa:la) /'pala/ pala); the minimum
pair is between the rst two examples not the third, which is a single consonant.
In some central areas, /L/ becomes /j/, or /lj/ becomes (L) (in this case it is not
self-geminant): ('paj:ja, 'pa;jja) /'paLLa/ paa ('O:Lo) /'Oljo/ olio (&paLLa'ti:v) /pallja'tivo/ palliativo On the other hand, in the north, generally, /L, lj/ are mixed into something intermediate that we can render quite generally here: (li'ta;-ja) that
stands for /li'talja/ l'Italia and for /li'taLLa/ li taa. Neutral Italian has /l/ = () +
/c, G, S/: ('a:ce) alce
Structures
3.3. We will now look at the characteristics which go beyond simple segments,
those with syllable, length, stress, and intonation relevance.
e intermediate vocoids, (, ), were mentioned in 3.1.1; for assimilations
and consonant taxophones, the relevant characteristics are shown, including the
most common regional peculiarities, in their respective paragraphs.
3. italian
133
It has already been stated that in Italian, diphthongs are biphonemic and are
formed by joining two vocalic phonemes/phones, retaining their normal typical
realizations, with no modication, in particular of the second element, that is always distinct, even in the case of /ai, au/ (ai, au), contrarily to many languages,
above all Germanic ( 8.26-8 of NPT/HPh).
Taxophonics
3.3.1.1. It is necessary and useful to speak out against the widespread practice,
led by journalists and unrened publishers, of yielding to laziness of various forms
regarding the spelling of grammemes, : la una della, -re (in innitives) we
refer to the linguistic convention, that places an asterisk (*) in front of undesirable forms, which report a decidedly inaccurate use like in *la esattezza, *una opportunista, *della universit, *voltare pagina for the more usual and denitely
more harmonious: (&lezat'teq:qa, &unop&poRtu'nis:ta, del&luni&veRsi'ta, &voltaR'pa:Gina)
/lezat'teqqa, unopportu'nista, delluniversi'ta*, voltar'paGina/ l'esattezza, un'opportunista, dell'universit, voltar pagina
As a matter of fact, Italian is a real native language only in central Italy (where
local dialects present substantially, the same structure as the Italian language, only with dierences of register); in Tuscany, Umbria, the Marches and Latium (although their linguistic borders do not coincide exactly with administrative borders, but are a little less widespread, for the four regions and above all for the
Marches, as even the whole province of Pesaro and Urbino do not come into the
linguistic Center).
In the center, the dropping of grammeme endings [-i -e -a -o {in clear phonic
order}), even for nouns, adjectives, conjunctions, adverbs and verbs, is a normal
and natural procedure, as seen in: ancor indietro sempr'avanti er'andato quant'altri mai cinqu'anni or'ott'e trenta (o&Rttet'tren:ta), Dant'Alieri .
For words in connected speech (unless there are particular reasons, such as highlighting, or emphasis, which should be present or risk lacking expressiveness), neutral pronunciation resorts to elision, the dropping of one of two identical V (including the dropping of the nal V of the articles and some pronouns {considered
by grammar, in front of even dierent V}): (lat'te:za, &unat&tivi'ta, le&zeRcitaq'qjo:ni,
aL&Lita'lja:ni, su&naRgo'men:to la'mi:ka, u'nal:tRa, u'nal:tRo, lin'seN:No) /lat'teza, unattivi'ta*, lezercitaq'qjoni, aLLita'ljani, sunargo'mento la'mika, u'naltra, lin'seNNo/
l'attesa un'attivit l'esercitazioni a'italiani s'un argomento l'amica un'altra
l'inseo.
If one is not inuenced by spelling without elision, and one is, at the same time,
free from non-central and non-neutral sounds, one can easily see that normal pronunciation really is not: *(laat'te:za, &unaat&tivi'ta, le&ezeRcitaq'qjo:ni, &aLLi&ita'lja:ni,
su&unaRgo'men:to laa'mi:ka, &una'al:tRa, liin'seN:No), even for the words written as:
le esercitazioni, a italiani, su un argomento, li inseo.
For this reason, it is more advisable to use writing which accurately considers
these facts, explicitly showing how closely linked genuine pronunciation and spell-
134
ing elision is (indicated by an apostrophe; whereas, it is dened as truncation, if

there is no apostrophe). e following is by no means a counterexample: (sa'peres
'kri:veR) /sa'peres 'krivere/ sapere &rivere, with no vocalic omissions, as it falls perfectly, into the genuine Italian structure (based on the use of central Italy), which
barely tolerates things such as (sa'pers 'kri:veR) /sa'pers 'krivere/ saper &rivere, but
prefers: (&sapes'kri:v{eR}) /sapes'krive{re}/ sap &rive(re), which are of local/dialectal use.
3.3.1.2. Amongst the bad habits we have just examined, unfortunately we can
nd, propagated by schools, the (humiliatingly mechanical, without the slightest consideration, resulting in a damaging conditioned reex) use of the notorious
euphonic d well-sounding d, however, the only euphonic part of it is its
high-sounding name as in: *ad Emilia, *ed an>e, *od altro, (not to mention: *ed
educazione, *ad Adele, *od odore!).
Once more, the central way of speaking, which is not ruined by schools or journalism, spontaneously and correctly, gives: (ae'mi:lja, e'a:ke, o'al:tRo) /ae'milja, e'anke, o'altro/ a Emilia, e an>e, o altro and even: e educazione, a Adele, o odore
(not including the really cacophonous d, forming sequences such as /dd/).
Contrarily, neutral language, prefers cases such as: (&ada'ko:na, e&deve'li:na, &odol'fat:to) /adan'kona, edeve'lina, odol'fatto/ ad Ancona ed Evelina od olfatto.
However, we obviously have: (e'E:va, o'Ot:to) /e'Eva, o'Otto/ e Eva o otto, as the V
in these cases are not the same at all (phonically speaking)!
e cases of the preposition da and of pronouns una and uno as well followed
by an adjective are clearly dierent. ey must never be elided (or truncated for
uno), as they are necessary to maintain an important distinction between the preposition di and the articles (written una uno as well): (dan'da:Re) /dan'dare/ d'andare is only di andare, not da andare; in the same way, (u&nita'ljana, &unan'ti:k)
/unita'ljana, unan'tiko/ un'italiana un antico are nouns; hence, dierent from: (&una&ita'ljana, &unoan'ti:k) /unaita'ljana, unoan'tiko/ una italiana uno antico, as in:
ne cerco (for instance: canzone, mobile) I'm looking for an Italian song/piece of
furniture.
A last reection, linked to spelling and school wrongdoings, has to be cast regarding the absurd complication of insisting on teaching the exception to the
rule, where the pronoun s is written with an acute (accent) in order to dier
from se (conjunction and also allotrope, or variant, of s in front of other weak pronouns or ne): (&sevveR'ra;i, &sela'ri:d, &sene'va) /sevver'rai, sela'ride, sene'va*/ se verrai
se la ride se ne va
It is not at all clear why, once the spelling rule is formulated, to distinguish s
from the other se, one must rack one's brains to excogitate the exception to s
stesso, s medesimo which some grammars and many pedants would rather we
wrote *se stesso and *se medesimo, with no good reason to do so. Indeed, we cannot exclude the possibility of nding (even in an ambiguous initial position) sentences such as: Se stessi male, non potrei venire If I were ill, I couldn't come, which
is dierent from: S stessi malediranno azzeccagarbu della grammatica e
pettifogging lawyers of grammar will curse themselves; or: Se medesime compae
3. italian
135
di gioco perdono, vengono e&luse If equal playmates lose, they are excluded,
dierent from: S medesime comparano a tutte le altre ey compare themselves
to all the others.
erefore, eliminating the accent from s is not only utterly pointless, but causes awkward ambiguity, as well as undue analogies that lead to many people writing *a se stante, instead of the only form possible: a s stante separate, apart.
Let us briey turn our attention to the improper use (decidedly incorrect
proudly lead by a fair number of intellectuals) of *e non *o non, as in: *italiani
e non. In Italian, the negative non /non/ not has to be followed by the term
which is denied, as in: giallo, non verde; cotto, non crudo; un lo, non due (li);
coa, non petto; essere, o non essere? Instead, the absolute negation (hence, nal in
the sentence) is only no /'nO*/ no (Eng. /'nOU/ ('n;, 'n;)) even if the given
term is not expressed or repeated. erefore, logically, we must even have: italiani
e no. Is it possible to imagine an Italic Hamlet saying *Essere, o non? or, *Io ti amo, ma tu non!? (for correct English To be, or not to be?, of course, and I love
you, but you don't). Such usage is much worse than English cases like: *I don't
want no bread *We ain't going (no more) *She don't love you (no more)
Syllabication and length
3.3.1.3. Distinctively, in Italian, length (or quantity) only concerns consonants
not vowels. rst and foremost, we must rmly repeat that this is about true gemination, not length or lengthening; hence, the only suitable way of rendering the
length of Italian C, consists even phonologically of the gemination (or doubling) of the symbol in question: (af'fat:to) /af'fatto/ aatto (&koRReG'Ges:se) /korreG'Gesse/ correggesse; certainly not: *(a'f:at:o, kor:e'G:es:e) */a'f:at:o, kor:e'G:es:e/
(even */kor:e'd:Zes:e/!), which have no reason or justication (neither theoretical
nor acoustic), and rather pose problems for the syllabication, which is: (af-'fat:to, &koR-ReG-'Ges:-se) /af-'fat-to, kor-reG-'Ges-se/.
On the subject of syllabication, apart from unsatisfactory modern phonological modes, one must recall that even traditional grammars are not the most objective quite the opposite, in fact! As well as the absurd graphic syllabic division of
-sC- (which, despite its obvious aws, has been entrusted to computers too, so, unfortunately there is no hope of changing it), from a phonic point of view, the only real division (and natural: just listen to it!) for /s0/ is after /s/, not before (as for
other sequences, which are heterosyllabic; therefore, dierent from /0j, 0w, 0r,
0l/, which are all tautosyllabic, in neutral Italian): ('pas:-ta) /'pas-ta/ pasta >pa-sta!
('par:-te) /'par-te/ parte >par-te ('al:to) /'al-to/ alto >al-to ('pas:-so) /'pas-so/ passo
>pas-so (but: ('ma:-Rjo) /'ma-rjo/ Mario >Ma-rio ('a:-kwi-la) /'a-kwi-la/ aquila >a-quila ('a:-pRo) /'a-pro/ apro >a-pro ('du:-pli-c) /'du-pli-ce/ duplice >du-pli-ce.
It is also true that in northern Italian, especially in the more typical, broader regional accents, the structures /0j, 0w, 0r, 0l/, after a stressed V are very often heterosyllabic: ('a;k-wi-la) for ('a:-kwi-la) /'a-kwi-la/ aquila ('a;p-Ro) for ('a:-pRo) /'a-pro/
apro ('ma;R-jo) for ('ma:-Rjo) /'ma-rjo/ Mario ('du;p-li-c) for ('du:-pli-c) /'du-pli-ce/
duplice.
136
is aspect can be hard to grasp, especially if accurate transcriptions are not

availed of (and obviously, if no keen listening is made); so it is quite a common
problem for many actors, dubbers and presenters of northern descent, who as
voice professionals are somewhat lacking; unfortunately they are more like
semi-professionals
As seen, phonically we have (bas-'ta:-Re) /'bas-ta-re/ bastare, but also (las-'tO:-Rja)
/las-'tO-rja/ la storia therefore, likewise (s'tO:-Rja) /s'tO:rja/ storia with (s't) /s't/, as
when there is a V in front the phono-syllabication is (s-'t) /s-'t/, as in the second
example [la storia] Even acoustic data conrm the fact that (|s't) /|s't/ (after a
pause, or silence) is part of the same syllable (a little particular, possibly, on the
scale of syllabicity, but nothing really surprising) whereas, obviously, (s't) /s't/
constitute two phono-syllables bordering between two C ( 12.2-6 of NPT/HPh]
3.3.1.4. From a phonetic point of view, neutral Italian, in an intoneme, undergoes a lengthening of the last element of the phono-syllable (with one exception,
which will be looked at presently): ('fa:-Re) /'fa-re/ fare (de-'ci:-z) /de-'ci-zo/ deciso
(Ri-'pE:-te-Re) /ri-'pE-te-re/ ripetere (&Ri-ka-&pi-to-'la:-Re) /ri-ka-pi-to-'la-re/ ricapitolare
('kan:-to) /'kan-to/ canto ('mol:-to) /'mol-to/ molto ('ver:-de) /'ver-de/ verde ('pos:to) /'pos-to/ posto (>po-sto!), (&i-kon-'tran:-do) /in-kon-'tran-do/ incontrando (aR-&Rive-'der:-ci) /ar-ri-ve-'der-ci/ arrivederci.
We must be careful not to misunderstand the meaning of in an intoneme,
which should be understood as in a prominent position in a sentence; this does
not mean only at the end of a sentence, but every time there is an intoneme
even in the middle of syntactic sentences, which after all has very little to do
with phonic sentences, which are decidedly richer and more variable than traditional-grammar sentences (and even generative-grammar ones).
e exception (to the rule of phonetic length) is made up of a single nal
stressed V, which is always short (except for possible emphatic or communicative
reasons): (ko'zi) /ko'zi*/ cos (kaf'fE) /kaf'fE*/ ca however: (poR'ta;i) /por'tai/ portai (&paRa'ti;a) /para'tia/ paratia, as in the middle of the word as well: ('ka;uza) /'kauza/ causa (in'trO;ito) /in'trOito/ introito.
A partial exception is formed by the nal stressed /r/ sequence in an intomene:
('fa:r{e}) /'far{e}/ far(e) (&Rive'de:r{e}) /rive'der{e}/ riveder(e) in other contexts, we
have: (pe'Ro:Ra, peR'di:R) /pe'rora, per'dire/ per ora per dire e other nal C in an
intoneme, are lengthened as they are in stressed checked syllables within a word:
(peR'don: peR'do:no) /per'don{o}/ perdon(o) (ka'nal: ka'na:le) /ka'nal{e}/ canal(e).
Even geminate C in an intoneme have the rst element lengthened: ('sas:so) /'sasso/ sasso ('vet:ta) /'vetta/ vetta ('son:no) /'sonno/ sonno ('brac:co) /'bracco/ braccio ('daq:qjo) /'daqqjo/ dazio ('peS:Se) /'peSSe/ pe&e In any other position, before the stress, there is no further lengthening: (sas'set:to, vet'to:Re, son'nam:bulo,
&peSSo'li:n) /sas'setto, vet'tore, son'nambulo, peSSo'lino/ sassetto vettore sonnambulo peolino, or in a preintoneme: ('sasso pe'zan:te, 'vetta ele'va:ta, 'sonno pRo'fon:do, 'bracco 'lu:g, 'daqqjo ob&bliga'tO:Rjo, 'peSSe 'frit:t) /'sasso pe'zante, 'vetta ele'vata, 'sonno pro'fondo, 'bracco 'lungo, 'daqqjo obbliga'tOrjo, 'peSSe 'fritto/ sasso
pesante vetta elevata sonno profondo braccio lungo dazio obbligatorio pe&e fritto
3. italian
137
In a dierent position from that in an intoneme, as also in a preintoneme, even

the vowel lengths change, dropping the semichrone ((;)): (kau'za:Re) /kau'zare/ causare ('kauze &natu'ra:li) /'kauze natu'rali/ cause naturali ( ('ka;uza) /'kauza/ causa).
3.3.1.5. However, in neutral Italian, not all C are geminable; in fact, /z, j, w/
are always single ( non-geminable, as also the xenophoneme /Z/, which has been
integrated into the Italian phonological system for centuries now): ('pO:za) /'pOza/
posa ('kwO:jo) /'kwOjo/ cuoio (a'Zu:r) /a'Zur/ jour Nevertheless, phonetically, in
the appropriate contexts, /z/ is lengthened: ('riz:ma) /'rizma/ risma.
ere are also ve self-geminant C (/N, S, L q, Q/), which, in a postvocalic position, are necessarily geminated; but with no chance of phonological opposition
with a single C: ('baN:No) /'baNNo/ bao ('laS:So) /'laSSo/ lao ('vOL:Lo) /'vOLLo/ voo ('viq:qi) /'viqqi/ vizi ('viq:qj) /'viqqjo/ vizio (oQ'QO:no) /oQ'QOno/ ozono
e reason for this is to be found in Latin; as a matter of fact, these ve phonemes
were not part of that language, and generally derive from two or more C (or from
other languages); therefore, by assimilation, the result is a geminate C ( 5.6.17 of MaPI 1.6 of DiPI]
We can observe that (only) phonetics can supply explanations about the grammar rules (which are troublesome for foreigners and for northern Italians), as the
articles (&uno, lo, Li) /&uno, lo, Li/ uno lo must be used in cases such as: (&unoS'Se:m, loQ'Qa;ino, LiQ'Qi;i, LiN'NOk:ki) /unoS'Semo, loQ'Qaino, LiQ'Qii, LiN'NOkki/
uno &emo lo zaino zii oc (as also (&unos'trac:co, los'kOp:pjo, Lis'kOL:Li)
/unos'tracco, los'kOppjo, Lis'kOLLi/ uno straccio lo &oppio &o). In fact, it is a
question of two heterosyllabic consonants, whereby the neutral Italian structure
could not tolerate more complicated sequences, as they would make these syllables too heavy to pronounce.
e remaining 15 C, /m, n p, b, t, d, k, g c, G f, v, s r l/, in a postvocalic position, can be single or geminate, distinctively, as shown in the example chosen:
('va:no) /'vano/ vano and ('van:no) /'vanno/ vanno ('fa:to) /'fato/ fato and ('fat:to)
/'fatto/ fatto ('mO:Go) /'mOGo/ mogio and ('mOG:Go) /'mOGGo/ moggio ('be:ve)
/'beve/ beve and ('bev:ve) /'bevve/ bevve ('ka:Ro) /'karo/ caro and ('kar:Ro) /'karro/ carro ('pa:la) /'pala/ pala and ('pal:la) /'palla/ palla We can see that ('ka:za) /'kaza/ casa
and ('kas:sa) /'kassa/ cassa do not constitute a minimal pair, in modern neutral pronunciation (but only in traditional pronunciation or as far as spelling is concerned).
3.3.1.6. In regional pronunciations in the whole of Italy, in checked syllables in
an intoneme, it is quite frequent to nd a length shifting from the consonant to
the vowel element: ('pa;sso, 'paasso) /'passo/ passo ('pa;sta, 'paasta) /'pasta/ pasta
('ka;nto, 'ka;to, 'ka;to, 'kaa-) /'kanto/ canto ('di;rti, 'diirti) /'dirti/ dirti ('mo;lto,
'moolto) /'molto/ molto for the neutral ('pas:so, 'pas:ta, 'kan:to, 'dir:ti, 'mol:to).
Above all in the south, it is typical to nd diphthongization, or doubling, of the
vocoid (in checked syllables and in a preintoneme as well), but over all in un-checked syllables (which are shown here generally; but can be seen in 9-15 of
MaPI and in the dialect phonosyntheses in 16 of NPT/HPh): ('pa;ane, 'pa;ne)
138
/'pane/ pane ('ve;Ido, 've;edo) /'vedo/ vedo ('do;Uve, 'do;ove) /'dove/ dove for the
neutral ('pa:ne, 've:do, 'do:ve)
In the north, geminates, do not often have sucient length, even where vowel
length does not present the lengthening mentioned at the beginning of this section: (af'fatto, af'fa;tto) /af'fatto/ aatto.
In the central-southern areas, there can be other self-geminant consonants (
3.3.1.5.); which takes place more widely for /b, G/ in parts of the centre for /j/
too: ('rOb:ba, 'rO;bba) /'rOba/ roba (lab'bar:ka, lab'ba;rka) /la'barka/ la barca ('viG:Gil, 'vi;GGil) /'viGile/ vigile (laG'Gak:ka, laG'Ga;kka) /la'Gakka/ la giacca
('nOj:ja, 'nO;jja) /'nOja/ noia (daj'jE:Ri) /da'jEri/ da ieri for the neutral ('rO:ba, la'bar:ka, 'vi:Gil, la'Gak:ka, 'nO:ja, da'jE:Ri).
In Rome (and other central areas), instead, we can nd the degemination of /r/,
in broad and typical accents: ('ko:ReRe) /'korrere/ correre (e'ro:Re) /er'rore/ errore (a'ro:ma) /ar'roma/ a Roma (with co-gemination, 3.3.2.1.), for the neutral ('kor:ReRe, eR'ro:Re, aR'ro:ma).
Co-gemination
3.3.2.1. We will now briey introduce an example of interlexical consonant
gemination, choosing the most signicant from other similar types, occurring in
neutral Italian. Traditionally it is (widely) known as raorzamento sintattico syntactic strengthening or raddoppiamento fonosintattico phonosyntactic doubling,
but it is best dened as co-gemination, the occurrence of which, in neutral pronunciation, gives us: (ak'ka:za) /ak'kaza/ a casa (faf'fred:do) /faf'freddo/ fa freddo
(v've:Ro) /Ev'vero/ vero.
In order to explain this, it is worth rst considering examples such as (Gak'ke,
tRep'pjE:de) /Gak'ke*, trep'pjEde/ giac> treppiede, which do not pose any problems as the pronunciation and the spelling correspond.
However, we can also nd cases such as (Gak'kjEs:to, tRep'pjE:di) /Gak'kjEsto,
trep'pjEdi/ gi esto tre piedi. Furthermore, we also have: (am'met:to) /am'metto/ ammetto and (am'me) /am'me*/ a me as well, which both derive from the Latin
sequences /dm/ admitto ad me giving /mm/, by assimilation, which operated
within words and between them.
erefore, one or more of the nal consonants assimilated to the rst consonant
of the following word, as the assimilation of certain dierent consonants in a word
occurred; indeed, even septem and octo gave sette and otto in Italian (/pt, kt/ = /tt/).
Hence, as we have had /am'me*/ from ad me, also /trek'kapre/ comes from tres capr. Even /E*/ (from Lat. est) causes co-gemination as in vero, as seen above.
Separately, in phonemic transcriptions, these words are indicated, as just seen,
by /*/: /'tre*, E*/ (here tre three has the phonic stress marked, because generally it
is stressed, as in (&sono'tre) /sono'tre*/ sono tre they're three, unless it immediately precedes another stress, as in tre capre three goats /trek'kapre/ (from a theoretical structure //'tre 'kapre//, unlike English /'<Ii 'gOUts/); whereas is is given with
no /'/, because it is not usually stressed in sentences (as in English).
3. italian
139
Not all Italian monosyllables have this characteristic as far as co-gemination is

concerned. For example the preposition di of does not co-geminate (and is not
stressed), /di/, whereas the noun d day co-geminates, /'di*/; on the other hand,
the imperative di' say! can co-geminate, as a second choice /'di, 'di*/: (di'kwes:ta)
/di'kwesta/ di questa of this (f], ('dik kwa'lu:kw) /'dik kwa'lunkwe/ (un) d qualunque just any day, ('di kwal'kO:za, 'dik k-) /'di kwal'kOza, 'dik k-/ di' qualcosa say
something.
3.3.2.2. Amongst the most frequent co-geminant monosyllables ( activating)
we nd a /a*/ /E*/ e /e*/ n /ne*/ se (conj.) /se*/ gi /Ga*/ pi /pju*/ l /la*/ l
/li*/ qua /kwa*/ qui /kwi*/ /ki*/ >e /ke*/ tre /tre*/: /ammi'lano/ a Milano
/Ek'kjaro/ aro, /ep'pOi/ e poi, /net'tun nel'lui/ n tu n lui, /sep'parti/ se parti,
/Gad'detto/ gi detto, /pjut'tEmpo/ pi tempo, /las'sopra/ l sopra, /kwikkon'lEi/ qui
con lei, /kikko'noSSi/ cono, /kettene'pare/ >e te ne pare, /'trep pun'tini/ tre
puntini.
e preposition da does not co-geminate in modern pronunciation: /dafi'rEnqe/
da renze (contrary to traditional {and Tuscan} pronunciation: /daffi'rEnqe/).
e following also co-geminate d do fa fu ha ho pu sa so sta sto va: /midarra'Gone/ mi d ragione, /fam'male/ fa male, /Oppa'ura/ ho paura, /pwOssa'lire/
pu salire, /sat'tutto/ sa tutto, /stOvve'nEndo/ sto venendo, /vas'solo/ va solo.
Amongst non-geminant monosyllables ( inactivating), the following must undoubtedly be included di /di/, de' /de/, i /i/, la /la/, le /le/, li /li/, lo /lo/,
/{*}Li/, e /{*}Le/, mi/me /mi, me/, ti/te /ti, te/, si/se /si, se/, ci/ce /ci, ce/
vi/ve /vi, ve/, ne /ne/, 'sta /sta/, 'sto /sto/, 'ste /ste/, 'sti /sti/, as well as the Latin
monosyllables (a de pro quo si tu v): /di'nOtte/ di notte, /de'mEdici/ de' Medici, /i'gatti/ i gatti, /la'lana/ la lana, /le'reti/ le reti, /li'prEndo/ li prendo, /losa'pevo/
lo sapevo, /Lirak'konta/ racconta, /cisene'rEnde 'konto/ ci se ne rende conto, /sta'sera/ 'sta sera /kwO'vadis/ quo vadis /vE'viktis/ v victis.
nal-stressed polysyllables ( with stress on the last syllable, or tron truncated co-geminate, even if they can lose their stress (for rhythmic reasons): (kaf'fEk
kolom'bja:no, sa&Rappar'ti:t, &toRnt'tar:di) /kaf'fEk kolom'bjano, sarappar'tito, tornOt'tardi/ ca colombiano sar partito torn tardi.
e systematic description of this phenomenon (and other similar ones, linked,
but dierent, often confused with co-gemination, in many previous treatments {or
in all those which uncritically perpetuate the hypothetical state of things}) is given in 5.6-9 (and 4.8.1.) of MaPI, and also entry by entry in DiPI
3.3.2.3. Co-gemination is part of neutral pronunciation, exactly as lexical gemination, which is marked in spelling, as in: ('Ot:to) /'Otto/ otto. However, this is not
the case in the north (natively, except in some common, set expressions, as vero,
ha detto, used by young people raised with high levels of exposure to the television). Too often, it is erroneously considered as if it were a regional characteristic
of the central-southern areas.
is opinion, but above all the fact that co-gemination does not gure in writing (apart from crystallized forms, such as giac>, davvero, soprattutto] is often
140
believed even by Italians from the central-southern areas that it is to be avoided; in that case, one should consider even normal lexical gemination (which is
distinctive, 3.3.1.5.) abnormal and unsuitable.
Clearly, in the central-southern areas, there are regional dierences which are
contrary to neutral use. In particular, the most systematic and widespread use is
met in Tuscany (which, however, is not totally homogeneous), followed by Rome;
the other central-southern areas follow, all of them with local, more or less strong
dierences.
e most normal type of co-gemination, in modern neutral pronunciation, is
generally represented by a sort of compromise or of average, based on Roman use
(including the cases of de-gemination of articles and pronouns, formed by /l/: la
lo le li l'] with further simplications. e kind of co-gemination which is more
similar to traditional pronunciation ( 3.3.5.3), follows Tuscan usage.
A few other examples follow, including one of de-gemination (given as the last
one): (pjuf'fOr:te) /pjuf'fOrte/ pi forte (cit&takkos'tjE:Ra) /cittakkos'tjEra/ citt costiera (&tRenta'trek koRRi'do:Ri) /trenta'trek korri'dori/ trentatr corridori (l'la:na)
/El'lana/ lana (&los'tes:so) /Elos'tesso/ lo stesso.
In modern neutral pronunciation, come like, as (appositive and comparative
with nouns and pronouns) is co-geminant, as well: (&komek'ka:ni, &komev've;Ri 'ka:ni, &komet'te) /komek'kani, komev'veri 'kani, komet'te*/ come cani come veri cani
come te
Stress
3.3.3.1. Generally, in Italian, stress is assigned to rhythm groups, according to
(lexical) word stresses. Rhythm groups are stress groups ( 3.2.7, 6.4.2, 12.1, 13.2-3
of NPT/HPh) made up of a syllable with strong or primary stress, and others
with weak or secondary stress, as in: (&peRfetta'men:te) /perfetta'mente/ perfettamente (in'trE:no) /in'trEno/ in treno (&peRil'ka:ne) /peril'kane/ per il cane (&vRsola&finedel'me:ze, &vRsola'fi;ne del'me:ze) /vErsolafinedel'meze, vErsola'fine del'meze/
verso la ne del mese
e last example shows that, according to the prominence given to certain parts
of sentences, the number of rhythm groups can change, as in (un&nwvo'li:bR)
/unnwOvo'libro/ un nuovo libro a new book, compared to (un'nwO;vo 'li:bR) /un'nwOvo 'libro/ a book (which is) new, written in the same way (un nuovo libro),
but with dierent prominence, corresponding to a semantic gradation, where in
the second case, nuovo takes on greater importance.
e communicative importance given to nuovo can be even greater, in an utterance that, in its written form, remains once more the same (given the considerable deciency of current punctuation): (un'nwO:vo2 'li:bR23) /un'nwOvo, 'libro./.
e introduction of the post-stress syllables of the continuative ((2) /,/) and conclusive ((23) /./) intonemes were necessary here, because the new greater prominence
is given, in this case, by the presence of two intonemes, for a single utterance. (Let
us also note the presence of the full chrone, even in the rst rhythm group, bear-
3. italian
141
ing a continuative intoneme, (O:).)

On the other hand, the communicative prominence would still increase, by using a conclusive intoneme, for the rst rhythm group too: (un'nwO:vo23 'li:bR23)
/un'nwOvo. 'libro./.
3.3.3.2. Going back to the examples of the rhythm groups given above, we can
observe that unstressed syllables (or rather, with weak stress), and those with half-stress ( with mid or secondary stress), alternate in such a way as to have one or
two weak phono-syllables amongst others with secondary (or primary) stress.
Secondary stress is assigned rhythmically, bearing in mind as far as possible
the position of the stress in the original forms ( fundamental forms, from which
they derived {certainly not in an evolutive sense, from Latin}), as occurs in lexical
compounds: (&pRtasa'po:ne) /pOrtasa'pone/ portasapone (pu&liSSis'kar:pe) /puliSSis'karpe/ puli&arpe.
In derivatives, instead, the origin counts for less than rhythmic reasons, even if,
in case of a double possibility (rhythmic or original), the stress of the original form
can exert a decisive inuence: (&pata'ti:n) /pata'tine/ patatine (notwithstanding
(pa'ta:ta) /pa'tata/ patata), (&atten'qjo:ne) /atten'qjone/ attenzione (notwithstanding (at'tEn:to) /at'tEnto/ attento]
For syllables that follow a stressed syllable of a word, furthermore, they behave
as follows: ('fab:bRo) fabbro, ('fab:bRika) fabbrica, ('fab:bRika&no, 'fab:bRika&mi) fabbricano fabbricami, ('fab:bRikame&lo) fabbricamelo (let us also add a rather improbable ('fab:bRika&mice&lo) fabbricamicelo the end of 6.4.2 in NPT/HPh]
3.3.3.3. As regards the syllables of a word that precedes a stressed syllable, the
spontaneous formation of a rhythmic alternation of weak and half-strong stresses
is formed, going back towards the beginning of the word, starting from the strong
syllable, of this kind: ({&}&&'). However, given that polysyllabic words are
generally derivatives or compounds, there is a clear and agreed tendency to place
secondary stresses on the same syllables of the simple forms of the lexeme which
are originally stressed.
e only exception, so to speak, is made up of the rhythmic tendency that interrupts both sequences of weak syllables that are too long, introducing some secondary stresses, and sequences of stressed syllables ( strong and half-strong), suppressing some secondary stresses or shifting them one or two syllables.
Some examples: (&soce'ta) societ, (ka&paci'ta) capacit, (pRo&babili'ta) probabilit, (be&ati'tu:din) beatitudine, (&Raqqjo&naliQ&Qabili'ta) razionalizzabilit, (aR&tifi&cali'ta) articialit, (u&tilita&Ristika'men:te) utilitaristicamente, (e&zeRcita'to:Re) esercitatore, (aR&tikolaq'qjo:ne) articolazione, (as&socaq'qjo:ne) associazione, (o&ceano'gra:fiko) oceanograco, (ka&RatteRiQ'Qa:bile) caratterizzabile, (in&telliGen'tis:sima)
intelligentissima, (oS&Sillo'gram:ma) ollogramma, (ef&feRveSSen'tis:sim) eerve&entissimo
Other examples: (do&loRo&zissima'men:te) dolorosissimamente, (as&tuta'men:te) astutamente, (im&maGinaq'qjo:ne) immaginazione, (mo&difikaq'qjo:ne) modicazione, (kom&poziq'qjo:ne) composizione, (ak&kjappafaR'fal:le) acappafarfalle, (e~&c-
142
falo'gram:ma) encefalogramma, (&ultRakon&seRva'to:Re) ultraconservatore, (&inteRkon&tinen'ta:le, -&konti-) intercontinentale, (&pRemedi&tata'men:te, pRe&me-) premeditatamente, (pRe&kce'men:te, &pReko-) precocemente.
3.3.3.4. Considering isolated words, normal ( neutral, spontaneous) secondary stressing complies to the rules of lexical composition and derivation. However,
obviously, there are also dierences in the distribution of secondary stresses, due
to the rhythmic context in which the words are placed at any given moment.
erefore, if in isolation a form like partitocrazia is (paR&titokRaq'qi;a) whereas (&paRti%tokRaq'qi;a) is a false note, deriving from the lack of compound analysis in actual sentences, there is some alternation: (&lapaR&titokRaq'qi;a, la&paRti&tokRaq'qi;a) la partitocrazia ('molto as&tuta'men:te) molto astutamente, (pRo'cE;deRe
&astuta'men:te) procedere astutamente ('vEgo &mRkole'di) vengo mercoled (can'drOm meR&kole'di) ci andr mercoled
We will not broach the subject of two possible stress patterns for some words,
: /dia'triba di'atriba/ diatriba /skle'rOzi s'klErozi/ &lerosi /te'zEo 'tEzeo/ Teseo
/e'dipo 'Edipo/ Edipo; the interested reader is referred to DiPI.
3.3.3.5. Neutral Italian does not in any way accept strong stress on two contiguous syllables (in an intoneme), but attenuates the rst, even by shifting it or, if necessary, it shifts it (without attenuation), as in the third (and fourth) variants given, which are possible, but less common: /fa'rOk 'kwesto = farOk'kwesto/ (&faRk'kwes:to), (fa&Rk'kwes:to), ('faRk 'kwes:to), ('fa;Rk 'kwes:to) far questo, /skoper'kjO
il'tetto = skoperkjOil'tetto/ (s&kopeRkjil'tet:to), (s&kopeR&kjil'tet:to), (s'kopeRkj il'tet:to), (s'ko;peRkj il'tet:to) &oper il tetto, /nonnepo'tep 'pju* = nonnepotep'pju*/ (&nonne&potep'pju), (&nonnepo&tep'pju), (&nonne'potep 'pju), (&nonne'po;tep
'pju) non ne pot pi, /bEn'kek k'reda = bEnkek'kreda/ (&bkek'kre:da), (b&kek'kre:da), ('bEkek 'kre:da) ben> creda, /me'tad 'dOze = metad'dOze/ (&metad'dO:ze),
(me&tad'dO:ze), ('metad 'dO:ze), ('me;tad 'dO:ze) met dose
One aspect of attenuation, owing to rhythmic reasons, can concern the moving
back of the reduced stress too, as seen in the following examples: /'vEngo 'dopo di'te* = 'vEngo dopodi'te*/ ('vEgo &dopodi'te), ('vEgo do&podi'te) vengo dopo di te
/'karo pa'pa* = karopa'pa*/ ('ka;Ro pa'pa), (&kaRopa'pa), (ka&Ropa'pa) caro pap /lo'facco 'anke per'te* = lo'facco ankeper'te*/ (lo'facco &akepeR'te), (lo'faccoa&kepeR'te) lo faccio an>e per te /si'kjama 'forse ko'zi = si'kjama forseko'zi*/ (si'kja;ma
&foRseko'zi), (si'kja;ma foR&seko'zi) si ama forse cos?
3.3.3.6. In the case of syntagms like //fuG'Gi z'vElto// /fuGGiz'vElto/ (&fuGGiz'vEl:to, &fuGGiz'vEl:to fuG&Giz'vEl:to) fugg svelto he ran away quickly, compared
to similar expressions as //'fuGGi z'vElto// /'fuGGiz 'vElto/ ('fuGGiz 'vEl:to) fuggi
svelto run away quickly!, there are slight phonetic and prosodic dierences,
which alone, however, are not always able to maintain the distinction, as even in
the case of fuggi svelto, attenuation of the rst stress is possible, above all in fast
pronunciation (&fuGGiz'vEl:to). Co-gemination does not apply with syllable structures like /_0/, which actually are syllabied /_0/.
3. italian
143
In the case of //fuG'Gis 'subito// /fuGGis'subito/ (&fuGGis'su:bit fuG&Gis'su:bit) fugg sbito he ran away at once, compared to /'fuGGi 'subito/ ('fuGGi 'su:bit) fuggi sbito run away at once!, the co-gemination contributes a better distinction, in neutral pronunciation: //pa'gOt 'tutto// /pagOt'tutto/ (&pagt'tut:t, 'pagt 'tut:t, 'pa;gt 'tut:t pa&gt'tut:t) pag tutto /'pago 'tutto/ ('pa;go 'tut:t, &pago'tut:t) pago tutto //losen'tim 'male// /losentim'male/ (lo&sentim'ma:le, lo'sentim
'ma:le &losen&tim'ma:le) lo sent male /lo'sEnti 'male, losEnti'male/ (lo'sEnti 'ma:le, lo&snti'ma:le) lo senti male Lastly: //lostrap'pOv 'via// /lostrappOv'via/ (los&tRappv'vi;a, los'trappv 'vi;a &lostRap&pv'vi;a) lo strapp via /los'trappo 'via, lostrappo'via/
(los'trappo 'vi;a, los&tRappo'vi;a) lo strappo via
Intonation
3.3.4. 3.3 shows the preintonemes and intonemes in neutral Italian (both
modern and traditional), which we will illustrate with simple examples:
/./: (Ri'partono 'tutti 'sa:bato23) /ri'partono 'tutti 'sabato./ Ripartono tutti sabato.
/?/: (Ri'partono 'tutti 'sa:bato21) /ri'partono 'tutti 'sabato?/ Ripartono tutti sabato?
//: (&seRRi'partono 'tutti 5sa:bato12| &mene'va;do a'ki;23) /serri'partono 'tutti 'sabato mene'vado an'kio./ Se ripartono tutti sabato, me ne vado an>'io.
In the north, regional intonation has consistent and varied shifts; in the south
they are rather repetitive; while, in the center, they are more contained (and, generally, more similar to neutral patterns). e major dierences are found in the suspensive intoneme, //, as can also be seen in the phonosyntheses of 16 of NPT/
HPh (and of 10-15 of MaPI).
3.3. Italian preintonemes and intonemes.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 5 1 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
Traditional neutral pronunciation

3.3.5.1. As already mentioned (in 3.0.1), traditional neutral pronunciation
mainly follows Tuscan use, in particular orentine, and dates back to the unication of Italy; therefore to about a century and a half ago, as quoted in dictionaries
144
which, at that time were dened as of Tuscan language, more often than of Italian.
e peculiarities of traditional pronunciation, therefore, are more rigid than
those of modern pronunciation (having less variations {and here, we relate above
all the strangest} and at the same time most capricious). The forms followed
by * are found only in some dictionaries or specic lists.
As regards the vowel phonemes /e, E o, O/, some examples of words and endings
follow. e more bizarre forms are marked with simple (orthoepical) written accents: Ase annsso aurra* Bertldo* co ccca corner, notch, colltta conntto crsima enrme esso flla Girgio gtta grppo lttera mccolo nsso
nrma* rgano* rcca dista, Rmolo sgmino site strpio strnna t&hio
as well as the endings: amarolo sttti stttero
As far as the distribution of /q, Q/ is concerned, the major traditional characteristic regards initial /q/ for many words like: zampa zio zoppo zuc>ero zucca
and internal /Q/, in cases such as: aguzzino amazzone brezza ribizzo intirizzito lazzi lezzo olezzo pettegolezzo ribrezzo rubizzo &orza arzo* sozzo*
3.3.5.2. e greatest dierence, however, concerns the use of single /s/, within
a word, between V (even if with /j, w/ before the second). e most signicant cases (including derivatives) are: asino casa esi uso cosa cos desidero desiderio (il) fuso mese naso peso Pisa posa raso riposo riso, and adjective endings
-ese -oso and verb endings -esi -osi\ /in'glese/ inese (but /fran'ceze/ francese!) /cinese'ria/ cineseria (but /borge'zia/ bor!esia!), /go'loso/ goloso /cellu'losa/ cellulosa
/kurjosi'ta*/ curiosit /'presi, -sero, -so/ presi presero preso (but /'lezi, -zero, -zo/ lesi
lesero leso!), /impre'sarjo/ impresario /'rosi, -sero, -so/ rosi rosero roso /ris'posi, -sero/ risposi risposero /ro'sikkjo/ rosico (but /ero'zjone/ erosione!).
However, in traditional pronunciation, we nd /z/ in cases such as: bisoo caso esa muso paese quasi sposa viso misi misero (io) fusi fusero (ho) fuso incisi incisero inciso
3.3.5.3. Traditional co-gemination is more systematic and more widespread
than modern co-gemination; indeed, it has a higher number of activating forms
(as da\ da Milano /dammi'lano/; for the modern: /dami'lano/), and does not provide for de-gemination la mia /Ella'mia/; modern: /Ela'mia/); what is more, the
following are always co-geminant, too, come dove qual>e sopra (unless they are
used as nouns): come si fa? dove vai? qual>e volta sopra quel tavolo /komessi'fa, dovev'vai, kwalkev'vOlta, soprakkwel'tavolo/ (but: il sopra del tavolo /il'sopra del'tavolo/); modern: /komesi'fa*, dove'vai, kwalke'vOlta, soprakwel'tavolo/).
As far as accentuation is concerned, there are no particular peculiarities; any
dierences are due to the fact that generations have passed, and, as one knows, the
accentuation of learned terms and classical nouns are subject to trends of preference, which can be agreed upon to a greater or lesser extent, but are not under discussion: they are merely documentable ( the end of 3.3.3.4).
Intonation does not change, except for the possible adoption (by some) of Tuscan-type preintonemes ( the relative phonosyntheses in 16).
3. italian
145
3.3.5.4. All characteristics of traditional pronunciation may be found, however, in MaPI and DiPI this pronouncing dictionary shows the variants of the center (the linguistic center; but Abruzzo belongs to the upper south) regarding the
distribution of /e, E o, O s, z q, Q/ (including some peculiar stress patterns), indicating Tuscany, Umbria, the Marches, Latium and Rome respectively with: t, u,
m, l, r (in square brackets).
Traditional variants are identied, in DiPI, by a full-stop which appears before
them, and which separates them from modern variants, given rst (if dierent,
otherwise they coincide), as, for example: lettera /'lettera. -Et-/ { E, e},
/'lEttera/ in Tuscany and /'lettera/ in the rest of the center.
ere are at times more intricate variants, : storpio /s'torpjo. -O-/ { O/o, o/O,
o}. However, there is full agreement throughout the center for bene /'bEne/.
Text
3.4.0. e story e North Wind and the Sun follows, given in eight dierent
normalized versions. We start with the (neutral) Italian pronunciations of (neutral British) English this is the rst step of the phonetic method (the written text
is given in 2.5.2.0). e Italian translation follows, in its neutral modern and traditional versions. ree further accents are given: two are mediatic (respectively
northern and central of Milan and Rome), which, for segmental and intonational aspects, are not as broad as those presented with vocograms and tonograms
in 10 12 of MaPI (in which there are 23 regional versions, in 10-14).
e respective (dialectal) phonosyntheses in 16 of NPT/HPh (looking at Milanese, 16.15, and Roman, 16.42) can also provide useful preliminary information. In northern mediatic pronunciation, stopstrictives often occur as sequences: (ts, dz TS, DZ) (special symbols would be better: ((tS, dZ))).
Lastly, the aected accent (snobbish, 3.4.9) is examined, as well, with a French
r (generally, an approximant: labiodental, (V), or uvular, (), or uvularized labiodental, (), which has been chosen as a representative type, having both components) and other characteristics linked to that kind of pronunciation, as tenser
but also laxer V, at the same time ( 3.4).
ey are realized basically by higher and more advanced vocoids than normal, which, in stressed syllables (and in preintonemes as well), are diphthongized,
but with second elements rather central in their vocogram boxes, therefore, ending in dierent opposite positions from the initial ones.
Furthermore, /c, G S, {Z}/ are postalveo-palatal over-rounded, (&, 1, , {});
there is also drawling and splitting of the tonic in the real tonetic sense and
posttonic syllables (sometimes even of the previous ones), in particular before a
pause, and above all for intonemes when they are dierent from the conclusive
ones.
Lastly, we note a bigger pitch extension (which we will indicate paraphonically, with ()@, at the beginning of each paragraph) and creaky voice in intonemes
( tonic and posttonic syllables), which is represented by the specic diacritic: ().
146
3.4. Aected realizations of the Italian vowels.

/i/ (i[i])
/e/ (e[e]),
({'ii/'uu))
/E/ (EE), ({'})
/u/ (u[u])
/o/ (o[o]),
({'ii/'uu))
/O/ (OO), ({'})
/a/ (a[a])
3.4.1. At the end, as always, there is the version which gives the English pronunciation of Italian, by neutral British speakers, uent in Italian (after prolonged contact with native speakers, but with no help from the phonetic method), who have
adequately learned the relative prominences, but who substantially use segmental
and intonation elements which are typical of neutral British English (for reference
purposes, although, of course, a neutral accent is not so common). Obviously, the
same principle is valid for the foreign pronunciation of English, given rst.
Speakers of American English could prepare their own version of the Italian pronunciation of American English, with /a/ even for English // ( ('Sat:$) instead of
('SOt:$), for shot) and so on, as an excellent exercise, by listening to native speakers,
best of all after recording them. Of course, speakers of other languages could prepare both their own pronunciation of Italian and the foreign Italian accent of their
language, as well. e author would be happy to receive their transcriptions and
recordings, both in case of help should they need it and to make their contribution known to others (possibly in our website on canIPA Natural Phonetics
0.12).
Italian pronunciation (of English)
3.4.2. (&denR'twin dende'san:$2 &weRdis'pju:tig$2 wic'wOz des'trO:geR23| 'wE na'trE:velleR2 &kima5lO:g$12| 'rEp$ti na'wOr:m 'klOk:$23| &deja'grid:$2 dtdi'wa nu'fErs[t] sak'si:dd$2 im'mEikin de'trE:velleR2 'tEi kisklOk5kOf:$12| Sud'bi kon'si:deRd$2 s'trOgeR 'dEndi
'a:deR23||
'dEn:2 denR'twin:d$ 'blu2 'zar dezi'kud:$23| 'bat$ de'mO:Ri 5blu12| de'mO:r2 'klOz:li2 'did
de'trE:velleR2 'fOl dis'klOk:2 ka'raundim$23| &ndt5las:t$12| &denR'twin:d$2| ge'vap$ djat'tEmp$t$23| 'dEn de'san:2 So'na;ut$2 'wOr:m$li23| &ndim'mi:djat$li2 de'trE:velleR2 tu'kOf fis'klOk:$23| end$'sO2 denR'twin:d$2 &wzo'blaiG$d$ &tukoM'fEs:2 'dEtde 'san$ wozdes'trOge
&Rvde'tu23|
'di;dju 'laik$21 des'tO:Ri2| &duju'wOn tu'i;Ri ta'gEn:$21|||)
3. italian
147
Italian text
3.4.3. Si bisticciavano un giorno il vento di tramontana e il sole, l'uno pretendendo d'esser pi forte dell'altro, quando videro un viaggiatore, >e veniva innanzi, avvolto nel mantello. I due litiganti decisero allora >e sarebbe stato pi forte fosse
riuto a levare il mantello al viaggiatore.
Il vento di tramontana cominci a soare con violenza; ma, pi soava, pi il
viaggiatore si stringeva nel mantello; tanto >e alla ne il povero vento dovette desistere dal suo proposito. Il sole allora si mostr nel cielo, e poco dopo il viaggiatore, >e
sentiva caldo, si tolse il mantello. E la tramontana fu costretta cos a ricono&ere >e
il sole era pi forte di lei.
T' piaciuta la storiella? La voamo ripetere?
Modern neutral pronunciation
3.4.4. (si&bistic'ca:va&no2 u~'Gor:no2| il'vEnto di&tRamon'ta:na2| eil'so:le23 'lu:n2
&pReten'dEndo &dsseRpjuf'fOr:te2 del'lal:tRo23 &kwando'vi;de&Ro uMvi&aGGa'to:Re23 &kevve'ni;va in'nan:qi2 av'vOlto &nelman'tEl:lo23|| i&dueliti5gan:ti12 de'ci:zeR2 al'lo:Ra2 &kessa&Rbbes'ta;to pjuf5fOr:te12| kif&fosseRiuS'Si:t2 alle'va;Re ilman'tEl:lo2 alvi&aGGa'to:Re23||
il'vEnto di&tRamon'ta:na2 &komi~'cO assof'fja:Re23 &koMvio'lEn:qa23| map'pjus sof5fja:va12| &pjuilvi&aGGa'to:Re2 &sistRi~'Ge;va &nelman'tEl:lo23\ 'tan:to2 ke&alla5fi:n12 il'pO;veRo
'vEn:to2 do'vette de'zis:teR23 dal&suopRo'pO:zito23|| il'so:le2 al'lo:Ra2 &simos'trOn ne'cE:lo23| ep&pko'do:po2 ilvi&aGGa5to:Re12 &kessen'ti;va 'kal:do2 si'tOl:se23 ilman'tEl:lo23| &ela&tRamon5ta:na12 &fukkos'tret:ta2 ko'zi2| aR&Riko5noS:SeRe12| keil'so:le2\ &Rapjuf'fOr:te23 di'lE;i23||
&tppja'cu:ta21 &lasto'rjEl:la2| &lavoL'La;mo Ri'pE:teRe21|||)
Traditional neutral pronunciation
3.4.5. (si&bistic'ca:va&no2 u~'Gor:no2| il'vEnto di&tRamon'ta:na2| eil'so:le23 'lu:n2
&pReten'dEndo &dsseRpjuf'fOr:te2 del'lal:tRo23 &kwando'vi;de&Ro uMvi&aGGa'to:Re23 &kevve'ni;va in'nan:qi2 av'vOlto &nelman'tEl:lo23|| i&dueliti5gan:ti12 de'ci:zeR2 al'lo:Ra2 &kessa&Rbbes'ta;to pjuf5fOr:te12| kif&fosseRiuS'Si:t2 alle'va;Re ilman'tEl:lo2 alvi&aGGa'to:Re23||
il'vEnto di&tRamon'ta:na2 &komi~'cO assof'fja:Re23 &koMvio'lEn:qa23| map'pjus sof5fja:va12| &pjuilvi&aGGa'to:Re2 &sistRi~'Ge;va &nelman'tEl:lo23\ 'tan:to2 ke&alla5fi:n12 il'pO;veRo
'vEn:to2 do'vEtte de'sis:teR23 dal&suopRo'pO:zito23|| il'so:le2 al'lo:Ra2 &simos'trOn ne'cE:lo23| ep&pko'do:po2 ilvi&aGGa5to:Re12 &kessen'ti;va 'kal:do2 si'tOl:se23 ilman'tEl:lo23| &ella&tRamon5ta:na12 &fukkos'tret:ta2 ko'si2| aR&Riko5noS:SeRe12| keil'so:le2\ &Rapjuf'fOr:te23 di'lE;i23||
&tppja'cu:ta21 &lasto'rjEl:la2| &lavoL'La;mo Ri'pE:teRe21|||)
148
Northern mediatic pronunciation (Milanese)

3.4.6. (i&bitit'tSa:vano2 ud'Zo;Rno2| ilveto di&tRamo'ta:na2| eilo:le3 3 'lu:no2
&pRetededo &djeRpju'fO;Rte2 della;ltRo3 3 &kwadovi;deRo u&vjaddZato:Re3 3 &keveni;va in'na;tsi2 avvO;lto &nlmatE;llo3 3|| i&duelitiga;ti32 det'Si:eRo2 al'lo:Ra2 &kea&Rbbe'ta;to pjufO;Rte32| ki&foeRiu'Si:to2 ale'va;Re ilma'tE;llo2 al&vjadZato:Re3 3||
ilveto di&tRamo'ta:na2 &komi'tSO aoffja:Re3 3 &kovjole;tsa3 3| ma'pju offja:va32|
&pjuilvjaddZa'to:Re2 &itRidZe;va &nlmatE;llo3 3\ 'ta;to2 ke&allafi:ne32 ilpO;veRo
've;to2 dovEtte dei;teRe3 3 dal&uopRopO:ito3 3|| il'o:le2 al'lo:Ra2 &imo'tRO ntSe:lo3 3| e&pko'dO:po2 il&vjaddZato:Re32 &keeti;va 'ka;ldo2 itO;le3 3 ilmatE;llo3 3| &ela&tRamota:na32 &fuko'tRE;tta2 ko'i;2| a&Rikono:SeRe32| keil'o:le2\ &eRapjufO;Rte3 3 dilE;i3 3||
&tjpjatSu:ta21 &lato'RjE;lla2| &lavoja;mo Ripe:teRe21|||)
Central mediatic pronunciation (Roman)
3.4.7. (sib&bistic'ca:vano2 u~'Go;rno2| il'vEnto di&Ramon'ta:na2| eil'qo:le23 'lu:no2
&Reen'dEndo &dsseRpuf'fO;rte2 de'la;ltRo2 &wando'vi;de&Ro uMvi&aGGa'o:Re23 &evve'ni;va in'na;nqi2 av'vOlto &nelman'tE;llo23|| i&duelii5ga;nti2 2 de'Si:seRo2 al'lo:Ra2 &essa&Rebbes'ta;op puf5fO;rte2 2| kif&fosseRiuS'Si:o2 alle'va;Re ilman'tE;llo2 alvi&aGGa'o:Re23||
il'vEnto di&Ramon'ta:na2 &omi~'cO assof'fa:Re23 &oMvio'lE;nqa23| map'pus sof5fa:va2 2| &puilvi&aGGa'o:Re2 &sistRi~'Ge;va &nelman'tE;llo23\ 'ta;nto2 e&ala5fi:ne2 2 il'pO;veRo 'vE;nto2 do'vette de'si;steRe23 dal&quoRo'O:sio23|| il'qo:le2 al'lo:Ra2 &simos'trO ne'cE:lo23| ep&po'dO:o2 ilvi&aGGa5o:Re2 2 &essen'ti;va 'a;ldo2 si'tO;lqe23 ilman'tE;llo23|
&ela&Ramon5ta:na2 2 &fukkos'tre;tta2 o'si;2| a&Rio5noS:SeRe2 2| eil'qo:le2\ &Rappuf'fO;rte23
di'lE;i23||
&tppa'Su:a21 &lasto'rE;lla2| &lavoj'ja;mo Ri'E:eRe21|||)
Aected pronunciation ( 3.4.0)
3.4.8. ()@ (si&bisti&'&a;aa&co2 u~'1oocoo2| il'vEnto di&tamon'ta;acaa2| eil'so;oVe23 'lu;uc2 &peten'dEEndo &dssepjuf'fOOte2 del'laaVto23 &kwando'vi;ide&o uMvi&a11a'to;oe23 &kevve'ni;iva in'naacqi2 av'vOOlto &nelman'tEEVVo23|| i&dueliti5gaacti12 de'&i;ie2 al'lo;oa2 &kessa&bbes'taato pjuf5fOOtee12| kif&fosseiu'i;it2 alle'va;ae
ilman'tEEVVo2 alvi&a11a'to;oe23||
()@ il'vEEnto di&tamon'ta;aca2 &komi~'&OO assof'fja;ae23 &koMvio'lEEcqa23| map'pjus sof5fja;aaa12| &pjuilvi&a11a'to;oe2 &sisti~'1e;eva &nelman'tEEVVo23\ 'taacto2 ke&alla5fi;ic12 il'pOOveo 'vEEcto2 do'veette de'ziiste23 dal&suopo'pO;Oito23|| il'so;oVe2
3. italian
149
al'lo;oa2 &simos'tOOn ne'&E;EVo23| ep&pko'do;opo2 ilvi&a11a5to;oe12 &kessen'ti;iva

'kaaVo2 si'tOOVse23 ilman'tEEVVo23| &ela&tamon5ta;aca12 &fukkos'teetta2 ko'zii2| a&iko5nooeee12| keil'so;oVe2\ &apjuf'fOOte23 di'lEEi23||
()@ &tppja'&u;uta21 &lasto'jEEVVaa2| &lavoL'La;amo i'pE;Eteee21|||)
British pronunciation (of Italian)
3.4.9. (s&bIisT5chA;v&n2 u~'G;[>]n2| 5vnT D&>mn'ThA;n2| 'slI3 3 5lun2 &p>ITn'DnD &Ds'phj;u 'f;[>]TI2 D'lA;>23 &kwA;nD5vIiD&>
uMvi&G'Th;>I23 &kIv5nIiv 'nA;ntsi2 5vT &nmn'Thl23| Ii&DuIlT'gA;nTi32 D'chIiz>2 'l;>2 &kIs&>bI5sTA;T &pju'f;[>]TI32| khIi&fsI>iu'SIiT2 &l5vA;>I &mn'Thl2 &vi&G'Th;>I3 3||
5vnT D&>mn'ThA;n2 &km~5ch; s'fjA;>I3 3 &kMvi'lnts3 3| m5phju s'fjA;v32| 'phj;u vi&G'Th;>I2 &sIis>~5GIv &nmn'Thl23\ 'ThA;nT2
kI&l'fIinI32 5phv> 'vnT2 D5vTI D'zsT&>I23 D&sup>'phz&T3 3||
'slI2 'l;>2 &sIim5s>; n'chl3 3| I&pk'Dp2 &vi&G'Th;>I32
&khIsn5ThIiv 'khA;D2 s'ThsI3 3 &mn'Thl3 3| &Il&>mn'ThA;n32 &fuk's>T2 k'zI;i2| &>Iik'nS&>I32 kh'slI2\ &>'phj;u 'f;[>]TI3 3 D'l;I3 3||
&Thpj'chuT21 &lsT>i'l2| &lvli5A;m >'phT&>I21|||)
American pronunciation (of Italian)
3.4.10. (s&bIisT'chA;v&n2 Uu~'G;<n2| 'vn[T] D&T<mO;n'ThA;n2| 'slI23 'lUun2 &p<ITn'DnD &Ds'phj;u 'f;<TI2 D'lA;T<23 &kwA;nD'vIiD&< UuMvi&G'Th;<I23 &kIv'nIiv 'nA;ntsi2 'vO;T &nEmn'Thl23| Ii&DUuIl[6gA;n[T]i2 2 D'chIiz<2 'l;<2 &kIs&<bI'sTA;[ &pju6f;<TI2 2| khIi&fO;sI<iUu'SIi[2 &l'vA;<I &mn'Thl2 &,vi&G'Th;<I23||
'vn[T] D&T<mO;n'ThA;n2 &kO;m~'ch; s'fjA;<I23 &kO;Mvi'lnts23| m'phju s6fjA;v2 2| 'phj;u vi&G'Th;<I2 &sIisT<~'GIv &nEmn'Thl23\ 'ThA;n[T]2 kI&l6fIinI2 2 'phO;v< 'vnT2 D'v[I D'zsT&<I23 D,&sUup<'phO;z[&23|| 'slI2 'l;<2 &sIimO;'sT<; nE'chl23| I&pk'Dp2 &vi&G6Th;<I2 2 &khIsn'ThIiv 'khA;D2 s'ThO;sI23 &mn'Thl23| &Il&T<mO;n6ThA;n2 2 &fUukO;'sT<[2 k'zI;i2| &<Iik6nS&<I2 2 kh'slI2\ &<'phj;u 'f;<TI23 D'l;I23||
&ThpjchUu[21 &lsT<i'l2| &lvli'A;m <ph[&<I21|||)
150
Italian dictionaries grammars and their (non) care for pronunciation

3.5.0. It is important to see how Italian dictionaries indicate pronunciation and
whether they are accurate and provide variants or not. In addition, a sure clue to
ascertain whether they care for pronunciation or assign it to hasty incompetents
(who follow dierent criteria do not take the trouble to verify what they are doing) consists in looking up the entry gliommero /'LOmmero/ (a kind of poem in endecasyllabics originally written in the Neapolitan dialect). ere are three kinds of
dictionaries: those which do not include the word or do not take care to distinguish between /L/ and /gli/, for the trigraph gli, as in glioma /gli'Oma/; then come
those which, because of false philological and etymological deductions, invent
*/gli'Ommero/, or copy it from some mistaken previous sources; lastly, there are
the dictionaries whose pronunciation characteristics are assigned to true experts,
who obviously give /'LOmmero/. In the dialect of Naples, a ball (of wool)/skein
(of cotton) is a glimmero /'LOmmr/, although it comes from Latin glomerum,
with /gl/, as on the other hand gland is glinnola /'Lannla/, from Latin glandulam (in Italian they are gomitolo and ghiandola, respectively, /go'mitolo, 'gjanda/, with /g/).
Even most grammars are not to be trusted excessively: it is sucient to thumb
through them with critical eyes. Besides, our judgment should not in any way be
inuenced by the fact that these texts are very widely used and well-known, even
abroad. However, when they deal with phonology and phonetics only because
they feel duty bound to do so, unfortunately we cannot be too optimistic. In fact,
some of them tend to invent nonexistent possibilities, which are frankly absurd,
as happens with some supposed dierences for such rst-person plural forms as alleviamo or spariamo. In actual fact, their pronunciation is identical, although they
belong to dierent verbs: allevare to breed or alleviare to alleviate, and sparare
to shoot or sparire to disappear, always /alle'vjamo, spa'rjamo/ (whereas certain authors try to argue that the second words in each pair are pronounced as */allevi'amo, spari'amo/).
It would be equally absurd to follow those who hypothesize the existence of
dierent pronunciations for the numbers sei, sette, otto six, seven, eight, if compared to (tu) sei you are, (le) sette (religiose) the religious sects, and Otto (a male
name)
4. French
4.0. As well as the modern neutral accent, we will be presenting the international accent, as an alternative teaching proposal.
Lastly, we will also consider the mediatic accent ( from television), based on
current Parisian, and two types of accents which are very dierent from these, in
so far as they are structurally very dierent: southern pronunciation, represented
by Marseilles, and Canadian, from Qubec.
Vowels
4.1.1.1. French vowels, even those represented in spelling by diphthongs, are
phonically, all short (or long, in certain contexts) monophthongs. Right from the
start one should avoid, the mistake many make (even authors of grammar books),
of confusing the written word with the phonic structure of a language: two very
dierent aspects. Contrary to opinion, which is as widespread as it is wrong,
sounds are the real essence of a language, not the trivial graphic signs used to x it
in writing.
Before looking at their qualities, we will immediately demonstrate the mechanisms for their duration: in an intoneme, vowels followed by nal /v, z, Z K, vK/
are lengthened, as are nasalized V followed by at least one phonic C ( pronounced, not only written); lastly, even /, o/ are lengthened when followed by
one or more phonic C. In a preintoneme, in all of these cases, there is a simple half-lengthening. Specic examples will not be given now, but will certainly be dealt
with later: therefore one should analyze each case, in the light of this knowledge.
In traditional pronunciation, and in Parisian (and mediatic) /A/ also comes
under this category, but with many exceptions and uctuations, as well as analogic
forms, which makes it impossible to try to establish complete and reliable lists;
however, dictionaries still show (obviously without agreeing) words with /A/, as
does Fouch (1959); whereas, Lerond (1980) adds them but with the label vielli
Paris.
If it is absurd to want to nd such words for traditional neutral pronunciation,
which is decidedly out-dated today, it could be interesting to do so for Parisian/mediatic (with some dierences for the suburbs). In the professionally used mediatic accent, there may rightly be the tendency to reduce the use of /A/, but, not yet
total substitution.
4.1.1.2. 4.1 shows the French vowel articulations, which have to be compared
to those of the other languages to show the similarities and, above all, the dierences. In our transcriptions, we use twenty-three elements (plus another less impor-
152
tant one). e front series presents ve vocoids: (i, I, e, , E), for three phonemes:
/i, e, E/.
Although these three, and even (), have the same symbols as in other languages,
they dier slightly: (mi'_i) /mi'di/ midi (e'te) /e'te/ t ('fEt) /'fEt/ fte ('E) /'E/ e
note: (s'm) /sE'mwa/ c'e moi (v'+) /vEK'ty/ vertu (_I'lP) /diK'lo/ dirlo (with
unstressed /E/, and with unstressed /i/ in a checked syllable in /K/). e French /a/
is denitely fronter than in Spanish, Italian, German, : (b':\) /ba'gaZ/ bagage
('m) /'am/ me (traditional ('A:m), mediatic/Parisian (':m)); furthermore: (p'+i:) /paK'tiK/ partir (&p+i') /paKti'K/ partirons (p'l) /paK'la/ par l
e phoneme /A/ of neutral traditional pronunciation was mentioned, which
is no longer up-to-date; for this reason it is not in 4.1 (however, its tongue position is slightly fronter than that of // (), in the vocogram of the international
accent, 4.4.1).
e back (and rounded; actually, back-central as can easily be seen in 4.1)
series is evidently dierent from that in most languages, although the phonemic
symbols are the same ((, , P, , ) for three phonemes /u, o, O/): (t'\:) /tu'ZuK/
toujours (dP'dP) /do'do/ dodo ('km) /'kOm/ comme note: (k'md) /kO'mOd/
commode (\'li) /ZO'li/ joli (k'se) /kuK'sje/ coursier (p'v) /puK'vu/ pour vo
(with unstressed /O/, and with unstressed /u/ in checked syllable in /K/).
4.1. Vowel elements of modern neutral French.
/i/ (i[:], I)
/y/ (y[:], Y)
/e/ (e), // ([:])
// (#)
/E/ (E[:], E[:]/, ), // ([:], 3)
// ([:], [:]/, #), /^/ (^/[:], }/)
/a/ ([:], )
/u/ ([:], )
/o/ (P[:]), // ([:])
/O/ ([:], [:]/, )
// ([:])
Front rounded vowels

4.1.1.3. e front rounded series (actually, front-central, 4.1) (y, Y, , #,
), for three phonemes /y, , /, besides schwa or, rather, unstable e (badly represented as //, and wrongly dened as mute e), which is realized as (#), like (an
unstressed) // and which we represent with // (#): (fy'+y:) /fy'tyK/ futur (')
/'K/ hSrSx (':) /'K/ hSre ('sl) /'sl/ sSl note: (#'te) /*K'te/ hSrt (s#l'm) /sl'm/ sSlement (&pYpy') /pyKpy'K/ purpurin (sY'ps) /syK'plas/ sur
place (l#'pi) /l'pKi/ le prix (m#'s) /m'sj/ monsiSr
Indeed, as far as the unstable e is concerned, one could adopt dierent
phonemic strategies and transcriptions. For example, instead of the still prevalent
phonemic //, one could simply resort to //, however, in theory, the possibility
of distinguishing between djSner (&de\#'ne) and djeter (&de\#'te, d/'te), or jSdi
(\'_i) and je d (\#'_i, \'_i), would nearly be lost, as one would have /deZ'ne/,
and /deZ'te/ or /deZ'te/ for djeter but /Z'di/, and /Z'di/ or /Z'di/ for je d.
4. french
153
e real dierence lies in the fact that normally, forms with unstable e are expressed with the zero phone and possible voicing assimilations, as seen in the second transcriptions. e solution of the zero phone might even leave natives in
doubt (for rarer words), who are frequently undecided, about the proper spelling
too, between /, / e S even if // S cannot be dropped and should not give
rise to confusion. For the moment, having avoided the unrealistic solutions,
//, and /, +/ too, which are less improbable diaphonemically, to play safe we will
adopt // (#), before transforming it denitively into //, with the possibility of
indicating its presence or absence with /[]/ or //, in opposition to an unstressed stable e, represented by //.
en // is stressed, as in d-le le e, it actually becomes the phoneme // (reinforcing further what has just been said): (_i'l) /di'l/, (l#') /l'/ (it is the writing that is capricious, not the phonic structure). In stressed checked syllables in
/K/, the articulation of /E, , O/ can be more or less back and lowered (((E/, /, /)),
but they are transcribed in the same way): ('mE:) /'mEK/ mre ('s:) /'sK/ sur
('f:) /'fOK/ fort
4.1.1.4. Modern neutral pronunciation keeps the opposition /'e/ >- -e -er -ez
V /'E/ >-s -et -Pe -a -P(en)t, which, however, is rather tiring, structurally, given that the others are neutralized: actually, there is no (longer) dierence between
p and pot they are both ('pP) /'po/, nor between pS and pSx\ ('p) /'p/, and
now, not even between bat and b\ ('b) /'ba/. We have, for example, ('ke) /'kle/
cl V ('kE) /'klE/ clPe ('pe) /'pKe/ pr V ('pE) /'pKE/ prs ('fe) /'fe/ fe V ('fE) /'fE/
fPt (v'le) /va'le/ valle V (v'lE) /va'lE/ vallet (p'le) /paK'le/ parler V (p'lE)
/paK'lE/ parla (s've) /sa've/ savez V (s'vE) /sa'vE/ savPt
It is prevalently through school and writing that such oppositions persist, as it
is above all useful to distinguish certain verb forms: (&\#p'le, /p-) /ZpaK'le, Sp-/
je parlP (past: I spoke) V (&\#p'lE, /p-) /ZpaK'lE, Sp-/ je parla (imperfect: I
spoke) and (\#&pl#'e, /&p-) /ZpaKl'Ke, Sp-/ je parlerP (future: I will speak)
V (\#&pl#'E, /&p-) /ZpaKl'KE, Sp-/ je parlera (conditional: I would speak).
As far as -ai is concerned, we have: ('\e) /'Ze/ j'P (but ('\E) /'ZE/ j'Pe] ('e, 'E) /'ge,
'gE/ gP ('e, 'E) /'ke, 'kE/ quP ('bE -e) /'bE -e/ bP ('mE -e) /'mE -e/ mP ('vE -e)
/'vKE -e/ vrP furthermore, we nd ('mE) /'mE/ ma (with (m, me) /mE, me/ in a
preintoneme), as also for ('E) /'E/ tu [il e] ('fE) /'fE/ je fa [il fPt] ('sE) /'sE/ je/tu
sa [il sPt] ('vE) /'vE/ je va with (, e f, fe s, se v, ve) /-E, -e/ in a preintoneme: (&il'l, -e-) /ilE'la, -e/ il e l (m's2 'l:2) {(me-)} /mE'sa, a'lOK,/ {/me-/}
ma a, alors!
Nasalized vowels
4.1.1.5. Obviously, the four nasal vowels remain (which are more precisely dened as nasalized] with their six taxophones, (', 3) //, ('^, }) /^/, () //, ()
//: (&3s't) /sEK't/ incertPn (}'b^) /^'bK^/ un brun (p'd) /p'd/ pendant (m'n) /m'n/ mon nom in stressed and checked syllables, we have:
154
('p:dX) /'pdK/ pRndre ('^:ba) /'^bl/ humble ('l:p) /'lp/ lampe ('n:bX) /'nbK/
nombre A funny example: (}'b 'v 'bl) /^'b 'v 'bl/ un bon vin blanc ese
phonemes can even be followed by N\ ('~i) /'ni/ ennui (n'nEtX) /n'nEtK/
non-tre (m'ne) /m'ne/ emmener.
It must be pointed out that, in current dictionaries and manuals, only four symbols are used (the same, for both phonemic or phonetic use): /, ^, , /; of
these, for modern neutral pronunciation, only /^/ can be used; the others reect
a pronunciation which is over a century old (exactly when the International Phonetic Association was founded), and lasted until the 1950's, as the neutral pronunciation.
As the neutral pronunciation is dierent today, it can still be found in various
regional pronunciations, even if, as we will see, /, ^, , / can be useful as
representatives of an international pronunciation, which is less tied to Paris and
to the neutral pronunciation of Parisian origin ( 4.2). Certain texts (and some
dictionaries), for the rst three, even use /e, J, /. In the transcriptions in the
French Dictionary (by R. Boch: Zanichelli, 1995), the present writer put /, ^,
, /, as here. In books which only use one kind of transcription (often a hybrid
of phonetic and phonemic), it is probably more suitable to give /, ^, , /.
Modern neutral pronunciation rmly keeps these four nasalized vowel phonemes, as do most regional pronunciations (even if with dierent, more traditional
timbres). Instead, for Paris (for the center and the west) /^/ merges with //, so,
there, /'bK/ corresponds to brin and even to brun (neutral ('b, 'b^), respectively). Consequently, even the mediatic accent loses a phoneme, merging both these
forms and other similar ones. is will all be dealt with later ( 4.4.2.1-2
4.5.2.3).
Other peculiarities of French vowels
4.1.2.1. As far as unstressed V are concerned, it would be best to give some indications to follow, in order to obtain the most tting results. Transcriptions in dictionaries and in manuals do not actually tally completely, as there are various factors to be considered. In order to simplify the description, let us say that independently of writing and dictionary transcriptions there are the following vowel adjustments: for /E/ ( the archiphoneme of /e, E/), we have (e) /e/, in an unchecked pretonic syllable, if it is followed by a close(r) V ( /i, y, u e, , o /}:
(e'te) /e'te/ t (me'z) /me'z/ maon (le'z,) /le'zj/ l ySx (&epe'te) /Kepe'te/ rpter
On the other hand, we have () /E/, in an unchecked pretonic syllable, when it
is followed by an open(er) V ( /E, , O a , ^, /) and in checked syllables (independently of the context) too: ('tE) /E'tE/ ta ('t) /E't/ tant (l'zm) /lE'zOm/ l homm (&ep'tE) /KepE'tE/ rpta and (&evn'm) /evEn'm/ vnement
(mt's) /mEt's/ mdecin (p'te) /pEl'te/ pelleter (p'_y) /pEK'dy/ perdu (d'mi)
/Ed'mi/ et demi (t'p:) /tEK'pOK/ t reports (s's) /sE's/ s leons (&lsp'tka)
/lEspEk'takl/ l spectacl (dp'n) /dEp'n/ d pne For (initial) ex- esC- there is
4. french
155
a strong tendency to have /e/: ('z:pa, e-) /Eg'zpl, e-/ exemple (&s',e, &e-) /Eska'lje, e-/ caliers
e examples of checked syllables, (intentionally) show, in these two cases, that
the timbre of the V that follows is of no importance (as it is, here, quite the opposite). One must always recall that spelling (which is not in any way a true transcription of phonic structure) can play dirty tricks For -err- we have /EK/, but we can
also nd the adjustment: (t'iba, te-) /tE'Kibl, te-/ terrible (s'e, se-) /sE'Ke, se-/ serrer but (&p'E) /pEKO'kE/ perroquet furthermore, (&Pp#&m, 't:\) /opK'mjE
KE'taZ/ au premier tage (&}l'\ '~i) /^lE'ZE K'ni/ un lger ennui (despite (p#'m,e, le'\e) /pK'mje, le'Ze/ premier lger] ere are possible uctuations for -er +
V- and for the spelling (towards /e/): (&}le'\e '~i, &Pp#&m,e e't:\) and for P
Z (towards /E/): (pe'zi:, p-) /ple'ziK/ plair
4.1.2.2. On the other hand, the isolated forms often inuence the contextualized ones, even in checked syllables, as in (&\evizi'tl n'vi:, -tel) /Zevizi'tEl na'viK,
-tel/ j'P vit le navire also for -ez: (&vl'v k'ny) /vula'vEK kO'ny, -eK/ vo l'avez
reconnu, and even for the proclitic -es monosyllables: (t'p:, te-) /tEK'pOK, teK-/
t reports (s's, se-) /sEl's, sel-/ s leons (&lsp'tka, les-) /lEspEk'takl, les-/ l
spectacl (dp'n, dep-) /dEp'n, dep-/ d pne also (d'mi, ed-) /Ed'mi, ed-/ et demi
Lastly, it is evident that the eects of vowel adjustment, in unchecked syllables,
can go back further than the pretonic syllable, as long as syllables with dierent V
timbres do not intervene: (&epe'te) /Kepe'te/ rpt (&p't) /KEpE'ta/ rpta (&bee',e) /bege'je/ bgZer (&b'm) /bEgE'm/ bgPement
For // too, (not to be confused with (`), zero), in unchecked syllables (phonically of course, and always independently of transcriptions easily found), we have
() // + a close(er) V and (#) // + an open(er) V\ (p#'v:) /pl'vwaK/ plSvoir (p'e) /pl'Ke/ plSrer; whereas for /O/ in unchecked syllables, we generally have () /O/, unless it is immediately followed by /z/, or by a syllable with /o/,
or it does not derive from /o/ and in -otion: (&mn'tn) /mOnO'tOn/ monotone (&sl'pEt) /salO'pEt/ salopette (b'se) /bO'se/ bosser but (\P'zEf) /Zo'zEf/ Jos#e (bP'bP)
/bo'bo/ bobo (P'ze) /Ko'ze/ rose (&gPs'te) /gKosjEK'te/ grossiret (&emP's) /emo'sj/ motion.
Even for the (unstressed) V spelt , au, the more frequent pronunciation is /O/:
(&pi'tl) /Opi'tal/ hpital (':) /O'KOK/ aurore (m'is) /mO'Kis/ Maurice ('v:) /OK'vwaK/ au revoir In aujourd'hui the preposition combined with the
denite article [au +le] holds; instead, it is the second syllable that gives more,
due to /uK/, as well: (&P\'_i, -\-) /oZuK'di, -OK-/; but we can hear (&\-, &\-) /OZOK-, uZuK-/ too.
For bcoup [b + coup] obviously, we have (bP'k) /bo'ku/, but the tendency to have /O/, is so strong that, as it would be quite absurd to have */bO'ku/, we
very often end up with having, (b'k) /bu'ku/. Even surtout and au fur et
mesure present us with the frequent colloquial pronunciation (s'tU, P&f#m'zy:), for (sY'tU, P&fyem'zy:) /syK'tu, ofyKeam[]'zyK/.
Likewise, for /, /, the base form is quite important: (d'z,Em) /d'zjEm/ dS-
156
xime (b#'e) /b'Ke/ bSrrer In both cases, in checked syllables, we nd (#) //,
() /O/: (s#l'm) /sl'm/ sSlement (ps'te) /pOs'te/ poer
Above, we have already dealt with /^, / and /iK, yK, uK/ ( 4.1.1: see given examples). In all other cases, in modern neutral pronunciation, in unstressed
syllables, we have, (i, y, , ) /i, y, u a , /.
4.1.2.3. Another interesting peculiarity of modern neutral French pronunciation is the tendency to devoice /i, y, u/, in certain contexts; this tendency is even
more consistent for the C, as will presently be shown.
erefore, between voiceless C, or between them and a subsequent pause, we
quite frequently nd (i, , U): (&pfi'te) /pKOfi'te/ proter (p&+iy',e) /paKtiky'lje/
particuliers (m'si) /mEK'si/ merci (t'pi) /t'pi/ tant p (p't#) /py'twa/ puto
(&kUs'+i) /akus'tik/ acouique (p'tU) /paK'tu/ partout as well as the colorful ('i,
', ') /'wi/ (/'wih/) oui! ('2 '\Em 5b,1 1 f's#; %z'_2) /'wi, 'ZEm 'bj fK'swa zaK'di,/ oui, j'Pme bien Franoe Hardy
Before a pause, devoicing can occur even after a voiced C, but only partially, (,
, ): (m'_) /maK'di/ mardi (&t'_) /t'dy/ entendu (d#'b) /d'bu/ debout
Even between a voiceless and a voiced C, /i, y, u/ frequently become devoiced: (&l#p'+ d[#]) /lpaK'ti d[]/ le Parti de (&+vi'te) /aktivi'te/ activit (&kf'z,)
/kfy'zj/ confion (&dek'p:\) /deku'paZ/ dcoupage Albeit more rarely, even between a voiceless C and a pause, /e, , o/ can become devoiced: (&k'tE) /Kak'te/
racont (le'd) /le'd/ l dSx (p't) /pal'to/ paletot
Consonants
n
(n)
m
F pb
t d (T D)
f v
_
s z
(l)
l
(n)
(~)
N
(+ _) ( )
(,)
uvular
velar
provelar rounded
postpalatal rounded
palatal
prepalatal
alveo-velar
alveolar
dental
labiodental
bilabial
4.2. Table of French consonants.
postalveo-prevelar
protuded
4.2.0. e table of 4.2 shows the French consonant articulations, which are
necessary for the right pronunciation of that language.
1.9-15, instead, give the orograms, grouped for manners of articulation, of all
the contoids given in the chapters of this volume (even as secondary, occasional,
or regional variants) for the 12 languages treated. is exposition makes the necessary comparisons between dierent languages more immediate.
{}
kg
(X )
/ \
() /j/
()
()
/K/
4. french
157
Nasals
4.2.1. In French there are three traditional phonemes: /m, n, N/. It is worth
keeping the last phoneme, even though it has lost most of its phonemicity, merging with /nj/ (as did /L/, which does not exist in French anymore, having merged
with /j/): (m'm) /ma'm/ maman (n'n) /na'n/ nanan (p'~,e) /pa'nje/ panier ('NP, '~,P) /a'No/ agn (&sN'm -~,#-) /sEN'm/ ensRgnement (m'tN
-~,) /m'taN/ montagne e transformation into /nj/ is more frequent (and for
many, by now, quite normal) in front of a V
If /N/ is losing ground, there is a xenophoneme (for English -ing] which is taking root (and, in theory possibly, given its limited use the balance of the phonological system could be restored, with the substitution between these two N] however, at the moment, the pronunciation of the English -ing, as in camping, varies
substantially; the most recommendable pronunciations are velar, (-i) (in an English-like way), or prevelar, (-i) (by partial assimilation): (k'pi, -); or even palatal, (-iN) (above all for the elderly); we also nd (-ig, -i, -iN) (a more autochtonous version).
For (tautosyllabic) /nj, n/, we have (~,, ~): (p'~,e) /pa'nje/ panier (y'~,l) /y'njOl/ une yole ('~i) /'ni/ ennui. (Popularly, we nd that /nj/ and /N/ tend to
merge into (N): (m'~,E:) /ma'njEK/ manire becomes (m'NE:, m-), but should
not be copied.)
Normally, /n/ does not assimilate a heterosyllabic C following it (as happens, instead, in most other languages), so we have (including ((yn$-)), nearly (&yn#-), with
quite an evident break): (yn'pm) /yn'pOm/ une pomme (yn'b:) /yn'bk/ une banque (yn'mE:) /yn'mEK/ une mre (&ynp'sn p's:t) /ynpEK'sOn pa'sjt/ une personne patiente (yn'fi,) /yn'fij/ une lle (&ynv'li:z) /ynva'liz/ une vale (yn'/E:z) /yn'SEz/ une >ae (yn'No:l) /yn'Nol/ une gnle (mn') /man'k/ mannequin (&yn'Es) /ynka'KEs/ une carse (yn'gt) /yn'gut/ une goutte and (&yn+y'lip) /ynty'lip/ une tulipe (yn&_,g'nl) /yndjagO'nal/ une diagonale (&yne'p:s) /ynKe'ps/ une rponse.
Only in (what is often considered to be) non-neutral pronunciation, or nearly,
is it possible to have coarticulations for /n/ before a dorsal C: (~) (with raised tip
of the tongue) + (N, , ) and (n) + (k, g , ): (y~'No:l, m8', &y8'Es, yn'gt,
&yne'p:s).
Near a voiceless C, as already seen in some examples, the N undergo devoicing,
and they reach total voicelessness before a pause: (}p'n) /^p'n/ un pnS ('pis))
/'pKism/ prme
Stops
4.2.2. ere are three (diphonic) pairs of phonemes: /p, b t, d k, g/, with important pairs of taxophones: prepalatal, /t, d/ (+, _), before /i, y j, / (and, less systematically, before /e, / too); a further (less important) one, is alveolar, for /t, d/
(T, D), before /S, Z/ (whereas, before /s, z/ they remain dental); and one palatal, or
158
rather, postpalatal, for /k, g/ (, ) (more appropriate symbols are ((, 8)), but it is
not absolutely necessary to use them), before front V (including /a, /), before /j,
/ and even at the end of syllables or rhythm groups, before a pause. In the other
cases, the velar articulation, (k, g) can also be prevelar, ((, )), but, again, it is
not necessary to use special symbols.
Let us take a look at some examples: (p'pe) /pu'pe/ poupe (be'be) /be'be/ bb
('tt) /'tut/ toute (d'dn) /du'dun/ doudoune (+y'_i) /ty'di/ tu d ('+) /'tj/ tiens
(k'_i:) /k'diK/ conduire (e'te e'+e) /e'te/ t ('d '_) /'d/ dSx (T'/U)
/kaut'Su/ caout>ouc (&D\'+if) /adZEk'tif/ adjectif (&s3m'its) /smO'Kits/ SPnt-Moritz (pid'z) /pid'za/ pizza (pi'ni) /pik'nik/ pique-nique ('k) /'kOk/ coq (') /ga'ga/ gaga ('l:) /'lg/ langue ('gP) /'gKo/ gros ('gi) /'gKi/ gr ('k) /'klak/
clac! Before tautosyllabic C (k, g) remain unchanged ({pro}velar), as can be seen
in some of these examples.
We nd complete voicing assimilation to the second element (if diphonic), in
cases such as: (&n'dt) /anEk'dOt/ anecdote (&T/P'se) /KEdSo'se/ rez-de->ase
(mt's) /mEd's/ mdecin (&ps'ly) /absO'ly/ absolu (&sd'dm) /sEt'dam/ cette dame ('kb d#/'pN) /'kup dS'paN/ coupe de >ampagne (/g'\:) /Sak'ZuK/ >aque jour (&v'v) /avEk'vu/ avec vo.
In these cases, a slowing down of the rate of speech can lead to only partial assimilation: (&n'dt, &/P'se, m's, &s'ly, &s'dm, 'k d#/'pN, &v'v).
is is the reason why, in phonemic transcriptions, we keep etymologic phonemes, whereas for V we indicate the actual timbres, as we use more precise symbols (despite dictionary even pronunciation-dictionary transcriptions which do
not, however, consider intermediate articulations!).
On the other hand, if the second segment is not diphonic (including /j/), the
aforementioned voicing assimilation does not occur: (&v'n) /avEk'vu/ avec no
(&v'1i) /avEk'li/ avec lui
Constrictives
4.2.3.1. ere are three (diphonic) pairs, /f, v s, z S, Z/, as well as two isolated
voiced phonemes, /j, K/ (instead of /J, /, the more legitimate symbols), which
we will see presently. We can observe that, usually, /s, z/ are articulated with the
tip of the tongue raised, therefore, if we wanted to highlight this aspect, for discussion and teaching, we could resort to supplementary symbols, ((s, z)); but, the most
important characteristic concerns /S, Z/, which, are generally, prevelarized postalveolar protruded, (/, \) (with a deeper timbre, caused by the lowering of the back of
the tongue between the two articulatory strictures, postalveolar and prevelar). Examples: ('fE:) /'fEK/ fPre ('vif) /'vif/ vif ('s) /'s/ cent ('v:z) /'vaz/ ve ('/) /'Sa/
>at ('p:\) /'paZ/ page
Voicing assimilation (to the second element) also concerns diphonic pairs of
constrictives: (nv'z) /nuf'z/ no faons ('P;s pfy'me) /'Koz paKfy'me/ rose
parfume ('v\ 'v,E,) /'vaS 'vjEj/ va>e viRlle (/'sE) /Z'sE/ je sa but we have (/f) /Sv/:
(/'fl) /S'val/ >eval (/'fe) /aS've/ a>ev Slowing down, we can even have (n'z,
4. french
159
'P; pfy'me, 'v? 'v,E,, ?'sE /'l, /'e). On the other hand, speaking quickly, we
can nd cases such as: (/'p) /Z{n}sE'pa/ je (ne) sa p.
About /j/, it must be said immediately that, more than a real constrictive, it is
a semi-constrictive, (,), indeed, it is placed half-way between the approximant,
(j), and the truly (voiced) constrictive, (J) (denitely rarer in world languages):
(',E:, i',E:) /'jEK, i'[j]EK/ hier (',e) /ka'je/ cahier ('pe) /'pje/ pied (,P',P) /jo'jo/ yo-yo (&t#',e) /at'lje/ atelier (f'mi,) /fa'mij/ famille (s'lE,) /sO'lEj/ solRl (f#'t)
/fj't/ fSilleton Before a pause, we often have (): (f'mi, s'lE). French phonotactics, contrary to that of other languages, also has /Sj, Zj/: ('/) >ien (e'\,) rgion
Sequences such as //0Ki', 0li'// are given with (i',), and therefore even the
most appropriate and most modern phonemic transcription gives /i'j/: (pi',e)
prier (pi',:) priSr (pi',) plia (&sbli',e) sablier Hence, (bi',) stands for
both brillant and Briand but we could possibly have (bi') /bKi'/ for Briand
in controlled pronunciation. Furthermore, we have: (pe'i, pe',i) /pe'[j]i/ pZs (&be'i,
-e',i) /abe'[j]i/ abbZe
4.2.3.2. As far as /K/ is concerned, neutral pronunciation has two taxophones
(with devoicing, and other possibilities, which we will point out): the voiced uvular constrictive (), before a stressed V after a (tauto- or hetero-syllabic) consonant
and after a pause; and the (voiced) uvular approximant, (), before an unstressed
V before a (heterosyllabic) consonant and before a pause.
Some examples of the constrictive: ('_,P) /Ka'djo/ radio ('y) /'Ky/ rue (p'i)
/pa'Ki/ Par ('tE) /'tKE/ trs (&pevi'z,) /pKevi'zj/ prvions (&pli'n:\) /pElKi'naZ/ plerinage ('tX) /'katK/ quatre ('sfX) /'sufK/ soufre (#'p:dX) /K'pKdK/
reprendre ese examples demonstrate typical devoicing and complete voicelessness, too, in (X), between (even voiced) C and a pause. Slowing down, or enunciating, we can even have (): ('li:vX, -v) /'livK/ livre Some examples of the approximant: (&i've) /aKi've/ arriver (&l'_,P) /laKa'djo/ la radio (p'+i:) /paK'tiK/ partir ('v:) /OK'vwaK/ au revoir
It must be immediately added that a frequent variant of () is the voiced uvular
trill, (K) (and this may explain why we use /K/, which generally indicates the uvular place of articulation, to help to avoid foreign pronunciation): ('Ky, p'Ki, 'tE,
&pevi'z,, &plKi'n:\, 't', K#'p:d', 'li:v', -K); this is frequent after tautosyllabic C, above all /p, t, k/, ('k:) /'kKwaK/ croire whereas after /b, d, g/ we also
have the (voiced) uvular tap: ('b^ 'br- 'bK-) /'bK^/ brun (d'pP dr- dK-) /dKa'po/ drap ('g 'gr- 'gK-) /'gK/ grand For emphasis (, K) can substitute ()
too: ('fE: - -K) /'fEK/ fPre Sometimes we can have the uvular constrictive trill,
or tap, (, ) (and the voiceless ()), especially after /p, t, k/: ('g, 'g, 'tE, 't).
On the other hand, we can also nd a voiced velar semi-constrictive (with a voiceless velar constrictive, in a devoicing context): ('gy, 't@E, 'tx).
160
Approximants
4.2.4. Apart from the taxophone () of /K/ (just treated with the constrictives),
we have two central approximant phonemes, () // (postpalatal rounded) and
() /w/ (provelar rounded, for which the symbol (w) of velar rounded phone
could quite easily be used, as in other kinds of pronunciation, treated at the end
of this chapter; but the opportunity to demonstrate not negligible gradation
would be lost): ('si) /'si/ su ('~i) /'ni/ nuit ('i) /'li/ lui ('li) /'lwi/ Lou
('m) /'mwa/ mo ('p#) /'pwa/ po ('s#:) /'swaK/ soir ese examples also
show assimilation of voicing and place of articulation, as well as the important fact
that // and /w/ are two dierent phonemes, dierent from /j/ too. Regarding /lw/,
we can have by assimilation, (]), with /l/ realized as semi-velar. e /0K, 0l/ + /y,
u/ sequences remain the same, with no insertion of approximants (contrary to
what happens for //'i, i'//, which normally become /'ji, i'j/, 4.2.3): (&psty'e)
obruer (gly') gluant (k'e) cloue (t') trouant
Laterals
4.2.5. ere is only one lateral phoneme (today, 4.2.1), (l) /l/, which assimilates for voicing (and, before /j, /, for place of articulation): ('lyn) /'lyn/ lune
('p#l) /'pwal/ poil ('le) /a'le/ aller ('bl) /'bl/ blS ('ke) /'kle/ clef ('fy) /'fly/ ux
('p) /al'p/ alpin ('z:pa) /Eg'zpl/ exemple (':ka) /'kl/ oncle (':ga) /'gl/ ongle ('sfa) /'sufl/ soue Slowing down, or enunciating, we can also have (l): (':ga
-l), between a voiced C and a pause. Sometimes, we can hear something in between, with (): ('k, ':g). In /lj, l/ the articulation is prepalatal: (s',e) /su'lje/
soulier (',) /'lj/ liS ('i) /'li/ lui. Often, in non-neutral pronunciation, /lj/ and
/j/ tend to merge into /j/: (mi', mi',) million (mi', mi',) miliS therefore (s',e) soulier and (&fyzi',e) filier can correspond to (s',e) souiller (&fyzi',e) filler
Structures
4.3.0. e greatest problems for segments are caused by the unstressed vowel
phoneme graphically expressed by e (apart from some exceptional cases, as monsiSr faons); whereas, with regard to phone groups or connected speech the
phenomenon of liaon is typical ( 4.3.3.1-3).
e (unstable) // phoneme
4.3.1.1. In neutral modern pronunciation, // is realized as an unstressed //
( 4.1). ere are various terms to indicate it, some less suitable than others, like:
schwa, e caduc, e muet. Its use and distribution constitutes one of the main characteristics of the French phonological system, even if its phonemic status could be
4. french
161
debatable. More often than not, it seems that // (#) is introduced, in pronunciation, to avoid long sequences of C which turn out to be dicult to pronounce.
erefore from a phonemic transcription such as //msj, dd, at'lje//, , we could
actually have: (m#'s, d#'d, &t#',e) monsiSr dedans atelier as isolated forms;
in connected speech, the most common forms are, for instance: (}'s) /^m'sj/
un monsiSr (ld'd) /lad'd/ l dedans whereas atelier remains the same.
We could, on the other hand, start from a full or isolated form, that retains
all these unstable e (which is the most common in the traditional reading of verse,
and is even stronger in the way of speaking in the south of France, the Midi] dropping all the possible //, without complicating the pronunciation with dicult
or impossible groups: //bOn'tKi// bonnetrie //ZtlK'di// je te le red for the
normal (bn'ti, &/t#l#'_i &\#t#'_i). Generally, in phonemic transcriptions in
dictionaries, the // that do not drop in isolated forms are retained, omitting the
others; hence, we can now see when, even those which are normally indicated, can
drop.
4.3.1.2. In practice, the dropping of one or more // can occur if the consonant groups, which come in contact, can occur within the word, /lst, ksj, kskl,
kspK, ksplw, Kkw, Ks, KstK, Kmn/, , as in: (ss'tis) /sOls'tis/ solice (&z's)
/Egzak'sj/ exaction (&ksk'me) /Ekskla'me/ exclamer (&kspi'me) /EkspKi'me/ exprimer (ks'p) /Eks'plwa/ exploit (p'k#) /puK'kwa/ pourquoi (p'sit)
/puK'sit/ poursuite (&syps't) /sypEKs'tKa/ superrat (&ipmne'zi) /ipEKmne'zi/
hypermnsie.
It is generally possible to increase the number of C in contact, if when dropping
//, constrictives, approximants, laterals and nasals (but also stops) are added
before or after: (is'sE;t 's &is#'s-) qui se sert de a? (\#n&l#_i'p, \&n#l#-) je ne le
red p (&ynp+it'fi,) une petite lle (&+yns'p) tu ne ser p ([&i],&bPkd'm:d)
il y a bcoup de monde (&np'lE #t's) on ne parlPt que de a (&s#&\#s'vE,
z&\#-) e-ce que je le sava? (\m'fi/) je m'en >e!
Furthermore: (/&k'b,) je cro bien (s&np'sy:) ce n'e p sr (s&i[]t#'fP,
sit'fP) ce qu'il te faut (s&t3'l) ce trPn l (/&sik'td l'v:) je su content de la
voir [de l'avoir] (/t#ld'm:d, /&t#ld#'m-) je te le demande (&s\#n&t#ldv'p, -d#v-,
s/&t#l) c'e que je ne te le deva p (lve'v, &l#-) levez-vo! (m&nem'l, m#&ne-)
menez-moi l! (\'_i kem'+iv dl'fn sp'b, -+if) je d que l motifs de l'enfant
ne sont p bons (&sYl#'b, syl-) sur le banc (si&\#nt#l_i'p, &si/t#l-) si je ne te le d
p (il&m#ldmt'p, in&m-, &inm#l&d#-) il ne me le demande p.
thin a word, we can nd: (bn'ti) bonneterie (s'l) cserole (m'ne)
amener ('pe) appeler (\'e) jouerP (pn'm) plRnement but: (&gl#'tE:) Angleterre (&mk#'_i) mercredi (&pl#'m) parlement (&ft#'m) fortement (&p#'m) amplement For parle-m'en, we often have (pl'm), which we can also
hear for parlement in fast or slow speech.
4.3.1.3. It is important to bear in mind the dierence that exists in French, for
the /0/ + /m, n, K, l/ + /j/ sequences, that would be too heavy, hence, a stable //
is inserted: (&sm#',e) sommelier (&ns#'m,) no semions (&st#'~,e) centenier
162
(&vt#'~,e) vo teniez (&ns#',) no serions (&v/t#',e) vo >anteriez (&i/#',) Ri>eliS (&nzp#',) no appelions but (s'pe) cse-pieds (st'pEs) cette
pice (bn'+e) bonnetier (/'+e) >arretier (p'+e) pelletier e same happens
for /, w/: (/e&s#i'si) >ez celui-ci (}&bd#'l) un bout de loi even if it is not unusual to hear: (&/esi'si) >ez celui-ci (&}b'dl), (&l#\'dl) le jS de l'oie and similar phrases.
e // does not generally drop in the rst syllable of names: ('li #'nP) /'lwi
K'no/ Lou Renault (&#'n:) /aK'naK/ Renard not even for de\ (&d#l'i:v)
/dla'Kiv/ De la Rive (m#&sd#'gP:l) /msjd'gol/ M. De Gaulle but if it is possible to simplify, it is done, on the radio and television too, even if it is a little stigmatized, as in: de De Gaulle which is nearly always (d#d'gP:l). Surnames usually
resist better than rst names, indeed, Renaud and Den, in appropriate contexts
are often: ('nP, d'ni).
However, we regularly nd: (}'n:) /^K'naK/ un renard (sn'l2 &v,d'gP:l)
{(s#n-)} /sn'la, vjd'gol/ ce nom-l vient de "Gaule& (\&nepd'gP:l) /Znepad'gol/
je n'P p de gaul (in a real or gurative sense), (P&bdl'i:v) /obOKdla'Kiv/ au
bord de la rive but: (i/'lE) /KiS'lE/ Ri>elet (obviously in the case of (&i/#',) /KiS'lj/ Ri>eliS, it is dierent). e same happens before /*/ ( disjunctive initial V usually represented by the so called aspirate h and by the names of numbers): (l#e'P) /le'Ko/ le hro (l#':z) /l'z/ le onze
Taxophonics
4.3.2.1. In a sentence, the way in which // behaves, retained or inserted or
dropped, makes its position within a word peculiar. Doubtlessly, in these cases,
pronunciation is inuenced by spelling, as well, with its internal e; whereas word-ending e, which are not usually pronounced in isolated forms, can make one
(think one should) prefer to drop it even in syntagms, compounds and common
phrases.
In certain words, // does not drop, even if the result of the drop would give a
rather simple consonant group: (&np#'z) no pons but (nv'z) no faons
(&l#'El) la querelle but (&}p't) un peloton (&def#'mEl) d femell but (lf'nEtX)
l fentr Furthermore, because of the disjunctive h, (&ynsy'+y:) une sculpture
but (&yn#':t) une honte (&stsp'd:) cette splendSr but (&st#'_,Es) cette hardise (we can have (&st-'_,Es), too).
ere are also cases such as (d'bEs 'i:z) d bell cer (&d#n#v'ni; #'s#:)
de ne revenir que le soir on the other hand, spelling without -e makes the pronunciation of the following seem superior: ('p d'p:s, 'p) Parc d Princ ('
d#ti',:f, ') arc de triom#e (&z'bl) ours blanc ('fil pl'nE) lm polona
(+i'/D '\P:n) T-shirt jaune instead of the more natural ('p# d'p:s, '# &d#ti',:f, 's# 'bl, 'film# &pl'nE, +i'/T# '\P:n), which are thus often considered
less good, because of the spelling, while, above all ('s# 'bl), for rhythmic reasons too, is more than legitimate; ('#t ti',:f) (with -C Ce (0#0)) is at times stigmatized as uneducated. Normally, the following cases can be seen in slow, careful
4. french
163
pronunciation; otherwise, in fast pronunciation, dropping prevails: ('pst s't:t,

-t# -) /'pOst KEs'tt/ poe reante ('p i'+im, -p# m-) /'alp maKi'tim/ Alp Maritim ('/l d#'gP:l, -l# d#-) /'SaKl d'gol/ arl de Gaulle
Let us further consider examples (due to rhythmic reasons) such as: (&pttP,
-t#m-) /pOKt[]m'to/ porte-mant (&ptke',, -t#k-) /pOKt[]kKe'j/ porte-crZon with /[]/, but (&pt#'pym) /pOKt'plym/ porte-plume and, hence, even
(&t#'d:) /kaKt'dOK/ carte d'or and certainly not *(kaR'dO:r) (also) in the Italian
advertisements for Carte d'Or (which in French, would be car d'or (&'d:));
however, (&d'd:) is possible, too.
e cases are increasing where, before a pause, one pronounces a non-etymologic //, which is not present in spelling, above all after voiced C, in particular the
sonants: (P'tEl, -l#, -) /o'tEl, O-/ htel (s'vi:, -#) /sEK'viK/ servir (b'\:, -#)
/b'ZuK/ bonjour!
4.3.2.2. en, in connected speech, a word ends in /0K, 0l/ and is then followed by another which begins with /0/, in slow and wary pronunciation, // is
inserted; but normally, even /K, l/ are dropped: ('t 'fm 't#) /'katK 'fam/ quatre femm (&ynPt'f# y&nPt#-) /ynotK'fwa/ une autre fo (&mddP'tEl 'mEt# d-)
/mEtKdo'tEl/ matre d'htel (l#'pP;v b'nm, 'p- l#'pP;v#) /l'povK bO'nOm/ le
pauvre bonhomme (i's;p #'n &ilm#'s;bl#) /ilm'sbl k'n/ il me semble que
non (&3p'sib d#'fE: -ibl#) /pO'sibl dl'fEK/ impossible de le fPre ('pb _it'li
l#'pp#) /l'ppl dita'li/ le pSple d'Italie
In cases such as /vOtKp'n/ votre pnS apart from a slow and wary (&vt#p'n), we also have (vp'n, -t'n, vtp'n), and (&vtp#'n), considered rather uneducated because of the dierence with its spelling. Decidedly uneducated (and
intentionally joky) is (&s#'pE, -e) for (ks'pE) exprs whereas (s'pE, -e) is rather
fast colloquial.
As seen (in the previous section), we have to have /0j/ in a word (// refers
to sonants: /m, n, K, l/), but it is not valid in sentences: (sl'm,) c'e le mien (&i[]fPl'~,e, fP-) il faut le nier (&i[l]z'+En ',) ils en tiennent liS (in'vl ',, &iln#-) ils
ne valent rien
However, one can denitely also have: (&d#n#', 'fE:) de ne rien fPre (&\#nd#'m;d# ',) je ne demande rien and even: (&sl#'m,), (&i[]fPl#'~,e, fP-), (&i[l]z'+En#
',), (in'vl# ',, &iln#-). Let us also consider: (&bnpt#'m '/P) bon appartement
>aud (&bn'pt '/P, -t#) Bonaparte man>ot If the context does not create
any ambiguity, we can easily have bon appartement >aud (&bnpt'm '/P).
4.3.2.3. For sequences of monosyllabic words with // [le je me te se ce de
ne] there are often clear, general preferences, but not absolute, such as: (\#n \n#)
je ne (\#m, \m#) je me (\#l, \l#) je le (d#l) de le (d#n) de ne (d#m) de me (d#t)
de te (d#s) de se (d#s, ts#) de ce (#l) que le (#n) que ne (#m) que me (#s) que
se/ce (#t) que te (s#) ce que\ (&d#n#v'ni; #'s#:) de ne revenir que le soir
(&m\#n&s#d'_i:) moi je ne sa que te dire (&\#ns',t s#ty'k) je ne sa rien
de ce truc l
For /Z/ in je me le demande we nd (&\#ml#d'm:d) and (\&m#ld#'m:d) the lat-
164
ter is at times considered less advisable; similarly, for the second forms in: (\l#'v,
\#l'v) je le vSx (\#'+, \m#-) je me tiens (/&tv'_id v#'ni;, -'_i d#v'ni:) je t'ava dit de venir (spelling inuences the choice of preferably retaining word-internal //).
e second forms here are, thus, also uneducated: (/'k #/'k) je cro
(\m'fU &#\-) je m'en fo (l'mE d#l'y, l'mE #dl'y) l mecs de la rue (&dek'vi; l#'m:d, &dek'vi; #l'm:d) dcouvrir le monde ('pd l#me'tP, 'p d[]#l-)
prendre le mtro (i'mEt #'pi, i'mEt []#'pi) y mettre le prix (if it could be confused
with ils mettent le prix (i'mEt #'pi), the /K/ does not drop), (i[]'/E/ s#i[l]'v, i[]'/E/#s i[l]'v) il >er>e ce qu'il vSt actually, the second forms are often more
natural, but alas dierent from the spelling!
In the case of forms such as (&b'm) aboiement (\n't#) je nettoie (i[l]'v)
ils voient (/pe'e) je pZerP [pPerP] (i[]'s#) qu'ils soient (&#+y'E, #'+E, #'tE)
que tu P forms such as (&b,'m, \n't#,, i[l]'v,, /p,'e, i[]'s#,) and (&#+y'E,, #'+E,, #'tE,) are certainly uneducated.
4.3.2.4. In some cases, in French, we have geminated C in derived words:
(&tzm'm) troimement (nt'te) nettet (&ek', &-) clPrera ( (&ek',
&-) clPra] in the future and conditional of courir mourir qurir (and prexed
forms, but not other verbs with -rr-): (/k'E) je courra ( imperfect (/k'E) je
coura and also, (/p'E) je pourra] furthermore, in cases such as: (ld'd) l-dedans ( (l'd) la dent], (+ym'm) tu me mens ( (+y'm) tu mens (&+ynm'p) tu
ne mens p], (&ll'_i) elle l'a dit ( (&l'_i) elle a dit] obviously, even: (&p#+it'tba)
petite table (&p#+it'blP) petit tabl Lastly we have gemination (or lengthening)
to give emphasis: (1sppfE3 3, 1sp:-) c'e parfPt!
Gemination is possible, to maintain distinction, in the case of the imperfect indicative and the present subjunctive, compared to the present indicative: (&nk,',) no croyions (&vk,',e) vo croyiez ( (&nk',) no croyons (&vk',e) vo croyez] to avoid ambiguity in cases such as: (&lssi'i) l'Assyrie (&lsi'i)
la Syrie to insist on a prex (especially negative): (&illi'ziba) illible Gemination
can even be found where it is not needed, for graphic geminates, in bookish words
(but spontaneous and not aected pronunciation carefully avoids such geminations): (vi'l, -l'l) villa (g'mE:, -m'm-) grammPre (&_i's, &d_i-) addition (i'lystX, il'l-) illure it is equally improper, though common, to geminate the pronoun l' (which neutral pronunciation obviously avoids) in cases such as: (&\#lle'vy, &+yll'_i, nl&lv's), for: (\le'vy) je l'P vu (&+yl'_i) tu l' dit (n&lv's)
no l'avons su
4.3.2.5. French presents consonant sequences with dierent places of articulation, which present considerable problems for many foreign people; it is useful to
look at the example we have just seen of (&_i's, &d_i-) addition as well as (&n'dt) anecdote (&+ivi'te) activit (/'k) je cro and many other previous examples and future ones.
e assimilation of voiced stops, between (either oral or nasalized) V and C,
which become N (except in very controlled pronunciation, dominated by spell-
4. french
165
ing) is considerable and typical: (}n&mii'lP) /^dmiki'lo/ un demi kilo (n'm) /ad'm/ demPn (tn'mEm) /tud'mEm/ tout de mme (&t3nm'\e) /tKdm'ZE/ en
trPn de manger (yn'g;n me'z) /yn'gKd me'z/ une grande maon (ln'm)
/ld'm/ lendemPn (d'g;n 'dm) /dE'gKd 'dam/ d grand dam (l'/;m d'mi) /la'Sb[K] da'mi/ la >ambre d'am (k'm,) /k'bj/ combien (∫'p, &it-,
il&n#-) /ilntb'pa/ il ne tombe p (l'l; m'dEn) /lE'lg mO'dEKn/ l langu
modern (yn'l; 'E:) /yn'lg 'gEK/ une longue guerre
For voiceless stops, we have assimilation of places of articulation in the same
way, while we can have voicing, devoicing, or voicelessness, for the type of phonation (again, as well as a slower or more controlled possibility, which corresponds
to the phonemic transcription ): (&}+i'k, &})-) /^pti'ku/ un petit coup (m3n'n,
-n'n, -'n, m3'n) /mt'n/ mPntenant (v3n'd, -n'd, -'d) /vt'd/ vingt-dSx
('b;~ pe&ife'i, -) p-) /'bk peKife'Kik/ banque pri#rique (\#n&m#tm'p, -'p, -)'p) /ZnmtKp'pa/ je ne me trompe p
Lastly, this assimilation can occur even before a V\ (p'n) /p'd/ pendant (tl&mn'l) /tulmdE'la/ tout le monde e l as even before a pause: (tl'mn) /tul'md/ tout le monde (m'l:) /ma'lg/ ma langue It can also be found between a
non-nasalized V and N\ (&mnm'zEl) /madmwa'zEl/ mademoelle (&mnm'zEl)
/mEdmwa'zEl/ mdemoell (&#nm'de) /Kdm'de/ redemander (&nmi'e) /admi'Ke/ admirer (&_,Nns'+i, -n-) /djagnOs'tik/ diagnoic (&\m'm) /Zb'm/
enjambement and adverbial -ment: (fn'm) /fKwad'm/ froidement (v'm)
/vag'm/ vaguement (&kpn'm, -n'm, -'m) /kplEt'm/ compltement
Let us also note cases such as (li'Em p'/, -mp p-) /lwi'kEnd pKO'S/ le week-end pro>Pn
4.3.2.6. In uent, familiar speech, there are certain reductions (even for stress),
which simplify speech, without compromising communication; instead, by slowing down speech, pronunciation can correspond to the phonemic transcription:
(s'tm) /sE'tOm/ cet homme (st#'fm) /sEt'fam/ cette femme (s't:) /asE'tK/ cette hSre (st'_i:) /sEta'diK/ c'e--dire ({m}p&tt, -tX) /{mE}p'tEtK/ [ma] pSt-tre
(&:) /a'vwaK/ avoir ('l, v'l) /vwa'la/ voil (vlPT'/P:z, &l-) /vwalaotK'Soz/
voil autre >ose (&s, )'s-, p's-) /m'sj/ monsiSr (&kte'm) /ekute'mwa/ coutez-moi (syz'm) /Ekskyze'mwa/ excez-moi (s&pE) /silvu'plE/ s'il vo plat
In a preintoneme, /swa'st/ soixante is currently reduced to ('s#;t), especially
in compounds: (s#t'sis, &s#d_iz'nf, -n_-) 66 79
Furthermore: (ps'#, &ps#, &ps, &s, s#, s, sk) /paKs[]k/ parce que (&s#, s#) /Esk, sk/ e-ce que {(p'k#s #&+y_i's) pourquoi e-ce que tu d a?]
(si'si) /sli'si/ celui-ci (&pi) /'pi/ pu (&pis#, -s, -sk) /'pisk/ puque ([]'b)
/[E]'b/ [eh] bien (bin bi'n would be better), ('p) /'ply/ pl (possibly: p'] (m&f) /mE'f/ ma enn [m'enn] (m&l;) /mEa'lOK/ ma alors [m'alors] (&b) /'b/
bon! (ban! would be better), (&n) /'n/ non! (nan! would be better), (', -E, -e, -)
/*'wE, -e/ oua! (&k;) /'kOK/ encore (st#&m, /t#-) /Zyst'm/ juement
Other examples: (dm&~,E;, t-) /dtutma'njEK/ de toute manire (tf&s) /dtutfa's/ de toute faon (&k#) /puK'kwa/ pourquoi (+yl&m) /natyKEl'm/ naturellement (s&) /sl'm/ sSlement (D&\:) /tu'ZuK/ toujours (s&p, &p) /nEs'pa/
166
n'e-ce p? (D&\) /de'Za/ dj (D\'ne) /deZ'ne/ djSner (s&pi's, s&p-) /Eksplika'sj/ explication (\'s) /ZEs'tj/ geion (&t) /'katK/ quatre ('f#) /kElk'fwa/
quelquefo ('/P:z) /kElk'Soz/ quelque >ose (t'kU) /tuta'ku/ tout coup (t'l:) /tuta'lK/ tout l'hSre (tP'py[s]) /tuto'ply[s]/ tout au pl ('v) /avEk'vu/
avec vo
More still: (/i'l, /si-, //i-) /Zsi'la/ je su l (\ie'_i \,e-) /Zlie'di/ je lui P
dit (&te'z) /tyaKe'z, ta-/ tu raon (t'f, te-) /tyE'fu, tE-/ tu fou (t&s) /ty'sE/ tu sa (&+yv'_i, &+-, &t-) /tyavE'di, ta-, ta-/ tu ava dit (l'b,) /ilE'bj/ il e
bien (v&zve'vy) /vuzave'vy/ vo avez vu (v*&ze'te) /vuvuzaKe'te/ vo vo arrtez? (n&zn'l) /nunuzna'l/ no no en allons (&is'b) /ils'b/ ils sont bons
(&iz'pi) /ilz'pKi/ ils ont pr (,) /ilja, ja/ il y a (&,n'vE) /ilina'vE, ilj-, j-/ il y en
avPt (uneducated even (&,n'vE)). In fast speech, y si ni tu ou, o before a V
often and usually, have consonant variants (often condemned by schools, all because of disguising spelling): (sil'v, s-) /siEl'v, sjE-/ si elle vSt (ni'n, ~,-) /nia'nu, nja-/ ni no (',, -) /ua'lj, wa-/ ou Lyon ('tEl, -) /uE'tEl, wE-/
o-e-elle? (i[]&fPi'le, -P,-, -P+-) /ilfoia'le, -oja-, -ot-/ il faut y aller
In rapid speech, between nasalized V, continuous C can be nasalized a little, but
it is not worth transcribing as it is barely perceptible: (&m\'b) /mZ'b/ mon
jambon ('v) /'v/ en vPn; the same occurs for oral V preceded and followed by
N\ (n'nEt) /nO'nEt/ nonnette (y&nnimi'te) /ynanimi'te/ unanimit On the other
hand, in other languages, such as English, Spanish, Italian, the nasalization of V
between N is more obvious, but this too, is hardly worth mentioning as it occurs
quite automatically.
An alveolar semi-lateral, /l/ (), is possible for the grammemes (articles or pronouns)
le la les lui (non-nal) preceded by the grammemes par pour sur vers: (&plf'nEtX, -- &pi'_i:, -i- &sYl'tba, -- &vl#'s:, -#-) /paKlef'nEtK, puKli'diK, syKla'tabl, vEKl'swaK/ par les fentres pour lui dire sur la table vers le soir
Liaison
4.3.3.1. For the all-important liaon, it must be said that it concerns, to dierent degrees, all types of pronunciation: from a minimum of linking in familiar
speech, to a maximum which can be found in classic poetry (a certainly more elaborate language). More or less in between, we can nd current conversation (real
language). Liaon only occurs within rhythm groups, between words which are
linked from a morphosyntactic and semantic point of view. Some are obligatory,
others impossible, some optional, depending on the style of diction and the choices of the speaker.
e most normal and frequent linking occurs with: /z/ [s x z] /t/ [t d] /n/ [n]\
(&lez'mi) l am ( m t s d c] (&d'z:) dSx hSr (&+il'le) e-il all? (&}g'tm) un grand homme (&pet'tE:) pied--terre (&ne'te) en t (&mn'mi) mon ami ( ton son once (&mn-) was neutral, but today it is no longer so,
although this pronunciation is still widely used), (&n't) on attend (&}nP'tEl) un
htel ( aucun htel] (&,3n'fE:) rien fPre (&b,3n'se) bien sez
4. french
167
Other forms with nasalized V do not link, except bon and adjectives such as
plRn vPn ancien certPn pro>Pn soudPn vilPn (but they lose their nasality):
(&bn'mi) bon ami (&p'nE:) en plRn Pr (l#&m,'n:\) le Moyen-ge
We never nd liaon after et, or before the names of numbers or a disjunctive
h, (or aspirated h, because it was pronounced like that centuries ago!) and generally, before w y\ (e'l:) et alors (ie'El) lui et elle (&is':z) ils sont onze (l#'^)
le un (l#'it) le huit (dee'P) d hros ('P) en haut (&}is'i) un whky (}',P:t,
}',t, }',) un ya>t and (}'i) un oui, as well.
Even in current conversation, it is obligatory to link a noun with the determinants that precede it: (le'zm) l homm (&sez'zP) c ox (te'z,) t ySx
(le&zPt#'zm, &lezPd'zm) l autr homm (&l#z'mi) lSrs ami (&dz'mi) dSx
am (&deg'zm) d grands homm (}&gt'mi) un grand ami
Subject pronouns and verbs are linked: (&nz'v) no avons (&vz'le) vo allez (il'zEm, i'zEm) ils Pment ( (i'lEm) il Pme] (&n'vE) on avPt (\&nep'le) j'en
P parl (&+ilv#'ny, &+iv'ny) e-il venu? (&+il'vy, &+i'vy) ont-ils vu? (p't) pSt-on?
4.3.3.2. ere is a case in which pronunciation does inuence spelling, and this
should be remembered. It occurs in nearly all imperatives, without -s when isolated, but with /z/ for the pronouns y and en: v-y (v'zi), pens-y (&ps#'zi), mang-en
(&m\#'z)
As well as forms such as va-t'en! (with elision of the pronoun te) we nd analogical euphonic t's in questions whith subject-auxiliary inversion: e-il? but Pme-t-on? viendra-t-elle? convPnc-t-il?
Similarly, the adverbs trs tout bien are linked to adjectives (or adverbs) modied by them: (&tzy'+il) trs utile (&tt'+e) tout entier (b,3&nt'+if) bien attentif
Instead, p pl moins trop fort sez jama can link or not; but in normal
conversation they do not usually do so: (&p[z]'k:) p encore (&tPett',
t&pe-) trop troitement In xed expressions they obviously link; for instance,
pl as in (&pyz'm) pl ou moins which is always so.
(Monosyllabic) prepositions and conjunctions link to the forms that follow
them: (&ni'vE:) en hiver (&dz}'m) dans un mo (s'zEl) sans elle (/e'z) >ez
Sx (/e&z}n'mi) >ez un ami (k&+ilv'ny) quand il e venu (but linking is only
possible with the interrogative adverb, as in (k&[t]silv'ny) quand e-ce qu'il e
venu? but not in ('k +ilv#'ny, +iv'ny) quand e-il venu? so as to avoid /tt/).
If the prepositions and conjunctions are polysyllabic, a liaon is equally only possible: (&p&v/'te, &pz&v/'te) aprs avoir >ant (d#&vynme'z, -+y-)
devant une maon ere is linking with dont en, too: (l#/'fl d+i&lp'le) le >eval dont il a parl and en y have a pre-liaison: (&p#n'z) prenons-en (&nzi'sm)
no y somm
th auxiliaries and semiauxiliaries, liaon, though possible, is becoming more
and more unusual in current conversation. It nearly always occurs between e (and
often sont ont] and a following adjective or past participle, especially with all\
(s&t3p'siba) c'e impossible (i[]&st'le) ils sont alls (&i[l]l'+) ils l'ont S (e ils l'ont
tu). ere are even some xed expressions which require linking: (le&/zeli'ze) l
amps-lys (&leze&tzy'ni) l tats-Un (&viz'vi) v--v (d#&tz't) de temps
en temps
168
4.3.3.3. On an uneducated level, liaon behaves in a particular fashion, in as

much as it is less frequent, but with analogical non-neutral extensions; furthermore, it generally marks the plural. erefore, monosyllables link with /z/: (le'z,)
l ySx (and so ySx is normally ('z,), even when isolated), (&iz'i:v) ils arrivent
Among the monosyllabic verbs, ont sont do not link, whereas su and e optionally do: (&iz'y) ils ont S ([]&s'le) ell sont all (i&li've, -t-) il e arriv (/ie'te, -ize-) je su t (which is uneducated, for j'P t but (/i&by'+i) je su abruti] an adjective links with the noun: (&bz'mi) bons am but sans may link or
not: (s'v:, &sz-) sans avoir tout and on mon ton son link: (&tt'b,) tout
e bien ('n) on a (&mn'mi) mon ami
ere are even analogical improper cases such as: *(&pza'p) /pa'p/ pS pS
*(i&vte'v,) /ilvae'vj/ il va et vient *(&ifP&dt'le) /ilfodKaa'le/ il faudra aller *(le&zi'kP) /leaKi'ko/ l haricots *(&st't) /sE't/ c'e hontSx *(/i&te3'te) /ZsieK'te/ je su rRnt *(tl'm zmy'z) /tEl'm amy'z/ tellement amant *('v
'zm) /'v 'tOm/ vingt homm *('d zfi'se) /'katK Ofi'sje/ quatre ociers *('s
'zf) /'s 't/ cent ufs
Considerations on stress
4.3.4.1. It is merely in theory that French only has a primary stress at the end
of each rhythm group. e previous examples have quite thoroughly shown the
use of secondary stress in French rhythm groups; they generally alternate, moving back from the primary stress.
Let us now consider some other examples and some structural dierences. We
normally nd: (&m\e'k:) mangez encore (yn&p#+it'fi,) une petite lle (&lmit'pE:) l'ami de Pierre unless there are two rhythm groups (to emphasize the rst
part, for a particular reason): (m'\e 'k:, &ynp#'+it 'fi,, l'mi d#'pE:). Let us also note such as cases: (&lm'sy:) la morsure (l&m'sy:, l'm; 'sy:) la mort
sre (&sde'gt) on s'en dgote ('s de'gt) on sent d goutt (&lk'l:) la
coulSr (&_yv3':\) du vin rouge (&lk'l; _yv3':\) la coulSr du vin rouge
(\v'dE _y'v ':\) je voudra du vin rouge
Traditional teaching describes the French stress in rhythm groups as follows:
with no emphasis, there is a primary stress on the last syllable, and, for the rest of
the rhythm group, the lexemes reduce their stress to secondary, whereas grammemes lose it completely.
To native speakers, this distribution sounds a bit too precise, text-book; it is
actually, a kind of mid-way between two kinds of transcriptions for the examples,
seen above, respectively with one or more intonemes: (m&\e'k:, ynp#&+it'fi,, l&mit'pE:). In reality, as we have seen, there are certain rhythmic needs, for which
the use of secondary stresses is quite dierent.
4.3.4.2. Besides, an emphatic stress is shown by the addition (rather than by
the shift) of a strong stress on the rst syllable of the word (which is dierent
from the stressed syllable) and the gemination (or lengthening, especially between
V] of the initial C; even if the word is spelled with an initial V, the C is there any-
4. french
169
way, inasmuch as we have V ().

Traditionally, we speak of accent d'insistance aective ( with an imperative
preintoneme and a half-high rst stressed syllable in the preintoneme): (1sff%midba3 3, 1sf:-) c'e formidable! (m:%Nifi3 3, 1mm-) magnique! (1stt_,3 3, 1s_,3 3, -t:-, -:-) c'e odiSx! (1k:et3 3 -kk-) quel crtin! and
of accent d'insistance intellectuelle ( with an emphatic preintoneme): (yn'Ega2\ ':p%sly3 3) une rgle absolue! (s#&sip'pf%t#"m %kEkt3 3, -'p:-) ceci e
parfPtement correct!
In cases such as c'e odiSx! we can also hear (1st-), above all as a ridiculous
habit which is typical of people in the media.
Intonation
4.3.5. 4.3 gives the preintonemes and the four intonemes of neutral French.
It can be noted that, compared to other languages (and variants of French itself),
instead of four preintonemes, there are ve. Indeed, for partial questions, as well
as the normal (echo-like) anticipation of the interrogative rise, we nd a general
falling movement, from half-high (similar to that of an imperative preintoneme,
except for the obvious and correct dierence of echo anticipation).
A more structured solution, rather than adding a fth preintoneme, would be
a split in two (something like // and //, or even // and //, or //, or
//), leaving the task of clarifying the oddity to observation explanations.
We prefer a more concrete approach ( 11.9). A careful examination of all the
preintonemes (and intonemes), as well as of the symbols is revealing.
It is worth carefully observing the position of the pretonic syllable of the continuative and conclusive intonemes: although it does not manage to be really high,
it is signicantly raised, and the auditory eect is noteworthy, as well (even if not
immediate), so much so that it contributes to giving the utterance a sort of prominence, (too) often mistaken for stress.
As we will see, for the mediatic accent, such prominence is increased by the
half-lengthening of the V in the pretonic syllable (again, without any particular
4.3 Modern neutral French preintonemes and intonemes.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (% 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 1 1)
/& / (& 1 1 2 2 ' 2 2 ' 2)
// (2 5 1 1)
/ / ( 1 1 2 2 ' 2 2 ' 2)
/,/ (% ' 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
170
stress).
We only give examples of the three marked intonemes (referring the imperative
and emphatic preintonemes to 4.3.4). As can be seen in the transcriptions of the
text ( 4.5), there is the mid parenthesis, as well ( 13.24 of NPT/HPh):
/./: (\v'le Psi%nem3 3) /Zva'le osine'ma./ Je vSx aller au cinma.
/?/: ([&s#]v&p%le'b, fsE1 1) /[Esk]vupaKle'bj fK'sE?/ [E-ce que] vo parlez bien frana? (p&le%v'b, fsE1 1) /paKlevu'bj fK'sE?/ Parlez-vo bien
frana? (&1km tlev3 3) /&kO'm tale'vu./ Comment allez-vo?
//: (&pi%'le v5+y:1 1 %'pe3 3) {(-%,-, %-)} /pia'le vwa'tyK ua'pje./ On
pSt y aller en voiture, ou pied.
Other accents
4.4.0. It will be interesting to compare what has just been said about neutral
pronunciation to other pronunciations which present more or less dierent characteristics.
International accent
4.4.1.1. For teaching purposes, the international accent could be even more
appropriate than the neutral accent seen so far, inasmuch as it is considerably less
inuenced by Paris and, therefore, undoubtedly nearer to many other varieties of
pronunciation, once any marked peculiarities have been eliminated.
erefore, an international pronunciation might even be more advisable (and,
generally, easier to pick up and command), giving unquestionably valuable results,
as long as there is coherence within all elements and no improvisation or variation
among dierent types (including the usual interference, not only phonological,
but due to spelling).
e vocogram in this section gives the international vowel articulations
which, as can be observed immediately from the comparison with 4.1, are less
marked; in particular, this refers to (u, o, O) which are back and not back-central;
even (a) is less peculiar, as it is central, even if fronted. e nasalized vowels are
still four, (, ^, , ). In the gure there are two grey markers, for /E, , O/ ( unstressed a third would be for /, ^/, which, however, coincides with //); but
could be removed to simplify the structure.
As a matter of fact, there are two satisfactory practical solutions, as, given the
lack of stress, either quite spontaneously we already have the lower-mid timbre ((,
, #, }, ), which we prefer to use); or the higher-low timbre ((E, , , ^, O)), if
weakened, could be acceptable. erefore, it is better not to introduce less important taxophones, keeping (i, y, u, a) in the context /K/: (&oZu'di) aujourd'hui
avoiding devoicing too.
Let us look at some examples, only for the cases in which there is a phonetic
dierence with the neutral (given in round brackets): ('lu) {('l)n} loup ('o) {('P)n}
4. french
171
('nOt) {('nt)n} note ('la) {('l)n} l (d#'d) {(d#'d)n} dedans ('bj) {('b,)n}
bien (bj'to) {(b,3'tP)n} bientt (pa'ti) {(p'+i)n} parti
4.4.1.2. e international accent presents more general and natural simplications for consonants as well; indeed, the preconsonantal nasals can be homorganic, and we can do without all the peculiar taxophones, such as the stop taxophones; furthermore, /j, K/ could always be approximants, (j, ) (or perhaps, even
(K); while substantial use of () is not as good; /w/ is velar rounded, (w). Even for
/S, Z/, of course, (S, Z) suce (instead of (/, \)). e frequent devoicing of sonants
is not necessary either, whereas for the devoicing of diphonic pairs something in
between would do: (yM'fam) {(yn'fm)n} une femme (&yka'vat) {(&ynk'vt)n}
une cravate (ty'di) {(+y'_i)n} tu d (kaut'Su) {(T'/U)n} caout>ouc ('kOk) {('k)n}
coq ('gid) {('id)n} guide ('pje) {('pe)n} pied ('kij) {('i,)n} quille ('a: 'Ka:K)
{(':)n} rare ('tu, 'tKu) {('t)n} trou ('mwa) {('m)n} mo (m's, &an'dOt)
{(mt's, &n'dt)n} mdecin anecdote
e length of vowels and consonants corresponds to the neutral accent, but can
be more attenuated, aiming at, above all, avoiding the dierences in length due to
interference from the mother tongue. Liaisons may be less frequent and intonation, less peculiar, is shown in the given tonogram.
/u/ (u[:])
/i/ (i[:]), /y/ (y[:])

/e/ (e), // ([:])
// (#)
/E, / (E[:], [:]) {(, )}
/, ^/ ([:], ^[:]) {(#, })}
/a/ (a[:])
/o, / (o[:], [:])

/O/ (O[:]) {/O/ ()}
// ([:])
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 5 1 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
Mediatic accent
4.4.2.1. is accent is spread by television and radio. It is based on the Parisian
accent, and shares quite a few characteristics with modern neutral, although it
shows further evolution which moves it further away from the international accent ( 4.4.1).
172
It also manifests some uctuation both towards and away from neutral; however, we show it here in its most typical form, indicating some internal dierences
too, for instance of an uneducated or suburban level (a peculiarity of the Parisian
banliSe).
In this section the vocogram is given of the vowels for which there are denitely gradations (comparing it to g 4.1, of neutral); here, we bring your attention
to the major dierences. e fronting of /u, O/ (%, @) (and /O/ (), unstressed)
is evident; the (anti-clockwise) rotation of /, , / (a, _, ), (and // ()), is just
as evident, as is the absence of /^/ which merges with // (a), or varies between
the two, even with an intermediate realization (not shown explicitly in the vocogram) (8), with slight rounding. (In the uctuations towards the neutral accent,
we have (), as well; or (), with a slight derounding).
e /EK, K, OK/ sequences often shift back one box: ([:], @[:], [:]
). Furthermore, note in the vocogram the values for a\ (even if with individual uctuations) the distinction is generally maintained between /a/ and /A/,
which traditional neutral had adopted. However, the timbres (of Parisian and) of
the mediatic accent are: /a/ (), but () for /aK[0], wa/, and /'A[0]/ (), but /A/
(A) (in the banliSe we have /wa/ (A, '), often /aK/ (:) ar(C) even more
often /aj/ (:,), but (s) for /Asj/ -ation of traditional neutral).
Some examples follow (where t indicates traditional neutral): ('t) {('tU)n} tout
('n@t) {('nt)n} note (b'n@:) {(b'n:)n} bonhSr (&P\T'_i, --) {(&P\'_i)n}
aujourd'hui ('b,a) {('b,)n} bien (\'t_) {(\'t)n} j'entend ('b) {('b)n} bon (P'a,
-8) {(P'^)n} aucun
Others: ('p:) {('pE:)n} pre ('p@:) {('p:)n} pSr ('p:) {('p:)n} port ('p@
's:) {('p 's:)n} Paul sort (p'i) {(p'i)n} Par (p'+i:) {(p'+i:)n} partir
(v'l) {(v'l)n} voil ('s#:) {('s#:)n} soir (':) {(':)n} car Also: (p'p)
{(p'p)n} papa (pA's:\) {(p's:\)n} psage ('g) {('g)n, ('gA)t} gr ('g:s)
{('gs)n, ('gA:s)t} grse ('_,:ba) {('_,ba)n, ('_,A:ba)t} diable
For the banliSe accent; (vA'l) {(v'l)n} voil ('s#A:) {('s#:)n} soir ('k:)
{(':)n} car ('p:,) {('p,)n} pPlle (t'v:,) {(t'v,)n, (-A:,)t} travPl (p&ns's) {(p&ns's)n, (-sA-)t} prononciation
ere is a strong tendency to merge /e, E/ into /e/ (but with many oscillations
due to hypercorrection and undecidedness; some even use the intermediate timbre, (), always or above all for -P): (p%'le) /pu'lE/ poulet
4.4.2.2. For the C, as well as for /K/ (as seen in some examples), which is typically (), but can be as well as in neutral even (), especially to give emphasis: ('y)
{('y)n} rue ('t%) {('t)n} trou ('tx) {('tX)n} quatre we must add that on an
uneducated level, the palatalization of /t, d k, g/ is much more evident, with articulations going from stops to stopstrictives, (, , k): (y'i) {(+y'_i)n} tu
d ('mEk) {('mE)n} mec ('id) {('id)n} guide; /w/ is provelar rounded, ():
('m) /'mwa/ mo.
e length in the mediatic accent, apart from what has already been said for the
neutral one, presents a typical half-lengthening of the vocoid of the unchecked pretonic syllable, the syllable that precedes the stressed syllable of an intoneme (of-
4. french
173
ten this prosodic phenomenon is incorrectly described as a shift of stress from the
last syllable to the penultimate in a rhythm group): (%p;'i) {(%p'i)n} Par (&p%i;'z,a) {(&p%i'z,)n} parien (in these examples, even for neutral French, we explicitly highlight the marked height of the pretonic syllable, which is almost half-high in the two accents, but in the mediatic one there is also the half-lengthening, seen above). e tonogram gives the intonation of the mediatic accent:
make all the necessary considerations, paying a (fair) bit of attention.
/i/ (i[:], I)
/u/ (%[:], T) (+/OK/)
/y/ (y[:], Y)
/e/ (e), // ([:])
/o/ (P@[:]), // (#[:])
/, / (#)
/O/ (@[:], ) {/OK/+([:], )}
/E/ (E[:], ) {/EK/+([:], )}
// (_@[:], @)
// ([:], #) {/K/+(@[:], )}
/A/ ([:0], A)
/aK{0}, wa/ ([:]), /a/ ([:]), /, ^/ (a[:])
/E/ (E, e), -ai (vb.) (e, , E)
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (% 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/& / (& 1 1 2 2 ' 2 2 ' 2)
// (2 5 2 2)
/ / ( 1 1 2 2 ' 2 2 ' 2)
/,/ (% ' 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
Southern accent (Marseilles)

4.4.3.1. We will now look at one of the accents which is furthest from the neutral accent (but also from the international and mediatic ones): Marseilles
French as a good representative of the pronunciation of the Midi As can be seen
from the vocogram, the V are few: seven plus //. ere is no phonemic opposition between /e, E , o, O/; least of all between /a, A/. For /E, , O/ we always
have (e, , o) (even in unstressed syllables), except in stressed syllables followed by
C (with or without //), where we nd (, #, ).
For /a/ there is a central timbre, (a). e articulation of // is unrounded central, (), except in contact with /K/, where there is rounding, (+); it corresponds to
every e in the spelling (except for Ve\ amie = ami (a'mi)}, and it often appears between C.
Furthermore, the nasalized V, /, ^, , /, are nothing but a sequence of oral
vowels and a nasal contoid homorganic to the following C but, before a pause,
174
we nd the semi-provelar, (). e timbre of the nasalized vowels is: (e, , o, )

in unstressed syllables, (, #, , a) in stressed syllables, but (e, #, o, a) in absolute nal (stressed) syllables.
e only kind of normal lengthening, apart from emphasis, is the half-lengthening, in an intoneme, of the V (followed by a C, with or without //), and the diphthongization of the nasalized V in absolute nal position before a pause. In less
broad pronunciation, the lengths can be more similar to those of neutral pronunciation (as the timbres and the distribution of /'E, ', 'O E0, 0, O0/, as well).
Some interesting examples follow: ('tu) {('tU)n} tout ('tre) {('tE)n} trs (te're)
{(t')n} terrPn (sl'ma) {(s#l'm)n} sSlement (&prom'nad) {(pm'nd)n}
promenade ('ve) {('v)n} vin (&bjene'me) {(&b,3ne'me)n} bien-Pm ('mns) {('m:s)n}
mince (o'k#) {(P'^)n} aucun (ln'di) {(l}'_i)n} lundi ('#mbl) {('^:ba)n} humble
(mo'no) {(m'n)n} mon nom ('nmbr+) {('n:bX)n} nombre (n'ta) {(\'t)n} j'entend ('lamp) {('l:p)n} lampe ('di;r) {('_i:)n} dire ('vwa;r) {('v:)n}
voir (p'n) {(p'n)n} pnS (&t&ld'mand) {(/&t#ld#'m:d)n} je te le demande
(&n&tl+&r+d'mand 'pa) {(\#n&t#l#d&mn'p)n} je ne te le redemande p
A less broad accent can have some partially nasalized V in stressed syllables:
('mens) {('m:s)n} mince (i'e) {('/)n} >ien ('#Jmbl) {('^:ba)n} humble
('br#J) {('b^)n} brun ('laAmp) {('l:p)n} lampe ('baA) {('b)n} banc ('nmbr+)
{('n:bX)n} nombre ('b) {('b)n} bon.
4.4.3.2. For the consonants, we can observe that all the taxophones of neutral
pronunciation are not present; furthermore: /n/+/0/ (m, M, n, ~, N, ), /nm, nK/
(m, r), /S, Z/ (, ), /j/ (j), /w/ (), /N, nj/ (~j), /Ni/ (ni), /lj, l/ (j, ), /'[0]/+/j,
, w/ ([0])+(i', y', u'), /K/ (r) (even (), especially before a consonant; and,
in less broad accent, also (K, )}: (M'v;r) {('vE:)n} envers ('bak) {('b:)n} banque ('ri) {('i)n} Henri ('r) {(\#'/E/)n} je >er>e (&trava'je) {(&tv',e)n}
travPller e banlieu accent has /t, d/ (+, _) + /i, j, y, /, and /K/ tendentially voiceless (3) (and (x, ', )).
Furthermore: (mon'ta~j) {(m'tN)n} montagne (&ma~je'rizm) {(&m~,e'is))n} manirme (je'zo) {(,e'z)n} liaon (ki'zi;n) {(i'zin)n} cuine
(&sy'i) {(\#'si, /'si)n} je su (ly'i) {('i)n} lui (lu'i) {('li)n} Lou (bi'e)
{('b,)n} bien ('ra;r+, 'a;+) {(':)n} rare Hence, (j) is maintained well only before a non-nal V; indeed, we generally have: ('pi) {('pE,)n} pZe (pe'i) {(pe',i)n}
pZs
Consonant clusters are simplied, as in: (di'r;k) {(_i'Ekt)n} direct (di'r;t) {(_i'Ekt)n} directe (os'ky;r) {(p'sy:)n} obscur (&aek'ti;f) {(&D\'+if)n} adjectif (se'tambr+) {(sp't:bX)n} septembre (a've) {('vE)n} avec but: (s) in words such as:
dix six alors lors gens Sx cSx cours vers jad av moins Roux Poux iers
Arno
ere is no devoicing, as can be seen in the examples given, neither for V, nor
for C. Liaon is rare, as a matter of fact, it only occurs with /z/ in articles, pronouns and plural masculine adjectives; however, there is no concept of a disjunctive h, hence, the use of linking and elision, : (le'z;m) {(l'zm)n} l homm
(&meza'mi) {(&mez'mi)n} m am (&noza'mi) {(&nPz'mi)n} nos am (&lrza'f;r+)
4. french
175
{(&l#z'fE:)n} lSrs aPr (&vuzi're) {(&vzi'e)n} vo irez (&i[l]za've) {(&i[l]z'vE)n}

ils avPent (&diza'mi) {(&_iz'mi)n} dix ami
Furthermore: ('gro zani'mo) {('gP zni'mP)n} gros animaux ('gran zM'fa)
{('g z'f)n} grands enfants (i[l]&nuzm'parl) {(i[l]&nz'pl)n} il no en parlent besides: (le&zari'ko) {(&lei'kP)n} l haricots (&ne'ro) {(}e'P)n} un hro
e identity of lexemes in sentences remains distinct, with no resyllabication,
contrary to what happens with grammemes: (m'bl wa'zo) {(}'bE l'zP)n} un bel
ois ('b ape'ti) {('b npe'+i)n} bon apptit (&-wa'zo) {(&}-n'zP)n} un ois (&lay're) {(&l\y'e)n} elle a jur e tonogram gives the intonation of Marseilles accent.
/i/ (i), /y/ (y)
/u/ (u)
/e, E/ (e, 0, 0, e)
/, / (, #0, #0, )
// (0, e, e)
/^/ (#0, #, )
/a/ (a)
/o, O/ (o, 0, 0, o)
// (, r+, +r)
// (0, o, o)
// (a0, a, )
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
Canadian accent (Qubec)

4.4.4.1. e accent we describe here is the normal, or rather neutral Canadian
pronunciation. ere are quite a few other variants (which ll two other vocograms, given in 4.6.3), which are both broader (and decidedly broader, or uneducated), and even less broad, that aim for neutral European or international
French; there could be many (and quite dierent) things to add, but it seemed better to simplify and present this accent homogeneously, as done with the others.
However, the variants will be included in a specic book on French pronunciations.
Before examining the timbres of the vowels, it must be said that in Canada both
/A/ and /E:/ still exist, and /e:/ can even be found. For /A/, apart from traditional
cases (still transcribed thus in dictionaries), as: ('pt) {('pt)n, ('pA:t)t} pte
('t/)n {('t/), ('tA:/)t} t>e ('ml) {('ml)n, ('mA:l)t} mle, compared to ('pt)
{('pt)n} patte ('t/) {('t/)n} ta>e ('ml) {('ml)n} malle (&pqi'se) {(&p+i'se)n,
176
(&pA-)t} ptsier we also have an expansion of its frequency, to practically all cases of /a|, wa|, av|, az|, as|, aZ|, aj|, a|, aK|, aK/ (with possible uctuation, especially
for /av|, az|/ and -ation which have both /A/ and /a/). ere is actual neutralization too, as for: ('k) which goes for both (') {('kA)t} c and (') {(')t} k however, the contextual distribution with /A/ in an intoneme leads to situations
such as: ('s '\E 's) a j'Pme a
ere is uctuation between (w|) and (w|) (the latter is generally considered
to be more uneducated), up to (w), decidedly dialectal: ('mw 'mw 'mw) moi
e old opposition between /E/ and /E:/ is stable, we have for example: ('mEt5)
/'mEtK/ mettre ('mEt5) /'mEtK/ {/'mE:tK/t} matre ('fEt) /'fEt/ fPte ('fEt) /'fEt/
{/'fE:t/t} fte ('sEn) /'sEn/ sPne ('sEn) /'sEn/ {/'sE:n/t} scne (which now, in neutral
European French, is only: ('mEtX, 'fEt, 'sEn)n); whereas for the new opposition
constituted by /e:/, we have examples such as: ('neI\) /'ne:Z/ nRge ('bReI)
/'bKe:k/ brque (or brake] (/'nEZ, 'bKEk/n ('nE:\, 'bE)n).
4.4.4.2. In checked stressed syllables in /v, z, Z, K, vK/, all recurring V are realized as diphthongs (here are only some examples): ('lIiv5) /'livK/ livre ('pYyR) /'pyK/
pur ('RUu\) /'KuZ/ rouge ('R\, 'R\) /'KaZ/ rage (('li:vX, 'py:, ':\, ':\)n). e
same occurs for /E:, e, , o, A/ in an intoneme for any checked syllables (without
repeating examples): ('fYt5) /'ftK/ fStre ('\oUn) /'Zon/ jaune (('f:tX, '\P:n)n), and
even in a preintoneme for unchecked syllables (again, without repetition): (&Re'te) /aKe'te/ arrter (neI'\e) /ne'Ze/ nRger (fY'te) /f'tKe/ fStr (\oU'nt5) /Zo'natK/ jauntre (f'/e) /fa'Se/ f> ((&e'te, ne'\e, f'te, \P'ntX, f'/e)n).
In the broader or more uneducated accent, there is even the fusion of these two
characteristics, whereby for an unchecked syllable in a preintoneme, the V often
diphthongize if followed by /v, z, Z, K, vK/ (the rst vocogram of 4.6.3 shows (e',
#', o')): (&qIi'Re) /ati'Ke/ attirer (\Yy\'mA) /ZyZ'm/ jugement (&epUu'ze) /epu'ze/
poe (te'Re) /tE'K/ terrPn (b#'Re) /b'Ke/ bSrre (Y'R) /'K/ hSrSx
(&Rlo'\ER) /OKlO'ZEK/ horlogre (&ek'ze, --) /ekKa'ze/ cr or if they are /e, ,
o/ (even when non-derived or in syntagms): (leI'pe) /le'pje/ l pieds (dY'me)
/d'm/ dSx mPns (soU'f) /so'fa/ sofa
Nasalized V diphthongize in an intoneme ( both in tonic and pretonic syllables), but they are short in a preintoneme: ('fe) /'f/ n (se'qYyR) /s'tyK/ cRnture
(o'}J) /o'k^/ aucun (}J'\UuR) /^'ZuK/ un jour (m9'n9) /m'n/ mon nom (\A'tA) /Z't/ j'entends (&fi'ni) /fi'ni/ inni (&}n'mi) /^na'mi/ un ami (&mn'mi)
/mna'mi/ mon amie (&ntA'dA) /nt'd/ en entendant (('f, s3'+y:, P'^, }'\:, m'n, \'t &3fi'ni, &}n'mi, &mn'mi, &nt'd)n).
4.4.4.3. Another notable peculiarity is that /i, y, u/ in a checked syllable
(stressed or not), are lower-high: ('vIf) /'vif/ vif ('lYn) /'lyn/ lune ('tUt) /'tut/ toute
{('vif, 'lyn, 'tt)n), as even in the rst unchecked syllable of a word or a rhythm
group: (fI'lIp) /fi'lip/ Philippe (Y'nI) /y'nik/ unique (kU'zIn) /ku'zin/ coine and
also in other subsequent unchecked syllables, in words or rhythm groups that end
in checked syllables: (&pImI'qIiv, pI'q-) /pKimi'tiv/ primitive (&\YRI'QI, \YR'Q-)
/ZyKi'dik/ juridique other examples: (&mInIs'tER, mIns-) /minis'tEK/ minire (&-
4. french
177
k.s'qI) /akus'tik/ acouique {(fi'lip, y'ni, k'zin &pimi'+i:v, &\yi'_i &minis'tE:, &ks'+i)n). e less broad accent can limit the occurrence of (I, Y, U) only in
checked syllables, or only in stressed checked syllables, as well; the decidedly less
broad accent, and more controlled, could have no (I, Y, U) at all.
Again, /i, y, u/ present two other peculiarities (except in slow or controlled
speech); indeed, when unstressed, they are devoiced between voiceless C, in
checked or unchecked syllables: (&kfN'qYyR) /kfi'tyK/ conture (&qNs'qI) /aKtis'tik/ artiique (&QIsp'te) /dispy'te/ dput (&deku'p\) /deku'paZ/ dcoupage
{(&kfi'+y:, &+is'+i, &_ispy'te, &dek'p:\)n).
ey can be dropped between a constrictive or sonant and another C, or between a stop and a non-stop (even other V above all, /e/): (&pfe's#R, pf's-) /pKOfE'sK/ profsSr (&Ynivsi'te, &Ynvs'te) /ynivEKsi'te/ universit (&bili'te, &bI'te)
/abili'te/ habilit (&mNI'fI, m'-, mJ'-, m9'-) /maNi'fik/ magnique (bI'zR, b'z-)
/bi'zaK/ bizarre (p&pyl's9, &pp-) /pOpyla'sj/ population (e&pyR's9, &ep-) /epyKa'sj/ puration (&buj'ne, bj-) /bujO'ne/ bouillonner (&pin'je, pn-)
/pina'je/ pinPller (&Re'je, R'je) /OKe'je/ orRller (&lk'mOd, l'Od) la commode
{(&pf's:, &yni&vsi'te, &bili'te, &mNi'fi, bi'z:, p&pyl's, e&py's, &b,'ne,
&pin',e, &e',e, &lk'md)n). Even // drops more often than in European French:
('lY s+p'mEn, 'lYks p-) Luc se promne {('ly s#p'mEn)n}, and presents even distributions which are often considered uneducated, like: ('fER+ 'f) fPre le fou
{('fE; l#'fU)n}. Furthermore, there is generally no vowel adjustment: ('me, 'mE)
in a broader accent (e-) {(e'me, 'mE)n} Pm Pma
/i/ (i, I, Ii), /y/ (y, Y, Yy)
/u/ (u, U, Uu)
/e/ (e, eI), // (, Y)

// ('e, e', ), // (+)
/E/ (E, E, )
// (, #, #), /^/ ('}J, }J', })
/a/ (, ), // ('A, A', )
/o/ (o, oU)

/O/ (O,
@ @O, )
@
// ('/9, 9'
/ , )
/
/A/ (, )
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
4.4.4.4. For the C, besides the examples already given, we can observe that /nj,
n/ are (~j, ~) (on a broad, uneducated level /nj/ becomes /N/): (p'~je) /pa'nje/
panier (A'~i) /'ni/ ennui {(p'~,e, '~i)n); /N/ varies between (N) and (J):
(m9'tN, -'tJ) /m'taN/ montagne (&sN'mA, &sJ'-) /sEN'm/ ensRgnement
178
{(m'tN, &sN'm)n). Furthermore, /w/ is (w): ('mw) /'mwa/ mo

Of all the stops, /t, d/ present a typical stopstrictive realization, (q, Q), before
/i, y j, /: ('qIp) /'tip/ type ('QYyR) /'dyK/ dur ('qe) /'tj/ tiens (k9'QIiR) /k'diK/ conduire {('+ip, '_y:, '+, k'_i:)n) It is possible for this articulation not
to extend beyond the rhythm group or word: (s'tIl, s'qIl) /sE'til/ Sept-l ( ('sE
'tIl) /'sE 'til/ sept l] (dI'm\, QI-, -\) /di'maZ/ d'imag ( ('Qi 'm\, -\) /'di
'maZ/ dix mag] On an uneducated level we can nd (+, _) also before /e, E, a, ,
/: (d'te, _'+e) dater {(d'te)n), and even (, ) before /j, /: ('e, k9'IiR).
In Canadian French too, we nd /k, g/ (, ) (with (k, ) on an uneducated
level) before front V (up to /, /) and before a pause: ('l) /'gl/ guSle ('kO)
/'kOk/ coq Before a pause, stops can be unreleased: ('p, -p) /'kap/ cap ('dt, -t)
/'dat/ date ('RO, -) /'KOk/ roque ('lA, -) /'lg/ langue ('RYd, -d) /'Kyd/ rude
At times, /p, t, k/ can be aspirated, after a pause or if initial in stressed syllables (which we mark here with (0)): (p'pIt5) /py'pitK/ pupitre (t&tli'te,
&tt'te) /tOtali'te/ totalit (&kR'kOl) /kaKa'kOl/ caracole {(py'pitX, t&tli'te,
&'kl)n). Even in the many English words, currently used in Canadian French,
/tS, dZ/ remain as sequences (T/, D\): (T'/Iip, T'/Ip) /t'Sip/ >eap (D'\Iin, D'\In) /d'Zin/
jeans the second variants are more assimilated, as also occurs with: ('qIim, 'qIm)
/'tim/ team
In Canadian French, /j/ is an approximant; /K/ is typically an alveolar tap, (R):
('RR) /'KaK/ rare, which constitutes traditional and neutral Canadian pronunciation; however, it is often uvular (approximant, trill or constrictive, (, K, ), especially in less broad pronunciation, or in the mediatic one, or even verging on modern, spreading from the city of Qubec).
In English words, we often nd the uvularized velar (or the prevelar one both
with a slight postalveolarization and labialization, as in the Canadian English pronunciation), (<, <): (<i'd#<, '<Iid+<) /Ki'dK/ reader (qi'/<t, qI-) /ti'SKt/ T-shirt
On an uneducated level, for French -Vr /K/, we nd the same (<), or vocalization
with complex diphthong or triphthong of various timbres (which are given in
4.6.3, grouped up for the moment). For the disjunctive/aspirated h, we can
have (h) /h/, above all in a broader accent: (A'ho) en haut {('P)n).
4.4.4.5. e groups with nal C are often simplied, above all in less controlled
language, even in the case of /0K, 0l/ (before a pause too): ('wEs) /'wEst/ oue (&/i'tE) /aKSi'tEkt/ ar>itecte (&ss'lis) /sOsja'lism, -ist/ socialme -ie ('mYs)
/'myskl/ mcle ('9, '9) /':gl/ ongle ('Es) /OK'kEstK/ or>ere ('9b, '9m)
/'bK/ ombre on the other hand, especially on an uneducated level, we can have
/`/ = /t/: (\i'jEt, \I-) /Zi'jE/ juillet (qi'Et, qI-) /ti'kE/ ticket (d+'bUt) /d'bu/ debout ('pEt) /'pKE/ prt ('~It) /'ni/ nuit ('lIt) /'li/ lit (k9'pEt) /k'plE/ complet
(bRy'nEt) /bKy'nE/ Brunet (&mRi'sEt) /mOKi'sE/ Morset (tl'bOt) /tal'bo/ Talbot
Cases such as (p'qi t'mi) /p'ti ta'mi/ petit ami and (p'qI t'mi, p'qIt 'mi) /p'ti
ta'mi/ petite amie maintain the distinction of the V in a checked syllable (of petite].
e tonogram gives the intonation of Canadian French.
4. french
179
Text
4.5.0. e story e North Wind and the Sun follows, given in seven dierent
normalized versions. We start with the (neutral) French pronunciation of (neutral British) English this is the rst step of the phonetic method (the written text
is given in 2.5.2.0). e French translation follows, in its neutral version; then
the international accent, followed by the mediatic/Parisian one and, lastly, the
Marseilles and Canadian versions.
At the end, as always, there is the version which gives the English pronunciation of French, by neutral British speakers, uent in French (after prolonged contact with native speakers, but with no help from the phonetic method), who have
same principle is valid for the foreign pronunciations of English, given rst.
Speakers of American English could prepare their own version of the French
pronunciation of English and of their pronunciation of French, as an excellent exercise, by listening to native speakers, best of all after recording them. Of course,
speakers of other languages could do the same thing. e author would be happy
to receive their transcriptions and recordings, both in case of help should they
need it and to make their contribution known to others (possibly in our website
on canIPA Natural Phonetics 0.12).
French pronunciation (of English)
4.5.1. (z#'n 'sin2 dnz#'s^n2 #_is'p+i 'i\ z#stg#3 3| 'E n#'tvl#2 'kE m#'l2 'Ep +in#' kk3 3| ze#'gid2\ zdz#' n'f s#k'si_i2 _i'mEi z#'tvl#2 'tE is'k 5kf1 1| &/bikn'sid#ts "tg# &znziz#3 3||
'zn2 z#'n 'sind#2 'bl2 'z2 dzikd3 3| &b#dz#'m i5bl1 1| z#'m 'kzli2 &_idz#'tvl#2\ 'fl _is'k k#n_im3 3| &ndt'st2\ z#'n 'sin2 'vb zi#tmt#3 3||
'zn2 z#'s^n /'nt3 3 mli3 3| &n_i'mi_,#ti2\ z#'tvl# Tkf3 3 iskk3 3|| n'sP2 z#'n 'sind#2 &z#'bli\d# tkn'fEs2| &zdz#s^n3 3 z#stg#3 3 &vz#t3 3||
&_i_,li1 1 z#s'ti2| _,'n tiit#En1 1|||)
French Text
4.5.2. La be et le solRl se dputPent un jour, prtendant l'un comme l'Qtre tre
le pl fort, lorsqu'ils virent s'avancer un voyagSr, envelopp dans son mant. L
dSx adversPr dcidrent, alors, que serPt dclar vPnquSr celui qui, le premier,
parviendrPt le lui fPre ter.
180
La be se mit, alors, souer de tout s forc; ma, pl elle souPt, pl le

voyagSr se serrPt dans son manteQ, tant et si bien, qu' la n, la pQvre be dut
renoncer s intentions. Le solRl se montra, alors, dans le ciel, et le voyagSr, qui Q
bout d'un moment commenPt avoir >Qd, retira son manteQ. C'e Pnsi que la
be dut reconnatre que le solRl tPt le pl fort d dSx.
Tu Pm cette hioire? Tu vSx la rentendre?
Neutral French pronunciation
4.5.3. (l'bi; z%s'lE,2\ s#&_ispy'tE2 }'\:2| &pe%t'd2 'l^ km'lP:tX2| &t#%pyf:3 3| &lsi[l]'vi; &s%v'se2 }&v%,\:3 3| &v%l'pe d&s%mtP3 3|| le&dzd%v'sE:2 &de%si'dE:3 3 t'l:2| #s&de%k'e v35:1 1| s#'i2 i%p#'m,e2 p%v,3'dE l[#]&if%Pte3 3||
l'bi:z2 s#mi3 3 'l:2\ %s'ft2 tt%sefs3 3| m&pys5fE1 1| 'pyl v%,'\:2 s#s'E d&s%mtP3 3|| &ttesi5b,1 1| &%l'f2 &l%pP;v'bi:z2\ &_Y%n'se se&z3%ts3 3|| &l#%s'lE,2 s#m't2 'l:2\ %dsEl3 3| l&v%,'\:2 iP&bd}%m'm2 &km'sE []v5/P1 1| &#%+i3 3 s%mtP3 3|| &st3'si k'bi:z2\ &_Yk'nEt #%s'lE,2 &tpyf:3 3
ded3 3||
+eme1 1 s%+is't#:2| +y&vlet:dX1 1|||)
International French pronunciation
4.5.4. (la'bi; zls'lEj2\ s#&dispy'tE2 }'Zu:2| &pet'd2 'l^ km'lo:t2| &tl#plyfO:3 3|
&lski[l]'vi; &sav'se2 }&vwajaZ:3 3| &vl'pe d&smto3 3|| le&dzadv'sE:2 &desi'dE:3 3 [t]a'lO:2| k#s&dekla'e v5k:1 1| s#'li2 kilp#'mje2 pavj'dE al[#]&lifote3 3||
la'bi:z2 s#mi3 3 a'lO:2\ asu'fl2 tutsefOs3 3| m&plylsu5flE1 1| 'plyl vwaja'Z:2 s#s'E
d&smto3 3|| &t[t]esi5bj1 1| &kala'f2 &lapov[#]'bi:z2\ &dyn'se ase&ztsj3 3|| &l#s'lEj2
s#m'ta2 a'lO:2\ dlsjEl3 3| l&vwaja'Z:2 kio&budm'm2 &km'sE a[a]vwa5So1 1|
&#tia3 3 smto3 3|| &st'si kla'bi:z2\ &dYk'nEt[#] k#ls'lEj2 &tlplyfO:3 3 ded3 3||
tae'me21 stis'twa:2| ty&vlae't:d21|||)
Mediatic, Parisian French pronunciation
4.5.5. (l'bi; z%s;'lE,2\ s#&_is%py;'te2 %a;'\%:2| &pe%t;'d_2 'la %km'lP:tx2| &t#%py;f:3 3| &lsi[l]'vi; &s%v;'se2 a&v%,;\@:3 3| &vl'pe d&s%m;tP3 3|| le&dzd%v's:2 &de%si;'dE:3 3 %t;'l:2| #s&dek'e %va;5@:2 2| %s#;'i2 i%p#;'m,e2 pv,a'de
l&if%P;te3 3||
4. french
181
%l;'bi:z2 %s#;mi3 3 %;'l:2\ %s%;'ft2 t%t%se;fs3 3| m&pys%;5fe2 2| 'pyl v%,;'\@:2

s#s'e d&s%m;tP3 3|| &ttesi;5b,a2 2| &%l;'fa2 &l%pP;v'bi:z2\ &_Y%n'se se&za%t;s3 3||
&l#%s;'lE,2 s#%m;'t2 %;'l:2\ %dsEl3 3| l&v%,;'\@:2 iP&b%da%m;'m_2 &km'se
[]v5/P2 2| &#%+i;3 3 s%m;tP3 3|| &sta'si %k;'bi:z2\ &_Yk'nEt #%s;'lE,2 &t%py;f:3 3 %de;d3 3||
+e;'me21 s%+is't#:2| +y&vle;'t_:dx21|||)
Southern pronunciation: Marseilles
4.5.6. (la'bi; zelso'l;j2\ s&dispy'te2 ~'u;r2| &pretn'da2 'l# kom'l;tr+2|
&t[r+]lply'f;r3 3| &lorski[l]'vi;r+ &savn'se2 M&vwaja'#;r3 3| &aMvlo'pe dn&somn'to3 3||
le&d[z]adver's;r+2 &desi'd;r3 3 ta'l;r[s]2| ks+&redekla're vek#;r2 2| s'li2 ki&lpr+'mje2 parvjen'dre al&lifro'te3 3||
la'bi;z2 s'mi3 3 a'l;r[s]2\ 2asu'fle2 d&tutse'frs3 3| &meply&lsufle2 2| 'ply lvwaja'#;r2 sse're dn&somn'to3 3|| &tntesibje2 2| &kala'fen2 la&pvr+'bi;z2\ dy&r+non'se
ase&zentn'sjo3 3|| &lso'l;j2 smon'tra3 3 a'l;r[s]2\ &dnl'sjl3 3| el&vwaja'#;r2 kio&budemo'ma2 &komn'se a[a]vwaro2 2| &r+ti'ra3 3 somn'to3 3|| &seten'si kla'bi;z2\ &dyr+ko'nEt[r+] k&lso'l;j2 e&telply'f;r3 3 ded3 3||
tae'me21 stis'twa;r+2| ty&vlaren'tandr+21|||)
Canadian pronunciation
4.5.7. (l'bIi zs'lEj2\ s+&QIsp'tE2 }J'\UuR2| &petA'dA2 'l}J km'loUt52| &t+p'fOR3 3| &lsi'vIi &svA'se2 }&vwj'\#R3 3| &vl'pe d&smA'to3 3|| le&dzdv'sER2 &desi'dER2 t'lOR2| +s&dek'Re ve#R2 2| s+'i2 Ip+'mje2 pRvje'dRE l[+]&ifRoU'te3 3||
l'bIiz2 s+'mi3 3 'lOR2\ s'ft2 t.tse'fOs3 3| m&pysfE2 2| 'pYl vwj'\#R2 s+s'RE
d&smA'to3 3|| &t[t]esibje2 2| &l'fe2 &lpoUv'bIiz2\ &QYRn9'se se&ztA's93 3|| &l+s'lEj2 s+m9't3 3 'lOR2\ d'sEl3 3| l&vwj'\#R2 io&budJm'mA2 &kmA'sE []vw/o2 2| &R+qi'R3 3 smA'to3 3|| &ste'si k'bIiz2\ &QYk'nEt +s'lEj2 &tp'fOR3 3
dedY3 3||
qe'me21 sqNs'tR2| qy&vlReA'tAd521|||)
English pronunciation of French
4.5.8. (l5bI;iz2 &s'l;I2\ s&Dspj'Th;I2 ;~'Z;2| &ph>IT'D;2 5l: &khm'l>2| &>l5phl;u 'f:3 3| 5l;sk 'v; sv's;I2 ;&vwA;j'Z:3 3| &vl5ph;I D-
182
&s;m'Th;3 3|| lI'D: Dv's;3 3 &DIs'D;3 3 'l:2| &khIs5>;I DIkl'>;I v'kh:32|

sl'I;i2 5khI;i lp>'mj;I2 &phA;vjn'>;I A;ll'I;i &f>'Th;I3 3||
l5bI;iz2 s'mIiT3 3 'l:2\ &s'fl;I2 D5ThuT sI'f;s3 3|| &mI5phl;u s'fl;I32| 5phl;u
l&vwA;j'Z:2 ss5>;I D&s;m'Th;3 3|| 5Th; I&sIibi';32| &khl'f;2 l5phv> 'bI;iz2\ 5Dju >n;'s;I &sIzTsi';3 3|| ls'l;I2 sm;'h>A:2 'l:2\ &Dlsi':3 3| Il&vwA;j'Z:2 &khIi5b;u D;m'm;2 &khm5s;I A;v'A: 'S;32| >T'>A:3 3
&s;m'Th;3 3|| &sIT5sI;i khIl'bI;iz2\ 'Dj;u >k5n> &khIls'l;I2 5T;I l'phl;u 'f:3 3 DI'D:3 3||
&ThjA;I'm;I21 &sTsT'A:2| Thju'v: l&>I'Th>21|||)
Appendix: further accents
4.6.0. We would like to add some further information: about the aected Parisian accent, in the form of phonosynthesis; and about the Parisian variants of the
banliS compared to mediatic Parisian ( 4.4.2.1-2 4.5.2.3); lastly about
Canadian variants ( 4.4.4.1-5 4.5.2.5). In NPT/HPh, you can nd the phonosyntheses of Acadian French (acadien east Canada, 21.10) and cadien, Cajun
(from Louisiana, 21.16).
Aected Parisian
4.6.1. is accent can be heard from representatives of the aristocracy and upper classes, but is not usually highly valued. We give the realizations of the vowels
/i/ (i{:}, I)
/y/ (y{:}, Y)
/e/ (e), // ({:})
/u/ (u{:}, U)
// ({:}), /o/ (o{:})
/O/ (O{:}, )
// ({:})
( ('a, a, a, wa))
/E/ (E{:}, ), /, ^/ ({:})

// ({:}, #), // (#)
/a/ ('A, 'A0, a', )
/ / (2 2 ' 2 2 ' 2 2 2)
/./ (% 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 2)
/?/ (2 2 1)
/& / (& 1 1 5 2 2 2 2 ' 2)
// (2 5 1 1)
/ / ( 1 1 5 1 2 ' 2 2 ' 2)
/,/ (% ' 2)
/ / ( 2 2 ' 2 1 ' 2 1 2)
4. french
183
and intonation, which should be carefully compared to the neutral realizations

(and, possibly, to the others).
Generally, the articulation has less tension than normal, above all in a preintoneme, which also has a higher speed than average, contrary to what happens in an
intoneme, which gives a lengthening of the initial contoid of the tonic syllable (
stressed in an intoneme), ('0:), and also for V there is often a doubling, in an intoneme, which adds to the normal length: (, ;, :) = (, ;, :). e tonograms
show the characteristics of intonation; the utterance is quite emphatic, with paraphonic expansions.
On the whole, the V are a little further back than the neutral (especially the
front and back ones, as the approximant /w/ (w), as well, instead of the neutral
()), and the traditional pronunciation /E:/ can be maintained; furthermore, for
/a/, we have ('A, 'A0, a', ) (but often we have (a) even in ('a, a, a, wa)); /,
^/ are (); /K/ () (and (), for (, , X)), often /t, d/ = (4, 7) and /l/ = (). In unstressed syllables, for /p, t, k f, s, S/ we have (p, , k , s, S); /N, j/ (N, ,) are maintained; we do not nd (+, _ , ), but (4, 7 , ).
Parisian variants of the banlieues
4.6.2. Compared to mediatic Parisian, the major vowel dierences concern: /
^, , / ( , ); /O/ (,, ); /EK, K, OK/ with an intermediate timbre, even
in stressed syllables, (, #, ); /a/ (), in all contexts, and /A/ (, ), including the
changes /wa/ (, ), often also /aK/ (:) >ar[C], very often /aj/ (:,), but
('s) -ation e lengthening of a V in a pretonic syllable is less marked or less
systematic: (2') (or even (2')), instead of (2;').
/i/ (i[:], I)
/y/ (y[:], Y)
/e/ (e), // ()
/E/ (E, , [:])
// (), // (@, , [:])
/a/ ([:])
/, ^/ ([:])
/u/ (%[:], T)
/o/ (P), // ([:])
// ([:], [:])
/O/ (,[:], , [:])
/A/ ([:], )
Canadian variants
4.6.3. As well as what has already been given in 4.4.4.1-5 4.5.2.5, we add
both some broader variants (, , ) and less broad variants (): /e, , o, A/ (ee',
', PP', AA'); /wA/ (w, w, w) (dialectal (w)); as also /EK0/ = /aK0/
is dialectal; /u/ (, , , P, P), /O/ (, , ). We can observe the use
of (, T, ) (considerably centralized), which is independent from the moderate systematic advancement of back phonemes (of type). We also indicate the
uvulo-postalveolarized velar rounded realization of /K/ (= (<)) and its vocaliza-
184
tion through (, +, P , , ) (as the last element at the end of the diphthongs
e triphthongs), in /iK, yK, uK EK, K, OK aK/, given in the rst vocogram (which
will be dealt with and illustrated with examples in the book in progress indicated
in the bibliography). Naturally, the vocograms will be less grouped up, and more
numerous, to properly demonstrate the various peculiarities (it was not possible
to place the variant (OP) of /O(:)0/). In rural accents, and above all for speakers
who are not so young, /S, Z/ (/, \) have dierent, peculiar and broad realizations,
as well as (S, Z): (, S, q , ', h, ', H). On an uneducated level, before a
rounded V /f/ = (). e French may have diculties in understanding Canadian
French, especially if the accent is very broad and uneducated.
/i(:)K/ (Ii<, Ii), /i/ ()
/y(:)K/ (Yy<, Yy+), /y/ (T)
/E:0/ (), /E'/+/v, z, Z, [v]K/ (e)
/E(:)K/ (R, <, )
/(:)0/ (#), /'/+/v, z, Z, [v]K/ (#)
/(:)K/ (R, #<, #)
/a/ (a), /a(:)K/ (<, )
/u(:)K/ (Uu<, UuP), /u/ ()

/O(:)0/ (0),
/O'/+/v, z, Z, [v]K/ (o)
/O(:)K/ (R, <, )
/A/ (A, A0, )
/A(:)K/ (<, )
/u/ (, , , P, P)
// (, / a3), /^/ (^})

/E/ (), /E:0/ (, a)
/a(:)0/ ()
// (, , As0, _, )
// (9)
/O/ (, , +, , )
/O(:)0/ ()
/A(:)0/ (, , )
5. German
5.0. e neutral pronunciation of (Germany) German is provided in this chapter. Afterwards ( 5.4.1), we will concisely present the accent of north-east Germany
( former East Germany, ex-), as used in professional and radio broadcastings, which diers only in a few nuances. We will also add the accents of Austria (
5.4.2), Switzerland ( 5.4.3) and South Tyrol or Alto Adige (in north-eastern Italy,
5.4.4), including some clearly non-neutral indications, which have been chosen
(because of their frequency and diusion) among those that will be given in a
specic monograph German Pronunciation* which will deal with the various regional accents of the whole German-speaking territory in greater detail.
Vowels
5.1.1. 5.1 shows the neutral articulations of modern neutral German, which
presents seven pairs of opposing short and long vowels in stressed syllables, with
the important addition of a timbric dierence too (except for /a, a:/, whose timbric dierence is now regional). As can be seen from 5.1, too, long vowels are
realized by closer vocoids, and a modern phonemic transcription must obviously
reect this characteristic: /i:, I e:, E a:, a/, /y:, Y :, /, /u:, U o:, O/.
But we have to add another inconvenient long phoneme, /E:/, which opposes
both /e:/ (because of timbre) and /E/ (because of length). Frequently, it merges with
/e:/, mainly in the North, but this pronunciation is not recommendable as neutral,
all the more so as current spelling clearly shows it. Indeed, for a modern neutral
pronunciation, the most appropriate realization is an intermediate one, (:), in fact
in 5.1 the other variants are shown as separate ones.
5.1. German monophthongs.
/i:, i/ (i:, i), /y:, y/ (y:, y)
/I/ (I), /Y/ (Y)
/e:, e/ (e:, e), /:, / (:, )
/E:/ (:, ;) {(E: e: e;)}
/E/ (E, ), // (, #)
/u:, u/ (u:, u)
/U/ (U)
/o:, o/ (o:, o)
// (, , , )
/O/ (O, )
// () { ()}
/a/ (a), /a:/ (a:)
In addition, we nd the typical unstressed phoneme, // (), which is realized

dierently when close to /K/: () (/K, K/ (, ), except for word-nal /K/ ()),
and in checked syllables with /K/ (as we will see shortly after, in the examples),
where we actually have // (); although it derives from //K//).
186
Let us now see some examples: ('mi:t) /'mi:t/ Mte ('mIt) /'mIt/ Mitte ('ve:k)
/'ve:k/ Weg ('vEk) /'vEk/ weg ('Sta:t) /'Sta:t/ Staat ('Stat) /'Stat/ Stadt ('hy:t) /'hy:t/
Hte ('hYt) /'hYt/ Htte ('h:l) /'h:l/ Hhle ('hl) /'hl/ Hlle In addition,
(': 'E:-) /'E:K/ hre ('e:) /'e:K/ Ehre and ('th:l 'thE:-) /'tE:l/ Tler ('thEl)
/'tEl/ Teller nally, ('zOn) /'zOn/ Sonne ('thUk) /g'tKUk/ getrunken ('a:d) /g'Ka:d/ gerade ('Unz) /'UnzK/ unsere ('Unz) /'UnzK/ unserer
e last example (as well as a previous one Tler] shows () // -er which is
more traditionally rendered phonemically as //K// (and even as //r//); but // is
more convenient, since its most normal and frequent realization is as an intense
faryngealized uvular approximant (). It is a fact that a vocalization is also possible, as () (and this symbol is certainly more appropriate than (), which however is found in many publications), that is represented by a broken-line white
marker ( 5.1); but it does not occur in a systematic way. As a matter of fact, more
often we nd (), or even (1, &); occasionally, even something like ((, , )) (
a back-central vocoid with traces of the contoids shown) is surely apter than a simple and pure vocoid.
5.1.2. We must add that, in unstressed syllables, /E, , O/ are realized as closer
vocoids, (, #, ) ( 5.1): (nt'SUldIgU,) /Ent'SUldIgU/ Enuldigu (k'tho:b) /Ok'to:b/ Oktober ('phk#lt) /'pKklt/ Prklt Again, in unstressed
syllables, also /i, e y, u, o/ are to be added, even in phonemic transcriptions:
(&miko'fo:n) /mikKo'fo:n/ Mikro#on (&eali'th:t -E:t) /Keali'tE:t/ Realitt (&yna'mi:t) /dyna'mi:t/ Dynamit (&kolo'gi:) /kolo'gi:/ kolog (&uni&vzi'th:t -E:t)
/univEKzi'tE:t/ Universitt (&fonolo'gi:) /fonolo'gi:/ Phonolog We must say at once
that, in current or familiar pronunciation, unstressed /i, e y, u, o/ easily become
/I, E Y, U, O/ (I, Y, # U, ) (except when word-nal): (&mIk'fo:n, &alI'th:t
{-E:t}, &Yna'mi:t, &#kl'gi:, &UnI&vzI'th:t {-E:t}, &fnl'gi:), and (nt'SUldI&gU,,
n-), but: ('Umi) /'gUmi/ Gummi ('zi:mi&le, -I-) /'zi:mile/ Simile ('pha:ty) /'pla:ty/
Platy ('qhe:bu) /'qe:bu/ Zebu ('mEksi&ko, -I-) /'mEksiko/ Mexiko
Unstressed /a, a:/ pose no problems to modern neutral pronunciation, since
they always become (a): (a'ga:v) /a'ga:v/ Agave (a'me:i&ka, -I-) /a'me:Kika/ Amerika ('a:fi&ka, -I-) /'a:fKika/ Afrika (&afi'kha:t, -I-) /afKi'ka:t/ Arikate Instead,
in certain transcriptions of traditional neutral pronunciation, we may nd /A/,
which somehow corresponds to /A:/; but there is not much coherence at all among dierent dictionaries and textbooks: (a'ga:v, A'gA:- a'me:i&ka, A'me:i&kA, -I- 'a:fi&ka, 'A:fi&kA, -I- &afi'kha:t, -A:t). Except for some loanwords, the
systematically gives /A/ ( /A/) for unchecked grapho-syllables (a strange
criterion indeed, to be frank), whereas Langenscheidt's dictionaries still using /A:/
mark /a/.
5.1.3. 5.2 shows the diphthongs of neutral German: three basic and two secondary ones (the latter are marked in grey and are used in English loanwords).
eir neutral duration, both in stressed and unstressed syllables, is always intermediate between the short (('al) /'al/ alle) and the long (('a:l) /'a:l/ Ahle) ones. e
three genuine diphthongs are (ae, ao, OY) /ae, ao, OY/ (often badly represented as
5. german
187
/ai, au, oyoi/): ('aen) /'baen/ BRn ('ael) /'ael/ Eile ('mae) /'mae/ MP
('aom) /'baom/ BQm ('aog) /'aog/ Auge ('thao) /'tao/ TQ ('lOYt) /'lOYt/ lTten ('OYl) /'OYl/ Eule ('nOY) /'nOY/ nS e two additional ones, in grey, in the
most spontaneous pronunciation type, merge into /e:, o:/ (but this depends on
words and speakers, too): ('le:di, 'lIdi) /'leIdi/ Lady ('e:, 'I) /'geI/ gay (&no:'hao,
&nU'hao) /noU'hao/ Know-how ('So:, 'SU) /'SoU/ Show
e broken-line white markers show two possible realizations of /ae, ao/, in reduced forms ( 5.3.1), (e, o), as in: (en) /aen/ Rn (os) /aos/ Qs (also (Y)
/OY/: (Y) /OY/ S>). e marker corresponding to () is also valid for reduced
forms with /a/, as a possible variant: (s) /das/ das. We have intentionally not
placed it in the monophthong vocogram ( 5.1), also to avoid confusion with the
quite frequent, but not recommendable use of () to indicate our (, ) (or,
at best, (), see 5.1 and the rst part of 5.1, and the second part of 5.2).
5.2. German diphthongs.
/eI/ (e:, I)
/ae/ (ae) {(e)}
/oU/ (o:, U)
/OY/ (OY) {(Y)}
{/a/ ()}
/ao/ (ao) {(o)}
5.1.4. In French loanwords with original nasalized vowels, we more normally

nd regular German timbres of /E, , a, O/ (according to stress force) followed by
//: ('thE) /'tE/ TRnt (pha'f) /paK'f/ Parfum (a'sab) /a'sab/ Ensemble ('bO) /bO'bO/ Bonbon Assimilation is possible, too, as in (an'samb,
m'bO)); although in more careful pronunciation real nasalized vocoids are possible, but generally followed by //: ('th, pha'f^, 'sb, 9'b).
In a more careful but less frequent kind of pronunciation, actual nasalized vocoids may occur, without any N. ey are long or half-long if stressed, and short
if unstressed: ('th:, pha'f^:, 's:b, 9'b:); in words ending in -on and -an even
a completely Germanized pronunciation is frequent with /o:n, a:n/: (al'khO, -o:n
- -:) /bal'kO, -o:n/ Balkon (e'kha, -a:n - -:) /e'kKa, -a:n/ Ekran. In addition, when nasalization is present, words with /a/ can show timbres that are closer to the French ones: ('s:b, 's:b, 'sb, 'sb), ('Sans[], -a-, --,
--, -:-, -:-) /'Sas[]/ ance (in words such as this, // can drop).
Consonants
5.2. 5.3 shows the necessary consonant articulations which are needed for a
good German pronunciation, including taxophonic variants. e examples will
even show the devoicing of voiced stops, after pauses or voiceless consonants and
of voiced constrictives only after voiceless consonants.
188
1.9-15, instead, give the orograms, grouped by manners of articulation, of all

the contoids given in the chapters of this volume (even as secondary, occasional,
or regional variants) for the 12 languages treated. is exposition makes the necessary comparisons between dierent languages more immediate.
(n) n
m (M)
F pb
47
q
c
(6) f v
_
s z
S Z
([)
(R) l
(~)
()
kg
(,) /x/
/j/
pharyngealized
uvular
laryngeal
uvular
velar
palatal
prepalatal
postalveo-palatal
protruded
alveolar
denti-alveolar
dental
labiodental
bilabial
5.3. Table of German consonants.
(,)
()
(X )
() h
/K/
()
+(, , n) /, / (, , , , , ? :, !, , , , /, , |, ) ( 1.9.1)
Nasals
5.2.1.1. ere are three phonemes for this manner of articulation, /m, n, /,
with a certain number of taxophones by assimilation ((m), (m, M, n, n, ~, , ,),
and (, ,)): ('man) /'man/ Mann ('hEmt) /'hEmt/ Hemd ('ve:m) /'ve:m/ wem
('naen) /'naen/ nRn (k'ni:) /k'ni:/ Kn ('qhe:n) /'qe:n/ zehn ('thIk) /'tKIk/
trinken ('am&bi;t) /'anbi:d/ anbten ('fYMf) /'fYnf/ fnf ('Uns) /'Uns/ uns (strictly (('Us)), but (n) is sucient), (i'thant) /di'tant/ d Tante (strictly ((i'4han4))),
('vYnS) /'vYnS/ wn(en, (khn'a:d) /kOn'Ka:d/ Konrade, ('mY~) /'mYn/
Mn>en ('an) /'an/ aeln ('lEst) /'lEst/ l ('lE) /'lE/ ler ('zI)
/'zI/ sien ('qhU) /'qU/ Zue ('dIU,) /b'dIU/ Bediu ('qhaetU,)
/'qaetU/ ZRtu ('qhaetU,?) /'qaetU/ ZRtuen (For /U, U/ we have
(U,, U,?).)
Please, note that the sequence /nj/ and /lj/ as well has two dierent realizations in stressed syllables, for tautosyllabic (after /:/) or heterosyllabic (after //)
elements: ('li:n,) /'li:nj/ Lin (Sam'pha~,) {((-a~-,))} /Sam'panj/ ampagner.
In lexical compounding, even for the negative prex un-, in slow accurate
pronunciation, people try to maintain (n), whereas currently it is assimilated
to the place of articulation of the following C
As an actual compromise, here we will show that it is possible to have articulations which maintain an apical contact but add secondary coarticulations (without
full contacts). ese coarticulations are: bilabial (), labiodental (), velar (n):
('Um&bndI, 'U-) /'UnbEndI/ unbndig ('UM&vIkI, 'U-) /'UnvIKklI/ unwirk-
5. german
189
li> ('Ug&e;gt, 'Un-) /'UngKe:gt/ ueregelt Below, we will nd that a couple

of these phones occur as intense too, (/, ).
In other cases, with less dierent coarticulations, assimilation (which is often
considered to be less recommendable, on pure graphical and morphological bases)
appears to be less noticeable, thus it is applied in a spontaneous though unconscious way.
5.2.1.2. Including two intense nasal phonemes too, /, / (instead of //m,
n//), we have ve of them, with all their various taxophones (() for //, and (,
, , , , ?) for //) ('a:t) /'a:t/ Atem ('zi:b) /'zi:b/ sben ('ne:m) /'ne:m/ nehmen ('Saf) /'SaKf/ (arfen ('hOs) /'hOs/ Hoens ('khaq) /'kaq/ Kaen ('u:d) /'du:d/ Duden (rigorously (('u:7~))), ('phanc) /'panc/ panen ('ap&En) /'apbKEn/ abbrennen ('m:+ 'mE:+-) /'mE:t/ Md>en ('fi:g) /'fli:g/ gen ('za:g) /'za:g/ sagen ('qhaetU,?) /'qaetU/ ZRtuen ('I) /'bKI/ brien ('vI?, -, -I,, -In) /'vIK/ wirren ('qha:l) /'qa:l/ zahlen
('e:) /'ge:/ gehen
Let us observe, however, that there are cases where intense nasals are less favored,
as results from such forms as ('Inddn) /'bInddn/ bindenden (with theoretical
sequences of (M), ('Indd), which are possible in faster speech, but mostly not
in intonemes).
As can be seen, it is more convenient to distinguish between /M/ and //. It is
sucient to know that, if pronunciation becomes slower or more controlled, it is
easy to go back to (), or (n): from ('zi:b, 'za:g) to ('zi:b, 'za:g) and ('zi:bn,
'za:gn), or from ('qha:l, 'e:) to ('qha:ln, 'e:n). However, generally, except in
intonemes, forms such as gehen can even become ('e:n).
5.2.1.3. In the previous section, we have shown up to six taxophones for the
two intense nasals, /, /. At rst sight, they might seem to be excessive; nevertheless, by concretely considering actual phonetic reality, it is clear that only in quite
approximate transcriptions could it be possible to renounce them. Besides, in order not to complicate things without reason, even nine could be useful, indeed:
(:, !, , , , /, , |, ). (All this without considering a denti-alveolar ((~)), for
which a simple () is sucient, as in (&aen'gu:t 'thak) ((&aen'gu:4~ '4hak))
/aen'gu:t 'tak/ Rnen guten Tank.)
However, we will show them only here, since these examples belong to actual
sentences taken from passages of connected speech, and we will put them in double square brackets that indicate an objective precision, but perhaps a less pressing
one. We are dealing with sequences where /, / and the tautosyllabic C before
them, /0M/, determine their exact place of articulation, (, , , , , ?), just
as in the examples seen above. Even the initial C of following syllables aect the
result, by coarticulation, in a more or less evident way, according to the places of
articulation directly involved. is happens every time two phones come into contact with one another; and, of course, this is done to render pronunciation more
uent and natural (certainly not to further complicate life!).
Among the taxophonically relevant actual cases in German, after experiencing
190
all possibilities, the following instances are worthwhile mentioning and exemplifying. As a matter of fact, should they not be applied, the result would be articulatory straining, indeed.
5.2.1.4. For practical reasons, we will start from /0/ (0). We can see that
when they are followed by labiodentals (/, f, v/), the coarticulation of () and
(, f, v) produces an intense labiodentalized bilabial ((:)): (('zi:b: 'he:7)) /'zi:b
'e:Kd/ sben Pferde (('zi:b: 'fE4)) /'zi:b 'fEt/ sben Vetter (('zi:b: 'va:g)) /'zi:b
'va:g/ sben Wagen
en /0/ (0) is followed by laminal C, /t, d, q, z, n, l/, we have a denti-alveobilabial ((!)): (('zi:b! '4ha:g)) /'zi:b 'ta:g/ sben Tage (('zi:b! '7e:&mak)) /'zi:b
'de:maKk/ sben DM (('zi:b! 'qhIm)) /'zi:b 'qIm/ sben Zimmer (('zi:b! 'nE4))
/'zi:b 'nEt/ sben N>te (('zi:b! 'lEn7)) /'zi:b 'lEnd/ sben Lnder
nally, for /0/ (0) + /k, g, K/, we nd an intense velarized bilabial nasal (()):
(('zi:b 'khaq)) /'zi:b 'kaq/ sben Kaen (('zi:b 'gE)) /'zi:b 'gE/ sben
Ge (('zi:b ':7 'E:-)) /'zi:b 'KE:d/ sben Rder
5.2.1.5. As far as /0/ (0) -em is concerned (which is not derived by assimilation, but is original, for //0m//), three groups deserve careful attention. In fact,
for /0/ (0) + /, f, v/, we nd a labiodental (): (mIt&i;z'he:t) /mItdi:z'e:Kt/ mit dsem Pferd (In&di;z'valt) /Indi:z'valt/ in dsem Wald
For /0/ (0) + /t, d, q, z, n, l/, once again, we have a denti-alveobilabial
((!)): ((qhu&7aen!'qhIm)) /qudaen'qIm/ zu dRnem Zimmer ((bae&7i;z!'lIt))
/baedi:z'lIt/ bR dsem Li>t For /0/ (0) + /k, g, K/, a velarbilabial (()):
((fo&maen'khO)) /fo:Kmaen'kO/ vor mRnem Ko ((mI4&aen'I)) /mItdaen'KI/ mit dRnem Ri
5.2.1.6. For /0/ (0) + /p, b, m/, for the third time we nd a denti-alveobilabial ((!)): (('u:4! 'phaes)) /'gu:t 'pKaes/ guten PrRs (('u:4! 'bal)) /'gu:t
'bal/ guten Ball (('u:4! 'mOg)) /'gu:t 'mOKg/ guten Morgen
For /0/ + /, f, v/, we have a labiodentalized alveolar ((/)): (('u:4/ 'hle:g))
/'gu:t 'le:g/ guten Peger (('u:4/ 'fa)) /'gu:t 'fa/ guten Fa (('u:4/ 'vaen))
/'gu:t 'vaen/ guten WRn For /0/ + /k, g, K/, we nd a velarized alveolar (()):
(('u:4 'khEl)) /'gu:t 'kEl/ guten Keller (('u:4 'gas4)) /'gu:t 'gast/ guten Ga
(('u:4 'o:4&aen)) /'gu:t 'Ko:tvaen/ guten RotwRn
Besides, for /0/ (0) + /p, b, m/, we again find a velarbilabial (()): ((&e;g'pho:l)) /ge:g'po:l/ gegen Polen ((&e;g'bElg,, -j)) /ge:g'bElgj/ gegen Belgn
((&e;g'mEksi&ko)) /ge:g'mEksiko/ gegen Mexiko
For /0/ (0) + /, f, v/, we nd a labiodentalized velar ((|)): ((&e;g|'han4))
/ge:g'ant/ gegen Pfand ((&e;g|'fak&ae)) /ge:g'fKakKae/ gegen FrankrR>
(('e:g|&va4)) /'ge:gvaKt/ Gegenwart
For /0/ (0) + /t, d, q, z, n, l/, we have a denti-alveolarized velar (()): (('e:g&4hael)) /'ge:gtael/ GegentRl ((&e;g'7ae)) /ge:g'dKae/ gegen drR ((&e;g'qhae))
/ge:g'qvae/ gegen zwR ((&e;g'zEks)) /ge:g'zEks/ gegen se7 ((&e;g'nOYn)) /ge:g'nOYn/ gegen nSn (('e:g&laes4U,)) /'ge:glaestU/ GegenlRu
5. german
191
5.2.1.7. In the case of /0/ (0) + /, f, v/, obviously, things do not change:
(mt'Saf 'fo:tos) /mIt'SaKf 'fo:tos/ mit (arfen Fotos equally, for /0/ (0) + /t,
d, q, z, n, l/: (mt'Saf 'qh:n 'qhE:-) /mIt'SaKf 'qE:n/ mit (arfen Zhnen
Instead, for /0/ (0) + /k, g, K/, we have a velarized labiodental (()): ((mt'Saf 'khI)) /mIt'SaKf 'klI/ mit (arfen Klien
For /0/ (0) + /p, b, m/, for the fourth time we nd a denti-alveobilabial
((!)): (('m:+!&pnz,o&na;4, -a-, --, --, -s,- 'mE:+-)) /'mE:tpenzjona:t, -a-, -sj-/
Md>enpensionat besides, + /, f, v/, we have a labiodentalized alveolar ((/)):
(('m:+/&fOYntI 'mE:+-)) /'mE:tfKOYntlI/ md>enfrSndli> then, + /k, g/, we
have a velarized alveolar (()): (('m:+&khas 'mE:+-)) /'mE:tklas/ Md>enklasse
For /0/ (0?) + /p, b, m/, we nd for the third time a velarbilabial (()):
(('qhae4U &publi'qhi:?, -, -,, -n)) /'qaetU publi'qi:K/ ZRtuen publizren
For /0/ (0?) + /, f, v/, we again have a labiodentalized velar ((|)): (('qhae4U|
f'khaof)) /'qaetU f'kaof/ ZRtuen verkQfen
Lastly, for /0/ (0?) + /t, d, q, z, n, l/, once again we nd a denti-alveolarized
velar (()): (('qhae4U 'le:z)) /'qaetU 'le:z/ ZRtuen lesen
In grammemes (and frequent lexemes), as in Rn von ins uns dann (on
Montag, in current speech, it is not at all rare that /n/ + /[0], [0]/ becomes (n,
n, ) (with progressively stronger nasalization, up to a complete dropping of /n/):
(Ins'khi:no, Is-) /Ins'ki:no/ ins Kino
Stops
5.2.2.1. ere are three diphonic pairs, /p, b t, d k, g/, which are peculiar because in word-nal position (and in postnuclear position, but with some oddities
that we will see) only the voiceless elements occur. In addition, /p, t, k/, after pauses and in stressed syllables (not preceded by /s, S/ that belong to the same lexeme),
are aspirated. Besides, /t, d/ are denti-alveolar (and rigorously the most appropriate symbols would be ((4, 7)), but it is sucient to use (t, d), unless it is thought to
be important to insist on this nuance).
Here are some examples: (pha'phi:) /pa'pi:K/ Papr ('phaq) /'plaq/ Pla
('SpE) /'SpKE/ spre>en ('halp) /'halp/ halb ('apt) /'apt/ Abt (I'phak) /I'paKk/ i> parke ('Unt) /'Unt/ und ('lu:p) /'lu:p/ Lupe ('Eb) /'Eb/ Ebbe ('It)
/'bIt/ bitte ('tho:t) /'to:t/ Tod ('the:t) /'tKe:t/ treten ('Sta:s) /'StKa:s/ Strae
('laet) /'laet/ LRd (s'thIkt) /s'tKIkt/ es trinkt ('laed) /'laed/ lRder ('ak)
/'dak/ danken ('khInt) /'kInt/ Kind ('khaes) /'kKaes/ KrRs (s'khi:no) /das'ki:no/ das Kino ({ae}k'nal) /{aen}k'nal/ (Rn) Knall ('bIg) /g'bIKg/ Gebirge
('I) /'gI/ gi
5.2.2.2. at is too often described as aspiration of stops before pauses, is actually nothing but the explosion of the phones (that we will show only here with
($)), not aspiration, which is a dierent thing: ('skalp$) /'skalp/ Skalp ('tha:t$)
/'ta:t/ Tat ('khEk$) /'kEk/ ke$ Some authors go as far as to indicate something like
192
('thaphth, 'Ythl), for ('thapt) /'tapt/ tappte ('Yt) /'KYtl/ rttle when they
simply want to hint at the fact that the transition from one stop to another, or to
any other phone, can be audible, that is exploded (('4hap$4$, 'Y4$)); but there
is certainly no need to mark such things.
By assimilation, in /pf, pv/, we have (]) + /f, v/: ('a]&fal) /'apfal/ Abfall (dierent from () //) ('a]&e;t) /'apve:Kt/ abwerten In addition, as we have seen
for /n/ ( 5.2.1), even /t, d/ assimilate to a following C: ('mIk&Ift) /'mItgIft/ Mitgi ('mIk&km) /'mItkOm/ mitkommen ('mIp&I) /'mItbKI/ mitbrien
('mIp&nS) /'mItmEnS/ Mitmen( ('o:k&khaof) /'bKo:tkaof/ Brot kQfen
('o:p&I) /'bKo:tbKI/ Brot brien Note that we have (+) in: ('m:+ 'mE:+-)
/'mE:t/ Md>en
5.2.2.3. Let us also consider: ('ap&aot) /'aplaot/ AblQt ('apa&ti;f) /'aplati:f/ Ablativ ('ap&u;f) /'apKu:f/ Abruf (at'Ep) /at'vEKp/ Adverb (&ati'a:l) /atmi'Ka:l/
Admiral ('Sto:f&aen) /'Sto:flaen/ StovlRn ('li:p&o;s) /'li:plo:s/ lblos ('e:k&o;s) /'Ke:klo:s/ reglos ('m:kI) /'m:klI/ mgli> ('ni:tI) /'ni:tlI/ ndli> ('hantU,) /'hantlU/ Handlu ('y:pI) /'y:plI/ bli> ('va:knIs) /'va:knIs/ Wagnis but: ('laebnIq)
/'laebnIq/ LRbniz ('e:dn) /'Ke:dn/ Redner ('e:gn) /'ge:gn/ Gegner ('va:gn)
/'va:gn/ Wagner ('a:dl) /'a:dl/ Adler ('a:dlI) /'a:dlI/ adlig ('a:dlIg) /'a:dlIg/
Adlige ('OdnU,) /'OKdnU/ Ordnu ('e:gl) /'Ke:gl/ Regler (zIg'na:l) /zIg'na:l/
Signal (&zIgna'thu:) /zIgna'tu:K/ Signatur ('e:bn) /'e:bn/ ebne ('y:bI) /'y:bKI/
brig
Actually, in this second group of forms, the consonantal sequences are voiced
because they belong to the same lexeme (with a possible diachronic drop of // -e-]\
reden gegen Wagen Adel Regel Orden eben ber On the contrary, in the rst
group we have combinations (which may be clear or opaque) with grammemes
( prexes or suxes), so we will consider only three cases: Ab-[lQt] [b]-li>
from ben and [Wag]-nis
Words of lofty or foreign origin generally maintain a voiced stop before sonants:
/bl/ Double moblren Problem publik Republik sublim Tablett Variable /bK/
Algebra Fabrik Kabrio Kobra Libretti Soubrette Vibration Zebra however
/bK, pK/ Abrasion Abrasit and /pK, bK/ abrupt /dK/ Adresse Hydrant Madrider
Madrigal Melodrama Quadrant
Other cases: /gm/ Dagmar Dogma Egmont Fragment Magma Paradigma
Phlegma Pigment Pragmatik Pygme Segment Sigma Stigma /gn/ Agnoiker
Diagnose ignorren indignrt kognitiv Magnat Magnol Magnus Physiognom prgnant Stagnation and /gn, n/ inkognito Magnesit Magnesium magneti( Signal signikant /gl/ eglomisren Hrogly#e Iglu Neglig and /l/ eli( Jli lli( and // Sile /gK/ agrari( Agronom Allegro Diagramm Emigrant Migrne Regress Saria Telegraf
But we have: /tm/ Admiral Kadmium In addition: /p, b/ Abort [= Abtritt] aboral abortren and /kn, gn/ ErRgnis in spite of /kn/ Befugnis Wagnis ZSgnis
5.2.2.4. From the examples provided, we see that German has a glottal stop
too (or glottal catch, but laryngeal stop is better) (): (&yb'al, &Y-) /y[:]b'al/ -
5. german
193
berall ('Ok) /g'OKk/ Georg (more frequently, ('e:k) /'ge:OKk/), ('ael) /b'ael/ beRlen (the'a:t) /te'at/ eater (f'aen) /f'aen/ VerRn ('vael []I'aos&abaet) /'vael I'aosaKbaet/ wRl i> QsarbRte ('Is 'aoX &aen[]'a) /'Is 'aox
aen'a/ iss Q> Rnen Ael
is phone precedes every V which is syllable-initial (thus with no C before that,
always in the same syllable), either in stressed or unstressed position, either in lexemes or grammemes, even in sentences. It is mostly noticeable after pauses and in
stressed syllables; if pronunciation becomes more relaxed, () may be missing but
only in completely unstressed syllables (and we indicate it here as ([])), provided
composition is not evident.
As can be seen, it also occurs within lexemes, in the context /'/ ('), but not
if the stress is weak: ('af) /'af/ Ae ('ap&a;t) /'apa:Kt/ Abart (f'aXt) /f'axt/
vera>ten ('aXt) /b'axt/ bea>ten (ds&avu'i:?, -, -n, -,) /dEsavu'i:K/
desavouren (e'EMfazIs) /de'EmfazIs/ Deem#asis (ant'a:qidUm) /ant'a:qidUm/
Antacidum (&ant[]a'qhi:t) /anta'qi:t/ Antacid ('aos≤z) /'aosle:z/ Qserlesen
('mUnt&a;t) /'mUnta:Kt/ Mundarten us we have: (thea'tha:lIS) /tea'tKa:lIS/
theatrali( but (&thea-) only in a kind of theatrical pronunciation.
In addition, we nd () even if the V is deep or assimilated, as in () //, for
//K//, provided it is an initial syllabic nucleus: ({qhu}'fa:?, -, -n, -,) /{qu}'fa:K/ [zu] erfahren
But we have: (ae'nand) /ae'nand/ Rnander with non-morphemic syllabication, which is acknowledged by the new orthography as well.
Further useful examples: (&'a&baet) /K'aKbaet/ erarbRten ('aof&Ste;U,)
/'aofKSte:U/ Auferehu ('vi:d&o;b,, -n) /'vi:dKo:bn/ wdererobern
('lat&aes) /'glataes/ GlattRs (la'ne:z&aes) /la'ne:zaes/ LaneseRs ('a&a;t)
/'gaa:Kt/ Gaart ('qhOl&h;U,) /'qOlh:U/ Zollerhhu (khlo'a:k) /klo'a:k/
Kloake (&ide'a:l) /ide'a:l/ ideal
5.2.2.5. e existence of pairs such as (f'aez) /f'aez/ verRsen and (f'aez)
/f'Kaez/ verrRsen could lead one to think that () is a phoneme, //; but it is not
so, even because it can be omitted, as generally happens in Switzerland, South Tyrol/Alto Adige, Austria, and Bavaria.
Even in neutral familiar speech, it is maintained better in fully stressed syllables:
(i[]'gEnqU,) /di'gEnqU/ d Ergnzu ('StUm []Un'da, []n-, -'-) /'StUKm
Unt'dKa/ Sturm und Dra (&I[]'a&baet) /I'aKbaet/ i> erarbRte However,
normally () never occurs before postverbal pronouns: ([I]'vaess) /I'vaess/ i>
wR es ('le:zI, --) /'le:z[]I/ les' i> On the other hand, in nal half-stressed syllables, after V it is certainly possible to have: ('mIa&l, -&e;l).
e banal mistake of not considering () a sound, simply because it does not
correspond to any grapheme, is to be avoided completely! For good phonetics,
clear ideas are needed, and above all it is necessary to use one's ears, not one's
eyes erefore, it is wrong to talk about a vocalic onset for (), in opposition
to an aspirated onset for (h), as for instance, in: (f'alt) /f'alt/ veralten (f'halt) /f'halt/ verhalten
Phonetically, they are not dierent from (f'fal) /f'fal/ verfallen\ they are all
194
(0), although the phonemic transcriptions do not use //, but just for the sake
of simplicity. Actually, for teaching purposes, it could be more useful to write it,
especially if no real phonetic transcription is added to the phonemic one. However, if only one type of transcription is used, it is decidedly more recommendable
to also write () or //, especially for important cases.
A possible variant of ({:/0}, ) is (, :, , 0): (&I'a&baet, &I'-),
(&I0'a&baet, &I'a-) /IK'aKbaet/ i> erarbRte (I'ant&t), (I'act-) /I'antvOKt/ i> antworte (&ide'a:l) /ide'a:l/ ideal erefore, we have () which fuses with
the voiced segments of a syllable, and realizes them as creaky. For native speakers,
this phonation type sounds similar to the sequences we have seen.
(In Danish, something similar occurs, although in the opposite direction: @d
{ 17.36 of NPT/HPh} entails a distinctive use of creaky voice for voiced phonemes, (, ); however, certain speakers change these phones into (, 0) sequences, with () in second place.)
5.2.2.6. It is interesting to note that () can also occur mainly as a substitute of
/t/ (but of /d, p, k/ as well), when preceded followed by /, M/. Even variants
with creaky intense nasal segments occur ((, =, )), which represent a frequent
compromise, as we will see. e cue and most examples are taken from Kohler
(1994), with integrations and changes, both for the transcriptions and contexts of
occurrence: ('aeMf&Stand, -nn=, -n) /'aenfKStand/ Rnveranden ('fo:&hand, -nn=, -n) /'fo:Rhand/ vorhanden ('StUnd, -nn=, -n) /'StUnd/ Stunden
('End, -nn=, -n) /'End/ Enden ('Ent, -n, -nn=) /'Ent/ Enten ('khnt,
-n, -nn=) /'knt/ knnten ('ak, -, -~) /'dak/ danken
Further examples: ('lOYp&pUkt, -t, -, -~) /'lOYtpUkt/ LS>tpunkten ('qhEntn, -nn) /'qEntn/ Zentner ('khlEmpn, -mn) /'klEmpn/ Klempner ('thaozp 'e;t, - '-, - 'm-) /'taoztme:t/ tQsend Meter ('amt, -mpt, -m, -m) /b'amt/ Beamten (not for ('amt, -mpt) /b'amt/ Beamte]
More: (&Inaen'nOYs 'amt 'aeM&fy;?, -ampt, -amp, -, -,, -n) /Inaen'nOYs
'amt 'aenfy:K/ in Rn mSes Amt Rnfhren ('phUkt, -t, -, -~) /'pUkt/
Punkten ('lOYtnant, -n-) /'lOYtnant/ LStnant (mIp'Indsts, mI-) /mIt'mIndsts/ mit mindeens ('hInt, -n, -nn=) /'hInt/ hinten (&mItn, -n) /&mItn/
mit 'nem ('aX&qhe;nt, -, -n=, -&qen-, -qn-, -qn-, -q-) /'axqe:nt/ a>ehnten
5.2.2.7. In addition: ('khOmt aeM'fOYnt&It, -mp&It, -m&-) /'kOmt daen'fKOYntmIt/ kommt dRn FrSnd mit? (e'khOmt 'nIt, -m, '-, '-) /e:K'kOmt
'nIt/ er kommt ni>t (&zIntaen'Su: 'hi:, -mp-, -m-) /zIntmaen'Su: 'hi:K/
sind mRne S>uhe hr? (&vIzt'nu: 'qhae, -'n-) /vIKzInt'nu:K 'qvae/ wir sind
nur zwR
erefore, we add the /t/ context, not preceded by /n/, but by another voiced
(or even voiceless) phoneme: ('hat, -) /'hat/ hatten ('It, -) /'bIt/ bitten
('i:t, -) /'bi:t/ bten ('vat, -) /'vaKt/ warten ('vOlt, -l) /'vOlt/ wollten ('vUst, -) /'vUst/ wusen ('lYft) /'lYft, -/ len
en (=) combine, they can become (): (Unt&IlUsta'qho:n, Un&I-)
/UntIlUstKa'qjo:n/ und Illurationen ('fae&tha;k 'a:bt, -a; 'a:-) /'fKaeta:k 'a:bt/
5. german
195
FrRtag abend (mIt&Int'sIti, mI&In-) /mItInt'sIti/ mit Intercity (mIt&i,'Elt,,

mI&-) /mIt'i:K 'Eltn/ mit ihren Eltern
Stopstrictives
5.2.3.1. German has three voiceless stopstrictives, (, q, c) /, q, c/. From a
descriptive and teaching point of view, it is not a good thing to exclude them from
the phonemic inventory and consider them as phonemic sequences such as /pf,
ts, tS/; also because these sequences do exist in the language (as we will see)! More
often, /q/ is articulated with lowered tip of the tongue, whereas /c/ is with raised
tip (but it is not necessary to use a special symbol (c)). e three of them are all
aspirated after pauses or in stressed syllables, except after a tautomorphemic /s/:
('hEf) /'Ef/ Pfeer ('hanq) /'lanq/ Panze ('SIM) /'SIm/ (imen ('khO) /'kO/ Ko ('qhaet) /'qaet/ ZRt (na'qho:n) /na'qjo:n/ Nation
('nIq) /'nIq/ ni>_ ('sqe:n) /'sqe:n/ Szene ('chE) /'cE/ T(e>e ('chElo 'SE-)
/'cElo 'SE-/ Cello ('OYc) /'dOYc/ dS ('phac) /'pac/ Pae
Let us now see some cases of /pf, tS/ (]f, tS): ('a]f&la) /'apfla/ abverlaen
(nt'SpIn, n-) /Ent'SpIn/ en_pinnen (nt'SUs, n-) /Ent'SlUs/ Enluss
However, in fast speech they can become (, c): ('a&la, nc'pIn, n-,
nc'Us, n-) (generally without aspirating the following stopstrictives or stops,
as also in: (nt'SUldIgU,, -n'cU-, n-) /Ent'SUldIgU/ Enuldigu 5.3.2.5 as
well).
Instead, German has no /G/, but /tZ/, which occurs in loanwords: (t'i:ns)
/t'Zi:ns/ Jeans (&khilimant'a:o) /kilimant'Za:Ko/ Kilimanaro ([Im]t'U)
/[Im]t'ZU/ [im] D(uel especially in current pronunciation, (t) /tZ/ can alternate with /tS, c/.
Constrictives
5.2.4.1. ere are ve diphonic pairs: (f, v s, z S, Z , , , X). Generally, /s, z/
are articulated with lowered tip of the tongue, whereas /S, Z/ are with raised tip (but
it is not necessary to use special symbols as (x, )). Strictly speaking, () could be
considered not to be a phoneme (by simply introducing a juncture phoneme,
as we will see shortly). In addition, (,) is actually a semi-constrictive phone (even
if it were indicated with (J)). On the other hand, functionally, () belongs to the
particular category of sonants, and to the vibrant group; as a matter of fact, one
possible realization is exactly (K) /K/.
us it is important to clearly know the relations among the various constrictive elements. erefore, it is not dicult in this case, thanks to spelling as well
to see that (, X) are in (almost) complementary distribution. On the other hand,
even (,) has close relationships (mostly in northern speech) with () and also (X,
k) (even with (g, ) as well, which in the North realize /g/ together with (,), that
belongs to the phoneme /j/ too).
196
But let us see some examples: ('fYMf) /'fYnf/ fnf ('vas) /'vas/ was ('aes) /'Kaes/ rRen ('aez) /'Kaez/ rRsen ('zo:n) /'zo:n/ Sohn ('S:n) /'S:n/ (n ('Staen)
/'Staen/ StRn ('thIS) /'tIS/ ( (Ze'ni:) /Ze'ni:/ Gen (e'tha:Z) /e'ta:Z/ Etage ('u:X)
/'bu:x/ Bu> ('y:) /'by:/ B>er ('aX) /'bax/ Ba> ('E) /'bE/ B>e ('u:I)
/'Ku:I/ ruhig ('aoIs) /'KaoKIs/ RQris
5.2.4.2. ere are three more approximant taxophones for slit constrictives,
which occur before /s, S/, and we will notate them with their committed variants
as well (which are more demanding), because they are not really necessary, although very frequent and typifying. It would not be useless to acquire them to
obtain a smoother pronunciation: (&, , ) /f, , x/: ('ao&&Sib, 'aof&Sib)
/'aofSi:b/ Qf(ben ('h:st, 'h:st, -t) /'h:st/ h> ('aoX&Spk, 'ao&Spk)
/'KaoxSpEk/ RQ7pe$
In preintonemes ( not in intonemes), forms such as Qf i> mi> di> si>
-ig -i> -li> -zig S> Q> do> no> in current speech, often have (&, , ) /f,
, x/, irrespective of the context just seen.
Quite rightly, German pronouncing dictionaries clearly indicate both () //
and (X) /x/ it is necessary to provide precise indications, not mysterious algebraic
formulae. In fact, unless through an appropriate morphonological knowledge,
there would be no way to tell the dierence between ('thao) /'tao/ TQ>en
[->en] small rope and ('thaoX?) /'taox/ tQ>en to plunge, or ('khu:) /'ku:/
Kuh>en [->en] calf and ('khu:X?) /'ku:x/ Ku>en cake.
5.2.4.3. e sux -ig is (I) /I/, even when followed by C (except for those belonging to suxes which contain ()): ('kh:nI) /'k:nI/ Knig ('aenIst, -st,
-t) /'aenIst/ Rnig (f'aenIt) /f'aenIt/ verRnigt ('qhanqIst, -st, -t)
/'qvanqIst/ zwanzige but: ('kh:nIkI) /'k:nIklI/ knigli> ('kh:nIk&ae)
/'k:nIkKae/ KnigrR> ('kh:nIgIn) /'k:nIgIn/ Knigin
However, it is possible to nd ('kh:nI&ae), and even fairly systematically
(Ik) /Ik/: ('kh:nIk, 'aenIkst, f'aenIkt, 'qhanqIkst), which is the only pronunciation really accepted in southern Germany, Austria, and Switzerland, in initial
position too: ('e:mIS, 'k[h]-) >emi( ('i:na, 'k[h]-) ina
e graphic sequence 7 has two dierent phonic structures: within lexemes
we nd (ks) /ks/: ('vEksn) /'vEksn/ we7eln ('vaeks) /'vaeks/ WR7el ('Oks)
/'Oks/ O7e ('zEks) /'zEks/ se7 ('fUks) /'fUks/ Fu7 ('daks) /d'daks/ der Da7
badger, while we have: (Xs) and (s, s, ) /xs, s/, when s is a grammeme: (s'aXs) /ds'daxs/ des Da7 of the roof, (s'phEs, -s, -) /ds'pEs/ des Pe7
(black) pitch.
/s, z/ are mostly articulated with lowered tip of the tongue; but, for /s/, we have
(): ('I) /'bIs/ bis(en and, except in slow speech, for /sS, sZ/, we have (SS,
S): ('aeS&Sak) /'aesSKak/ Eis(rank ([I]US'SnEl 'vEk) /ImUs'SnEl 'vEk/ i> muss
(nell weg (&aSe'ni:) /dasZe'ni:/ das Gen. /S, Z/ are with raised tip (((x, ))).
In addition, in the /kv, qv, Sv/ sequences, /v/ is more often realized as bilabial,
(6): ('khac) /'kvac/ Qua ('qhae) /'qvae/ zwR ('SEst) /'SvEst/ S>weer In
neutral German, /j/ is semi-constrictive, in all contexts: (',a:) /'ja:K/ Jahr (&ma,o-
5. german
197
'n:z -E:-) /majo'nE:z/ Majonse/MayonnPse (In&tona'qho:n) /Intona'qjo:n/ Intonation

5.2.4.4. For /K/, it is necessary to add a few observations, both on our notation
and phonemic interpretation. e generic phonemic symbol seems to be useful
in indicating its normal place of articulation, which is uvular and the most recommendable, for teaching purposes. As far as its manner (of articulation) is concerned, the voiced constrictive one, (), is normal (and the most recommendable
as well).
For emphasis, or for individual peculiarities, it is certainly possible to have a
(voiced) uvular trill: (K). On the other hand, among native speakers it is not rare
to nd (voiced) alveolar articulations: a tap, (R), or a trill, (r) (mostly in the South,
and Austria and Switzerland). However, before stressed or unstressed V we will
transcribe only () /K/: ('aoIs) /'KaoKIs/ RQris ('I) /'bKI/ brien.
After tautosyllabic V instead, we will always transcribe () (voiced pharyngealized uvular approximant) for /K/, which is intense after C or diphthongs, () //:
('vi:) /'vi:K/ wir ('vI) /'vIK/ wirr ('mae) /'mae/ MRr (like MRer Meyer] ('Es)
/'bEs/ besser (h'fo:) /h'fo:K/ hervor.
Many phonemic traditions have considered //K//, mostly in parallel with //n,
m, l//, which we prefer to transcribe and interpret as (, , ) /, , /, although
we are perfectly aware that, in slower or emphatic speech, (n, m, l) are possible,
and ( K ) as well even (R), or rather (R, R) (in Switzerland) and (, ,
) (in Austria).
Contrary to what we do and Kohler (1977, 1995) and others as well several
authors distinguish between () and (:) ( /r/ and /:/, above all in teaching textbooks), but we do not think it is the best proposal, also because it is more
complex and far from being the most widespread pronunciation ( 5.1.1 too).
For the prexes er- her- ver- zer- we will transcribe ({0}) /{0}/, since it is the
normal and most recommendable pronunciation, although in pronouncing dictionaries (except the ) we still nd /{0}Er, {0}E/. A slow, careful, or emphatic, pronunciation with ({0}) /{0}EK/ can be safely inferred from our notation as well. Diaphonemically, though, we might use /{0}/.
Approximants
5.2.5. For this manner of articulation, the most important German phoneme
is certainly (h) /h/, which often has the voiced variant, among voiced phones (especially in non-primary-stressed syllables): ('haos) /'haos/ HQs ('haem) /g'haem/
gehRm ('da:&ha;b/vI, -H-) /'da:ha:bvi:K/ da haben wir ('Un&hael, -H-) /'Unhael/
UnhRl ('vIlhlm, -H-) /'vIlhElm/ Wilhelm Notice, however, cases like: ('e:) /'ge:/
gehen ('u:I) /'Ku:I/ ruhig ('za:) /'za:/ sah
Graphically, h is a device often used to indicate that the preceding V is long,
mostly before m n l r (the so-called Dehnungs-h): ('ne:m) /'ne:m/ nehmen
In addition to the variant of /v/, (6), occurring in the sequences /kv, qv, Sv/ (
198
5.2.4.3), we will briey say something about other possible variants for /0j, y,
u', o'a/: ('fe:,, -j) /'fe:Kj/ Fern (na'qho:n, -'qho:n, -qi'o:n) /na'qjo:n/ Nation (qhya'no:z, qha-) /qya'no:z/ Zyanose (&manu'El, -u'El, -'njEl, -'nwEl) /manu'El/ manuell (thoa'lEt, thja-, thwa-) /toa'lEt/ Toilette
Laterals
5.2.6. In German neutral pronunciation we nd only (l) /l/ (with the corresponding intense one, () //) and the variant (), before (, ,) (the last one is heterosyllabic only): ('lant) /'lant/ Land ('fOl) /'fOl/ voll (fa'mi:l,) /fa'mi:lj/ Famil
(va'nI,) {((-nI-,)) and (-Il)} /va'nIl[j]/ Vanille ('mI) /'mIl/ Mil> ('hElt) /'hElt/
Held ('e:z) /'e:z/ Esel ('u:dn) /'du:dn/ dudeln
Structures
5.3.0. In this part, we will make the necessary observations for pronouncing
words in actual sentences. us, we will deal with assimilation, reduction, stress,
and intonation.
Reduced forms
5.3.1.1. In spontaneous speech, grammemes, in addition to gradually reducing
their stress, may even attenuate vowel timbres and lose some consonants or syllables. Grammemes include: articles, personal pronouns, possessive adjectives, particles, axes, prepositions, short conjunctions and adverbs and some other similar forms, such as geworden spazren and greetings.
We will give some examples, mostly taking them from Kohler (1977, 1995,
6.3.2), by modifying and completing both their inventory and variants, including
the transcriptions. In the original publications, these are limited to the segments
in question, whereas we will adapt them to our own style. e examples come in
useful to show mechanisms and possibilities, rather than absolute necessities.
(&mI[t]'hUnt) mit dem Hund
('gu:t 'gast) den guten Ga
(&In'gat) in dem Garten
(&Ist#'vIkI) i es wirkli>?
(&hass'ze:) ha du's gesehen?
(&hatdI'qhaetU g'Saot) er hat in d ZRtu ge(Qt
(&zld'fa:t&u;f) er soll den Vater rufen
('man []ndI'fao) der Mann und d FrQ
([I]'as 'e:d) i> lass ihn reden
([I]&[h]ab g'ze:, amg-) i> hab' ihn gesehen
5. german
199
([I]'phakm&ma;l) i> pa$ ihn mir mal

('las ',U,? &dX'e:d) lass den Juen do> reden
('i:p 'khIndn&dX vsqU'Es) gib den Kindern do> was zu essen
('phak 'khIndn&dX vsqU'Es&aen) pa$ den Kindern do> was zu essen Rn
([I]&kan nI[t]'se:, -'-) i> kann ihn ni>t sehen
(Is'albn, -,) er i albern ("Is 'albn, -,) er albern
(Is&nIIn'la:g) er i ni>t in der Lage (Is"nI[t ]In'la:g) er i n>t in der
Lage
(vs&[h]as[t]g'za:kt) was ha du gesagt?
(&h[t]'le: g'SpOX?) hat der Lehrer gespro>en?
(zI'vo:nt &In[d]'Stat) s wohnt in der Stadt
(&vIz&In[d]I'Stat&fa;?, -, -n, -,) wir sind in d Stadt gefahren
(&vIz&aof[]'Sta:s) wir sind Qf der Strae
('mo:ni&kas 'khOm) Monika i gekommen
('phe:ts 'khOm) Peter i gekommen
(&s'khOm) er i gekommen
('khUtIs 'khOm, -t#) Kurt i gekommen
('haenIIs 'khOm, -#) HRnri> i gekommen
(zo;&hatg'SpOX?) so hat er gespro>en (zo;&hatg'SpOX?) so hatte er gespro>en
(zo;&ha;ptg'za:kt) so habt ihr gesagt (zo;&hatIg'za:kt) so hat er ihr gesagt
(&haptM&vas'h:t) habt ihr denn was gehrt?
(&has&Id&vas'ge:b) ha du ihr denn was gegeben?
(&has&dg'ze:) ha du ihn denn gesehen?
(&hasdM&vas'ge:b) ha du ihm denn was gegeben?
('ve: ht&dg'ze:) wer hat ihn denn gesehen?
(&vas'Olz&dn, &vas'Olz) was wllen S denn? (&vas'Ol 'zi:&dn) was wollen
Se denn?
(&vas'Olzdn&da;, &vas'Olz&da;) was wllen S denn da? (vs&l'zi:d&da;,
vs&l'zi:n&da;) was wollen Se denn da?
(vs'aX 'phe:t) was ma>t denn Peter? (vs'aX[p] 'phe:t&dn) was ma>t
Peter denn?
(vs'Oltm d'ge:b) was wollt ihr ihm denn geben?
(vs'sOlII d'ge:b, vs'OlI 'ge:b, -s'-, -'s-) was soll i> ihr denn geben?
(&has'maXt) er hat das gema>t
(zI&has'maXt) s hat das gema>t
(&hat&as'maXt) hat er das gema>t?
(&hats&as'maXt) hat s das gema>t?
('le: &hats'hOYt 'nIq 'aof≥b) der Lehrer hat uns hSte ni>_
Qfgegeben
('thIS 'StElvI IndI'Ek2| 'Sak dI'and3 3) den ( ellen wir in d
E$e, den S>rank in d andere
('thIS dI'Ek&Ste;t) der (, der in d E$e eht
('thIS MvIdI'Ek g'StElt&ha) den (, den wir in d E$e geellt haben
200
(vs'aX?&didn&da;) was ma>en d denn da?

(s'Use;&zaen, &asUs'e:&zaen) das muss der sRn
(I&hapt'So: g'tha:n) ihr habt (on getan
('ha;pt &das'maXt) habt ihr das gema>t?
([I]&apIds'ge:b, [I]&ap-) i> hab' ihr das gegeben
([I]&abI'fo:g&Stlt) i> hab' ihn ihr vorgeellt
(zI&hatI'khaet&vaS) s hat ihr KlRd gewa(en
('h:?vIns 'Og 'fy:, 'h:v-, -s) hren wir uns morgen frh
('vOlzi [z]I'thEf) wollen S si> treen?
(vi'ge:ts, -q) w geht's? [-t es?]
(&[]IstInd'khEl&ae~g&dU,?, &s[]I-) er i in den Keller Redruen
(&[]IstIn'khEl&ae~g&dU,?, &s[]I-, --) er i in Rnen Keller Redruen
(&ha&,a;g'za:kt2 s'khOmt) er hat dir ja gesagt, dass er kommt
(s&hatg'thOYmt) das hat er getrTmt
(&a;h&apa'a:t&Staekt) da hat der Apparat gerRkt
(a&hapt&vas'zOYmt) da habt ihr was versTmt
('thy:n 'tho:) Tr und Tor
('le:b 'tho:t, -) Leben und Tod
('vaen g'za) WRn und Gesa
(s'ha:p 'gu:t) das Hab und Gut
('E 'bIn) el und Birnen
(aen'thISn []aem'bEt, -Un) Rn ( und Rn Bett
('fYMfn 'qhanqI, -Mv, -m) fnfundzwanzig
('nOYnn 'qhanqI, -n) nSnundzwanzig
('zi:bm 'qhanqI, -m) sbenundzwanzig
('qhae HUn'qhanqI, n-) zwRhunderwanzig
(&vns5e:gnt1 1 M'fa:vI2 &mI'ba:n3 3) (vns-) wenn es regnet, dann fahren
wir mit der Bahn
5.3.1.2. Here is a list of reduced forms; it is fairly complete, which could even
seem to be too analytical. However, it is of paramount importance to carefully examine all the modications given, in order to grasp the mechanism.
It is obvious that the most normal forms, in this list, are not the rst given,
but some of the succeeding ones, in suitable conditions.
aber ('a:b, &a;b, &ab, &ab, &a6)
also ('alzo, &alzo, &alzo, &azo)
an ('an, &an, an, n), (-m, -M, -~, -,
-,)
am an dem ('and, &and, &ad,
'a, 'am, &am, am, m)
ans an's an das ('ands, &ands, &ands, 'ans, &ans, ans, ns)
an'n an den ('and, &and, &and,
'an, &an, &an, a, an:, n:)

an'r an der ('and, &and, &and, 'an, &an, &an)
an die ('andi, &a-, &a-, -ani, -I, -)
auf ('aof, &aof, &aof, aof, of)
aufs -f's -f das ('aofas, 'aofs, &aofs, &aof#, &aofs, &aofs, aofs, aofs,
ofs)
aufm auf'm -f dem ('aof, 'aof,
5. german
&aof, &aof)
aufn auf'n -f den ('aof, 'aof,
&aof, &aof)
auf der -f'r ('aof, 'aof, &aof,
&aof)
auf die -f'ie ('aofi, -fi, &ao-, &ao-, -I,
-)
auf Wiedersehen! (aof'i:d&ze;,
-&ze;n, -zen, of-, 'id-, 'id-, 'i-)
aus ('aos, &aos, &aos, aos, os)
ausm -s'm -s dem ('aos, 'aos,
&aos, &aos)
ausn -s'n -s den ('aos, 'aos, &aos, &aos)
aus der -s'r ('aos, 'aos, &aos, &aos)
bR ('ae, &ae, ae, e), (b-)
bRm -i dem ('aed, &aed, 'ae,
'aem, &ae, &aem, aem, em),
(b-)
bR'n -i den ('aed, &aed, &ae,
&aen, ae, &aen, aen, en), (b-,
-m, -M, -~, -, -,)
bR'r -i der ('aed, &aed, 'ae,
&ae, ae, e, ), (b-)
bin ('In, &In, In, n, ), (b-)
bist ('Ist, &Ist, Is, s, #, 6Is, 6s,
6#), (b-)
da ('a:, &a;, &a, a, ), (d-)
da- (a-, -, -), ('a:&-, 'da:&-), (d-)
[-bR -na> -von -vor -zu )
(a'-, '-, '-), ('a:&-, 'da:&-), (d-)
[-ran -rauf -raus -rin -rum )
dann ('an, &an, an, n, n), (d-,
-m, -M, -~, -, -,)
das 's ('as, &as, as, s, s, s, #, s),
(d-), ([-) (only after V]
das (rel.) ('as, &as, as, s, s), (d-)
das (dim.) ('as, &as, as), (d-)
dass ('as, &as, as, s), (d-)
dRn ('aen, &aen, aen, en), (d-,
-m, -M, -~, -, -,)
dRne ('aen, &aen, &en, &n),
(d-)
dRnem dR'm ('aen, &aen, &aem, &en, &em, &n, &-
201
m, aem:, em), (d-)

dRnen dR'n ('aen, &aen, aen:,
aen, en), (d- -m, -M, -~, -, -,)
dRner ('aen, &aen, &en, &n,
&n, &), (d-)
dRnes ('aens, &aens, &ens, &ns,
&ns, &s), (d-)
dem 'm ('e:m, &e;m, &em, em,
m, m, , , ), (d-), ([-)
(only after V]
dem (rel.) ('e:m, &e;m, &em, em,
m, m, ), (d-)
dem (dim.) ('e:m, &e;m, &em, em),
(d-)
den 'n ('e:n, &e;n, &en, en, n,
n, , , ), (d-, -m, -M, -~, -,
-,), ([-) (only after V]
den (rel.) ('e:n, &e;n, &en, en, n,
n, ), (d-, -m, -M, -~, -, -,)
den (dim.) ('e:n, &e;n, &en, en),
(d-, -m, -M, -~, -, -,)
denen ('e:n, &e;n, &en, en:,
en, n, n), (d- -m, -M, -~, -, -,)
denn ('En, &n, n, n, , , n),
(d- m, M, ~, , ,) (the most reduced
forms are not used when preceded by
words in N whereas, in nal
position, they are used when they
occur after unstressed pronouns;
instead, in non-nal position they
occur after stressed pronouns too)
der 'r ('e:, &e;, &e, e, , ),
der (rel.) ('e:, &e;, &e, e, ),
(d-)
der (dim.) ('e:, &e;, &e, e), (d-)
deren ('e:?, &e;?, &e?, -, e,,
,, -n, ,, n), (d-, -m, -M, -n,
-~, -)
des 's ('Es, &s, s, s, #, s, #, s),
des (rel.) ('Es, &s, s, s), (d-)
des (dim.) ('Es, &s, s), (d-)
dessen ('Es, &s, &s), (d-)
desto ('Esto, &sto, &so), (d-)
di> ('I, &I, I, I), (d-), ([-) (on-
202
ly after V]
die ('i:, &i;, &i, i, I, ), (d-), ([-)
(only after V]
die (rel.) ('i:, &i;, &i, i, I), (d-)
die (dim.) ('i:, &i;, &i, i), (d-)
diese ('i:z, &i;z, &iz, &Iz), (d-)
diesem ('i:z, &i;z, &iz, &Iz),
(d-)
diesen ('i:z, &i;z, &iz, &Iz), (d)
dieser ('i:z, &i;z, &iz, &Iz), (d-)
dieses ('i:zs, &i;zs, &izs, &Izs), (d-)
dir ('i:, &i;, &i, I, ), (d-), ([-)
(only after V the most reduced
forms do not occur in nal position)
do> ('OX, &X, X, X), (d-)
du ('u:, &u;, &u, u, U, ), (d-), ([-)
(only after V the most reduced
dur> ('U, &U, U), (d-)
dur>s -h das ('Us, 'Us,
'U#, 'Us, &Us, Us, -,
-s), (d-)
dur>'n -h den ('U, &U,
'U, &U), (d-)
dur> die ('Ui, &-, -I, -), (d-)
eben ('e:b, &e;b, &eb, e,
em, em)
Rn 'n ('aen, &aen, aen, n, ), (-m,
-M, -~, -, -,)
Rne 'ne ('aen, &aen, &aen, &n, &n,
, n)
Rnem R'm 'nem ('aen, &aen,
&aen, &aem, &n, &m, aem:,
aem:, aem, em, m, nm, m, )
Rnen R'n 'nen ('aen, &aen, &aen,
aen:, aen, aen, en, n, nn, n,
), (-m, -M, -~, -, -,)
Rner 'ner ('aen, &aen, &aen, &en, &n, &n, , n)
Rnes 'nes ('aens, &aens, &aens, &ens, &ns, &ns, s, ns)
Rnmal ('aen&ma;l, &aenmal, -l, -,
-mm-, &m-, &m-)
ent- (nt-, nt-, nt-, t-) [-wi$eln
-gegen )
er ('e:, &e;, &e, e, , ) (() only
when not preceded by (, , ) /, K,
/ the most reduced forms do not
occur in nal position)
er- ('-, '-, '-) [-arbRten -brigen )
('-, '-, '-) [-klren -saufen )
es 's ('Es, &s, s, s, s, #, s) (the
most reduced forms occur in nal
position too, in many cases)
eu> ('OY, &Y, Y, Y, Y,
Y, Y)
euer ('OY, &Y, Y, Y, Y)
eure ('OY, &Y, Y, Y, Y)
eurem ('OY, &Y-, &Ym, Y-,
Y, Y-, Y-)
euren ('OY?, -, &Y-, &Y,, -n,
Y-, Y-, Y-)
eurer ('OY, &Y-, Y, Y, Y)
eures ('OYs, &Y-, &Ys, Ys, Y-,
Y-)
fr ('fy:, &fy;, &fy, fY, fY, f)
frs -r's -r das ('fy:ds, 'fy:ds, 'fy:d#, 'fy:s, &fy;s, &fys, fYs, fYs, fs)
fr den -r'n ('fy:d, 'fy:?, 'fy:,,
&fy;,, &fy,, fY,, f,, fYn), (-m, -M,
-n, -~, -)
fr die ('fy:di, &fy;-, &fy-, &fY-, -I, -)
gar ('a:, &a;, &a, a, a, ), (g-)
gegen ('e:g, -g, ≥-, &ge-, e), (g-)
gegens -n's -n das ('e:gds, 'e:gds,
'e:gs, -gs, ≥-, &ge-, es), (g-)
gegen die ('e:gdi, -n-, --, ≥-, &ge-,
-I, -), (g-)
geworden ('vOd, &vd, &v,), (g- -m, -M, -n, -~, -)
guten Abend! ('u:t 'a:bt, &u;t'-,
&ut-, &Ut-, &Ud-, Un'-, '-, '-,
'-, [-]'a:mt, [-]'amt), (g-)
guten Morgen! ('u:t 'mOg, &u;t'm-, &ut-, &Ut-, &Ud-, -'m-, Un'm-,
-m'm-, 'm-, 'm-, 'm-, 'm-, 'm-,
[-]'mO,, [-]'mO,), (g-)
guten Tag! ('u:t 'tha:k, &u;t'tha:k,
&ut-, &Ut-, &Ud-, Un'th-, 'th-, 'th-,
5. german
'th-, 'th-, -ak, -a), (g-)

habe -b' ('ha:b, &ha;b, &hab, &ha6,
hab, hb, &a6, ab, b) (forms with
/h/ = (`) do not occur in rhythm-group initial position)
haben ('ha:b, &ha;b, &hab, ha,
ham, am, m, , m) (forms with
/h/ = (`) do not occur in rhythm-group initial position; the last,
more reduced, forms occur after
wir]
habt ('ha:pt, &ha;pt, &hapt, hapt, hpt,
apt, pt) (forms with /h/ = (`) do
not occur in rhythm-group initial
position)
hast ('hast, &hast, hast, has, ast, as, st,
s) (forms with /h/ = (`) do not occur in rhythm-group initial position)
hat ('hat, &hat, hat, ht, at, t, t)
(forms with /h/ = (`) do not occur
in rhythm-group initial position;
the last, more reduced, form occurs
after er]
her- (h'-, h'-) [-an -auf -aus -Rn
-ber -unter )
(h'-, h'-) [-bR -na> -vor -zu )
Herren mRne Damen und (&maen'da:m Unt'hE?, -, men-, mn-, n'hE,, -n)
hin- (hI'n-, h'n-) [-an -auf -aus -Rn
-ber -unter )
('hIn&-) [-arbRten )
(hIn'-, hn-') [-zu ) (-M'-) [-weg )
('hIn&-) [-rR>en ) (-m&-) [-passen
) (-&-) [-kommen )
hier- (hi'-), ('hi:&-) [-an -auf -aus
-in -ber -unter )
(hi'-), ('hi:&-) [-bR -dur> -her -mit
-vor -zu )
hier ('hi:, &hi;, &hi, &hI, hi, hI)
hinter ('hInt, &hInt)
hinterm -r dem ('hIntd, 'hInt,
'hIntm, &hIntm)
hintern -r den ('hIntd, 'hInt?,
'hInt,, &hInt,), (-m, -M, -~, -)
hinters -r das ('hIntds, 'hIntds,
203
'hIntd#, 'hInts, &hInts)

hinter die ('hIntdi, -I, -)
-hundert (-&hUndt, -hUnt, -HUnt,
-HUnt, -Unt)
-hundert- (-&hUndt-, -hUnt-, -HUnt-,
-HUn-, -Un-, -n-, -n-, -n-)
i> ('I, &I, I, ) (() + verb: i>
wR; verb + (I): wR i>)
ihm ('i:m, &i;m, &im, im, Im, m, )
(the most reduced forms do not
occur in nal position)
ihn ('i:n, &i;n, &in, in, In, n, ), (-m,
-M, -~, -, -,) (the most reduced
ihnen I- ('i:n, &i;n, &in, in:, in,
in, In), (-m, -M, -~, -, -,)
ihr ('i:, &i;, &i, I, ) (() only
when not preceded by (, , ) /, K,
/; in addition, () only when enclitic, immediately after verbs; the
most reduced forms do not occur in
rhythm groups which are in intonemes, in neutral pronunciation)
ihr (poss.) ('i:, &i;, &i, I)
ihre ('i:, &i;, &i, &I)
ihrem ('i:, &i;, &i, &I,
im, Im)
ihren ('i:?, &i;?, &i?, &I?, -,
i,, i,, I,, -n), (-m, -M, -~, -)
ihrer I- ('i:, &i;, &i, &I)
ihres ('i:s, &i;s, &is, &Is)
in ('In, &In, In, n, ), (-m, -M, -~, -)
im in dem ('Ind, &Ind, &In,
&Im, 'I, 'Im, &Im, Im, m, )
ins in das ('Inds, 'Inds, 'Ins,
'Ins, 'In#, 'Ins, &Ins, Ins, ns, s)
in den in'n ('Ind, &Ind, &Ind,
'In, &In, &In, I, In:, n, ), (-m,
-M, -~, -, -,)
in der in'r ('Ind, &Ind, &Ind, 'In, &In, &In, I, )
in die ('Indi, &I-, &I-, &Inni, &Ini, -I, -,
I)
ist ('Ist, &Ist, Ist, Ist, Is, s, #, s) (only
V or intense C + (s))
ja (',a:, &,a;, ,a, ja, j)
204
jedo> (,e'dOX, ,-, ,-, j-)

jet (',Eqt, &,qt, ,q, ,q, j-)
mal ('ma:l, &ma;l, &mal, mal, ml, ma,
m)
-mal (-ma;l, -mal, -ml, -ma, -m)
mehr ('me:, &me;, &me, m)
mRn ('maen, &maen, maen, men),
(-m, -M, -~, -, -,)
mRne ('maen, &maen, &men, &mn)
mRnem mR'm ('maen, &maen,
&maem, &men, &mem, &mn,
&mm, maem:, mem)
mRnen mR'n ('maen, &maen, maen:,
maen, men), (-m, -M, -~, -, -,)
mRner ('maen, &maen, &men,
&mn, &mn, &m)
mRnes ('maens, &maens, &mens,
&mns, &mns, &ms)
mi> ('mI, &mI, mI, mI)
mir ('mi:, &mi;, &mi, mI, m) (the
rhythm groups which are in intonemes, in neutral pronunciation,
not even when they are enclitic after verbs)
mit ('mIt, &mIt, mIt, mt)
mit dem -t'm ('mIt, 'mIt, &mIt,
&mI, &mId, &mI[, &mIp,
&mI, &mIb, mI, mIm)
mit den -t'n ('mIt, 'mIt, &mIt,
&mI, &mId, &mI[, mI, mIn), (-m,
-M, -~, -, -,)
mit der -t'r ('mIt, 'mIt, &mIt,
&mI, &mId, &mI[, mI, mI)
musste ('mUst, &mUst, &mUs)
na> ('na:X, &na;X, &naX, naX, nX)
na> dem -h'm ('na:X, 'na:X, &na;X, &naX)
na> den -h'n ('na:X, 'na:X?, &na;X?,
&naX?)
na> der -h'r ('na:X, 'na:X, &na;X,
&naX)
neben ('ne:b, ≠b, &neb, &ne,
nem)
nebens -n's -n das ('ne:bds, 'ne:bds, 'ne:bd#, 'ne:bs, ≠bs, &ne-
bs, &nes, nems)

neben dem -n'm ('ne:bd, 'ne:bb, 'ne:b, 'ne:bm, ≠bm,
&nebm, &nem, ne, nem)
neben den -n'n ('ne:bd, 'ne:b,
'ne:bn, ≠bn, &nebn, &nen,
nem, ne, nem), (-m, -M, -~, -,
-,)
neben der -n'r ('ne:bd, 'ne:b,
≠b, &neb, &ne, &nem)
neben die ('ne:bdi, ≠bdI, &nebI, &neI, &nemI, -)
nRn ('naen, &naen, naen, nen), (-m,
-M, -~, -, -,)
ni>t ('nIt, &nIt, nI)
ni>_ ('nIq, &nIq, &nIs) (nix ('nIks)
/'nIks/ is a familiar lexical variant,
not a reduced form of ni>_)
nie ('ni:, &ni;, &ni, ni, nI)
nun ('nu:n, νn, &nun, nun, nUn,
nn, nU), (-m, -M, -~, -, -,)
nur ('nu:, ν, &nu, nu, nU)
oder ('o:d, &o;d, &od, &od, &d,
&d, -[)
ohne ('o:n, &o;n, &on, &on)
(on ('So:n, &So;n, &Son, Son, Sn, Sn,
S), (-m, -M, -~, -, -,)
sRd ('zaet, &zaet, zaet, zet)
sRn ('zaen, &zaen, zaen, zen), (-m, -M,
-~, -, -,)
sRne ('zaen, &zaen, &zen, &zn)
sRnem sR'm ('zaen, &zaen, &zaem, &zen, &zem, &zn, &zm,
zaem:, zaem, zem)
sRnen sR'n ('zaen, &zaen, zaen:,
zaen, zen), (-m, -M, -~, -, -,)
sRner ('zaen, &zaen, &zen, &zn, &zn, &z)
sRnes ('zaens, &zaens, &zens, &zns,
&zns, &zs)
sRt ('zaet, &zaet, zaet, zet)
si> ('zI, &zI, zI, zI, I, I) (the last
two forms are used after sie/Sie]
sie S- ('zi:, &zi;, &zi, zI, z)
sind ('zInt, &zInt, zIn, zn, z), (-m, -M,
-~, -, -,)
5. german
so ('zo:, &zo;, &zo, zo, z, z)

soll ('zOl, &zl, zl, z)
sonst ('zOnst, &znst, zns, zns)
-tgigen (-&th;gIg, -tg-, -gg-, -gg,
-g) [zwR- drR- )
ber (unstressed) (&yb-, &Y-, -6-)
berm -r'm -r dem ('y:bd, &y;bd, 'y:b, 'y:bm, &y;bm, &ybm, &Ybm, -6m)
bern -r'n -r den ('y:bd, &y;bd,
'y:b?, 'y:b,, &y;b,, &yb,, &Yb,, -6,), (-m, -M, -n, -~, -)
bers -r's -r das ('y:bds, 'y:bds,
'y:bd#, 'y:bs, &y;bs, &ybs, &Ybs, -6s)
ber die ('y:bdi, -I, -, &y;-, &y-, &Y-,
-6-)
um ('Um, &Um, Um)
ums um's um das ('Umds, 'Umds,
'Umd#, 'Ums, &Ums, Ums)
um'n um den ('Umd, 'Um, &Um, U, Um:, Um)
um die ('Umdi, &U-, U-, -I, -)
und ('Unt, &Unt, &Un, Un, n, ), (-m,
-M, -~, -, -,)
-und- (-&Unt-, -&Un-, -Unt-, -Un-, -n-,
--), (-m-, -M-, -~-, --, -,-)
uns ('Uns, &Uns, &Uns, ns, s, ns) (the
rhythm groups which are in intonemes, in neutral pronunciation)
unser ('Unz, &U-, &U-)
unsere ('Unz, &U-, &U-, -z)
unserem ('Unz, &U-, &U-, -z-, -m,
-z)
unseren ('Unz?, -, &U-, &U-, -z-,
-,, -z)
unserer ('Unz, &U-, &U-, -z, -)
unseres ('Unzs, &U-, &U-, -zs, -zs)
unter ('Unt, &Unt, &Unt)
untern -r den ('Untd, 'Unt,,
&Unt,, &Unt,), (-m, -M, -n, -~, -)
unterm -r'm -r dem ('Untd, 'Untm, &Untm, &Untm)
unters -r's -r das ('Untds, 'Untds, 'Untd#, 'Unts, &Unts, &Unts)
205
ver- (f'-, f'-) [-Rsen -arbRten )

(f'-, f'-) [-rRsen -lassen )
von ('fOn, &fn, fn, fn, f), (-m, -M,
-~, -, -,)
vom -n dem ('fOnd, 'fOn, 'fOm,
'fOm, &fm, fm, fm, f)
von'n -n den ('fOnd, 'fOn, &fn,
f, fn:,), (-m, -M, -~, -, -,)
von'r -n der ('fOnd, 'fOn, &fn)
vor ('fo:, &fo;, &fo, fo, f, f)
vorn -r den ('fo:d, 'fo:,, &fo;,,
&fo,, fo,, f,, f,), (-m, -M, -n,
-~, -) ( adv. ('fOn, &fn)}
vors -r das ('fo:ds, 'fo:ds, 'fo:d#,
'fo:s, &fo;s, &fos, fos, fs, fs)
vor'm -rm -r dem ('fo:d, 'fo:n,
'fo:m, 'fo:m, &fo;m, &fom,
fom, fm, fm)
vor'r -r der ('fo:d, 'fo:, &fo;, &fo,
fo, f)
vor die ('fo:di, &fo;-, &fo-, -I, -)
vor- (fo'-) [-an -auf -aus -in -ber
)
(fo'-) [-ab -Rnst -arbRten -Rlig
)
(fo'-) [-bR -weg )
(fo'-), ('fo:&-) [-her -hin -zu )
whrend ('v:?t, &v;-, -t, 'v:,t,
-nt, &v;-, v- 'vE:-)
was (non-interrogative) ('vas, &vas,
vas, vs)
wegen ('ve:g, &ve;g, &veg, ve)
wRl ('vael, &vael, vael, vel, vl)
wem ('ve:m, &ve;m, &vem, vem, vm)
wen ('ve:n, &ve;n, &ven, ven, vn), (-m,
-M, -~, -, -,)
wenn ('vEn, &vn, vn, vn), (-m, -M,
-~, -, -,)
wer ('ve:, &ve;, &ve, ve, v)
werden ('ve:d, &ve;d, &ved,
&ve:n, ven, vn, -,), (-m, -M, -n,
-~, -)
werde ('ve:d, &ve;-, &ve-, vet)
werdet ('ve:dt, &ve;-, &ve-)
wider ('vi:d, &vi;d, &vid, &vId)
wider- (unstressed) (&vid-, &vId-)
206
wie ('vi:, &vi;, &vi, vi, vI)

wie viel ('vi:fil, vi'fi:l, &vifIl, vI&fil, &vif)
wieder ('vi:d, &vi;d, &vid, &vId)
wieder- (unstressed) (&vid-, &vId-)
will ('vIl, &vIl, vIl, vI)
wir ('vi:, &vi;, &vi, vI, v, v, v)
(the most reduced forms do not occur in rhythm groups which are in
intonemes, in neutral pronunciation, not even when they are enclitic after verbs)
wird ('vIt, &vIt, vIt, vt, vt, vt)
wirst ('vIst, &vIst, vIst, vst, vst,
vst, -s)
wo ('vo:, &vo;, &vo, vo, v)
wo- (vo'-, v'-, v'-) [-bR -her -hin
-vor -zu )
(vo'-, v'-, v'-), ('vo:&-) [-ran -rauf
-raus -rber -runter )
worden ('vOd, &vd, vn, -,),
(-m, -M, -~, -)
wurden ('vUd, &vUn, vUn, -,),
(-m, -M, -~, -)
wurde ('vUd, &vU-, &v-)
wurdest ('vUdst, &vU-, &v-, -s)
wurdet ('vUdt, &vU-, &v-)
wrden ('vYd, &vYn, vYn, vTn,
-,), (-m, -M, -~, -)
wrde ('vYd, &vY-, vYt, vTt)

wrdest ('vYdst, &vY-, -s)
wrdet ('vYdt, &vY-)
-zehn (-qe:n, -qe;n, -qen, -qn, -qn,
-q), (-m, -M, -~, -)
-zehnte (-qe:nt, -qe;n-, -qen-, -qn-,
-qn-, -q-)
zer- (q-, q-) [-fahren -rinnen )
zu ('qhu:, &qu;, qu;, qu, qU, q)
zum zu dem ('qhu:d, &qu;d, &qud, &qU, 'qhUm, &qUm, qUm, qm,
q)
zur zu der ('qhu:d, &qu;d, &qud,
'qhu:, &qu;, qu;, qu, qU, qw,
q)
zu'n zu den ('qhu:d, 'qu;d, 'qud,
'qhu:, 'qhu:n, &qu;n, qu;n, qun,
qUn, qn, q), (-m, -M, -~, -, -,)
-zu- (-qu-, -qU-, -q-) [abholen )
zusammen (qhu'zam, qu-, qU-, q-)
zwar ('qha:, &qa;, &qa, qa,
qa, q)
-ren -r'n (after short or long
stressed V where /K/ (?, ) becomes /Kn/ (,), through /n/)
(-'[:]?, -'[:], -'[:],, -'[:]n),
(-m, -M, -~, -).
Taxophonics
5.3.2.1. ere are various other ways of simplifying words and above all sentences which make speech easier and more uent, but less clear to foreigners.
It is important, thus, to know the mechanisms implied, both to understand better and move closer to the native speakers' genuine pronunciation, and to be understood better. Following Kohler (1977, 1995, 6.3.1), we will give several cases, by appropriately integrating and completing them.
We have already seen that sequences of (MM) /MM/, in fast speech, are possible
only in preintonemes: (I'apS&nIt 'o:z) d abge(nittenen Rosen but: ('nIm
dI'apS&nItn) nimm d abge(nittenen
5.3.2.2. Often -e () // of the present-tense rst person singular is dropped, provided this does not produce any (M) ( intense nasal): ([I]'aX) /I'max[]/ i>
ma>' ('maXI) /'max[]I/ ma>' i> ([I]'habs 'ze:, --) /I'hab[] sg'ze:/ i>
5. german
207
hab' es gesehen ('ha:bI, -I) /'ha:b[]I/ hab' i> ('le:zI, -I) /'le:z[]I/ les' i> and
('fnI) /'fn[]I/ n' i>, ('a:tI) /'a:tm[]I/ atm' i> but: ([I]'fn) /I'fn/ i> ne, ([I]'a:t) /I'a:tm/ i> atme; in the past tense, -e () // may be
dropped, before an enclitic i>, provided the verb does not end in -ete (t) /t/:
('maXtI) /'maxt[]I/ ma>t' i> ('UftI) /'dUKft[]I/ dur' i> but: ('fnt&I)
/'fntI/ nete i> ('a&baett&I) /'aKbaettI/ arbRtete i> Even -e () // belonging to nouns may often be dropped while maintaining the C voiced or half-voiced: ('li:bUn 'laet, --) /'li:b[] Unt'laet/ Lb' und LRd
We have already dealt with the assimilation of place of articulation to a following C for /t, d, n, , s/ ( 5.2.1-2 5.2.4). We have also mentioned some alternations, especially for approximants and /K/ ( 5.2.4-5).
5.3.2.3. As already seen, // assimilates to a preceding C but /n/ remains unchanged when -e- // () is dropped (and a voiced syllable-nal but lexeme-internal C does not become voiceless): ('khOm) /'kOm/ kommen ('le:bt) /'le:bt/
lebend ('e:b) /'e:b/ eben (but: ('e:bn) /'e:bn/ ebne unchanged); ('fa) /'fa/
faen (',u:gt) /'ju:gt/ Jugend (f'lo:g) /f'lo:g/ verlogen (but: (f'lo:gn) /f'lo:gn/ verlogne unchanged). Even double assimilation occurs (but not in intonemes): (aof'e:b 'Sta:s) /aof'e:bn 'StKa:s/ Qf ebenen Straen (I'aeg&Sa:g 've:g) /di'aegeSla:gn/ d Re(lagenen Wege
Even after nasals (which may be intense or not), /t, d/ can assimilate, if they are
unexploded: ('hEmt, -mp) /'hEmt/ Hemd ('hEmd, -mb) /'hEmd/ Hemden (I',u:gt 'aM&fOYn, I',u:gk) /di'ju:gt 'anfOYn/ d Jugend anfSern (I'ge:gt
'an&ze:, I'ge:gk) /di'ge:gt 'anze:/ d Gegend ansehen (I'ge:gd 'an&ze:, I'ge:gg) /di'ge:gdn 'anze:/ d Gegenden ansehen ('amt, -mp) /be'amt/
Beamten ('haopt, -pp, -p) /b'haopt/ behQpten (I'le:bd 'laed, -bb) /di'le:bdn 'laed/ d Lebenden lRden
5.3.2.4. Frequently, the sequence /gn/ assimilates as well (and also /bn, dn/,
even if preceded by N with a subsequent reduction): (#'e:gnt, #'e:nt) /s'Ke:gnt/ es regnet ('agns, 'ans) /'agnEs/ Agnes (mag&nifi'qhEnq, ma&n-, -IfI-) /magnifi'qEnq/ Magnizenz (zIg'na:l, zI'na:l) /zIg'na:l/ Signal, ('va:gn, 'va:n) /'va:gn/ Wagner ('e:bn, 'e:mn) /'e:bn/ ebne ('le:bt, -mt) /'le:bt/ lebend ('e:dn,
'e:nn) /'Ke:dn/ Redner ('ve:d, -n) /'ve:Kd/ werden ('fInd, 'fIn) /'fInd/ nden ('hEmd, -mb, -m) /'hEmd/ Hemden ( above).
Even sequences of N + a voiced stop can undergo nasal assimilation: (no'vEmb,
-mm) /no'vEmb/ November ('Umb&halt, 'Umm-) /'Umbhalt/ umbehalten
(qhUm'bae&Spi;l, -m'm-) /qUm'baeSpi:l/ zum BRspl ('Unds&ak, 'Unns-) /'bUndsbak/ Bundesbank ('vand, 'vann-) /'vandK/ Wanderer ('vUnd&ba; 'vUnn-) /'vUndba:K/ wunderbar (qhU'mIndst, -Innst) /qu'mIndst/ zuminde ('Ug&vIs, 'U-) /'UngvIs/ uewiss ('ag≥b, 'a-) /'agge:b/ aegeben
('a≥b, 'a&e;-) /'age:b/ aeben ('Uga, 'Ua) /'UgaK/ Uar
5.3.2.5. A /t/ occurring between C in non-slow speech, is generally dropped:
('EsI) /'KEstlI/ reli> ('SIfI) /'SKIftlI/ (rili> ('EI) /'KEtlI/ re>tli>
208
('aXI) /b'axtlI/ bea>tli> ('EnlI) /'EntlI/ endli> ('vEl&khu;g) /'vEltku:g/

Weltkugel (but: ('halp&khu;g) /'halpku:g/ Halbkugel ('halp&qhaet) /'halpqaet/
HalbzRt] ('vItSaf Mfi'nanq) /'vIKtSaft Untfi'nanq/ Wira und nanzen
('EM&vUf) /'EntvUKf/ Entwurf ('aens 'StIt, -nS 'S-, -n 'S-) /'aenst 'StKIt/ Rn ritten (qhU'mInds 'qhae) /qu'mIndst 'qvae/ zuminde zwR (',Eq qU'haoz) /'jEqt
qu'haoz/ jet zu HQse ('EtIt, -I) /b'KEtIt/ bere>tigte In unstressed
positions, we frequently nd reduced forms ( 5.3.1.2) without /t/ for i ni>t
und
Even /q/, after C can become (s): ('glanq, -ns) /'glanq/ Glanz ('hElqt, -lst) /'hElqt/ erhlt ('sqe:n, 'sse:-, s'se:-) /'sqe:n/ Szene (qhU'mInds 'sae) /qu'mIndst
'qvae/ zuminde zwR (',Eq sU'haoz) /'jEqt qu'haoz/ jet zu HQse (&asqU'StEndIg, &assU-) /dasqU'StEndIg/ das zundige ('h:st qU'fi:d, -s sU-) /'h:st qu'fKi:d/ h> zufrden
e most frequent case occurs after //, for stops and stopstrictives: ('hEltst,
-lqt, -lst) /'hEltst/ erhlt in addition: ('khaM, -M]f, -Mf) /'kam/ Kam
('glanq, -nts, -ns) /'glanq/ Glanz ('ganq, -nts, -ns) /'ganq/ ganz (M'ha, M]'fa-, M'fa-) /Em'a/ emaen ('thEnc, -nS) /'tKEnc/ Tren> more: ('mUmps,
-ms) /'mUmps/ Mumps ('phOmpt, -mt) /'pKOmpt/ prompt (nt'SUldIg, n'S-)
/Ent'SUldIg/ enuldigen ( 5.2.3.1, too), ('sfIks, -s) /'sfIks/ S#inx ('phUkt,
-t) /'pUkt/ Punkt
5.3.2.6. In fast speech, between vowels and in unstressed syllables, /b, g/ can be
realized as (6, ): ('ha:b, -6) /'ha:b/ habe ('le:g, -) /'le:g/ lege /t/ can become
([): ('It, 'I[) /'bIt/ bitte Under the same conditions, voiceless stops and constrictives may become half-voiced or totally voiced: (s&hat'dOX 'maXt, --, -d-)
/dashat'dOx g'maxt/ das hat er do> gema>t (S'Sa:fI 'OX&nIt, --, -v-, 'S-) /das'SafI 'dOxnIt/ das (a' i> do> ni>t (s&UsI'aX?, --, -z-) /dasmUsI'max/
das muss i> ma>en (s'aXI 'OX&nIt, --, --) /das'maxI 'dOxnIt/ das ma>' i>
do> ni>t (obviously, because of ambiguity for native speakers which, however,
does not coincide with foreigners' ambiguity! this typical fast-speech neutralization is avoided).
After pauses, voiced constrictive phonemes remain unchanged; whereas, after
voiceless C they become half-voiced (but more often /z/ becomes voiceless): ('vi:n)
/'vi:n/ Wn (s'as) /das'vas/ das Wasser ('qhanqI) /'qvanqI/ zwanzig ('zi:b) /'zi:b/ sben ('a:t&sa;m, -t'-) /'Ka:tza:m/ ra_am (&ZUna'lIst) /ZUKna'lIst/ Journali (t'U) /t'ZU/ D(uel (',a:) /'ja:/ ja (s'a:) /das'ja:K/ das Jahr ('o:t)
/'Ko:t/ rot ('Sak) /'SKak/ 6rank After voiceless stops, /K/ may become (X)
(while phonotactically remaining /FK/): ('phaes, 'phXaes) /'pKaes/ PrRs ('the:t,
'thXe:t) /'tKe:t/ treten ('khaes, 'khXaes) /'kKaes/ KrRs
Voiced stops are realized as half-voiced, both after pauses and after voiceless C:
('Uk) /'bUKk/ Burg ('mEs&a;) /'mEsba:K/ messbar ('e:nU,) /'de:nU/ Dehnu
(&asu'Et) /dasdu'Et/ das Duett ('u:t) /'gu:t/ gut ('aos&e:) /'aosge:/ Qsgehen
5.3.2.7. Geminates, whether normal or with an intense element, can be simplied. us, N in nal position: ('khOm, -m:, -m) /'kOm/ kommen ('e:b, -m,
5. german
209
-m:, -m) /'e:b/ eben ('nEn, -n:, -n) /'nEn/ nennen ('hE, -:, -) /'hE/ hen
('le:g, -, -:, -) /'le:g/ legen Please, note that kommen and nennen can become
identical with komm! and nenn! however, in case of ambiguity, obviously, distinct forms are used.
Geminates derived by assimilation can be reduced when they occur between
stressed and unstressed syllables: ('li:bs&e;t, -m-, --, -m-) /'li:bsve:Kt/ lbenswert ('UndsinIs&te;,Um, -nn-, -n-, -mI-) /'bUndsminIste:KjUm/ Bundesminierium ('ands, -nn-, -n-) /'ands/ anders ('ag≥b, --, --) /'angge:b/ aegeben In addition to ('Sto:&m) /'StKo:mE/ Strohmee we can also have
('Sto:m&m, -&m-) /'StKo:mmE/ Strommee with possible reduction; however,
in case of ambiguity, no reduction occurs.
Geminate stops can be reduced to their second element, when they are very similar: ('ap&Ilt, 'a&Ilt) /'apbIlt/ Abbild ('a:t&aM, 'a:&-) /'Ka:tdam/ Raddamer ('mIt&thael, 'mI&th-) /'mIttael/ mittRlen ('mIt&qh;l, 'mI&qh-) /'mItqE:l/
mihlen ('vEk&khm, 'vE&kh-) /'vEkkOm/ wegkommen ('vEk&e:, 'vE&-) /'vEkge:/ weggehen e same holds for grooved constrictives: (s'sElb, s'-, 's-)
/das'zElb/ dasselbe (s'SIf, S'S-, 'S-) /das'SIf/ das 6i ('aes&Sak, -S&S-, -&S-) /'aesSKak/ Eis(rank Instead, the /s/ sequence, as we have already seen, has a couple
of possible variants: ('h:st, -st, -t) /'h:st/ h> whereas /s/ is only slightly
modied: ('I) /'bIs/ bis(en
5.3.2.8. From the above, a phonation-type dierence can become almost functional in three ways: ('ap&Ilt, 'a&Ilt) /'apbIlt/ Abbild in comparison with ('a:p)
/'a:p/ Apel and ('a:b) /'a:b/ Abel Let us also consider: ('mEnd&zo;n) /'mEndzo:n/
Mendelsohn ('mEnds&so;n, -s&-, -&s-) /'mEndszo:n/ Mendelssohn
If the geminates are produced by assimilation of place (of articulation), they can
be reduced (provided the syllable of the rst element is unstressed): (&Immu'ze:,
&Imu-) /Immu'ze:/ in Museen (mIk'khal, mI'kh-) /mIt'kaKl/ mit Karl (&magnifi'qhEnq, &man-, &ma-, &ma-, -IfI-) /magnifi'qEnq/ Magnizenz (qhUm'bae&Spi;l,
-m'm-, -'m-) /qUm'baeSpi:l/ zum BRspl (qhu&mInds'as&a;, -Inn-, -In-) /qu'mIndst
'dasda:/ zuminde das da
We have, though: ('am&mld) /'anmEld/ anmelden ('mIk&khm) /'mItkOm/ mitkommen ('o:k&khaof) /'bKo:tkaof/ Brot kQfen ('a:k&fa:?, -,
-&fa;,, -n) /'Ka:tgfa:K/ Rad gefahren ('am&bInd, 'am&m-) /'anbInd/ anbinden
('agns, -n-, --) /'agnEs/ Agnes ('a≥b, 'a&-, -e;m, -e;, -e;m) /'ange:b/
aeben ('Unds, -nn-) /'bUnds/ Bundes ('vand, -nn-) /'vandK/ Wanderer
5.3.2.9. In sequences of /n, l/ + voiceless constrictives, homorganic (and homo-phonic) stops can be inserted (although this characteristic denotes a less careful
pronunciation which it is better not to follow): ('amt, -mpt) /'amt/ Amt ('ams,
-mps) /'gams/ Gams ('amS, -mpS) /'KamS/ Ram( ('zEMf, -M]f) /'zEnf/ Senf ('ans,
-nts, -nq) /'gans/ Gans, ('i:nst, -ntst, -nqt) /'di:nst/ Dn ('mEnS, -ntS, -nc) /'mEnS/
Men( ('ma~, -~+, -~k) /'man/ man> ('hEt, -kt) /'hEt/ ht ('hEst, -kst)
/'hEst/ He ('hElst, -ltst, -lqt) /'hElst/ erhell is frequently occurs for /nz,
lz/ as well: ('Unz, -ndz, -nQ) /'Unz/ unser ('alzo, -dzo, -Qo) /'alzo/ also
210
5.3.2.10. Certainly, the transcriptions seen thus far are useful to conveniently
separate German spelling from its phonic structure, which might seem rather
strange to foreigners. But this is reality. Here are some examples for reection:
('ap&aez) AbrRse ('th:kI 'thE:k-) tgli> ('mUnt&at) Mundarten ('Un&aofItI, -f&I-) unQfri>tig (f'ap&e;dU,) Verabredu ('a:p, -t) beatmen
It is obvious then that German has very complex consonant sequences; here we
will give just one example: ('hEps[t]&lt) Herbbltter Please, note carefully even
the spelling sh /sh/: ('lands&haoshalt) /'landshaoshalt/ LandeshQshalt.
Stress
5.3.3.1. In long words and in sentences, there are unstressed syllables that alternate with stressed or half-stressed ones. Even rhythm groups follow the same principle; and all this is already clear from the transcriptions we have given.
German compounds have particular patterns. e most normal and widespread
one shows a primary stress on the prominent syllable of the rst lexeme and reduced stresses on the ex-prominent syllables of succeeding lexemes.
Here are some examples of two-lexeme compounds: ('SIf&fa;t) 6ifahrt ('aM&SIf) Dam(i ('a:n&ho;f) Bahnhof ('aez&ba;n) Eisenbahn ('ho:X&OYc) Ho>dS ('fy:&StYk) Frh$ ('aoto&ba;n) Autobahn (fa'mi:l,&na;m) Familnname ('mUnt-ha&mo;ni&ka) Mundharmonika ('Su:&maX) 6uma>er
It is useful to note that phonemic transcriptions, which do not mark secondary
stresses, naturally imply such a reality. erefore, pronunciations as the following
for four of the compounds just seen are typically foreign-like (in this case Italian):
('Sif:faRt, 'dam:Sif, 'ba:nof, 'Ok:dic). We cannot help ending with an observation
on the compound 6uhma>er and its derived family name 6uma>er which
is generally pronounced by Spanish-, Portuguese-, and Italian-speaking people with
peculiar stress patterns that we will illustrate here using the Italian ones: currently (Su'ma:keR), up to an ugly hybrid, which is neither German nor Romance: ('Su:maxeR).
5.3.3.2. Let us also consider: (e'mi:&fa;z) emfaser (e'mi:&vk) emwerk (':t&Insti&thu;t) GoethRnitut ('fa:&pha;n) Fahrplan ('an&Stalt) Analt
('mIt&tha;k) Mittag ('fa:t&lant) Vaterland ('fElt&aSal) Feldmar(all (and often
('fElta&Sal, &fElt'aSal) for rhythmic reasons).
With three lexemes (paying attention that the kind of stress we mark here as ()
is weaker than (&) and is not normally written, even because it can be completely
reduced, although its syllable maintains the original characteristics of length, vowel timbre, and consonant consistency, including possible aspiration): (e'mi:fa;z&vk) emfaserwerk ('mIttha;ks&qhaet) MittagszRt ('fa:tlants&li;b, -nq-) Vaterlandslbe
More: ('vIltSaes&kh) Wild(wRnsko ('SIm&anStalt) 6wimmanalt
('na:X&Ittha;k) Na>mittag ('alt&ho;XOYc) Altho>dS ('haopt&aofa;b)
5. german
211
HQptQfgabe ('StOY&aenne;m) StSerRnnehmer ('vElt&anSaoU,) Weltan(Qu ('Elt&anla;g) Geldanlage ('na:X&abaet) na>arbRten

5.3.3.3. However, rhythm often produces changes, especially in intonemes.
us, stress patterns as the following are actually more frequent, although they are
not always fully perceived (or accepted): ('SIman&Stalt, 'na:XIt&tha;k, 'altho:X&OYc, 'haoptaof&a;b, 'StOYaen≠m, 'vEltan&SaoU,, 'Eltan&la;g, 'na:Xa&baet), and (&vIlt'Sae&kh), too.
With four lexemes we can have: ('aezba;M&fa;pha;n) Eisenbahnfahrplan
('na:XIttha;ks&qhaet) Na>mittagszRt but rhythm often makes ('na;XIt&tha;ksqhaet), or even (na;X'Ittha;ks&qhaet) preferable.
With ve or six (lexemes): ('khaft&fa;&qhOYk 'haft&hItf&zIU,) KrafahrzSghapi>tversi>eru ('Spi:l&va;? 'aos&hands&zlSaft) SplwarenQenhandelsgesell(a ('ho:X&akuUm 'lIt&o;g&Slqo;f) Ho>vakuum-Li>tbogen-6melzofen ey generally attenuate the beginning, becoming: (&khaftfa;'-,
&Spi;lva;?'-, &ho;XakuUm'-).
5.3.3.4. ere are even cases where secondary stresses actually determine the
meaning of certain compounds: ('aMSIf&fa;t) Dam(ifahrt steamboat voyage, ('aM&SIffa;t) Dam(ifahrt steam navigation. In cases such as these,
of course, rhythmic uctuations are less frequent, although context must (and can)
help very much.
We also nd some compounds which do not bear primary stress on their rst
lexeme, as: (&,a;'qhe:nt) Jahrzehnt (&o;st'mo:n&ta;k) Oermontag (&Saq'aes)
(warzwR (&o;t'khOYq&Sst) RotkrSz(weer (&za;'bYk) Saarbr$en
Others have more than one primary stress: ('lao 'gy:n) blQ-grn, ('Um 'StOlq)
dumm-olz ('Se:sI 'hOlStaen) 6leswig-HolRn however, they can become
(&lao'gy:n, &Um'StOlq, &Se;sI'hOlStaen), mostly in preintonemes.
Let us observe: ('lao 'gy:n) blQ-grn blue and green ( two colors put alongside, as in a ag or sports -shirt) and ('lao&gy:n) blQgrn bluish green, water
green ( two colors blended together, as painters do).
5.3.3.5. e faster speech becomes, the more non-primary stresses can be reduced dropped, mostly in preintonemes. On the contrary, if speech becomes
slower, the stresses can be restored. In particular in intonemes, the last non-strong
stress can often be strengthened (we will mark it only here with ()): ('aotoba;n
'anStalt 'alt&ho;XOYc 'haopt&aofa;b) besides, ('aezba;M&fa;pha;n, -fa;pha;n &Spi;l&va;?'aos&hands&zlSaft, -zlSaft &ho;X&akuUm'lIt&o;g&Slqo;f, -Slqo;f).
ere are frequent cases of (homographic) compounds with (stressed or unstressed) prex grammemes, which (in their unitary non-inected forms) is the
most important element of semantic dierentiation: berseen ('y:b&zq) to
pass over, (&yb'zEq, &Y-) to translate, umfahren ('UM&fa;?, -, -a;,, -a;n)
to run down, (UM'fa:?, -, -a:,, -a:n) to drive/sail round.
Generally, compounds with unstressed monosyllabic suxes have /I, , , , ,
212
/, even without secondary stresses (except for rhythmic reasons, when several unstressed syllables are arranged in sequence, as happens for (-laen) -lRn (-U,) -u
as well): ('vtI) w>entli> (('vt&I, -I) w>entli>en].
5.3.3.6. Other unstressed monosyllabic suxes always bear a secondary stress:
[(-&ba;) -bar (-&haft) -ha (-&haet) -hRt (-&kaet) -kRt (-&la) -la (-&lI) -li (-&lo;s)
-los (-&ma;l) -mal (-&za;l) -sal (-&za;m) -sam (-&Saft) -(a (-&tu;m) -tum (-&fl) -voll
(-&vq) -wr_] for example: ('aen&haet) RRnhRt ('Ent&o;s) endlos
Let us further consider (and notice that not fully stressed grammemes have no
aspiration): (-I&kaet) -igkRt (-&haftIkaet) -haigkRt (-&lo;zIkaet) -losigkRt
ere are further rhythmic oscillations: ('am&mkU,?, 'amm&kU,?) Anmerkuen ('a&baet&In, 'abaet&-) ArbRterinnen mostly with un-\ ('Un&anStndI, 'Unan&StndI) unanndig ('UM&fozItI, 'UMfo&zItI) unvorsi>tig as
for unQfri>tig (at the end of 3.2), for which emphasis or emotion often produce: (&Unan'StndI, &Unaof'ItI, &UMfo'zItI).
Let us nally recall that German words are not all stressed on their rst syllable
(as too hastily certain textbooks describe or prescribe), mostly for prexed
forms or (even old) loanwords: (zo'fOt) sofort ('nao) genQ ('fUnd) gefunden
(mo'mEnt) Moment (ma'Si:n) Ma(ine (pho'fEso) Professor (with its troublesome (&phof'so:?, -, -,, -n) Professoren] In addition, we nd forms such as:
(a'bae) dabR (hI'naos) hinQs (vo'he:) woher which for emphasis become: ('a&bae, 'hI&naos, 'vo&he;).
5.3.3.7. Another characteristic of German stressing is nal destressing of verbal forms, mainly auxiliaries and modals (and suchlike), innitives and past participles. As a matter of fact, instead of an expectable stress pattern, which can
normally be found in various languages at the end of intonemes, we nd this
kind of destressing which corresponds to the following typology.
With compound verbal forms (thus including: future, conditional, passive, modals with dependent innitives): das 'dre 'ri>tig &sRn s 'mag 'Re>t &haben s
'soll &sehr 't>tig &sRn i> &habe 'Cola be&ellt er i '(lafen ge&gaen i> &habe es
'ni>t ge&wollt i> &bin na> 'Rom ge&fahren er wird 'wohl zu 'HQse &sRn er &i ver'haet &worden
Also: er &hat ihn 'fahren &lassen s &hatten mir den 'Brf '(rRben &helfen das 'Auto
&wird bis 'morgen repa&rrt i> muss 'jet na> 'HQse &gehen i> &wrde 'gern 'Rot&wRn &trinken i> &habe ihn 'fahren ge&lehrt &will du ins 'Kino &gehen?
Naturally, the strange eect of destressing increases when there is more than
one innitive form: 'Karin hat den 'Brf &(rRben &mssen s &htten 'ni>t &kommen &sollen i> &habe um '3 &Uhr '(lafen &gehen &mssen
5.3.3.8. Cases such as the following are interesting too: 'ma>en S d 'Re>nu &fertig 'halten s d 'Psse be&rRt where the non-conjugated form of the verb
is an indeclinated adjective; the old orthography used to give: fertigma>en berRthalten instead of: fertig ma>en berRt halten
For subordinate clauses, we have conjugated forms in the last position, innitive
5. german
213
ones with um zu: s 'kann 'ni>t &kommen, &wRl s 'morgen &sehr 'frh 'Qf&ehen
&muss (except when dependent innitives are more than two: s i &sehr 'mde, &wRl
s, 'hSte 'Morgen, 'sehr &frh hat 'Qf&ehen &mssen i> 'hoe, &dass er d 'Re>nu
&hat be'zahlen &knnen i> 'hoe, &dass s &Rne 'nSe 'Wohnu hat 'nden &knnen]
i> 'fr>te, &dass er 'Re>t &hat der 'Arzt &wollte 'wissen, &ob er 'ber ge&habt &habe s
'sagt, dass &ihre 'Kinder im 'Garten &splten i> 'glQbe, &dass s 'ni>t 'hr &i i>
'wR, dass er vor &zwR 'Jahren &ope'rrt &worden &i i> 'brQ>e 'ni>t zu &antworten
es i 'immer 'besser, 'pnktli> &anzu&kommen um 'pnktli> &anzu&kommen, &mssen
wir d 'U-&Bahn &nehmen e constructions with modals and innitives in the last
position are to be noticed too: du 'sollte zum 'Arzt &gehen
With inected separable verbs, postponed particles generally bear primary
stresses ((')); however, it is often weaker (up to a secondary degree), thus we will
indicate it here with (): s 'kommt 'morgen {an i> 'lege es {bR du 'ruf {an er 'gab
den 'Kam {Qf das 'Da> 'sprit {ber
Let us add some typical examples, since the concepts expressed by some words
can (or must) come out: 'Anna darf 'ni>t &mitkommen der 'Brf &wurde von 'mir
ge&(rben. More: &kann man 'hr 's>wimmen? (two concepts), however: 'knnen
S Kla'vr &splen? something like to piano; 'morgen &muss i> 'gen i> muss
'morgen &gen [ gen is already known).
en no complement or predicate is present, destressing certainly does not occur, as in the following sentences, where verbs are preceded only by words (and
concepts) with very low information power: 'kann i> 'mal telefo'nren? er 'fuhr
in d 'Fern, um &si> zu er'holen er 'l das ReQ'rant, 'ohne zu be'zahlen 'bitte,
'blRben S &do> 'sien Let us further consider the following examples: das i be'immt 'unab&si>tli> ge&(ehen er hat ge'sagt, dass es 'unab&si>tli> ge&(ehen &i
er hat ge'sagt, dass es 'unab&si>tli> ge&ma>t &worden &i er hat ge'sagt, dass es 'unab&si>tli> ge&ma>t &worden sRn &kann er hat ge'sagt, dass es 'unab&si>tli> ge&ma>t
&worden sRn &knnte Here, the particular syntactic structure of German subordinate clauses, with verb phrases at the end of sentences, makes it possible for the
last primary stress to be followed by a considerable number of syllables.
Intonation
5.3.4. 5.4 shows neutral German intonation. Each tonogram must be analyzed very carefully. Here we will simply provide examples for the three marked
intonemes. In 5.3.1.1, the third and fourth utterance (of the second group of examples, before the reduced-form list) show mid-pitch parentheses ( 13.24 of
NPT/HPh), which is also typical of German and French. Here is another example:
und, &wenn es 'mgli> zu 'ma>en &wre, Others may be found in the transcriptions of the text ( 5.5):
/./: (I'su:X aem'm:bg&Sft3 3) /I'zu:x aen'm:bgSEft./ I> su>e Rn Mbelge(.
/?/: ('khOmzi aos':st&ae1 1) /'kOmzi aos':stKae?/ Kommen S Qs errR>?
214
//: ('fa:?zi naX5Ys&df1 1| {&o;d'mak&bUk3 3}) /'fa:Kzi na:x'dYsdOKf {o:d'makdbUKk.}/ Fahren S na> Dsseldorf oder Magdeburg?
5.4. German preintonemes and intonemes.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 1 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 5 1 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
Other accents
5.4.0. As we have already said, we think it important to present, in addition to
neutral pronunciation, at least some accents, not only for useful comparisons, but
also to be ready to grasp the pronunciation characteristics of German outside the
former West Germany. We will start from former East Germany (ex ,
the present-day north-eastern part of reunied Germany), to Austria, Switzerland,
and South Tyrol (or Alto Adige), to understand dierent accents better and more
quickly.
ese observations could even be useful in order to acquire a more typical accent among the last three, mainly if somebody wanted (or had) to be in preferential contact with those inhabitants, more than with those of Germany. It could also be interesting to see in NPT/HPh the phonosyntheses of Pennsylvania German or Pennsylvania Dub (as it is called, 21.12) and various Germanic dialects spoken in Germany (NPT/HPh 17.27), Austria (NPT/HPh 17.28), Switzerland (NPT/HPh 17.20), Luxembourg (NPT/HPh 17.22), France (NPT/HPh
17.21), and Italy (NPT/HPh 16.2, 16.12, 16.15-6, 16.18).
North-eastern-Germany German
5.4.1.1. e rst vocogram shows the vowels of the ex- mediatic (and neutral) accent. It is not the local pronunciation of Berlin or of Saxony. As will be seen
shortly, by comparing this vocogram with that of 5.1, long stressed V are actually monotimbric diphthongs, or vocalic doublings, with a slight upward movement. Besides, /E, E:, , O/ are slightly closer (suciently so to use dierent phonetic symbols: (, , #, )).
We just give some examples to directly show the dierence, thanks to phonetic
transcriptions: ('fi;il) /'fi:l/ vl ('ze;e) /'ze:/ sehen ('Sp;t) /'SpE:t/ spt and:
('a;an) /'ba:n/ Bahn ('zo;o) /'zo:/ so ('u;ut) /'gu:t/ gut ('thy;y) /'ty:K/ Tr ('S;n)
/'S:n/ (n
5. german
215
e two realizations of // are slightly lower: ('Unz, -ndz-, -nQ-) /'UnzK/ unsere whereas, the possible vocalization of /K, / is closer: ('vas, -x, -x, -x, -x) /'vas/ Wasser
5.4.1.2. By comparing the vocogram provided, it can be seen that the second
elements of /ae, ao, OY/ are more central (and even lower for /OY/): (a, aP, +). In
addition, especially the rst element of /ao/ is slightly backer and the rst of /OY/
is slightly higher, and indeed it crosses the edge towards the upper box: ('as) /'aes/
Eis ('laP) /'blao/ blQ ('n+) /'nOY/ nS ere is little dierence for the two xenophonemes /eI, oU/; while, for the possible rst-element variants of /ae, ao, a OY/
they coincide.
We add for C that there is quite a frequent change of voiceless constrictives (and
/z/), after /n, l/, into homorganic sequences of stops + constrictives, or even into
real stopstrictives ( 5.3.2.9). nally, before pauses, /k/ can be realized as a (pre)velar stopstrictive (w, ): ('khi;ik, -w, -) /'kKi:k/ Krg ('tha;ak, -w, -) /'ta:k/
Tag ('qhu;uk, -w, -) /'qu:k/ Zug However, all these realizations are not at all recommendable.
e tonogram shows intonation, which sounds slightly less peculiar than the
western neutral one, approaching what may be called international.
/u:, u/ (u;u, u)
/U/ (U)
/o:, o/ (o;o, o)
// (, , , )
/O/ (, )
// () { (x)}
/i:, i/ (i;i, i), /y:, y/ (y;y, y)

/I/ (I), /Y/ (Y)
/e:, e/ (e;e, e), /:, / (;, )
/E/ (, ), // (#, #) /E:/ (;, ;)
/a/ (a), /a:/ (a;a)
/oU/* (o;o, U)
/OY/ (+) {(+)}
{/a/ ()}
/ao/ (aP) {(P)}
/eI/* (e;e, I)
/ae/ (a) {()}
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 5 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
216
Austrian German
5.4.2.1. is accent is the mediatic one, which prevails in radio and television
broadcastings and is half-way between the slightest and broadest accents. ese observations are drawn from our (growing) sound archives.
Long monophthongs are realized as geminate or doubled vocoids and have a
shorter duration although gemination gives an impression of greater fullness,
which almost equalizes true length: not (:), but ().
It is at once clear even that the articulations corresponding to /I, Y, U E[:], , O/
are closer than in neutral German, so much so that they are represented as (i, y, u
[], #, ) (in less broad accents we nd (I, Y, U E[E], , O), with the unstressed
variants (, #, ), slightly closer than in neutral pronunciation): ('fiS) /'fIS/ (
('fyMf) /'fYnf/ fnf ('hunt) /'hUnt/ Hund ('lf) /'Elf/ elf ('Spt) /'SpE:t/ spt ('q6#lf)
/'qvlf/ zwlf ('ft) /'Oft/ o Certain words have /e:/ (ee) for /E:/.
5.4.2.2. Both /a/ and /a:/ are back-central: ('StAt) /'Stat/ Stadt ('StAAt) /'Stat/ Staat
(in less broad accents we nd (a, aA) (and (, ) in rened, or aected, ones),
whereas in broader ones we have: (, ) and (, ) as well (which will certainly be noted at once).
Austrian traditional dialects generally present phonemic opposition between a
back a and a fronter one, both short and long. It is possible to nd this timbre distinction in language too, where the back timbre is used in traditional words,
whereas the fronter one is used in loanwords and neologisms, as happens for instance in Bank\ /'bAk/ ('Ak) bench, /'bak/ ('ak) bank, gambling table.
A very broad accent has /e:, :, o:/ (e, #, o): ('vee, 've) /'ve:/ weh (',
'#) /'b:/ B ('soo, 'so) /'zo:/ so
5.4.2.3. // is realized as (, ), with the possible vocalization of /K, / () (with
consonantal accompaniments too); thus we have: ('uns) /'UnzK/ unsere and
('vAs, -, -, -, -) /'vas/ Wasser e symbol () corresponds to the one currently used, though less rigorously, in neutral German as well, for (), which on the
contrary is appropriate for less broad Austrian accents that have () (and even
(x)); whereas the broadest accents have (a[]). It is to be noted that () // is front-central, considerably fronter than in neutral German. e most rened or aected accents may reach (, ) for /K, / ( the last vocogram but one).
Moving to diphthongs, we can see that /OY/ has just a higher rst element, while
/ae, ao/ are decidedly dierent, (, ): ('s) /'aes/ Eis ('l) /'blao/ blQ ('nY)
/'nOY/ nS Less broad accents have, (, A, O), respectively, while the broadest
ones have (E, , Y) (as in Vienna, where even (, , ) occur, and even (O,),
as a compromise, and (@, ) as well). Xenophonemes are slightly closer than in
neutral German: /eI, oU/ (ee, i oo, u).
5.4.2.4. In the second vocogram, two further typical diphthongs are evident (although marked in grey, since they only occur in local family and place names), (i,
u) /i, u/: ('iks, 'iiks) Dx ('uq, 'uuq) Rue No variant of /ae, ao, OY/ in
5. german
217
reduced forms are given, because they tend to remain as indicated; nevertheless,
we have shown /a/ (), which is possible in reduced forms.
In comparison with neutral German, there are some dierences in phonemic
length too, as in: ('Alt&tum) /'alttu:m/ Altertum ('Spiil&a) /'Spi:lba:K/ splbar
('hAq) /'ha:Kq/ Harz ('AS) /'ba:KS/ Bar( ('Sust) /'Su:st/ 6uer ('Sloos) /g'SlOs/ Ge(loss/-o
/u:, u/ (uu, u)
/U/ (u)
f
/o:, o/ (oo, o)
// (, , , )
/O/ (, )
// () { ()}
/a/ (A), /a:/ (AA)
/i:, i/ (ii, i), /y:, y/ (yy, y)

f
/I/ (i), /Y/ (y)
/e:, e/ (ee, e), /:, / (, )
/E/ (, ), // (#, #), /E:/ (, ;)
/u/* (u)
/i/* (i)
/oU/* (oo, u)
/OY/ (Y)
{/a/ ()}
/ao/ ()
/eI/* (ee, i)
/ae/ ()
/i:, i/ (ii, i), /y:, y/ (yy, y)

/I/ (I), /Y/ (Y)
/e:, e/ (ee, e), /:, / (, )
/E/ (E, ), // (, #)
/E:/ (EE, ;)
/e:/ (e, ee), /:/ (#)

/ae/ (, E, )
// (, x, a)
/ / (2 2 2 2 2 2 2)
/./ (2 ' 2 3)
/ / ( 2 2 2 2 2 2 2)
/?/ (2 1 2)
/ / ( 2 2 2 2 2 2 2)
// (2 ' 2 1)
/ / ( 2 2 2 2 2 2 2)
/,/ (2 ' 2)
/u:, u/ (uu, u)
/U/ (U)
/o:, o/ (oo, o)
// (, , , )
/O/ (O, )
// {() ) ()
/a/ (a), /a:/ (aA)
/K-/ (-)
/o:/ (o)
/OY/ (O, O,, Y, , @, )
/ao/ (A, , )
/a[:]/ ([], [], [])
218
e last two vocograms show both less broad (()) vowel realizations (up to
some aected ones, ()), and on the contrary some of the broadest realizations
(()).
5.4.2.5. Moving to C we nd that often /, 0/ become /k, k0/ (but they
are not recommendable): ('qtu, -k) /'qaetU/ ZRtu (however, even without
(k), the nasal is always velar, not uvular, in /K/ as well: (SpA'qii, -ii, -iin)).
/t, d/ are more dental than denti-alveolar: (e'tkto) /de'tEktoK/ Detektor often /k, kl, gl/ can be prevelar or retracted palatal (and /k/ may even become stopstrictive, but these variants are best avoided): ('tAAk, -, -, -, -k) /'ta:k/ Tag ('ukl, -l-, -N-) /'dUkl/ dunkler ('eel, -"l-, -l-) /'Ke:gl/ Regler
Voiceless stops and stopstrictives are not aspirated (except in less broad pronunciations), as we have just seen for /q/: ('pool) /'po:l/ Pol ('toon) /'to:n/ Ton
('kint) /'kInt/ Kind ('unt) /'Unt/ Pfund ('cAko) /'cako/ T(ako
Voiced stops are half-voiced, (, , ); but they can also be (b, d, g), between
voiced phones, again in less broad accents; however, in the broadest ones, we generally nd (p, t, k): ('uu, -6) /'bu:b/ Bube (often we have (6) /b/), ('uu) /'du:d/ Duden ('ee) /'ge:g/ gegen.
On the other hand, within words, b d g followed by heterosyllabic /n, l/ can
become /b, d, g/ (instead of /p, t, k/, as in neutral German pronunciation, which
has neutralization): ('nu) /'OKdnU/ Ordnu ('mli) /'m:klI/ mgli>.
No glottal stop is used, except in less broad accents: (i's) /I'Es/ i> esse (te'AAt) /te'a:t/ eater
5.4.2.6. Among constrictives, we nd a greater variation (than in neutral German) for v mostly when initial: ('vs, 'f-) /'fEKs/ Vers ('fiip, 'v-) /'vi:p/ Viper Normally, /z/ is /s/ (s) (while we nd (, z) only in less broad accents, and more easily
after C]\ ('sAAg) /'za:g/ sagen ('Also) /'alzo/ also ('s) /'Kaez/ rRsen (then identical to ('s) /'Kaes/ rRen] In words such as Stil Strateg unlike neutral
German, /st/ is preferred to /St/: ('stiil) /'Sti:l/ (&stAte'gii) /StKate'gi:/ (although they
are both used, in both accents).
In addition, /j/ is an approximant, (j): ('jAA) /'ja:/ ja; generally, /x/ is velar, (x):
('nAAx) /'na:x/ na>. For /K/ we have /Kx/: ('ux) /'dUK/ dur>; for /-I/ we have
/-Ik/: ('q6Anqik) /'qvanqI/ zwanzig for // we nd /k/: ('kiinA) /'i:na/ ina
but: (me&lA~o'lii) /melako'li:/ Melan>ol Often // is realized as a postpalatal,
(), mostly in Vienna.
e most widely used articulation for /K/ is a uvular constrictive, (), with the
possibility of postnuclear vocalization (as seen in the rst part of this section),
which coexists with a fairly widespread alveolar trill or tap, (R, r) (even velarized,
(R, 5)). is is mainly a non-urban characteristic (but also widespread in Vienna,
although alternating with (, K)), which is best avoided, even if it does not sound
peculiar: ('AA 'RAAR, 'rAAr, -AA) /'Ka:K/ rar
Although, of course, it is certainly not recommendable, we want to mention
that the typical broad cennese l is a velarized alveolar unilateral, (), which may
occur in every position, although it normally alternates with (l), even in the broad-
5. german
219
est and most typical speakers. In the broadest accents (as in the traditional local
dialect see the cennese phonosynthesis, 17.28 of NPT/HPh] we nd velar (,
), in the sequences /kl, gl k, g/. In cennese pronunciation again, in addition
to () (and ()), we often nd it vocalized in (,) (as in the dialect): ('Spii, -,, 'ees, -s,) /'Spi:g, 'e:z/ Spgel Esel.
5.4.2.7. In non-neutral pronunciation, simple word-internal voiceless intersyllabic /0/, after (short) stressed V, are realized as restrained (but evident) geminates,
which we indicate with superscript symbols of the rst element, in the variant
form notated only here, (00): ('Stp 'Stpp) /'StOp/ oppen ('vt 'vtt) /'vEt/ Wetter ('tA 'tA) /'ta/ taer ('siq 'siqq) /'zIq/ sien ('vAs 'vAss)
/'vas/ Wasser ('vAS 'vASS) /'vaS/ wa(en ('si 'si) /'zI/ si>er ('mAx
'mAxx) /'max/ ma>en
e insertion of a homorganic stop between /n, l/ and voiceless constrictives or
/z/ ( 5.3.2) is possible, although not particularly widespread.
ere are some dierences in stressing: (kA'fee) /'kafe ka'fe:/ Kaee ('lAA, lA'oo) /la'bo:K/ Labor (tA'Ak) /'ta[:]bak ta'bak/ Tabak (p'tii, p'tjee) /pOK'tje:/ Portr Motor ('moot, mo'too) /'mo:to:, mo'to:/
Austrian intonation is easily recognized because of its (slightly rising) half-low stressed
syllables and (falling) mid unstressed syllables, occurring between them in preintonemes; mostly the interrogative intoneme (and the suspensive one too) have particular
movements, which can be seen in the tonogram.
Swiss German
5.4.3.1. Even for this accent we will add, when relevant but without further
gures, some more non-neutral characteristics, taken from our sound archives. e
main accent illustrated is the mediatic one something in between the broadest
most local ones and a neutralized, unmarked, accent expressly acquired.
e rst vocogram shows the short and long monophthongs; actually, in this
accent long vowels are diphthongs with narrow movements: ('fiil) /'fi:l/ vl
('f5yy) /'fKy:/ frh ('uut) /'gu:t/ gut ('eet) /'be:t/ Beet ('l) /':l/ l ('oo) /'vo:/
wo ('SpEEt) /'SpE:t/ spt (which is very stable and typical) ('StaAt) /'Sta:t/ Staat e
last two have non-neutral broad variants: (E) and () (or (A) as well, which is
typical of Bern, and (), typical of Zurich). (e phonosynthesis in 17.20 of
NPT/HPh shows the characteristics of Zurich 6wyzert {which is Alemannic}, with still more dierent realizations.)
Among short V, /I, Y, U/ have closer realizations (i, y, u): ('fiS) /'fIS/ ( ('fyMf)
/'fYnf/ fnf ('hun) /'hUnt/ Hund /E, , O/ are relatively open: ('Elf) /'Elf/ elf
('q6lf) /'qvlf/ zwlf ('Oft) /'Oft/ o (with unstressed taxophones (, #, )). For
the spelling we can also nd a broad non-neutral realization (): ('fElt, 'flt)
/'fElt/ fllt /a/ is back-central (A) (which, in broad marked pronunciation, is
rounded too, (), or fully back, ()): ('StAt) /'Stat/ Stadt
220
5.4.3.2. For // we have (), in every context, since /K/ is not uvular: ('5aA)
/g'Ka:d/ gerade ('unsR) /'UnzK/ unsere ('unsR) /'Unz/ unser Besides, we generally nd (-n, -m, -l) for /, , /: ('faARn) /'fa:K/ fahren ('loom) /'lo:z/ losem ('eel) /'e:z/ Esel Only in very careful pronunciation, can we nd (, , ),
and (, , , ) as well, even for speakers lacking /K/ ().
Neutral Swiss diphthongs are /ae, ao, OY/ (a, Ao, OI): ('a) /'aes/ Eis ('lAo)
/'blao/ blQ ('nOI) /'nOY/ nS in the broadest accents we nd: /ae/ (EI, I), /ao/ (U,
U), /OY/ (I, o). Xenophonemes are decidedly closer than in neutral German: /eI,
oU/ (ee, ei oo, ou).
/i:, i/ (ii, i), /I/ (i)
/y:, y/ (yy, y), /Y/ (y)
/u:, u/ (uu, u), /U/ (u)

/o:, o/ (oo, o)
/e:, e/ (ee, e), /:, / (, )

// (, R, R), // (R)
/E/ (E, ), // (, #), /E:/ (EE, ;)
/O/ (O, )
/a:/ (aA), /a/ (A)
/u/ (u)
/i/ (i), /y/ (y)

/eI/* (ee, ei)
/oU/* (oo, ou)

/OY/ (OI) {(I)}
{/a/ ()}
/ao/ (Ao) {(o)}
/ae/ (a) {()}
/I/ (I), /Y/ (Y)

/E/ (), // (#)
/ae/ (EI, I, i)
/E:, E/ > (E, )
/ / (2 2 2 2 2 2 2)
/./ (2 2 3)
/ / ( 2 2 2 2 2 2 2)
/?/ (2 1 2)
/ / ( 2 2 2 2 2 2 2)
// (2 2 2)
/ / ( 2 2 2 2 2 2 2)
/,/ (2 ' 2)
/U/ (U)
/OY/ (i, i, I, o)
/O/ ()
/ao/ (U, , o, U)
/a:/ (, , A)
/a/ (, )
In the second vocogram again, three further typical diphthongs are evident (although marked in grey, since they only occur in local family and place names), (i,
y, u) /i, y, u/: ('it) Dth ('5y) Regg ('hup) Huep Except in less marked
5. german
221
or aected accents, often /y:, y, y/ become /i:, i, i/. In the same gure, possible realizations of reduced forms of /ae, ao, OY, a/ are given: (, ao, , ).
In comparison with neutral German, there are some words with short V, as in:
('A5t) /'a:Kt/ Art ('flOs) /'flo:s/ o [oss\ in Switzerland is not used], ('jA) /'ja:kt/
Jagd ('nun) /'nu:n/ nun ('E5) /'e:Kt/ Pferd ('fO5) /'fo:K/ vor ('E5n) /'ve:Kd/
werden ('E5) /'e:Kd/ Erde ('uks) /'vu:ks/ Wu7 ('qE5tliX, -) /'qhE:tlI/ zrtli>
e last vocogram shows broad realizations of /E, E:, a, a: ae, ao, OY/ (and less
broad ones of /I, Y, U/).
5.4.3.3. Moving to C we have to say that /t, d/ are decidedly dental: (e'tEktoR)
/de'tEktoK/ Detektor whereas, /k/ is (k, w, , k): ('kin, 'w, '-, 'k-) /'kInt/ Kind
('sAk, -w, -, -k) /'zak/ Sa$; please, note that in the transcription of 5.5.2.4
(and 5.5.1.2), we write (), to insist on this typical realization (although it tends
to be avoided in neutral-like pronunciations, perhaps resorting to a velar stop-semi(con)strictive, (), which is half-way between (k) and the other stopstrictives indicated).
/p, t, k , q, c/ are not aspirated (unless aspiration is voluntarily used):
('pool) /'po:l/ Pol ('toon) /'to:n/ Ton ('kuu, 'w, '-, 'k-) /'ku:/ Kuh ('un) /'Unt/
Pfund ('qeen) /'qe:n/ zehn ('cAkko) /'cako/ T(ako.
Furthermore, () is lacking, except in less marked and less typical accents: (te'aAtR) /te'a:t/ eater (iX'Ess, i-) /I'Es/ i> esse as can be seen, though, lexeme
and grammeme boundaries are respected; only in very broad pronunciations resyllabication is possible: (i'XEss, i'-).
Voiced C of diphonic pairs are actually half-voiced, (, , , , ), even before
pauses or when syllable-nal (where, in neutral German, they are neutralized and
change to /p, t, k f, s, S/). erefore, we generally nd /s/ for /z/, after pauses or C\
('uu) /'bu:b/ Bube ('uu) /'bu:p/ Bub ('uu) /'du:/ du ('un) /'Unt/ und ('5een) /'Ke:d/ reden ('oon) /'bo:g/ Bogen ('taA) /'ta:k/ Tag ('in) /g'vIn/ Gewinn ('leen) /'le:z/ lesen ('saAn) /'za:g/ sagen ('ins) /'bInz/ Binse ('paA)
/'pa:Z/ Page
5.4.3.4. Often, /f/ occurs instead of (neutral) /v/: (no'fEmR) /no'vEmb/ November (fn'tiil) /vEn'ti:l/ Ventil (ful'kaAn, -w, --, -k-) /vUl'ka:n/ Vulkan
For sp internal or nal we have /sp, st/, but in broad accents we nd /Sp, St/:
(k'nOsp, -Sp) /k'nOsp/ Knospe (ist, -iSt) /bIst/ bi ('EstRn, -St-) /'gEstn/
geern Generally, for internal /S/ (after stressed short V, like for the other C] we
have (SS): ('muSSl) /'mUS/ Mu(el for 7 /ks/, in marked pronunciation it is fairly possible to have /xs/: ('fuks, -Xs, -s) /'fUks/ Fu7
Only in less broad speech, can we nd // (but articulated as postpalatal (), or
prevelar ()); whereas, normally, we have /x/ (X, ): ('milX, - -, --) /'mIl/
Mil> ('AoX, -) /'aox/ Q> -ig is regularly /Ik/ ( /Ig/; while /I/ is decidedly
rare and only voluntarily used): ('q6Anqi) /'qvanqI/ zwanzig and // is normally /k/: ('kiinA, 'w-, '-, 'k-) /'i:na/ ina /j/ is an approximant, /j/: ('jaA) /'ja:/ ja
222
5.4.3.5. e typical articulation of /K/ is uvularized alveolar, a trill in stressed

syllables (5), and a tap in unstressed syllables (R), in every context, even after V\
('5aA5) /'Ka:K/ rar (Re'Akto[o]R) /Ke'akto:K/ Reaktor In regional pronunciation (or,
on the contrary, intentional pronunciation, under the inuence of neutral German pronunciation), uvular realizations are possible: (, K, ).
In neutral pronunciation /l/ is always alveolar (whereas in non-neutral pronunciation we often nd (], ), before pauses or C]\ ('All) /'al/ alle ('leen) /'le:b/ leben ('Olf) /'vOlf/ Wolf ('hEl) /'hEl/ hell
5.4.3.6. In word-internal position, in neutral Swiss pronunciation, simple intersyllabic /0/, after short stressed V (both voiceless and voiced, and spelled as geminates >pp bb tt dd gg ss mm nn rr ll, or represented with digraphs >$
> or tri-/quadri-graphs >( ), are realized as restrained (but perceptible) geminates, which we indicate with superscript symbols of the second element, (00), thus: ('immR) /'Im/ immer ('innn) /'In/ innen ('huR) /'hU/
Huer and ('StOppn) /'StOp/ oppen ('k5A) /'kKab/ Krabbe ('EttR) /'vEt/
Wetter ('iR) /'vId/ Widder ('5On) /'KOg/ Roggen ('sOkkn, --, -ww-,
--, -kk-) /'zOk/ So$en ('tAR) /'ta/ taer ('siqqn) /'zIq/ sien ('5Eccn) /'gKEc/ gren ('Aff) /'af/ Ae ('AssR) /'vas/ Wasser ('ASSn) /'vaS/
wa(en ('siXXR) /'zI/ si>er ('mAXXn) /'max/ ma>en ('hA5R) /'haK/ harren
('k6Ell) /'kvEl/ Quelle
For stressing, we have cases such as: ('AotooR, Ao'too5) /'aoto:K/ Autor ('lAoo, lA'oo) /la'bo:K/ Labor ('mootR, mo'too5) /'mo:to:, mo'to:/ Motor ('ufft, uf'fEt)
/bY'fe:/ Buet ('hOttl, ho'tEl) /ho'tEl/ Hotel (nOI'jaA5) /'nOYja:K/ NSjahr In addition,
there are interesting words such as: (A&nn'mEnt) /abOn'ma, -:/ Abonnement
Swiss intonation is easily recognized because of its (slightly rising) low stressed syllables and mid unstressed syllables, which continue the pitch movement at least in normal preintonemes; even the marked intonemes have particular movements a close examination of the tonogram is highly recommended.
South-Tyrol/Alto-Adige German
5.4.4.1. e rst vocogram shows the short and long monophthongs of South-Tyrolese mediatic pronunciation. Even for this accent it is immediately clear that
the articulations corresponding to /I, Y, U E[:], , O/ are closer than in neutral German so they are represented with (i, y, u [:], #, ) (in less broad accents we nd
(I, Y, U E[:], , O), and unstressed (, #, ) only slightly closer than in neutral
pronunciation): ('fiS) /'fIS/ ( ('fyMf) /'fYnf/ fnf ('hunt) /'hUnt/ Hund ('lf) /'Elf/
elf ('Sp:t) /'SpE:t/ spt ('qV#lf) /'qvlf/ zwlf ('ft) /'Oft/ o
5.4.4.2. /a:/ is diphthongized: ('Stat) /'Stat/ Stadt ('StaAt) /'Stat/ Staat (in less
broad accents /a, a:/ are more similar to one another: (a, a:); whereas, in the broadest accents, near the dialects, we can also have (, :) (which will certainly be noted at once; 16.15 of NPT/HPh, where the phonosynthesis of the South-Tyrolese
5. german
223
dialect koin is given).

Generally, Tyrolese traditional dialects have phonemic opposition between a
non-back a and a back rounded one, both short and long. It is possible to nd this
timbre distinction in the language too, where the back timbre is used in traditional
words, whereas the non-back one is used in loanwords and neologisms, as happens
for instance in Ball\ /'pl/ ('pl, 'pAl) ball, /'pal/ ('pal) dance.
Very broad accents have /e:, o:/ (eI, oU) (as in the dialect): ('ve:, 'veI) /'ve:/ weh
('soo, 'soU) /'zo:/ so (to these /:/ (Y) is added: ('p, 'pY) /'b:/ B which does
not belong to the genuine dialect that has historically merged the front-central
rounded series with the front one). On the other hand, the bourgeois dialect of
Bolzano (Bozen) has restored /y:, Y, :, , OY/ (y:, y, :, #, Y), under the inuence
of neutral German pronunciation, including /y/ (y, i), by structural analogy.
5.4.4.3. // is realized as (, ), with the vocalization of /K, / () (with consonantal accompaniments too), and in the most broad pronunciations even (), as
in the dialects ( the phonosynthesis); thus we have: ('uns) /'UnzK/ unsere and
('vas, -, -, -, - 'vss, -, -, -) /'vas/ Wasser It is important to note that
(, ) //, in less broad accents, given in the last vocogram, are realized slightly
dierently (as other V are too) although they are represented by the same symbols.
nal -e (not only of verbs) and the rst one in -ere(C) tend to be dropped (except
in Val Pusteria, or Pustertal), especially in non-controlled pronunciation: ('ty:p,
'ty:p) /'ty:p/ Type (which can coincide thus with ('ty:p) /'ty:p/ Typ]
e second vocogram shows the diphthongs of South Tyrolese German. ile
/OY/, in comparison with neutral German, diers mostly in its second element that
is unrounded, (O); /ae, ao/ are decidedly less wide, (a, aP): ('as) /'aes/ Eis ('plaP)
/'blao/ blQ ('nO) /'nOY/ nS In the dialect, for /ae, ao/, we have (a, AP), but also
(aP), which are even less wide; only in Bolzano the same is true for /OY/ (), since
in traditional dialects it has merged into /ae/ (a), or it is represented by /ui, oi/:
('nui) /'nui/ nui {= nS}, ('foiR) /'foiK) foir {= FSer}. e xenophonemes are quite
rarely used (and, if they are, it is in a completely intentional way): /eI, oU/ (e:, ei
o:, ou).
5.4.4.4. In the second vocogram, two further typical diphthongs are evident (although marked in grey, since they only occur in local family and place names), (i,
uX) /i, u/: ('tit, 'iit) Dtl ('wuXns, 'ku:ns) Kuens In the same diagram, the
possible realizations of /ae, ao, OY, a/ (, P, , ), in reduced forms, have been
marked.
As already said, the last vocogram shows less broad vowel realizations (()),
which are rather similar to the neutral German ones; however, this pronunciation
is still recognizable, since the prosodic characteristics have typical dierences.
ere are dierent distributions too, as for instance /E/ in ('st, 'E- 'St) /'e:Kst/
er ('t, 'E- 't) /'e:Kt/ Pferd (which in neutral German have /e:/: ('e:st,
'he:t)).
224
ff
/i:, i/ (i:, i), /I/ (i)
/y:, y/ (y:, y), /Y/ (y)
/u:, u/ (u:, u), /U/ (u)

f
/e:, e/ (e:, e), /:, / (:, )
/o:, o/ (o:, o)
/O/ (, )
// () {//K// ()}
/a/ (a), /a:/ (aA)
/E/ (, ), // (#, #), /E:/ (:, ;)

// (, , -, -)
/u/* (uX)
/i/* (i)
/eI/* (e: ei)
/oU/* (o: ou)
{/a/ ()}
/OY/ (O) {()}
/ae/ (a) {()}
/ao/ (aP) {(P)}
/i:, i/ (i:, i), /y:, y/ (y:, y)

/I/ (I), /Y/ (Y)
/e:, e/ (e:, e), /:, / (:, )
/E/ (E, ), // (, #)
/E:/ (E:, ;)
/a/ (a)
/a:/ (a:)
/ / (2 2 2 2 2 2 2)
/./ (2 ' 2 3)
/ / ( 2 2 2 2 2 2 2)
/?/ (2 ' 2 1)
/ / ( 2 2 2 2 2 2 2)
// (2 ' 2 2)
/ / ( 2 2 2 2 2 2 2)
/,/ (2 ' 2)
/u:, u/ (u:, u)
/U/ (U)
/o:, o/ (o:, o)
// (, , -, -)
/O/ (O, )
// () { (, )}
5.4.4.5. Moving to C we nd that generally internal // becomes /g/ (but it

is not so in less broad accents): ('pin, 'I) /'bKI/ brien (however, even
without (g), as in less broad accents, the nasal is always velar, not uvular). Especially in checked syllables with N (above all in stressed positions), V are nasalized
in broad accents: ('vi:n, 'vi:n) /'vi:n/ Wn ('fa, 'f) /'fa/ Fa
/t, d/ are dental: (te'tkto) /de'tEktoK/ Detektor often /k, k, k/ are stopstrictives, or even stop-semi(con)strictives, velar (or postpalatal, near front V]\ ('km,
'w-, '-) /'kOm/ kommen ('sk, 'swn, --) /'zOk/ So$en ('tik, -w, -, -k,
-%) /'dIk/ di$
Voiceless stops and stopstrictives are not aspirated (generally even in less broad
accents) as we have just seen for /k/: ('po:l) /'po:l/ Pol ('to:n) /'to:n/ Ton ('kint, 'w-,
'k-) /'kInt/ Kind ('unt) /'Unt/ Pfund ('cako) /'cako/ T(ako
Phonemic voiced stops are voiceless in typical pronunciation, but half-voiced,
5. german
225
(, , ) in less broad accents. However, between voiced phones, they are half-voiced, (, , ) (except, usually, in the dialects which mostly have (p, t, k/w),
as in the broadest accents. Instead, in less broad accents they are voiced, (b, d, g)):
('pu:, -b) /'bu:b/ Bube Half-voiced realizations are possible before internal /n,
l/, too: ('Onu, 'm:li) /'OKdnU, 'm:klI/ Ordnu mgli>. (In Bolzano, we
can nd (B, , ) /b, d, g/: ('u:B).)
ere is no (), except in less broad accents, where however it is less evident than
in neutral German (and, in this case, we could mark it with a special symbol, (,)):
(i's, ,I',Es) /I'Es/ i> esse (te'a;At, te',a:-) /te'a:t/ eater
5.4.4.6. For constrictives, it is to be noted that /z/ is normally /s/ (s) (generally,
even in less broad accents, in Bolzano as well): ('sa;A) /'za:g/ sagen ('also) /'alzo/ also ('as) /'Kaez/ rRsen (which becomes identical to ('as) /'Kaes/ rRen] In initial position, before C /s/ practically never occurs (substituted by /S/);
and, as far as internal and nal /st/ is concerned, in broad accents, as in the dialects, we typically nd /St/: ('li:pst, -St) /'li:pst/ lben ('snst, -nSt) /'zOnst/
son However, in case they are heterolexemic, we have /st/: ('sams&taAw) /'zamsta:k/ Samag (as in the dialects: ('sams&tiw, 'sms&tik) /'samstik/ samig]
For (tautosyllabic) /0v/, we have (0V): ('SVst) /'SvEst/ 6weer ('qVa)
/'qvae/ zwR ('kVl) /'kvEl/ Quell for /f/, it is possible to have /v/, besides in forms
such as Vize (which, even in neutral German, may have the variant with /v/, although not prevailing), even for v against neutral phonemic rules, in cases such
as positiv\ ('vi:q, &posi'ti). In addition, /j/ is approximant (j): ('ja;A) /'ja:/ ja; generally, /x/ is velar (x): ('na;Ax) /'na:x/ na>; // is maintained (even before /K/, contrary
to what happens in Austria): ('tu) /'dUK/ dur>; but, for /-I/ we have /-Ik/ (except in less broad accents) ('qVanqik, -w, -k) /'qvanqI/ zwanzig for // we systematically nd /k/: ('ki:na, 'w-, 'k-) /'i:na/ ina However, in the dialects, except
in Bolzano, () // does not exist at all.
e most widely used articulation for /K/ is a uvular constrictive, (), which coexists with a fairly widespread uvular trill, (K), with the possibility of postnuclear
vocalization (that, as we have already seen, is back-central () (or back, (), in the
dialects and in the broadest accents). In absolute nal position, after stressed V, in
the dialects and in the broadest accents we have (R): ('a;A -a;A -a;AR -:R)
/'Ka:K/ rar ('hi: 'hi: 'hi:R) /'hi:K/ hr
5.4.4.7. In non-neutral South-Tyrolese pronunciation, simple word-internal intersyllabic (voiceless, nasal and lateral) /0/, after (short) stressed V are realized as
restrained (but fairly evident) geminates, which we indicate with superscript symbols of the rst element, in the variant form notated only here, (00): ('im 'imm)
/'Im/ immer ('Stp 'Stpp) /'StOp/ oppen ('vt 'vtt) /'vEt/ Wetter ('ta
'ta) /'ta/ taer ('siq 'siqq) /'zIq/ sien ('vas 'vass) /'vas/ Wasser ('vaS 'vaSS) /'vaS/ wa(en ('si 'si) /'zI/ si>er ('max 'maxx) /'max/ ma>en ('kVl 'kVll) /'kvEl/ Quelle
e insertion of a homorganic stop between /n, l/ and voiceless constrictives or
/z/ ( 5.3.2) is possible, although not particularly widespread.
226
In lofty or foreign words, there is a frequent use of Italian or Italian-like structures: /gn/ = /nj/: (po~'jo:s) /pKog'no:z/ Prognose (si~'ja;Al) /zIg'na:l/ Signal (&esi~'ji:, -, -'ji:n, -n) /KezIg'ni:K/ resignren /t/ = /q/: (ti&ploma'qi:) /diploma'ti:/ Diplomat (&kaan'qi:) /gaKan'ti:/ Garant /q/ = /c/: (m'ce:s) /mEK'qe:dEs/ Mercedes (&pace'lo:na) /baKqe'lo:na/ Barcelona /S/ = /sk/: ('ske:ma) /'Se:ma/
S>ema (&skiqo'fe:n) /Siqo'fKe:n, si-/ S>izo#ren /Z/ = /dZ/: (&tuna'list) /ZUKna'lIst/ Journali /kv/ = /kw/ (kj) (in addition to (kV)): (kja'tt) /kvaK'tEt/ Quartett For V we nd: /y/ = /u/: (pu&oka'qi:) /byKokKa'ti:/ Brokrat /Oy/ = /eu/:
(eu'o:pa) /OY'Ko:pa/ Europa (neu'ta;Al) /nOY'tKa:l/ nStral
South-Tyrolese intonation is easily recognized because of its half-low stressed syllables, except the rst one which is half-high; mostly the interrogative intoneme (and the
suspensive one too) have particular movements, which can be seen in the tonogram.
Text
5.5.0. e story e North Wind and the Sun follows, given in some dierent
(normalized) versions. We start with the (neutral) German pronunciation of
(neutral British) English this is the rst step of the phonetic method (the written text is given in 2.5.2.0). e German translation follows, in its neutral version; then the north-eastern accent, followed by the Austrian, Swiss, and South-Tyrolese ones. ese transcriptions would have been (visually) more dierent, if
we had used the special (denti-alveolar) symbols ((4, 7, )), and instead of (, )
(for /K, /) respectively the following possible variants: () (neutral), (x) (north-eastern), (, a) (Austrian), (5, R) (Swiss), (, ) (South Tyrolese).
At the end, as always, there is the version which gives the English pronunciation of German, by neutral British speakers, uent in German (after prolonged
contact with native speakers, but with no help from the phonetic method), who
have adequately learned the relative prominences, but who substantially use segmental and intonation elements which are typical of neutral British English (for
reference purposes, although, of course, a neutral accent is not so common). Obviously, the same principle is valid for the foreign pronunciation of English, given rst.
Speakers of American English could prepare their own version both of the German pronunciation of English and of their pronunciation of German, as an excellent exercise, by listening to native speakers, best of all after recording them. Of
course, speakers of other languages could do the same thing. e author would be
happy to receive their transcriptions and recordings, both in case of help should
they need it and to make their contribution known to others (possibly in our
website on canIPA Natural Phonetics 0.12).
German pronunciation (of English)
5.5.1. (z'nOs 'Int2 nt'zan2 v#dI'spu:tI 'vIc s'StOg3 3| 'vn e'thEfl2
'khe:m 'lO2 'Ept Ine'vOm 'khlo:k3 3| &ze'gi:t2\ &zt'van hu'fst ak'si:dt2
5. german
227
Im'me:kI z'thEfl2 'the:k hIs'khlo:k 5Of1 1| &SUtikn'zi:d "StOg &znzi'az3 3||

'zn2 z'nOs 'Int2 'lu:2 'hat2 shi'khUt3 3| &at'mO hi5blu:1 1| z'mO 'khlo:sli2
&dIt'thEfl2\ 'fo:lt hIs'khlo:k 'aonthIm3 3| &nt'la[:]st2\ z'nOs 'Int2 'e:f 'ap i'thEmpt3 3|| 'zEn2 z'zan 'SOn 'aot3 3 'Omli3 3| &ntI'mi:d,tli2\ z'thEfl 'thUk 'Of3 3
hIs'khlo:k3 3|| nt'o:2 z'nOs 'Int2 so'blaec tukM'fEs2| &zt'zan3 3 vs'StOg3 3
&f'thu:3 3||
&Itu'laek1 1 z'StOi2| ju'vOn tu'hi:It 'ge:n1 1|||)
German text
5.5.2. Ein ritten si> Nordwind und Sonne, wer von ihnen bRden wohl der
Strkere wre, als Rn Wanderer, der in Rnen warmen Mantel gehllt war, des Weges
daherkam. S wurden Rnig, dass derjenige fr den Strkeren gelten sollte, der den
Wanderer zwien wrde, sRnen Mantel abzunehmen.
Der Nordwind bls mit aller Ma>t, aber je mehr er bls, deo feer hllte si>
der Wanderer in sRnen Mantel Rn. Endli> gab der Nordwind den Kam Qf. Nun
erwrmte d Sonne d Lu mit ihren frSndli>en Strahlen, und (on na> wenigen Augenbli$en zog der Wanderer sRnen Mantel Qs. Da muse der Nordwind zugeben, dass d Sonne von ihnen bRden der Strkere war.
Hat dir d Ge(i>te gefallen? Wollen wir s wderholen?
Neutral German pronunciation
5.5.3. ('aenS2 'StItzI 'nOt&Int2 Un'zOn2| &vefni!'baed/&vo;l d'StEk&v;3 3| &alseM'vand2 &deInaen/'vam 'mant g'hYlt&a;2 s'e:gs2 'he:&kha;m3 3|| zI&vUd'aenI2| as'e:&,e;nIg2 &fY[]d'StEk 'gElt&zlt2 &ded/'vand 'qhI|&vYd2 &zaen!'mant 'apq≠m3 3||
'nOt&Imp 'li:s2\ mIt'al2 'maXt3 3|| &ab,e'me: 5bli:s1 1| &sto'fEst2 'hYltzI
d'vand2 In&zaen!'mant 'aen3 3|| 'EntI2 'a:p 'nOt&In2 'khaM 'aof3 3||
'nu:n2\ 'vEmt dI'zOn2 dI'lUft2| mItiM'fOYntI2 'Sta:l3 3|| Un'So:n2 na;X'e:nIg
'aog&blIk2| 'qho:k3 3 'vand2 &zaen!'mant 'aos3 3|| 'a:2\ &mUstd'nOt&In2
'qhu:≥b2\ &dasI'zOn2| &fni!'baed2| 'StEk&va;3 3||
hat&IdIg'SIt2 g'fal1 1| 'vOl/vI zI'vi:d&ho;l1 1|||)
Neutral north-eastern German pronunciation
5.5.4. ('anS2 'StItzI 'nt&Int2 Un'zn2| &vefni!'bad/&vool d'Stk&v23| &alsM'vand2 &deInan/'vam 'mant g'hYlt&aa2 s'e;egs2 'he;e&khaam23|| zI&vUd'anI2| as'e;e&,eenIg2 &fY[]d'Stk 'glt&zlt2
228
&ded/'vand 'qhI|&vYd2 &zan!'mant 'apq&neem23||

'nt&Imp 'li;is2\ mIt'al2 'maXt23|| &ab,e'me;e 5bli;is21| &sto'fst2 'hYltzI d'vand2 In&zan!'mant 'an23|| 'ntI2 'a;ap 'nt&In2 'khaM
'aPf23|| 'nu;un2\ 'vmt dI'zn2 dI'lUft2| mItiM'f+ntI2 'Sta;al23|| Un'So;on2
naaX'e;enIg 'aPg&blIk2| 'qho;ok23 'vand2 &zan!'mant 'aPs23|| 'a;a2\
&mUstd'nt&In2 'qhu;u&geeb2\ &dasI'zn2| &fni!'bad2| 'Stk&vaa23||
hat&IdIg'SIt2 g'fal21| 'vl/vI zI'vi;id&hool21|||)
Neutral Austrian pronunciation
5.5.5. ('nS2 Stitsi 'nt&vint2 un'sn2| &vefni!'/&vool 'Stk&v23| &AlsM'vAn2 &einn/vAm mAnt 'hylt&vAA2 s'vees2 'hee&kAAm23|| si&vu'nik2| As'ee&jeeni2 &fy[]Stk 'lt&slt2 &e/vAn 'q6i|&vy2 &sn!mAnt 'Apq&neem23||
nt&vimp 'liis2\ mit'Al2 'mAxt23|| &Ajemee 'liis21| &sto'fst2 hyltsi
'vAn2 in&sn!'mAnt 'n23|| 'ntli2 AAp 'nt&vin2 'kAM 'f23|| 'nuun2\ vmt i'sn2 i'luft2| mitiM'fYntli2 'StAAl23|| un'Soon2 nAAxveeni
'&lik2| 'qook23 'vAn2 &sn!'mAnt 's23|| 'AA2\ &must'nt&vin2
'quu&ee2\ &Asi'sn2| &fni!'2| 'Stk&vAA23||
hAt&ii'Sit2 fAl12| vl/vi sivii&hool12|||)
Neutral Swiss pronunciation
5.5.6. ('anS2 St5ittsiX 'nO5t&vint2 un'sOnn2| &veRfni!'a/&vool RStE5R&vR23| &AlsM'vAnRR2 &eRinan/vA5m mAnt 'hylt&vaAR2 s'vees2 hee5&aAm23|| si&vuR'ani2| As'ee5&jeeni2 &fyRStE5R 'Elt&slt2 &eR/vAnRR 'q6i|&vyR2 &an!mAnt Apq&neem23||
RnO5t&vimp 'liis2\ mit'AllR2 mAXt23|| &ARjemee5 Rliis2 2| &sto'fEstR2 hyltiX R'vAnRR2 in&san!mAnt 'an23|| 'EntliX2 aAp R'nO5t&vin2 AM
'Aof23|| 'nun2\ RvE5mt i'Onn2 i'luft2| mitiRM'f5OIntliX?2 St5aAl23|| un'Soon2 naAXveeni 'Ao&li?2| qoo23 R'vAnRR2 &san!mAnt 'Aos23|| 'aA2\ &mustR'nO5t&vin2 'quu&ee2\ &Asi'Onn2| &fni!'a2| RStE5R&vaAR23||
hAt&iRi'SiXt2 fAll12| vOll/viR siviiR&hool12|||)
5. german
229
Neutral South Tyrolese pronunciation

5.5.7. ('anS2 'Stitsi 'nt&vint2 un'sn2| &vefni!'a/&vo;l 'Stw&v;3 3| &alsM'van2 &einan/'vam 'mant 'hylt&vaA2 ts've:s2 t'he:&waAm3 3|| si&vu'anik2| tas'te:&je;ni2 &fy[]'Stw 'lt&slt2 &e/'van 'qVi|&vy2 &san!'mant 'apq≠m3 3||
t'nt&vimp 'pli:s2\ mit'al2 'maxt3 3|| &a,e'me: 5li:s1 1| &tsto'fst2 'hyltsi 'van2 in&san!'mant 'an3 3|| 'ntli2 'wa;Ap t'nt&vin2 t'waM 'aPf3 3|| 'nu:n2\
'vmt i'sn2 i'luft2| mitiM'fYntli2 'Sta;Al3 3|| un'So:n2 naAx've:ni 'aP&lik2| 'qo:w3 3 t'van2 &san!'mant 'aPs3 3|| 'ta;A2\ &must'nt&vin2 'qu:&e;2\ &asti'sn2| &fni!'a2| t'Stw&vaA3 3||
hat&tii'Sit2 'fal1 1| 'vl/vi si'vi:&ho;l1 1|||)
English pronunciation of German
5.5.8. ('ansT2 5s>TnzS 'n;D&vnD2 nD'zn2| 5v; fn&Iinm'baDM&v D'sTk>&v3 3| &A;saM'vA;nD>2 &D>nanM5vA;mm 'mnT g'h;T&vA;2 Ds'vIgs2 D'h;&khA;m3 3|| zIi&vDn'anS2| Ds'D;&jIng2 &f;Dn5sTk> 'gT&zT2 &DDM5vA;nD> 'thsvM&v;D2 &zanm5mnT 'ts&nImn3 3||
D5n;D&vmb 'blIis2\ mT'l2 'mA;xT3 3|| &A;bj5m;> 'blIis32| &DsT'fsT2 5h;TzS
D'vA;nD>2 n&zanm'mnT 'a;n3 3|| 'nDlS2 'gA:b D'n;D&vn{D}2 D5khA;mpf
'af3 3|| 'n;un2\ 'vmT D'zn2 D'lfT2| mT&'f>nTlS 's>A;ln3 3|| n'S;n2 nA;x5vIngn 'agm&blkn2| 'tshk3 3 D'vA;nD>2 &zanm5mnT 'as3 3|| 'DA:2\
&msTD'n;D&vnD2 'tsu&gIbn2\ &DsD'zn2| &fnIinm'baDn2| D'sTk>&vA;3 3||
&hTDDg'SST2 g'fln21| 5vlMv zIi'vIiD&hln21|||)
6. Spanish
6.0. e Spanish accents that will be dealt with are the neutral Iberian and the
(central-southern) American versions. ere are really very few dierences in neutral pronunciation; therefore a diaphonemic transcription is not necessary (whereas for Portuguese and English it is); the American variant simply has two phonemes less: /, L/, which become /s, J/. e former two symbols, however, are used
diaphonemically, as they distinguish the two accents.
Requiring informative symbols, to recall this phenomenon more clearly, one
could resort to /, L/, or if inclined to a more international type of transcription
to /s, J/. On the other hand, again, to help foreigners, it is worth indicating the
phonemes /b, d, g/ with /b, , g/, in contexts where they are realized as (B, , ).
Placing /B, , / (as seen in some text-books and dictionaries with transcriptions)
is no good usage at all: because they are not phonemes, and because in the inection of lemmata things do not remain constant.
Although, even for Spanish, the number of speakers with American accents is
decidedly more consistent than for the Iberian ones, we prefer to consider the latter primary, because it is closer to the written word; therefore it is more advisable, for teaching purposes too, also for more coherent spelling, which, on the other hand, is a safer guide to pronunciation.
Obviously, even the American accent has many internal variants, which we here
consider to be non-neutral, even if, they are often more widespread than neutral
pronunciation, which (as happens in every language) is decidedly minimal.
Generally, these variants are not stigmatized in the way easily localized pronunciations are, nor do they have negative connotations, as each nation has its own
national accent as well as regional variants. A more in-depth study of Spanish
pronunciation will give the macro-koins (at least seven for America), to be considered neutral, as well as more specic, regional variants.
On the other hand, the non-neutral characteristics of Spanish are shared by most
Hispanic American nations, and in Europe as well, so as to make the non-neutral
accents more alike, even if they remain recognizable. For this reason, regarding
Spanish, other accents will not be presented, other than the two neutral ones (as
done, instead for German, which has more distinct accents, above all in Switzerland and Austria).
However, recordings are being collected for a systematic description of the actual Spanish accents. In the meantime, the phonosynthesis of eastern Andalusian can
be seen (in 17.9 of NPT/HPh).
6. spanish
231
Vowels
6.1.1.1. Spanish has only ve vowel phonemes realized as shown in the rst vocogram in 6.1 (i ', e a ', o u) /i, e, a, o, u/. For e, o in the neutral Iberian-American accent, in stressed syllables, (, ) are normal, but, in unstressed syllables, we nd (e, o).
We therefore have examples such as: ('i)i ('si)a /'si/ s ('kRi;i)i (-sis)a /'kRisis/
cr ('mil) /'mil/ mil (eR'nR)i (s-)a /eR'neR/ cerner (pa'pl) /pa'pel/ papel ('tjRr:a) /'tjeRr:a/ tierra ('i)i ('sis)a /'seis/ se ('r:to)i ('r:s-)a /'r:esto/ rto ('t;Xa)i
(-xa)a /'texa/ teja (de'XaR)i (-x-)a /e'xaR/ dejar ('jmpRe)i ('s-)a /'sjempRe/ siempre
(a'tnto) /a'tento/ atento (u't)i (us-)a /us'te/ ted ('p)i (-s)a /'pe/ pez ('de)i (-z-)a /'ese/ dde ('k;o)i (-so)a /'keso/ o
And: ('p;Co) /'peco/ pe>o (kom'pR) /kom'pRe/ compr ('anda) /'anda/ anda
(pa'ta;ta) /pa'tata/ patata ('gRr:a) /'goRr:a/ gorra ('i) /'oi/ hoy ('r:;a)i (-sa)a /'r:osa/ rosa (';Xa)i (-xa)a /'oxa/ hoja (mo'XaR)i (-x-)a /mo'xaR/ mojar (fa'BR) /fa'boR/
favor ('l)i ('s-)a /'sol/ sol (r:a'n)i (-s-)a /r:a'on/ razn ('glpe) /'golpe/ golpe
('p;Lo)i (-,o)a /'poLo/ poo (La'm)i (,a-)a /La'mo/ am (de'k;Ro) /e'koRo/ decoro ('tu;Bo) /'tubo/ tuvo (r:u'mR) /r:u'moR/ rumor
6.1.1.2. More sophisticated descriptions, as those by Navarro Toms, are, in
actual fact, excessive; indeed, even acoustic data (which, often indicate dierences
not heard by the human ear {not even natives'}), generally agree that there are only ve realizations. In the pre-phonemic time, Navarro Toms represented the
stressed and unstressed vocoids with dierent symbols, including gradations dependent on their position in a checked or unchecked syllable, or because of contextual inuences, leading back to only ve elements. Normally, however, the hyper-dierentiated gradations come under the stressed ((i, , a, , u)) and unstressed
((i, e, a, o, u)) phones ( 6.1 again).
6.1. Spanish vowels (see text for the second vocogram).
/i/ (i)
/e/
(, , e)
/a/ (a)
/u/ (u)
/o/
(, , o)
/i/ (i)
{((I))}
/e/ ()
{((e, E))}
/a/ (a)
{((a, q ))}
/u/ (u)
{((u))}
/o/ ()
{((o, O))}
/we/ (w, w,
w, j, we,
w, w, j+, +)
Resorting to special symbols ( 8.12 of NPT/HPh), which indicate the intermediate gradations between our fundamental vocoids, however, we could usefully prot from some important indications, not only with reference to the second
vocogram given (for which they would be undoubtedly more useful), but also for
(i, u) ((I, u)) and for (a@, a# a) ((a, q )) (as it would be excessive to use (I, U ,
A ), and, for those which follow, (e, E o, O)).
For the articulation of e, o, stressed in pronunciation, there is considerable varia-
232
tion (for speakers words; the second vocogram of 6.1); indeed, sometimes
(E O) ((E, O)) can be found, especially in checked syllables, or in the /ei, oi r:e, r:o
ex, ox/ sequences; or, (/e) ((e)), (/o) ((o)), above all in unchecked syllables; and
also for /e/ in checked syllables in /m, n d , s/. However, it is not necessary to
point these out, as their regular timbre, in a stressed position, (, ), is normal and
safe, used by real natives (despite the variations). In unstressed syllables, in the
same contexts, instead of ((E, O)), we obviously nd (, ).
For the sake of curiosity, and only here, we give some adapted cases which are
pertinent to the aforementioned examples: ((R'nER)) /eR'neR/ cerner ((pa'pEl)) /pa'pel/ papel (('tjERr:a)) /'tjeRr:a/ tierra (('Ei))i (('sEis))a /'seis/ se (('r:Eto))i (('r:Es-))a /'r:esto/ rto (('tE;Xa))i (-xa)a /'texa/ teja ((d'XaR))i (-x-)a /e'xaR/ dejar (('jempRe))i ('s-)a
/'sjempRe/ siempre ((a'tento)) /a'tento/ atento ((u'te[]))i (us-)a /us'te/ ted (('pe))i
(-s)a /'pe/ pez (('dee))i (-z-)a /'ese/ dde (('ke;o))i (-so)a /'keso/ o
And: (('pe;Co)) /'peco/ pe>o ((kOm'pRe)) /kom'pRe/ compr (('gORr:a)) /'goRr:a/ gorra (('Oi)) /'oi/ hoy (('r:O;a))i (-sa)a /'r:osa/ rosa (('O;Xa))i (-xa)a /'oxa/ hoja (m'XaR)i
(-x-)a /mo'xaR/ mojar ((fa'BOR)) /fa'boR/ favor (('Ol))i ('s-)a /'sol/ sol ((r:a'On))i (-s-)a
/r:a'on/ razn (('gOlpe)) /'golpe/ golpe (('po;Lo))i (-,o)a /'poLo/ poo ((La'mo))i (,a-)a
/La'mo/ am ((de'ko;Ro)) /e'koRo/ decoro ((r:u'mOR)) /r:u'moR/ rumor
Also for /i, a, u/: ((en'tIR)) (en'tiR)i (s-)a /sen'tiR/ sentir, ((&aI'naR)) (&ai'naR)i (-s-)a
/asig'naR/ ignar, (('I;Xo)) ('i;Xo)i (-xo)a /'ixo/ hijo, (('mIRr:)) ('miRr:a) /'miRr:a/ mirra,
(('r:I;ko)) ('r:i;ko) /'r:iko/ rico; (('ka;N)) ('ka;Na) /'kaNa/ caa, (('ma;Co)) ('ma;Co) /'maco/ ma>o, (('ma;,o)) ('ma;,o) /'maJo/ mayo, (('ka;Le)) ('ka;Le)i (-,e)a /'kaLe/ cae, (('baile)) ('baile) /'baile/ baile; (('qlto)) ('alto) /'alto/ alto, (('mq;X)) ('ma;Xa)i (-xa)a /'maxa/
maja, ((kq'XOn)) (ka'Xn) /ka'xon/ cajn, (('kqu)) ('kaua)i (-sa)a /'kausa/ caa, ((q'un)) (a'un) /a'un/ an, ((bIl'Bqo)) (bil'Bao) /bil'bao/ Bilbao, ((q'O;R)) (a';Ra) /a'oRa/ ahora; (('tuRko)) ('tuRko) /'tuRko/ turco, (('Xunt)) ('Xunta)i ('x-)a /'xunta/ junta, (('lu;Xo))
('lu;Xo)i (-xo)a /'luxo/ lujo, ((tuR'r:On)) (tuR'r:n) /tuR'r:on/ turrn, ((r:u'mOR)) (r:u'mR)
/r:u'moR/ rumor.
6.1.1.3. After /'i, 'u/, /e|, o|/ remain (e, o), even if at times we can nd (, )
(in neutral pronunciation too): (fe'li;e)i (-ses)a /fe'lies/ felic ('r:i;ko) /'r:iko/ rico Even in an unstressed syllable, above all next to /x, r:/, we can have the (, )
timbres, but, for neutral pronunciation, (e, o) are regular and adequate: (&Xela'ti;na)i (&x-)a /xela'tina/ gelatina (koR'r:o) /koR'r:eo/ correo.
We also indicate a pronunciation which can be heard, above all in words of frequent usage, in not slow speech, the sequence /we/ can be pronounced (w) (realizing a front-central vocoid); whereas in faster or less controlled speech, the sequence can be reduced to a single vocoid ( a central rounded ()), through progressive shifts: ('pw;Blo, 'pw;- 'pw;- 'pj;- 'p;-) /'pweblo/ pueblo ('nw;Be, 'nw;-
'nw;- 'nj;- 'n;-) /'nwebe/ nueve (&ata'lw;o, -w;- -w;- -j;- -;-)i (&as-)a /asta'lwego/ hta luego (kwe'tjn, -w- -w- -j+- -+-)i (-s-)a /kwes'tjon/ cuestin. However, it is not appropriate to actively adopt this particular kind of pronunciation
(which is shown in the central part of the second vocogram in 6.1).
6. spanish
233
Diphthongs
6.1.2.1. e various possible diphthongs are biphonemic, with phonetic realizations corresponding to those of monophthongs, joined together (as in Italian):
('li) /'lei/ ley (pei'na;o) /pei'nao/ peinado ('io) /'oigo/ oigo (&boiko'to) /boiko'teo/ boicoteo ('baile) /'baile/ baile ('aula) /'aula/ aula (au'a)i (-s)a /au'a/ audaz ('dua) /'eua/ deuda (eu'X;njo)i (-x-)a /eu'xenjo/ Eugenio ('bu) /'bou/
bou including (di'Ria) /i'Ria/ dira (na'Bio)i (-s)a /na'bios/ navos (&konti'nuo)
/konti'nuo/ contino
As far as diphthongs are concerned, we must be absolutely resolute because
(strange though it may seem, in the third millennium), there are deep-rooted incorrect convictions, dragged through centuries, which are particularly valued
even by Hispanic phoneticians. A look at how things really are, would ( could)
be easy, by simply considering what is phonetic, exclusively in phonetic terms.
Instead, the range of mixtures of omnipresent (and interfering) spelling and
grammatical (not to speak of metrical and diachronic) considerations, still loom,
resulting merely in the creation of chaos of a subject which sets itself apart in being
clear and objective.
6.1.2.2. As said, Hispanic literature (not that it is alone unfortunately!) dedicates too much eort in complicating what is quite simple. Indeed, instead of
three very common structures, real diphthongs ((', &, )), hiatuses ((',
&)), and heterophonic sequences ((0), such as, (j), (w), and the like), they continue to consider only two of them: diphthongs (with fusion: syneresis) and
hiatuses (with separation: dieresis) but with strained interpretations of medieval origin, of a graphic-grammatical and graphic-metrical nature.
Indeed (unless one is a magician and can do phonetics based on graphic-grammatical categories), in phonetic terms, it is absurd to speak about diphthongs in
the case of (j, w) (('bjn) /'bjen/ bien ('gwa;pa) /'gwapa/ %apa].
As a matter of fact, only (i, u) (('aiRe) /'aiRe/ aire ('kaua)i (-sa)a /'kausa/
caa] are real diphthongs, as any (', &, ) sequences are (('auto) /'auto/ auto (&auto'Bu)i (-s)a /auto'bus/ autobs (au'tntiko) /au'tentiko/ autntico] and it
is just as absurd to speak of hiatuses for ('i, 'u), given that only (i', u') are real hiatuses, as any other (', &) sequence ((pa'i)i (-s)a /pa'is/ pas] compared to
(pai'a;no)i (-s-)a /pai'sano/ paano a real diphthong. One can, therefore, not trust
literature that only uses two categories ( diphthong and hiatus) and, what is
more, who dangerously mix them, so as to include heterophonic sequences in
diphthong, and the real diphthong in hiatus
Obviously, (real) triphthongs are sequences of three vocoids ('), with prominence on the rst element (certainly not (', '), nor even ('j, 'w), or
(j, w)), which, instead, occur in ('bwi) /'bwei/ buey (&paRa'wai) /paRa'gwai/
Para%ay
234
Consonants
6.2.0. 6.2 gives the consonant articulations, of the two neutral accents, which
are necessary for satisfactory Spanish pronunciation.
Instead, 1.9-15, give orograms, grouped by manners of articulation of all the
contoids given in the chapters of this volume, even as secondary, occasional or regional variants of the 12 languages dealt with.
C {}
i ()i
sa (z)a ()i ()i
()
R|r:
(l)
l
()
k g
()
/J/|(,)
uvular
velar round.
velar
(~)
prevelar
alveolar
palatal
(n)
t d
postalveopalatal
m (M)
F pb
f
_
(B)
dental
labiodental
bilabial
6.2. Table of Spanish consonants.
(,)i
xa ()
(X)i
()
j|()
()
(F)
w|(j)
Li
+ (N, M, M, , ) text
Nasals
6.2.1.1. ere are three nasal phonemes, /m, n, N/, with various taxophones for
/n/ (m, M, n, ~, N, , , ,) ( 1.9.1-2; () is semi-provelar without full contact; we
could add dental (()), before /t, d /, for which, however, (n) is sucient; below,
we add ve more taxophones, (N, M, , M, ), for nasal heterorganic sequences):
('ma;no) /'mano/ mano ('ni;No) /'niNo/ nio (um'pRr:o) /um'peRr:o/ un perro (im'bjRno) /im'bjeRno/ invierno (iM'fjl) /in'fjel/ inel (&konten'dR) ((-te-)) /konten'deR/ contender (kon'to) ((-'-))i (-s-) ((-'s-))a /kon'ebto/ concepto (kon';Xo)i
(-s;xo) ((-'s-))a /kon'sexo/ consejo ('nr:a) /'onr:a/ honra ('a~Co) ((-nC-)) /'anco/ an>o ('kNuXe)i (-xe)a /'konJuxe/ cnyuge (uN'uke) /un'Junke/ un yune (&koNLe'BaR)i (-N-, -NJ-)a /konLe'baR/ conevar (u'w;Bo) /un'webo/ un huevo ('bako)
/'banko/ banco ('tgo) /'tengo/ tengo ('fRa,Xa)i (-xa)a /'fRanxa/ franja
e articulation of /N/ is palatal, as in Italian, but short (not self-geminant, as in
neutral Italian; even if, at times, it can geminate a little (NN), after a stressed V
which is then short): ('ba;No 'baNNo) /'baNo/ bao ( Italian ('baN:No) /'baNNo/ bagno]
As for /nw/, even with the prexes cons-, ins-, trans-, the most normal and suitable articulation is with (): (&kotRu'jn)i (-stRu'sj-)a /konstRug'jon/ construccin (i'tante)i (-s-)a /ins'tante/ instante (&tRafoR'maR)i (-s-)a /tRansfoR'maR/ transformar obviously forms such as (&iRku'tanja)i (&siRkus'tansja)a /iRkuns'tanja/ circunstancia are included in this case.
For the rst elements of /mn, nm/ sequences, we nd some further taxophones,
6. spanish
235
with double articulations ((N), alveolarbilabial, in a more precise pronunciation

style, whereas a performance corresponing to the phonemic structure would sound
extremely pedantic, "&), or with coarticulations ((M, ), respectively alveolarized
bilabial and labialized alveolar) or with semi-articulations ((M, )), in a more spontaneous and less controlled kind of pronunciation ( 1.9.1-2): ('iNno, 'iMno, 'iMno
'imno) /'imno/ himno (koN'mi;o, koM-, ko-, ko- kon-) /kon'migo/ conmigo
For /n/, neutral pronunciation gives (n), even if the velar realization, which remains non-neutral, is very common, above all in America; for word-nal written
-m, we can also nd (N, M, M, , ) (the labialized taxophones are due to the spelling), even if (n) is neutral and considered more traditional (probably because of
the spelling): (a'jn)i (-sj-)a /ag'jon/ accin ('alBun, -uN, -uM, -uM, -u, -u)
/'albun/ lbum It is important to note that before, a pause, the vibrations of the
vocal folds stop at the same time as the oset of the tip of the tongue from the alveolar ridge (therefore, something like (-n, -n, -nO) is not at all acceptable).
Stops
6.2.2.1. Spanish has three diphonic pairs of stops, (p, b t, d k, g) /p, b t, d k,
g/: ('pa;o)i (-so)a /'paso/ po ('b;o)i (-so)a /'beso/ bo (um'b;o)i (-so)a /um'beso/ un bo (tu'to) /tu'teo/ tuteo (do'lR) /o'loR/ dolor (&kondo'lR) /kondo'loR/
con dolor ('kaldo) /'kaldo/ caldo ('k;Ce) /'koce/ co>e (go'Ri;la) /go'Rila/ gorila
(&ugo'Ri;la) /ungo'Rila/ un gorila
However, the voiced stops, are realized as such only after a pause, after a nasal,
and in the (homorganic) sequence /ld/, as the previous examples demonstrate. As
a matter of fact, in all other contexts, the normal realization is approximant, (B,
) /b, d/, or constrictive, () /g/ (unless one speaks slowly, with precision or emphasis): ('l;Bo) /'lobo/ lobo (e&ta'Bjn)i (es-)a /es'ta 'bjen/ t bien ('bRBo) /'beRbo/
verbo ('alBa) /'alba/ alba ('p;BRe) /'pobRe/ pobre (&ao'lu;to)i (-s-)a /abso'luto/ absoluto (u'Bli;me)i (s-)a /su'blime/ sublime (&uB-le'BaR)i (&s-)a /sub-le'baR/ sublevar
(in this example, the prex still sounds as if it is separate), (o'taR) /ob'taR/ optar
(both b and v are always /b/ (b, B): Spanish does not have /v/).
More examples: ('t;o) /'too/ todo (peR'R) /peR'eR/ perder ('ma;Re) /ma'Re/
madre ('de)i (-z-)a /'ese/ dde (lo';o)i (loz';os)a /los'eos/ los dedos
(&ami'RaR) /ami'RaR/ admirar (a'mfeRa)i (-s-)a /a'mosfeRa/ atmsfera (&aXe'ti;Bo)i (&axe-)a /axe'tibo/ adjetivo ('i;o) /'igo/ higo (la'Rr:a) /la'geRr:a/ la %erra
('a;wa) /'agwa/ a%a (el'lpe) /el'golpe/ el golpe ('kaRo) /'kaRgo/ cargo ('i;lo)i
('s-)a /'si-glo/ siglo ('dino, 'dino) /'ig-no/ digno ('tnika, -n-) /'tegnika/ tcnica
(for /gn, kn/, (-n-) is also possible), (a'tR) /ag'toR/ actor (&dijo'a;Rjo)i (-s-)a
/igjo'naRjo/ diccionario
Before a front V and /j/, /k, g, g/ realize as prevelar, by normal assimilation, but
it is not necessary to systematically use the special symbols ((, , )): (ki'taR) /ki'tar/ itar ('kj;Ro) /'kjeRo/ iero ('gia) /'gia/ %a ('a;ila) /'agila/ %ila In a true
intervocalic position, we currently have a semi-constrictive articulation, ((y)) (and
((g))): ('a;o) (('a;yo)) /'ago/ hago (and ('a;ila) ((-gi-)) /'agila/ %ila]
236
6.2.2.2. In word-nal position, () // is weak ( articulated with less tension,

((d)), as well as devoiced, ((D)), unless it is followed by voiced phones): (')i ('s-)a
/'se/ sed (a'Bla) /a'bla/ hablad often it drops in: (u't[])i (us-)a /us'te[]/ ted
(ma'Ri[]) /ma'Ri[]/ Madrid, and in nouns with // (not monosyllabic nouns):
(beR'a[]) /beR'a/ verdad (biR'tu[]) /biR'tu/ virtud
Furthermore, in the masculine ending -ado(s) the articulation is just as attenuated (up to (`), zero, in familiar pronunciation, above all the Iberian one, but not
systematically; generally, in American pronunciation, the drop is considered to be
non-neutral): (Le'a[;]o)i (,e'a;o)a /Le'gao/ egado (ol'da[;]o)i (sol'da;os)a
/sol'daos/ soldados
As seen, before a voiceless C (or before a possible pause), /b, , g/ are devoiced:
(o'taR, &aXe'ti;Bo, a'tR) (from the previous section). Due to an excessive inuence of writing, above all in the American accent, some articulate /b, , g/ as (p,
b t, d k, g) (with voicing in relation to the spelling), before heterosyllabic C
(which, instead, represent a completely normal neutralization): (o'taR, op-),
(&uB-le'BaR, -b-)i (&s-)a, (&ami'RaR, &ad-), (a'mfeRa, at-)i (-s-)a, (&aXe'ti;Bo, &ad-)i
(-xe-)a ('dino, 'dig-), (a'tR, ak-).
Again, for the same reason, with a further (and more serious) removal from the
real phonic structure, due to improper spelling inuence (or because of regional
accents, from Valencia, (v), and from Paraguay, (V)), especially in American pronunciation, some introduce the /v/ phoneme in Spanish, which has not existed
for centuries: (&embi'aR &eMvi'aR) /embi'aR/ enviar
It is evident that the use we make of /b, , g/ is not diaphonemic, as the two accents substantially coincide; it is moreover, interphonemic, because it is supposed
to show where their realization is not a stop, to help foreigners use it correctly,
without strained deductions (and, often, incorrect and therefore, the source of
endless problems).
Stopstrictives
6.2.3. ere is only one stopstrictive phoneme, voiceless postalveo-palatal, (C)
/c/ (which, compared to postalveo-palatal protruded (c) /c/, of English or neutral
Italian does not have labial protrusion: ('l;Ce) /'lece/ le>e (mu'Ca;Co) /mu'caco/
mu>a>o (&CaCa'Ca) /caca'ca/ >a->a-> e regional change from /c/ to () is
typically Andalusian and Caribbean.
Phonetically, there is another palatal stopstrictive, (), which is voiced, and realizes the constrictive phoneme /J/, which only occurs after a pause or after /n, l/:
([koN]'Rr:o) /[kon]'JeRr:o/ (con] hierro (&iN'jn)i (-sjn)a /inJeg'jon/ inyeccin ([eL]'uke) /[el]'Junke/ (el) yune Often, in familiar (and neutral) pronunciation, after pauses, we also nd (J) (real constrictive) or a semi-stopstrictive, (W),
or even a stop-semi-strictive, (): ('JRr:o, 'Juke 'W-, '-); the same can occur, in
the order (, , W, J), after /n, l/: (koN'Rr:o, koN'-, koN'W-, koN'J-), (&iN'jn,
&iN-, &iNW-, &iNJ-)i (-sjn)a, (eL'uke, eL'Ju-).
6. spanish
237
Constrictives
6.2.4.1. ere are ve constrictive phonemes; four are voiceless: (f) /f/, (i, sa)
//, (i, sa) /s/ and (Xi, xa) /x/; whereas (,) /J/ is voiced (palatal) and, if the truth be
told, only semi-constrictives, as it is half-way between an approximant, (j), and the
real constrictive ((J); but rarer in various languages).
(f) /f/ does not pose any problems, even if, often, American and Iberian speakers realize it as a bilabial (constrictive, (), or approximant, (F)): ([&uM]fa'BR
[&um]a- [&um]Fa-) /[un]fa'boR/ (un) favor
6.2.4.2. e others need further explanations. Indeed, () //i is neutral only
in the Iberian accent, whereas in the American accent it becomes /s/: (a'pa;to)i
(sa-)a /a'pato/ zapato ('j;lo)i ('sj-)a /'jelo/ cielo ('di;e)i (-se)a /'ie/ dice ('lu)i
('lus)a /'lu/ luz Starting from an American, or international type of transcription,
it could be more appropriate to use the diaphoneme /s/: /sa'pato, 'sjelo, 'ise, 'lus/.
Before voiced C the articulation becomes voiced: (Xu'aR)i (xuz'aR)a /xu'gaR/
juzgar ('lu o'Ra;a)i ('luz)a /'lu o'Raa/ luz dorada naturally before sonants, in
the American accent, (s) is preferred (as for /s/; the following section): ('djmo)i
(-smo)a /'jemo/ diezmo
6.2.4.3. For /s/ the place of articulation changes, from one accent to another,
as /s/ is (apico-)alveolar in Iberian Spanish, ()i, but, (lamino-)dental in American
Spanish, (s)a: (e'ta;o)i (es'ta;os)a /es'taos/ tados ('l)i ('s-)a /'sol/ sol ('pi;o)i
(-so)a /'piso/ po (After a stressed V as well as the normal and more appropriate
(';)i (';s)a, one can also have (')i ('ss)a: ('pio)i ('pisso)a.) Before voiced
diphonic heterosyllabic C (/b, d, g/), /s/ becomes voiced, ()i (z)a: (lo'BuRr:o)i
(loz'BuRr:os)a /los'buRr:os/ los burros ('de)i ('dze)a /'ese/ dde (di'uto)i
(-z'us-)a /is'gusto/ dgto In the /st, s/ sequences, in neutral Iberian pronunciation, (, ) remain, as can be seen in the given examples; only the denti-alveolar
articulation (dental with a raised tip, ((s, z))) is possible, while it is necessary in the
(still Iberian) pronunciation of the sequence /s/: (es';na) /es'ena/ cena (which
in American pronunciation is (e's;na) /e'sena/, from //es'sena//).
However, before heterosyllabic non-diphonic C (/m, n, N J w l, L/; for /s/ + /r:/,
see the end of this section), the behavior pattern is dierent, even if complementary: in the Iberian accent voicing prevails, (, , ); whereas in the American accent
voicelessness prevails, (s, , z) (even if we transcribe only the rst realization for
each accent): ('mimo)i (-s-)a /'mismo/ mmo ('ila)i (-s-)a /'isla/ la (lo'w;o)i
(-s'w;sos, -'sw-)a /los'wesos/ los huos ('d ',Rr:o)i ('d)a /'os 'JeRr:os/ dos hierros
e last example shows the normal articulation of the postalveo-palatal kind, ()i
()a, before /J/, as also before /L, N/: (la'Lu;Bja)i (la',u;Bjas)a /las'Lubjas/ l uvi
(lo'N;No)i (lo-, -os)a /los'NoNos/ los oos.
If the voiced C that follow are tautosyllabic ( when they are part of the same
syllable), /s/ remains voiceless, as even before V /s/ (being initial in the phono-syllable): ('jta)i ('sjs-)a /'sjesta/ sita ('w;lo)i ('s-)a /'swelo/ suelo (&loa'mi;o)i
(-sa'mi;os)a /losa'migos/ los amigos (&mieR'ma;no)i (-seR'ma;nos)a /miseR'manos/
238
m hermanos (In the Catalan pronunciation of Spanish, instead, it is voiced, as

in the Catalan language: (&loa'mi;o, &mieR'ma;no).)
For /sr:/ (also /sr:/) the most normal articulations are (r:, Rr:, r:, r:, :, R:)
(however, we only transcribe (r:), but /sr:/, even if its actual pronunciation with
(i, sa) is decidedly excessive, as if spoken by a foreigner {both non-Hispanic and
Hispanic}): (la'r:w;a)i (-s)a /las'r:weas/ l rued (&ir:a'l) /isr:a'el/ Israel
In the phonemic transcription, we use /s/, since, when speaking in a slow or emphatic manner, for (i, za), the actual pronunciation can undoubtedly be voiceless
even in Iberian pronunciation.
6.2.4.4. e criterion adopted here, only considers two kinds of neutral pronunciation; however, while describing the pronunciations of each single nation,
we will inevitably broaden our criteria so as to adequately include the characteristics of every single country (though, with all the non-neutral variants, which
obviously each area presents). erefore, let us mention a non-neutral variant, for
(American, Canary, and Andalusian Spanish) /s/, a very weak laryngeal approximant ((h), voiceless; and (H), voiced before voiced phones), which is generally
found together with what is traditionally (but incorrectly), dened as aspirate(d)
/s/ (which, from a strictly phonetic point of view would mean (sh), as (th)), before a pause, or a C or even before a V It is realized as if it were /h/. For example: (lah'kwhtah) /las'kwestas/ l cut ('dHe) /'ese/ dde ('ihlah, 'iH-) /'islas/ l (&loha'mi;oh) /losa'migos/ los amigos (in neutral pronunciation: (la'kwta)i (las'kwstas)a, ('de)i (-z-)a, ('ila)i ('islas)a, (&loa'mi;o)i (-sa'mi;os)a]
Such pronunciation can even get to transform (h, H) into (`), zero.
6.2.4.5. Again, referring to non-neutral pronunciation, there can also be colorings of (h) (voiceless), depending on the timbre of the preceding vowel. is
often happens, in Argentinian Spanish (where the phenomenon is normally heard;
and, only in more formal conversation, can one avoid it; but, before a pause, it is
considered very uneducated and, thus, more carefully avoided): ('lita) (palatal)
/'lista/ lta ('kwhta) (laryngeal) /'kwesta/ cuta ('pata) (velar) /'pasta/ pta
('tta) (laryngeal rounded) /'tosta/ tosta ('guWta) (velar rounded) /'gusta/ %sta
('iHla, 'dHe, 'aHma, 'Hmosis, tuH'jntes) /'isla, 'ese, 'asma, 'osmosis, tus'jentes/ la dde ma smos t dient In neutral pronunciation we have: ('lita,
'kwta, 'pata, 'tta, 'guta)i (-s-)a and ('ila, 'de, 'ama, 'moi, tu'jnte)i
('is-)a
In Argentinian neutral pronunciation, however, only rarely is /s/ fully (s, z), in
/s, s/ contexts, but it is rather a dental grooved semi-constrictive, (, d): ('dde,
lo'pRr:o) /'ese, los'peRr:os/ dde los perros. (e two neutral pronunciations
treated here, have (, )i (z, s)a. A less neutral but controlled Argentinian pronunciation has () for (, d) 1.9-18.)
Another, even more marked characteristic, which is typical, above all, of parts
of Andalusia and America (in particular, in the Caribbean and southern areas), /s/
fuse together into (=): (miz';os miH';oh mi+';oh mi'- mi'- -o)
/mis'eos/ m dedos (&tRez'Bailes &tReH'Baileh &tRe='Bai- &tRe'ai- &tRe'Fai-) /'tRes 'bai-
6. spanish
239
les/ tr bail (laz'B;tas laH'B;tah la'B- la'- la'F- -a) /las'botas/ l bot
(loz'a;,os loH'a;,oh lo'a- lo'xa- lo'a- -o) /los'gaJos/ los gaos (tuz'a;tos
tuH'a;toh tu'a- tu'xa- tu'a- -o) /tus'gatos/ t gatos ((=) and derivatives have
an intermediate type of phonation between (h) and (H)). Neutral pronunciation
is: (mi';o, &tRe'Baile, la'B;ta, lo'a;,o, tu'a;to)i (-z-, -s)a
Otherwise, before son(or)ants (/m, n l r:/), /s/ can change into (0, =0): ('mizmo, 'mi-, 'mis- 'miH-, 'mi+- 'mi-, 'mi)-) /'mismo/ mmo ('azno, 'a-, 'as- 'aH-,
'a- 'an- 'a-) /'asno/ no ('izla 'i-, 'is- 'iH- 'i+- 'i- 'ia-) /'isla/ la (&ir:a'l, &i-5) /isr:a'el/ Israel Neutral pronunciation: ('mimo, 'ano, 'ila)i (-s-)a (&ir:a'l)
6.2.4.6. e voiced palatal semi-constrictive, (,) /J/ (already introduced in
6.2.4.1; 6.2.4.3, as well), occurs between vowels, in words or sentences ( in contexts which are dierent from the strong ones in 6.2.3, but, as said there, it is
also possible in those contexts): ('ba;,a) /'baJa/ vaya (';,e) /'oJe/ oye
For /J/, there is a pronunciation which could come under the neutral one (from
a familiar to an energetic kind), which is very common in both accents: (). Articulatorily it corresponds to (C) /c/, which comes to form a diphonic pair, thus rendering the consonant system more natural and coherent. Although it is still not
completely neutral, it can be used in a kind of international accent, simpler and
more functional (even with (s, ,) /s, J/ for /, L/; therefore indicated by the diaphonemes /s, J/, 6.0); this pronunciation is justied and supported by the actual
use of many Iberian (including Madrilenian) and American speakers: ('ba;a, ';e).
According to the criteria adopted in this chapter, the transformation of /J/ to (,
) is doubtlessly regional (and typical, for example, of Argentinian pronunciation,
whose neutral local pronunciation has, however, () for /J, L/).
Again, in American and Andalusian pronunciations, above all, /J/ can often be
realized as an approximant, (j), but such pronunciation barely comes under neutral (however small the dierence may be, since (,) is only a semi-constrictive); a
systematic use of (j) is regional or foreign.
6.2.4.7. For /x/ too, the place of articulation changes from one accent to another (even within the neutral accent), as /x/ is, respectively, uvular, (X)i, and velar,
(x)a: (Xa'mn)i (xa-)a /xa'mon/ jamn (Xe'miR)i (xe-)a /xe'miR/ gemir ('di;Xe)i (-xe)a
/'ixe/ dije when word-nal it is weaker, as it becomes an approximant of the same
place of articulation, or even laryngeal: ('b, -h)i ('b, -h)a /'box/ boj it is currently lost in: (r:e'l[], -[h])i (r:e'l[], -[h])a /r:e'lox/ reloj
In the two accents, there can be some variants for /x/, which are included in neutral pronunciation: respectively, a more vigorous articulation (()i, voiceless uvular constrictive trill), or less vigorous, (()a, voiceless velar approximant), which
we refer to here only. On the other hand, its transformation into (h) (laryngeal),
which is very common in America and Andalusia, cannot be considered neutral,
according to the criteria adopted here.
240
Approximants
6.2.5.1. e Spanish approximants are /j, w/: ('j;lo)i ('sj-)a /'jelo/ cielo ('w;Bo) /'webo/ huevo (&awe'kaR) /awe'kaR/ ahuecar devoiced realizations after voiceless C are dialectal (or foreign): ('tj;ne 't-) /'tjene/ tiene ('kwa;tRo 'k-) /'kwatRo/
cuatro Generally, /j/ only occurs after a tautosyllabic C as in the previous examples, and in ('r:j;o) /'r:jego/ riego. Instead, at the beginning of a syllable, we have
/J/ exclusively ( 6.2.4.6, 6.2.4.1, 6.2.3), except in certain areas of America, such as
Argentina, where it is found for hiV-: (';lo, ',-)i ('-, ',- 'j-)a /'Jelo/ hielo ('RBa,
',-)i ('-, ',- 'j-)a /'JeRba/ hierba against ('RBa, ',-)i ('-, ',-)a /'JeRba/ yerba (independently of the non-neutral reduction of /J/ to (j), in certain areas).
For /[]bw, []gw/, in familiar pronunciation, there can be a simplication,
through a velarized bilabial approximant, (), for the rst case; or a constrictive,
()) (or (), 9.14 of NPT/HPh), or a semi-constrictive, (m), up to the approximant (w) (both velar rounded): (a'Bw;lo, a'm-, a'-, a'w-) /a'bwelo/ abuelo (la'Bwlta, la'm-, la'-, la'w-) /la'bwelta/ la vuelta ('a;wa, 'a;)a, 'a;ma, 'a;wa) /'agwa/
a%a (la&wape't;na, la&)a-, la&ma-, la&wa-) /lagwape'tona/ la %apetona
Instead, for /w, w/, in familiar pronunciation, a more vigorous realization is
more frequent (compared to the phonemic transcription): (');Bo, 'm-, '-) /'webo/, (&a)e'kaR, &ame-, &ae-) /awe'kaR/. Even in the /nw/ combination ( 6.2.1.1),
we can have this more vigorous articulation: (u'w;Bo, u')-, u'm-, u'-) /un'webo/ un huevo In all these cases, we can hear even syntagmatically divided realizations, (w, Bw) (and even, (gw, mbw), for the last case), however it is doubtlessly better to avoid them.
6.2.5.2. As well as (j, w) and (B, ) (/j, w/, /b, /), there are three more approximant (taxo)phones, which, in normal, not at all slovenly spontaneous conversation, are the realization of /e, o, a/, in the sequences /0e, 0o, 0a/; therefore we
have, (semi-palatal) /e/ (), (semi-velar rounded) /o/ (j) and (semi-prevelar) /a/ (F):
('pR) /pe'oR/ peor ('ta;tRo) /te'atRo/ teatro ('pj;ta) /po'eta/ poeta (&kjau'laR)
/koagu'laR/ coa%lar (u'nF;Xa)i (-xa)a /una'oxa/ una hoja ('lF;tRa pa'Ra;a) /la'otRa
pa'Raa/ la otra parada In slower, or more solemn speech, we doubtlessly have
(pe'R, te'a;tRo, po';ta, &koau'laR, &una';Xa, la';tRa pa'Ra;a)i (-xa)a, as the phonemic transcription indicates.
Furthermore, there are pronunciations which are currently considered to be uneducated: ('pjR, 'tja;tRo, 'pw;ta, &kwau'laR); there is a dierence, and it is more
than enough to distinguish the three dierent realizations, even if the dierence
between (, j) and (j, w) might seem negligible (not only for foreigners, but also
for natives who write articles and books, limiting themselves to only two extreme
possibilities, also because of the lack of adequate symbols).
6. spanish
241
Trills
6.2.6.1. Spanish has an alveolar trill /r:/ (r:), with three rapid tappings of the tip
of the tongue against the alveolar ridge (so, (r;) would be enough, but experience
prefers (r:)), and a tap /R/ (R), with a single rapid alveolar contact: ('r:a;Ro) /'r:aRo/ raro In interphonemic transcriptions (concerning several languages), it is important
to mark the chron(em)e, because otherwise its nature could be hidden, and Spanish /r:/ might seem to be a simple trill (with two tappings, as Italian /r/ in stressed
syllables: ('ra:Ro)it /'raro/ raro]
e two Spanish types ((R, r:)) join together too, giving, for examples: ('tjRr:a)
/'tjeRr:a/ tierra ( Italian ('tEr:Ra) /'tErra/ terra] (laR'r:a;jo) /laR'r:ajo/ la radio ( Italian (la'ra:djo)it /la'radjo/ la radio] Furthermore, in Spanish, the two types are distinctive, between V (even if with length dierences, for vowels too): ('ka;Ro) /'kaRo/
caro ('kaRr:o) /'kaRr:o/ carro ('p;Ro) /'peRo/ pero ('pRr:o) /'peRr:o/ perro (&ene'RaR)i
(-s-)a /ene'RaR/ encerar (&eneR'r:aR)i (-s-)a /eneR'r:aR/ encerrar
In our phone(ma)tic analysis, (Rr:) /Rr:/ are heterosyllabic sequences, constituted by (syllable-nal) (R) /R/ or (word-nal) (R) /R/ + (r:) /r:/, with one + three (or
four, to give more emphasis) tappings. ey are not simple segments such as /R,
r/ (or r r in the Hispanic tradition).
6.2.6.2. We now (systematically) examine the distribution of the two types. Apart from intervocalic context (just seen), with (Rr:) /Rr:/, where the use is phonemic, we also nd (r:) after a pause or after heterosyllabic C (/n, l, s/): (r:a'tn) /r:a'ton/ ratn ('nr:a) /'onr:a/ honra (un'r:a;mo) /un'r:amo/ un ramo (al&r:ee'R) /alr:ee'oR/ alrededor (el'r:i) /el'r:ei/ el rey (&ir:ae'li;ta) /isr:ae'lita/ raelita (mi'r:;pa)i (-s)a /mis'r:opas/ m rop
In the other contexts, (R) is normal, even before a pause, where the vibrations
of the vocal folds stop at the same time as the tip of the tongue is removed from
the alveolar ridge (therefore the following are not at all acceptable (-R, -R, -RO)):
('mi;Ra) /'miRa/ mira ('miRlo) /'miRlo/ mirlo ('Ren) /'oRen/ orden ('tRn) /'tRen/
tren (a'BRiR) /a'bRiR/ abrir (&ofRe'R)i (-sR)a /ofRe'eR/ ofrecer (peR'R) /peR'eR/
perder (&poRfa'BR) /poRfa'boR/ por favor ('iR po'Ra;wa) /'iR po'Ragwa/ ir por a%a
In familiar pronunciation, (R) /R/ can be weakened in all cases, substituting it
with (): ('mi;a, 'milo, 'en, 'tn, a'Bi, &ofe')i (-s)a, (pe', &pofa'B,
'i po'a;wa). On the contrary, in a more energetic or emphatic pronunciation,
/R, R/ can be strengthened into (r): ('mirlo, 'ren, a'BRir, &ofRe'r)i (-sr)a, (per'r, &porfa'Br, 'ir po'Ra;wa). However, neither of these two phones are indispensable for genuine pronunciation.
Laterals
6.2.7. ere are two lateral phonemes in neutral Iberian Spanish, (l, L)i /l, L/;
whilst in neutral American Spanish, the second merges with /J/ (l, ,)a /l, L/; moreover, /l/ has taxophones which are appropriately used, (, L) (it is not necessary to
242
mark explicitly ((l)), as long as the articulation is dental): ('la;o) /'lao/ lado (ka'lR) /ka'loR/ calor (kol'aR) /kol'gaR/ colgar ('alto) {((-lto))} /'alto/ alto (el'i;ne)i {((el'-))} (el'si;ne)a {((el's-))} /el'ine/ el cine (ko'Cn) ((-L'C-)) /kol'con/ col>n (eL'N;No) /el'NoNo/ el oo (eL'uke, eL'J-) /el'Junke/ el yunke ('kla;e)i (-se)a /'klase/
cle ('d;Ble) /'oble/ doble (e'la;wa) /e'lagwa/ el a%a (La'maR)i (,a-)a /La'maR/
amar ('ba;Le)i (-,e)a /'baLe/ vae (&koNLe'Bando)i (-N-, NJ-)a /konLe'bando/ conevando Phonetically, before palatals, even the American accent has (L), although
it does not have the phoneme /L/. In the same context, /J/ is articulated as a real
constrictive, (J), or as a stopstrictive, ().
Before a pause, even for /l/ (as for /n, R/ and for //), the vibrations of the vocal
folds stop at the same time as the tip of the tongue is removed from the alveolar
ridge (therefore, the following are not at all acceptable (-l, -l, -lO)): ('l)i ('s-)a
/'sol/ sol (lau'Rl) /lau'Rel/ laurel
Structures
6.3. We treat the characteristics of the combination of words in connected
speech, and then, above all, sentence-stress (as Spanish writing is quite explicit
about word-stress, though not exactly without doubts and uncertainties, including possible oscillations).
Taxophonics
6.3.1.1. th regard to consonant assimilations, they have been dealt with in
the sections of the respective parts. e combination of vowels within words, and
between words in sentences, will be dealt with above all here.
Within a word, two same vowels tend to be reduced to only one, except in formal, slow or controlled speech: (&alBa'a;ka, al'Ba;ka) /alba'aka/ albahaca (aa'aR,
-'aR)i (-s-)a /aa'aR/ azahar (&akRee';Re, &akRe'-) /akRee'oRes/ acreedor (bee'mnja, be'-)i (-sja)a /bee'menja/ vehemencia (nii'lita, ni'-) /nii'lista/ nihilta (&alko'l, al'kl) /alko'ol/ alcohol (&koope'RaR, ko&o-, &kope-) /koope'RaR/ cooperar (&oolo'Xia, &olo-)i (&so-, -'xia)a /oolo'xia/ zoologa
However, certain words, that otherwise would not be easily recognized, are not
reduced:(kRe'nja)i (-sja)a /kRe'enja/ creencia (mo';o)i (-so)a /mo'oso/ mohoso
(lo'R) /lo'oR/ loor other words can be reduced except in an intoneme: (le'R, 'lR)
/le'eR/ leer ('le, 'l) /'lee/ lee (kRe', 'kR) /kRe'e/ cre (pae';mo, pa';-, pa';-) /pase'emos/ peemos
th dierent V we have: (ko&au'la;o, &kja-) /koagu'lao/ coa%lado (aR'Xnteo, -to)i (-x-)a /aR'xenteo/ argnteo (leo'paRo, lo-) /leo'paRo/ leopardo
6.3.1.2. On the contrary, in formal, slow or controlled speech, (j, w, j) can
change into the vocoids (i, e u, o): ('bja;Xe bi'a-)i (-xe)a /'bjaxe/ viaje ('wa;Be
u'a-)i ('swa-)a /'swabe/ suave ('bju;a bi'u-) /'bjua/ viuda ('r:wi;o r:u'i-) /'r:wio/
6. spanish
243
ruido (al'da;no &alde'a-) /alde'ano/ aldeano (lal'ta leal-) /leal'ta/ lealtad ('li;na
-nea) /'linea/ lnea ('tja;La to'a-)i (-,a)a /to'aLa/ toaa (&kjau'laR ko&a- &koa-) /koagu'laR/ coa%lar ('r:al r:e'al) /r:e'al/ real
It must be remembered that there is a dierence between normal pronunciation
(not slow): ('ta;tRo, &empo'RaR, 'pj;ta, &kjau'laR) {/te'atRo, empeo'RaR, po'eta, koagu'laR/ teatro empeorar poeta coa%lar] and uneducated pronunciation: ('tja;tRo,
&empjo'RaR, 'pw;ta, &kwau'laR), in addition to slow pronunciation: (te'a;tRo, &empeo'RaR, po';ta, &koau'laR, ko&a-).
Except in particular cases, for rhythmic reasons (as will soon be seen), the pronunciation with unauthorized diphthongs is decidedly uneducated (although it
is very widespread, especially in Latin America): (pa'i, 'pai)i (-s)a /pa'is/ pas (ma'i, 'mai)i (-s)a /ma'i/ maz (ba'ul, 'baul) /ba'ul/ bal e transformation from
/ae, ao/ to /ai, au/ is equally uneducated: ('kaen 'kain) /'kaen/ caen (tRae'Ran
tRai-) /tRae'Ran/ traern (bil'Bao -au -aU) /bil'bao/ Bilbao (&baka'lao -au -aU)
/baka'lao/ bacalao in cases such as (&e[]tRa&oRi'na;Rjo, e[]&tRaoR-)i (-s-)a /e[k]stRaoRi'naRjo/ extraordinario we also have the possibility of: (&e[]tRFoR-, e[]&tRFoR-)i
(-s-)a.
6.3.1.3. It is not easy to assess the vocalic (/i, u/) or consonantal value (/j, w/)
of i u in CiV CuV sequences; /i, u/ are denitely more probable near a stress and
at the end of a word: (r:e'fRio)i (-s-)a /r:es'fRio/ rfro (&r:efRi'aR)i (-s-)a /r:esfRi'aR/ rfriar but: (&eMfRja'mnto) /enfRja'mento/ enfriamento (and also: (r:e'fRja;o)i (-s-)a
/r:es'fRjao/ rfriado] (flu'tuo) /fluk'tuo/ ucto (&flutu'aR) /fluktu'aR/ uctuar
(and also: (&flutu';o)i (-so)a /fluktu'oso/ uctuoso as well as suntuoso virtuoso
which have only (-'tw;o)i (-so)a /-'twoso/) but: (&flutwa'jn)i (-'sj-)a /fluktwa'jon/ uctuacin
e innitives with /i, u/ (which are prevalently short, and their derivatives,
even when there is no longer the simple, original form) are: (re-, mal-)criar ar
rfriar enfriar liar (ex-)piar (-, -)triar %iar (d)viar enviar reenviar puar
ruar uctuar concluir excluir incluir ocluir recluir (a-, in-, re-)uir diluir
(re)huir ()muir (re)construir instruir tatuir (re)constituir dtituir instituir
rtituir stituir intuir For ui the pronunciation with /'wi/ is frequent, as well.
Other cases of /i, u/ appear for short or compound forms: (&aBi'n) /abi'on/ avin (bi';njo) /bi'enjo/ bienio (biu'ni;Boko) /biu'niboko/ biunvoco (tRi';njo) /tRi'enjo/ trienio (tRi'agulo) /tRi'angulo/ trin%lo (&djei';Co, &dj-)i (-si-)a /jei'oco/ diecio>o (&beinti';Co, &bi-) /beinti'oco/ veintio>o (gi'n) /gi'on/ %in (i'a;to) /i'ato/ hiato (pRi'R) /pRi'oR/ prior (kRi'ana)i (-sa)a /kRi'ana/ crianza (fi'ana)i
(-sa)a /fi'ana/ anza (fi'a;o) /fi'ao/ ado (fi'ambRe) /fi'ambRe/ ambre (pi'a;no)
/pi'ano/ piano but: (&tRjagu'laR) /tRjangu'laR/ trian%lar (kRja'tu;Ra) /kRja'tuRa/ criatura (fja'R) /fja'oR/ ador (koM'fjana)i (-sa)a /kon'fjana/ conanza (fjam'bR;Ra) /fjam'bReRa/ ambrera (pja'nita)i (-sta)a /pja'nista/ pianta Let us note,
logically: ('Cja;pa)i (-s)a /'cjapas/ iap (with no dropping of /j/ after /c/).
We also have: (u'i;a) /u'ia/ huida (as huir] (di'uRno, 'dju-) /i'uRno, 'ju-/ diurno (but: (dju'tuRno) /ju'tuRno/ diuturno] (Xu'i;jo, 'Xwi-)i (xu'i;sjo, 'xwi-)a /xu'ijo,
'xwi-/ juicio (but: (Xwi'j;o, 'Xw)i (xwi'sj;so, 'xws)a /xwi'joso, 'xwe/ juicioso juez] (tRi'uMfo) /tRi'unfo/ triunfo (tRiuM'fal, tRju-) /tRiun'fal, tRju-/ triunfal For
244
muy we normally have ('mwi) /'mwi/, but also ('mui) /'mui/ is frequent (even if it
is often considered as dialectal).
Preferably, words like the following have /'wi/, but pronunciation with /u'i/ is
common, too: circuito fortuito gratuito suizo cuta ruido ruin ruina arruino
However, in fast speech, forms with /u', i'/ easily become /'w, 'j/: (&detRu'iR,
de'tRwiR)i (-s-)a /estRu'iR/ dtruir (&kotRu'i;o, ko'tRwi;o)i (-s-)a /konstRu'io/
construido (fi'a;mo, 'fja;mo)i (-s)a /fi'amos/ amos (&atu'aR, a'twaR) /agtu'aR/ actuar (&kaRi'a~Co, ka'Rja~Co) /kaRi'anco/ carian>o (&beinti'u;no, bein'tju;no) /beinti'uno/ veintiuno
6.3.1.4. Between words clusters of V are more varied and more numerous, but
the criteria are the same. th identical V reduction is greatly favored: (laaR'r:;lo,
lFaR-, laR-) /laaR'r:eglo/ la arreglo (lee[]'pli;ko, le-, le-)i (-s-)a /lee[k]s'pliko/ le explico (&loolBi', &ljo-, &lo-) /loolbi'e/ lo olvid ('agulo o'tuo) /'angulo ob'tuso/ n%lo obto (&impla'ka;Ble e'k;no, &impla'ka; Ble-) /impla'kable en'kono/ implacable encono (la'pR;a 'a;e u'na~CoR r:e'mano, la'pR; 'a;e)i (-sa, -se, -so)a /la'pResa
'ae u'nancoR r:e'manso/ la pra hace un an>o remanso (e'laiRe 'entRa il'Bando, e'lai 'Ren-)i (sil-)a /e'laiRe 'entRa sil'bando/ el aire entra silbando (&ma'B2 'kwa;tRo ';Xo2 ke', &kwa'tR;-)i (&maz-, -xos, -s)a /mas'ben 'kwatRo 'oxos ke'os/ ms ven cuatro ojos e dos
If the V are dierent, we have: (loaR'r:;lo, ljaR-) /loaR'r:eglo/ lo arreglo (la&oeR'B, &lFo-)i (-s-)a /laobseR'be/ la observ (loe[]'pli;ko, lje-)i (-s-)a /loe[k]s'pliko/ lo
explico (lae[]'pli;ko, lFe-)i (-s-)a /lae[k]s'pliko/ la explico (loim'pli;ko, ljim-)
/loim'pliko/ lo implico (laim'pli;ko, lFim-) /laim'pliko/ la implico ('t;o a'k;Lo,
't; ja-)i (-,o)a /'too a'keLo/ todo aeo ('tRite o'ka;o, 'tRi to-)i (-is-, -so)a /'tRiste
o'kaso/ trte oco (&entReilu'j;ne, &entRi-)i (-sj;nes)a /entReilu'sjones/ entre ilion ('pu;o auen'taRe, 'pu; jau-)i (-sen'taRse)a /'puo ausen'taRse/ pudo aentarse
(';Roe iNmoR'tal, ';Rje, i-, i-, '; Rji-) /'eRoe inmoR'tal/ hroe inmortal (pa'la;jo
au'uto, pa'la; jjau-)i (-sjo, -sto)a /pa'lajo au'gusto/ palacio au%sto (e'un e&ano'ta;o, ano-)i (s-, s-)a /se'gun seano'tao/ segn se ha notado (de'ambo 'm;o,
'dam-)i (-s)a /e'ambos 'moos/ de ambos modos
Obviously, the function of the vocalic elements that come into contact also
count. In fast speech, a grammeme-nal V seems rather redundant; thus, it can be
dropped, even where its morphological function may seem important (however,
there are all the other elements which compensate adequately).
us, one could quite easily achieve even zero, even with dierent V\ (la&aBRi'R, &lFa-, &laB-) /laabRi'Re/ la abrir (lo';jo, 'lj;-, 'l;-) /lo'ojo/ lo odio (mi'i;Xo,
'mi;-, 'mi;-)i (-xo)a /mi'ixo/ mi hijo (lae'p;a, lFe-, le-)i (-s'p;sa)a /laes'posa/ la
posa (la';Ra, 'lF;-, 'l;-) /la'oRa/ la hora (lai'Xi;ta, lFi-, li-)i (-x-)a /lai'xita/ la hijita
(lo'u;niko, 'lju;-, 'lu;-) /lo'uniko/ lo nico (lou'nj;Ron, lju-, lu-) /lou'njeRon/ lo unieron (me'i;Ba, 'mi;-, 'mi;-) /me'iba/ me iba (me&ima'Xi;no, &mei-, &mi-, &mi-)i (-x-)a
/meima'xino/ me imagino
6.3.1.5. Even clusters of various V are frequent in these examples (taken from
Navarro Toms, but retranscribed and completed): (&eka'l;Ra aR'r:i;Ba, -'l; RaR'r:i-)i
6. spanish
245
(&eska-)a /eska'leRa aR'r:iba/ calera arriba (a'ma;a e'p;a, a'ma; Fe-, e-)i (-s-)a
/a'maa es'posa/ amada posa (o'fRta i,'Xuta, o'fR tFi,-, -ti,-)i (-'x-)a /o'feRta in'xusta/ oferta injta (pa'la;BRa o'a;a, pa'la; BRFo-, BRo-)i (-s-)a /pa'labRa o'saa/ palabra osada ('ka;a u'milde, 'ka; Fu-, u-)i (-s-)a /'kasa u'milde/ ca humilde ('kj;Re a'BlaR, 'kj; Ra-, -Ra-) /'kjeRe a'blaR/ iere hablar ('pw;e ekRi'BiR, 'pw; e-, e-)i
(-s-)a /'pwee eskRi'biR/ puede cribir
Also: ('nmbRe i'lutRe, 'nm bRi-, bRi-) /'nombRe i'lustRe/ nombre iltre ('tj;ne
oR'u;Lo, 'tj; no-, no-)i (-,o)a /'tjene or'guLo/ tiene or%o ('Xnte u'milde, 'Xn tu-
tu-)i ('x-)a /'xente u'milde/ gente humilde ('ka;i apa'a;o, 'ka; ja-)i (-si, sj-)a /'kasi apa'gao/ ci apagado (mi&epe'Rana, &mie-, &mje-)i (-es-, -sa)a /miespe'Rana/ mi
peranza ('ka;i impo'i;Ble, 'ka; im-)i (-si, si-)a /'kasi impo'sible/ ci imposible
(mio&Blia'jn, mjo-)i (-'sj-)a /miobliga'jon/ mi obligacin (&niuna'B, ni&u-,
&nju-)i (-s)a /niuna'be/ ni una vez
And also: ('gRi;to a'u;o, 'gRi; tja-) /'gRito a'guo/ grito a%do ('p;ko e'fwRo,
'p; kje- ke-)i (es-, -so)a /'poko es'fweRo/ poco fuerzo ('n;Ro iM'fjRno, 'n;
RjiM-, RiM-) /'negRo in'fjeRno/ negro inerno ('kwaRto o'ku;Ro, 'kwaR tjo-, to-)i
(-s-)a /'kwarto os'kuRo/ cuarto oscuro (e'ga;No u'ma;no, e'ga; Nju-, Nu-) /en'gaNo
u'mano/ engao humano (&uami'ta, &wa-)i (&s-, &s-)a /suamis'ta/ su amtad ('impetu &epan't;o, -&tu e-, -pe &twe-)i (-es-, -so)a /'impetu espan'toso/ mpetu pantoso ('tRi;Bu i'gRa;ta, 'tRi; Bwi-) /'tRibu in'gRata/ tribu ingrata (&poRuo'nR, -wo-)i
(-s-)a /poRsuo'noR/ por su honor (e'pi;Ritu u'ma;no, -Ri tu-)i (-s-)a /es'piRitu u'mano/
pritu humano ('L;a aao'RaR, 'L;a ao-, 'L; ao-)i (',-)a /'Lega aao'RaR/ ega
a adorar ('i;Ba aenen'dR, 'i;Ba en-)i (-s-)a /'iba aenen'deR/ iba a encender
Further examples still: (be'gana ai'Ra;a, -an Fai-, -an ai-)i (-sa, s-)a /ben'gana ai'Raa/ venganza airada (e'ta;Ba ao'a;a, -e'ta; BFao-, Bao-)i (es-)a /es'taba ao'gaa/ taba ahogada (r:o'a;a au'R;Ra, r:o'a; Fau-, au-)i (-s-)a /r:o'saa au'RoRa/
rosada aurora ('kulta eu'R;pa, 'kul tFeu-, teu-) /'kulta eu'Ropa/ culta Europa (a'pRnde aa'BlaR, a'pRn daa-, da-) /a'pRende aa'blaR/ aprende a hablar ('auRea e'pa;a, 'auRa, &au RFe'pa;a)i (-s-)a /'auRea es'paa/ urea pada ('mwRte ai'Ra;a, 'mwR tai-,
tai-) /'mweRte ai'Raa/ muerte airada (pRe'u;me aon'daR, pRe'u; maon-)i (-s-)a /pRe'sume aon'daR/ prume ahondar ('fRnte au'uta, 'fRn tau-, tau-)i (-s-)a /'fRente au'gusta/ frente au%sta (biR'Xi;neo e'kanto, -njo, -no)i (-x-)a /biR'xineo en'kanto/
virgneo encanto
6.3.1.6. Examples of vowel clusters in sentences continue: (no'ti;ja a'l;Re, no'ti; ja'l-)i (-s-)a /no'tija a'legRe/ noticia alegre ('r:Xja e'tiRpe, 'r:; XjFe-, Xje-)i (-x-,
-s-)a /'r:exja es'tiRpe/ regia tirpe ('gl;Rja iNmoR'tal, i-, i-, 'gl; RFi-) /'gloRja inmoR'tal/ gloria inmortal (e'tanja o'kulta, e'tan jFo-, jo-)i (es-, -sja)a /es'tanja o'kulta/ tancia oculta (Xu'ti;ja u'ma;na, Xu'ti; jFu-, ju-)i (xus'ti;s-)a /xus'tija u'mana/ jticia humana ('na;je a'ku;e, 'na; ja-) /'naje a'kue/ nadie acude ('X;njo
a'tu;to, 'X; njja-)i ('x-, -s-)a /'xenjo as'tuto/ genio tuto (i'lnjo elo'kwnte, i'ln jje-)i (si-, -sjo)a /si'lenjo elo'kwente/ silencio elocuente (';jo i'nu;til, ';
jji-)i (-sjo)a /'ojo i'nutil/ ocio intil
Also: ('n;jo oR'u;Lo, 'n; jjoR-, 'joR-)i (-sj-, -,o)a /'nejo oR'guLo/ necio or%o
('i;tjo um'bR;o, 'i; tjjum'-)i ('si-, -so)a /'sitjo um'bRoso/ sitio umbroso ('bwlBo aa-
246
'taRlo, -Bo a't-, 'bwl Bja't) /'bwelbo aa'taRlo/ vuelvo a atarlo ('bgo aempe'aR,
'b gjaem-, gjFem-, gjem-)i (-saR)a /'bengo aempe'aR/ vengo a empezar ('kwRpo
ai'R;o, 'kwR pjFi-, pjai-) /'kweRpo ai'Roso/ cuerpo airoso (di'pwto ao&Bee'R,
di'pw tjFo-, tjao-)i (-s'pws-, -sR)a /is'pwesto aobee'eR/ dputo a obedecer
('fauto au'pi;jo, 'fau tjFu-, tjau-)i (-st-, -s'pi;sjo)a /'fausto aus'pijo/ fato apicio
And also: (an'ti;wa alti'B, an'ti; wal-)i (-s)a /an'tigwa alti'be/ anti%a altivez ('lgwa e[]'tRa;Na, 'l gwFe-)i (-s-)a /'lengwa e[k]s'tRaNa/ len%a extraa (e'ta;twa iN'm;Bil, i-, i-, e'ta; 'twFi-)i (es-)a /es'tatwa in'mobil/ tatua inmvil ('a;wa olo'R;a, 'a; wFo-) /'agwa olo'Rosa/ a%a olorosa ('fa;twa u&mani'a, 'fa; twFu-)
/'fatwa umani'a/ fatua humanidad ('mu;two a'mR, 'mu; twja-) /'mutwo a'moR/
mutuo amor ('aRwo em'p;No, 'aR wjem-) /'aRdwo em'peNo/ arduo empeo (peR'p;two im'p;Rjo, peR'p; twjim-) /peR'petwo im'peRjo/ perpetuo imperio (kon'ti;nwo e'l;Xjo, kon'ti; nwje-)i (-xjo)a /kon'tinwo e'loxjo/ continuo elogio
Lastly: ('fa;two oR'u;Lo, 'fa; twjoR-, twoR-)i (-,o)a /'fatwo oR'guLo/ fatuo or%o
('mntRwo u'ma;no, 'mn tRwju-, tRwu-)i (-s-)a /'monstRwo u'mano/ monstruo
humano ('r:;Xja au&toRi'a, 'r:; XjFu-, Xjau-)i (-x-)a /'r:exja autoRi'a/ regia autoridad (koR'r:j aepe'RaRlo, Fe-, e-)i (-s-)a /koR'r:jo aespe'RaRlo/ corri a perarlo (pa'la;jo au'uto, pa'la; jjau-, jjFu-)i (-sjo, -us-)a /pa'lajo au'gusto/ palacio au%sto (i'ni;kwo au'u;Rjo, i'ni; kwjau-, kwjFu-) /i'nikwo au'guRjo/ inicuo au%rio (em'bi;jo aeu';Bjo, em'bi; jjaeu-, jaeu-, jFeu-, jjeu-)i (-s-)a /em'bijo aeu'sebjo/
envidio a Eebio
6.3.1.7. e conjunctions y o (and their variants e u], between V in dierent
words (even if with /w/), are normally realized as (j, w) /j, w/ (but also with (,
j)): ('ka;a 'jwRta, 'w-)i (-sa)a /'kasa 'jwerta/ ca y huerta ('a;wa 'jaiRe, 'a-) /'agwa
'jaiRe/ a%a y aire (a'gRjnta 'ja~Ca, 'a-) /san'gRjenta 'janca/ sangrienta y an>a
(a'pa;a jen'jnde, e-)i (-sj-)a /a'paga jen'jende/ apaga y enciende ('ka;Le je'ku;Ce, e-)i (-,e jes-, e-)a /'kaLe jes'kuce/ cae y cu>e (o'BRi;na 'ji;Xa, 'i-)i (so-, -xa)a
/so'bRina 'jixa/ sobrina e hija (r:i'k;a jin'dutRja, in-)i (-sa, -st-)a /r:i'kea jin'dustRja/ rieza e indtria ('ta wa'k;La, ja-)i ('s-, -,a)a /'esta wa'keLa/ ta o aea ('blaka wa'ul, ja-)i (-sul)a /'blanka oa'ul/ blanca o azul ('j;te 'w;Co, 'j-)i
('sj-)a /'sjete u'oco/ siete u o>o ('u ;no 'w;tRo, 'j-) /'uno u'otRo/ uno u otro
e real language ( the spoken language) shows how grammar futilely complicates things, given that the normal pronunciation of the four written forms [y e
o u] is the same (for e there is (), too). Naturally, schools have then reinforced the
need for distinction, imposing it on pronunciation too; thus slowing down emission, or paying attention (to spelling), or for the sake of clarity, /i, e o, u/ can
come up again.
6.3.1.8. Usually (except in slow speech), a nal C followed by an initial V in
words without breaks, resyllabies: (e'l;Ro) /e'loRo/ el oro (u'nmbRe) /u'nombRe/
un hombre ('muCo o'n;Re)i (so-, -es)a /'muco so'noRes/ mu>os honor ('lu ama'Ri;La)i (sa-, -,a)a /'lu ama'RiLa/ luz amaria
Hence, there is no dierence between: helado and el hado (e'la;o) /e'lao/, elegi-
6. spanish
247
do and el ejido (&ele'Xi;o)i (-x-)a /ele'xio/, hele>o and el he>o (e'l;Co) /e'leco/, heleno and el heno (e'l;no) /e'leno/, la sab and l av (la'a;Be)i (la'sa;Bes)a /la'sabes/, ena% and en a% (e'na;wa)i (-s)a /e'nagwas/, enojo and en ojo (e'n;Xo)i
(-xo)a /e'noxo/
In emphatic or vigorous pronunciation, or in singing, after stressed V the continuous non-diphonic C (voiced: /m, n, N l, L/, and voiceless: /f, , s, x/ but not
/R, Rr:/, for which length is essential), instead of the normal structure, with (';0),
often have ('00): ('u ;mo, "ummo) /'umo/ humo ('a;na, "anna) /'ana/ Ana ('ba;le,
"balle) /'bale/ vale ('X;fe, "Xffe)i (x-)a /'xefe/ jefe ('di;e, "die)i (-se, -sse)a /'ie/
dice ('ka;a, "kaa)i (-sa, -ssa)a /'kasa/ ca ('i;Xo, "iXXo)i (-xo, -xxo)a /'ixo/ hijo
6.3.1.9. Phonetically, two same C are realized as a contained gemination, /00/
(00): (i&nnume'Ra;Ble) /innume'Rable/ innumerable (&inne&ei'a)i (&sinne&se-)a /sinneesi'a/ sin necidad ('BBjo) /'obbjo/ obvio (&uBBen'jn)i (&su-, -'sj-)a /subben'jon/ subvencin (e'a i'C;a)i (-sa)a /e'a i'cosa/ edad di>osa (&XuBen'tu
o'Ra;a)i (xu-)a /xuben'tu o'Raa/ juventud dorada (koR'l li'X;Ro)i (-'sl li'x-)a
/koR'el li'xeRo/ corcel ligero (el'l;Bo) /el'lobo/ el lobo ('lu eni'tal)i (-s s-)a /'lu eni'tal/ luz cenital ('dj i'aRr:o)i (-s s-, -os)a /'je i'gaRr:os/ diez cigarros (&lo e'N;Re)i (-s s-, -es)a /losse'NoRes/ los seor ('d o'BRi;no)i (-s s-, -os)a /'os so'bRinos/ dos sobrinos
e dierence between (0) and (00) is more than enough (and what is necessary) to distinguish cases such as: (&uno'Bi;Lo)i (-,o)a /uno'biLo/ un ovio and (&unno'Bi;Lo)i (-,o)a /unno'biLo/ un novio (o'nmbRe)i (s-, -s)a /so'nombRes/ son hombr and (on'nmbRe)i (s-, -s)a /son'nombRes/ son nombr (a&k'l;Ro) /a'ke 'loRo/
ael oro and (a&kl'l;Ro) /a'kel 'loRo/ ael loro ('ma ';BRan)i ('s-)a /'ma 'sobRan/
ms obran and ('ma ';BRan)i (-s 's-)a /'mas 'sobRan/ ms sobran
e prex /sub/ sub- before /r:/, is obviously heterosyllabic: (&uB-r:a',aR) /subr:a'JaR/ subrayar the same occurs, with other C when the ax is clearly perceived:
(&uB-limi'nal)i (&s-)a /sub-limi'nal/ subliminal
For -dl- the syllabication is /-l/: (mi'Ra-lo) /mi'Ralo/ miradlo for -tl- /t-l/i
( /-l/i) and /-tl/a prevail: ('ala)i ('a;tlas)a /'alas, 'atlas/ atl (a'l;ta)i (a'tl;ta)a
/a'leta, a'tle-/ atleta
6.3.1.10. As has already been mentioned, spontaneous and normal pronunciation for heterosyllabic (voiced or voiceless) stops has non-stopped realizations:
voiced before voiced C and half-voiced before voiceless ones: ('ato) /'abto, -pt-/
apto (do'tR) /og'toR, -k't-/ doctor (&ote'nR) /obte'neR/ obtener (&aki'RiR) /aki'RiR/ adirir (&aBi'kaR) /abi'kaR/ abdicar (&aBeR'tiR) /abeR'tiR/ advertir (&ekli'aR)i (-s-)a /eklib'saR, -p's-/ eclipsar (&kone'jn)i (-se'sjn)a /koneb'jon, -p'-/
concepcin (&uB-,u'aR)i (s-)a /subJu'gaR/ subyugar (&a-,a'nte)i (-s-)a /aJa'ente/
adyacente (a'jn)i (-s-)a /ag'jon, ak-/ accin (e'a;men)i (-s-)a /eg'samen, ek-/ examen (aB&nea'jn)i (-sjn)a /abnega'jon/ abnegacin (a'mfeRa)i (-s-)a /a'mosfeRa, at'm-/ atmsfera (&ami'Ra;Ble) /ami'Rable/ admirable ('tnika, -n-)
/'tegnika, -kn-/ tcnica ('dino, -no) /'igno/ digno (&uBma'Ri;no) /subma'Rino/
submarino (fully voiced or, respectively, voiceless) stop realizations are typical of
248
emphasis attention (to spelling).

Lastly, we see some cases of complex consonant clusters: (o'ta;kulo)i (-s-)a /obs'takulo/ obstculo (a'kRi;to)i (-s-)a /as'kRito/ adscrito (ko'tante)i (-s-)a /kons'tante/
constante
Stress
6.3.2.1. Spanish spelling is quite satisfactory regarding the indication of word
stress; however, some doubts remain concerning clusters of graphic VV above all
with i u furthermore, there are possible oscillations, and dierences with respect
to Portuguese or Italian.
e position of stress is distinctive (and we can also nd threefold phonemic
contrasts, the rst of which are nouns, indicated graphically too), as in: ('a;nimo)
/'animo/ nimo (a'ni;mo) /a'nimo/ animo and (&ani'm) /ani'mo/ anim; (kon'ti;nwo) /kon'tinwo/ continuo (&konti'nuo) /konti'nuo/ contino and (&konti'nw)
/konti'nwo/ continu; ('li;mite) /'limite/ lmite (li'mi;te) /li'mite/ limite (&limi't)
/limi'te/ limit (de'p;ito)i (-s-)a /e'posito/ depsito (&depo'i;to)i (-s-)a /epo'sito/
deposito (de&poi't)i (-s-)a /eposi'to/ deposit
For stress oscillations, some examples follow: acn acne aerstato aerostato
ambrosa ambrosia atmsfera atmosfera atriaco atraco balatre balastre
cartomancia cartomanca (but only farmacia] cclope ciclope conclave cnclave
dinamo dnamo elixir elxir etope etiope (but only miope] gladolo gladiolo
bero ibero mil msil olimpiada olimpada orga orgia smos osmos pelcano pelicano pensil pnsil pentagrama pentgrama perodo periodo policromo polcromo polgloto poligloto radar rdar reptil rptil reuma rema termostato termstato utopa utopia varic vric
6.3.2.2. Lexical compounds and adverbs ending in -mente keep quite a strong
stress even on the rst element (according to the scale: ('), (), (&)): (kRta'Bla)i
(-sas)a /'koRta'bolsas/ cortabols ( (&koRta';Ra) /koRta'oRa/ cortadora), (ika'pj)
/'inka'pje/ hincapi (entRe')i (-s)a /'entRe'os/ entreds ( (&entRe' mu'Ca;Co)i
(-os)a /entRe'os mu'cacos/ entre dos mu>a>os), (r:e&al'mnte) /r:e'al'mente/ realmente (&beRaRa'mnte) /beRa'eRa'mente/ verdaderamente (fail'mnte)i (-s-)a
/'fail'mente/ fcilmente Also: (ta'Bia) /'toa'bia/ todava (a&i'mimo)i (a&si'mis-)a /a'si'mismo/ immo (as can be seen in the last example, if a rst /'/ is immediately followed by another /'/, it becomes (&); on the other hand, however, if the
form were with only one /'/, /asi'mismo/, we should have (&ai'mimo)i (&asi'mis-)a).
To begin to acknowledge the dierent stress patterns, compared for instance
to Italian, we give some useful examples: (nal stressed) acento amplo varo
(stressed on the last but one syllable) abdico altero animo celebro certico computo convoco denomino deposito dputo edico indico integro interrogo limito medito modico modulo penetro signico /signi'fiko/, termino venero violo
/'bjolo/, vito /bi'sito/, limpio /'limpjo/. Furthermore: (a'n;mja) /a'nemja/ anemia
(bi'a;mja) /bi'gamja/ bigamia (&kaRam'b;la) /kaRam'bola/ carambola (di'pu;ta)i
6. spanish
249
(-s-)a /is'puta/ dputa (e'l;na) /e'lena/ Elena (en&iklo'p;ja)i (-&si-)a /eniklo'peja/ enciclopedia (faR'ma;ja)i (-s-)a /faR'maja/ farmacia ('mj;pe) /'mjope/
miope (pa'R;ja) /pa'Roja/ parodia (te'Ra;pja) /te'Rapja/ terapia obviously, forms
such as (a'pndie)i (-se)a /a'pendie/ apndice are less risky, at least when reading.
Let us also note: (r:a'n)i (-s-)a /r:a'on/ razn (r:a';ne)i (-'s;nes)a /r:a'ones/
razon ('biRXen)i (-x-)a /'biRxen/ virgen ('biRXene)i (-xenes)a /'biRxenes/ vrgen
and lastly: (ka'RateR) /ka'RagteR, -kt-/ carcter (&kaRa't;Re)i (-s)a /kaRag'teRes, -k't-/
caracter ('r:;Ximen)i (-x-)a /'r:eximen/ rgimen (r:e'Xi;mene)i (-x-, -s)a /r:e'ximenes/ regmen
6.3.2.3. Even in Spanish, in connected speech, there are normally words which
are destressed; the articles are among these: (e'lmbRe) /e'lombRe/ el hombre (&elal'kalde) /elal'kalde/ el alcalde (la'ka;a)i (-sa)a /la'kasa/ la ca (la'Ci;ka)i (-s'Ci;kas)a
/las'cikas/ l >ic (&una'mi;o) /una'migo/ un amigo (&una'tRinta peR';na)i
(&unas-, -'s;nas)a /unas'tReinta peR'sonas/ un treinta person
We then have forms (plurisyllabic too), such as salvo excepto mediante durante and phrases such as rpecto a junto a encima de delante de\ (a'ma;laa) /a'malaga/ a Mlaga (ko'n;La)i (-,a)a /ko'neLa/ con ea (&empa'Ri)i (-s)a /empa'Ris/ en
Pars (&inom'bR;Ro)i (&sinsom-)a /sinsom'bReRo/ sin sombrero (&ante't;o) /ante'too/ ante todo (&baXolo'aRBole)i (-xolo's-, -es)a /baxolo'saRboles/ bajo los rbol
(&kontRa&mia'mi;o)i (-sa'mi;os)a /kontRamisa'migos/ contra m amigos (&dee'i)i (-z-)a /ese'oi/ dde hoy (&entRelo';Xo)i (-'s;xos)a /entRelo'soxos/ entre los ojos
Furthermore: (&ata'ki, -Fa-, -aa-)i (&as-)a /astaa'ki/ hta a (&ajael'pwRto,
-jFel-)i (-sj-)a /ajael'pweRto/ hacia el puerto (&paRa&nwetRo'wpee)i (-stRos'wspees)a /paRanwestRos'wespees/ para nutros husped (&oBRel';BRe)i (&soBRel's-)a
/sobReel'sobRe/ sobre el sobre (e&eto')i (-&seto's)a /egebto'os, ekepto-/
excepto dos (&Xuntoala'pwRta, -tja-)i (&xu-)a /xuntoala'pweRta/ junto a la puerta
Also: (en&ima&ela'ka;a)i (-si-, -sa)a /enimaela'kasa/ encima de la ca (de&lantee'mi) /elantee'mi/ delante de m Even in more or less long series: (&paRa&entReno';tRo, pa&RFen-)i (-'s;tRos)a /paRaentReno'sotRos/ para entre nosotros (&poRen&imae't;o)i (-'si-)a /poRenimae'too/ por encima de todo (&poRe&lante&elXaR'in)i (-x-)a /poRelanteelxaR'in/ por delante del jardn (&deepo&RentRelo'aRBole)i (&dez-, -'saRBoles)a /esepoRentRelo'saRboles/ dde por entre los rbol
Generally, conjunctions too are destressed (even in phrases, such as en cuanto
(e) puto e suputo e), except with ora ya bien (disjunctive), (consecutive), no obstante con todo fuera de (adversative), en efecto por tanto por consi%iente e (consecutive), apen an no no bien ya e luego e dpus
e en tanto e (temporal), a no ser e dado e con tal e (conditional), por
ms e a par de e mal e ya e (concessive), (&pwetoke&noloa'Bia)i (-sto-,
-sa-)a /pwestokenolosa'bia/ puto e no lo saba At the beginning of elliptic questions y is stressed: ('i tupa;Re21)i (-'pa;-)a /'i tu'paRe?/ Y tu padre?
6.3.2.4. It will not be superuous to observe that there is a dierence, in stress
too, between: (e&koe'Ca;o)i (-s-)a /ekose'cao/ he cose>ado and (';ko o'n;Ro) /'e-
250
ko so'noRo/ eco sonoro (&aLe'a;o)i (&a,-)a /aLe'gao/ ha egado and ('a;No 'ma;lo) /'aNo 'malo/ ao malo (&eto'ma;o)i (&es-)a /esto'mao/ tomado and ('to 'pi;o)i
('s-)a /'esto 'pio/ to pido (o&nami'Ra;o)i (s-, -s)a /sonami'Raos/ son admirados and (' naRmo'nj;o)i ('s-, -so)a /'so naRmo'njoso/ son armonioso
Object pronouns are destressed, la le lo l l los me nos os se te\ (me'pa;Ro
ami'RaR, me'pa; Rja-) /me'paRo ami'RaR/ me paro a mirar (o'Bi;mo Be'niR)i (oz'Bi;moz)a /os'bimos be'niR/ os vimos venir (e&lepeR'j)i (se&les-)a /selespeR'jo/ se l
perdi instead, subject pronouns and indirect pronouns are stressed:(' ')i ('s)a
/'Jo 'se/ yo s ('tu 'pw;e)i (-s)a /'tu 'pwees/ t pued ('l 'di;e)i (-se)a /'el 'die/ l
dice (poR'mi) /poR'mi/ por m (&paRa'ti) /paRa'ti/ para ti
Possessive adjectives are destressed: (&mieR'ma;no)i (-se-, -os)a /miseR'manos/ m
hermanos (tu';Xo)i (-'s;xos)a /tu'soxos/ t ojos (u'ma;Re)i (s-)a /su'maRe/ su
madre (&nwetRo'tjmpo)i (-s-)a /nwestRo'tjempo/ nutro tiempo
Generally demonstrative adjectives, are destressed (but, from a pragmatic point
of view, they can be considered potentially stressable): (&ete'li;BRo, -)i (-s-)a /este'libRo/ te libro (&ee'pRr:o, -) /ese'peRr:o/ e perro (a&keLamu'XR, ak-)i (-,amu'x-)a /akeLamu'xeR/ aea mujer
6.3.2.5. Relative pronouns [e ien{} cual{} cuyo{s}] are destressed (contrary to interrogative and exclamatory ones): (e'lmbRe ke'Bi;mo)i (-s)a /e'lombRe
ke'bimos/ el hombre e vimos (e&ku,o'ka;o)i (-so)a /enkuJo'kaso/ en cuyo co
cual cual (with the article) and tal tal are stressed: ('tgo uneR'ma;no2 el'kwaL
'L;a 'i2 ke)i (-L ',-)a /'tengo u'neRmano, el'kwal 'Lega 'oi/ tengo un hermano, el cual
ega hoy, e (le'i;Xo 'tal 'k;a)i (-xo, -sa)a /le'ixo 'tal 'kosa/ le dijo tal cosa
Relative adverbs [como cuando cuanto donde] are also destressed, contrary to
interrogative and exclamatory ones: (&kwandoe'laiRe e'kalma, -dje-) /kwandoe'laiRe se'kalma/ cuando el aire se calma (la'pla;a &onde&tatu'ka;a)i (-sa, -s&t-, -sa)a /la'plaa ondeestatu'kasa/ la plaza donde t tu ca
ereas the indenites [algo algn al%no al%ien nadie ningn nin%no otro] are stressed, cada generally is not: (&kaa'ia) /kaa'ia/ cada da Even tan is
not stressed, contrary to tanto tanta
en used as conjunctions, the following adverbs are destressed: luego mientr an (which becomes /'aun/), ms menos ci the adverb medio is too (contrary to its corresponding adjective).
6.3.2.6. e forms of tratamiento, don doa fray sor san santo santa are always destressed before a name: (do'kaRlo)i (-s)a /on'kaRlos/ don Carlos (&doNao'l;Re)i (-s)a /oNao'loRes/ doa Dolor (&antoo'migo)i (&s-)a /santoo'mingo/
santo Domingo vocative forms are destressed too, seor seora seorita padre madre hermano hermana to ta\ (e&NoRmaR'ti;ne)i (s-, -s)a /seNoRmaR'tine/ seor
Martnez (e&NoRama'Ria)i (s-)a /seNoRama'Ria/ seora Mara (&paRean'dR, -Ran-)i
(-s)a /paRean'dRes/ padre Andrs (tio'Xwan)i (-x-)a /tio'xwan/ to Juan however,
when not used in forms of tratamiento, we have the regular: (&una'anta mu'XR)i
(-s-, -x-)a /una'santa mu'xeR/ una santa mujer ('bi;no ele'NR maR'ti;ne)i (-s-, -s)a /'bino else'NoR maR'tine/ vino el seor Martnez (ae'kRi;to el'tio 'Xwan)i (aes-, 'x-)a /aes-
6. spanish
251
'kRito el'tio 'xwan/ ha crito el to Juan

Even in vocative phrases, there is destressing: (bwe'nmbRe) /bwe'nombRe/ buen
hombre! (&mala'lgwa) /mala'lengwa/ mala len%a! (gRam'pi;kaRo) /gRam'pikaRo/
gran pcaro! (djo'mio)i (-s-)a /djos'mio/ Dios mo!
Even in compound names (of people or places), there is attenuation of the rst
element: (Xwa'kaRlo)i (x-, -s)a /xwan'kaRlos/ Juan Carlos (ma&RiaXo';fa)i (-xo's-)a
/ma&Riaxo'sefa/ Mara Josefa (&antoo'migo)i (&s-)a /santoo'mingo/ Santo Domingo (&toRr:eel'knde) /toRr:eel'konde/ Torre del Conde
6.3.2.7. Certain common phrases are also destressed: (&boka'BaXo)i (-xo)a /boka'baxo/ boca abajo (&pataaR'r:i;Ba) /patasaR'r:iba/ pat arriba
In numbers formed with y the rst element is destressed: (&tReintai'iko)i (-'s-)a
/tReintai'inko/ treinta y cinco (kwa&Rentai'j;te)i (-'s-)a /kwaRentai'sjete/ cuarenta y
siete also cien(to) before mil is destressed: (jeN'mil, -M'm-, -'m-, -M'm-, -'m-)i
('s-)a /jen'mil/ cien mil (&oCo&jenta'mil pe';ta)i (-&sj-, -as-, -'s;tas)a /ocojentas'mil pe'setas/ o>ocient mil pet Simple numbers are stressed, even if monosyllabic and near a stress: ('d 'a;to)i (-z 'a;tos)a /'os 'gatos/ dos gatos ('tR
ka'Ba;Lo)i (-s ka'Ba;,os)a /'tRes ka'baLos/ tr cabaos ('un 'li;BRo) /'un 'libRo/ un libro
(compared to the article: (un'li;BRo) /un'libRo/ un libro).
6.3.2.8. Obviously in long words, as in sentences, some secondary stresses are
introduced, for rhythmic reasons: ('kmpRame&lo) /'kompRamelo/ cmpramelo (a'Na;ae&le)i (-s-)a /a'Naasele/ adele (&entRe'andoe&lo)i (-s-)a /entRe'gandoselo/
entregndoselo
Words such as aune cone pore sino are normally /'aunke, a'unke 'konke 'poRke 'sino/, however, in sentences, their composition can also bring to stress
patterns such as: (&auke, a&uke, au&ke, au'k &koke, ko&ke, ko'k &poRke,
poR&ke, poR'k &ino, i&no, i'n)i (s-)a, although, they are often considered incorrect, for spelling reasons.
Emphasis, particularly in imperatives with enclitic pronouns, can modify the
structure considerably (above all, for stresses), as in: ('da;melo, "da:melo, "da;me&lo,
"da;me'l, 'da;me'l, &dame'l) /'amelo/ dmelo!
6.3.2.9. e words (a';Ra) /a'oRa/ ahora (a'i) /a'i/ ah (a'un) /a'un/ an/aun
when they are not in an intoneme and are linked to what follows, have the frequent variants ('aoRa, 'ai, 'aun): (a';Ra 'Bj;ne, 'aoRa) /a'oRa 'bjene/ ahora viene (a'i
e'ta, 'ai)i (-s-)a /a'i es'ta/ ah t (a'um mi 'pa;Re, 'aum) /a'un mi'paRe/ aun mi padre (a'un no'ale, 'au)i (-s-)a /a'un no'sale/ an no sale
In the case of /'j[s]/, in an intoneme (more often when followed by a pause), it
is quite frequent to nd, even in the neutral accent, a pronunciation which, often,
seems as if it were /'i[s]/, while, it is generally only /i'[s]/: (paR'tj, &paRti')
/par'tjo/ parti (a'j, &ai')i (-s)a /a'jos/ adis (lim'pj, &limpi') /lim'pje/
limpi (em'pj, &empi') /em'pje/ en pie however, occasionally, we can doubtlessly have the structure ('i&), even in neutral pronunciation: (paR'ti&, a'i&, lim'pi&,
em'pi&)i (-s)a.
252
6.3.2.10. In fast and familiar speech, common words and proper names in a preintoneme with ('ia) /'ia/ a frequently become ('ja) /'ja/: (al'dia i'jnte, al'dja, al&djai'-) /al'dia si'gjente/ al da si%iente ('p;ko 'ia e'pw, 'ja, 'p;ko&ja)i
(-z, -z, -s-, -s)a /'pokos 'ias es'pwes/ pocos d dpus (u'tia o'l;Re, u'tja, u&tjao'-)i (s-, -s)a /su'tia o'loRes/ su ta Dolor (&doNama'Ria maR'ti;ne, &doNama'Rja, &doNama&RjamaR'-)i (-s)a /doNama'Ria maR'tine/ doa Mara Martnez (gaR'ia u'tjRr:e, gaR'ja, gaR&jau'-)i (-s-, -s)a /gaR'ia gu'tjeRr:e/ Garca Gutirrez (&noeR'Bia
paRa'na;a, -'Bja, -&BjapaRa'-)i (-s-)a /noseR'bia paRa'naa/ no serva para nada
Other examples: (ea&Bia'pwto e'pj, ea&Bja'-, a-)i (s-)a /seabia'pwesto e'pje/
se haba puto de pie (&eta'Ria kan'ao, -'Rja, -&Rjakan'-)i (&es-, 'sa;os)a /esta'Rian
kan'saos/ taran cansados (&nopo'RiaN Le'a; Ra'tjmpo, -'RjaN, -&RjaNLe'-)i (,-)a
/nopo'Rian Le'gaR a'tjempo/ no podran egar a tiempo
Intonation
6.3.3. 6.3 shows the preintonemes and intonemes of the neutral Iberian and
American Spanish language. erefore, let us simply look at the fundamental examples. It is important to make comparisons both between them and with those
of other languages:
/./: (me'Bi ma'Na;na poR&lama'Na;na3 3)i (me'Bi ma'Na;na poR&lamaNa;na3 3)a /me'boi ma'Nana poRlama'Nana./ Me voy maana por la maana.
6.3. Iberian-Spanish preintonemes and intonemes.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 ' 1 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
6.4. American-Spanish preintonemes and intonemes.

/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
6. spanish
253
/?/: (e'ta nu't;e kontnto21)i (es'ta nus't;es kon'tntos21)a /es'ta nus'tees

kon'tentos?/ Estn ted contentos?
//: (&ima'Na;na &noRr:e'i;Bo no'ti;ja1 1| le&kRiBi'R e'nw;Bo3 3)i (&sima'Na;na &noRr:e'si;Bo noti;sja2 2| les&kRiBi'R enw;Bo3 3)a /sima'Nana noRre'ibo no'tija| leeskRibi'Re
e'nwebo./ Si maana no recibo noticia, le cribir de nuevo.
Text
6.4.0. e transcribed passage, e North Wind and the Sun is given in four
normalized versions. We start with the (neutral) Iberian pronunciation of (neutral British) English this is the rst step of the phonetic method (the written text
is given in 2.5.2.0). e Spanish translation follows, in its neutral version.
At the end, as always, there is the version which gives the English pronunciation of Spanish, by neutral British speakers, uent in Spanish (after prolonged contact with native speakers, but with no help from the phonetic method), who have
same principle is valid for the foreign pronunciation of English, given rst.
Speakers of American English could prepare their own version both of the Spanish pronunciation of English and of their pronunciation of Spanish, as an excellent exercise, by listening to native speakers, best of all after recording them. Of
Iberian Spanish pronunciation (of English)
6.4.1. (de'nR 'wind2 ande'an2 wRdi'pju;tin 'gwiC we'tRgeR3 3| 'wn a'tRa;BeleR2 'kim a'lg2 'r:a tina'wRm 'kluk3 3| dia'Ri{}2\ ai'wa xu'fR a'i;e2 iN'mikin de'tRa;BeleR2 'ti ki'ku 'kf1 1| &juBikan'i;eR e"tRgeR &andi'aeR3 3||
'dn2 de'nR 'win2 'blu2 a'xaR2 ai'ku3 3| &ba{}e'mR xi'Blu1 1| de'mR 'kluli2 &ie'tRa;BeleR2\ 'ful di'kluk a'Rauxin3 3| &an[d]a'la{t}2\ de'nR 'win2 'gi 'Ba ja'tmt3 3|| 'dn2 de'an 'j 'naut3 3 'wRmli3 3| &andi'mi;jali2\ de'tRa;BeleR 'tu 'kf3 3 xi'kuk3 3|| an'u2 e'nR 'win{d}2 &wo'Blai tukoM'f2| &dae'an3 3 we'tRgeR3 3
&Be'tu3 3||
&dijulaik21 de't;Ri2| dju'wn tu'xi;Ri tan21|||)
254
Spanish Text
6.4.2. El viento norte y el sol poraban sobre cul de eos era el ms fuerte, cuando acert a par un viajero envuelto en an>a capa. Convinieron en e ien ant
lograra obligar al viajero a itarse la capa sera considerado ms poderoso.
El viento norte sopl con gran furia, pero cuanto ms soplaba, ms se arrebujaba
en su capa el viajero; por n el viento norte abandon la empra. Entonc bri el
sol con ardor, e inmediatamente se dpoj de su capa el viajero; por lo e el viento
norte hubo de reconocer la superioridad del sol.
Te ha %stado el cuento? Vamos a repetirlo?
Neutral Iberian pronunciation
6.4.3. (el'Bjnto 'nRte2 jel'l2\ poR'fja;Ban2 &oBRe'kwal 'd;Lo2\ &eRael'ma 'fwRte3 3|
&kwandoaeR't apa'a; Rumbja'X;Ro2 em'bwlto e'na~Ca 'ka;pa3 3| &kombi'nj;Ron2 e&kekje'nante lo'Ra;Ra2 oBli'a; RalBja'X;Ro2 aki'taRe la'ka;pa1 1| e'Ria kon&ie'Ra;o2
'ma poe'R;o3 3||
el'Bjnto 'nRte2\ o'pl ko'gRaM2 'fu;Rja3 3| &peRo'kwanto 'ma o'pla;Ba1 1| 'ma aR&r:eBu'Xa;Ba2 enu'ka;pa3 3 [e]lBja'X;Ro3 3|| poR'fin2\ el'Bjnto 'nRte2 a&Bando'n lFem'pR;a3 3|| en'tne1 1\ BRi'L el'l3 3 konaR'R3 3| eiNmejata'mnte2 e&epo'X3 3 eu'ka;pa3 3 [e]lBja'X;Ro3 3|| &poRlo&kel'Bjnto 'nRte1 1\ 'u;Bo eR&r:ekono'R3 3 la&upe&RjoRi'a el'l3 3||
&tauta;o21 el'kwnto2| 'ba;mo aR&r:epetiRlo21|||)
Neutral (central-south-) American pronunciation
6.4.4. (el'Bjnto 'nRte2 jel'sl2\ poR'fja;Ban2 &soBRe'kwal 'd;,os2\ &eRael'mas fwRte3 3|
&kwandoaseR't apa'sa; Rumbja'x;Ro2 em'bwlto e'na~Ca ka;pa3 3| &kombi'nj;Ron2 e&kekje'nantes lo'Ra;Ra2 oBli'a; RalBja'x;Ro2 aki'taRse laka;pa2 2| se'Ria kon&sie'Ra;o2
'mas poeR;so3 3||
el'Bjnto 'nRte2\ so'pl ko'gRaM2 fu;Rja3 3| &peRo'kwanto 'mas sopla;Ba2 2| 'mas saR&r:eBu'xa;Ba2 ensuka;pa3 3 [e]lBjax;Ro3 3|| poR'fin2\ el'Bjnto 'nRte2 a&Bando'n lFempR;sa3 3|| entnses2 2\ BRi', elsl3 3 konaRR3 3| eiNmejata'mnte2 se&espox3 3
esuka;pa3 3 [e]lBjax;Ro3 3|| &poRlo&kel'Bjnto nRte2 2\ 'u;Bo eR&r:ekonosR3 3 la&supe&RjoRi'a elsl3 3||
&taus'ta;o21 el'kwnto2| 'ba;mo saR&r:epe'tiRlo21|||)
6. spanish
255
English pronunciation of Spanish

6.4.5. (&vi5nT 'n;TI2 i's:2\ &ph;fi'A;vn2 5s;b>I 'khwA: 'Djs2\ &>5ms f'TI3 3| 5khwA;nD s'Th; A;p'sA: uMvi'h>2 &Mv5T n'A;~c
'khA;p3 3| &khMvni'>n2 5kh;I ki'n 'A;nTs l'g>A;>2 &bl5gA:> vi'h>2 k5ThA;sI l'khA;p2| s5>Ii kn&sD'>A;D2 5ms pD'>s3 3||
&vi5nT 'n;TI2\ s5phl; k'g>A;M2 'f>i3 3| 5ph> 'khwA;T 'ms s'phlA;v32| 5ms si>b'hA;v2 nsu'khA;p3 3 &vi'h>3 3|| ph;'fI;in2\ &vi5nT
'n;TI2 &bnD5n; lA;m'ph>s3 3|| n'Th;nss32\ b>5j; 's:3 3 khnA;'D:3 3| &InmDi&A;T'mnTI2 sI&Dsp'h;3 3 Dsu'khA;p3 3 &vi'h>3 3|| &ph;l5kh;I vi'nT 'n;TI32\ 5uv D&>kn's;3 3 l&sup>i>5DA;D D's:3 3||
&ThIg'sTA;D21 &'khwnT2 5vA;ms &>p'Thl21|||)
7. Portuguese
7.0.1. We present both Brazilian and Lusitanian neutral Portuguese, in this order, although not separated, for various good reasons. e most suitable interphonemic transcription (and useful for both accents), as far as V are concerned,
has to use seven phonemes and four (or ve) diaphonemes. erefore, the seven
Brazilian phonemes, /i, e, E, a, O, o, u/, plus /e, , A, , /.
th this transcription, if we ignore or delete the dots under the diaphonemes
(even for given C] we also obtain the transcription for an international pronunciation, which is closer to the spelling and without the specic Brazilian or Lusitanian peculiarities, as we will see. e examples show pronunciation, not usage.
It seems correct to prefer neutral Brazilian Portuguese (which has its regional
variants, as Lusitanian Portuguese also has, but they will not be treated here), since
it is simpler, with fewer phonemes and fewer surprises, closer to spelling and,
therefore, more easily understandable, as well as decidedly more widespread, in
a ratio of 18 to 1! On the other hand, the Brazilians themselves may nd it dicult
to understand the Lusitanian pronunciation.
7.0.2. Should we instead proceed inversely, preferring Lusitanian pronunciation, again for the V we would need nine phonemes: /i, e, E, a, O, o, u, , /, including /E, a, O/, even in unstressed syllables, and with a massive occurrence of
/, /, almost exclusively above all // in unstressed syllables.
at is traditionally merged into //, for e ( /e/ + /N, L, S, Z, j/, and for ei
/ei/ as well) albeit very similar (particularly for some people), actually has a tiny
dierence. is can clearly be seen in our vocogram, where on the contrary it
does not appear to be so tiny (contrary to traditional practice with a rougher
quadrilateral: /e/ in these contexts (', ) and // (', )).
erefore, central Lusitanian by now neutral has /e/ ('3I, I, 'I, I ',
'N, 'L, 'S, 'Z) (compared to /e/ ('I, eI 'eI 'eJ), in the rest of Portugal and in
Brazil), that foreigners too often render as /E/, losing a functional opposition,
which is not negligible for native speakers of both accents.
Let us conclude these preliminary remarks (necessary to dene the problem
well), underlining the absurd and pointlessly complicated steps made trying to explain pronunciation, above all Lusitanian pronunciation (we restrict ourselves to
this language {obviously avoiding talking about English, or Danish, or even
French}), making use of graphemes instead of transcriptions, as is still done. On
the other hand, it must be said that no real pronouncing dictionary for Portuguese
has been written; therefore, one must make the best of it.
7.0.3. As can be seen from phonemic transcriptions, our analysis does not consider V nasalization distinctive, although it must be marked since it is clearly perceptible (above all in Brazilian pronunciation) and typical of Portuguese (but, in
7. portuguese
257
Lusitanian pronunciation, only in checked syllables in N] However, if nasalization is lacking as is often the case in foreigners' pronunciation comprehension
is not compromised; all the more so because, not rarely, in Portugal vowel nasalization may be hardly perceptible, while the nasal contoid is surely there.
As a matter of fact, we always have sequences of a (nasalized) V plus N the latter may be missing (phonetically, but not phonemically) only when a word-nal
() is followed by a word-initial () or (), instead of being dropped, since the
word ends in C\ //. e nasal contoid is homorganic in the case of stops ((m, n,
), and of stop-strictives; in Brazilian pronunciation we also nd (~)+(C, ),
which are taxophones of /t, d/), while it is semi-provelar (()) before other contoids or before a pause (as will be seen systematically in the sections on N 7.2.1).
Vowels
7.1.1.1. 7.1 shows the vocalic realizations of neutral Brazilian Portuguese.
is language also has several diphthongs, but, since their starting and nishing
points correspond to some vocoids already present, it may be sucient to list them
and give some suitable examples. However, /ei, o/ have been put in the vocogram, with their stressed and unstressed variants, not because they are considered
monophonemic, but owing to their movements which are a little particular.
erefore, there are seven phonemes, /i, e, E, a, O, o, u/ (i, e, E, a, O, o, u), with
some taxophones, such as (Au, A, [s]) for /'au, 'a, Au, A, a[s]/ and (i e,
' A '9, u) for /i, e, a, o, u/ for VN sequences, both in
checked syllables (/0/) and in unchecked ones (//).
As to /a[s]/, it is to be noted that ([s]) holds for -a(s) in nal rhythm groups,
due to a sort of attenuation, with or without a following pause. On the other
hand, however, the variant (a) is even possible, although it is more typical of traditional pronunciation (which is possible as well, provided it is not realized too low).
Anyway, () is more recommendable, and this is what we use in this chapter.
ere are three broken-line white markers as well, to show possible but not necessary articulations, also belonging to the neutral accent, for /i, a, u/ (I, x, U), occurring in rhythm-group nal unstressed syllables (with or without -s). Besides, in
the same position of (x), there may additionally be (), as a possible variant of
(stressed or unstressed) (A) /a/.
7.1.1.2. Here are some rst examples (for the time being, let us ignore the diaphonemic undersigned dots): (i'fi;si)b /e'fisi/ dicil ('e;]i)b /'ee/ ele ('E;])b /'Ea/ ela (ba'ta;t)b /ba'tata/ batata ('nO;v)b /'nOva/ nova ('no;vu)b /'novu/ novo
('u;v)b /'Suva/ >uva However, it is worthwhile reecting upon (A): ('pAu) /'pau/
pau (Au'das)b {(Au'aS)l] /Au'as/ audaz ('mA)b {('mA)l] /'ma/ mal (A'tu;R)b
{(A'tu;R)l] /A'tuRa/ altura
It is important to distinguish: seca ('se;k)b /'seka/ dry (f.) and ('sE;k)b /'sEka/
dries, drought, nuisance, cerco ('seku)b /'seKku/ siege and ('sEku)b /'sEKku/
I enclose, lobo (']o;bu)b /'obu/ wolf and (']O;bu)b /'Obu/ lobe, fosso ('fo;su)b
258
/'fosu/ trench and ('fO;su)b /'fOsu/ I dig.

th nasalization, we have: ('si)b /'sin/ sim (&]ega']g)b /enga'enga/ lengalenga ('bI)b /'bein/ bem ('9;meI)b /'mein/ homem (&AmA'NA)b /Ama'Nan/ ama (bm'b9U)b /bom'boun/ bombom ('9)b /'onKa/ honra (A'gus)b /A'guns/ al%ns (kA'mi;Nu)b /ka'miNu/ camio (fe'n9;menu)b /fe'nmenu/ fenmeno (&te]ef'n;m)b /te]efo'nema/ telefonema ('u;niku)b /'uniku/ nico
7.1. Brazilian vowel elements.

/i/ ({'}i) {(I)}, /i{}/ (i)
/e/ (e), /e{}/ (, e)
/ei/ (eI), /ei/ (I, eI)
/E/ (E, )
/a/ (a, Au, A, [s]) {/a/ ()}
/u/ ({'}u) {(U)}, /u{}/ (u)

/o/ (o), /o{}/ (9, )
/o/ (oU, o), /o/ (9U, U)
/O/ (O, )
/a[i/u]{}/ (A[i/u]{}) {([i/u]{})}
7.1.1.3. Notice that nasalization slightly changes some timbres according to the
diaphonemic symbols used: in particular /a/ (A) (for which, as already seen,
() is possible as well), and /'e, ', 'o, '/ (, 9) (when stressed, sometimes, (e, ) may be heard too).
Besides, in word-nal position, we have real diphthongs for what too often and
too hurriedly is transcribed simply as /e, /, /ein, oun/ ('I, eI '9U, U).
Instead, in the same position, /an/ opposes the diphthong /aun/: ('OfA) /'OKfan/
rf ('OfAu) /'OKfaun/ rfo Later on, we will see other diphthongs, too, both
oral and (phonetically) nasalized.
It has to be noted that in modern neutral Brazilian pronunciation, the nasalization of vocoids is more and more evident (even in diphthongs). As already said,
nasalization occurs when a V is followed by a nasal consonant [N), in the same syllable or not, both in checked and unchecked syllables; and, above all, both in
stressed and unstressed syllables: (&AmA'NA)b /Ama'Nan/ ama ('bA;Nu)b /'baNu/
bao ('fi;nu)b /'finu/ no
ereas, in traditional neutral Brazilian pronunciation, nasalization occurs only in checked syllables (in N] either stressed or not; but a troublesome occurrence of nasalization is possible (or not), in unchecked stressed syllables followed
by /N/ (and with /'a/ (') as well): (&ama'NA)b;t /Ama'Nan/ ama ('bA;Nu,
'b;Nu)b;t /'baNu/ bao ('fi;nu)b;t /'finu/ no
Lastly, in mediatic pronunciation, nasalization occurs, in addition to (stressed
or unstressed) checked syllables (in N), even in unchecked stressed syllables followed by N but, again, a troublesome occurrence of nasalization is possible even
in /N/ ( an unchecked, unstressed syllable, followed by /N/): (&amA'NA, &ama-)b;m /Ama'Nan/ ama ('bA;Nu)b;m /'baNu/ bao ('fi;nu)b;m /'finu/ no
7. portuguese
259
Lusitanian peculiarities
7.1.2.1. Let us now consider 7.2 to see the vocalic realizations of neutral Lusitanian Portuguese. It is a good idea to make a close comparison of the two gures; as a matter of fact (neglecting the broken-line white markers, of the unnecessary Brazilian variants, (I, U, x ): ('e;]i, -I 'E;], -x 'no;vu, -U &AmA'NA, &m'N)b] it
is clear that Lusitanian Portuguese has a higher number of both taxophones and
phonemes: at least two more (which, in an exclusively Lusitanian phonemic transcription, would be /, / or, deviating from ocial symbols, they should rather
be /, /; but, more traditionally rendered as /, /) for (, '), (', ) in
addition to () /A/ and (A) /'a/, with occasional oppositions between them
(that, in a Lusitanian way, we ought to mark as /a, /).
Besides, (, a, ) /, A, / are needed, above all, but not only, in some pre-stress
syllables (by diachronic fusion, which is no longer visible in present-day spelling:
// = //, deriving from a prior /0/ structure, with C dropping), in addition
to /, , A; K, K, AK/, as we will see (and /Au/, and /Ai/ as well). Furthermore, /-e2 -2 -e, -/ are special cases, as in: (ab'd9;meI)b (B'O;mI,
-n)l /ab'me2/ abdmenb -menl ('ipsi]U)b (-]U, -n)l /'ipsi2/ psilon
(&ik]u'zi;vi)b (&i~k9u-, -v[], -v)l /inku'zive, -/ inclive
7.2. Lusitanian vowel elements.
/e, e/ (), () in contact with (N, L, S, Z, ) (, ), but /e/ () between (], ) (S, Z)
/i/ (i), /i/ (i)
/u, o/ (u), /u/ (u)
/'e/ (e), // (), /e/ (, e)

/o/ (o), // (), /o/ (9, )
/ei/ (I, I), /ei/ (3I, I)
/o/ (o, oU), /ou/ (9U, U)
/'e/+/N, L, S, Z, j/ ()
/O/ (O, )
/E/ (E, )
/a, a', a%/ (), /a/ (, |), /'a/ ()
/a[i/u]/ (A[i/u], 4[i/u])
/A/ ('a), /A/ ({'}), /A/ (a, {'}ai, {'}Au, {'}A)
7.1.2.2. e conspicuous Lusitanian reduction of timbres is striking in unstressed syllables, where we have (, ) /e, e/ (or, in the exclusively Lusitanian phonemic transcription, //) (u) /u/ (, |) /a/ ( 7.4.0); and, more rarely, (i) /i/
(which, however, has the complication of oscillating with (, ), as we will see).
If we thoroughly analyze the phonemic transcription, it becomes clear that the
contradiction between the diaphonemes with underwritten dots and actual phonemes is only outward; as a matter of fact, in Lusitanian pronunciation, for /a/ we
nd (), in unstressed syllables, but (), in stressed syllables, when it is followed by
N in checked syllables (with nasalization (A)) or unchecked (without nasalization ()).
Every exception is marked with /A/ (besides its other not fundamental, though
deducible use, in contact with /u, /, seen above): ('E;])l /'Ea/ ela (b'ta;t)l /ba'tata/ batata ('nO;v)l /'nOva/ nova ('su;kaR, -R)l /a'sukAK/ {(a'su;ka)b} acar (&am-
260
'NA)l /Ama'Nan/ ama ('sAmb) /'samba/ samba ('k;m) /'kama/ cama

However, there is one exceptional case where ('a;) /'A/ opposes normal
/'a/ (';), as in: (f'];mS) /fa'amus/ falamos we speak (f']a;mS) /fa'Amus/
falmos we spoke; of course, in Brazilian pronunciation, they are both (fa']A;mus).
7.1.2.3. Here are some other remarkable cases of /A/, in Lusitanian pronunciation, by contraction: (a) /A/ [a a] (a'ke;]) /A'kee/ ele (ka'vIR) /kA'veiRa/ caveira ((a, a'ke;]i, ka'veIR)b] after CC (simplied or not in their pronunciation
and in spelling): (a'sAu)l /A'saun/ a(c)o (fa'tu;R)l /fA'tuRa/ fa(c)tura ((a'sAu, fa'tu;R)b] Also (ka'm9iS)l /kA'moins/ Cames.
Before looking at the two further phonemes, let us consider the Lusitanian
pronunciation of the examples already given in Brazilian pronunciation: (d'fi;si)l
/i'fisi/, ('e;])l /'ee/, ('E;])l /'Ea/, (b'ta;t)l /ba'tata/, ('nO;v)l /'nOva/, ('no;vu)l /'novu/, ('Su;v)l /'Suva/ ('pAu)l /'pau/, (Au'aS)l /Au'as/, ('mA)l /'ma/, (A'tu;R)l /A'tuRa/ ('se;k)l /'seka/, ('sE;k)l /'sEka/, ('seku)l /'seKku/, ('sEku)l /'sEKku/, (']o;Bu)l
/'obu/, (']O;Bu)l /'Obu/, ('fo;su)l /'fosu/, ('fO;su)l /'fOsu/.
Furthermore (for both accents): (ka'da;ve)b (k'a;vR)l /ka'davK/ cadver ('sO;Ro)b (-R)l /'sORK/ sror (a'ma;ve)b ('ma;v)l /a'mav/ amvel ('Akjo)b ('Ak)l /'akw/ lcool (vo'ta)b (v'taR)l /v'taK/ voltar Of course, if only Brazilian pronunciation were taken into account, even in these cases, no diaphoneme
with underwritten dots would be necessary, since there is no dierence in the realization of /e, o, / (e) and (o).
7.1.2.4. Bearing in mind that nasalization, in Lusitanian pronunciation, is
found only in checked syllables (but, sometimes, it is so slight and hardly perceptible that it could be marked as (), instead of ()), we have: ('si)l /'sin/ sim,
(&]eg']g)l /enga'enga/ lengalenga, ('b3I)l /'bein/ bem, ('O;mI)l /'mein/ homem, (&am'NA)l /Ama'Nan/ ama, (bm'b9U, -U)l /bom'boun/ bombom,
('9)l /'onKa/ honra, (A'uS)l /A'guns/ al%ns (k'mi;Nu)l /ka'miNu/ camio,
(f'nO;mnu)l /fe'nmenu/ fenmenol (&t]f'ne;m)l /teefo'nema/ telefonema, ('u;niku)l /'uniku/ nico One further Lusitanian trouble, by diachronic assimilation
of two V adds // (besides /A/), in stressed or unstressed syllables: ('A;fR)l (a'AfoR)b /a'anfoRa/ a nfora (;n'ti;)l (aA~'Ci;g)b /aan'tiga/ antiga
7.1.2.5. Here are some specic examples to show the dierences between the
two accents more clearly. In order to account for the pronunciation of central Portugal, with Lisbon as its focal point, therefore as a model of neutral Lusitanian pronunciation, it is important to know that /e/, in stressed syllables followed by /N, S,
Z, L, j, i/, takes on the pronunciation (), which is close to the realization of the
Lusitanian phoneme // (but we must lay great stress on the fact that they are
not alike {even if, occasionally, for some words or specic speakers, the dierence
is only very slight}). However, we must point out that even the more regular and
etymological pronunciation corresponding to the Brazilian one, is neutral ( acceptable neutral] with /e, ei/ (e, eI), which is kept quite well outside the areas in-
7. portuguese
261
uenced by Lisbon, with Coimbra leading the way.

Some examples: ('v;Nu)l ('v;Nu)b /'veNu/ veo ('f;Su)l ('fe;u)b /'feSu/ fe>o
('v;Zu)l ('ve;u)b /'veZu/ vejo ('t;Zu)l ('te;u)b /'teZu/ Tejo (?S'p;Lu)l (is'pe;Lu)b /es'peLu/ peo ('s;u)l ('se;u)b /'seju/ seio ('sIS)l ('seIs)b /'seis/ se besides ('s3I)l
('seI)b /'sein/ sem ('k;Zu kU']It 'v;L)l ('ke;u kU']eICi o've;L)b /'keZu
koun'leite eo'veLa/ eijo com leite de ovea In unstressed syllables /ei/ (I)l (eI)b
remains: (']Itu)l (']eItu)b /'eitu/ leito (]I'toR)l (]eI'to)b /ei'toK/ leitor
7.1.2.6. e other pre-eminently Lusitanian characteristic, in purely Lusitanian
transcription, would resort to the phoneme // for (), with the taxophone () in
contact with /N, L, S, Z, j/ or followed by /s, a/. However, in our diaphonemic interpretation, we have word-internal /e/ corresponding to Brazilian /e/ and to written
-e- (but neither for initial e- {= /i/}, in both accents, nor for nal -e {= /e/ ()l (i)b}): (n'Nu)l (ne-)b /ne'Nun/ neum (m'LOR)l (me'LO)b /me'LOK/ meor (f?'SaR)l (fe'a)b
/fe'SaK/ fe>ar (SZ'aR)l (e'ga)b /Se'gaK/ >egar (&tZZ'i;Lu)l (&tea'-)b /teZa'iLu/ tejadio ('o;Z)l ('o;i)b /'oZe/ hoje (]';mS)l (]e'A;mus)b /e'jamus/ leiamos (&i]'Ant)l (&i]e'gA~Ci)b /ie'gante/ elegante (dZ&vt'BRaR)l (dez&vete'bRa)b /eaveKte'bRaK/
dvertebrar ('pi;RZS)l ('pi;Ris)b /'piRes/ pir (i'E;], 'E;])l (i'E;], 'E;])b /i'Ea/ e ela
However, if the sequences /e/ () + /N, L, S, Z s, a/ are preceded by /, K/ (] , )
(given their dorsal component), the taxophone that should be used is (): (&i]'ZeR)l
(&i]e'e)b /ie'ZeK/ eleger ('to;S)l ('to;is)b /'toKes/ torr In these contexts, /i/ -i- generally remains the same (/i/, but it currently becomes /e/): (]iZ'Bo, ]-)l (]iz'bo)b
/ia'boa/ Lboa ('fE;]S, -]ZS, -]ks)l ('fE;]is, -ks)b /'fEis/ Flix (mi'LAu, m-)l (mi'LAu)b
/mi'Laun/ mio (di'NIRu, d-)l (i'NeIRu)b /di'NeiRu/ dieiro We have /e/ for
() as well, while /e/ is (e): (&tn'BRo;zu)l (&tene'bRo;zu)b /tene'bRozu/ tenebroso (&tep'Ra;u)l (&tempe'Ra;du)b /tempe'Rau/ temperado
7.1.2.7. Another Lusitanian oddity consists of the change of /i/ (i) -i- sequences (not /i/ (i)) into //, /e/, in adjacent syllables (except for the last one,
whether stressed or not, and the rst one, if absolute initial, with no C-]\ (&dv'iR,
dv'-)l (&ivi'i)b /eve'iK/ dividir (m[]'niStu)l (mi'nistRu)b /me'nistRu/ mintro
(&in[]'mi;u)l (&ini'mi;gu)b /ine'migu/ inimigo (p&vi][]'Za;u)l (pRi&vi]e'a;du)b /pRevi'Zjau/ privilegiado (in the very last example, the sequence (-]'Z-) /-'Z-/ results
from what has just been said previously).
However, we nd the following exceptions, where i /i/, corresponding to an
original /'i/, remains /i/ in derivatives, as in the conditional mood of verbs in -ir:
(d[]&vi'RimS)l (i&vii'RiAmus)b /evei'Riamus/ dividiramos and in superlatives: (&df'si;]imu)l (&ifi'si;]imu)b /efi'siimu/ diclimo
For rare /u/ sequences, the same is true, so we have to resort to the diaphoneme
// to be able to account for this possible Lusitanian phenomenon: (f'tu;Ru, f-,
f-, fu-, f-, fu-)l (fu'tu;Ru)b /f'tuRu/ futuro
7.1.2.8. In Lusitanian pronunciation, /o/ (not /o/ ()) corresponds to /u/:
(&upu'BRo;zu, &-)l actually, [h]o[CV]- can be either (u)l or ()l (&opRo'bRo;zu)b /opRo'bRjozu/ oprobrioso (&munut'ni)l (&mnot'ni)b /monoto'nia/ monotonia
262
(m'sAu)l (m'sAu)b /mon'saun/ mono however, in bureaucratic lexical

compounding, for -o- we nd //: (&mns']a;Bku)l (&mnosi']a;biku)b /mnsi'abiku/ monossilbico (']uz f4'seS)l (']uzo fRA'ses)b /'uzfRan'ses/ lo-francs
7.1.2.9. A diaphonemic use of /, / regards the Lusitanian possibility to have
distinct timbres, (, ), before the stressed syllable; as already seen for a this mainly occurs for diachronic contraction or after CC (simplied or not {in pronunciation or writing}): (v'se)l (vo'se)b /v'se/ voc (from vossemec vossa merc {as in
some southern Italian dialects vossa from vos(tra) signora your Lordship}), (k'oR)l (kRe'do)b /kR'oK/ credor (&k'seR)l (&ake'se)b /ak'seR/ aecer (&f'ti;vu)l
(&afe'Ci;vu)b /af'tivu/ afe(c)tivo (f9k'sAu)l (f]ek'sAu)b /fk'saun/ exo (k'Ra;u)l (ko'Ra;du)b /k'Rau/ corado (mR'o;mu)l (mo'd9;mu)b /mK'omu/ mordomo (&'ti;vu)l (&ado'Ci;vu)b /a'tivu/ ado(p)tivo (p'sAu)l (op'sAu)b /p'saun/
opo also: (bo'aR)l (bo'a)b /b'jaK/ boiar (from /oi/).
Besides, this phenomenon arises in compounds and some derivatives, with semantically more distinct elements (as seen above for o as well): (p'tO;nku)l (pRe't9;niku)b /pR'tniku/ pretnicol pret-b, but also in (more) scholarly words: ('tO;Rk)l (e'tO;Rik)b /K'tORika/ retrica
nally, note the following Lusitanian minimal pairs: (p'aR)l /pRe'gaK/ to
nail (p'aR)l /pR'gaK/ to preach both pregar ((pRe'ga)b) and (&kR'sAu)l /koRa'saun/ heart, (&kR'sAu)l /kRa'saun/ coloring corao ((&koRa'sAu)b]
Diphthongs
7.1.3.1. As far as grammatical or traditional diphthongs are concerned, let us
clarify that we consider them biphonemic, and not as unitary phonological entities, since their extreme points generally coincide with the usual Portuguese vocalic elements.
Furthermore, we want to specify, right from the start, that true diphthongs are
only those formed by two vocoids, //, as /ai/, in ('pai) /'pai/ pai certainly not
sequences of contoid plus vocoid, /0/, as /ja/, in ('pa)b ('paR)l /'pjaK/ piar On
the other hand, sequences like /'/ are not diphthongs either: (pa'is)b (p'iS)l /pa'is/
pas against ('pais)b ('paiS)l /'pais/ pa
After conrming this, let us also stress the fact that it is not good to transcribe
true diphthongs as if they were sequences of a vowel and a consonant, /0/, as
/aj/, since they are not at all comparable with sequences like /as, a, aK, an/,
which are structurally very dierent. Least of all, are we allowed to consider sequences of vowel and consonant plus vowel, /0/, as diphthong + vowel, or
vowel + diphthong, like /aja/, in (ka'a)b (k'aR)l /ka'jaK/ caiar ('sa;)b (-)l /'saja/ saia
7.1.3.2. erefore, with constant reference to 7.1-2 (for both accents Brazilian and Lusitanian), we now show the Portuguese diphthongs, even if in the vocograms only /ei, o/ appear as (eI, eI, I, oU, U, 9U)b (I, I, 3I, I oU, 9U, U 4i,
7. portuguese
263
4u)l. As a matter of fact, their second elements are (I, U), which are not the exact
realization of the phonemes /i, u/; for the Lusitanian accent, even (4i, 4u) are
placed in the vocogram, because their rst element is closer, as occurs however for
/a/ (4) as well).
rst of all, we list them phonetically and diaphonemically: (eI)b (I, I)l /ei/,
(Ei, i) /Ei/, (ai) /ai/ (ui) /ui/, (oi) /oi/, (Oi, i) /Oi/ (iu) /iu/, (eu) /eu/, (Eu, u)
/Eu/, (Au) /au/, (oU, o)b (o, oU)l /o/ ((o) unstressed //).
Here are some actual examples: ('eIs)b ('IS)l /'Keis/ re (]eI'9is)b (]I'S9iS)l
/ei'Soins/ Leix ('Eis)b ('EiS)l /'KEis/ r (&i'zi;tus)b (&i'zi;tuS)l /'KEi'zitus/ reizitos ('vai) /'vai/ vai ('fui) /'fui/ fui ('sois)b ('soiS)l /'sois/ so ('sOis)b ('sOiS)l /'sOis/ s
(iRikA'm~Ci)b (iRik'mnt)l /i'ROika'mente/ heroicamente ('viu) /'viu/ viu (and
vi-o as well, since the supposed dierence between the two is not at all phonic in
neutral pronunciation but derives exclusively from a wish to keep dierent morphological forms separate), ('seu) /'seu/ seu (&apu'zi;Nu)b (&Spu'zi;Nu)l /SapEu'ziNu/ >apeuzio (a'pEu)b (S-)l /Sa'pEu/ >apu ('mAu) /'mau/ mau ('voU, vo)
/'vo/ vou Lusitanian pronunciation prefers the reverse order, ('vo, 'voU), for the
last one.
Instead, the simplication of ('eI)b ('I, I)l /ei/ changing into ('e) does not belong to neutral pronunciation, although it is quite widespread, mainly in Brazilian
(and southern Lusitanian) pronunciation.
Of course, there are other actual diphthongs, as the following: (i'f]uu)b (i'f9-)l
/in'fuu/ inuo ('mou) /'mou/ moo (']u)b (-)l /'ua/ lua ('i)b ('i)l /'Kia/ ria
('bo)b (-)l /'boa/ boa including combinations as in the true triphthongs resulting
from /i+ei, i+au/: ('vieIs)b ('viIS)l /'vieis/ veis, ('ieIs)b ('iIS)l /'Kieis/ reis, ('ieI)b
('iI)l /'Kiein/ riem, and ('iAu)b ('i4u)l /'Kiaun/ riam.
7.1.3.3. e diaphonemic transcription has to adequately show the characteristics in unstressed syllables as well, seen that, in Lusitanian pronunciation, the diphthong elements do not undergo reduction. Note (besides reizitos heroicamente
>apeuzio]\ ('faseIs)b (-IS)l /'faseis/ fce (pai'zi;Nu)b (pai'zi;Nu)l /pAi'ziNu/ paizio (f]ui'des)b (f9ui'eS)l /fui'es/ uidez (oi'ta;vu) /i'tavu/ oitavo (miu'di;Nu)b
(miu'i;-)l /miu'iNu/ miudio (&adeu'zi;Nu)b (&eu'zi;-)l /aeu'ziNu/ adeuzio
(kAu'tE;])b (-])l /kAu'tEa/ cautela in Lusitanian pronunciation, for // the monophthong (o) is preferred, without reduction to (u): (o'ba)b (o'BaR)l /K'baK/
roubar
e only seeming exception is proclitic ao(s) which is not /Au[s]/, but simply
/a u[s]/ its actual nature is a little hidden by the spelling that unies the two grammemes, which could even be a o(s); on the other hand, it is not *au(s)!
is goes for both Lusitanian and Brazilian Portuguese; as a matter of fact, pronunciations like (Au) are excessive and pedantic, against the normal realization
(au)b (u)l. Even in its Lusitanian nasalized form we have (Au, 4u), by coarticulation; or rather, in current Luso-Brazilian pronunciation, we generally nd ()
/O/: (au']a;du, ao-, -)b (u']a;u, o-, -)l /au'au/ ao lado (&auzA'mi;gus, &ao-, &-)b
(&uz'mi;S, &o-, &-)l /auza'migus/ aos amigos
erefore, it is important to mark /'au, Au 'ai, Ai/, so as to be mistaken no lon-
264
ger and avoid people believing that, in unstressed syllables, in Lusitanian pronunciation they can be reduced to /u, i/ (as some texts say).
Let us briey observe (and without showing it in vocograms {although, of
course, this is unavoidable in a book entirely dedicated to the pronunciation of
Portuguese}), that when /E, O/ are half-lengthened in unchecked syllables (or, for
emphasis, even in checked syllables), besides as being realized as normal monophthongs, (E;, O;), they can also be realized as doubling, or as very narrow diphthongs
more frequently so in Brazilian pronunciation. In comparison with the articulations shown in 7.1-2, these may start from slightly raised points and reach the
indicated ones, (EE, OO), or else they may start from those and lower themselves
a little, crossing the border of the phone below (in the vocogram), ((E, O)) including intermediate shifts (however, again of the opening type, although limited), that is ((Ep, OQ)).
7.1.3.4. Moving on to the nasalized versions, instead, we nd (Ai)b ('Ai,
4i)l /ai/, (Au)b ('Au, 4u)l /au/: ('mAi)b ('mAi)l /'main/ me (&mAi'zi;N)b (&mAi'zi;N)l /'main'ziNa/ mezia (&fa]a'RAu)b (&f]'RAu)l /faa'Raun/ falaro (fa']a;RAu)b (f']a;R4u)l /fa'aRaun/ falaram (th regard to the vocalic timbres in nasalized diphthongs, note that in compounds (&) as a realization of /'/,
followed by another stress, corresponds to /'/, while a rhythmic (&) / / corresponds
to //.) As already seen ( 7.1.1.3), the diphthong /aun/ opposes /an/: ('OfAu)b
('Of4u)l /'OKfaun/ rfo ('OfA)b ('Of4)l /'OKfan/ rf
It is to be noted that, in Lusitanian pronunciation as well, there is a (not slight)
dierence between /ai/ and /ei/; and it is more remarkable in unstressed syllables (even if /ai/ is never completely unstressed, since it has a low occurrence in
the Portuguese vocabulary, and always in lexemes, not in grammemes): ('kAis)b
('kAiS)l /'kains/ c (&mAi'zi;N)b (&mAi'zi;N)l /'main'ziNa/ mezia ('sI)b
('s3I)l /'sein/ sem (seI'vi)b (sI'viR)l /sein'viK/ sem vir In its stressed Lusitanian
form, the dierence between /ai/ and /ei/ is not only in the rst elements,
(A) (higher-low central) and (3) (lower-mid front-central) respectively, but also in
the second (front) ones: (i) (high) and (I) (lower-high).
Instead, for /ei/ and /'ei/, only the rst (front-central) elements are dierent,
() (higher-mid) and ('3) (lower-mid); in Brazilian pronunciation, the latter is
(e) or ('), which have the same height, but are front, instead of front-central,
while for the rst element of Brazilian /ai/ we nd (A) (higher-low, with the possible variant ()), against Lusitanian (A, 4) (lower-mid when unstressed), which are
both central (but () is back-central).
7.1.3.5. e other nasalizable diphthongs (besides ('Ai) /ai/, ('I, eI)b
('3I, I)l /ei/) are\ ('ui) /'ui/, ('9i) /'oi/, ('9U, U) /ou/: ('vIs)b
('v3IS)l /'veins/ vens ('va;eIs)b ('va;ZIS)l /'vjaZeins/ viagens ('muintu)b ('muintu)l /'muintu/ muito (just one example, varying with ('mi-)), (]i's9is)b (-S)l /i'soins/ li ('s9U) /'soun/ som (&kUka']o)b (-~k']oR)l /kounka'oK/ com calor
We have already seen marginal cases as: (ab'd9;meI)b (B'O;mI, -n)l /ab'me2/ abdmenb -menl ('ipsi]U)b (-]U, -n)l /'ipsi2/ psilon
7. portuguese
265
e following forms, with true triphthongs and quadriphthongs, are a bit

odd: ('p9ieI, 'p9i)b ('p9iI, 'p9i)l /'poi[ei]n/ pem ('vIeI, 'veI, 'vI)b
('v3II, 'v3I, 'v3I)l /'vei[ei]n/ vm ('veIeI, 'veeI, 'vI)b ('veII, 'veI 'v3I)l
/'v[ei]ein/ vem
Of course, the other triphthongs of grammars and teaching tradition are simply sequences of the two types /0, 0/ (even + //), as for instance: ('feIs)b
('fIS)l /'fjeis/ e (']ais)b (']aiS)l /'jais/ lea ('pjeIR)b ('pIR)l /'pweiRa/ poeira
('sa;u) /'saju/ saio ('me;)b ('m;)l /'meja/ meia (']qAu)b (']qAu)l /'jaun/ leo
('pq9is)b ('p'9iS)l /'pjoins/ pe
Consonants
7.2.0. 7.3 is the table of the Luso-Brazilian consonants, which we will examine systematically, according to manners of articulation. Instead, 1.9-15 give the
orograms, equally grouped by manners, of all the contoids given in this book, even
as secondary, occasional, or regional variants, for the 12 languages dealt with.
uvular
velar rounded
velar
provelar
palatal
postalveo-palatal
protruded
postalveo-palatal
velarized alveolar
velarized alveolar
rounded
alveolar
dental
labiodental
bilabial
7.3. Table of Portuguese consonants.
(~)
N
() ()
(n) n
m
F pb
( )b
t d
kg
(C )b
f v
()l
()l
_
s z
( )b Sl Zl
/w/(j) ()b
(B)l
()l
/j/()
R
/K/
(])| ()b
L
/b/ (b, Bl), /t, d/ (t, Cb d, b, l), /k, g/ (k, b g, b, l), /K/ (b, l)
Nasals
7.2.1.1. Portuguese has three nasal phonemes: (m, n, N) /m, n, N/: ('kA;ms)b
('k;mS)l /'kamas/ cam ('kA;ns)b ('k;nS)l /'kanas/ can ('kA;Ns)b ('k;NS)l /'kaNas/ ca
In addition, there are some taxophones for /0, , /, and it is useful to linger
over them, since too many descriptions keep on ignoring them completely or in
part. As a matter of fact, the only result of transcriptions like ('Ki, 'esu, 'A, 'bRAku,
's, 'us/'uS) is to make people think that Portuguese sounds almost like French
which is decidedly misleading since they persist in groundless phonological
characteristics.
266
As we have already said, Portuguese vowel nasalization is an important phonetic aspect; it is stronger in the Brazilian accent (so much so that it even occurs in
unchecked syllables followed by NV] than in the Lusitanian accent, which presents nasalization only in checked syllables in N where it is very slight and sometimes very hard to perceive (although instruments may nevertheless indicate it, as
happens in languages for which nobody would ever imagine to mark it, being automatic).
7.2.1.2. erefore, transcriptions like those just mentioned are not trustworthy,
unless they presume that foreigners will inevitably add a nasal consonant (although it is hard to believe), producing something similar to what is expected (excluding northern French people, who would then have other problems indeed).
On the other hand, this would be a very shoddy way of teaching phonetics.
erefore, taxophones are absolutely necessary for exact transcriptions of Portuguese (and for satisfying phonic reproductions), including devoiced variants especially for the Lusitanian accent (in front of voiceless C]\ (m) + /p, b/: ('tmpu)b
('tpu)l /'tempu/ tempo ('s9mbR)b ('s9mbR)l /'sombRa/ sombra (n) + /t, d/ (dental, which could be transcribed with (()), but the normal symbol is quite sucient) and (~) (for Brazilian Portuguese, before (C, ) + (i, ), /t, d/ + /i, e, j/): ('A~Cis)b ('Ant?S)l /'antes/ ant ('vnd)b ('vnd)l /'venda/ venda ('9~i)b ('9nd)l /'onde/ onde () + /k, g/: ('nuk)b ('nu~k)l /'nunka/ nunca (']igj)b (']igj)l /'ingwa/ ln%a
7.2.1.3. Lastly, the semi-provelar taxophone is very important; it occurs before any other consonant (among the phonotactically permissible ones, from
which /R/ is excluded). ey are realized as constrictives (f, v s, z b, Sl, b, Zl ):
(i'fi e-)b (-'f-)l /in'fin en-/ enm ('dsu)b ('dsu)l /'densu/ denso ('kAis)b
('kAiS)l /'kains/ c ('9)b ('9)l /'onKa/ honra (possibly even before the traditional or regional alveolar trill realization of /K/: ('9r[:])b ('9r[:])l] approximants ( j ): (']A Agu'dAu)b (']A Au'Au)l /'an iAgu'aun/ l e algodo (&seIjA'mi;gu)b (&sIj'mi;u)l /sein-ua'migu/ sem o amigo ( honra] and
laterals (]): (u']a;du)b (-u)l /un'au/ um lado it is the same before a pause as
well\ ('si) /'sin/ sim ('nAu) /'naun/ no
Instead, for /[]/ followed by N or V(N) the actual realization is (, ,
), where () is dropped, but there is no vowel elision ( 7.3.2.2, Taxophonics]
Stops
7.2.2.1. ere are three diphonic pairs, already seen in various examples, (p, b
t, d k, g) /p, b t, d k, g/, with dental (t, d) which, in Brazilian pronunciation,
becomes stop-strictives ((C, ), seen previously), before /i, e, j/: ('p9mb)b (-)l
/'pomba/ pomba ('kAg)b ('kAg)l /'kanga/ canga ('tnd)b (-)l /'tenda/ tenda
('Ci;midu)b ('ti;miu)l /'timiu/ tmido (sAu'da;i)b (sAu'a;)l /sAu'ae/ saudade
('Ca;tRu)b ('ta;tu)l /'tjatRu/ teatro Before front V or /j/, /k, g/ are realized as pre-
7. portuguese
267
velar, ((, l)), by coarticulation, but it is not necessary to use these symbols, except for Brazilian pronunciation before /i, e, j/, where we nd (, ): ('intu)b ('kintu)l /'kintu/ into ('tO;i)b ('tO;k)l /'tOke/ toe (iN'E;tu)b (i~'kE;tu)l /in'kjEtu/ inieto ('i)b ('gi)l /'gia/ %ia
e most interesting Lusitanian peculiarity (which is more complicated for foreigners, but also for Brazilans who might try to speak like Lusitanians) is constituted by voiced stops, /b, d, g/, which are realized as actual stops, (b, d, g), only after
pauses, after N (as can be seen in previous examples), and for /d/ (d), with an apical contact (since both of them are homorganic articulations), or for emphasis or
precision: ('b9U) /'boun/ bom ('da) /'a/ d ('ga;tu)l (-u)b /'gatu/ gato ('kAdu)l
(--)b /'kadu/ caldo (p'i;u)l (pe'i;du)b /pe'iu/ pedido with emphasis: (p"di:du)l
7.2.2.2. In all other cases, continuous realizations occur, (B, , ): two approximants and a constrictive respectively. e diaphonemic transcription shows /b, ,
g/, just to underline the dierence and to make their distribution adequately understood (avoiding forced inferences, which risk strengthening wrong concepts).
Of course, in Lusitanian Portuguese, the (word-initial) notation /b, , g/ is potential, since it corresponds to (b, d, g) or (B, , ) according to actual contexts,
not in absolute terms. In addition, we must admit that in the case of /b, g/, and
of /Rb, R, Rg/, and of (more) scholarly consonant clusters as well, stop realizations
are not rare, (b, g Rb, Rd, Rg), even without emphasis or desire for precision.
Some examples: ('a;B)l ('a;b)b /'aba/ aba (u'BRi~ku)l (u'bRiku)b /u'bRinku/ o
brinco (]iZ'Bo, ]Z-)l (]iz'bo)b /ia'boa/ Lboa ('ABu, 'Ab-)l ('Abu)b /'abun/
lbum (R'Bi;tu, R'b-)l (a'bi;tRu)b /aK'bitRju/ arbtrio (&Bi'kaR, &bd-)l (&abi'ka)b /abi'kaK/ abdicar (&um'a;)l (&umade'da;d)b /umae'aa/ uma dedada (p'RAu)l (pa'dRAu)b /pa'Raun/ padro ('aRju, 'aRd-)l ('adju)b /'aKwu/ rduo (dZ'3I)l (dez'dI)b /ea'ein/ ddm
More: ('a;u)l ('ga;gu)b /E'gagu/ gago (A'OS, A'g-)l (A'gOs)b /A'gOs/ algoz
('veZu)l (-zgu)b /'veagu/ vgo ('Ra;s)l (a'gRa;s)b /a'gRasa/ a graa (&R'ma;s,
&Rg-)l (&agA'ma;s)b /aKga'masa/ argamsa
7.2.2.3. Besides, in Brazilian (even neutral) pronunciation, /e/ (i I) is typically
added in order to separate word-nal stops, or stops + C (dierent from /R, /): (op'ta, &opi'ta, &op-)b (p'taR)l /p'taK/ optar (&advo'ga;du, &aiv-)b (&vu'a;u)l /avo'gau/ advogado ('itmu, 'i;Cimu)b ('itu)l /'Kitmu/ ritmo (p'neus, pi'-)b
(p'neuS)l /p'neus/ pne Other examples: ('kamu, -dm-)l ('kadmu, -imu)b /'kamju/ cdmio ('mi], -gd-)l (A'migda] -id-)b /a'migaa/ amgdala ('dinu,
-gnu)l ('ignu, -inu)b /'ignu/ digno ('dOm, -gm)l ('dOgm -im)b /'Ogma/
dogma (i'ze;m, ig-)l (ig'z;m, &ii-)b /ig'zema/ eczema
In Lusitanian pronunciation, there are some similar cases: (ob'E;tu, &obi-)b (B'ZE;tu &B'Z-)l /ob'ZEtu/ obje(c)to (bi']ak, -']a;i, -)b (bi']ak -k)l /bi'ak/ Bilac. Lastly, among complicated consonant clusters for Brazilians, we also nd cases like:
('aft, 'afit, -ft)b ('aft)l /'afta/ afta but ('k]a;Ru, i']-)b ('k9a;Ru)l /'kaRu/ claro as
well (due to dierences in the primary place of articulation).
268
Constrictives
7.2.3.1. ere are three diphonic pairs of constrictive phonemes, plus a couple
of diaphonemes (/s, a/, which we will see presently): (f, v s, z b, Sl, b, Zl) /f, v s,
z S, Z/: ('fa;su)b (-su)l /'fasu/ fao ('po;vu) /'povu/ povo ('su)b (-)l /'su/ sul (te'sAu)b (--)l /ten'saun/ tenso ('vAs)b (-s)l /'vasa/ valsa ('pEs)b ('pEs)l /'pERsa/ persa (pa'se;u)b (p's;u)l /pa'seju/ pseio ('va;zu) /'vazu/ vo (a'za)b ('zaR)l
/a'zaK/ azar (u'zi~us)b (u'zindS)l /u'zindjus/ os ndios (e'Res)b (S'ReS)l /Se'Res/ xerez ('Eis)b ('SERS?S)l /'SEKSes/ Xerx ('o;i)b ('o;Z)l /'oZe/ hoje
In neutral Brazilian, the diaphonemes /s, a/ behave as /s, z/, with distribution:
/s|, s[]=, z[], z/, while, in neutral Lusitanian, they correspond to /S|, S[]=, Z[],
z/ (as well as in the Carioca accent, of Rio de Janeiro, but of course realized as
(|, []=, [], z)): ('tRas)b ('taS)l /'tRas/ trs (us'pais)b (uS'paiS)l /us'pais/ os pa (i'sEsu)b (?S'sEsu)l /is'sEsu/ excelso (bas'ta)b (bS'taR)l /bas'taK/ btar (pes'ka;d)b
(p?S'ka;)l /pes'kaa/ pcada (az'mAis)b (Z'mAiS)l /aa'mains/ m ('azm)b
('aZm)l /'aama/ ma (az']As)b (Z']AS)l /aa'ans/ ls (further examples occur in
other parts of this chapter).
Approximants
7.2.4.1. ere are two approximants, (, j) /j, w/ (semi-palatal, and semi-velar
rounded), corresponding to prevocalic unstressed >i e and >u o: ('mo;]u) /'mjou/ miolo ('g]O;R)b (-)l /'gORja/ glria ('pq9;n)b ('pO;n)l /'pjnja/ peniab pe-l
('pqAu)b ('p'Au)l /'pjaun/ peo (i'dE;)b /i'Eja/ idia (i';)l /i'eja/ ideia (ko'e;u)b (k';u)l /ko'Keju/ correio (km'bO;u) /kom'bOju/ comboio (sAm'pa;u)b
(sA-)l /sam'paju/ Sampaio ('a;gj)b ('a;j)l /'agwa/ %a ('pjE;t)b ('pE;t)l
/'pwEta/ poeta ('va;kju)b (-ku)l /'vakwu/ vcuo (a'gji)b (R'jiR, R'g-)l /aK'gwiK/ ar$irb -%irl ('kimbR)b ('kimbR)l /'kwimbRa/ Coimbra
If pronunciation is slowed down, either for clarity or emphasis, (, j) /j, w/ may
change (through (j, w)) to (i, u) (and to (e, o) as well, according to spelling): ('g]O;Ri)b (-i)l, (pi'9;ni, pe-)b (pi'O;ni, pe-)l (pi'Au, pe-), (i'dE;i)b (i';i)l, (ko'e;Iu)b
(k';Iu)l, (km'bO;iu), (sAm'pa;iu)b (sA-)l ('a;gu)b ('a;u)l, (pu'E;t, po-)b (-),
('va;kuu, -u), (&agu'i)b (&Ru'iR, -g-), (ku'im-, ko-).
Several examples show that, in Lusitanian pronunciation, by assimilation, we
have devoicing after voiceless C and nasalization before nasalized V (, ', q,
); in Brazilian pronunciation, we generally nd only nasalization, (q, ).
Trills
7.2.5.1. Under this manner of articulation, in addition to the alveolar tap, (R)
/R/, we will treat the theoretical uvular trill, /K/, and the diaphoneme /K/ as well.
For the former, there is not much to say, except that it has a single alveolar contact, in fact it may even become an alveolar approximant (); in Luso-Brazilian pro-
7. portuguese
269
nunciation, it occurs between V (in the same word), where it distinctively opposes /K/ and after tautosyllabic C\ ('ka;Ru) /'kaRu/ caro ( ('ka;u)b (-u)l /'kaKu/ carro] ('bRa;su)b (-su)l /'bRasu/ brao ('fRiu)b ('f-)l /'fRiu/ frio
Besides (as a realization of /K/) (R) occurs, in Brazilian pronunciation, in word-nal position too, when followed by a subsequent word-initial V with resyllabication and behavior as if in word-internal position ( /K/=/R/), while in Lusitanian pronunciation it also occurs in nal position, even before a pause or a C\
(']e 'RAgu)b (']e 'RAu, -gu)l /'eK 'agu/ ler algo (']e 'poUku, 'po-)b (']e 'poku, 'poU-)l
/'eK 'poku/ ler pouco (nAu']e)b (n4u']eR)l /naun'eK/ no ler
7.2.5.2. e phoneme /K/ occurs in word-initial position, even after C or V and
word-internally after / n, , s/ ( heterosyllabic C] in neutral Brazilian, it is realized as a voiceless uvular approximant (), independently from context, with the
following variants, again independent from neighboring phones, shown in order
of frequency: voiced uvular constrictive (), or voiced uvular trill (K), even devoiced (, ). In addition, there are two further variants, which are less neutral:
voiceless uvular constrictive trill () and, at last, voiced alveolar trill (r:), which after /[]/ may become (Rr:): ([R]r:).
In Lusitanian pronunciation, this phoneme is (), with the following variants
(again in order of frequency): (, K, , , ) (where () is the voiced counterpart of
()), in addition to ([R]r:), which a little more than a century ago was the only
neutral pronunciation (while it is in the minority today, either provincial or rural).
Some examples: ('u)b ('u)l /'Kua/ rua (az'us)b (Z'uS)l /aa'Kuas/ ru
(u'a;tu)b (u'a;tu)l /un'Katu/ um rato ('tE;)b ('tE;)l /'tEKa/ terra ('mEu)b
('mEu)l /'mEKu/ melro ('tu)b ('tu)l /'tenKu/ tenro
Speakers who have (r[:]) /K/ may also have /sK, aK/ (Rr[:]), and (zz) as well: (R'r[:]uS, z'zuS)l ru; assimilation or dropping of /s, a/ is frequent with neutral
/K/ too: (a[]'us)b ([]'uS)l.
7.2.5.3. e diaphoneme /K/ also occurs in word-internal syllable-nal position
and corresponds to /K/, in Brazilian pronunciation, but to /R/, in Lusitanian
pronunciation (where it can be realized as (r) before /m, n, /); also in the Brazilian
accent /K/ can have the variant (R), which is acceptable (or even (), which however is not neutral): (']agu ']aR-)b (']aRu -gu)l /'aKgu/ largo ('pOt 'pOR-)b
('pOt)l /'pOKta/ porta ('am 'aR-)b ('aRm 'ar-)l /'aKma/ arma ('kani 'kaR-)b
('kaRn 'kar-)l /'kaKne/ carne ('ka]s -R]-)b ('kaR]uS -r]-)l /'kaKus/ Carlos
In current rather uneducated Brazilian Portuguese, /K/ () may drop (possibly lengthening a little the preceding vocoid), above all in innitives; instead, in
Lusitanian Portuguese, /K/ is /R/, often devoiced, (, ), or followed by () (more
or less short devoiced): (fa'ze -e; -e)b (f'zeR, -, -, -R, -R)l /fa'zeK/ fazer ('ma)b
('maR, -, -, -R, -R)l /'maK/ mar
Several examples show that, in Lusitanian pronunciation, we have () in contact with voiceless C
270
Laterals
7.2.6.1. ere are two lateral phonemes, in both accents, /, L/. For the former
we prefer the symbol //, to a more generic /l/, since even before V it is realized as
a semi-velarized alveolar, (]) (or () as well, completely velarized alveolar, even
if it can also be (l), which used to be the traditional neutral Lusitanian pronunciation, but only optional nowadays).
After V before a pause or a C in the Brazilian accent, velarized alveolar rounded () occurs; it has a frequent vocalized variant, (u) which, however, is hardly
neutral often mistaken for ().
In the Lusitanian accent, we have () (with the possible uvularized variant, (l)):
(']eICi)b (']It)l /'eite/ leite ('k]a;Ru)b ('k9a;-)l /'kaRu/ claro (a'zu ]is'ku;Ru)b ('zu ]ZS'ku;Ru)l /a'zu is'kuRu/ azul curo ('sAtu 'sAutu)b ('sAtu)l /'satu/ salto (e'RA
-Au)b (Z'RA)l /Ze'Ra/ geral ('ta;Lu) /'taLu/ tao
ile in Brazilian pronunciation, as already said, /, 0/ may become (u); in
Lusitanian pronunciation, // may be followed by a more or less short devoiced
() in contact with voiceless C
Structures
7.3.0. e interesting subjects, for this part, are: metaphony, words in connected speech, stress and intonation.
Taxophonics
7.3.1.1. In the Portuguese vowel system, there is an important phenomenon to
consider: vowel adjustment (or metaphony), which concerns the timbre of
stressed vowels according to the vowels occurring in the endings.
It is a phenomenon of diachronic origin, going back to the archaic phase of Portuguese, under the inuence of Latin endings. e native speakers, either Brazilian
or Lusitanian, use it automatically and coherently, even if it operates in an incomplete and sectorial way, further complicated by a number of exceptions.
For foreigners, it is one of the major obstacles to achieving a good Portuguese
pronunciation; actually, it would be of fundamental importance to be able to rely on a pronouncing dictionary, with diaphonemic transcriptions (in order to deal
with both accents simultaneously), considering metaphony as well.
7.3.1.2. Simplifying our exposition a little, we may say that metaphony operates in partially dierent ways with verbs and non-verbs ( nouns adjectives and
some pronouns)
Furthermore, a distinction is to be made between e and o For non-verbs with
stressed e the endings -o os /-u, -us/ may cause the closing of timbres: (ka'pe;]u[s])b
(k'pe;]u, -S)l /ka'peu[s]/ capelo(s) but (ka'pE;][s])b (k'pE;][S])l /ka'pEa[s]/ capela(s)
7. portuguese
271
However, not all feminine forms have /E/; actually, very often it is not so: ('ze;bR)b (-BR)l /'zebRa/ zebra ('se;d)b (-)l /'sea/ seda on the other hand, not all
masculine forms have /e/ either, so we nd: ('bE;]u) /'bEu/ belo compared to ('ne;gRu)b (--)l /'negRu/ negro (plural and feminine as well).
7.3.1.3. For nouns with stressed o only -o /-u/ (m. sg.) may cause closing: ('poku)b ('poku)l /'poKku/ porco while we have ('pOkus)b ('pOkuS)l /'pOKkus/ porcos
and ('pOk[s])b ('pOk[S])l /'pOKka[s]/ porca(s) as well; besides: ('no;vu) /'novu/ novo but ('nO;vus)b (-S)l /'nOvus/ novos and ('nO;v[s])b (-[S])l /'nOva[s]/ nova(s) (fo'mo;zu)b (fR-)l /foK'mozu/ formoso but (fo'mO;zus)b (fR-, -S)l /foK'mOzus/ formosos and (fo'mO;z[s])b (fR-, -[S])l /foK'mOza[s]/ formosa(s) however, we nd: (is'po;zu[s])b (?S-, -u, -S)l /es'pozu[s]/ poso(s) and (is'po;z[s])b (?S-, -[S])l /es'poza[s]/ posa(s) but (is'pO;zus)b (?S-, -S)l /es'pOzus/ posos for the couple.
Nevertheless, we also nd several cases with no variation, such as: (a'do;bu[s])b
('o;Bu, -S)l /a'obu[s]/ adobo(s) ('godu[s], -[s])b (-Ru[S], -[S])l /'goKu[s], -a[s]/
gordo(s) -a(s) Of course, there are cases with /O/ as well: ('mO;du[s])b ('mO;u, -S)l
/'mOu[s]/ modo(s) ('fO;ku[s])b (-ku[S])l /'fOku[s]/ foco(s) and even feminine forms
with /o/: ('go;t[s])b (-[S])l /'gota[s]/ gota(s) ('fos[s])b ('fos[S])l /'foKsa[s]/ fora(s)
7.3.1.4. For verbs the endings which cause closing are -o -a - -am /-u, -a, -as,
-aun/ (for second-conjugation forms with stress on the stem): ('de;vu, 'mo;vu {-,
-s, -Au})b ({-, -S, -4u})l /'evu, 'movu {-a, -as, -aun}/ devo movo
In checked syllables in N the eect is neutralized: ('vndu, 'v~i)b ('vndu,
'vnd)l /'vendo, 'vende/ vendo vende however, in unchecked syllables, Lusitanian
pronunciation maintain the dierence (contrary to Brazilian pronunciation due
to nasalization): ('t;mu, 't;mi)b ('te;mu, 'tE;m)l /'temu, 'tme/ temo teme ('k9;mu,
'k9;mi)b ('ko;mu, 'kO;m)l /'komu, 'kme/ como come
For the same reason, nouns behave in the same way: ('s;no)b ('sE;nR)l /'snjK/ snior/snior ('k9;miku)b ('kO;mku)l /'kmiku/ cmico/cmico (with far from
unquestionable consequences on spelling, which is still overestimated).
Lastly, we have the pronouns ('e;]i[s], 'E;][s])b (-, -S, -[S])l /'ee[s], 'Ea[s]/ ele(s) ela(s) ('esCi[s], 'Est[s])b ('eSt, -?S, 'ESt[S])l /'este[s], 'Esta[s]/ te(s) ta(s) and their derivatives. For other indications, and exceptions, good grammars answer the purpose
quite exhaustively (especially if they are less recent).
7.3.1.5. In colloquial Brazilian Portuguese (in common not lofty words), another type of vowel adjustment is frequent; it is a synchronic phenomenon which
may cause pre-stressed e o to be realized as /i, u/ (i, u), often (I, U) ( 7.1), when
the next stressed vowel is /i, u/: (me'ni;nu, mi-, mI-)b (m'ni;nu)l /me'ninu/ menino
(&a]e'gRi, &a]i-, &a]I-)b (&]'Ri)l /ae'gRia/ alegria (so'i;zu, su-, sU-)b (s'i;zu)l /su'Kizu/ sorro (ve']u;du, vi-, vI-)b (v']u;u)l /ve'uu/ veludo
However, the same vowels may be realized as (, ) when the stress is on opener V\ (e']O;u, -)b (']O;Zu)l /Ke'OZju/ relgio (ko'E;tu, k-)b (k'E;tu)l /ko'KEtu/ correto/correcto
A systematic use of /e, o/ may give the impression of meticulous attention (or,
272
perhaps, formality); on the other hand, a methodical use of (i/I u/U , ) would
certainly produce something excessive, and strange or foreign, since lofty or rare
words must remain unchanged. Furthermore, we have for instance: foia /fo'LiNa/ (fu'Li;N)b (f'Li;N)l calendar, but (fo'Li;N)b (f'Li;N)l small leaf, which
is felt as a derivative, contrary to the other.
Words in connected speech
7.3.2.1. Clusters such as // (where the rst V is word-nal {even in unstressed
monosyllables: me te se e e e de o do, no} + initial V] produce some simplications within intonation groups, not only in rhythm groups.
Generally, /e, i/ and /u/ are realized as /j, w/, respectively, or are dropped: (en'tRe;]is)b (en'te;]S)l /entRe'ees/ entre el ('da;gj, 'a;-)b ('da;j)l /e'agwa/ de %a ('oUtR 'koiz, 'o;-)b ('o;t 'koiz, 'oU-)l /i'otRa 'koiza/ e outra coa (si'k]i;n)b (s~'k9i;n)l /sein'kina/ se inclina
Further examples: ('to; d[j]esCi'tmpu)b ('to; [j]eSt'tpu)l /'tou este'tempu/
todo te tempo (sa'vi;Ris)b (s'vi;RZS)l /sia'viRes/ se a vir (&ist[j]u'oUbu, -'o;-)b
(?St[]u'o;Bu, -oU-)l /estuEun'Kobu/ to um roubo (A'mi; gjita']A;nu)b (j'mi; jt'];nu)l /ua'migu ita'janu/ o amigo italiano
More: (nAum'po; '[]i)b (n4u'po; '[]iR)l /naumpoe'iK/ no pde ir ('muin
t[j]a']E;gRi)b (k'muin t[]']E;R)l /keE'muintu a'EgRe/ e muito alegre ('i;si L[]a'e;]i)b ('di;sZ L[]'e;][])l /'iseLe a'ee/ dse-e a ele ('deI[is] 'ta)b ('dIS[?S] 'taR)l /'eiSe
is'taK/ deixe tar ('u; R[]i'd9;nu)b ('Zu; R[]i'O;nu)l /'ZuRi i'nju/ jri idneo/idneo
Besides: ('be;b[j]u 'to;du)b ('be;B[j] 'to;u)l /'bebuu 'tou/ bebo-o todo ('t[j]u;zz
'muin 't[j]i;su)b ('t[]u;zZ 'muin 't[]i;su)l /tu'uzas 'muintu 'isu/ tu muito so
('gRA~ o'o, -n do-)b ('gRAn d[]'oR)l /'gRande o'KoK/ grande horror (An't9;nu)b
(j4n'tO;nu, 4n-)l /uan'tnju/ o Antnio/Antnio
7.3.2.2. en the V in contact are two /a/, in Lusitanian pronunciation we
have /aa/ = (a) /A/ (with greater lengthening, especially if one of them is stressed;
however it is kept only for clarity): ('to;d [a]'~Ci)b ('to;a 'Znt)l /'toa a'Zente/
toda a gente (aA'mi;g, A;'m-, ;'m-)b (a'mi;)l /aa'miga/ a amiga (a'a;gj)b ('a:j,
'a;-)l /a'agwa/ a %a (a[;]'a;gj)b ('a:;j, a'a:-)l /A'agwa/ %a ('vi[] An'da)b ('vi
n'daR)l /'via an'daK/ vi-a andar (a'Am)b ('A;m, 'A-)l /a'ama/ a alma (a'Am,
a;'A-)b ('A:m, a'A;-)l /A'ama/ alma
Further cases: (&]a'E;R, ']E;R)b (&]'E;R, ']E;R)l /Ea'ERa/ ela era (u'mO;R, &uma'O-)b (u'mO;R, &um'O-)l /uma'ORa/ uma hora (&aoRa'sAu, &Ra-)b (&R'sAu, &R-)l /auRa'saun/ a orao ('da;vau, -v)b ('da;vu, -v)l /'avau/ dava-o ('p9;N u&seua'pEu,
'p9; N&seu-)b ('po;N u&seuS'pEu, 'po; N&seu-)l /'poNa useuSa'pEu/ poa o seu >apu
('nuk o'vi fa']a 'ni;su, 'nu k'vi)b ('nu~k o'vi f']aR 'ni;su, 'nu~ k'vi)l /'nunka 'vi fa'aK 'nisu/ nunca ouvi falar nso
As already mentioned ( 7.2.1), /[]/ followed by /, , / are respectively realized as ([], [], []), where () is dropped but nevertheless protects the V from elision: (']A natu'RA)b (']A nt'RA)l /'an natu'Ra/ l natural (&kU-
7. portuguese
273
ni'gI)b (-3I)l /kounnin'gein/ com nin%m (']A A~'Ci;g, ']A;~ 'C-)b (']A 4n'ti;,
']A;n 't-)l /'an an'tiga/ l antiga (kU'As)b (kU'As)l /koun'ansja/ com nsia (']A
a'zu, ']A; 'z-)b (']A 'zu, ']A; 'z-)l /'an a'zu/ l azul (kU'i;su)b (-su)l /koun'isu/ com so (u'9;meI 'Atu, '9;-)b (u'O;mI 'Atu, 'O;-)l /un'mein 'atu/ um homem alto (se'qE;])b (s'qE;])l /sein'Ea/ sem ela (nA'E)b (n4'E)l /naun'E/ no
7.3.2.3. en the syllables in contact have identical or similar C we can nd
some geminate, due to the dropping of the vowel element: ('fi;[i] k'mi;gu)b ('fik[]
k'mi;u)l /'fike ko'migu/ e comigo (da&ke][i]']a;du)b (d&ke][]']a;u)l /a'kee 'au/ daele lado ('k9;m[i] 'muintu)b ('kO;m[] 'muintu)l /'kme 'muintu/ come muito ('k9mpRuCi 'tu;du, -t 't-)b ('k9put[] 'tu;u)l /'kompRute 'tuu/ compro-te tudo
('kAmp[u] peRi'go;zu)b ('kAp[u] pRi'o;zu, -p pi-)l /'kampu peRi'gozu/ campo perigoso
More: (us'peIi zis&tAuna'dAndu, -eI is-)b (uS'pISZ zZS&t4un'Andu, -IS S?S-)l /us'peiSe zistaunna'andu/ os peix to nadando ('oUv[i] 'vO;zis, 'o;-)b ('o;v[] 'vO;zZS, 'oU-)l
/'ove 'vOzes/ ouve voz ('pa;s[i] 'se;du)b ('pa;s[] 'se;u)l /'pase 'seu/ pse cedo ('pE;[i] 'tu;du, - 't-)b ('pE;[] 'tu;u, - 't-)l /'pEe 'tuu/ pede tudo ('s~C[i] 'bu;L, -n
'b-)b ('snt[] 'Bu;L, -n 'b-)l /'sente 'buLa/ sente bua
Chiey when the syllables contain /t, /, one of them may drop completely: (&faku'da;[i] i']e;tRs, -a i-)b (&fk'da;[] ']e;tS, -a -)l /faku'dae e'etRas/
faculdade de Letr ('pO;[iz &]dez]i'ga, 'pOd d-)b ('pO;[Z &]Z]i'aR, 'pO -)l /'pOea
eai'gaK/ pod dligar ('kA[du] i'kA;n)b ('kA[du] d'k;n)l /'kadu e'kana/ caldo de cana
More examples: ('~[Ci] i'fO;R)b ('Znt 'fO;R, -n d-)l /'Zente e'fORa/ gente de
fora ('~[Ci] i'ReIt)b ('Znt 'RIt, -n d-)l /'Zente e'Reita/ gente direita also:
('ba;[Ci]Ci, -CCi)b ('ba;[t]t, -tt)l /'batete/ bate-te ('vEs[Ci]Ci, -sCCi)b ('vES[t]t, -tt)l
/'vEstete/ vte-te
7.3.2.4. Most of all, in Lusitanian pronunciation, even in connected speech, /e/
tends to drop, after voiced C as well: (f']i;p[])l (fi']i;pi)b /fe'ipe/ Filipe (']a;v[])l (']a;vi)b /'ave/ lave on the other hand, /u/ has the same tendency: ('ti;p[u])l ('Ci;pu)b
/'tipu/ tipo (']a;v[u])l (']a;vu)b /'avu/ lavo
However, generally a fairly perceptible dierence is maintained between forms
like /'ave/ and /'avu/, since /u/, although dropped, labializes the preceding C even
if it is bilabial or labiodental. In actual fact, strictly speaking, we have: (']a;, 'ti;()l;
the same is true of all other admitted C including /s, z/, which phonetically
may then occur even before a pause (with or without lip rounding): ('sintz[])l
('sintezi)b /'sinteze/ snte (u'mAs[])l (o'mAsi)b /Ko'manse/ romance (&duvi'o;z[u], -z)l (&duvi'do;zu)b /uvi'ozu/ duvidoso ('pa;s[u], -s)l ('pa;su)b /'pasu/ pso compared to ('pa;s[])l ('pa;si)b /'pase/ pse Of course, we generally use a simpler
transcription, except when we want to lay special stress on the fact.
e sequences /0je, je/, in the Lusitanian accent, in addition to having a pronunciation which corresponds to its spelling, preserve the once, more systematic possibility of blending the last two elements: ('sE;R, -R, -R, -Ri)l ('sE;Ri)b /'sERje/
srie ('ka;, -, -a;, -a;i)l ('ka;i)b /'kaje/ caie (where lengthening may be the only
274
dierence as to ('kai) /'kai/ cai] In Lusitanian pronunciation, again, the rare instances of /i/ oscillate between /i/ (which is more modern and closer to spelling)
and (the more traditional) /e/: ('taksi, -s)l ('taksi)b /'taksi/ txi forms like /'ZuRi/ jri and /'ZuRe/ jure (which are alike in Brazilian pronunciation: ('u;Ri)b], in the Lusitanian one, may either be alike: ('Zu;R)l, or dierent: ('Zu;Ri)l (-R)l (respectively).
7.3.2.5. en, at word boundaries, /KK, / come together, they normally simplify: ('ma[] 'A)b ('ma[R] 'A)l /'maK 'Kja/ mar real ('ma[] umo'Ro;zu)b ('ma[R]
umu'Ro;zu)l /'maK Kumo'Rozu/ mar rumoroso (kA'nA[] ']impu)b (k'nA[] ']ipu)l
/ka'na 'impu/ canal limpo (kA'nA[] ]imi'ta;du)b (k'nA[] ]m'ta;u)l /ka'na emi'tau/ canal limitado
e same occurs to /ss, az, sS, aZ/, when they are alike: ('dE[s] 'sE;ku]us)b ('dES
'sE;k]S)l /'Es 'sEkuus/ dez sculos ('dEs 'a;]is)b ('dE[S] 'Sa;]S)l /'Es 'Sais/ dez xal
(a[z]'z9;ns)b (Z'zo;nS)l /aa'zonas/ as zon (&azA'nE;]s)b (&[Z]Z'nE;]S)l /aaZa'nEas/
janel however, we nd (Ss)l (s)b /ss/ in cases such as: (&?Ss']nt)l (&ise']~Ci)b
/isse'ente/ excelente (kZS'seR)l (kRe'se)b /kRes'seK/ crcer
7.3.2.6. In the Lusitanian accent, the sequences /0, 0R/ are often realized as
(0], 0R) (even if it is better to avoid such a pronunciation): ('f9oR f']- f']-)l
('f]o)b /'foK/ or ('k9a;Ru k']- k']-)l ('k]a;Ru)b /'kaRu/ claro ('g]O;R g']- g']-)l
('g]O;R)b /'gORja/ glria on the other hand, the sequences /0e, 0eR, 0o, 0oR/,
while behaving in exactly the opposite way, produce similar results, even if in this
case one can use that pronunciation without hesitation (provided one's speech
rate is not slow).
Some examples: (k'Ri k'i 'ki)l (ke'Ri)b /ke'Ria/ eria (ko'Ro k'Ro 'ko)l
(ko'Ro)b /ko'Roa/ coroa (f'ROS, f'OS, 'fOS)l (fe'ROs)b /fe'ROs/ feroz (&mR'seR, m'seR)l
(&meRe'se)b /meRe'seK/ merecer in current speech it is normal to reduce the prepositions combined with the denite article pelo(s) pela(s) to /p, p/, also in Brazilian Portuguese [p'lo po]\ (p]A'mAi, &pe]A-)b (p9'mAi, &p]-)l /p[e]a'main/ pela
me ('p]ja, p]u'a, &pe]u-)b ('p9jaR, p9u'aR, &p]u-)l /p[e]u'aK/ /p[e]u'aK/ pelo ar
Compare also: pra ('pa;R)b (-)l /'paRa/ and para (&paRa, pRa0, pa0, paR,
pR)b (&pR, p0, p0, pR, p)l /&paRa, pRa0, pa0, paR, pR/: (&paRa'ka,
pRa'ka &paRa'i, pRa'i)b (&pR'ka, p'ka &pR'ki, p'ki)l /para'ka, pRa'ka para'ki,
pRa'ki/ para c para aqui; besides, cada is generally (&kada)b (&k)l /&kaa/. ese
are restressed forms, when they occur in isolation (as it happens in metalinguistic
usage), in comparison with normal forms, which have no primary stress, but secondary or weak ones. e same holds true for a(s) da(s), na(s) (a{s}, da{s},
na{s})b ({S}, d{S}, {S})l /a{s}, a{s}, na{s}/, and even for mas (mas)b (mS)l /mas/
(with a possible secondary stress for rhythmic reasons, (&mS)l).
In Lusitanian pronunciation again, also /0es, 0es/, mainly in the ending - may
be quite reduced, up to becoming intense syllabic (however the dropping of
the V is an uneducated feature): ('f9o;RZS, -R -S)l ('f]o;Ris)b /'foRes/ or; combining
what we have just seen above, in a fairly uneducated type of Lusitanian pronunciation, we may also nd (f']o;S)l), (u'mAs?S, -s)l ('mAsis)b /Ko'manses/ romanc (&StAu'RAnt?S, &t-, -nt)l (&estAu'RA~Cis)b /Kestau'Rantes/ rtaurant
7. portuguese
275
In the Lusitanian accent, the sequences /peK, pRe/ often receive an intense C
and confuse with one another (as has happened to per%ntar in comparison to
Spanish pre%ntar] therefore (p'fItu)l may stand for (pe'feItu, pRe'feItu)b /peK'feitu, pRe'feitu/ perfeito prefeito (with additional variants (p'fItu, p'fItu)l].
7.3.2.7. In both accents, the preposition com /koun/, followed by the articles,
is reduced to /kon-/ (k), and to (ku, kj, k) as well: (ku'pai, ku-, kju-)b (ku'pai, ku-, ku-)l /kon-u'pai/ com o pai (k&umAmu'LE, ku-)b (k&ummu'LER,
ku-)l /kon-umamu'LEK/ com uma muer (kaz'mAus, kaz-, kjaz-)b (kZ'mAuS, kZ-, kZ-)l /kon-az'mauns/ com mos (kuz']o;bus, kuz-)b (-Z']o;BS,
k-)l /kon-uz'obus/ com os lobos
Even em is considerably reduced, up to /in/: (eIm&putu'ges, im-)b (I&pt'eS, i-)l /impuKtu'ges/ em portu%s (eI'ka;z, i-)b (I~'ka;z, i~-)l /in'kaza/ em
casa (&eqAu'si;]u, iAu-, q-)b (qAu'si;]u, iAu-, q-)l /in-Au'siju/ em auxlio e dash put
before a V in the diaphonemic transcription, could even be dispensed with, without creating problems, since those forms are recognizable, somehow, thanks to the
dot under the symbols too, /kon, in/. Otherwise, we could use //, as a diaphoneme (but in a systematic way, then, for all cases of //); however, this would
make the transcription heavier and the phonemic analysis more complicated.
In rhythm groups, sequenze // = (0): (u&mejA'mi;gu)b (u&mej'mi;u)l
/umeua'migu, umewa-/ o meu amigo.
7.3.2.8. In the neutral Lusitanian accent, as the various examples have shown,
we systematically nd (in addition to (, n, ~, , , 9, , , ), except in slow and
very precise pronunciation) even (, ? u), when between voiceless C, or between
them and pause, and vice versa. It is quite common to hear things like: (uS&pf'so;RZS &pt'e;zZS)l (us&pRofe'so;Ris &potu'ge;zis)b /uspRofe'soRes poKtu'gezes/ os profsor portu%
In Lusitanian pronunciation, /e, u/ are fairly regularly devoiced, even completely, up to their dropping (after voiced C as well) which occurs, for /e/, even
within words or rhythm groups: (S'kup, d?S'kup)l (is'kupi)b /es'kupe/ dculpe (&tvi'zAu, &tv-, &t]-)l (&te]evi'zAu)b /teevi'zaun/ televo (p'tE;s, &p'tE;s)l
(&ape'tE;si)b /ape'tEse/ apetece ('pOt, ?S'pOtuS)l (is'pOtus)b /es'pOKtus/ dportos
More: (S'paR, &?Sp'RaR)l (&espe'Ra)b /ispe'RaK/ perar (St'tA, &?S-)l (&ista'tA)b /ista'ta/ tatal (Sk'seR, &k-, &?S-)l (&iska'se)b /iska'seK/ cser (&'B;Nu, )l (e'bA;Nu)b
/Ke'baNu/ rebao (&supuRi'a[] u'sO, &sp&Ru-)l (a&supeRoRi'da[i] du'sO, -'dad
du-)b /asupeRjoRi'ae u'sO/ a superioridade do sol (u'fin s';n, s-, ds-)l (u'fi~ ise'mA;n)b /u'fin dese'mana/ o m-de-semana (p'RE;s um'Bo i';, p'RE;s
u-, p'RE;s m-, p'RE;s mu-)l (reading: (pa'RE;semi uma'boa i'dE;, -se mqu-)b} /pa'REseme uma'boa i'Eja/ parece-me uma boa ideial/idiab
7.3.2.9. We will now consider some examples of /0u/ = () (the symbols for
/0e/ are more normal, as in ('knt)l ('k~Ci)b /'kente/ ente]\ ('a;t[[])l ('a;Cimu[s])b /'atimu[s]/ timo(s) ('p9;[])l ('p]A;nu[s])b /'panu[s]/ plano(s) ('b;[])l ('bA;-
276
Nu[s])b /'baNu[s]/ bao(s) ('t([])l ('tmpu[s])b /'tempu[s]/ tempo(s) ('ti;[])l ('tRi;bu[s])b /'tRibu[s]/ tribu(s) ('paT[])l ('patu[s])b /'paKtu[s]/ parto(s) ('sa;k[])l ('sa;ku[s])b
/'saku[s]/ saco(s) (']a;)[])l (']a;gu[s])b /'agu[s]/ lago(s)
Furthermore: ('tu;5[])l ('tu;fu[s])b /'tufu[s]/ tufo(s) ('po;, 'pO;)l ('po;vu, 'pO;vus)b
/'povu, 'pOvus/ povo(s) ('po;s, 'pO;s)l ('po;su, 'pO;sus)b /'posu, 'pOsus/ poo(s) ('ka;z[])l
('ka;zu[s])b /'kazu[s]/ co(s) ('o;[])l ('o;u[s])b /'KoSu[s]/ roxo(s) ('bI[])l ('beIZu[s])b
/'beiZu[s]/ beijo(s) ('m;[])l ('me;u[s])b /'meju[s]/ meio(s) ('ka;[])l ('ka;Ru[s])b /'kaRu[s]/ caro(s) ('ka;R[])l ('ka;u[s])b /'kaKu[s]/ carro(s) ('o;3[])l ('o;]u[s])b /'Kou[s]/ rolo(s) ('fi;[])l ('fi;Lu[s])b /'fiLu[s]/ o(s)
In neutral Brazilian pronunciation, devoicing is quite rare; nevertheless, it may
occur, mainly after voiceless C before a pause, in particular for /e, u/: ('k~Ci, C,
-Ci)b ('knt)l /'kente/ ente ('po;su, -s, -su)b (-su)l /'posu/ poo
7.3.2.10. An oral V followed by a nasalized V may be nasalized, especially in
current pronunciation: (&impRe'sAu, &Aim-)b (&ip'sAu, &4i-)l /aimpRe'saun/ a
imprso ('impu, im-)b (-pu)l /E'impju/ mpio
Stress
7.3.3.1. e position of stress is fairly well indicated in spelling, once we know
its rules, which are explained in all grammars. From a phonetic point of view, some
secondary stresses occur. ey are distributed alternatively, for rhythmic reasons,
in a similar way as they occur in Spanish or Italian.
Generally, monosyllabic grammemes (and the like) are unstressed (except for
rhythmic secondary stresses, of course): articles [o a os um uns] personal pronouns [me te se o a e nos vos os and combinations mo ta os {};
eu tu as well, if weak), possessive adjectives [meu te sua {}), prepositions [a
com de em por sem sob] prepositions with articles [ ao da do na no num
{}, pro/pr = para o {}, plo/p'lo = pelo {}), conjunctions [e ou m nem
e se] the relative pronoun e auxiliaries [sou s so so tens tem tm
(es)tou (es)ts (es)t] forms of tratamento [dom frei so seu = senhor), others [cem gro o to impersonal h]
7.3.3.2. en there are (actual) lexical compounds, which maintain a fairly strong
stress on their rst element, uctuating between a primary and a strengthened secondary one (according to the scale ('), (), (&)); so we have a rst /'/ followed by another which is realized as () (or (&), if adjacent to the primary stress): (gjada'oUp, -o;-)b (-R'o;p, -oU-)l /'gwaKa'Kopa/ %arda-roupa (agja'fOCi)b (aj'fOt)l /'agwa'fOKte/ %a-forte (ACi'baiu)b (At'BaiSu)l /'ati'baiSu/ altibaixo (&mA'kRa;du)b (&mA'ka;u)l /'ma'kRjau/ malcreado (pAu~i']O)b (pAund']O)l /'paunde'O/ po-de-l (]uzobRazi']eIRu)b (-BRzi']I-)l /'uzbRazi'eiRu/ lo-brileiro
ere are even some particular compounds, still considered as such (rather than
crystallizations), which maintain the two elements fairly independent; rst of all,
we nd adverbs in /-'mente/ -mente: (ustA'm~Ci)b (ZuSt'mnt)l /'Zusta'mente/ j-
7. portuguese
277
tamente (]indA'm~Ci)b (]ind'mnt)l /'inda'mente/ lindamente (&fRiA'm~Ci)b

(&fi'mnt)l /'fRia'mente/ friamente (ko&tez'm~Ci)b (k&teZ'mnt)l /koK'tea'mente/ cortesmente (fasi'm~Ci)b (fas'mnt)l /'fasi'mente/ facilmente
7.3.3.3. In addition, nouns with the diminutive inx /-z-/ -z- follow the same
pattern: (mu&L'zi;N)b (mu&LR'zi;N)l /mu'LEK'ziNa/ muerzia (ku&L'zi;N)b
(k&LR'zi;N)l /ku'LEK'ziNa/ coerzia (fA'zi;N)b (f4'zi;N)l /'OKfan'ziNa/ orfzia (avena'zi;N)b (Savn'zi;N)l /'Savena'ziNa/ >avenazia (a&v'zi;N)b (&v'zi;N)l /a'vO'ziNa/ avozia (a&vo'zi;Nu)b (&vo'zi;Nu)l /a'vo'ziNu/ avozio (&pai'zi;Nu)b (-'zi;Nu)l /'pai'ziNu/ paizio (9meI'zi;Nu)b (mI'zi;Nu)l /'mein'ziNu/ homenzio (9meIza'Au)b (mIz'Au)l /'meinza'Kaun/ homenzarro (a&pa'ze;Lu)b (&pa'z;Lu)l /Ka'pa'zeLu/ rapazeo (]e&s'zi;tu)b (]e&s'zi;tu)l /en'sO'zitu/ lenolzito (Once, a graphic grave accent was used in such compounds.)
As to the (phonic) stressing of verbs, it is important to examin the following
examples well; they are very signicant and certainly not free from serious doubts
when consideredmerely from an orthograc point of view; they are given in simple diaphonemica transcription: /li'mitu, li'mita{s}, li'mitaun/ limito, limita(s)
limitam; /'bailu, 'baila{s}, 'bailaun/ bailo, baila(s), bailam; /'kauzu, 'kauza{s}, 'kauzaun/ causo, causa(s), causam.
Besides: /aK'kwaK/ arcuar, /aK'kuu/ arcuo, /aK'kua{s}/ arcua(s); /in'flwiK/ inuir,
/in'fluu/ inuo, /in'flui{s}/ inui(s), /in'flwi/ inu, /in'flwiu/ inuiu, /in'flwia{s}/
inua(s); /pa'sjaK/ passear, /pa'seju/ passeio, /pa'seja{s}/ passeia(s), /pa'sejaun/ passeiam; /'KiK/ rir, /'Kiu/ rio, /'Kiu/ riu, /'Kia{s}/ ria(s), /'Kiein/ riem, /'Kiaun/ riam; /i'zia/ dizia, /i'Ria/ diria, /i'Rei/ direi.
Now, let us observe very carefully the following: /sa'iK/ sair, /sa'iu/ sado, /sa'indu/ saindo, /'sai{s}/ sai(s), /sa'i{s}/ sa(s), /'saju/ saio, /'saja{s}/ saia(s), /sa'iu/ saiu,
/sa'ia{s}/ saa(s), /sa'iaun/ saam, /'saein/ saem, /sa'iRaun/ saram, /sai'Riaun/ sairiam,
/sai'Raun/ sairo, /sai'Ria{s}/ sairia(s), /sai'Rei{s}/ sairei(s), /sa'iReis/ sareis, /sai'Rieis/
saireis.
Lastly, we nd inxed futures and conditionals (which are quite odd {for
foreigners or Brazilians, too}): (&se']jI)l /'se'wei/ s-lo-ei (t&ta'sa)l /tRa'taK'sja/
tratar-se- (&diR'LjaS)l /'iK'Lwas/ dir-o-s (kn&ta']jAu)l /kon'ta'waun/ cont-lo-o
(&faR'L[]i)l /'faK'Ljia/ far-e-ia (pu&e's[]i)l /po'eK'sjia/ poder-se-ia (&di']jiS)l /'i'wias/ di-lo-i (kn&ta']i4)l /kon'ta'wian/ cont-lo-iam (reading: (&se']jeI, tRa&ta'sa, &i'Ljas, kn&ta']jAu, &fa'L[]i, po&de's[]i, &i']jis, kn&ta']iA)b).
Intonation
7.3.3.4. 7.4-5 show the preintonemes and intonemes of the two neutral Portuguese accents. It is important to make comparisons both between them and with
those of other languages:
/./: (t;Nu Ami;gu muintu simpa;Ciku3 3)b ('t;Nu 'mi;u muintu sipa;tiku3 3)l /'teNu un-a'migu 'muintu sim'patiku./ Teo um amigo muito simptico.
278
/?/: (&useNo; RmE;iku12)b (&usZ'No; R'mE;ku21)l /use'NoK E'mEiku?/ O seor

mdico?
//: (&AmANA sa;badu2 2| odmigu3 3)b (&am'NA 'sa;Bu2 2| oumigu3 3)l /Ama'Nan E'sabau oo'mingu./ Ama sbado, ou domingo?
7.4. Brazilian preintonemes and intonemes.
/ / (2 2 2 2 2 2 2)
/./ (2 3 3)
/ / ( 2 2 2 2 2 2 2)
/?/ (2 1 2)
/ / ( 2 2 2 2 2 2 2)
// (2 2 2)
/ / ( 2 2 2 2 2 2 2)
/,/ (2 ' 2)
7.5. Lusitanian preintonemes and intonemes.

/ / (2 2 ' 2 2 ' 2 2 2)
/./ (2 3 3)
/ / ( 2 2 ' 2 2 ' 2 2 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 2)
// (2 ' 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 2)
/,/ (2 ' 2)
International accent
7.4.1. We will now provide the international pronunciation of Portuguese.
Of course, it is based more on the Brazilian accent (rather than on the Lusitanian
one, with all its phonetic idiosyncrasies), also if we consider the actual number of
its native speakers. However, even in respect of the Brazilian accent, its most peculiar characteristics are to be put aside (within Brazil itself), from a more international point of view or, rather, point of hearing.
erefore, an International pronunciation of Portuguese does not relate to the
Carioca pronunciation of Rio de Janeiro, but more generally to that of So Paulo
City and other cities in the southern states of Brazil. As a matter of fact, this kind
of pronunciation moves less away from spelling, which is a great advantage internationally.
7.4.2. Obviously, native speakers (Lusitanian as well) will understand it without problems. But it is not the same to foreigners faced with Lusitanian pronuncia-
7. portuguese
279
/i/ (i, i)
/u/ (u, u)
/e/ (e, e)
/o/ (o, )
/E/ (E, )
/O/ (O, )
/a/ (a, A)
/ai, au/ (ai, au Ai, Au)
/ / (2 2 ' 2 2 ' 2 2 2)
/./ (2 3 3)
/?/ (2 ' 1 2)
// (2 2 2)
tion. However, in teaching and learning, it is important to have the opportunity

to be able to show a kind of pronunciation which is not the mere result of approximative and partial study (subject to so many interferences from other languages,
also because of very dierent orthographic rules, as generally happens when foreigners study the Portuguese language with no adequate phonic method). On the
contrary, this results from a long reection, based on phonic naturalness, but also
from true typicalness (which does not sound foreign).
Perhaps, this accent might add some local peculiarities from the two neutral
accents (either Brazilian or Lusitanian, including some regional variants), thus
making it more native-like, although many native speakers do not nd the international one to be strange at all. To Lusitanians it might sound rather Brazilian-like, but with no typical Southern, Central, or Northern peculiarities.
7.4.3. As shown in the gure, the vowels are (i e 'E, a 'O, o u) /i, e, E, a,
O, o, u/ (we still use the diaphonemic transcription as in the preceding chapters):
('vi, 'e;li, &la'dE;vi, 'ga;ta, 'pO;su, &psu'daR, 'o;ku, 'u;va) /'vi, 'ee, Ea'Eve, 'gata, 'pOsu,
pOsu'aK, 'oku, 'uva/ vi ele ela deve gata posso posso dar oco uva; including their
nasalized taxophones (i, e, A, , u): ('si, pen'denti, 'kAntu, 'knta, 'mundu) /'sin,
pen'dente, 'kantu, 'konta, 'mundu/ sim pendente canto conta mundo. Of course,
the phonemes /e, E O, o/ must be keep distinct, although they are troublesome
for foreiners, including the other thorny vocalic problem metaphony otherwise,
we would really have a foreign pronunciation. Written e o, in unstressed syllables, always correspond to /e, o/ (even for esC-); while, only for nal -e(s) -o(s), we
have /e, u/: ('lentis, 'mu;Rus) /'entes, 'muRus/ lentes muros
7.4.4. e diphthongs are simpler, as well. In fact, they all have (i, u) as second
elements: (ei, Ei, ai, Oi, oi, ui iu, eu, Eu, au, ou), /ei, Ei, ai, Oi, oi, ui iu, eu, Eu, au,
ou, o/: ('rei, 'vou) /'Kei, 'vo/ rei vou; with their nasalized taxophones (ei, Ai, i,
ui Au): ('bei, 'mAi, 'pi, 'muintu, 'nAu) /'bein, 'main, 'poin, 'muintu, 'naun/
bem me pe muito no. Certainly, () would not be a feasible proposition for
the nasalized diphthongs, or for nasalized /a/. ey are too far away from native-speaker's reality (although this would certainly not prevent understanding).
On the other hand, a kind of pronunciation which is still international, but a
little less genuine, might easily renounce the nasalization caused by // in checked
280
syllables all the more so that, in many types of Lusitanian pronunciations, this
nasalization is really very reduced, so that it corresponds to the merely phonetic one
which is present in // sequences, in most languages without distinctive/phonemic nasalization. Normally, this slight degree of nasalization is not indicated.
7.4.5. Beside the ocial phonemic diphthongs, we nd a new series of diphthongs, derived from the vocalized realizations of //l0, l// sequences (diaphonemically indicated as /0, /). By explicitly adding only the one which
does not already coincide with the ocial diphthongs, we have (Ou): as in
(bRa'ziu, 'feutRu, 'mEu, 'sau, 'sOu, 'soutu, 'suu) /bRa'zi, 'fetRu, 'mE, 'sa, 'sO, 'sotu,
'su/ Brasil feltro mel sal sol solto sul.
Actually, these diphthongal realizations are more convenient, for their articulatory simplicity and for better understanding spoken Portuguese, because they prepare foreigners for the interpretation dilemma of // as (u) (coinciding with
many phonological /u/ diphthongs).
7.4.6. As to consonants, as already said, our international pronunciation does
not make use of particular taxophones, which are peculiar of native accents of Portuguese, but are not sustained by a suciently widespread natural universality.
us, we regularly nd the assimilation of /n0/, always with full nasal contoids
(not attenuated, as neutral semi-nasals are, instead), before any consonants: ('sAmba, 'fRA~Za, 'fRAngu) /'samba, 'fRanZa, 'fRangu/ samba franja frango. In addition, in
word-nal position, () is normal, even before vowels or nasals at the beginning
of a following word: (nAu'E, 'lA a'zuu, u'o;vu, u'mau) /nau'E, 'a a'zu, un'ovu, un'ma/ no l azul um ovo um mal Note that in this only case the neutral
pronunciations, both Brazilian and Lusitanian, have ().
Compare also ('wo;vu, u'mau) /u'ovu, u'ma/ o ovo o mal.
7.4.7. In the international pronunciation of Portuguese, we do not have either
the Brazilian taxophones of /t, d k, g/ + /i, j/, (C, , ): (di'kinta) /de'kinta/ de
quinta; nor the continuous taxophones of /b, d, g/ (of Lusitanian, (B, , ), indicated as /b, , g/): (&obRi'ga;du) /obRi'gau/ obrigado. Even more so, /s{}0, z{}0/ e
/s/ (diaphonemically /s, a/) are exactly realized as (s, z), without changing their articulations: ('pas, 'dezdi) /'pas, 'eae/ paz desde (which are so remarkable in Lusitanian: ('paS, 'deZ) and in Carioca: ('pa, 'dei)). us, we only have (S, Z) /S, Z/
for written ch x j g: ('Sa, 'Si;kaRa, 'Za, Ze'Rau) /'Sa, 'SikaRa, 'Za, Ze'Ra/ ch xcara j
geral
7.4.8. Also the approximants, more naturally, are such, (j, w) /j, w/ (not semi-approximants, (, j)): ('glO;Rja, 'ligwa) /'gORja, 'ingwa/ glria lngua As to /R/,
of course, we have (R) (even in word-nal position before a pause or C, where in
neutral Bazilian we nd () /K/). ile, for /K/, a more natural (r) is better (after C\ /n, l, s/, diaphonemically /n, , s/): ('nra, &ura'pas, 'biuru, ra'laR) /'onKa, unKa'pas, 'biKu, Ka'aK/ honra um rapaz bilro ralar, and (Rr), after V: ('kaRru, &uRra'pas) /'kaKu, uKa'pas/ carro o rapaz en, it occurs even in word-initial position;
7. portuguese
281
while we have ()b, ()l (with so many other variants, 7.2.5.2).

Besides, /L/ is maintained ('vE;Lu) /'vELu/ velhu; but, more naturally, we have
both (l) and (u) (as already seen), which diaphonemically are //): ('ljau) /'ja/
leal.
Stress(es) and duration coincide with the neutral ones; while intonation is a
compromise between the two neutral types. To simplify a little, the gure gives
only the unmarked preintoneme and the three marked intonemes.
Text
7.5.0. e transcribed passage, e North Wind and the Sun, given in six dierent normalized versions. We start with the (Brazilian and Lusitanian) neutral
pronunciations of (neutral British) English this is the rst step of the phonetic
method (the written text is given in 2.5.2.0). e Portuguese translation follows,
in its neutral Brazilian, Lusitanian, and international Portuguese versions. e
variants given in brackets might need slight adjustments, on which the reader is
invited to reect, according to the occurring segments.
In the text, for the Lusitanian accent, we indicate (|) /a/, before a pause (even
though in the isolated examples given we simply put (), in order to better show
the dierence with the Brazialian accent).
At the end, as usual, the version showing the English pronunciation of Portuguese is given; it represents the habits of neutral British speakers, who are uent
in Portuguese (after prolonged contact with native speakers, but with no help
from the phonetic method), who have adequately learned the relative prominences, but who substantially use segmental and suprasegmental elements which are
typical of neutral British English (although, of course, a neutral accent is not so
common). Obviously, the same principle is valid for the foreign pronunciations
of English, given rst.
Speakers of American English could prepare their own version both of the Portuguese pronunciation of English and of their pronunciation of Portuguese, as an
excellent exercise, by listening to native speakers, best of all after recording them.
Of course, speakers of other languages could do the same thing. e author would
be happy to receive their transcriptions and recordings, both in case of help
should they need it and to make their contribution known to others (possibly
in our website on canIPA Natural Phonetics 0.12).
Brazilian pronunciation (of English)
7.5.1. (i5nOs 'ji~i2 ~i'sA2 jis'pu;Ci 'j jzistR9ge3 3| 'jn A'tRE;ve]e2 'kI mA']92 'Epi CinA'jOm k]oUi3 3| de'gRid2\ di5jAn u'fE sk'sii2 i'meIi~ i'tRE;ve]e2 'teI is'k]oU kOf2 2| &Ubikn'side istR9ge d~a;ze3 3||
'd~2 i'nOs 'ji~i2 'b]u2 'zai2 zikui3 3| &bi'mO Rib]u2 2| i'mO 'k]oUz]i2
282
&ii'tRE;ve]e2\ 'foU is'k]oU kRAuni3 3| &ndt']asCi2\ i'nOs 'ji~i2 5geI 'vap

iatmpiCi3 3|| 'd~2 i'sA~ '9U AuCi3 3 jOm]i3 3| &~i'mi;at]i2\ i'tRE;ve]e
'tu kOf3 3 isk]oUi3 3|| n'soU2 i'nOs 'ji~i2 &jzob]ais tukM'fEs2| &disA3 3 jzistR9geR3 3 &vitu3 3||
&iu]aii12 is'tO;Ri2| u'j9n 'tji;Ri tag12|||)
Lusitanian pronunciation (of English)
7.5.2. (d'nOs 'jind[]2 nd'sA2 jRZS'pu;t jiS jZZSt9gR3 3| 'jn 'tE;v]2
'kI m']92 'Ep tnjOR k9oUk3 3| d'Ri2\ dE'jAn fES sk'si;2 i'mIkn d'tE;v]2 'tI k?Sk9oU 'kOf2 2| &Sbk'si;R ZSt9gR nda;R3 3||
'dn2 d'nOs 'jim2 'b]u2 'zaR2 zku3 3| &ba'mO Ri'B]u2 2| d'mO 'k9UZ]i2 &i'tE;v]R2\ 'fU dZSk9oUk RAundi3 3| &nd't9aSt2\ d'nOs 'ji2 'gI vap itmpt3 3|| 'dn2
d'sA S9U Aut3 3 jORm]i3 3| &ndi'mi;t9i2\ d'tE;v] tu kOf3 3 ?Sk9oUk3 3|| 'soU2
'nOs 'jind[]2 &jzu'B]aiS tuk'fEs2| &dsA3 3 jZZSt9gR3 3 &vtu3 3||
&diu']aik21 dZS'tO;Ri2| &duj9n'ti;R t'21|||)
Portuguese text
7.5.3. O vento norte e o sol poravam sobre al dos do era o ma forte, ando sucedeu psar um viajante envolto numa capa. Ao v-lo, pem-se de acordo em como aele e primeiro conse%se obrigar o viajante a tirar a capa seria considerado o ma forte.
O vento norte comeou a soprar com muita fria, m anto ma soprava, ma
o viajante se embruava na sua capa, at e o vento norte dtiu. O sol briou
ento com todo o plendor, e imediatamente o viajante tirou a capa. E sim o vento norte teve de recoecer a superioridade do sol.
Voc gostou da htoria? Queremos repet-la?
Brazilian pronunciation
7.5.4. (uvntu nOCi u'sO2 pofa;vAu &sobRikjA duz'dois2\ R9&maisfOCi3 3|
kAndu &susedeu pasa; Ruva'A~Ci2\ evotu &numaka;p3 3| 've;]u2 'p9i[I]si
a'kodu2 eIk9; mjake;]i &ipRimeIRu2 &ksei; sobRiga; Ruva'A~Ci2 aCiRa;
Raka;p2 2 se&Riak&sideRa;du2 u&maisfOCi3 3||
uvntu 'nOCi2 &kmesoU aso'pRa2 kmuint fu;R3 3| maskAntu mai[s] sopRa;v2 2| mai zuva'A~Ci2 &simbRuLa;v &nasjaka;p3 3| a'tE2 uvntu 'nOCi2 &izisCiu3 3|| usO bRiLoU en'tAu2 knto;du [u]sp]endo3 3| &imeatA'm~Ci2 uvaA~[Ci] CiRoU3 3 aka;p3 3| a'si2 uvntu nOCi2 2\ te;vi &ie&kNe'se2\ a&supe&RoRida;[i]3 3 dusO3 3||)
vo'se gostoU12 &daisto'Ri;N2 keR;mu[s] &epeCi;]12|||)
7. portuguese
283
Lusitanian pronunciation
7.5.5. (u'vnt nOt 'sO2 p'fa;v4u &soBRkA duZ'oiS2\ R&maiSfOt3 3|
'kAndu &sus'eu psa; Ruv'ZAnt[]2\ i'vot &numka;p3 3| 've;]u2 'p9i[I]s[] 'koRu2 I~'ko; mjke;][] &kpi'mIRu2 &ks[]'i; &sBRia; Ruv'ZAnt[]2 t'Ra; R'ka;p2 2 s&Rik&s'Ra;u2 u&maiSfOt[]3 3||
u'vnt[] 'nOt[]2 &km'so su'pa2 k'muint fu;R3 3| mS'kAnt maiS su'pa;v2 2| 'mai zuv'ZAnt[]2 &simbRu'La;v &nska;p3 3| 'tE2 ku'vnt[] 'nOt[]2 &zZStiu3 3||
'sO BRiLo en'tAu~2 kn'to; [j]Sp9endoR3 3| &imat'mnt[]2 uv'ZAn[t]2 tRo3 3
ka;p3 3| 'si2 u'vnt[] 'nOt[]2 2\ 'te;v[] &kNZ'seR2\ &sup&RoRia;[]3 3 sO;3 3||
vo'se S'to21 &[]ZSt'Ri;N2| k'Re;mu[Z] &p'ti;]21|||)
International pronunciation
7.5.6. (u'ventu 'nORti ju'sOu2 poR'fja;vAun &sobRi'kwau duz'dois2\ Rau&maisfORti3 3|
'kwAndu &suse'deu pa'sa; RuMvja'ZAnti2\ eM'voutu &numaka;pa3 3| au've;lu2 'pinsi dja'koRdu2 ei'ko; mwa'ke;li &kipRi'meiRu2 &knse'gi; sjobRi'ga; Ruvja'ZAnti2 ati'Ra; Raka;pa2 2
se&Riakn&side'Ra;du2 u&maisfORti3 3||
u'ventu 'nORti2 &kome'sou aso'pRaR2 k'muinta fu;Rja3 3| mas'kwAntu 'mais sopRa;va2 2| 'mai zuvja'ZAnti2 &sjimbRu'La;va &naswaka;pa3 3| a'tE2 kju'ventu 'nORti2 &dizistiu3 3|| u'sOu bRi'Lou en'tAu2 kn'to;du usplendoR3 3| &jimedjata'menti2 uvja'ZAnti2 tiRou3 3 aka;pa3 3| ja'si2 u'vntu nORti2 2\ 'te;vi &diRre&koNe'seR2\ a&supe&RjoRida;di3 3 dusOu3 3||
vo'se gos'tou12 &daisto'Ri;Na2| ke'Re;muz &repe'ti;la12|||)
English pronunciation of Portuguese
7.5.7. (u5vnT 'n;Ti ju's;2 ph;fi'A;vn &sb>i'khwA: Dz'Ds2\ &>u&mas'f;Ti3 3| 5khwnDu &sus'DI p'sA:> uMv'ZnTi2\ M5vTu &num'khA;p3 3|
a5vIlu2 'phnsi Di'kh;Du2 I5khm 'khIli &khIip>'m>u2 &kns5gIisi b>'gA:> uv'ZnTi2 T5>A:> 'khA;p32 s&>kn&sD'>A;Du2 u&mas'f;Ti3 3||
u5vnT 'n;Ti &km5s; s'ph>A:2 k5munT 'fu>i3 3| ms5khwnT 'mas
s'ph>A;v32| 5ma;z uv'ZnTi2 &smb>u5ljA;v &ns'khA;p3 3| 'Th;I2 ku5vnT
'n;Ti2 &Ds'sTIiu3 3|| u5s b>'j; In'Than2 kn5ThD spln'D:3 3| Ii&mDT'mnTi2 uv'ZnTi2 T'>;3 3 'khA;p3 3| Ii'sI;in2 u5vnT 'n;Ti32\ 5ThIvi D&>k~j's2\ &sup>i&;>'DA;Di3 3 D's;3 3||
v5s;I g'sT;21 &DasT'>Iinj2| kh5>Ims &>p'ThIil21|||)
8. Russian
8.0.1. e Russian pronunciation given in this chapter is the modern neutral
one, which is decidedly more relevant and less far from spelling (which we certainly cannot dene as friendly, in particular for traditional pronunciation, which
had an even greater number of inconsistencies).
In any case, we will show the dierences between modern and traditional pronunciation as well ( 8.3.2). e latter mostly coincides with the old Moscow
pronunciation. We will also give other more recent variants, which may be found,
in a more or less (un)systematic way, even among people believed to be neutral
speakers, who can oscillate among modern, traditional, and alternative variants.
However, modern neutral pronunciation is clearly denable and recognizable, and
appreciable indeed, although frequently mixed with the other two types.
Furthermore, our own phonemic interpretation diers from Russian-inspired
ones, because it looks more to concrete things than to (more) abstract ones, which
are further away from the true linguistic intuition of native speakers, unless they
are deviated by archaic Soviet preconceived patterns, to the point of accepting
inappropriate solutions without questioning.
at is more, this analysis of ours is closer to the ocial writing, with obvious
advantages which compensate for many other discordances between orthography
and pronunciation.
Obviously, some other authors have preferred this kind of analysis, as Tromov
Jones (1923), which we had not yet discovered at the time of our early interpretation, presented for the rst time in 1975. We maintain it here, with some renements (even phonetic ones, with respect to previous choices which turned out to
coincide with those of Tromov Jones with six vowel phonemes, although they
treated even coronal palatalized consonants in a phoneme-like way).
Unfortunately and strangely, the structuring in Jones Ward (1969) is of a fanta-Soviet type (given the direct involvement of Ward with the Soviet/Russian
world and an unexplainable to us rethink on the part of Jones, who died two
years before the publication of the book, about the adequacy of the 1923 analysis).
Nevertheless, all in all, Jones Ward's book remains the best extensive treatise
on the subject; it is decidedly better than the very heavy and (almost) unchanging
Soviet and Russian productions.
Most examples we provide are taken from the various texts given in the bibliography, by retranscribing them (or transcribing them entirely, when they were only graphemic or partial). We always do this both phonetically and phonemically,
according to our principles and methods.
Let us focus attention on another notational dierence, as well: between our
transcriptions (which are phonetically and phonemically canIPA) and the oIPA
(which is ocial after the 1993/96 reform): ('e8) /'djenj/ {/'dJenJ/o} den (='e)
8. russian
285
/z'djesj/ {/'zJdJesJ/o} dec (=';x) /z'djelatj/ {/'zJdJEltJ/o} cdelat (&ii'vi;i~i[]I) /tili'vidiniji/ {/tJIlJI'vJidJInJIjI/o} teleb$den$e (x&jiIst'v; -i'v;) /ajistist'vje/ {/ajIsJtJI'sJtJvJE/o;t, ocial and traditional) o ectectbe It seems quite obvious to us that an excess of (J) is certainly harmful to readability too.
8.0.2. At the beginning, we think it useful to show the Cyrillic alphabet, both
roman and italic. e latter especially, since the shape of some letters can be misleading at rst, by comparison with the Roman (or Latin) ones typical of certain
kinds of transliteration, which are given in brackets. It is not recommendable to
use the ambiguous ch for /x/, since kh is decidedly more appropriate, while keeping ch for /c/ (in case the most satisfying choice of using /c/ and x /x/ is not
made).
For Cyrillic writing, it is important to pay particular attention to the roman >b,
e, n, p, c, y, x, and the italic >e, $, , p, c, t, y, x.
a, a [a] ('Fa;) /'a/
, [b] ('bF;) /'be/
b, b [v] ('vF;) /'ve/
g, g [g] ('gF;) /'ge/
d, d [d] ('dF;) /'de/
e, e [e je ye] ('j;) /'je/
e/, e/ [ jo yo] ('j;) /'jo/
, [` zh] ('qF;) /'qe/
, [z] ('zF;) /'ze/
$, $ [i] ('i;) /'i/
=, = [j ] (&ik'RFatkx[]I) /ik'ratkaji/ $
kpatkoe
k, k [k] ('kFa;) /'ka/
l, l [l] ('F1, -) /'el/
m, m [m] ('F) /'em/
n, n [n] ('Fn) /'en/
o, o [o] ('j;) /'o/
, [p] ('pF;) /'pe/
p, p [r] ('F) /'er/
c, c [s] ('Fs) /'es/

t, t [t] ('tF;) /'te/
y, y [u] ('ju;) /'u/
f, f [f ] ('Ff) /'ef/
x, x [x 9 h >! { u}) ('Fa;) /'xa/
z, z [c _] ('qF;)/'qe/
u, u [ >] ('C;) /'ce/
w, w [ ] ('SFa;) /'Sa/
, [3 ( >] ('a;) /S'Sa/
, [] (t'vRdiz&nak,' j) /t'vjordiznak, 'jer/ tbepdv= nak
v, v [y y] ('F;, jI'RF;) /, ji'r/
, ['] ('miiz&nak, 'maii-, 'je)
/'mjaxkiiznak, 'mjakkii-, 'jerj/ mrgk$= nak
<, < [e ] ('F;, &Fb'Rjtnx[]I) /'e,
eaba'rotnaji/ < oopotnoe
, [ju yu] ('ju;) /'ju/
r, r [ja ya] ('ja;) /'ja/
8.0.3. In Russian, the normal phonetic syllabication, although not absolutely, considers that /r, l/ preceded by /0[j]/ are heterosyllabic, unless they are preceded by another C (as in (&xts'tRa;dx) /aftas'trada/ abtoctpada and other examples
in the text). e same occurs for /0j-j/ (J-j): (kp'jun) /kap'lun/ kalyn, ('jutRx) /'utra/ ytpo, (kvd'RFat) /kvad'rat/ kbadpat, ('pFak-x) /'paklja/ aklr, (v'Rj;vi) /lav'rovi/ labpobv=, ('j-Rx) /'oxra/ oxpa, (zvi'j;) /zvirj'jo/ bepe is
may not be very surprising within words; but perhaps it could seem to be a bit
strange in word-initial position. Actually, if we adequately consider not single
words but the phrases and sentences which contain them, the logic behind this de-
286
cision is (almost) evident. In fact, if we listen carefully, this syllabication is generally conrmed, although it is possible to nd contrary opinions: (t'i;) /t'ri/ tp$,
(s'pj) /s'por/ cop, (t'vRdi) /t'vjordi/ tbepdv=, (&ik'RFatkx[]I) /ik'ratkaji/ $
kpatkoe, (s'tRFa) /s'trax/ ctpax, (s'pFaf) /s'plaf/ clab
Vowels
8.1.1. Russian orthography is clearly based on morphonological principles,
which certainly privilege lexical and grammatical structures (even diachronically!)
in comparison with phonic ones. e latter are considered more in the abstract,
more in phonological terms, but still chiey morphologically. Nevertheless, dierently from most languages, in Russian orthography we nd that the phonology of
vowels and consonants is presented fairly spontaneously, as felt by nave speakers; not in the sophisticated way linguists do, with useless complications both
for teaching and descriptive purposes.
Indeed, the Russian orthography uses six vocalic graphemes which correspond exactly to the number of vowel phonemes of this language: $, <, a, o, y, v /i, e, a, o,
u, / (clearly, >e, r, , do not indicate vowels, but sequences of /j/ + /e, a, o, u/,
that is (j) or (J) + (, a, , u), or (e, , P, ) when they are also followed by = /J,
i/). By using these six vowel phonemes (obviously listed in a phonic order), twenty-two consonant phonemes (including /Z/) are sucient, in contrast with over thirty
( from thirty-three up to thirty-nine) used in other analyses, which persist in positing only ve vowels. Of course, this excessively complicates both the whole phonological inventory and the relation between pronunciation and writing.
If a comparison is made, the choice is simple and even obvious. In fact, it seems
to be more logical and natural to conclude by saying that, in Russian, C have taxophones which are patalalized by /i, j/, rather than to imagine some soft C
which would (automatically!) make the Russian speakers produce soft(ened)
V whereas hard C would make them produce hard(ened) V by advancing the
false excuse that Russian-speaking people are unable to pronounce a clear i (i)
sound after k (k).
8.1. Russian vowel elements.
/'i, 0iJ/ (i), /i'/ (I), /i/ (I)
/'/ (), /'/ (), // ()
/'JeJ/ (e)
/'e/ () {/'e, "e/ ([])}
/'JaJ/ (', "{})
/u/ (u), /JuJ/ ()

/'JoJ/ (P)
/'o/ {/o/} () {/'o, "o/ ([])}
/a/ (x), /a'/ ()
/'a/ (a) {/'a, "a/ (a[])}
/i/ (i) diphthongs /j/ ('j, ), /ji/ ('i, []I), /'i, 'e, 'a, ', 'o, 'u/ ('i, 'F, 'Fa, 'F, 'j, 'ju)
In modern pronunciation, even the oddity of (ii) /kii/ >-k$=, traditionally

pronounced as /ki/, disappears. erefore, by denitively reversing the prob-
8. russian
287
lem, everything becomes simple. As a matter of fact, Russian-speaking people say

(i) and (k) (in stressed syllables ('i;), ('kF;)), since phonemically they are /ki, k/
not /i, ki/; and the same is true of (bI, 'bi; b, 'bF;) /bi b/ (not /bi bi/), .
In addition, for /J, J0/, the very spelling indicates /0j/ through >C, except
in cases where the pronunciation changed centuries ago (whereas the orthography
has remained unchanged, still keeping sequences such as w ). Furthermore,
there are (more or less incomplete) assimilations for /00j/, which become (0J),
or (JJ), depending on the type of pronunciation (more) modern or (more) traditional, as we will soon see ( 8.3.2.0-11).
8.1.2. In stressed syllables, the six vowel phonemes are realized as the black
markers shown in 8.1. Obviously, those with a white center also correspond to
certain unstressed contexts that we will see later on: (t'i;) /t'ri/ tp$, ('F;Rx) /'era/
<pa, ('kFak) /'kak/ kak, ('tjt) /'tot/ tot, ('jum) /'um/ ym, ('bFt) /'bt/ vt; besides, we also have: ('qF) /'qex/ zex, ('SFak) /'Sak/ wag, ('qjuk) /'quk/ yk
As some examples have already shown, there are some interesting dierences in
our phonetic transcriptions in comparison with phonemic ones, apart from symbol variations such as () /e/, () /o/. ile (i, I) /i/ remain unchanged (except in
the sequences (i, I) /ji/, which represent >$ after V and >e in unstressed syllables), the most important element is instead the prevelar semi-approximant (F), occurring before stressed /e, a, /, and the velar rounded semi-approximant (j), before stressed /o, u/ (all of them preceded or not by a dierent C from /j, c, SS, ZZ/).
erefore, the V which are dierent from /i/ are realized as sequences of (F) + (, a,
), or (j) + (, u); in particular this happens in stressed syllables, even in compounds and in sentences where distinct timbres occur; that is with /e, o/ too.
Obviously, (F, j) are less evident than fully prevelar, or velar rounded, approximants, (j, w); (w) occurs in English ('wT, 'khwT, 'w:{<}) /'wIt, 'kwIt, 'wO:/ wit,
quit war. It is to be noted that (F, j) are quite dierent from a phonic zero, (`)
(although nave Russian speakers would swear they cannot hear anything but (`)),
as in English ('T, 'khT, ':{<}) /'It, 'kIt, 'O:{}/ it kit oar). However, in unstressed syllables, the approximants (F, j) do not occur, as will be clear from some examples
given below.
It is important to explain from the very beginning the palatalizing function of
the phonemes /i, j/ on the C which precede them. us we have to pre-empt, here,
the triple division of the Russian C. We nd normal C, which are palatalizable\
/m, n r l/ (m, m n, ~ R, , ), /p, b t, d k, g/ (p, p b, b t, d, k, g, ),
/f, v s, z x/ (f, f v, v s, z, = , ). en there are the palatal ones, in the general sense that they have a real palatal component, in spite of their actual realizations: /j, c, S[S], ZZ/ (j, C[], [], [] ). nally, we have non-palatal C\ /q S,
q/ (q S, q), which are never palatalized (although spelling sequences like >z$, we,
exist: (S'ju;) /S'ju/ w). We must always keep in mind, then, that */qi Si, qi/
do not occur (any longer) but are substituted by /q S, q/, in spite of historical
writings with >z, w, followed by >$, e, r, ,
8.1.3. After all, /i, j/ (and /c S, Z/) would be sucient to adequately explain
how the Russian phonemic system works, which is so typical because of palatali-
288
zation erefore, we can represent (in general, as a functional category, especially in our concise phonemic formulae) those C which have a palatal component (
(j) /j/, (C) /c/, ([]) /S[S]/, () /ZZ/) by means of /J/. In our formulae, this also includes the typical diphthongs, (i) /i/ >V=] as if they were /j/, which is a
more abstract representation, often used but not by us. In fact, our only concession to abstraction is the use of /J/ in our formulae, even for /i/. Another (more
abstract) possibility, which is traditional and frequent in transliterations, would
give //Sc// for /S[S]/.
So if normal C are followed by /i, j/, they are realized as palatalized (and absorb /j/): (m, ~ ), (p, b /+, /_ , ) ((, ) occur before V and heterorganic C; (+, _) occur before homorganic C], (f, v , = ) obviously, without the addition of the semi-approximants (F, j), before V
erefore, we have: ('mi) /'mir/ m$p, ('vj;i)) /'vosimj/ bocem, ('~;bx) /'njeba/ neo, (k8'i;) /kanj'ki/ konk$, ('is) /'ris/ p$c, ('zjkx) /'zorjka/ opka,
('kj) /'korj/ kop, ('s) /'ljes/ lec, (b1'Sji) /balj'Soi/ olwo=, ('sj1) /'solj/
col, ('p) /'pjel/ el, ('tjp) /'topj/ to, ('bust) /'bjust/ ct, (';x) /'tjela/
telo, ('mFa) /'matj/ mat, (+'a;) /t'lja/ tlr, ('e;I) /'djeti/ det$, (d'v;, '-)
/d'vje/ dbe, (_I'nFa;) /dli'na/ dl$na, ('it) /'kit/ k$t, (&mx~i'ju) /mani'kjur/ man$kp, ('nj;I) /'nogi/ nog$, ('fi;~Ik) /'finik/ f$n$k, (k'Rjf) /k'rofj/ kpob, ('ve)
/'vjesj/ bec, (';) /f'sje/ bce (also (';) /f'sjo/ bc) ('gju) /'gusj/ gyc, (=I'mFa;)
/zi'ma/ $ma, ('i;ii) /'tixii/ t$x$=
Naturally, even (C, 'C, ) /c/, ([]) /S[S]/ (and () /ZZ/) have the value of :
('Ck) /'cek/ uek, ('Cu;) /xa'cu/ xouy, (C'n) /c'ljen/ ulen, ('bFa;) /alc'ba/
alua, ('it) /S'Sit/ $t, (I'u;) /iS'Su/ $y, (t'vFa;i[]) /ta'variS[S]/ tobap$
e phoneme () /ZZ/ is particular and typical only of a kind of traditional and
old-fashioned pronunciation: ('vjI) /'voZZi/ bo$, ('jeu) /'jeZZu/ ey. e
most modern and recommendable pronunciation has (qq) /qq/, in almost all words:
('vjqq, 'jqqu) /'voqq, 'jeqqu/.
Initial /j/ is par excellence, and even after V (keeping well in mind that >e, r,
, being simply C + V sequences, not diphthongs correspond to /j/ + /e, a,
o, u/): (jI'zFk) /ji'zk/ rvk, ('j) /'jel/ el, ('ja;) /'ja/ r, ('jS) /'joS/ e, ('juk) /'juk/ g,
(m'ja;) /ma'ja/ mor, (s'jus) /sa'jus/ co, (d'vj[;]I) /d'voji/ dboe, ('nj;vxx) /'novaja/ nobar, ('i;~i[]I) /'siniji/ c$nee
8.1.4. e four vowel phonemes /e, a, o, u/ may occur between and are realized closer fronter, as indicated by the grey markers, (e, , P, ) (again in 8.1):
('je1) /'jelj/ el, ('p) /'pjatj/ rt, ('P;x) /'tjotja/ tetr, ('fak) /tju'fjak/ tfrk
e advancement and raising of /[0]iJ/ is decidedly more limited, (i) (and, in
the vocogram, there is no special marker dierent from the one for normal (i), so
as not to render the gure too heavy): ('bi) /'bitj/ $t
On the contrary, // does not occur in this context; and this phenomenon has
cultivated the myth of the complementarity of (i, ). So they are grouped in the
same phoneme, /i/, under the delusion that a better phonological system has
been found while it is exactly the contrary. Deviant schooling is hard to re-
8. russian
289
move, especially in regard to ocial writing (with its known and less known problems) and respelling, which is even used in schools in addition to in scientic
works, with phonemic ambitions. ey would like to nd almost alchemist's
solutions, by using complex formulae and esoteric iconograms (which we will be
careful not to reproduce).
In 8.1, three potential phonetic diphthongs, (, a, ), are also given,
which can substitute (;, a;, ;) (with an interpalatal variant, (), for (;)). ey
occur mainly in intonemes, in unchecked (nal or internal) syllables: (';, '
-;, -) /f'sje -jo/ bce, ('F;x, 'Fx) /'exa/ <xo, ('dFa;, 'dFa) /'da/ da, ('Fa;tx, 'Fatx) /'atam/ atom, (';x, 'x) /'djadja/ drdr, (d'nj;, d'nj) /d'no/ dno,
('gj;xs, 'gjxs) /'golas/ goloc ey also occur for emphasis, in which case they
are longer; thus we have in unchecked syllables (; a;, ; ;): ("; -;)
/f'sje -jo/ bce, ("F;x) /'exa/ <xo, ("dFa;) /'da/ da, ("Fa;tx) /'atam/ atom, (";x) /'djadja/ drdr, (d"nj;) /d'no/ dno, ("gj;xs) /'golas/ goloc; and in checked
syllables ( a, ): ("~t) /"njet/ net!, ("qFa1) /"qalj/ al!, ("p) /"pjatj/
rt!, ("dj) /"dom/ dom!
8.1.5. Russian also has some ocial diphthongs, all with a front second element /i/ (i). Here we will consider those occurring in stressed syllables: ('ii)
/'kii/ k$=, ('peiI) /'pjeiti/ e=te, ('mFai) /'mai/ ma=, ('Ci) /'cai/ ua=, ('bji)
/'boi/ o=, (z'Pi) /za'rjoi/ ape=, ('djuiI) /'duiti/ dy=te, ('i) /'djuim/ d=m,
('vFidu) /'vidu/ bv=dy. As can be seen, the orthography uses >= for diphthongs
consecrated by grammar. However, mainly in loanwords, others occur, as >ay:
('fFaust) /'faust/ Fayct
In unstressed syllables, we have (again with (i)): ('ju;ii) /'uljii/ yle=, (Cii'kjsii) /cii'kofskii/ Ua=kobck$=, (ti'gFa;) /tai'ga/ ta=ga, (';xiI) /'djelaiti/ dela=te, (ui'dju;) /ui'du/ y=dy, (i'mjkx) /djui'mofka/ d=mobka, ('b;i)
/'bjeli/ elv=
Instead, purely graphemic diphthongs like >$$ $r e$ ee ae ar a oe ve
do not correspond at all to phono-diphthongs, except when the semi-approximant
(which is shown in round brackets, ([])) falls completely. In this case, however, if the last vocalic element is /i/, the phone (I) is maintained (instead of having
(i), as in phonemic diphthongs, unless it is followed in sentences by (J) or /i/):
(vI'tFa;i[]I, v-) /vi'taliji, v-/ b *tal$$, (vR'i[;]I) /vras'siji/ b Pocc$$, ('i;i[]x) /'lilija/ l$l$r, (mu'=e[;]I) /mu'zjeji/ mye$/-ee ( (mu'=ei) /mu'zjei/ mye=], ('e~~i[]I)
/a'sjenniji/ ocennee, (6'ma[;]I) /v'maji/ b mae, (&~IRu'SF;mxx) /niru'Smaja/ nepyw$mar, (z'nFa;u) /z'naju/ na, ('nj;vx[]I) /'novaji/ noboe ('juqn[]I) /'juqnji/ nve
e following examples are bisyllabic: (p'juk) /pa'uk/ ayk, (m'i;) /ma'ji/
mo$, (p'Ft) /pa'et/ o<t, (du'F1) /du'elj/ dy<l
8.1.6. As far as V in unstressed syllables are concerned, in addition to what we
have already seen in some examples, in modern pronunciation we have ( 8.3.2,
for traditional and alternative pronunciations): (I', I, iJ) /i/ (including (Jii)
/0ii/ thus, /i/ is (i) when it is followed by (J) or by /i/, even in sentences, with-
290
out a pause), (', , x) /a/, (u) /u/, (', ) //: (CI'sF;) /ci's/ uacv, (pi'i;) /pi'ti/ rt$, (&Iz'kFa;) /jiz'ka/ rvka, ('bF;Cii) /a'bcii/ ovua=, (&jIR'FaS) /jira'laS/ epalaw, (&piimi'i;) /pirimi'li/ epemel$, (&IIk'ti;CIstvx) /ilik'tricistva/ <lektp$uectbo, ('pj;I) /'poli/ ole (&gx'vFa;) /gala'va/ goloba, (&v'gFat, -n'g-) /avan'gart/
abangapd, (k'RFasnxx) /k'rasnaja/ kpacnar (u'Rjk) /u'rok/ ypok, ('dj;mu) /'domu/ domy, (u'dFa;) /sju'da/ cda, ('u;kx) /sju'sjukatj/ cckat (q&viI'zFa;qx) /qvili'zatsja/ z$b$l$az$r, (q'nFa;) /q'na/ ena, ('dF;R) /'dr/ dvpv
For >r in unstressed syllables (and for >a too), it is important to explain that
neutral pronunciation has only /i/, whereas a transversal use of (, ) (or even (x,
) /a/) is very widespread, although decidedly non-neutral. is occurs both in traditional and alternative pronunciation (and indeed even for speakers who are neutral, one way or another!). is can occur both within lexemes, and in inected
forms with >r+C consonantal endings (inuenced by the forms ending in -r, with
/ja/, (Jx, x)).
us we have: (pIt'nj;, pt-, pt-) /pit'no/ rtno, (jI'zFk, j-, jx-) /ji'zk/ rvk, (&jIz'i;, &j-, &jx-) /jiz'ki/ rvk$, (&pxgi', --, --) /pagli'djel/ oglrdel,
('nFa;CIt, -Ct, -Cxt) /'nacit/ nauat, ('bF;Cii, -Ci, -Cxi) /a'bcii/ ovua=,
('kjRmIt, -mt, -mxt) /'kormit/ kopmrt, ('e;I, -, -x) /'djetim/ detrm, ('zFa;~I, -, -x) /'zanil/ anrl, (&bIspi'i 'p, &bsp'i 'p, -p'i) /bispi'ti 'pjatj/ e rt$ rt. (Please note the variants of -$k, as well: (s'pju+~Ik,
-~k, -~xk) /s'putnik/ cytn$k.)
Normal traditional pronunciation, for (pretonic) >e', r', a' ( occurring in
syllables immediately before the stressed or tonic one in a intoneme) has /e/ (',
'+J), which correspond to modern pronunciation /i/ (I, iJ): (&bIsp'i 'p)t, as
also happens in: (bi';zx)m (b-)t /bi'rjoza/ epea.
In vowel sequences which contain /a/, we generally nd an extension of the use
of (): (&vb-R'qFa) /vaabra'qatj/ boopaat, (nu'gFat) /nau'gat/ naygad, (&pdn'mju;) /paadna'mu/ o odnomy, (&nstR'vFa) /naastra'vax/ na octpobax, (u&dn'vj;) /uadna'vo/ y odnogo, (su'Cas[t]vxvx) /sau'cas[t]vaval/ coyuactbobal,
(st'vs[t]vxvx) /saat'vjes[t]vavatj/ cootbetctbobat.
Consonants
8.2. e phonemic structuring we prefer for Russian only has twenty-two consonant phonemes (including /ZZ/), which combine with the six vowels, as we have already seen in the previous sections. Here we will systematically consider Russian
C's according to their manners of articulation.
Let us observe that no voiced diphonic C occurs before a pause or a voiceless C:
('bjp) /'bop/ o, ('mjsk) /'mosk/ mog, (t'RjupI) /t'rupki/ tpyk$, (s'dju;)
/fsa'du/ b cady. In addition, no voiceless C occurs before a diphonic voiced C (/v/
is a bit particular, 8.3.1.7): (z'bF) /z'btj/ cvt, (zg'RF;) /zga'r/ c gopv.
nally, in general, son(or)ants are partially devoiced before either voiceless C
or pauses; usually, between a voiceless C and a pause, they become voiceless, unless they become intense, in which case they are voiced or half-voiced (according
8. russian
291
laryngeal
velar rounded
velar
prevelar
palatal
prepalatal
postalveo-palatal
velarized alveolar
alveolar
dental
palatalized
labiodental
labiodental
labio-palatal
bilabial
8.2. Table of Russian consonants.
velariz. postalveo-pal.
protruded (non-pr.*)
to their phonic context): ('qF) /'qr/ $p, ('tFa;) /r'ta/ pta, ('kFatx) /'karta/ kapta, ('bjb, -;b) /'bobr/ op, (_'~p5, -;p) /d'njepr/ Dnep.
e table in 8.2 shows the consonantal articulations of Russian, which are
necessary for an appropriate pronunciation of this language.
Instead, 1.9-15 show the orograms of all the contoids treated in the chapters
of this book, by grouping them according to manners of articulation, including
secondary, occasional, or regional variants, for the 12 languages dealt with. is
way of displaying them makes the necessary comparisons between dierent languages more straightforward.
m (m) (M) () (n) n

(~) (2)* (~)
(N)
()
F p b (p b)
t d
(+ _) ( )
k g
q (Q)
C ()
( )
( 6)
f v (f v)
_
s z
{} S q ( =)
(,|J)
(y)
() j|() (F) ()
(j) {H}
R
()
()
(L)
/c S, Z x r l/ (C , R ) /m, n r l/+(), 5 , ) + ((, , , , n))
Nasals
8.2.1. Russian has only two nasal phonemes, /m, n/, with various taxophones,
especially for the latter. us: /m/ (m, m, M, ) and /n/ (m, m, M, , n, ~, 2, N,
) (more specic symbols like ((, n, )) are not necessary, but as we will shortly
see (, , n , ) can be used, for a less-assimilated kind of coarticulation,
which maintains their front lingual contact alveolar/prepalatal while adding a
secondary articulation with no contact: bilabial, labiodental, velar).
For /m/: ('mFa;mx) /'mama/ mama, ('Fapx) /'lampa/ lama, ('njR) /'norm/
nopm, ('it), 'i;tM) /'ritm/ p$tm, ('i), 'i;M) /'rifm/ p$fm, (m'nji) /m'noi/
mno=, ('Fa;) /m'xa/ mxa, (mg'Fa;) /mg'la/ mgla, ('mik) /'mik/ m$g, ('m)
/'mjatj/ mrt, ('e)) /'sjemj/ cem, (Im'bi) /im'birj/ $m$p, (M'vjn) /am'von/
ambon, ('FafxRx) /'amfara/ amfopa. /m/ can occur befor heterorganic C: (m'gFa;)
/m'gla/ mgla, (m'nji) /m'noi/ mno=, ('Fa;) /m'xa/ mxa; however, for /mj/, and
/m/ + /i/, we regularly have (m): ('mi) /'mir/ m$p, ('e)) /'sjemj/ cem.
For /n/: (&jmbi't, &j-) /ombi'rjot, on-/ on epet (kM'vFa;, -'v-) /kan'va/
kanba, ('fFas, -) /an'fas/ anfac, ('njs) /'nos/ noc, (n'RFaf) /n'raf/ npab, (k'mFa~I) /ka'mandi/ komande, ('~i) /'nitj/ n$t, ('kj8) /'konj/ kon (bi~'=in)
292
/bin'zin/ en$n, (&bxR'bFa8Ik) /bara'banSik/ apaan$k, (I2'qFa) /kin'qal/

k$nal, ('RFaNS\) /'ranjS/ panwe, ('tjii, -8- -~ki, -ki) /'tonkii/ tonk$=,
('bFa~k, -k) /'bank/ ank
us, /n0, n0/ regularly assimilate. Also /nj/ becomes (~), but then it keeps its
phonetic autonomy (even as a phonemic sequence, with a distinctive value), without neutralizations before other C. It must be said, though, as we have already seen
from some examples, that especially because of spelling inuence (and mainly in
less colloquial and more traditional pronunciation) a complex coarticulated way
of pronouncing is widespread, with partial assimilation: (b, v, 8, k) ,
which would try to maintain (nb, nv, n, nk) . In these cases, people oscillate
considerably; but, on the other hand, should it be important to emphasize a kind
of pronunciation with heterorganic ((n0)), the orthography would use >n, at
least within words; but it does not do so
Stops
8.2.2. ere are three diphonic pairs which do not oppose before a pause or a
voiceless C: /p, b t, d k, g/ (p, p, b, b t, , +, d, , _ k, , g, ). e actual articulation of palatalized /t, d/ is prepalatal which are stopstrictives, but stops before homorganic (or similar) C. Sometimes, we nd (+) before pauses as well): ('pjt) /'pot/
ot, ('jp) /'lop/ lo, ('peC) /'pjec/ eu, ('tjp) /'topj/ to, ('gj;up) /'golupj/
goly, ('bFt) /'bt/ vt, ('bF;tx) /a'betam/ o <tom, ('bi) /'bitj/ $t (t'Rjut)
/t'rut/ tpyt, (q'qFa;) /at'qa/ otza, (t'kju;) /t'ku/ tky, ('vit) /'vit/ b$d, ('mFa;)
/tj'ma/ tma, ('fju;) /tj'fu/ tfy!, (pi+'a;, pIt-, 'pe+-x) /pit'lja, 'pjetlja/ etlr, (R'bj+~Ik) /ra'botnik/ paotn$k, (C'Ct) /at'cot/ otuet, ('djup) /'dup/ dy,
('tjdq) /'todq/ tot e, ('e8) /'djenj/ den, (_'~a;) /d'nja/ dnr, ('pj_I) /'podli/
odle, (d've, '-) /d'vjerj/ dbep, ('kFak) /'kak/ kak, ('pFakt) /'pakt/ akt, ('k)
/'ljok/ leg, ('kF;txmu) /'ketamu/ k <tomy, (Is'j) /kias'kjor/ k$ockep, ('ipii)
/'gipkii/ g$k$=, ('gjt) /'got/ god, (g'nju) /g'nutj/ gnyt, (n'i;) /na'gi/ nog$
Stopstrictives
8.2.3. Russian has two voiceless stopstrictive phonemes, /q, c/, but four realizations, by voicing assimilation (in addition to the palatalized variants of /t, d/,
(, ), which we have already seen in 8.2.2). Besides, before a stressed V dierent from /i/ /c/ is (C): ('qFk, -%) /'qkl/ z$kl, ('q) /a'tjeq/ otez, (b'RFaqqx)
/b'raqqa/ patza, (pQ'dFaR) /plaq'darm/ lazdapm (u'Ci;i1) /u'citilj/ yu$tel, ('Cas) /'cas/ uac, (pi'C;) /pli'co/ leuo, ('juC) /'luc/ lyu, (C'n) /c'ljen/
ulen, (n'if) /nac'dif/ naud$b
8. russian
293
Constrictives
8.2.4. For the time being, let us introduce the three constrictive diphonic pairs,
(f, f, v, v, 6) /f, v/, (s, z, =) /s, z/, (S, q) /S, q/; apart from the palatalized taxophones (of the rst two pairs, but not of the last one), the labiodentals have a bilabial realization when they do not precede either /j, r, l/ or V (including the voiced
approximants (F, j) which are added, from a phonetic point of view, in front of
stressed vocoids): ('fjn) /'fon/ fon, ('Rjf) /'rof/ pob, (&t'mFat) /afta'mat/ abtomat, (f'fimI) /f'filjmi/ b f$lme, ('vf) /'vjerfj/ bepf, ('vjk) /'volk/ bolk,
(S'kFa6q) /S'kavq/ wkaf e, ('sjn) /'son/ con, (R'sFa;) /ra'sa/ poca.
More examples: ('~k) /s'njek/ cneg, ('j) /'osj/ oc, ('zjup) /'zup/ y, (z'ji)
/z'loi/ lo=, (z'b;) /z'bjelm/ c elvm, (I'=u) /i'zjum/ $m, (&bi=i'ei) /bizdi'tjei/ e dete= ('SF) /'Sestj/ wect, ('jS) /'loS/ lo, ('qFst) /'qest/ ect,
(q'dFa) /q'datj/ dat, ('qju;) /xa'qu/ xoy
Lastly, there is another quite particular voiceless constrictive, ([]) /S[S]/, which
is lengthened before //, and has a traditional variant constituted by the corresponding constrictive followed by the (sometimes very weak) stopstrictive, (C).
is is more complicated and not at all necessary in modern pronunciation (indeed, as we have already said, it more aptly belongs to traditional pronunciation).
Before a stressed V (when dierent from /i/), a () is inserted: ('it)m ('Cit)t /S'Sit/
$t, (p'a;dx)m (pC'-)t /paS'Sada/ oada, ('bj)m (-C)t /'borS/ op.
e possible sequence (C) /Sc/ is dierent but quite normal indeed as in:
(' 'C) /S'Sem/ c uem ( 8.3.2.5).
Approximants
8.2.5.1. From a phonemic point of view, Russian has two approximants. Palatal
/j/, which is realized as fully palatal either before a stressed V, ('j), or in initial position in a rhythm group before a (stressed or unstressed) V, (j). Instead, /j/ is realized as a palatal semi-approximant after a (stressed or unstressed) V before another unstressed V\ () (and also in ('i), for /'ji/ >V$\ (m'i;) /ma'ji/ mo$] after
C it changes the sequence into a palatalized C, /0j/ (J) ( 8.1.2-3): ('j) /'jatj/
rt, (ji'vi) /ji'vitj/ rb$t, (&mxI'kjsii) /maji'kofskii/ Markobck$=, (b1'SFa;x) /balj'Saja/ olwar
For emphasis, or precision, we can have (j) = (,) (semi-constrictive) and, respectively, () = (j). However, it is more important to note that, in non-slow speech,
the /ij, ji/ sequences are realized as (i, i): ('FaRmix, -mix) /'armija/ apm$r,
(m'i;, m'i;) /ma'ji/ mo$ Sometimes, even in /j/ () sequences (without /i/),
it is not easy to adequately perceive (), especially in fast speech: (b1'SFa;x, -a[;]x)
(above all in preintonemes).
8.2.5.2. Current Russian orthography still presents some old-fashioned ways of
writing >w z followed by >e r $ and even by > But in these cases,
nowadays, the value of those V is /e, a, o, u, / (obviously, > is only used in dic-
294
tionaries and teaching texts): ('qF) /'qex/ zex, ('Sjk) /'Solk/ welk, (&pxR'Sjut)
/para'Sut/ apawt, ('qFk) /'qrk/ z$pk, ('qF) /'qtj/ $t, ('jS) /'loS/ lo
However, they can be followed by /j/ (j) inserted before the V (and in writing
> appears too): (S'jt) /S'jot/ wet, (S'ju;) /S'ju/ w, ('jqju) /'loqju/ lo
Besides, the palatal approximant can occur after (J), too, then it is phonemically geminated /jj/, and may have a distinctive value: (s'j;) /srj'jo/ cvpe, (p'ju;)
/pj'ju/ , ('ja;vx) /dj'javal/ drbol, (v'jut) /vj'jut/ bt, (IL'jiC) /ilj'jic/ *l$u
Even /sjj, zjj/ can occur, but indicated with > (although not always, as in ce$tcr, given at the end of this section): ('j;x) /sj'jexatj/ cexat, (I='j) /izj'jatj/ $rt
e same occurs for /Jj/: (C'ja;) /c'ja/ ur ( ('Cat) /'cat/ uad, where the dierence is clear also because of the dierent syllabic structure). e /ljj, njj/ sequences
are realized as (Lj, Nj) (or even (j, ~j), which are already suciently dierent):
(L'jt) /lj'jot/ s/he pours let ( ('t) /'ljot/ ice led or ight let] (L'ju;)
/lj'ju/ l, (vRN'j;) /vranj'jo/ bpane, (sviN'ja;) /svinj'ja/ cb$nr
ere are even cases where a (0) remains separated and uninuenced by a following /j/; thus, if no stress sign is already present, it is necessary and sucient to
introduce a hyphen in the phonemic transcription (as the orthography uses >):
(b'j;x) /ab'jexatj/ oexat, (s'j;qqx) /s'joqqa/ ce$tcr, (d&vu'ja;Rusni)
/dvux'jarusni/ dbyxrpycnv=, (tRansjI6R'peisii) /trans-jivra'pjeiskii/ tpancebpoe=ck$=, (&mq'ja;Rusni) /mjeq'jarusni/ merpycnv=
8.2.5.3. e other approximant, /x/, in spite of its (constrictive) ocial symbol, is realized as (, ) (voiceless, velar or palatal, respectively); in addition, by assimilation to a following voiced C, we have (y), (velar) semi-constrictive, which
diers from the (less evident and less intrusive) non-phonemic element (F), automatically inserted before /'e, 'a, '/ in the typical and genuine pronunciation (as
happens with (j), before /'o, 'u/, 8.1.2): ('ju;q) /'xuq/ xye, ('mj) /'mox/
mox, ('it-Ri) /'xitri/ x$tpv=, (g'Rjyq) /ga'roxq/ gopox e
us, phonetically, Russian has six approximant (or semi-approximant) phones:
(j, ) /j/ (, ) /x/; (F, j) /`/, and two semi-constrictives: (y) /x/ and (,) (a possible
variant of /j/, 8.2.5.1).
Trills
8.2.6. In the Russian phonemic system we nd only one trill phoneme, which
is realized either as an alveolar, (R), or a prepalatal, (), tap according to contexts.
For emphasis, or speaking in a slow and clear way, as in teaching, it is possible to
hear the corresponding trills, (r, D), which may be used in these situations. But,
for a good pronunciation, these are not necessary; instead, it is more important to
respect the normal degrees of devoicing ( 8.3.1.7): ('RFat) /'rat/ pad, ('pFak)
/'park/ apk, ('pt5, -;t) /'pjotr/ etp, ('is) /'ris/ p$c, ('at) /'rjat/ prd, (f'nFa) /fa'narj/ fonap, (i~'ab, -;bh) /sin'tjabrj/ centrp, ('vi[, -;) /'vixrj/ b$xp, ('jbRxs) /'obras/ opa, ('Fad-Is) /'adris/ adpec
8. russian
295
Laterals
8.2.7. ere is only one lateral phoneme, /l/, with three taxophones, (, , L)
(independently of devoicing 8.3.1.7); () is a velarized alveolar, but it becomes (velarized) dental (which is possible to transcribe with (())), by assimilation
before /t, d q s, z/: ('F;q) /'lq/ lv$, (s'j;vx) /s'lova/ clobo, ('dju) /'dul/ dyl,
('ig, 'i;g) /'igl/ $gl, (s'mFs, -;s) /s'msl/ cmvcl, ('i;qx) /'liqa/ l$za, ('qFa1) /'qalj/
al, ('tj1kx) /'toljka/ tolko, ('pFamx) /'paljma/ alma, ('Rjub1, -;b) /'rublj/
pyl, ('mF, -;) /'mslj/ mvcl, (&pj'it-Rx) /pol'litra/ ol-l$tpa (here, lexical
compounding shows how it prevails against phonology, since /l+l/ remain separated, (), instead of fusing into ()), (L'jut) /lj'jut/ lt Some speakers can have a
velarized dental (or denti-alveolar) () in any case, instead of velarized alveolar, not
only before dental articulations.
Structures
8.3. In these sections we will deal above all with the dierences between the
phonic structure and writing, which inevitably presents surprises, since pronunciation changes, whereas orthography remains unchanging, except for ocial reforms (which however are always insucient and partial, as those of Peter the
Great {1708-1710}, the Academy of Sciences {1735, 1738, 1758}, and Lenin {1917}
('e;~In) /'ljenin/ Len$n). We will treat some dierent distributions of certain
phonemes, in the formation of Russian words.
Taxophonics
8.3.1.1. Let us start by contrasting some (similar, but not identical) structures,
which foreigners are likely to confuse, thus compromising what they say.
Russian grammar considers the division into grapho-syllables to be quite exible, for instance: /sis'tra/ (Is'tRFa;) ce-ctpa cec-tpa cect-pa us, it is quite surprising that for the division into phono-syllables, instead, the prevailing approach
perseveres in proposing (or imposing) */si'stra/, 8.0.3).
In the case of graphic geminates, for instance, we have: An-na /'anna/, but phonetically ('Fannx).
8.3.1.2. Obviously, Russian has its particular phonotactics ( the grouping of
phonemes). We will give some examples of dierent structures: ('vjtkx) /'votka/
bodka, (g';) /g'dje/ gde, (s'vFabx) /s'vadjba/ cbada, (t'kFaC) /t'kac/ tkau, (p'pFaCkx) /ap'packatj/ oaukat, (p'i;qx) /p'tiqa/ t$za, (pIt'nj;) /pit'no/
rtno, (s'pju+~Ik) /s'putnik/ cytn$k, (pi+'a;, pIt-, 'pe+-x) /pit'lja, 'pjetlja/ etlr, (pd'mtkx) /pad'mjotka/ odmetka, ('ep) /s'tjepj/ cte, (tk'nju)
/tk'nutj/ tknyt, (d'qjut) /d'qut/ dyt, ('bjukx) /'butjka/ yd-ka, (t've 'v-)
/t'vjerj/ Tbep, (+'a;) /t'lja/ tlr, (d'v; 'v;) /d'vje/ dbe, (_'~a;) /d'nja/ dnr, (k'vu) /k'vjerxu/ kbepxy, (S'kju;Rx) /S'kura/ wkypa, (k'Fa) /k'lastj/ klact, ('jatx)
296
/'kjaxta/ Krxta, (t'jt) /t'kjot/ tket, (s'kj;I) /fs'kori/ bckope, (I'nu) /fti'nutj/ btrnyt, (&pt'fe) /part'fjelj/ optfel, (v'v) /v'vjol/ bbel, (v'jt) /vj'jot/
bet, ('ju;gu, 'kju-) /'kjugu/ k gy, (k'~i;gx) /k'niga/ kn$ga, (6z'gat) /vz'gljat/
bglrd In traditional (and traditional-like) pronunciation, the palatalization of
/k, g/ also spreads to cases such as the last two: (k'~i;gx)m ('~i-)t /k'niga/ kn$ga,
(6z'gat)m (-at)t /vz'gljat/ bglrd
More examples: (s'RFa;zu) /s'razu/ cpay, (s'pRFa;vx) /s'prava/ cpaba, ('zFatRx)
/'zaftra/ abtpa, (zR'Ck) /zra'cok/ pauok, (';ii) /f'sjakii/ bcrk$=, ('j) /sj'jel/ cel, (6'=) /v'zjatj/ brt, (='iqqx) /z'liqqa/ l$tcr, (=';x) /z'djelatj/
cdelat, (dRu='ja;) /druzj'ja/ dpyr, (S'ju;) /S'ju/ w, (&tS'gFa) /atSa'gatj/ otwagat, (q'dFa) /q'datj/ dat, (q't) /q'djot/ det, (q'qFa) /q'qatj/ cat,
(C'Ctnx) /c'cetna/ tetno, ('PI) /x'ljoSSi/ xlectue, (IS'Sjkx) /iS'Solka/ $
welka, (i'i;) /sti'xi/ ct$x$, ('vjst) /x'vost/ xboct, (Rs'qvt) /ras'qvjet/ paczbet, (C'n) /c'ljen/ ulen, (kp'C;ni) /kap'coni/ kouenv=, (CIz'mRnx) /criz'mjerna/ upemepno, (&CI'vj;) /atci'vo/ otuego, ('mFa;) /tj'ma/ tma
Some further examples: (m'gFa;) /m'gla/ mgla, (m'nji) /m'noi/ mno=, (m'FatSi) /m'latSi/ mladw$=, (m'~e;~i[]I) /m'njeniji/ mnen$e, ('e;~i[]I) /m'Seniji/
men$e, (m'RFa;mx) /m'ramar/ mpamop, ('Caqqx) /m'caqqa/ muatcr, (I'iz, -;z) /ria'lizm/ peal$m, (If'RFag, -;g) /diaf'ragm/ d$afpagm, ('ve,
-;) /'vjedjm/ bedm, (d'RFa), -;M) /d'raxm/ dpaxm, ('me;Iq) /'mjesiq/ mecrz,
('pm) /f'prjamj/ bprm, ('kFa=8, -;=) /'kaznj/ kan, ('pe}, -;) /'pjesnj/ ecn, ('bFa;) /l'ba/ la, ('gFa) /l'gatj/ lgat, ('ig, 'i;g) /'igl/ $gl, ('RF, -;%) /'rxl/
pvxl, (b'dni) /b'ljedni/ lednv=, (S'a;px) /S'ljapa/ wlra, ('tFa;) /r'ta/ pta,
(R'qF;) /r'q/ p$, (R've;~i[]I) /r'vjeniji/ pben$e, ('~g, -;g) /'njegr/ negp, (pi't)
/fpi'rjot/ beped, ('ja;nx) /rj'janastj/ prnoct, (6'njut[, -;t) /v'nutrj/ bnytp
nally, let us also observe: (spk'nju) /fsplak'nutj/ bclaknyt
8.3.1.3. Generally, in Russian (but not necessarily, as for instance in quick pronunciation) stops and stopstrictives, in heterorganic sequences, are audibly released ((0$0)), whereas those in homorganic sequences are inaudibly released
((00); only here we will use the special diacritics): (k'Rjp$kx) /ka'ropka/ kopoka, (p$'i;qx) /p'tiqa/ t$za, (kup$'qju;) /kup'qu/ kyzy, ('jt$pusk) /'otpusk/
otyck, (t$'kFaC) /t'kac/ tkau, ('t$kx) /S'Sotka/ etka, (d$'gFat$kx) /ad'gatka/
otgadka, (s'vFa$bx) /s'vadjba/ cbada, (ig$'dFa;) /fsig'da/ bcegda, (g$';) /g'dje/ gde, (pd$'mt$kx) /pad'mjotka/ odmetka, (zt$'me;~i[]I) /zat'mjeniji/ atmen$e, ('ib$nu) /'gibnutj/ g$nyt, (k$'~i;gx) /k'niga/ kn$ga, (i$'mji) /sidj'moi/ cedmo=, (d$'ve, $'ve) /d'vjerj/ dbep (kg$'dFa;) /kag'da/ kogda, ('g$b)
/'tjogb/ tek v
More: (pIt'nj;) /pit'no/ rtno, ('bdni) /'bjedni/ ednv=, (&vxR+'~ik) /varat'nik/ bopotn$k, (pi+'a;, pIt-, 'pe+x) /pit'lja, 'pjetlja/ etlr, (fu+'a) /fut'ljar/ fytlrp. In addition: (p'pFaC$kx) /ap'packatj/ oaukat, (t'tju;dx) /at'tuda/ ottyda, (q'qFa;) /at'qa/ otza, (&I'nju) /atti'nul/ ottrnyl, ('u;dx) /at'sjuda/
otcda (also (t'u-), and even (t'su-) /at'su-/), (&CCI'vj;) /atci'vo/ otuego (phonemic geminates and similar homorganic sequences are realized as phonetic
geminates with a limited duration of the second element: ((00))).
8. russian
297
8.3.1.4. In addition to /q, c/ (q, Q C[], ), we also nd sequences like /ts, dz

tS, dq/ (ts, dz tS, dq) (but generally /ts, dz/ (ts, dz) become /q/ {(q) and (Q)}, except in compounding, especially with prexes): (&ts'Fa;u) /ats'laju/ otcvla,
(pts'ju;Sx) /pats'luSal/ odclywal, (nts'mjt5, -;t) /nats'motr/ nadcmotp,
('jdzf) /'odzf/ otvb, (&pxdzxg'j;vxk) /padzaga'lovak/ odagolobok, (nd'zj) /nad'zor/ nadop, (t'SF~Ik) /at'Seljnik/ otweln$k, (&bvIt'SFa;i) /abvit'Sali/ obetwalv=, (&pxdq'gFa) /padq'gatj/ od$gat, ('tjdq) /'todq/ tot
e, (d'qjugI) /d'qungli/ dyngl$ (&gxRq'kji, -ts'k-) /garats'koi/ gopodcko=, (s'vqii, -ts-) /sa'vjetskii/ cobetck$=, (ku'pFaqqx, -atsx) /ku'paqqa, -atsa/ kyatcr, ('qtvx, -tst-) /'djetstva/ detctbo (only in careful speech the variants given after a comma are possible, and are due above all to the inuence of spelling). Let us
notice the dierences between /vj, vvj, v-j/ (v, vvj, vj), in: ('v) /'vjol/ bel, (v'v)
/v'vjol/ bbel, (v'jku, 'v-) /v'jolku/ b elky; (vj) easily becomes (v). However, they
can remain distinct in careful speech above all in (the most) traditional one.
8.3.1.5. Nowadays, the basic forms with non-palatalizing endings, such as
('Fapx) /'lampa/ lama, ('bjmbx) /'bomba/ oma, make people produce even
('FapI) /'lampi/ lame, ('bjmbI) /'bombi/ ome, whereas in traditional pronunciation ('Fa)pI, 'bjmbI) were normal. In any case, we always nd: (Im'bi)
/im'birj/ $m$p, (m'bi;qx) /am'biqja/ am$z$r
8.3.1.6. Consonant gemination is phonemic, even though it is realized as a simple fairly limited phonetic lengthening, which may be absent in fast(er) speech,
unless it is thought important to keep the dierence (in the last two examples /n/
(n) corresponds to nn): ('SF) /'Stj/ w$t and (S'SF) /S'Stj/ cw$t, or (p'Rqxni) /pa'djerqani/ odepannv= and (p'Rqxni) /pad'djerqani/ oddepannv=
is kind of lengthening only occurs in word-initial position (even after a
pause) or between V\ (f&fIv-R';) /ffivra'lje/ b febpale, (ss'Fa) /ss'latj/ ccvlat,
(Iq'qj;gx) /iq'qoga/ $oga, ('qFa) /q'qatj/ cat, (quq'qFa) /quq'qatj/ yat,
('dFanni) /'danni/ dannve, ('jipi1) /'ottipilj/ otteel, (z'bjuI) /za'but[j]ti/ aydte, ('eni) /ad'djeljni/ otdelnv=, ('jdd) /'oddx/ otdvx.
For /S[S]/ ([]) we have: ('it) /S'Sit/ $t, (I'u;) /iS'Su/ $y, ('bj) /'borS/ op,
(t'vFa;i[]) /ta'variS[S]/ tobap$; thus, after C it is short; when nal after V it is
short (or slightly geminated).
8.3.1.7. As we have seen from various examples, the Russian voiced diphonic
phonemes are substituted by the corresponding voiceless ones, before a pause, or
before voiceless C: (d'Rjuk) /d'ruk/ dpyg, ('vjtkx) /'votka/ bodka, (&bIskxI'ba;~ii)
/biskali'banii/ e kolean$= On the contrary, the voiceless phonemes are substituted by the corresponding voiced ones (or by voiced phones, in the case of /q, c/
(Q, )), before a voiced diphonic C: (&~Ig'djt) /anig'dot/ anekdot, (&6gx~Is'tFan) /avganis'tan/ Afgan$ctan, (g'dj;mu) /g'domu/ k domy, ('Qb) /a'tjeqbl/ otez vl
However, >b /v/ alone is not sucient to change a preceding voiceless C into a
298
voiced one; in fact, it has to be followed by a voiced diphonic consonant (such as

/z, d, g/): (d6'zga;dx) /adv'zgljada/ ot bglrda (but (t'vFaI) /at'vlasti/ ot blact$] (g6d'v;) /gvda'vje/ k bdobe (but (k6'nju;ku) /kv'nuku/ k bnyky, (k'vFa)
/k'vam/ k bam]
e only voiced C to occur before a pause, or before voiceless C, are sonants,
/m, n, r, l/ (even if partially devoiced, unless they are preceded by voiceless C, in
which case they are fully devoiced; but only partially so if they are intense): ('tFa)
/'tam/ tam, (s'mjt5, -;t) /s'motr/ cmotp, ('tFp) /'temp/ tem, (1'i) /lj'stitj/
lct$t It is useful to insist on the fact that this kind of devoicing is not and it
must not be complete, if the previous C is voiced. Before a pause, it is important
not to pronounce a kind of () so phonation must stop before the hold of the C
in question is completed. Another fairly useful device to show this fact could consist in using a non-explosion diacritic (), in order to counter the fact that a non-phonemic () is added, by actually halting the articulation. But this is less frequent
among native speakers.
Even unstressed V (between voiceless C, or between these and a pause) may be
devoiced, but less frequently than C and above all in fast speech. us we will
mark them only here (and with only partial devoicing): ('vFstkx, -txk) /'vstafka/ bvctabka, (pt'mjuSt, &pxt-) /pata'muSta/ otomy uto, (';vuSk)
/'djevuSk/ debywek. is may happen to /i/ >= in diphthongs as well, before a
pause (although a {voiced} V may precede, instead of a voiceless C), as well as to
V exclusively followed by voiceless C: ('=im~i) /'zimnii/ $mn$=, ('Ce;p) /'cerip/
uepe is also occurs in stressed syllables (before a pause): (mu'=e) /mu'zjei/ mye=, (tRM'va) /tram'vai/ tpamba=.
8.3.1.8. Written geminate C (and >) are phonically short before a pause or
before a C: (g'RFa) /g'ram/ gpamm, ('vFan) /'van/ bann, (d'qFn) /d'qn/ d$nn,
(k'Fas) /k'las/ klacc, (p'Fa) /p'laS/ la, (pRg'RFamni) /prag'ramni/ pogpammnv=, ('Rjusii) /'ruskii/ pycck$=, (g'Rjupkx) /g'rupka/ gpyka, ('mjni)
/'moSni/ monv=
thin lexemes, written geminate C can be realized as phonetically slightly
geminated C (00): (&ss'nFans) /assa'nans/ acconanc, ('vFannx) /'vanna/ banna,
('kFassx) /'kassa/ kacca, ('tjnnx) /'tonna/ tonna
However, more often CC are pronounced short: (&Is'tFat) /atis'tat/ attectat, (b'jn) /ba'lon/ allon, (b'ein) /ba'sjein/ acce=n, (gR'mFa;Ikx) /gra'matika/ gpammat$ka, (d&iI'Rjkx) /drisi'rofka/ dpecc$pobka, (I';=i[]x) /i'ljuzija/ $ll$r, (&kxIk'if) /kalik'tif/ kollekt$b, (&mii'mt5, -;t) /mili'mjetr/ m$ll$metp, (&ku'pFa;qx) /aku'paqja/ okkyaz$r, (&pxR'e1) /para'ljelj/ apallel,
(pI'Rjn) /pi'ron/ eppon, (&Iq') /riq'sjor/ pe$ccep, ('tF;~Is) /'tenis/ tenn$c,
(&iI'tj;i[]x) /tiri'torija/ tepp$top$r, (tR'eibus) /tra'ljeibus/ tpolle=yc, (I'nqx) /i'sjenqja/ <ccenz$r, (I'fkt) /i'fjekt/ <ffekt
8.3.1.9. In mid-fast speech, unstressed syllables are reduced (in particular the
non-initial ones occurring immediately before a stress, or the nal ones after a
stress), especially near /r, l/: (px&mx', &pxm-, &pxm-) /pamala'djel/ omolo-
8. russian
299
del, (mx&xd'qj;n, &mxd-, &mxd-) /malada'qon/ molodoenv, (px&xR'~i,

&px-, &pxR-) /paxara'nitj/ oxopon$t, (&kxR'ju;i, kx'j-, kRx'j-) /kara'ulitj/
kapayl$t, (&kxR8'in, k8-, kR8-) /karan'tin/ kapant$n, (&pxIk'mFa;I, phIk-,
pIk-) /parik'maxir/ ap$kmaxep, (&txR'kFan, t-, tR-) /tara'kan/ tapakan, (mx&xk'sjs, &mxk-, &mxk-) /malaka'sos/ molokococ, (gx&xs'vFa, &gxs-, &gxs-)
/galasa'vatj/ golocobat, (px&RxSu'ist, &pxSu-, &pxSu-) /paraSu'tist/ apawt$ct,
(bi&Is'vFa;ti, &bis-, &bi1s-) /bilisa'vati/ elecobatv=.
Other examples: (dx&Rxg'vFa;ti, &dxg-, &dxRg-) /daraga'vati/ dopogobatv=,
(=i&In'vFa;ti, &=in-, &=in-) /zilina'vati/ elenobatv=, (pi&Id'vji, &pihd-, &pid-) /pirida'voi/ epedobo=, (pi&ivIR'nju, &pihvIR-, &pivIR-) /pirivir'nutj/ epebepnyt, (&dmi&~IstR'i6ni, -s-, -8s-) /administra'tivni/ adm$n$ctpat$bnv=, (I&~Iq'i;vx, &Iq-, &Inq-) /iniqa'tiva/ $n$z$at$ba, (I&~Iq'Fa, &Iq-, &Inq-) /iniq'al/ $n$z$al.
More: ('i;kxxtkx, -xtkx, -xtkx) /S'Sikalatka/ $kolotka, ('sju;dxRxgx, -dxgx,
-dxRgx) /'sudaraga/ cydopoga, (p'i;txxkx, -txkx, -txkx) /'pritalaka/ p$toloka,
('qFa;vxRxnxk, -vxnxk, -vxRnxk) /'qavaranak/ aboponok, ('j;xdnx, -dnx,
-dnx) /'xoladna/ xolodno, (z';xx, -x, -x, -x, ='-) /z'djelala/ cdelala,
(&bI'zFa;i1stvx, -Istvx, -stvx, -#tvx, -astvx) /abi'zatiljstva/ oratelctbo,
(ds'tFa;txCnx, -atCnx, -aCCnx) /das'tatacna/ doctatouno, (mI'nju;txCkx, -tCkx,
-Ckx) /mi'nutacka/ m$nytouka, ('n;vxvx, -*vx, -vvx) /'novava/ nobogo, ('sj;vvx, -*vx, -vvx) /f'sovvatj/ bcobvbat, (d'mj;vxvx, -*vx, -vvx) /da'movava/
domobogo, (xs'i;vxvx, -i*vx, -ivvx -'-) /SSast'livava/ cuactl$bogo, (nxk';vvxqqx,
-*vx-, -vvx-) /nak'ljovvaqqa/ naklebvbatcr, (&zxSpk';vvx, -*vx, -vvx)
/zaSpak'ljovvatj/ awaklebvbat, ('qFastvxvx, -t*vx, -tvx) /'qarstvavatj/
zapctbobat.
8.3.1.10. In the formation of words, phrases, and sentences with prexes or
prepositions there are various particular clusters of V's which otherwise would
not be normal in Russian. ere are frequent reductions, since these positions are
mainly those before a stress: (pg'iisi, pg-, -ng-) /paang'liiski/ o-angl$=ck$,
(z'e, z'-) /zaa'ljetj/ aalet, (nk'~;, nk-) /naak'nje/ na okne, (vp';,
vp-) /vaap'Se/ booe, (st'vqtvxvx, st-, -t*vx, -tvx) /saat'vjeqtvavatj/ cootbetctbobat, (nd'nji, nd-) /naad'noi/ na odno=, (zs'ti, zs-) /zaas'tritj/
aoctp$t, (&stn'i, &st-, &sxt-) /saatna'sitj/ cootnoc$t, (&sb-R'qF;~i[]I,
&sb-, &sxb-) /saabra'qeniji/ coopaen$e, (v&duS6'e;~i[]I, v-, vx-) /vaaduSv'ljeniji/ boodyweblen$e.
More: (&vIx8'Ce1, &vI8-) /vialan'celj/ b$olonuel, (s&pIq'Fani, -'qFa-) /spiq'aljni/ cez$alnv=, (&sxq'iz, -q-, -;z) /saqa'lizm/ coz$al$m, (&~Idnk'RFatni, &~Ixd-, &~Id-) /niadnak'ratni/ neodnokpatnv=, (~I&bkn'vnni, ~iI-, ~I-)
/niabkna'venni/ neovknobennv=, (&~i[]Iz'vsnx, ~Iz-) /nijiz'vjesna/ ne$bectno, (&nug'jk, &nxu-, &nu-) /nauga'lok/ na ygolok, (&pug'vj;Ru, &pxu-, &pu-)
/pauga'voru/ o ygobopy, (&udv'ka;tx, &uxd-, &ud-) /uadva'kata/ y adbokata, (&ug'Rj;dx, &uxg-, ug-) /uaga'roda/ y ogopoda, (&pImI'nFa, &pxIm-, &pIm-, &pm-) /paimi'nam/ o $menam.
300
8.3.1.11. Some personal pronouns have reduced forms: (i'ba;, i'a;, i&a, ,
x) /ti'bja/ ter, (i'b;, i';, i&, I) /ti'bje/ tee, ('vFas, &vas, vs, vxs) /'vas/ bac,
('vFa, &va, v, vx) /'vam/ bam, ('j i&b'dFa, 'j I'dFa) /'ja ti'bje 'dam/ r
tee dam, ('ja &vam'dFa, 'ja vxm'dFa) /'ja 'vam 'dam/ r bam dam.
Certain numbers have reduced forms too (although we only provide the models, which are used for similar forms as well): (&ii'i;, i'i;) /disi'ti/ decrt$, ('i;nxqqx, -nxqx, -nqx) /a'dinaqqatj/ od$nnadzat, ('i;nxqqxti, -nxqxti,
-nqxti, -nxqti) /a'dinaqqati/ od$nnadzatv=, (d'vFaqqx, d'vFa;qx, d'vFaq$)
/d'vaqqatj/ dbadzat, (d&vFaqq'i;, d&vFaq'i;, dvaq'i;, dvq'i;) /dvaqqa'ti/
dbadzat$, (&pii'at, &pii'-, pi[]i'-) /piddi'sjat/ rtdecrt, (pi'i;iiI, -'i;iI, -'i[;]iI) /pi'tidisiti/ rt$decrt$, (&S=i'at, &S=i'-, Si'-) /Szdi'sjat/ wectdecrt, (S'i;iiI, -'i;iI, -'i[;]iI) /Ss'tidisiti/ wect$decrt$, ('emixt,
'e)xt) /'sjemdisjat/ cemdecrt, (i'mi;iiI, -'mi;iI, -'mi[;]iI) /si'midisiti/
cem$decrt$, ('vj;imixt, 'vj;i)xt, 'vj[;]i)xt) /'vosimjdisjat/ bocemdecrt, (v'mi;iiI, -'mi;iI, -'mi[;]iI) /vasj'midisiti/ bocm$decrt$.
8.3.1.12. rst and patronymic names are usually reduced by frequent use, more
than by particular phonemic rules: ('j;IpxviC, -ipiC, -IpC, -pviC, -pC,
-pC) /'osipavic/ Oc$ob$u, (b'i;sxviC, -'iviC, -'i;iC, -'i;sC, -'iC, -'i) /ba'risavic/ op$cob$u, ('pFavxviC, 'pFa;vviC, -aviC, -avC, -a;iC, -a;C, -a1C, -aC) /'pavlavic/ ablob$u, (&Ik'sFandRxviC, -andviC, -andviC, -andiC, -andRC, -a;~iC, -a;nC, &Ik'sFa8C, Ik'sFa8C, k'sFa8C) /alik'sandravic/ Alekcandpob$u, (v'i;mIRx6nx, -mIR6nx, -mIRxnx, -mIRnx) /vla'dimiravna/ Blad$m$pobna, ('f;dxRx6nx, -dxRxnx, -dR*nx, -dxRnx, -dnx) /'fjodaravna/ Fedopobna, (b'i;sx6nx, -'isnx) /ba'risavna/ op$cobna, (kR'~e[;]i6nx, -'~6nx) /kar'njejivna/ Kopneebna, (f'e[;]i6nx,
-'i6nx, -'6nx) /fa'djejivna/ Fadeebna, (p'fi-jI6nx, -'fi;I6nx, -'fi;Inx, -'finx)
/par'firijivna/ opf$p$ebna, (s've;LI6nx, -'ve;I6nx, -'ve;I6nx, -'ve;Inx, -'venx)
/sa'vjeljjivna/ Cabelebna
In addition: ('Fannx mI'Faix6nx, -*nx, -xnx, -nx) /'anna mi'xailavna/ Anna
M$xa=lobna, (&Ik'sFand RIk'sFandRxviC, &Ik'sFand RIk'sFandRC, Ik'sFand RIk'sFandRC, Ik'sFand RIk'sFa8C, Ik'sFan Ik'sFa8C, Ik'sFan 'sFa8C, 'sFan 'sFa8C) /alik'sandr alik'sandravic/ Alekcandp Alekcandpob$u, (&kxnst8'i; ns'Fa;kxviC, -'i;
nI-, &kxns'i; ~I'sFakC) /kanstan'tin sa'akavic/ Konctant$n *caakob$u, ('pFa;vI
'vFa;nxviC, -vi lI-, -vFa8C, 'pFa 'vFa8C) /'pavil i'vanavic/ abel *banob$u
8.3.1.13. Currently, /ji/ >e r becomes /i/: (jI'mju;, I-) /ji'mu/ emy, (jI'qF;, I-) /ji'qi/ e$, (jIv'aqqx, Iv'aqqx, Iv'aq[q]) /jiv'ljaqqa/ rblrtcr, (&~i[]i'stvInnx, ~ii'svInx -'ev-) /nijis'tjestvinna/ neectectbenno.
In general, /ij/ = /i/ ( 8.2.5.1): (vi'i;x, -'i;x, -'ix) /vi'tija/ b$t$r, ('FaRmix, -mix, -mIx) /'armija/ apm$r, (&pi'iqqx, &pi-) /priju'tiqqa/ p$t$tcr,
(pi'j+~iI, pi'+~iI) /pri'jatniji/ p$rtnee, (&qi'j;, -i';) /qti'je/ $t$e, (&bi'ju;, &bi'u;) /'btiju/ vt$
Besides, usually /0jj/ = /0j/ and /ii0/ = /i0/: (&pi-jI'vji, &piI'vji) /pirjji'voi/ epebo=, (b'CC-jI, -;CI) /b'cacjix/ vuau$x, ('kFa=N-ju, 'kFa=~-ju,
'kFa=~u) /'kaznjju/ kan, (u'bjv-ju) /lju'bovjju/ lob, (Svii'qFa, Svi'-, SvI'-)
8. russian
301
/Svii'qar/ wbe=zap, (iist'vi;inx, Ist- ii- i-) /diist'vitiljna/ de=ctb$telno.

In the case of prepositions + /j/, we have /`/ too: (vjI'vj 'dj;mI, v'vj)
/v-ji'vo 'domi/ b ego dome, (sjI'pjnqxmI, s'pjn-) /s-ji'ponqami/ c ronzam$;
however, by now we frequently hear: (vI'vj 'dj;mI, I'pjnqxmI), and also:
('ju;gu) /'kjugu/ k gy, instead of ('kju;gu).
8.3.1.14. In traditional pronunciation, /i/ >$- preceded in sentences by C
(which is dierent from /c, S, 0j/ >u t) becomes // (less systematically so
in the case of /k, g, x/, thus we also nd (k, g, ) + (i, I)). However, in modern
pronunciation /i/ remains unchanged, even without changing the C (, not (J)),
but with /k, g, x/ (, , ) + (i, I)): (s'm I'gj;I)m ('m 'gj;, i-)t /s'mjex i'gori/ cmex $ gope, ('i;gxu)m ('kF-, 'kji;-)t /'kigarju/ k *gop, ('dF mI't)m (m-)t
/'dm i'djot/ dvm $det, ('tjivi[]Ivx)m (a'tFv-)t /a'tivlijiva/ ot *bl$eba, ('kj tI'pj;vx)m (t-)t /'kot i'povar/ kot $ obap, (&jnIz'djyb)m (&jnz-)t /on-iz'doxb/ on $dox v, (I'vFa;nx)m (s-)t /s-i'vanam/ c *banom, (I'vFa;nu)m (k-, ki-)t
/k-i'vanu/ k *bany, (t'vFa;i I'vFan)m (- C-)t /ta'variS Si'van/ tobap$ *ban.
Traditionally, Italia, *tal$r behaves in the same way, so much so that even
for Alitalia ( Ali ngs + Italia] we have: (--, -I-)t rather than (-I-)m.
Again in traditional pronunciation, even /i/ ><- preceded in sentences by C
(which is dierent from /c, S, 0j/ >u t), becomes //, although it often remains /i/. In modern pronunciation, instead, we have (I) /i/ (and, above all in alternative pronunciation, () /e/), both in absolute initial position and when preceded by C: (&tII'vFa;txRx, &tI-)m (&ti-, &tIi-)t /at-ili'vatara/ ot <lebatopa In
absolute initial position we nd respectively: (&II'vFa;txRx)m (&-, &-)a (&i-, &Ii-)t /ili'vatara, -/ <lebatopa (I'pj;x)m (-, -)a (-, i-)t /i'poxa, -/ <oxa
For the non-autoctonous >=enwte=n, we have: (&i=INS'tFin, &I-, -z-) /eizinS'tein, i-, -ze-/; and (&si=INS'tFinx, &si=-, &sIz-, -N-) /seizinS'teinam, si-, -ze-/ c >=enwte=nom
Pronunciation variants: modern, traditional and alternative

8.3.2.0. It is undeniable that Russian pronunciation is in general poorly known,
both by native and foreign speakers, even if we refer to experts. In addition, there
are dierent methods of description and evaluation, apart from dissimilar applications of dierent criteria on the subject. As we have already said, even ocial
model speakers, of the ocial radio and television broadcasting corporation too,
uctuate between more or less modern or traditional pronunciations (or alternative as well), most probably because of the lack of specic information, too. us
comparisons and convenient choices are dicult to make, since the kind of transcription used in handbooks and dictionaries (even in ocial and scientic
ones) is nothing but a graphemic respelling mixed with a few other signs.
However, and with very good reasons, what has been described thus far, belongs
to modern neutral Russian pronunciation, which is more and more accepted and
302
widespread (and also relatively less far away from spelling). We will now indicate
the dierences which mainly fall within traditional pronunciation, according to
the old Moscow pronunciation which was acquired orally. It used to include
some strange exceptions, which in some respect contributed to dierentiate it
from (St) Petersburg pronunciation, generally closer to orthography. Since then,
this city has been called Petrograd, then Leningrad and now Saint Petersburg: so
to say rename it and you'll get over it!.
8.3. Vowel elements of traditional (and alternative) Russian.
/JuJ/ (%)t, /JuJ/ (T)t
/'i, 0iJ, i'/ (i)t, /i/ (I)t
/'u, u'/ (u)t, /u/ (U)t
/', '/ ()t, // ()t, /'/ ()t
{/u/ (, , )a}
/...'/ (, )t:a /'..., / ()t:a
/'JeJ/ (e)t
/'JoJ/ (+)t
/'o/ {/o/} () {/'o, "o/ ([])}t
/'e/ () {/'e, "e/ ([])}t
/a'/ (a)t, /a/ (x)t
/'JaJ/ (, "[])t
/'a/ (a), /'a, "a/ (a[])t
>e /i/ ()t, >e, r, a /i'/ (', J'J)t >e /i[]/ ()a, /0iJ/ ()a, >< /i/ ()a
8.3.2.1. In interpalatal contexts, traditional pronunciation presents decidedly fronter and closer articulations ( 8.3 8.1.4, although the symbols (i, e)
are identical, but stand for (i, e)): ('bi) /'bitj/ $t ('je1) /'jelj/ el, ('p)
/'pjatj/ rt, ('+;x) /'tjotja/ tetr, (%'fak) /tju'fjak/ tfrk
In pre-tonic syllables ( immediately preceding a stressed syllable), traditional
pronunciation has closer realizations than the modern ones for unrounded /i, /.
Instead, the rounded /u/ is opener in all unstressed positions, except again for
pre-tonic syllables ( 8.3, again): /i', ', u'/ (i', ', u')t (I', ', u')m: (&bi'kjs)t (-I-)m
/abri'kos/ ap$koc, (b'i;)t (b-)m /b'ki/ vk$, (u'qF;)t:m /u'qe/ ye. Pre-tonic /a'/
is (a')t (')m: (&Uga'jk)t (&ug-)m /uga'lok/ ygolok.
Most typically, traditional pronunciation also has /e'/ (', '+J) for pretonic >e,
r (including >ua a, excepting >$, which has /i'/ (i')). All of these in modern
pronunciation have /i'/ (I', i'+J)): ('i)t (i'i)m /di'litj/ del$t, (~'i;)t (~i'i;)m /nis'ti/ nect$, (t'i;)t (ti'i;)m /tris'ti/ tprct$, (C'sF;)t (CI'sF;)m /ci's/
uacv Let us thus consider (i'sFa;)t (I'sFa;)m /li'sa/ l$ca and ('sFa;)t:a (I'sFa;)m /li'sa/
leca In /J'/ contexts, only another modern phoneme occurs, /u/ (u, ): (u'dFa;)m:t /sju'da/ cda, ('bi)m (%'bi)t /lju'bitj/ l$t
Modern pronunciation has /i, ji, ji/ for >e in unstressed syllables, and /i/ for
(initial) ><. In all these cases, alternative pronunciation, strengthened by orthography, can have /e/ (, J): (b';zx)a (bi-)m /bi'rjoza/ epea, (&b'qj;ni)a (&biI-)m /biri'qoni/ epeenv=, (&m'na;)a (&imI-)m /simi'na/ cemena, (p&m'i;)a (pi&imi-)m /pirimi'li/ epemel$, ('be;k)a (-Ik)m /'bjerik/ epeg, (&bI'zFa;nx)a (-i-)m /abi'zatiljna/ oratelna, ('vF;)a (-ii)m /'vtiritj/ bvtepet, ('tFaS, -)a (I'tFaS)m /i'taS/ <ta, ('mj;)a (-)t (-I)m /'mori/ mope ('pj;)a (-)t (-I)m /'polji/ ole (for >e an alternative pronunciation with ()t is particularly frequent for neutral grammatical gender).
8. russian
303
8.3.2.2. Modern pronunciation has () for //, whereas alternative (and often
traditional) pronunciation presents opener variants, (, ). Most frequently, we
nd ()a:t before a stress; ()a:t after it; and ()a ()t in absolute nal position: (&bta'vji)t (&bt-)a (&bt-)m /bta'voi/ vtobo=, (&va'i)t (&v-)a (&v-)m /vxa'ditj/ bvxod$t, ('vF;m%)ta (-m%)m /'vml/ bvmvl, ('j;pt)t (-pt)a (-pt)m
/'opt/ ovtv (q'vjtnx)t (-nx, -)a (q'vjtnx[]I)m /q'votnaji/ $botnoe.
As we have already said, the timbres of // (, ) are fairly similar to those of /a/
(x), thus it is easy to think of them as the same sound and to declare their neutralization (as quite a few authors have done). Actually, in general, there is such a
dierence, so that they are rarely confused, even in spelling, except for evident cases of poor education.
For >e, in traditional pronunciation we often nd ()t even for // (in addition to /i/, most surprisingly, as we have seen at the end of 8.3.2.1): ('juCS)m
(-)t /'lucS/ lyuwe ('ju;q)m (-)t /'xuq/ xye Here we may actually speak of possible neutralization between /i/ and //, which was more typical in former times.
For /u/, alternative pronunciation may have (0, 0, 0 J, JI, JJ, JiJ) (in
8.3, () is indicated by a thin broken line and glossed in brackets): (&kv'kFaqqx, &k-, &k-)a (&kuv-)m /kuvr'kaqqa/ kybvpkatcr, (&sm'tj;x, &s-, &s-)a
(&su-)m /suma'toxa/ cymatoxa, ('zFa;mq, -m-, -m-)a (-uq)m /'zamuqm/ amyem, (kx&m~i'j;, -mi-)a (-m~i-)m /kamjuni'kje/ kommn$ke, (I'me;&iix,
-[]i&-)a (-&-)m /i'mjejuSSiisja/ $me$=cr, ('dFa;, I-)a (-)m /sju'da/ cda
In traditional pronunciation, we can even nd: (mid've;mI)t (mId've;imI)m
/mid'vjedimi/ medbedrm$ (mi've;)t (mId've;i)m /mid'vjedim/ medbedem, ('vF;~s)t (-Is)m /'vnis/ bvnec and, on the other hand, also: (i'vj_~I)t (I'vjd~x)m /si'vodnja/ cegodnr.
8.3.2.3. Adjectives ending in -k$=, -g$=, -x$=, in traditional pronunciation have
/i/ (i), but also (i) (usually rendered as /j/, as if it were (xi) /ai/ {while being
slightly dierent, 8.1}), when they are preceded by (k, g, ), instead of the
modern rendering as /ii/ (ii, ii, ii) (according to spelling): (z'vj~ki, -ki, -k-)t
(-ii, -8ii)m /z'vonkii/ bonk$=, (s'tRj;gi, -gi)t (-ii)m /s'trogii/ ctpog$=, ('i;i, -i)t (-ii)m /'tixii/ t$x$=
Nowadays, this traditional pronunciation sounds non-neutral, or lofty. Also
verbs in -$bat behave in this way: (pa'mFakvx)t (p'mFaIvx)m /pa'malkivatj/ omalk$bat, (na'a;gvx)t (n';Ivx)m /na'tjagivatj/ natrg$bat,
(pa'mFa;vx)t (p'mFa;Ivx)m /pa'maxivatj/ omax$bat
For unstressed endings with -V=, modern pronunciation has (ii) /ii/ -$=, (ii) /ii/
-e=, (i) /i/ -v=, (xi) /ai/ -o=; the traditional one has /i/ (i) for -v=, and the alternative one has /ei/ (i) for -e= (and, in general, also /i/ (i) -v=). ese are
often described as neutralization between the rst two (and it is the same in modern pronunciation) or between the second two (but we have: (i)m/(i)a V (xi)):
('i;~ii) /'sinii/ c$n$=, ('ju;ii)m ('ju;i)a /'ulii/ yle=, (t'Rjudni)m (-ni)t /t'rudni/
tpydnv=, (t'Rjudnxi) /t'rudnai/ tpydno=
For pretonic wa, a, za (even with >o), modern pronunciation has /0a'/ (0'),
whereas the traditional one had /0'/ (0') (which is now lofty): (S'i;)m (S-)t /Sa'gi/
304
wag$, (q'RFa;)m (q-)t /qa'ra/ apa, (d&vxqq'i;)m (-qq-)t /dvaqqa'ti/ dbadzat$

However, for certain words the traditional type of pronunciation is still frequent:
akt acmn abl ewaml dbdzat tpdzat lowad= In general, these sequences even have an alternative pronunciation with (', ') for /'/;
the same occurs in pretonic we, e, ze as well (even with >o). But modern
pronunciation has /'/ = ('): (Sp'tFa) /Sp'tatj/ wetat (q'nFa;) /q'na/ ena
(q'nFa;) /q'na/ zena.
8.3.2.4. Among the greatest dierences in comparison with orthography, in traditional pronunciation, we nd unstressed verbal endings -rt, -r$=, which have
/u/ (that however nowadays is rightly considered to be lofty or non-neutral) instead of /a[0], i/ in modern pronunciation: ('j;Ut)t ('j;xt)m /'xodjat/
xodrt, (s'tRj;Tii)t (s'tRj[;]iii)m /s'trojiSSii/ ctpor$=
Another non-neutral pronunciation, which is possible to hear (this time with
the opposite timbre), concerns feminine nouns such as ('djb-Ruu -xu) /'dobruju/ dopy, ('i;~u -i[]u) /'sinjuju/ c$n
Traditional pronunciation can have (x) for (non-pretonic initial) /a/ () and
() for (nal) /a/ (x): (&xdn'vj;)t (&d-)m /adna'vo/ odnogo, ('pjzn)t (-nx)m
/'pozna/ odno
8.3.2.5. Traditional pronunciation has (C)t (or, as we have already said, (C)t or
(C)t) for (, , )m /S[S]/, but in Moscow the latter more agile pronunciation
has always prevailed: (Ci'kFa;)t (I-)m /SSi'ka/ eka, (ta'vFa;iC)t (t'vFa;i[])m /ta'variS[S]/ tobap$ Contrary to the most frequent transliteration type (), the grapheme > might seem to suggest a closer articulatory relation with >w (S) /S/; and
indeed a constrictive pronunciation is decidedly more recommendable: ('jI)
/S'Sastjji/ cuacte, (&pii'piIk) /piri'piSSik/ epe$cu$k (the -u$k sux is not interpreted as a separation element).
However, in modern pronunciation also (C) /Sc/ is normal, when a clear morphemic boundary which is shown in writing too is present, with >cu u wu
u ctu du, never >: ('C) /S'cem/ c uem, (&ICI'vj;) /iSci'vo/ $ uego /SS/
or /Sc/ occurs in: (vIs'njuIti, -C-) /vis'nuSSiti, -Sc-/ becnywuatv=, (b'RjIti,
-C-) /ba'roSSiti, -Sc-/ opoduatv=.
But, between a lexeme and a sux, we generally nd () /SS/: (Iz'vjIk) /iz'voSSik/ $bou$k, (&pii'beIk) /piri'beSSIk/ epeeu$k, ('qjI) /'qoSSi/ ectue Between a prex and a lexeme, (C) /Sc/ is most frequent: (bi&CI'veCni) /biScila'vjecni/ ecuelobeunv=, (&ICI'i) /iScir'titj/ $cuept$t, (&RxCI'aqqx) /raSci'xaqqa/ pacu$xatcr. In most commonly used words and when a prex is no longer thought of as something separated, we usually nd () /SS/: (Rx'skx) /raS'Soska/ pacuecka, ('jI) /S'Sastjji/ cuacte, ('t) /S'Sot/ cuet, (I's) /iS'Ses/ $cue
nally, the phoneme /ZZ/ () (which occurs in few words and only within lexemes) is more typical of traditional than modern pronunciation: ('t)t (q'qjt)m
/q'qot/ et, ('jeU)t ('jqqu)m /'jeqqu/ ey, (&U[]i')t (&u[]Iq'qFa)m /ujiq'qatj/
yeat, ('pjI)t ('pjqq)m /'poqq/ oe for derivative forms of ('djC)t
('dj, 'djS)m /'doS, 'doStj/ dod we have: ('djIk)t ('doZZik/ ('djqIk)m
8. russian
305
/'doqdik/, dod$k, (da'i;)t /daZ'Zi/ (dq'i;)m /daq'di/ dod$, (da')t /daZ'Zom/ (dq')m /daq'djom/ dodem
8.3.2.6. As we have seen, in (even modern) Russian, by voicing assimilation,
we have (a voiced semi-constrictive) (y) for /x/: (&jnIz'djyb)m (&jnz-)t /oniz'doxb/ on $dox v. In a few words belonging to a particular area, traditional pronunciation has: ('bj;yx)t ('bj;gx)m /'boga/ oga, (yas'pj)t (gs'pj)m /gas'potj/ gocod, (buz'yFa1I)t (buz'gFa1I)m /bjuz'galjtir/ ctgaltep (besides,
/gi/ (JI)t, with a true constrictive): (a'bj;JI)t ('bj;I)m /a'bogi/ o oge). e same
occurs for the following forms, which still remain the same even in modern pronunciation: ('bj)t (-, -k)m /'box, -k/ og, ('yjspxI)t (y-, g-)m /'gospadi/ Gocod$! however, it is thought that before long, (y) for /g/ will eventually disappear
completely (followed by /ZZ/ = /qq, qd/, as we have seen).
For the preposition k followed by g- we have: (y'gj;RxdU)t (g'gjRxdu)m /g'goradu/ k gopody. Lastly, also in modern Russian we nd (H, y) for the /h/ phonostyleme in exclamations and onomatopoeic words: ('HFa;, 'yFa;) /a'ha/ aga!
In traditional pronunciation /k, g/ are (, y) before stops: (tay'dFa;)t (tg'dFa;)m
/tag'da/ togda, ('tj;)t (k'tj;)m /k'to/ kto, (ka'mju;)t (kk-)m /kka'mu/ k komy
For -g before the ending -k$= and derivatives, we have (, k , ): ('mii)m,
('mki)t /'mjakkii, -xki/ mrgk$=, ('makkxmu)m, ('makxmU)t /'mjakkamu,
-xk-/ mrgkomy, ('kki)m, (-k-)t /'ljokki, -xk-/ legkv=, ('makCI)m, (-CI)t
/'mjakci, -xci/ mrgue, ('Pii)m, ('ki)t /'ljokkii, -xki/ legk$=, ('kCI)m ('CI)t /'ljekci, -xci/ legue. But (k) will prevail, even though it is often still considered to be a non-neutral pronunciation. In fact, for instance, trgua=w$= is already (Ik'CiSi, I-) /tik'caiSi/, and otrgu$t is (&Ik'Ci, -'C-) /atik'citj/.
In gt, kk sequences we have: ('kjkI)m (-I)t /'kokti/ kogt$, ('njkI)m (-I)t /'nokti/ nogt$, ('kx)m (-x)t /'djoktja/ degtr
e pronoun uto is always (S'tj;) /S'to/, as its compounds are, except neuto
('~eCtx) /'njecta/; whereas n$uto can have two pronunciations: (~IS'tj; ~iC'tj;) /niS'to nic'to/. In addition, >u stands for /S/ in feminine patronymic names:
('v;Rx IL'ji;~ISnx) /'vjera ilj'jiniSna/ Bepa *l$n$una, and in: (k'~Snx) /ka'njeSna/ koneuno, (n'RjSnx) /na'roSnx/ napouno, (s'kjuSni) /s'kuSni/ ckyunv=, (ji'[]iS~Iqx) /ji'jiSniqa/ r$un$za
e graphemic sequence un used to be /Sn/, in the traditional pronunciation of
common words. In modern pronunciation, we have /cn/, even though for some
words both pronunciations are used. However, in general, we nd: ('bju;xCnxx)m (-Sn-)t /'bulacnaja/ ylounar, ('lFa;vxC~Ik)m (-S~-)t /'lavacnik/ laboun$k, (m'jCni)m (ma'jSni)t /ma'locni/ molounv=
8.3.2.7. In traditional pronunciation, the C's that precede ( palatal or palatalized C] are more consistently palatalized than in modern pronunciation: (za'me;~i[]I)t (zt-)m /zat'mjeniji/ atmen$e, (s'kjp)t (s'kjp)m /s'korpj/ ckop,
('CeI)t ('CI)m /'certi/ uept$.
More: ('eI)t ('I)m /'djefki/ debk$, ('ve)t (d'v-)m /d'vjerj/ dbep,
('min)t (t'min)m /t'min/ tm$n, ('mit-ii)t (d'm-)m /d'mitrii/ Dm$tp$=, ('m;-
306
nx)t (s'm-)m /s'mjena/ cmena, ('me)t (s'm)m /s'mjertj/ cmept, (pi'nFa;)t

(s'pI-)m /spin'a/ c$na, ('f;Rx)t (s'f-)m /s'fjera/ cfepa, ('vt)t (s'vt)m /s'vjet/ cbet,
('~i;gx)t (k'~i-)m /k'niga/ kn$ga, (pI&sxqa'i=mI)t (-'izmI)m /prisaqa'lizmi/ p$
coz$al$me. A pronunciation with a generalized palatalization, although quite
widespread, is not at all neutral: (&kxmu'~iz, -;z, -mm- -=))t (&kxmu-)m /kamu'nizjm, -mm-/ kommyn$m Not even in traditional pronunciation is /l/ palatalized
by assimilation: ('mjvI) /'molvji/ molbe, nor by gemination: (&pj'it-Rx) /pol'litra/ ol-l$tpa
Traditional palatalization is more tenacious within words than at the beginning:
(&nxvi'~;Ci)t (&nxsvi'~;Ci)m /nasvi'njacitj/ nacb$nru$t, (viN'ja;, sv-)t (sviN'ja;)m /svinj'ja/ cb$nr
In addition, palatalization due to assimilation is normal, even in modern pronunciation, for homorganic phonemes (except for /l, r/), although a pronunciation without palatalization is spreading. In fact, it further shows that a phonemic
approach like the one adopted here is more useful: ('SF, -s) /'Sestj/ wect, (k'~e, -s'~-) /kas'njetj/ kocnet, ('ep, s'-) /s'tjepj/ cte, (='e, z'-) /z'djesj/
dec, (mi'i;, smis'i;) /smis'ti/ cmect$
For heterorganic phonemes, indirect palatalization is no longer current: (Iz'b;)m (i='b;)t /iz'bje/ $e, (Rz'm)m (Ra='m)t /raz'mjatj/ pamrt, (s'm)m
('m)t /s'mjex/ cmex en palatalization is required, spelling shows it too: (i='b;) /rizj'bje/ pee, (v='mi;) /vazj'mi/ bom$, (i'm;) /tisj'mje/ tecme
8.3.2.8. Please, note the dierences between modern and traditional pronunciation, as far as /rJ/ is concerned: ('RmIn)m ('emIn)t /'tjermin/ tepm$n, (I'pit)m (i-)t /tir'pit/ tep$t, ('Rbi[]x)m ('eb-)t /'sjerbija/ Cep$r, ('vf)m ('vef)t
/'vjerfj/ bepf, ('CRvI)m (-vI)t /'cervi/ uepb$, ('R~i[]I)m ('e~-)t /'tjerniji/ tepn$e, ('qF)m (-)t /'qertj/ epd, (u'Ri[]I)m (-ei[], -[])t /u'sjerdiji/ ycepd$e, ('Ri)m (-e-)t /s'tjerlitj/ cteplrd, ('vi[]x)m ('ve-)t /'vjersija/ bepc$r,
('vR=Ix)m ('ve=Isx)t /s'vjerzilsa/ cbep$lcr, (k'mCIsii)m (ka'meCiii)t /ka'mjerciskii/ kommepueck$=
Even in traditional pronunciation, as in the modern one, we nd (RJ) after ('0)
(with non-front V]\ ('pFai[]x) /'partija/ apt$r, ('jRI) /'ordir/ opdep, ('djuR~I) (-)t /'durni/ dypne, ('mFaRx) /'marlja/ maplr, ('kjuI) (-)t /'kursi/ kypce e
same occurs after unstressed V (even if they are front ones): (vI'e) /vir'tjetj/ beptet, (IR'vis) /sir'vis/ cepb$, (vIR'=i;x) /vir'zila/ bep$la, (=IR'~Isti) /zir'nisti/ epn$ctv=
C + () are not palatalized: (v'e;mx) /v'rjemja/ bpemr (f'skx) /f'rjeska/ fpecka
(p'tt) /par'trjet/ optpet (n'di) /xan'dritj/ xandp$t (pIz'it) /priz'rit/
pep$t
For reexive forms (in -cr -c] traditional pronunciation has /s/ (not /sj/), except for nal-stressed gerunds: (ba'j)t (b'j)m /ba'jasj/ orc (against (&sxbi'RFa;xs)t (-bI-)m /sabi'rajas/ co$parc] On the contrary, in modern pronunciation reexive verbs have /sj/ (), except in the third person (in -tcr) and innitives (in
-tcr): (~i'qqx) /ni'sjotsa, -qqa/ necetcr, (k'djuqqx) /kla'dutsa, -qqa/ kladytcr, (u'Ciqqx) /u'citsa, -qqa/ yu$tcr
8. russian
307
In particular, front-lingual C undergo palatalization by assimilation, in modern pronunciation too: ('mjIk) /'mostik/ moct$k, (i'~ik) /lis'nik/ lecn$k,
(u='eCkx) /uz'djecka/ ydeuka, (R'bj+~Ik) /ra'botnik/ paotn$k, (_'~i;) /ad'ni/ odn$, ('bFa8Ik) /'banSik/ an$k. Before (), after (, R), or in prexed words,
or in word-initial position, both pronunciations occur (though palatalization is
more traditional): (pIt'a;, pi+-, 'pe+-x)m (p+-, 'pe+-)t /pit'lja, 'pjetlja/ etlr, ('jsIk, --) /'oslik/ ocl$k, (ts'ak, -'-)m (ta'-)t /tals'tjak/ tolctrk, ('mRz~It,
-=~-) /'mjorznit/ mepnet, (sI'nFa;, -)m (-)t /sti'na/ ctena, (z'e, ='-) /z'djes/ dec, (s'~k, '~-) /s'njek/ cneg, (&RxsI'RFa, -I-)m (-i-)t /rasti'ratj/ pact$pat, (pd'~s, -_'~-)m (pa_'~-)t /pad'njos/ odnec.
8.3.2.9. e pronouns $x $m $m$ are more regularly /ix, im, 'imi/ ('i, i, I
'i, i, I 'i;mI, &imI, &ImI), whereas traditionally they had an exceptional pronunciation (for >$) with /ji/, by analogy with ego /ji'vo/, emy /ji'mu/ (but today this is
lofty, or non-neutral).
A stronghold of traditional pronunciation had >C$ as /0/: (v'tFa;i[]U)t /v'taliju/ b *tal$, ('kF;I)t /'kri/ k *pe, by distinguishing them from (vi'tFa;i[]U)t
/vi'taliju/ B$tal$, ('i;I)t /'kiri/ K$pe However, native speakers spontaneously
make things natural, by eliminating artices, thus today a palatalized pronunciation prevails in both pairs of examples, without doubt because of orthography too.
is is a further demonstration that Russian actually has six vowel phonemes, indicated in spelling (with the only problem of unstressed-syllable reductions and historical changes). Purists (of course!) still consider this spontaneous pronunciation
non-neutral, but they will have to change their minds sooner or later. As a matter
of fact, a well-balanced position gives: /'kiri, 'kri/ (where indicates a kind of intentional pronunciation used because it should be used).
8.3.2.10. Ocial Russian orthography maintains quite a few letters which do
not correspond (any longer) to actual sounds. e most common case is that of
>t, d between other C\ ('msni) /'mjesni/ mectnv=, (u'Cas~Ik) /u'casnik/ yuactn$k, (Ss'sjt) /Ss'sot/ wectcot, (tu'issii, -sqii) /tu'risskii, -sqkii/ typ$ctck$=, (kIs'qFa;) /kris'qa/ kpectza, (&pxdus'qF;) /padus'q/ od ydzv, (g'Fanq) /ga'lanq/ gollandzv, ('q) /'sjerq/ cepdze, (I'CiSkx) /sir'ciSka/ cepdu$wko, ('pjznx) /'pozna/ odno, (p'RFaz~Ik) /p'raznik/ padn$k, (Is'iCIk,
i-) /SSis'lifcik/ cuactl$bu$k, ('sj;vIsIvi, -vi-) /'sovislivi/ cobectl$bv=, (I'gFansii -nqs-) /gi'ganskii/ g$gantck$=, (g'Fansii -nqii) /ga'lanskii/ gollandck$=, (g'Fa~kx, -kx) /ga'lanka/ gollandka, (St'Fa~kx, -kx) /Sat'lanka/ wotlandka.
Nowadays, indeed, in certain words the graphic C can be restored: ('bz[d]nx)
/'bjez[d]na/ edna, (z'vz[d]ni) /z'vjoz[d]ni/ bednv=, (ks[t]'a;vi, -[+]'-) /kas[t]'ljavi/ koctlrbv=, (ps[t]'Fa) /pas[t]'latj/ octlat, (Iz'vs[t]kx) /iz'vjos[t]ka/ $bectka, (ms[t]'i;) /mas[t]'ki/ moctk$, (p'js[t]kx) /pa'jes[t]ka/ oedka, ('qjs[t]ii) /'qos[t]kii/ ectk$=.
Other words show correspondence between writing and pronunciation: (stu'ntkx) /stu'djentka/ ctydentka, (&fIq'antkx) /afiq'antka/ of$z$antka.
308
Some other particular cases: ('sjnq) /'sonq/ colnze, ('Custvx) /'custva/ uybctbo, ('tF;iCx, 'tFx) /'tsica/ tvcrua, (i'Cas, 'as, as) /si'cas/ ce=uac (in the
sense of now, at once), (p'qFa;ustx, -a;stx, -astx, b'q-) /pa'qalusta, -l[]sta/ oaly=cta, (z'dRFastvuiI, -astui-, -asI, -a, z&dRa, za) /z'drastvuiti/ dpabctby=te.
As a useful reminder, and as a drill (thus in the examples we only mark stress,
through accents), we can say that >t = (`) in the graphic sequences >ctd ctd
ctl ctn ctc ctck (the last one is /sk/): wectdect nebctka cuactlbv= wectndzat $bctnv= wectct /Ss'sot/, typctck$=; >d = (`) in
>dk dn\ odka pdn$k dno
Besides, the sequences >tc dc correspond to /q/, in the pronominal endings
-t()cr: (smi'jqqx) /smi'joqqa/ cmeetcr, (smi'jaqqx) /smi'jaqqa/ cmertcr, and
when >t d are followed by the endings -ck$= -ctbo -ctb$e\ ('qii) /'djeqkii/
detck$=, (t'sjuqtvi[]I) /at'suqtviji/ otcytctb$e, (&pR[]Iz'vjqtvx) /prajiz'voqtva/ po$bodctbo, ('qtvi[]I, s'-) /s'ljeqtviji/ cledctb$e
Lastly, in the genitive endings -ogo, -ego, >g corresponds to /v/: ('b;xvx) /'bjelava/ elogo, (jI'vj;) /ji'vo/ ego us the oddity of (I'vj_~x) /si'vodnja/ cegodnr
today (= of this day) is only seeming.
8.3.2.11. Here we will present the dierences between modern and traditional
pronunciation, in reference to poststress-syllable V with // () for /i/ (>e generally described as //, exactly like /a/, of unstressed >a o, greatly falsifying phonetic reality). e transcribed or retranscribed examples are taken from Shapiro
(1968).
Nominative and accusative singular of neuter nouns, -e: ('pj;I)m (-)t /'poli/
ole, ('mj;I)m (-)t /'mori/ mope, ('ve;CI)m (-C)t /'vjeci/ beue, (q'iI)m (-C)t
/q'liSSi/ $l$e, (z'dFa;~i[]I)m (-[])t /z'daniji/ dan$e.
Genitive, dative, and locative singular of nouns and adjectives, -e-: ('i;~Ivx)m
(-~-)t /'siniva/ c$nego, ('i;~Imu)m (-~-)t /'sinimu/ c$nemy, ('i;~I)m (-~)t /f'sinim/ b c$nem.
Nominative and accusative, singular and plural, of neuter adjectives and pronouns, -e: (z'lj[;]I)m (-[])t /z'loji/ loe, (z'F[;]I)m (-[])t /z'lji/ lve, (t'kj[;]I)m
(ta'kj[;])t /ta'koji/ takoe, (t'i[;]I)m (ta'i[;])t /ta'kiji/ tak$e, (i'ba;qjI)m (-j)t
/li'bjaqji/ lere, ('vjCjI)m (-j)t /'volcji/ bolue.
Collective numerals, -e\ (d'vj[;]I)m (-[])t /d'voji/ dboe, (t'Rj[;]I)m (-[])t /t'roji/ tpoe
Comparatives, -ee: (smi'e[;]I)m (mi'e;)t /smi'ljeji/ cmelee, (vIR'~e[;]I)m (viR'~e[;])t /vir'njeji/ bepnee.
Instrumental singular non-feminine and genitive plural, -em, -eb: (u'Ci;iI)m
(-)t /u'citilim/ yu$telem, ('pFa;CI)m (-)t /'placim/ lauem, (q'iI)m (q'i C)t /q'liSSim/ $l$em, (b'RFa;jIf)m (-f)t /b'ratjif/ pateb, ('SFL-jIf)m (-f)t
/'Sljjif/ w$leb.
Instrumental of plural nouns, -am$: ('kFap-imI)m (-mI)t /'kaplimi/ kalrm$,
('tju;CimI)m (-mI)t /'tucimi/ tyuam$, ('RjimI)m (-mI)t /'roSSimi/ poam$.
8. russian
309
Stress
8.3.3.1. As far as the actual stess force is concerned on the various syllables of
words in Russian sentences, we have to make the situation clear since on this aspect too confused and misleading ideas freely circulate. As a matter of fact, the
prominence of a certain syllable does not necessarily coincide with its stress(ing)
As is well-known, prominence (which makes a given syllable stand out from nearby ones) is not constituted by stress alone. Rather it is a complex play of stress
force, length, pitch, and timbres of the various segments of the syllables.
Now, in Russian, the vocoids in a pre-tonic syllable ( the one that immediately precedes the stressed or tonic syllable in a rhythm group, that is the rst pretonic in Russian texts) are articulated in a fairly clear and distinct way, although
slightly less so than in stressed syllables. is fact is most evident for /a/, as a matter of fact we have for instance: (&gx'vFa;) /gala'va/ goloba
Nevertheless, we must not confuse a vowel timbre with the degree of stress,
since for rhythmic reasons a secondary stress falls on the rst syllable of a word,
not on the second one. Each syllable has a peculiar prominence: (&gx) because of
rhythmic stress (in spite of a decidedly more attenuated timbre less peripheral
in the vocogram); () because of its semi-attenuated timbre and a signicantly
dierent pitch from that of the tonic (syllable) which follows it, although not
marked by a particular symbol ( 8.4). Lastly, (vFa;) is prominent because of
primary stress, a full timbre, and a half-lengthening too (in unchecked syllable, in
addition to the insertion of (F)).
All too often, even in specic handbooks, we come across statements which
claim that the pretonic syllable, (), would have the second degree of stress,
whereas all the others are weaker, above all those after the stress. However, in (s'tFa;Rxx) /s'taraja/ ctapar for the last two syllables, we have a timbre attenuation (as
for (&gx)) and a stress weakening (as for ()), which might make them seem one
degree lower in comparison with the other two but it is not so. eir weakening
if any is due to the fact that, generally, words given as examples are considered
to occur in an intoneme (even if unintentionally), with a consequent reduction of
articulatory and intonational intensity on posttonic syllables.
8.3.3.2. In Russian words, stress undergoes variations even determined by
grammar. In fact, the dierent paradigms (of nouns, adjectives, and verbs) are certainly dicult, especially for foreigners, who can laboriously nd assistance in
grammars and dictionaries.
Here are some examples of distinctive stress ( phonemic minimal pairs): ('Rju;I) /'ruki/ hands and (Ru'i;) /ru'ki/ of the hand pyk$, ('Fat-xs) /'atlas/ atlas and
(t'Fas) /at'las/ satin atlac, ('mju;kx) /'muka/ torment and (mu'kFa;) /mu'ka/
our myka, ('pFa;i) /'paritj/ to steam and (p'i) /pa'ritj/ to plane ap$t,
('pFa;u) /'parju/ I steam and (p'u;) /pa'rju/ I plane ap.
e following are sub-minimal pairs (phonemically), but equally functional (linguistically): ('jRgxn) /'organ/ organization and (R'gFan) /ar'gan/ organ opgan,
(s'vjistvx) /s'voistva/ characteristic and (svist'vj;) /svaist'vo/ property
310
cbo=ctbo, ('zFa;mxk) /'zamak/ castle and (z'mjk) /za'mok/ lock amok,

(k'Rj;u) /k'roju/ I cover and (kR'ju;) /kra'ju/ I cut (a dress) kpo, ('pj;tx)
/'potam/ sweat (instrumental case) and (p'tj) /pa'tom/ then otom, ('ju;q) /'uq/ narrower and (u'qF;) /u'qe/ already ye, ('as) /'xaos/ confusion and
('js 'as) /xa'os 'xaos/ chaos xaoc, ('pFanu) /'paxnutj/ to smell and
(p'nju) /pax'nutj/ to blow axnyt.
Let us now consider some cases where the stress variation is between neutral and
dialectal or slangy pronunciation: (d'bF;Cx) /da'bca/ and ('dj;bCx)d /'dobca/ dovua, (&bIk'Ci) /ablik'citj/ and (b'kCi)d /ab'ljekcitj/ olegu$t,
(&mii'mt5, -;t) /mili'mjetr/ and (mi'i;mIt5, -t)d /mi'limitr/ m$ll$metp; or between neutral and uneducated pronunciation: (kR'i;vi[]I) /kra'siviji/ and (&kRxi've[;]I)p /krasi'vjeji/ kpac$bee, (zv'~iS) /zva'niS/ and (z'vj;~IS)p /z'voniS/ bon$w.
Otherwise, the dierence may be of literary stylistic levels, such as a folk one
(used in traditional tales and stories): (&mx'q) /mala'djeq/ and ('mj;xIq)f
/'moladiq/ molodez, (i'vi;qx) /di'viqa/ and ('e;vIqx)f /'djeviqa/ deb$za, (&iIb'Rj;) /sirib'ro/ and ('e;Ib-Rx)f /'sjeribra/ cepepo, ('Sjkxvi) /'Solkavi/ and (S'kj;vi)f /Sl'kovi/ welkobv=, ('sFa;xRni) /'saxarni/ and (s'FaRni)f /sa'xarni/
caxapnv=, (&bxg'tF) /baga'trj/ and (b'gFa;t)f /ba'gatrj/ ogatvp; or a lofty
level: ('nFa;CIti) /'naciti/ and (n'Ca;ti)a /na'cati/ nauatv=, ('izbRxnni) /'izbranni/ and (Izb'RFanni)a /izb'ranni/ $pannv=, (';biii) /'ljubiSSii/ and ('bii)a /lju'bjaSSii/ lr$=, ('vFa;iii) /x'valiSSii/ and (v'ii)a /xva'ljaSSii/ xbalr$=, (mS'e;~i[]I) /mS'ljeniji/ and ('mFSi~i[]I)a /'mSliniji/ mvwlen$e.
Lastly, we also nd cases of double neutral stress(ing): (t'vj;Rxk, tv'Rjk) /t'vorak, tva'rok/ tbopog, (I'nFa;CI, 'i;nxcI) /i'naci, 'inaci/ $naue, ('j;bu, 'bju) /'obux, a'bux/ oyx, (p'Rj;bI, pR'bi) /p'robil, pra'bil/ po$l
By considering paradigm variation too, we can see: ('e;Ivx) /'djeriva/ depebo,
(i'ev-jx) /di'rjevjja/ depebr, (&iI'vFa;) /diri'va/ depeba including less recommendable pronunciations (which appear after ;), or are marked as wrong in
pronunciation dictionaries (here marked with only in phonetic transcriptions):
('jddx, d'dFa) /'oddal, ad'dal/ otdal, (&dd'Fa;, 'jddxx, d'dFa;x) /adda'la/
otdala, ('jddxx, d'dFa;x &dd'j;) /'oddala, ad'dala/ otdalo, ('jddxI, d'dFa;I) /'oddali, ad'dali/ otdal$
8.3.3.3. In compounds, every element (before the last one, which rmly maintains its primary stress, /'/ (')) tends to keep a certain degree of stress, which we will
mark with // () (or (&), if in contact with ('), corresponding to a possible rhythmic
stress). Even the vocalic timbre is rmly maintained, with no neutralizations,
which are typical of unstressed syllables): (&sxmtxstR'je;~i[]I) /samaljotastra'jeniji/ camoletoctpoen$e, (ktjfiik'pFakx) /kartofilika'palka/ kaptofelekoalka, (bjtpRxv_'~ik) /bortpravad'nik/ optpobodn$k, (spjtkRu'qjk)
/sportkru'qok/ coptkpyok, (FRxfjt'~i;mxk) /aerafotas'nimak/ a<pofotocn$mok, (IktRxpFaRxpx&dxgI'vFa;i1) /iljektraparapadagri'vatil/ <lektpoapoodogpebatel, (gjsz'dFat, -sI-) /gos-z'dat/ Goc$dat, (&mjz'bFa~k, &mjs-,
-k) /moz'bank, mos-/ Mocank, (&ik'RFatkx[]I) /ik'ratkaji/ $ kpatkoe, (pjCI-
8. russian
311
'sFa;) /polci'sa/ oluaca, (&pj'gj;dx) /pol'goda/ olgoda (but ('pji8) /'poldinj/ olden has to be noted).
8.3.3.4. Obviously, in actual sentences there are unstressed words too. Mostly
they are functional monosyllables (not lexical ones) such as prepositions, conjunctions, and some particles: (&tst'Fa;) /atsta'la/ ot ctola, (&pxdk'nj) /padak'nom/ od oknom, (km'~;) /kam'nje/ ko mne, (b'RFa-jx i'stR) /b'ratjja i'sjostr/
patr $ cectpv, (~Iz'nFa;u) /niz'naju/ ne na, (sk'qF;kx) /ska'qka/ cka$-ka,
('j2q) /'onq/ on e, (pi'+-I) /pri'djotli/ p$det l$.
Also pronouns and monosyllabic adjectives can be destressed: (mjib'RFat)
/moib'rat/ mo= pat, (='jn, -xn) /z'djesj-on, -an/ dec on, (ku'dFa vbI'qF;I)
/ku'da vbi'qti/ kyda bv e$te, (&vii'j8, &ve-) /visji'junj/ bec $n also <to:
(k'tj Ft=';x, It-) /k'to etaz'djelal, i-/ kto <to cdelal, ('vj;njn, -xn) /'vonon, -an/ bon on, ('vj;tjn, -xn) /'voton, -an/ bot on, (vjt'tjut, v-) /vot'tut,
va-/ bot tyt, (vjn'tFa, v-) /von'tam, va-/ bon tam, (jnz'nFa) /onz'nal/ on
nal, (S&tjjn'tFa, S&tj-) /Stoon'tam, Stao-/ uto on tam, (&njjm'bFta,
-'b-, &nj-) /noom'bltam, -n'b-, nao-/ no on vl tam, (tk'vjt) /tak'vot/ tak
bot (conjunction, whereas the adverb tak does not reduce), ('jn d'ja;) /'on da'ja/ on da r (a conjunction again, whereas da, the adverb yes is generally stressed
('dFa;); nally: (I'mju xbS'tj;) /i'mu xadjbS'to/ emy xot v uto.
Here is a list of the most frequent (functional) monosyllables with reduced
forms (realized with unstressed vowel taxophones, required by the context): v,
vl, bv, bam, bec, da, kak, l$, mne, mv, ne, n$, o, on, ot, o, od, cem, to,
tv. e pronoun r I can be reduced immediately after a stressed V: (m'gu jxpxvI'dFa) /ma'gu japavi'datj/ mogy r ob$dat, (g'e jxm'gu pxu'Ci) /g'dje jama'gu palu'citj/ gde r mogy olyu$t, (kg'dFa jxnCI'nFa) /kag'da janaci'nal/
kogda r nau$nal
However, monosyllables can be more independent about vowel timbre than unstressed syllables in polysyllables. Consider, for instance: ('vj;Rxn) /'voran/ bopon
and ('vjR-jn) /'voron/ bop on, (&iI'sFa;) /tili'sa/ teleca and (&eI'sFa;) /tjeli'sa/
te leca, (jIs'~ei) /jis'njei/ rcne= and (js'~ei) /jas'njei/ r c ne=.
8.3.3.5. On the contrary, certain prepositions and monosyllabic particles can
receive a rhythm-group stress, when the word which follows them has the original
stress on the rst syllable: ('bzviI) /'bjezvisti/ e bect$, ('pjd-RuI) /'podruki/
od pyk$, ('~;bx) /'njebla/ ne vlo, ('pjt-I) /'potri/ o tp$, ('nFa;px) /'napal/
na ol, ('nFa;gxRu) /'nagaru/ na gopy, ('nFaInu) /'nastinu/ na cteny, ('zFa;gxRxdx, -gxd-, -gRxd-) /'zagaradam/ a gopodom, (&Rx=i'i| 'nFadvx, -&i'n-) /razdi'litj
'nadva/ padel$t na dba, ('pjdnxsx) /'podnasam/ od nocom, ('zFa;nxs) /'zanas/ a noc, ('jCCIsu) /'occisu/ ot uacy, ('pjtpxx) /'potpalam/ od olom,
(&japxi'a jIvj'izvidu) /japati'rjal jivo'izvidu/ r oteprl ego $ b$dy, (I'mu 'zFa;sxRxk) /i'mu 'zasarak/ emy a copok, (pm'nj;q 'em| 'nFa;pi) /pam'noqtj 'sjemj|
'napitj/ omno$t cem na rt.
Also the negative particles ne, n$, which are generally unstressed, may have this
structure, above all with the verb vt, ('bF) /'btj/, to be: (jI'vj;tFam
312
'~;bx) /ji'vo-tam 'njebla/ ego tam ne vlo, (&jntFam'~;q) /on-tam'njeql/ on

tam ne $l.
Here are the prepositions that can have this stress pattern (most typical of colloquial speech): e, bo, do, a, $, na, o/o, ot, o, od, p$. However, it is not
sucient that the noun or number that follows is stressed on its rst syllable, as
happens, for instance, in: (&jmpx&qFk'~i;gu ns'tj, &j-) /ompala'ql k'nigu
nas'tol, on-/ on olo$l kn$gy na ctol, (&n';x ns'tju) /ana'sjela nas'tul/ ona
cela na ctyl.
e exact application depends on actual usage, without being possible to provide a precise and absolute rule. A serious pronouncing dictionary, with real transcriptions (at least in ocial IPA] ought to show these forms accurately.
Intonation
8.3.4. 8.4 shows the preintonemes and intonemes of neutral Russian then
we just give some illustrative examples, to use for comparisons:
/./: (F;tx mjibi;mi &kxp=i;tx) /'eta moi-lju'bimi kampa'zitar./ >to
mo= l$mv= komo$top.
/?/: (vF; jI&vjvi;iI2 2) /'v jivo'vidili?/ Bv ego b$del$?, (&FtxIkkj;2 2, -k-)
/etalik'ko?, -x'ko?/ >to legko?
//: (jeI &j~~Itve;It2 2 &janxpI'Sju; jI&vjqqju;3 3) /'jesli on-niat'vjetit ja-napi'Su jivoaq'qu/ Ecl$ on ne otbet$t, r na$wy ego otzy.
Elliptic questions ( those without a verb) have the pattern / /: (aIgzFa;mIn2 2) /aig'zamin/ A <kamenv?
8.4. Russian preintonemes and intonemes.
/ / (2 2 2 2 2 2 2)
/./ (2 3 3)
/ / ( 2 2 2 2 2 2 2)
/?/ (2 2 2)
/ / ( 2 2 2 2 2 2 2)
// (2 2 2)
/ / ( 2 2 2 2 2 2 2)
/,/ (2 ' 2)
Text
8.4.0. e transcribed passage, e North Wind and the Sun, gives the neutral
accent; as always, according to the phonetic method, rst comes the English text
( 2.5.2.0) with a broad Russian accent, then the Russian translation, with neutral pronunciation; then follows the traditional accent (not recommendable any
8. russian
313
longer today).
At the end, as usual, the version showing the English pronunciation of Russian
is given; it represents the habits of neutral British speakers, uent in Russian (after prolonged contact with native speakers, but with no help from the phonetic
method), who have adequately learned the relative prominences, but who substantially use segmental and intonation elements which are typical of neutral Russian
(although, of course, a neutral accent is not so common). Obviously, the same
principle is valid for the foreign pronunciations of English, given rst.
Speakers of American English could prepare their own version both of the Russian pronunciation of English and of their pronunciation of Russian, as an excellent exercise, by listening to native speakers, best of all after recording them. Of
Russian pronunciation (of English)
8.4.1. (InjRt 'vint2 nd'sjn2 &vRispu;Ik vic v=IstRjgxR3 3| vn
tRFavxR2 m 'jk2 RFap InvjRmk jk3 3| g'it2\ dxIvjn
ufRst sk'i;i2 i~me;I It'RFavxR2 e Iskjk jf2 2| &SudbIkni;dxRt
stRjgxR &dxnij;dxR3 3||
'e~2 InjRt 'vint2 bju;2 s'FaRt2 Ikjut3 3| &bxImjR Ibju;2 2| ImjRk 'ji2 &iIt'RFavxR2\ fjt Iskjk RFau~Im3 3| &xntt'Fast2\ InjRt 'vint2 v jp Impt3 3|| 'e~2 Isjn Sjn aut3 3 vjRmI3 3| &~i'mi;i+I2\ ItRFavxR tjuk jf3 3 Iskjk3 3|| nt'sj;2 InjRt 'vint2 vzbFaiC tukM'fs2| &dxIsjn3 3 v=IstRjgxR3 3 &vItju;3 3||
&iuFaik2 2 Is'tj;RI2| juvjn tui;It n2 2|||)
Russian text
8.4.2. Odnadv cebepnv= betep $ colnze cop$l$, kto $ n$x c$lne=.
Kak pa b <to bpemr on$ amet$l$ akytannogo b la ytn$ka, $dyego
o dopoge, $ opew$l$, uto tot $ n$x ydet cu$tatcr camvm c$lnvm,
komy panwe ydactcr actab$t ytn$ka cnrt la.
Tyt cebepnv= betep p$nrlcr dyt $o bcex c$l; no uem c$lne= on dyl,
tem c$lne= kytalcr ytn$k b cbo= la; tak uto b konze konzob cebepnv= betep dolen vl ot cboe= adau$ otkaatcr. Togda ac$rlo colnvwko, ytn$k onemnogy otogpelcr $ bckope cnrl cbo= la. Tak$m opaom,
cebepnv= betep bvnyden vl p$nat, uto colnze c$lne= ego.
Tee onpab$lac ckaka? obtop$t ee?
314
Modern Russian pronunciation

8.4.3. (d'nFaqd2 e;vIRni 've;IR2 I'sjnq2\ spj;iI2 2 ktjIz~I i~ei3 3| kkRFas vF;txv e;mx2 2 ~i z'me;iI2 zkju;txnxvx 'pFa2 pju+~Ikx3 3 IdjuIvx
&pxd'Rj;I2 I&pxISF;I2 2| S'tj;2 tjt Iz~iy bju;It2 tFaqqx sFa;m in2 2|
k&mu'RFa8S2 u'dFasqx2\ zstFa;vi pju+~Ikx2 2| ~ pFa3 3||
tjut 'e;vIRni2 ve;I2 2 &pi~Ia; 'dju2| &Iz i3 3|| &njC i~ei jndju2 2 i'~ei2 kju;txx 'pju+~Ik2 s&vjipFa3 3|| takS'tj2\ knqF knqjf2 2| e;vIRni ve;IR2 2\ 'djqm&b2| &tsvjei zdFa;CI2 &tkzFaqqx3 3|| tgdFa;
zxija;x sjnSkx2 2| pju+~Ik2 &px~Im'nj;gu2 &tgx2 2| Iskj;i ~a
svjipFa3 3|| tim jb-Rxzx2 2\ e;vIRni 've;IR2 vF;nuqImb2 pIzna2 2 Stsjnq3 3 i~ei3 3 Ivj;3 3||
ib pnRFa;vIx2 2 skFaskx2| &pxtii[]2 2|||)
Traditional Russian pronunciation
8.4.4. (ad'nFaqd2 e;vIRni 've;IR2 i'sjnq2\ spj;iI2 2 ktjIz~I i~ei3 3|
kakRFas vF;txv e;mx2 2 a~i za'me;iI2 zakju;txnxvx 'pFaC2 pju+~Ikx3 3 idjuCIvx &pxda'Rj;I2 I&pxiSF;I2 2| S'tj;2 tjt iz~iy bju;It2 CtFaqqx sFa;m in2 2| k&mu'RFaNS2 u'dFasqx2\ zastFa;vi pju+~Ikx2 2| ~ pFaC3 3||
tjut 'e;vIRni2 ve;I2 2 &pi~isFa; 'dju2| &Iza i3 3|| &njC i~ei jndju2 2 i'~ei2 kju;txsx 'pju+~Ik2 s&vjipFaC3 3|| takS'tj2\ kanqF kanqjf2 2| e;vIRni ve;IR2 2\ 'djqb2| &tsvajei zadFa;CI2 &tkazFaqqx3 3|| tagdFa;
zxija;x sjnSkx2 2| pju+~Ik2 &px~im'nj;gu2 &tagsx2 2| iskj;i ~a
svjipFaC3 3|| taim jb-Rxzx2 2\ e;vIRni 've;IR2 vF;nuqIb2 pizna2 2
Stsjnq3 3 i~ei3 3 ivj;3 3||
ib panRFa;vIxs2 2 skFaskx2| &pxtaii[]2 2|||)
English pronunciation of Russian
8.4.5. (D'nA;ZD2 si5vnIi vi'c2 's;nts2\ 'sp;>li32 k5Th; z'nx sI'n;I3 3|
khk5>A;s 'vT >i'mj32 5nI;i zmi'tsli2 z5khuTnv 'phlA;S2 'phuTnk3 3 5DuScv pD'>gi2 p>'S;li32| 'ST;2 5ThT z'nx 'budzT2 S5ThA;ts 'sA;mm 'snim32| &khm'>A;nS2 'DA;sts2\ zes5ThA;vts 'phuTnk32| 5snA;ts 'phlA;S3 3||
5ThuT si'vnIi2 &ph>Iin5sA: 'Duts2| &zfsi5x 'si;3 3|| &n5chm s'n;I n'D;u32 tsi5m sni';I2 5khuTs 'phuTnk2 fsv'phlA;S3 3|| &Thk'ST;2\ fkn'ts;I
kn'tsf32| si5vnIi vi'c32\ 'D;Zmb2| &tsv5j;I z'DA;ci2 &Tk'zA;ts3 3|| Thg5DA:
8. russian
315
zsi'A;l 's;nSk32| 'phuTnk2 &pnm'ngu2 &Tg>i'si32| f5sk;>i sni'A; sn'phlA;S3 3|| Th5khI;im 'b>zm32\ si5vnIi vi'c2 'v;nuZ[d]zm&b2 p>z'nA;ts32
ST's;nts3 3 &sni';I3 3 Ii'v;3 3||
&tsbi5;I pn'>A;vlS21 'skA;sk2| &phft'>Iitsi&21|||)
9. Arabic
9.0.1. In this chapter, we will deal with the neutral modern pronunciation of
Arabic, corresponding to the supraregional language used in ocial radio and
television broadcastings, free from local characteristics (which are typical of every-day life in each of the Arabic countries). In fact, as for the other languages treated in this handbook, this is the most suitable form for general teaching and learning purposes. It constitutes the basis for local varieties too (which are more or less
important from a numerical and socio-cultural point of view), with some simplications, modications and additions.
On the other hand, of course, if substantial contact in particular areas is forseen,
it is advisable to take on some local lexical, semantic, and morphosyntactic characteristics, as well as the phonic ones (which mainly regard some phonemes).
Our transliteration diverges from the more traditional ones used by Arabists
(which are far from being homogeneous, however), especially as far as long vowels and diphthongs are concerned: ii aa uu (not as i a u) ai au (but ay aw only when nal, or followed by y w therefore, they are not always ay aw). Furthermore, we use (for /S , X, /, not digraphs: sh th dh kh gh {not even
for the last two}) and ` (instead of j for /h, Z/), while keeping =
(for /t, d, , /) and q ' as well (for /, , H/).
Vowels
9.1.1. Neutral Arabic has three short and three long vowels, with some remarkable taxophones, due to the inuence of certain consonants and of syllable structure. ere are more variations for /a[:]/ and less for /u[:]/, while /i[:]/ is in an intermediate position. ere are two diphthongs as well, /ai, au/, which are best considered as sequences of /a/ + /i, u/, since their neutral realization is obtained precisely by juxtaposing the two elements, which are subject to the inuence of consonants and syllable structure. e inuence of the local ways of speaking is very
strong, even if unintentional, in teaching recordings as well, especially for /ai, au/,
but also for the basic V
However, the information we will give in this chapter represents the neutral pronunciation, which does not necessarily correspond to everything one may hear
even from good native speakers. Nevertheless, the reader who follows exactly what
is indicated certainly achieves the neutral pronunciation (not a regional one),
even if for V this usage is quite close to that of Levantine Arabic ( 19.2 of
NPT/HPh], in particular for /ai, au/, seen that elsewhere they are generally realized as monophthongs ((e:, o:) or, at most, as narrow diphthongs, (I, U)); consider also Gulf Arabic and Egyptian Arabic ( 19.1 18.25 in NPT/HPh] while
9. arabic
317
Maghreb Arabic is dierent still ( Moroccan Arabic, 18.2 in NPT/HPh] ese

four dierent variants of Arabic are not simply accents of the same language, but
four partially dierent dialects, which normally aect the language itself.
However, here we will consider only supranational (or international) Arabic
pronunciation (although a few major dierences will be dealt with as well).
9.1.2. Always bearing in mind that the unmarked value of /i{:}, a{:}, u{:}/ is (i{:},
{:}, u{:}) (shown by the black markers in the vocogram of 9.1), we should note
well and constantly recognize the list of the Arabic realizations, which have the
following distributions (besides, /:/ in unstressed syllables = ()):
/i/
/i:/
1.1 (), in contact with /t, d, , , /,

1.2 (I), in contact with /h, H, X, /,
1.3 (I), in checked syllables (with dierent C than in 1.1),
1.4 (i), in unchecked syllables (except if 1.1-2);
1.5 (in /ai/) as /i/, for 1.1-4;
1.6 (:), between /t, d, , , / (in checked or unchecked syllables),
1.7 (I:), in contact with /t, d, , , / (in checked or unchecked syllables),
1.8 (i:), in all other cases (in checked or unchecked syllables);
/a{:}/ 2.1 ([:]), between /t, d, , , /,

2.2 (A[:]), in contact with /t, d, , , / (and ()),
2.3 (a[:]), in contact with /h, H, X, , r/,
2.4 (a[:]), utterance-nal unstressed (except 2.2),
2.5 ([:]), in contact with other C (also /, h/),
2.6 ([:]), utterance-internal word-nal unstressed even in monosyllables
(except 2.1-3),
2.7 () /a[h]/, for -a-h (taa' marbuu=a) in pausal position (with no inuence
as in 2.2-4),
2.8 (in /ai, au/) as /a/, for 2.2-3 2.5;
/u/
3.1 (U), between (or in contact with) /t, d, , , , h, H/,

3.2 (U), in checked syllables,
3.3 (u), in unchecked syllables (except 3.1),
3.4 (in /au/) as /u/, for 3.1-3;
/u:/ 3.5 (U:), in checked syllables with /t, d, , , , h, H/,
3.6 (u:), in all other cases.
9.1. Arabic vowel elements.

/i/ (i, I, )
/i:/ (i:, I:, :)
/a/ (, a, A, )
/u/ (u, U)
/u:/ (u:, U:)
/a:/ (:, a:, A:, :)
318
9.1.3. According to the distributions just seen, 9.1 shows the realizations of
the Arabic short and long vowels, (i[:], I[:], [:]) /i[:]/, ([:], a[:], A[:], [:]) {and ([:]),
Koranic variant, 9.2) /a[:]/, (u[:], U[:]) /u[:]/. Here are some examples: ('f) /'if/
qif ('A:n) /i:'a:n/ qiiqaan ('I:ni) /'i:ni:/ iinii (A'dI:) /a'di:/ adiiq ('bInt)
/'bint/ bint ('fi:l) /'fi:l/ il ('IH -) /a'iH/ aqi ('Af:) /'aff/ a ('tA:h) /'ta:h/
=aaa ('baHda) /'baHda/ bada ('5a:HIn) /'ra:hin/ raahin ('H:a) /'ha:a:/ haaaa
('wld) /'walad/ walad ('b:b) /'ba:b/ baab ('U:) /'u:/ uuq ('Huna) /'huna:/
hunaa ('u:f) /'u:f/ uuf (U'u:m) /Xu'u:m/ uuum ('fUndU) /'fundu/ funduq
9.1 does not show the dierent realizations of /ai, au/, which result from the
combination (-, a-, A-) + (-i, -I, -) or + (-u, -U), according to context: ('bIt) /'bait/
bait ('ina) /'aina/ aina ('Haini) /'Haini:/ ainii ('AIl) /'ail/ qail ('fUz) /'fauz/
fauz ('lUn) /'laun/ laun ('HaUd) /'Hauda[h]/ auda-h (mU'Auwm, -Awwm)
/mu'awwam/ muqawwam (u'l:d) /au'la:d/ aulaad
9.2. Some Arabic vowel variants.
/ai/ (e:, I) frequent

colloquial variants
/au/ (o:, U) frequent

colloquial variants
/a[:]/ ([:]) (between (t, d, , , ), in

the Koranic kind of pronunciation)
9.1.4. 9.2 shows some frequent realizations of /ai, au/: (e:, I o:, U), which
are very widespread outside neutral and Levantine Arabic. However, one's pronunciation may still be considered neutral, although colloquial, even if it uses such
variants, provided all other articulations are appropriate; in fact, this pronunciation is quite common indeed: ('be:t, 'bIt) /'bait/ bait ('e:na, 'I-) /'aina/ aina
('He:ni, 'HI-) /'Haini:/ ainii ('e:l, 'Il) /'ail/ qail ('fo:z, 'fUz) /'fauz/ fauz ('lo:n,
'lUn) /'laun/ laun
9.3. Colloquial variants and neutralizations.
/i/ (), /a/ (), /U/ (P)

frequent colloquial variants
{/i, a, u/ () // possible
colloquial neutralization)
9.3 shows further vocalic articulations, all of which are in the intermediate
realizational area (typically unused in neutral pronunciation, as can be seen in
9.1), as happens for the variants of /ai, au/ as well. e white markers indicate unstressed realizations of /i, a, u/, (, , P), which are considerably centralized. e
9. arabic
319
broken-line white marker, in turn, indicates the realization of a frequent neutralization of unstressed /i, a, u/ (unied into ()), which is more typical of quick and
familiar speech (here given as a third realization): (si'j:Z, s-, s-) /si'ja:Z/ siyaa`
(tIm':l, tm-, tm-) /tim'a:l/ timaal (s'l:ma, s-, s-) /sa'la:ma/ salaama (sH'w:n, sH-, sH-) /sah'wa:n/ sahwaan (su'Hu:la, sP-, s-) /su'hu:la/ suhuula (mUS'tA:,
mPS-, mS-) /muS'ta/ mutaqq
Another feature frequent in some colloquial pronunciations tends to avoid
realizations like ([:]) in whole (even long) words, if they have /t, d, , , , h, H, X,
, r/: ('mH-Rab, 'maH-) /'mahrab/ mahrab ('tAlb, -lab) /'talab/ =alab ('tmdI,
'tam-) /'tamdi:/ tamii (mn'u:b, man-) /man'u:b/ manuub
nally, 9.2 also shows the rounded realization of /a[:]/ between /t, d, , , /,
which is more typical of Koranic pronunciation, as already said. Scattered examples of it can be found here and there, and in 9.1.3.
9.1.5. Actually, not every single realization given in 9.2-3 is necessary for a
good neutral pronunciation of Arabic. Nevertheless, if they are rationed and used
in a natural way (speaking uently), they help to give greater spontaneity, similar to natives' speech again within a kind of colloquial neutral pronunciation. is
will depend on words themselves for instance, ([:]) is more likely to occur in
lofty words.
Furthermore, still within the neutral accent (but slightly more international),
there is another possibility taking a dierent approach, towards a reduction in
the number of vowel taxophones, as regards the realizations of /a[:]/. In fact, they
may reduce to two articulations, in stressed or unstressed syllables: ([:]), for 2.37, and (A[:]), for 2.1-3; thus with a slight overlapping for 2.3, variably resolved by
single speakers.
e vocalic realizations belonging to this kind of international pronunciation
are shown in 9.4. is clearer pronunciation generally uses: (i[:], [:] [:], A[:]
U[:]) (it may also have (:, :) for /ai, au/), but not necessarily (, , P) (and least of
all ()), seen in 9.1.4.
e systematic and complete shortening of unstressed long vowels belongs to
modern and international pronunciation (and to dialects as well, generally).
However, in Koranic pronunciation, /:/ (appearing in phonemic transcriptions) are realized as half-long, (;). It is not dicult to achieve them, if we consider both types of transcription used in this chapter: ('A:n, ;-) /i:'a:n/ qiiqaan
('I:ni, -ni;) /'i:ni:/ iinii ('H:a, -a;) /'ha:a:/ haaaa ('Huna, -na;) /'huna:/ hunaa
9.4. International vowels.
/i/ (i, ), /i:/ (i:, :)
{/ai/ (:)}
/u/ (U), /u:/ (U:)

{/au/ (:)}
/a/ (, A), /a:/ (:, A:)
320
Geographic variants
9.1.6. Among the main variations and deviations from the neutral form (
9.1.3), in certain areas, we nd that /a:/ never has the (:) timbre, in any context
(as often happens in Iraq and northern Lebanon). Furthermore, in an almost general way, in several colloquial variants, the diphthongs /ai, au/ reduce to (e:, I o:,
U) ( 9.2), from Maghreb to the Persian Gulf, except in the Levant. However,
the diphthongs are kept, generally, when they are in absolute nal position or followed by /j, w/: ('bIt, 'bIt, 'be:t) /'bait/ bait ('lUn, 'lUn, 'lo:n) /'laun/ laun ('HaIn,
'HIn, 'He:n) /'Hain/ ain ('sijId, 'sjjId) /'sajjid/ sayyid (mu'Suw, -ww-) /mu'Sawwi/ muawwiq ('ni, 'nj:, -J:) /'nai/ nayy ('lu, 'lw, 'l)) /'lau/ law
Especially in Maghreb, besides (unstressed) /i, u/, even /a/ may be dropped (with
possible consequent stress shifts): ('nta, 't, n't) /'anta/ anta ('Alm, 'Alm)
/'alam/ qalam (li's:n, 's:n) /li'sa:n/ lisaan (s'l:m, s'l:m) /sa'la:m/ salaam
Consonants
9.2.0. 9.5 gives the articulations of modern neutral Arabic. e main geographical variants of some phonemes are given in round brackets.
1.9-15, instead, show gathered in groups according to manners of articulation the orograms of all contoids treated in the chapters of this book, even as secondary, occasional, or regional, variants for the 12 languages dealt with.
F pb
(M)
f
n
()
t d td
s z
(~)
{(G)} {()}
{(D)}
()
k {(g)}
laryngeal
pharyngeal
uvular
velar rounded
velar
palatal
postalveo-palalatal
protruded
postalveopalatal
uvularized alveolar
alveolar
uvularized dental
dental
bilabial
labiodental
9.5. Table of Arabic consonants.
(,)
S Z
5|(R)
H (h) H
l {} ()
(l)
// (, {D}), /Z/ (Z, {G}), /h/ (h, H), /r/ (5, R), /X, / (, )
Nasals
9.2.1. ere are two nasal phonemes, /m, n/ (m, M, n, ~, , ,): (mu'ml)
/mu'maal/ mumaal (tm'Zi:d, -'Gi:d) /tam'Zi:d/ tamìid ('Sms) /'Sams/ ams
9. arabic
321
('ma5-jm) /'marjam/ Maryam (mU'hammd) /mu'hammad/ Muammad ('nImnIm) /'nimnim/ nimnim ('Han-w) /'Hanwa[h]/ anwa-h ('Zmb, 'G-) /'Zanb/ ànb
(mIm'b:b) /min'ba:b/ min baab ('Mf) /'anf/ anf ('wI~S) /'winS/ win ('k)
/'ank/ ank ('A,) /'Xan/ anq (m,'u:wa) /min'u:wa/ min quuwa However, the Koranic pronunciation tends to avoid assimilating /n/ to the following
C\ ('Gnb, mIn'b:b, 'nf, 'wInS, 'nk, 'an, mIn'u:wa).
Stops
9.2.2.1. Arabic has no /p/, but only /b/, which however is realized as (p) when
followed by voiceless C\ ('b:b) /'ba:b/ baab ('lbn) /'laban/ laban ('haps) /'habs/
abs On the other hand, neutral Arabic has no /g/, either; while it has two voiceless stops, velar and uvular, /k, / (k, ) (in phonemic opposition): ('k[h]Uskusu,
kUs'k[h]usu) /'kuskusu/ kuskusu (mIk'a:5) /mik'a:r/ mikaar ('mlIk) /'malik/
malik (A'di:m) /a'di:m/ qadiim (A'd:m) /a'da:m/ aqdaam ('sU:) /'su:/ suuq (&AUR':n) /alur'a:n/ al-Qur'aan (sA':tA -':-) /sa'a:ta[h]/ saqqaa=a-h
() enjoys great prestige, even among speakers who do not use it, although very
frequently it is substituted by other articulations (as will be seen in 9.2.7). Note:
('klb) /'kalb/ kalb dog and ('Alb) /'alb/ qalb heart.
Furthermore, we have the diphonic pairs /t, d/ (t, d) and /t, d/ (t, d) (uvularized), which is quite peculiar. ey are dental; and /t/ may be a little aspirated,
when at the beginning of a stressed syllable (but we will mark it only here). Often,
/t, d/ are denti-alveolar if nal before a pause (but it is not necessary to use ((4, 7))):
(t'd:wUl) /ta'da:wul/ tadaawul (Si't[h]:) /Si'ta:/ itaa' ('had:) /'hadd/ add (bA't:ts -'t:-) /ba'ta:tis/ ba=aa=is ('dA5t, 'd5t 'd5t) /'dart/ ar= ('dAZIR, -GIR) /'daZir/ aìr
Although neutral Arabic has no (g), this sound is frequent in dierent dialects,
as a variant of other phonemes, mainly of /Z/ and //. erefore, it is natural that
the isolated phonemes may currently be brought to normalization, by changing
their articulations, in order to form a more homogeneous and coherent system. Even the shift of /Z/ to (Z) (instead of the more Koranic and ancient (G)), or to
(g), is a part of this trend.
9.2.2.2. e last neutral Arabic stop is // (), which may occur in every position, as the other C do, and may be geminated as well. In word-initial position, it
automatically occurs when no other C is present: ('mi:n) /a'mi:n/ amiin ('s:Ih) /'sa:ih/ saa'i ('m:) /'ma:/ maa' ('znn) /'zanna/ zanna' (&Ru's:) /rua'sa:/ ru'asaa' (l':n) /al'a:n/ al-aan ('5as) /'ras/ ra's ('bd) /'bad/ bad'
(s':l) /sa'a:l/ sa''aal
erefore, in Arabic, any word traditionally beginning with a V ( vowel), phonically, begins with // (), because in this language all phono-syllables begin with
a C\ ('b) /'ab/ ab (i'd:n) /i'da:na[h]/ idaana-h ('Um:) /'umm/ umm On the
other hand, Arabic does not tolerate syllables beginning with more than one C (except rare cases of loanwords not adapted to Arabic phonology). Consequently, in
actual pronunciation, a short V is prexed; generally, it is /i/ (but, sometimes, /u,
322
a/, in accordance with the following V] and /a/ in neutral pronunciation, for the
article while, in the dialects, we more often nd (, I)); of course, in isolated
forms, // is prexed as well.
However, when words with CC- ( with an initial consonant cluster) occur in
connected speech after a word ending in a V it is not necessary to add the V (nor
//), therefore the two words are linked. If, instead, the preceding word ends in a
C then the V is added, but // is not. e reader is referred to grammars, where this
phenomenon (called wala) is generally treated quite widely. It is to be found in
connection with the article, certain verbal forms, the imperative and a dozen
nouns. Among these, the most important are: ('Ibn, -b, -bn) /'ibn/ ibn ('ImRU)
/'imru/ imru' ('Ism, -s, -sm) /'ism/ ism (I'n:ni) /i'na:ni/ inaani\ (Ra'itUb 'ni:) /ra'aitu b'ni:/ ra'aitu bnii ('b:bUl 'bIt) /'ba:bu l'bait[i]/ baabu al-baiti
Constrictives
9.2.3. Among the phonemes belonging to this articulation manner, we nd /f/
(f) (but not the corresponding voiced phoneme): (fa'5i:d) /fa'ri:d/ fariid (If's:d)
/if'sa:d/ ifsaad In addition, there are two diphonic pairs, which pose no problems,
/, s, z/ (, s, z): ('l:) /a'la:/ alaa ('mIR) /'mair/ mair ('m:)
/'amm/ amm ('da5s) /'dars/ dars ('za:5) /'za:r/ zaar ('knz) /'kanz/ kanz
However, there are two more diphonic pairs, with variations that may pose
some phonemic dilemmas; they are /, / ( /D) and /S, Z/ (S Z/G). We prefer (,
Z) for their voiced members, as they are more modern and more integrated in the
phonologic system than their more Koranic variants (D, G), which are considered more prestigious (even by those who do not use them), but have dierent
place or manner of articulation. erefore, they complicate the phonemic system.
However, they can be used especially in a kind of pronunciation which aims at
the traditional more than at the international accent: (mah'U:, -DU:D) /mah'u:/
mauu ('A:mI, 'D-) /'a:mi/ aami' ('URUR, UR'u:R) /'urur, ur'u:r/ uru(u)r (Ra'I:) /ra'Xi:/ raii (IS'h:d) /iS'ha:d/ ihaad (S'j:) /aS'ja:/ ayaa'
(mu'SuwS, -wwS) /mu'SawwaS/ muawwa ('Sd:) /a'Sadd/ aadd (RaS'S:S)
/raS'Sa:S/ raaa (S'SaRu) /aS'Sarw/ aarw (Z'mi:l, G-) /Za'mi:l/ àmiil ('ZmaH,
'G-) /'aZmaH/ a`ma ('t:Z, -G) /'ta:Z/ taa`
In the pharyngeal place of articulation, we nd the voiceless constrictive /h/ (h)
(currently, the corresponding voiced sound is considered to be constrictive as well,
but in neutral pronunciation, it is clearly an approximant, /H/ (H), as we will see below, 9.2.4): ('hUbbi) /'hubbi:/ ubbii (mah'tu:m) /mah'tu:m/ matuum (mU'hAddR) /mu'haddir/ muair ('faRIh) /'farih/ fari (fah'ha:S) /fah'ha:S/ faaa
Arabic also has a diphonic pair of uvular constrictive trills, (, ) (as will be
seen), which phonemically may be represented with the ocial symbols /X, /.
Approximants
9.2.4.1. Let us rst consider the least peculiar ones (although there are rather
free occurrences), /j, w/ (j, w), even realized as (i, u), for /0j, 0w/ (and, possibly,
9. arabic
323
for /j0, w0/, in colloquial variants, as no doubt in the dialects) and (ij/jj,
uw/ww), for /jj, ww/): ('j:wIR) /'ja:wir/ yaawir (wA'I:j) /wa'i:ja[h]/ waiiya-h (wU'u:l) /wu'u:l/ wuuul ('sijId, 'sjjId) /'sajjid/ sayyid (nu'w:m, nw'w-)
/naw'wa:m/ nawwaam ('ni) /'najj/ nayy (mn'Hi:) /man'hi:/ manhiyy ('bu)
/'abw/ abw ( ('bu) /'abu/ abu).
e Koranic pronunciation prefers (-jj-, -ww-), as in the cases seen above. Let
us also consider these further examples, which show us the dierences between
modern pronunciation and traditional Koranic pronunciation. In principle,
they coincide even with the most important cases where, even in colloquial accents
and dialects, /ai, au/ do not change into monophthongs ((:, :)).
is happens when they are in absolute nal position, and when in front of /j,
w/ or after /i:, u:/, respectively, or when nal, after C as the following examples
will show: ('ki, 'kj, -J) /'kai/ kay ( ('kij, 'kj:, -J:) /'kaij/ kayy), ('lijn, 'ljjn)
/'lajjan/ layyan (ju'mi:j, -J) /jau'mi:ja[h]/ yaumiiya-h ('Humi, 'Humj, -mJ) /'Humj/
umy ('lu, 'lw, 'l)) /'lau/ law ('duwaR, 'dww-) /'dawwar/ dawwar (Ha'du:w,
-u:)) /Ha'du:wa[h]/ aduuwa-h ('sHu, 'sHw, -H)) /'sahw/ sahw ('Hafu, 'Hafw, -f,
-f) /'Hafw/ afw
9.2.4.2. As we have said above ( 9.2.3), the Arabic phoneme /H/ (H) is an approximant (and generally it is laryngealized as well, (()), which is fairly easy to detect by its lower intrinsic tonality and creaky voice): ('HaIn) /'Hain/ ain ('maHa)
/'maHa/ maa ('baHda) /'baHda/ bada ('naHnaH, naH'na:H) /'naHnaH, naH'na:H/ nana(a)
(&faHHa'li:j) /faHHa:'li:ja[h]/ faaaliiya-h ('slH) /'salH/ sal It is occasionally possible
to hear some peculiar regional non-neutral variants, as that with a slight simultaneous laryngeal stop, (?), or else a pharyngealized laryngeal stop, (), :
('na?na?, na?'na:?, 'nAnA, nA'nA:).
e last Arabic approximant is a true laryngeal phone, /h/ (H, h), and has a very
free occurrence; it is typically (lenis) voiced (H), but it becomes (lenis) voiceless,
when near a pause, or a voiceless C or when geminate: (&Itti'Z:h, -G-) /itti'Za:h/
ittiàah ('mUhtaR) /'muhtar/ muhtar ('h:iHi) /'ha:ihi:/ haaihii (mH'bu:l)
/mah'bu:l/ mahbuul (&InHi'z:m) /inhi'za:m/ inhizaam (hf'n:f) /haf'na:f/ hafnaaf ('hija) /'hija/ hiya ('ka5h) /'karh/ karh ('bIh) /'bih/ bih ('AH-w) /'ahwa[h]/
qahwa-h (wh'h:Z, -G) /wah'ha:Z/ wahhaa`
In our transliteration, taa' marbuu=a is indicated by a hyphen, -a-h (while another only seemingly more recommendable and less complicated solution could be
-a); it is pronounced (, h) /a[h]/ (but (a, A), if preceded by C which are capable
of modifying timbres). It is to be noted that a realization with (h) represents a very
careful, Koranic pronunciation, while the normal realization of -a is (a): ('hzz) /'hazza[h]/ hazza-h (movement) but: ('hzza) /'hazza/ hazza ((he) shook).
Trills
9.2.5. Typically the voiced alveolar /r/ is realized as a velarized trill, (5), generally, in stressed syllable, and as a tap, (R), in unstressed syllables. ey inuence the
324
timbre of /a/ (a): ('5Ip) /'rib/ ribq (maR'bu:H) /mar'bu:H/ marbuu ('maRIh) /'marih/ mari (mIR'5i:h) /mir'ri:h/ mirrii (mi'5a:h) /mi'ra:h/ miraa ('fU5fUR, fUR'fu:5)
/'furfur, fur'fu:r/ furfu(u)r It is possible to hear (), mainly for nal /r/, but this pronunciation is not recommendable.
As already said above ( 9.2.4), Arabic has a diphonic pair of uvular constrictive
trills, /X, / (, ): (ba'Si:S) /baX'Si:S/ baii ('aU) /'XauX/ au (fa'a:Ri) /faX'Xa:ri:/ faaarii ('adn) /'adan/ adan (A'i:R) /a'i:r/ aiir ('b:lI) /'ba:li/
baali (mS'u:l) /maS'u:l/ mauul (t'waUl) /ta'waul/ tawaul
Laterals
9.2.6. ere is one lateral phoneme, /l/ (l, , ) (the third taxophone occurs in
contact with /t, d, , , /; it occurs as a phonostyleme ( a stylistic phoneme),
too, in Allaah and derivatives (provided that it is not preceded by /i/): (&mut'llI) /muta'lali/ mutala'li' (tl'bi:s) /tal'bi:s/ talbiis ('lIl) /'lail/ lail (m'j:n)
/mal'ja:n/ malyaan (zl'lA:A) /zal'la:a[h]/ zallaaqa-h (tA'I:h) /tal'i:h/ talqii
(&bIsmIl'l:h) /bismil'la:h/ bismillaah (A'A:h) /al'la:h/ Allaah For the typical
complete assimilation of /l/ in the article al see below ( 9.3.1.1).
Geographic variants
9.2.7. Very often, /, / are pronounced like /t, d/, mainly in big cities in North
Africa, or like /s, z/ as well, particularly in the Middle East. In each one of these
cases a phonemic distinction is lost. In Iraq, // is (D), as in traditional and Koranic pronunciations; elsewhere it is often realized as (), as in Egypt and Syria,
but it may even be confused with /t, d/, above all in Maghreb.
e grapheme ìim /Z/ (Z), has very many geographical, social, and religious,
variants. e normal (Z) prevails in the Middle East and in North Africa, while
(G) (typical of Koranic reading) is used in Jordan, Saudi Arabia (typical of Bedouins) and Iraq. But in some areas, as in Egypt (Cairo), Sudan and Oman, we nd
(g); elsewhere, even (), as in Luxor (southern Egypt), and (J).
For instance, /'Za:b/ àab is ('Z:b) in the Levant (Lebanon, Palestine, and Syria)
and in Africa (except Egypt and Sudan, as we have just said); it is ('G:b) in Arabia,
the Persian Gulf, Jordan, Iraq and among the rural and the nomadic peoples in
Morocco.
e grapheme qaaf // (), very often becomes (), above all in Cairo and other big cities; but it becomes (H, k) as well, or even (g), particularly in central-southern Egypt, in rural areas of Morocco, and among Saudi Arabia Bedouins (who often, typically, change /k/ into (c)). For instance, /'alb/ qalb may be: ('Alb 'lb,
'Halb, 'klb, 'glb).
In the Levant, // is generally quite weak, therefore we could transcribe it as (,)
(instead of ()). In Maghreb, // may not even be present at all.
9. arabic
325
Structures
9.3.0. Here we will deal with assimilation, quantity, and some typical reductions of colloquial speech, still within neutral pronunciation (although with geographical variants).
Taxophonics
9.3.1.1. e denite article, (l) /al/ al obligatorily assimilates completely to
all apical or laminal C following it. erefore, in such cases we may say that the article is /a0/ with a following homorganic /0/ something like /a*/, as in the
neutral-Italian pronunciation phenomenon called co-gemination. th /Z/, assimilation takes place if it is pronounced (Z), but there is no change if it is pronounced
(G), or (g), : (S'Sms) /al'Sams/ al-ams (aR'5A, -5) /al'ra/ al-raq (t'tutUn) /al'tutun/ al-tutun (z'zll) /az'zalal/ al-zalal (Ad'd:) /ad'di:/ al-iiq (Z'Zml, 'G- l'g-) /al'Zamal/ al-àmal (l'kUSk) /al'kuSk/ al-kuk
(l'bahR, -ah{) /al'bahr/ al-bar (l'wld) /al'walad/ al-walad
Even voicing assimilations are quite common indeed, therefore a voiced nal C
may become voiceless, or devoiced, mainly after voiceless C while voiceless C may
become fully (or partially) voiced, between voiced phones. Besides, many cases of
assimilation (regarding place or manner of articulation) are possible, even accompanied by the dropping of some phones: (As'smIHa) /ad'samiHa/ qad samia
('lm 'juRIS 'Sin) /'lam 'jurid 'Saian / lam yurid ai'an ('IbHa ':lika) /'IbHa
'a:lika/ iba aalika ('HUttu) /'Hutu/ utu ('IhfaZ 'Za:Raka, -G 'G-) /'ihfa 'Za:raka/ ifa àaraka
In Arabic both vowel and consonant quantities are distinctive. Consonant quantity is shown by gemination ((00)) between V but by lengthening ((0:)) elsewhere: ('b) /'ab/ ab (':b) /'a:b/ aab (Z'm:l, G-) /Za'ma:l/ àmaal (Zm'm:l,
'G-) /Zam'ma:l/ àmmaal ('Um:) /'umm/ umm ('Atla) /'atala/ qatala ('A:tla)
/'a:tala/ qaatala ('Attla) /'attala/ qattala
Unstressed /:/ are (;) only in the Koranic kind of pronunciation, since they
are generally shortened, () ( 9.1.5).
9.3.1.2. In colloquial speech, unstressed /i, u/ tend to drop whenever this does
not form three-C clusters. On the other hand, when in connected speech two
word-nal C are followed by a word-initial C then a short V generally /i/, is inserted to divide the cluster. In certain cases, the added V may even bear stress; and
sometimes a V inserted in a certain place may cause or allow the dropping of
another short V
Furthermore, long V tend to shorten a little in checked syllables (but less than
in unstressed syllables); and short nal V in polysyllables tend to drop (above all
grammemes, and /n/ as well, except for accusative /an/, which may become
/a:/, in any position, even non-pausal). Verbs and pronouns are more likely to
maintain the V since endings have distinctive and pragmatic values.
326
Let us see some examples: ('h: ki't:b, 'h:k 't:b) /'ha:a: ki'ta:b[un]/ haaaa kitaabun (l'hIb-RU 'tAijIb, 'tAjjIb) /al'hibru 'tajjib[un]/ al-ibru =ayyibun (k'tptu&biHi, -tUb&Hi) /ka'tabtu 'bihi/ katabtu bihi ('k:n fIl'bIt) /'ka:na fi:albait[i]/
kaana i al-baiti (Ra'itUl 'bIt) /ra'aitu al'bait[a]/ ra'aitu al-baita (k'tpt) /ka'tabt[u]/ katabtu (&sk'ki:n) /saka:'ki:n/ sakaakiin
9.3.1.3. As far as pausal forms are concerned, let us take from Mitchell (1990)
some examples (which we will transcribe again appropriately), keeping in mind
that this is an orthological phenomenon people use to pronounce words in intonemes, including preceding words, semantically and syntactically linked (to the
exclusion of grammemes).
is category includes nal short V (with or without innitive -n) ('ktb)
/'katab[a]/ kataba ('jktUb) /'jaktub[u]/ yaktubu ('li~ 'jktUb) /'lian 'jaktub[a]/
li'an yaktuba (&fim'da:RIs) /fi:ma'da:ris[a]/ i madaarisa (fIl'bIt) /fi:l'bait[i]/ i al-baiti ('bIt) /'bait[un]/ baitun (fi'bIt) /fi:'bait[in]/ i baitin ('mUnu 'zmnI
w'Zi:z, -'Gi:z) /'munu 'zamanin wa'Zi:z[in]/ munu zamanin waìizin However,
(accusative) -an /an/ may be substituted by /a:/: (aH'tA:HU HIS'5i:n di'na:Ra) /aH'ta:hu HiS'ri:na di:'na:ra:/ a=aahu iriina dinaaran except for adverbs like: ('ha:ln)
/'ha:lan/ aalan ('dAn) /'aidan/ aian
Lastly, taa' marbuu=a [ -atin -atan -atun] is substituted by /a[h]/, which is
generally pronounced () (or (h) in a {more} Koranic style): (Z'mi:l)
/Za'mi:la[h]/ àmiila-h ((Z'mi:ltUn, &Zmi'ltUn, %G-) /Za'mi:latun/ àmiilatun]
('ftahat m'da:Ris li'li:j, &mda'5is) /'ftahat ma'da:risa lai'li:ja[h]/ fataat
madaarisa lailiiya-h ((li'li:jtn, &lili'jtn) /lai'li:jatan/ lailiiyatan] (&fIlZu'nin,
-Gu-) /fi:lZu'naina[h]/ i al-ùnaina-h ((Zu'nin&ti, &Zuni'nti, %G-) /Zu'nainati/
ùnainati]
9.3.1.4. Colloquially, very often // is not maintained when it occurs within
words or at the end of words it is dropped or changed into /j, w/, or else it
lengthens a possible preceding V\ ('mi, 'mi, 'mij, 'mi:) /'mia[h]/ mi'a-h ('juu, 'ja:uu, ja'uu) /'jaXuu, ja'Xuu/ ya'uu ('5as, '5a:s) /'ras/ ra's (s'm:, s'm:) /sa'ma:/ samaa' en two hamzas occur in contiguous syllables,
the rst one is maintained: ('Z: &[]Z'luHUm, '[]Z&luHUm, []'ZlHUm, 'G:,
-G-) /'Za:a aZa'luhum, 'aZaluhum/ àa'a aàluhum
In /0j, 0w, 0r, 0l/ sequences, Arabic syllabication is heterosyllabic, /0j, 0w,
0r, 0l/: (mIt'5a:s) /mit'ra:s/ mitraas ('tlfa, t'lfa) /'atlafa/ atlafa ('mdRas,
md'Ra-) /'madrasa[h]/ madrasa-h ('d-la) /'adla/ adla ('ma5-jm) /'marjam/ Maryam ('mUsRIH) /'musriH/ musri (mIs'w:k) /mis'wa:k/ miswaak ('mS-wi) /'maSwi/
mawi ('fRaza, f'Raza) /'afraza/ afraza ('mk-wa) /'makwa/ makwa ('-jn)
/'aljan/ alyan ('H-wa) /'ahwa/ ahwa
9.3.1.5. In nal position, after C the sonants (/m, n, r, l/) may be realized in
dierent ways, according to how accurately one speaks. From a phonemic point
of view, they are just C but phonetically they may be plain (or devoiced, mainly in front of a voiceless C] or intense (syllabic), or even with a short epenthet-
9. arabic
327
ic V (like (I, )): ('s, -s, -sm) /'ism/ qism ('ldn, -d, -dn) /'ladn/ ladn ('ftn,
-t, -tn) /'fatn/ fatn ('dUHn, -hn, -H, -Hn) /'duhn/ duhn ('bdR, -d{, -dR) /'badr/
badr ('mR, -, -{) /'mir/ Mir ('ft, -t, -tl) /'fatl/ fatl ('5At, -t, -tl) /'ratl/ ra=l
9.3.1.6. Here are some examples of typical Arabic phonotactics: (ta'bi:n) /ta'bi:n/ ta'biin (m'u:R) /ma'u:r/ ma'uur ('mtHUb) /'matHu:b/ matuub ('Al) /'aal/ aqal ('mdal) /'madXal/ madal (m'Hu:R) /ma'Hu:r/ mauur
('tzHu) /'tazhu/ tazhu ('msZId, -GId) /'masZid/ masìd ('mSta) /'maSta/ mata
(mS'u:l) /maS'u:l/ mauul ('mA5aH) /'maraH/ mara ('AdZaHa, Ad'ZaHa) /'adZaHa/ aàa ('AdHafa, Ad'Hafa) /'adHafa/ aafa ('Alma, A'lma, -D-) /'alama/ alama (maH'Zu:n, -Gu:n) /maH'Zu:n/ maùun (maH'5u:f) /maH'ru:f/ maruuf
(aH'm:l) /aH'ma:l/ amaal (ma'Su:S) /ma'Su:S/ mauu ('mAha) /'maha/ maqha (ml':n) /mal'a:n/ mal'aan ('jisu, ji'su) /'jaiasu/ yai'asu ('uHma,
u'Hma) /'auhama/ auhama
Stress
9.3.2.1. For words of two or more syllables, we have the patterns given below.
It should be kept in mind that // stands for light syllables (/0/); /9/ for
heavy ones, consisting of heavy sequences like /00, :0, 0/, which are
located counting from their syllabic nucleous onwards, up to the next (belonging
to another sequence). erefore, in this kind of calculation, we do not consider
them to be true phono-syllables. In addition, // indicates a light or heavy syllable indierently with no direct inuence on stress assignment. e symbol //
indicates alternative stressing, which is substitutive (certainly not simultaneous
as a matter of fact, one is free to choose either stress pattern from sequences showing both /'/ and //):
2 syll. /', '9, '9/,
3 syll. /', '9, '9, '9/,
4 syll. /', '9, '9, '9, 9'/,
5 syll. /', '9, '9, '9, 9', 9'/,
6 syll. /', '9, '9, '9, 9', 9', 9'/.
ey function as a useful point of reference, for analyses and to nd correspondences; in fact, it would not be easy to try to memorize them.
9.3.2.2. e solution to the formulae with variants, to read along columns, provides:
/'9, 9'/
/', '/,
/'9, 9'/,
/9', '9/
/', '/,
/'9, 9'/,
/9', '9/,
/9', 9'/
/', '/,
/'9, 9'/,
/9', '9/,
/9', 9'/.
328
In patterns with two possible stressings, the variants may be of two types: Egyptian (Cairo), as: /'9, '9, '9, '9/; or literary ( dialectal, like
those of southern Egypt), with the following possibilities: /', 9'/, /',
9', 9'/, /', 9', 9'/.
For /9'/, we may nd the patterns /'9/, /9'/ as well (which
constitute loftier stressings, as can be seen in the story in 9.4.2.1-2).
All the stress patterns we give belong to modern neutral pronunciation. erefore, one is free to choose, provided certain structural homogeneousness is maintained.
A dialectal peculiarity found in Lebanon (which should not be followed) puts
nal stress on words ending in /:, / (which, in neutral pronunciation, are not
heavy enough to bear the stress): ('mInHuma, mIn'Huma) (and (&mInHu'ma:))
/'minhuma:/ minhumaa
9.3.2.3. Prexes do not inuence the application of the stress rule, in dierent
actual words; nor does the article even when it assimilates completely. But the rule
is modied by the dropping of some (vocalic or consonantal) phones, because this
changes the syllabic structure, both within words and at their end, before a pause.
As far as secondary stress in polysyllabic words is concerned, it tends to occur
on alternate syllables, but preferring the heaviest ones, whenever possible.
Lastly, here are some examples: (Ra'su:l) /ra'su:l/ rasuul (s'f~Z, -~G) /sa'fanZ/ safan` (mu'5a:sIl) /mu'ra:sil/ muraasil (ta'5assUl) /ta'rassul/ tarassul ('HA:m, HA'-)
/'Ha:ima[h]/ aaima-h (&mut'fuw, -ww) /muta'fawwi/ mutafawwiq (mu'fa:RAA, &mufa'5AA) /mu'fa:raa[h]/ mufaaraqa-h ('md-Ra&s, md'5as) /'madrasa[h]/ madrasa-h ('daRa&ki) /'daraki:/ darakii (&kt'bHu, 'kt&bHu) /kata'bahu/ katabahu (mU'tli&f, 'mUt&lif) /muX'talifa[h]/ mutalifa-h (&Z'luHUm, 'Z&luHUm, 'ZlHUm, -G-) /aZa'luhum, a'Zalhum/ aàluhum
9.3.2.4. We give further useful examples (again completed and retranscribed
from Mitchell, 1990), both with pausal and pre-pausal forms ( 9.3.1.3): ('5amt) /'ramat/ ramat (Ra'mthu) /ra'mathu/ ramathu ('ahad) /'ahad/ aad (&aha'duHUm, 'aha&duHUm) /aha'duhum/ aaduhum ('Sdda) /'Sadda/ adda ('SddHu, Sd'd-) /'Saddahu/ addahu ('a5dA) /'arda:/ ardaa (a5'dA:Hu) /ar'da:hu/
ardaahu (k'tpti) /ka'tabti/ katabti (k'tptiHi, &ktp'tiHi) /ka'tabtihi/ katabtihi (Is'taA) /is'tala:/ istalqaa (&Ista'A:Hu) /istal'a:hu/ istalqaahu ('k:t&ba,
k'tba) /'ka:taba:/ kaatabaa (&kt'b:Hu) /ka:ta'ba:Hu/ kaatabaahu (&kt'bta,
'kt&bta) /kata'bata:/ katabataa (&kt'bt:&Hu) /kataba'ta:hu/ katabataahu (k'tb&ta, 'k:t&bta) /ka:'tabata:/ kaatabataa (&ktb't:Hu, k&t-) /ka:taba'ta:hu/
kaatabataahu (&SZa'5atUn, 'SZa&RatUn) /SaZa'ratun/ aàratun
Here are some more longer examples: (&dwij'tuHu, d'wij&tuHu) /adwija'tuhu/ adwiyatuhu (mUR&tbi'tAtUn, mUR'tbi&tAtUn) /murtabi'tatUn/ murtabi=atun (&SZa'5atu&Hu, S'ZaRa&tuHu) /SaZa'ratuhu/ aàratuhu (&SZa&Ratu'Huma, &SZa'5atu&Huma) /SaZaratu'huma:/ aàratuhumaa (&dwij'tuHu&ma, d'wi-) /adwija'tuhuma:/ adwiyatuhumaa (mu&tZn'nib&tUn, &mut'Znni&btUn, -G-) /mutaZan'nibatun/ mutaànnibatun (mu&tAA'tiltUn, &mutA'Ati&ltUn) /mutaa:'tilatun/
mutaqaatilatun
9. arabic
329
Intonation
9.3.3. 9.6 shows the preintonemes and intonemes of modern neutral Arabic.
e continuative intoneme is not very dierent from the suspensive one; therefore, generally, the pause which follows the suspensive type has greater duration.
e interrogative intoneme in the Levant is of the rising-falling type, /?/ (2 ' 1 2), instead of that given here. It will be interesting to compare the tonograms (but also
the V and C] referring to the four geographic Arabic varieties ( 9.1.1 which
dier in grammar as well, but most of all in their vocabulary). We will merely give
some illustrative examples, for neutral intonation, to be used for comparisons:
/./: ('l: 'aHRIf m&din'tkUm23) {(m'di:n&tkUm23)} /'la: 'aHrif[u] madi:na'takum./ Laa arifu madiinatakum.
/?/: ('hl lki't:b Z'di:d21) {(G-)} /'hal alki'ta:b[u] Za'di:d[un]?/ Hal al-kitaabu
àdiidun?
//: ('jUmkIn &nnA'u:m bi'Zul2 2) {(bi'Gu-)} /'jumkin[u] anna'u:m[a] bi'Zaula[h]/ Yumkinu an naquuma bi-àulatin
9.6. Arabic preintonemes and intonemes.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 ' 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
Text
9.4.0. e story e North Wind and the Sun follows. According to the Phonetic
method we start with the version in Arabic pronunciation of (neutral British) English (the written text is given in 2.5.2.0). e Arabic translation follows, in its
neutral version.
At the end, as always, there is the version which gives the English pronunciation of Arabic, by neutral British speakers, uent in Arabic (after prolonged contact with native speakers, but with no help from the phonetic method), who have
and suprasegmental elements which are typical of neutral British English (for reference purposes, although, of course, a neutral accent is not so common). Obviously,
the same principle is valid for the foreign pronunciations of English, given rst.
Speakers of American English could prepare their own version both of the Arabic pronunciation of English and of their pronunciation of Arabic, as an excel-
330
lent exercise, by listening to native speakers, best of all after recording them. Of
Arabic pronunciation (of English)
9.4.1. (d'n5s 'wInd2 nd'sn2 w5dIs'bjutI 'wIS wzzst'5g523| 'wn t'5fl52 'k:m 'lk2 '5bt in'w5m 'kl:k23| z'g5i:d2\ &dtz'wn Hu'f5s sk'si:dId2
In'm:kIn dt'5fl52 't:k hIsk'l:k 'f2 2| &SUdbikn'sid5ds "t5g5 &zndi'z523||
'dn2 d'n;5s 'wInd2 'blu:2 z'H5d2 zHi'kUd23| &btd'm5 HIb'lu:2 2| z'm5 'kl:sli2
&dIdzt'5fl52\ 'f:ld HIsk'l:k '5UndHIm23| &ndt'l{:}st2\ z'n5s 'wInd2 'g:f 'b
di'tmt23|| 'dn2 d'sn 'Sn 'Ut23 'w5mli23| &ndi'mi:djtli2\ dt'5fl5 'tUk 'f23
hIs'kl:k23|| nd's:2 z'n5s 'wInd2 wzb'lIZd tukM'fs2| &ztd'sn23 wsdst'5g523 &fz'tu:23||
&didju'lIk21 ds't:5i2| &duju'wn t'Hi5It 'gn21|||)
Arabic text
9.4.2. is passage highlights the fact that the modern classical Arabic language is a rather articial concept. In fact, the currently unwritten short V have
very uctuating realizations, due to both their presence or absence and to their
timbres (themselves) (/i, a, u/). As the number of recordings (of dierent speakers)
increases, the number of variations also increases (even for stressing and orthology, the use of intonemes and pauses).
9.4.3. Kaanat riiu al-amaali tataàadalu wa al-amsu i ayyin minhumaa kaanat aqwaa min al-uraa, wa i bi-musaarin ya=lau mutalaan bi-abaa'atin samiikatin. Fa ittafaqataa alaa itibaari al-saabiqi i i`baari al-musaari alaa ali abaa'atihi al-aqwaa.
aafat riiu al-amaali bi-aqaa maa ista=aa'at min quuwatin. Wa laakin kullumaa izdaada alafu, izdaada al-musaaru tadauran bi-abaa'atihi, ilaa an usqi=a
i yadi al-rii fataallat an muaawalatihaa. Bada'iin sa=aati al-amsu bi-dif'ihaa, famaakaana min al-musaari illaa an alaa abaa'atahu alaa al-tauu. Wa
hakaaa i=arrat riiu al-amaali ilaa al-itiraafi bi-anna al-amsa kaanat hiya
al-aqwaa.
Hal kaanat al-qiatu àmiila? Hal turiidu an nuraddidahaa?
9.4.4. ('k:nt2 '5i:hUS S'm:l[i]2 &tt'Z:d&lu wS'Sms[u]2 2| fi'ij[In] 'mInHu&ma2|
'k:nt 'Awa2 &mInl'uRa23| w'I bimu's:fIR[In]2| 'jAtAHU &mut'lffI&Ham2 &bI-
9. arabic
331
Ha'b:[tIn] s'mi:k23|| 'ft tfA'Ata2 2| 'HalaH ti'ba:RIs 's:b[]2[ &fiIZ'ba:RIl mu's:fIR[i]2| 'Hala 'alHI Ha&b'tiHIl 'A-wa23||
'HAAft2 '5i:hUS S'm:l[i]|| bi'AA 'm: IstA'tA:t m,'u:wa23|| w'l:kIn2 'kUllmz 'd:dl 'HAf[u]2 2|| Iz'd:dl mu's:fIR[u] t'du&Ran2] &biHa'b:&tIh2 2| 'il
n'Us&tA2 fi'jdIR '5i:h23|| &fta'allt 'Ham mU&hawl'tiHa23|| baH'di&In2| &sAtA'tIS 'Smsu bi'dIfi&Ha2| &fm'k:n mIn&lmu's:fIR[i]2| 'Ill a,'ala&Ha Ha&b'tHu23[ 'Halt 'tu23|| 'w 'Hk& d'tA5Rat2 '5I:hUS S'm:l[i]2 2| 'ill &IHti'5a:f[i]2 bi'nnS 'Sms[]2| 'k:nt2 'hijl 'A-wa23||
'hl 'k:nt A'Atu2 Z'mi:l21|| 'hl tu'5i:d[u]2 'n &nuRad'did&Ha21|||)
mIn'Huma2 jAt'AHU 'tfA&Ata2 2| [ s'b2| Ha'b:&tiHIl
kUl'lmz ] &td'uRan2 bI&Hab'tIh2 2| &nUs'tA2 mUha'wl&tiHa23||, mU'ha:w&lti&Ha23||, mU&haw'lti&Ha23|| -I2|, 'baHd&iIn2|, -&I2|
'sAtA&tIS 'a,a&laHa [ Ha'b:&tHu2 H'k nu'5addid&Ha21|||
English pronunciation of Arabic
9.4.5. ('khn2 5>IihuS S'mli2 &TT5GD&lu wS'Smsu32| fIi5am 'mnh&mA;2| 5khnT 'A;kw2 &mn'uk>3 3| w5 &bIim'sf>n2| 5jA;tl&hu &muT'lfhm2 &bh5bTn s'mIik3 3|| 5fT[ Tf'khA;T32| 5hl T'bA;>s 'sbk[]2
&fIiG5bA;> m'sf&>Ii2| 5hl 'khhi h&b'ThIih 'A;kw3 3||
'hA;s&fT2 '<IihuS S'mli|| bi'A;ks 'mA: st'ThA:T m'kuA;23|| w'lkn2 'khlmz] 'DD 'hA;fu32|| z'DD mu5sf&>u t'D&>n2 &bIih'b&TIi32| 5Iil n'usk&TA;2 fIi5D '>Iix3 3|| &fT5khlT 'hm m&hA;wl'ThIih3 3|| b'Da&n2| &sA;TA;5TS 'Smsu bIi'Dfh2| &fm5khn mn&m'sf&>Ii2| 5 l'khlh A;&b'Thhu3 3 5hlT 'Tha;3 3|| 5w 'hk D'ThA;>T2 5>IihuS S'mli32| 5Iil
&T'>A;fi2 bi5nS 'Sms2| 'khnT2 5hIi 'A;kw3 3||
5h 'khnT A;'khsT2 G'mIil21|| 5h Tu'>IiDu2 5n &n>'DIiD&h21|||)
-'m;2 mn'hum jT'lA;h [ 5faT s'bIik[i] -'mA;li2 ]
k'lA;m &n'skIiT2 h'kh -'mA;li32| &T'>I;iD2 5n
10. Hindi
10.0.1. Neutral Hindi pronunciation is presented in detail (whereas regional
variants are given only for the /EE, OO/ ae ao phonemes, which constitute the major peculiarity 10.2.). Our transliteration diverges from more traditional
one(s) and from English orthography as well, which neglects vowel duration. It
shows long vowels by doubling their graphemes (ii aa uu), as it does for consonants (pp nn), even when there is no opposition between short and long (ee oo),
for coherence. On the contrary, the historical indication of diphthongization is
kept, for the two vocalic phonemes mentioned at the beginning (ae ao), instead
of using a more complicated spelling (even if a little less inaccurate, perhaps, such
as ]
We can still nd and quite often, even if it is, rightly, losing ground a kind
of transliteration which shows the three short vowels, /i, a, u/ (I, , U), as i a u
and the three corresponding long, /ii, aa, uu/ (ii, aa, uu), as i a u but, at the
same time, it shows /ee, oo/ (ee, oo), just as e o (since there are no corresponding
short phonemes). en transcriptions are not added, in that kind of system, it
would be decidedly better to write e, o. However, we are left with the problem of
/EE, OO/, which if transcriptions are still lacking could perhaps coincide more
with the last indications given above, if one chose to write them as , .
10.0.2. Nasalized vowels are marked with the tilde sign (i q ); however, in
accordance with the system of marking long vowels with a macron, awkwardly,
the tilde should be put over the macron ( > <] as some transcribe. Nevertheless,
others use less satisfactory digraphic devices (a a a a).
Postalveolar consonantal articulations are indicated by a dot underneath (=
); postalveopalatal ones with an inverted circumex (or haek\ ; instead of
traditional, but ambiguous, c j ] for the uvulars of Arabic and Persian loanwords
we use particular signs (q ; instead of digraphs or other more complex combinations, or less evident, from a visual point of view, as well). ese have been mentioned in order to avoid both signs and also digraphs which could be too generic.
e latter are reserved for aspirated consonants (ph bh h). Furthermore, we
decidedly prefer w (to v] for (6, j) /w/.
10.0.3. Traditional devanagari script [deewnaagrii /deew'naagrii/ (de6'naag-Ri))
is not of great help, unless one has already learnt it. On the other hand, it is more
syllabic than alphabetic (and with a lot of monographs, indeed more than 150, in
addition to the over 40 basic signs), for sequences of polyphonemic graphemes.
erefore, it is not suitable to indicate actual phonic structures in a simple way,
with a further complication constituted by (short) inherent a, which is not written when preceded by C except in (some) transliterations. ese transliterations
10. hindi
333
are, basically, of two kinds: those that indicate /a/ only when it is really pronounced, and those that show every a (or very many), as for instance in ('mt-lb)
/'matlab/ matlab (or matalaba]. However, the actual phonetic situation may be
something in between, as can be seen in 10.3.1.1 as well.
Vowels
10.1.1. 10.1 shows Hindi vowels, which are ten: three short, (I, , U) /i, a, u/,
and seven long. ese are actually diphthongs: ve monotimbric, with narrow
movements, but nevertheless perceptible, (ii, ee, aa, oo, uu) /ii, ee, aa, oo, uu/, and
two more evident ones (even if, generally, they are simply described as long V
themselves), (, ) /EE, OO/ (because native speakers think they are monophthongs, including the variants given, and use them as such when learning foreign
languages).
Let us see, for now, some examples of the three short V\ ('dIn) /'din/ din ('pR)
/'par/ par ('kUl) /'kul/ kul the three corresponding diphthongs are: ('diin) /'diin/
diin ('paaR) /'paar/ paar ('kuul) /'kuul/ kuul the other four diphthongs are: ('beeR)
/'beer/ beer ('bR) /'bEEr/ baer ('booR) /'boor/ boor ('bR) /'bOOr/ baor We will
deal again with the geographic variants of /EE, OO/ ( 10.2.) again, only here,
('bR) /'bEEr/ baer ('bR) /'bOOr/ baor\ ('bR, 'bR) (east: Bihar, West Bengal),
('bER, 'bOR) (west: Rajasthan), ('bER, 'bOR) (northwest: Panjab)
10.1. Hindi vowels.
/ii/ ('ii, &i, i)
/i/ (I)
/ee/ ('ee, &e, e)
/uu/ ('uu, &u, u)
/EE/ (', &, )
/OO/ (', &, )
/a/ (', &, 0, )
/u/ (U)
/oo/ ('oo, &o, o)
/aa/ ('aa, &a, a)
10.1.2. Each Hindi vowel may be (distinctively) nasalized: (t'hI) /t'hi/ thi ('H,
&H, H) /'h/ hq (b'H) /b'h#/ bh (k'h_9) /k'h/ kh let us consider, for
instance ('Rg) /'rag/ rag vein ('RAg) /'rg/ rg color. Currently, words such as
('Hs) /'hans/ hans tend to be confused with ('HAs) /'hs/ hs but, in neutral pronunciation, they must be accurately distinguished, even if () has no full contact
with the alveolar ridge ( 10.2.1.1-2).
en ae ao are followed by /j, w/, they correspond to (I, U) /ai, au/: ('mIa)
/'maijaa/ maeyaa (tI'jaaR) /tai'jaar/ taeyaar ('HUja) /'hauwaa/ haowaa e same
should hold for Sanscrit words with ae ao (ai au); but, currently, these become
/EE, OO/ ( 10.2.1.2).
Intra-lexemic sequences /eeh0, ooh0/ may be realized as short vowels: ('seeH-Ra,
'seH-) /'seehraa/ seehraa ('mooH-lt, 'moH-) /'moohlat/ moohlat when there is grammemic derivation, no shortening occurs: ('leeH-j) /'leehja/ leehya ('mooH-na)
334
/'moohnaa/ moohnaa Even in English loanwords we nd (more or less evident)

shortenings: ('pee) /'pee/ pee= stomach, but ('pe;, 'pe, 'p) /'pee/ pee= pet (animal).
e sequence /ah/, in front of a C or in front of short /a/, or at word boundary
(/ah0, aha, ah/), is realized as ('H, H) (which could be marked as /E/): (kH'naa) /kah'naa/ kahnaa (pH'laa) /pah'laa/ pahlaa (l'HR) /la'har/ lahar ('tH) /'tah/
tah In the other cases, /a/ remains ( the following section) with the following realizations (, , , , x, ): (kx'Haa) /ka'haa/ kahaa (d'Hii) /da'hii/ dahii in Sanscrit
loans we nd /a/ before /ha/ as well: (R'Hs-j, 'RHs) /ra'hasja/ rahasya (gx'Hn)
/ga'han/ gahan (m'Ht:j) /ma'hattwa/ mahattwa
10.2. Regional variants () and unstressed neutral taxophones of /a/ ().
. Regional variants of ae ao /'EE, 'OO/: ('E,

'O) west (Rajasthan); ('E, 'O) northwest
(Panjab); (', ') east (Bihar, West Bengal).
. Attenuated taxophones of /a/: (),

((0)), (j, j), (()): (m, p[h],
b[H], , 6), ((x)): (k[h], g[H], , , r).
10.1.3. In unstressed syllables, inherent a is pronounced only when it is necessary to give substance to a phono-syllable. us, it is often not pronounced at all,
except in the case of consonant clusters which are not so easy to produce. Consequently, at the end of words (or word-internally, mainly after /h/), some unchecked phono-syllables are generally produced with a fading vocoid of the ()
type ( 10.1). However, this timbre, although attenuated (in length as well),
changes in accordance with the contoids preceding it (even in /0h/ sequences).
erefore, as shown in 10.2., after labials ((m, p[h], b[H], , 6)) it has rounded
lips, (); after velars or uvulars ((k[h], g[H], , , r)), it moves back, (x) (besides, after (j), we have ()); after /j/, it moves forwards, () (this phone also occurs in
('H0, H0) /ah0/ ahC and with /h/, near /ee, EE/, as well as an echo).
It is a good thing to manage to use these ve vocoids (in their attenuated, rather
than full, forms), since the coarticulatory logic is quite evident. However, a rather
satisfying result can be achieved, if we systematically use an attenuated () (while
a full () would sound too pushy).
Let us see some examples (but it should be noted that, often, in various published texts, we nd () for (), even if stressed): ('kN-h) /'kanh/ ka=h ('p~C) /'panc/ pan ('kRm) /'karm/ karm (karma), ('bagx) /'baang/ baa;g
(('ba) /'baa, 'baang/ is possible as well), ('muuRk-hx) /'muurkh/ muurkh (M'6) /an'waj[a]/ anway(a) ('gM-j) /a'gamj[a]/ agamy(a) ('n-j) /'anj[a]/ any(a) ('g-j) /a'gaNj[a]/ agay(a) ('mooH&na) /'moohnaa/ moohnaa ('mHl) /'mahl/
mahl ('meeH&tR, &meH'tR) /'meehtar/ meehtar
10. hindi
335
10.1.4. In unstressed syllables, the phonetic diphthongs are normally realized

as short monophthongs (or, in slow or more accurate pronunciation, as half-lengthened monophthongs): ('HIndi -di;) /'hindii/ hindii (I'daaRa -Ra;) /i'daaraa/
idaaraa ('jaad -d;) /'jaad/ yaad (a'baa a;-) /Saa'baaS/ aabaa According
to stress strength in sentences, for /EE, OO/, as in hae we have: ('H, &H, H) /hEE/.
Unfortunately, Hindi grammars persist in placing among the vowels the
grapho-syllable which in Sanscrit indicated /(/ () and traditionally is transliterated , but today it only stands for /ri/ (RI) ( a /0/ sequence): ('RI) /S'ri/ ri
Consonants
10.2.0. 10.3 shows Hindi consonants, including phono-stylemes for words
coming from Sanscrit (corresponding to ; ) and Arabic, Persian, Turkish and
English (q f z , ), and taxophones as well.
1.9-15, instead, show the orograms (gathered by articulation manners) of all
the contoids given in the chapters of this book even as secondary, occasional, or
regional, variants for the 12 languages dealt with. is exposition makes necessary comparisons more direct.
m|(M)
F p5 b
{} 6
_
(n) n|() {N|()}

t5 d
5
s {z}
{}
(~|)
(n) ()
C5
j|()
(l)
R
l
($)
{}|()
k5 g
{(,)}
{}
/w/
(j)
{}
laryngeal
uvular
velar
rounded
velar
provelar
palatal
postalveopalatal
postalveolar
alveolar
dental
bilabial
10.3. Table of Hindi consonants.
{}
(h) H
{r}
()
/c, G/ (C, ), /S/ (), /j/ (j, ), /F/ (), /w/ (6, j), /r/ (R), /#/ ()
Nasals
10.2.1.1. ere are two fundamental nasal phonemes, /m/ (m), and /n/ (m, n,
N, ~, , ,), since the NC sequences are homorganic. Besides, (M) (()} occurs before /w, F/; () (()) before /s/; (n) ((q)) before /j/; and () {()} before /h/ with
a rare //, which would give /h/ (and this would show the advantage of considering sequences of Ch biphonemic, even for stops) or /ngh/ (gH, H). (It is worth
observing that (M, , , n, ) are nasals with no full contact ( semi-nasals): semi-labial, semi-alveolar, semi-postalveo-palatal, semi-palatal, and semi-provelar, respectively, 9.9 of NPT/HPh and () {()} as well, given in 10.2.1.2.)
336
On the other hand, it is not necessary to use the specic symbols ((, n)):
('mooH) /'mooh/ mooh ('smbl) /'sambal/ sambal ('tiim ma'tae) /'tiin maa'taaee/
tiin maataaqq (sM'jaad) /sam'waad/ samwaad ('nbH) /'nabh/ nabh ('kNh)
/'kanh/ ka=h ('k~) /'kanG/ kan ('6) /'wanS/ wan ('n-j) /'anja/ anya
('pk) /'pank/ pa;k (&I,x'laab) /ina'laab/ inqalaab ('sIH[x]) /'sih[a]/ sinh(a)
In some words, we nd /m, n/ before heterosyllabic C\ (b'Ramda) /ba'raamdaa/ baramdaa (m'Ruud) /am'ruud/ amruud ('gUmi) /'gumii/ gum=ii ('CmCa)
/'camcaa/ amaa ('Umka) /'Gumkaa/ humkaa ('meenka) /'meenkaa/ meenkaa
(indicated in writing by the segments m n instead of by anuswaar a diacritic
sign).
10.2.1.2. Furthermore, in lofty Sanscrit words, also /N, / (N, ) occur, and
have specic graphemes, ;, but generally change into /n/. e more frequent
genuine realization of /N/ is (), a nasalized postalveolar ap, or even a nasalized
postalveolar approximant, (), so it is better to use the symbol of the diaphone,
(), which incorporates these values, although it generally becomes /n/. Also (~)
has a particular grapheme, n, even if it does not represent a phoneme (as was the
case instead in Sanscrit): ('bam) /'baam/ baam ('ban) /baan/ baan ('ba, 'baN,
'ban) /'baaN/ baa (gx'ee, gx'Nee, gx'nee) /ga'NeeS/ gaee ('gN-j, 'g-j)
/a'gaNja/ agaya ('6amI, -m, -m-, -nm, -mm, -mm) /'waamaj/ wa;may ('ba[gx]) /'baang/ baa;g.
Furthermore, we nd the sequences /mh, nh/ (mH, nH), which (together with
/lh/ (lH) and, possibly, /h/, 10.2.1.1) have no ocial devanagari graphemes,
but combinations. In fact, they did not occur in Sanscrit, contrary to aspirated
stops and stopstrictives, including /#h/ (H), which comes from /h/ (H): (kUm'HaaR) /kum'haar/ kumhaar ('kanH) /'kaanh/ kaanh
Stops
10.2.2.1. ere are four diphonic pairs, /p, b t, d , k, g/ (p, b t, d , k,
g), in addition to the voiceless uvular phonostyleme // () (which, currently, becomes /k/ (k)): (pI'taa) /pi'taa/ pitaa ('b) /'ab/ ab ('Raat) /'raat/ raat (n'dii) /na'dii/ nadii ('hiik) /'hiik/ =iik ('pIN) /'pin/ pi ('k9n) /'kOOn/ kaon ('gana)
/'gaanaa/ gaanaa ('k) /'kEE/ kae (') /'EE, 'kEE/ qae
10.2.2.2. e most remarkable peculiarity of stops is that the elements of diphonic pairs may occur in sequences, with /h/, producing /ph, bh th, dh h, h
kh, gh/ (ph, bH th, dH h, H kh, gH) (it is not rare to nd that the aspirated
voiced ones are, actually, devoiced, (H, H, H, H), although a fully voiced pronunciation is perfectly acceptable too). Generally, they are considered to be unitary phonemes: aspirated opposed to the corresponding non-aspirated. No
doubt, this opinion derives from traditional writing as well, since particular graphemes are in use.
We prefer to consider the aspirated sounds as phonemic (and phonetic) se-
10. hindi
337
quences constituted by plain stops + /h/, which is realized as (h), after voiceless C
or as the normal (in Hindi) voiced laryngeal approximant, (H), after voiced C (all
the more so because they are realized as heterosyllabic, instead of tautosyllabic; this
can be seen by the placement of stresses, which is more evident and logical within a word or a rhythm group): ('pl) /'pal/ pal (p'hl) /p'hal/ phal ('baala) /'baalaa/
baalaa (b'Haala) /b'haalaa/ bhaalaa ('tl) /'tal/ tal (t'hl) /t'hal/ thal ('daa6a) /'daawaa/ daawaa (d'Haa6a) /d'haawaa/ dhaawaa ('ppa) /'appaa/ =appaa ('hppa)
/'happaa/ =happaa ('iil) /'iil/ iil ('Hiil) /'hiil/ hiil ('kaal) /'kaal/ kaal
(k'haal) /k'haal/ khaal (gI'Raa) /gi'raa/ giraa (gHI'Raa) /ghi'raa/ ghiraa
10.2.2.3. It is not unusual for /b, bh/ to be realized as (6, 6H). Besides, attenuations are also possible for /k/ (, , ), /kh/ (xh, x), /g/ (, y, ), /gh/ (H). Before
front vowels (and before /j/), /k[h], g[h]/ are realized as prevelar, but no special symbol (([], [])) is necessary. In nal position, the stops may have inaudible realizations: ('naak, 'naak) /'naak/ naak ('b, 'b) /'ab/ ab this is contrary to what happens to aspirated sequences, which do not simplify, even if the laryngeal element
may be less evident, in this nal position (for other C too), but it may not be missing, because it is distinctive\ ('siikh, -kh) /'siikh/ siikh ('nbH, -bH) /'nabh/ nabh
(and: ('booH, -H) /'booGh/ booh ('baaH, -H) /'baa#h/ baah] Only within sequences like /0h0/, may /h/ be dropped ( 10.3.1.5). Also word-nal /h/ may be
attenuated (and even disappear completely): (b'jaaH, -aaH, -aa) /b'jaah/ byaah
Stopstrictives
10.2.3. We only nd the postalveopalatal diphonic pair, /c, G/ (C, ) (with the
corresponding aspirated sequences, /ch, Gh/ (Ch, H), and the possibility of
(H), as for the stops): (C'tUR) /ca'tur/ atur ('Raa) /'raaG/ raa ('Caal) /'caal/
aal (C'haal) /c'haal/ haal ('aal) /'Gaal/ aal ('Haal) /G'haal/ haal In various
contexts, there may be frequent attenuated realizations ( constrictives or approximants), even after a pause (or, instead, sharper ones, stops), for /c/ (, T), /ch/
(h, h Th), /G/ (, D), /Gh/ (H, H DH).
Constrictives
10.2.4. In actual fact, we nd two voiceless constrictives: /s, S/ (s, ) (sometimes,
the latter is realized as velarized postalveolar, (), () with velarization) and the
voiced /w/ (6, j) (whose rounded semi-velar approximant variant, (j), mainly occurs after C after /uu, u/, and sometimes after a pause but, in actual fact, the
two types alternate quite freely): ('sR) /'sEEr/ saer ('kooI) /'kooSiS/ kooi (&R6I'6aaR) /rawi'waar/ rawiwaar ('nii6, 'nii, 'niiM) (with possible nasalization, near
nasalized V]\ /'niiw/ niiw (6'Rt) /w'rat/ wrat ('HUja) /'hauwaa/ haowaa (s'jRg)
/s'warg/ swarg (6'H, |j-) /wa'h/ wah
In Sanscrit loans, we also nd // () (which currently becomes /S/ (): ('bI, 'bI)
338
/'bi/ bi; on the other hand, /s/ becomes (), before //: ('k) /'kas/ ka= (shown
in writing as well). In Persian, Arabic, and English loans, we nd /F, z/ (, z) too
(which currently become /ph, G/ (ph, )): ('iiR, ph-, -'kiiR) /Fa'iir, -'kiir/ faqiir ('Ut, p'hUt) /'Fut/ fut (ba'zaaR, ba'aaR) /baa'zaar/ baazaar
Approximants
10.2.5. In this articulatory manner, there are two fundamental phonemes, /j,
h/ (j, H, h); (the semi-palatal) () occurs between V in unstressed syllables: ('jee)
/'jee/ yee (dH'jan) /dh'jaan/ dhyaan ('lIe, 'lie, 'lIe) /'lijee/ liyee en in unstressed
nal position (with an inherent a] its sound is (-, -); when preceded by i it
drops, often closing /i/ {(I) = (i)}, except in careful pronunciation: (s'm, s'm)
/sa'maj/ samay (k', -, k'-) /k'aj/ kay ('Caa, 'Caa) /'caaj/ aay (k't-RI,
kt'RI, -I, -i, k-) /k'atrij/ katriy
Generally, the /h/ phoneme is (H) while (h) occurs with voiceless aspirated
consonants. Near nasalized V /h/ becomes nasalized: ('Haathi) /'haathii/ haathii
('gRH) /'garh/ garh (&pH'laa) /pah'laa/ pahlaa ('teeRH, te'RH) /'teerah/ teerah
(HAs'naa) /hs'naa/ hsnaa ('bH) /'bh/ bh (kx'H) /ka'h/ kah
ere are two further approximants, for Persian and Arabic loans, for which the
ocial uvular constrictives symbols are often used, /X, / (or, even less precisely,
the velar ones, /x, /); they are uvular approximants (, ) (but, for the voiced one,
more often, the uvular tap, (r), occurs). However, currently they become /k, g/ (k,
g): (d'l, d'kl) /da'Xal/ daal ('aana, 'kaana) /'Xaanaa/ aanaa ( (k'haana)
/k'haanaa/ khaanaa] ('baar, 'baa, 'baag) /'baa/ baa ( ('baag) /'baag/ baag]
Trills
10.2.6. In this category, we have an alveolar tap, /r/ (R), and a postalveolar ap,
/#/ (), which oppose distinctively (and sometimes the former may be stronger, a
true trill (r), or on the contrary weaker: an approximant (); often, the latter is
weaker than normal: an approximant (), similar to British English (>), but with
no lip rounding at all): (H'R) /ha'ra#/ hara (l'kaa) /la#'kaa/ lakaa ere is also the sequence /#h/ (H, H): (b'Haa, -'Haa) /ba#'haa/ bahaa ( (b'aa, -'aa) /ba'#aa/ baaa]
Laterals
10.2.7. ere is only one lateral phoneme, /l/ (l, $, ); besides, there is the sequence /lh/ (lH): ('laat) /'laat/ laat (m'laal) /ma'laal/ malaal ('k C'loo) /'kal ca'loo/ kal aloo ('k j'H 'aao) /'kal ja'h 'aaoo/ kal yah aaoo ('aa$a) /'aalaa/
aalaa ('kuula) /'kuulaa/ kuulaa ('kuulHa) /'kuulhaa/ kuulhaa ('deelHi) /'deelhii/
Deelhii
10. hindi
339
Structures
10.3.0. We will consider here the relationships between segments, syllabic structure, stress, and intonation. Most examples re-transliterated and transcribed are
based on recordings expressly made (taken from Shukla, 2000).
A typical Hindi pronunciation shows a particular kind of paraphonic setting
with breathy voice @, or, less positively, with tense voice @.
Taxophonics
10.3.1.1. As for V which oppose as short and long, or rather (very) narrow
diphthongs (generally represented by geminating symbols, 10.1.1), gemination is distinctive for C too: (p'taa) /pa'taa/ pataa ('ptta, pt'taa) /'pattaa/ pattaa
(b'Caa) /ba'Caa/ baaa ('bCCa, bC'Caa) /'baccaa/ baaa (U'see) /u'see/ usee
('Usse, Us'see) /'ussee/ ussee is fact changes the syllabic structure, consequently,
often even stress changes (although without phonemic relevance).
We have already mentioned and demonstrated in our examples, that in
Hindi sequences of two or three C syllabify moving the last one to the beginning
of the next syllable: ('st-j) /'satja/ satya ('gd-Ha) /'gadhaa/ gadhaa (bH'Rk) /abh'rak/ abhrak ('Uk-l) /'Sukla/ ukla (kh'aa) /akh'#aa/ akhaa. Obviously, in case
of isolated initial sequences, the syllabicity scale joins the elements in one syllable,
although some slight dierence is maintained; however, if they are internal, they
divide into two syllables, including the preceding V\ ('Hiil) /G'hiil/ hiil. e
geminates are (00) between V but (0:) + C even for /j, w, h, r, #, l/ (in this last
context, especially short stops and stopstrictives are often realized as (0:) + C thus
neutralizing the dierence with geminates): ('bU:Ha) /'buhaa/ buhaa (6Id[:]'aal, -l, -l) /wid'jaalaj/ widyaalay If the number of C in the sequences exceeds three, the last two move to the beginning of the second syllable: (st'jaaRthpR&ka) /sat'jaarthprakaaS/ Satyaarthprakaa.
Another interesting phenomenon, that complicates the description and the acquisition of this language, concerns the epenthesis of a vowel; therefore, an attenuated vocoid is inserted ( 10.1.3, although here we indicate it simply as an audible oset, ($)), not only in complex consonant clusters, but also in not quick pronunciation at the end of words, even after a single consonant: ('gR$) /a'gar/ agar
('uupR$) /'uupar/ uupar (Cl$'naa) /cal'naa/ alnaa ('phuul$) /'phuul/ phuul ('R$bt) /'Sarbat/ arbat (bd$'maa) /bad'maaS/ badmaa (l$'kaa) /la#'kaa/ lakaa
(khI$'kii) /khi#'kii/ khikii (bH'jaas$) /abh'jaas/ abhyaas (s'm$) /sa'maj/ samay
(Rakh$'naa) /rakh'naa/ rakhnaa (Ug$'naa) /ug'naa/ ugnaa ('uugH$na;) /'uughnaa/
ghnaa ('p$) /'pa#/ pa ('pH$) /'pa#h/ pah
10.3.1.2. Hindi phonotactics presents some peculiar consonant clusters, both
at the beginning and end of words. In fact, we may nd, for instance: (m'RIg)
/m'rig/ mrig (m'lan) /m'laan/ mlaan (n'jaa) /n'jaaj/ nyaay (nRI's) /nri'Sans/
nrians (p'jaala) /p'jaalaa/ pyaalaa (b'R) /b'raG/ bra (b'laak) /b'laak/ blaak
(bH'Rm) /bh'ram/ bhram (tj'Raa) /twa'raa/ twaraa (t'Raas) /t'raas/ traas (d'juut)
340
/d'juut/ dyuut (dH'jni, -'nii) /dh'wani/ dwani (dH'RU6) /dh'ruw/ dhruw ('jooHi, jo'Hii) /'joo#hii/ yoohii (C'jUt) /c'jut/ yut ('jooti) /G'jooti/ yooti (k'jaa)
/k'jaa/ kyaa (k'jaath) /k'waath/ kwaath (kh'jaab) /kh'waab/ khwaab (g'Ram)
/g'raam/ graam (g'lani) /g'laani/ glaani (gH'Ra, -N, -n) /gh'raaN/ ghraa.
More: ('jaal, kh'j-) /X'jaal/ yaal (H'Raas) /h'raas/ hraas (m'an) /Sma'Saan/
maaan ('Rant) /S'raant/ raant ('lth) /S'lath/ lath (s'kndH) /s'kandh/ skandh
(skhx'ln) /skha'lan/ skhalan (s'tR) /s'tar/ star (s'thl) /s'thal/ sthal (s'mIt) /s'mit/
smit (s'nan) /s'naan/ snaan (s'jaaH) /s'jaah/ syaah (s'jRg) /s'warg/ swarg (s'Ra)
/s'rasaa/ sra=aa (z'jaada) /z'jaadaa/ zyaadaa (6jt'haa) /wjat'haa/ wyathaa (6'Rt)
/w'rat/ wrat In words of Sanscrit origin, we nd /k/ as well (but people currently change it into (k)): (k', k'N, k'n) /k'aN/ ka (k'iiR, k'-) /k'iir/ kiir
Furthermore, for initial sequences with /s0[0]/, current pronunciation prexes
an epenthetic vowel, generally (I, I) but (, ) is possible as well (and also (, ),
with the other variants seen, 10.1.3), before non-front vowels: (s'tRii, Is-, Is-)
/s'trii/ strii (sp'huuRti, Is-, Is-, s-, s-, s-, s-) /sp'huurti/ sphuurti (s'nan, Is-, Is-, s-,
s-, s-, s-) /s'naan/ snaan (also with /S/: ('look, I-, I-, -, -, -, -) /S'look/ look]
10.3.1.3. As far as word-nal position is concerned, we nd consonant clusters
such as: ('gUpt) /'gupt/ gupt ('bd) /'Sabd/ abd ('lUbdH) /'lubdh/ lubdh ('ps)
/'aps/ =aps ('nbz) /'nabz/ nabz ('kUb) /'kubG/ kub ('mUt) /'muFt/ muft ('Uts)
/'uts/ uts ('Rkt) /'rakt/ rakt ('dgdH) /'dagdh/ dagdh ('ks) /'aks/ aks ('mook, -k)
/'mook/ mook ('st, -kht) /'saXt/ sat ('s, -khs) /'SaXs/ as ('b, -kh) /'baXS/
ba ('CUst) /'cust/ ust (s'jsth) /s'wasth/ swasth ('k) /'kas/ ka= ('6aap)
/'waaSp/ waap
In nal position, geminates are possible as well, realized as long: ('gp:) /'gapp/
gapp ('Cm:) /'camm/ amm ('CIt:) /'citt/ itt ('Rd:) /'radd/ radd ('n:) /'ann/ ann
('sR:, -r[:]) /'sarr/ sarr (pRp'hUl:) /prap'hull/ praphull (b'H:) /b'ha/ bha== (U':)
/u'Ga/ ua ('UC:) /'ucc/ u (nI'l:) /ni'laGG/ nila ('dg:) /'dagg/ dagg
sequences which include a laryngeal are also found: ('Ut:h) /'Gutth/ utth ('jUd:H)
/'juddh/ yuddh ('sIk:h) /'sikkh/ sikkh (s'jC:h) /s'wacch/ swah
nal consonant clusters, in morphological derivation, resyllabify in accordance
with the phonic structure of the Hindi language: ('paap) /'paap/ paap ('paa-pi)
/'paapii/ paapii ('Rkt) /'rakt/ rakt ('Rk-tIm) /'raktim/ raktim
10.3.1.4. Short (I, , U) /i, a, u/, in nal position, have the peculiarity of fading
( 10.1.3), to a point in which they drop as well, as happens to /a/, ( , , , x,
`), or strengthen, (i, u), simply as closer realizations of /i, u/, or actually becoming /ii, uu/. In this case, stress shifts are also possible, according to the weight
of the syllables constituting given words: ('R6i, R'6ii) /'rawi/ rawi ('tIthi, tIt'hii)
/'tithi/ tithi ('Iu, I'uu) /'SiSu/ iu ('6su, 6'suu) /'wasu/ wasu e change V =
VV is not considered completely neutral, in spite of being very widespread indeed
(with many supporters too).
nally, as many examples have already shown, sequences of /'/ + /, , 0/
are realized as ('): ('Hani) /'haani/ haani ('kam) /'kaam/ kaam (ak'Rant)
/aak'raant/ aakraant
10. hindi
341
10.3.1.5. Dealing with consonants, we have already seen the assimilatory characteristics within words. e same holds in sentences, for words linked by meaning, especially if joined in rhythm groups, as in: ('tiim ma'tae) /'tiin maa'taaee/ tiin maataaqq ('tiim p'piite) /'tiin pa'piitee/ tiin papiitee ('tiiN 'aal) /'tiin 'aal/ tiin aal ('tii kx'ml) /'tiin ka'mal/ tiin kamal ('tii 'gaa) /'tiin 'gaaj/ tiin gaay
('tii~ 'Caa) /'tiin 'caaj/ tiin aaj ('tii~ 'aal) /'tiin 'Gaal/ tiin aal ('tiin 'Raab) /'tiin Sa'raab/ tiin araab ('tiin 'jaaR) /'tiin 'jaar/ tiin yaar ('tiiM 'jR, -R) /'tiin 'war/
tiin war ('tii, x'saai, - kx-) /'tiin a'saaii/ tiin qasaaii ('tii 'Haath) /'tiin 'haath/
tiin haath
Besides, both within words and sentences, in not slow nor emphatic speech,
for diphonic C voicing assimilation is regular, to the second element of a sequence
(while a possible /h/ is lost): ('gbR, g'bR) /'akhbar/ Akhbar (tz'biiH) /tas'biih/
tasbiih ('CUb 'bho, 'CUb b'hoo) /'cup 'bEEhoo/ up bae=hoo ('Chb gx'jaa)
/'cap ga'jaa/ hap gayaa ('saad 'bho, 'saad b'hoo) /'saath 'bEEhoo/ saath bae=hoo ('p 'tk) /'ab 'tak/ ab tak ('Haad 'doo) /'haath 'doo/ haath doo ('aakpR) /'aagpar/ aak par ('eeg 'dIn) /'eek 'din/ eek din ('pki) /'Gabki/ ab ki ('baa 'kaa, -k
'k-) /'baa 'kaa/ baa kaa
Stress
10.3.2.1. In Hindi, stress position is not distinctive; in fact, the same speaker,
in dierent occasions, may stress dierent syllables of the same word. Furthermore,
these uctuations also depend on the placing of words in sentences, on nearby
words, on orthological highlighting and emphasis. Still more important is the fact
that stress is distributed among rhythm groups, usually moving back from the end,
according to syllabic weights. is holds for isolated words as well, but always
with a certain exibility. For instance, we normally have ('HIndi) /'hindii/ hindii
but (HIn'dii&ke, -&ka) /hin'diikee, -kaa/ hindii kee hindii kaa
On the other hand, given its non-distinctiveness, native speakers use stress as
something uctuating (without fully realizing it), frequently alternating, for rhythmic reasons, within rhythm groups. After all, it is the same thing for segmental
duration and for syllabic pitch, in languages where they are not distinctive: they
may change quite freely, without real problems. In Hindi intonemes (both for the
three marked and the unmarked one, or continuative), the terminal posttonic syllable, generally, bears a secondary stress, which complicates the (already unsettled)
dierentiation and identication of stress strength on the dierent non-light syllables (but sometimes on the light ones too).
10.3.2.2. However, some rules may be formulated in order to produce a coherent eect, if rigorously applied (even if they might be considered excessively precise or even mechanical as regards current language).
Regardless of the C that may precede a V we dene a light syllable one which
contains a short V (I, , U) /i, a, u/ not followed by any C (in the same syllable):
(kI, ki) /ki/ ki (sU'mtI, -ti) /su'mati/ sumati within a word, the attenuated reali-
342
zations of /a/ (, , x, , ) do not count ( 10.1.3), being usually more elusive,

(, , x, , ), and are just a mere physical support, necessary to make a word pronounceable, but they may drop completely.
Furthermore, we have mid syllables constituted by VC ( a short V and a C]
or by VV\ ('bl) /'bal/ bal (k'jaa) /k'jaa/ kyaa ('HIndi) /'hindii/ hindii ('jja) /'Sajjaa/ ayyaa including disyllabic syllables in counting morae, as (b'Haai) /b'haaii/
bhaaii (/aa/ + /ii/ = four morae).
Besides, there are heavy syllables formed by VCC or VVC (more rarely VVCC
as well): ('nt) /'ant/ ant ('am) /'aam/ aam ('aaRt) /'aart/ aart ('6aap) /'waaSp/
waap
10.3.2.3. Stress assignment, in isolated words, is done on the basis of the identication of the heaviest syllable, as in: (U'paadHi) /u'paadhi/ upaadhi (U'pant)
/u'paant/ upaant (&ad-jo'pant) /aadjoo'paant/ aadyoopaant (kx'lii) /ka'lii/ kalii
(kx'man) /ka'maan/ kamaan ('kamna) /'kaamnaa/ kaamnaa (ko'aaRk, ko'NaaRk, ko'naaRk) /koo'Naark/ kooaark ('andani) /'Xaandaanii/ aandaanii
('jamIti) /G'jaamiti/ yaamiti (&tabe'daaR) /taabee'daar/ taabeedaar (&tIgU'naa)
/tigu'naa/ tigunaa (p&RIthji'Raa-Ra&so) /prithwii'raaGraasoo/ prithwiiraaraasoo.
More: (bn'duugba&zi) /ban'duukbaazii/ banduukbaazii (mR'dagi) /mar'daangii/ mardaangii ('mandHa&ta) /'maandhaataa/ maandhaataa (&st-jpR'kaa)
/satjapra'kaaS/ Satyaprakaa (st'jaaRth-pR&ka) /sat'jaarthprakaaS/ Satyaarthprakaa
(&sago'pag) /saangoo'paang/ saangoopaang (sa'man) /saa'maan/ saamaan ('Hani) /'haani/ haani
If a word has more than one non-light syllables of the same weight, there are two
possibilities: the stress preferably hits either the last syllable but one or the last but
two (or even the very last, especially if constituted by VV as happens more often
within a sentence, with particles and postpositions): ('kUnt&la, &kUnt'laa) /Sa'kuntalaa/ akuntalaa (gx'eeRI&a, gx&e;RI'jaa) /ga'#eerijaa/ gaeeriyaa (b'HaadU&Ri,
b&HadU'Rii) /ba'haadurii/ bahaadurii (&mHab'HaaRt) /mahaab'haarat/ mahaabhaarat (jUd'HIhIR, &jUdHI'hIR, --, -'-) /jud'hiir/ Yudhi=hir (s'HuulI&t, s&HulI'jt) /sa'huulijat/ sahuuliyat.
More examples: (a'iiR-jad, &aiR'jaad) /aa'Siirwaad/ aaiirwaad ('xl&mnd, &xl'mnd) /'almand/ aqlmand (&ana'kani, a&naka'nii) /aanaa'kaanii/ aanaakaanii
(dH'jni, -'nii) /dh'wani/ dhwani (pR'jRti, &pRjR'tii) /par'wartii/ parwartii
('bRtn, bR'tn) /'bartan/ bartan ('bs-ta, bs'taa) /'bastaa/ bastaa (b'Haala, bHa'laa) /b'haalaa/ bhaalaa (s'mIti) /sa'miti/ samiti (sa'lana, &sala'naa) /saa'laanaa/
saalaanaa (s~'Caaln, &s~Ca'ln) /san'caalan/ sanaalan (&HeRap'heeRi, He&Raphe'Rii) /heeraap'heerii/ heeraapheerii
10.3.2.4. e same stress pattern occurs in inected and derived words as well:
(&bd-HI'k) /badhi'k/ badhik (&lg-HU'tR) /laghu'tar/ laghutar (&UCI'tm) /Suci'tam/ uitam (ms'leega, &msle'gaa) /mas'leegaa/ masleegaa ('laap&ta;, &lap'taa)
/'laapataa/ laapataa (&gxlI'jaaRa, gx&lIa'Raa) /gali'jaaraa/ galiyaaraa (pa'glpn, &pa;gl'pn) /paa'galpan/ paagalpan (sUn'dRta, &sUndR'taa) /sun'dartaa/ sundartaa
(bH'naapa, &bHna'paa, b&Hna'paa) /bah'naapaa/ bahnaapaa
10. hindi
343
Among Hindi monosyllables, lexemes bear a stress, even in sentences, while

grammemes ( postpositions, conjunctions, auxiliaries, enclitics) are unstressed
or half-stressed: ('b) /'ab/ ab ('km) /'kam/ kam ('kam) /'kaam/ kaam ('R) /'OOr/
aor ('dee) /'deeS/ dee but (ka) /kaa/ kaa (H, H) /h/ hq
10.3.2.5. Lexemic compounds, normally, have the following stress pattern (' )
/' / (since the second stress, which is generally more attenuated, may sometimes
reach a degree of prominence which is relatively slightly more perceptible than a
secondary stress): (R'sooigHr) /Ra'sooiighar/ rasooiighar ('deenIka[a]la) /'deeSnikaalaa/ deenikaalaa ('6Ijko[o]) /'wiSwakooS/ wiwakoo ('duuRdRi[i]) /'duurdarSii/ duurdarii ('dI-la[a]) /'dilGalaa/ dilalaa ('nImnlIkhIt) /'nimnlikhit/
nimnlikhit ('CndRka[]nta) /'candrakaantaa/ andrakaantaa ('CndRka[]ntasntti) /'candrakaantaasantati/ andrakaantaasantati.
More: (m'HaasbHa[a]) /ma'haasabhaa/ mahaasabhaa ('HIndumHa[a]sbHa[a]) /'HIndumahaasabhaa/ mahaasabhaa ('tuu 'kRmHi[i]M j'jkti&H) /'tuu
'karmhiin w'jaktihEE/ tuu karm-hiin wyakti hae ('tuu 'kRmHi[i]&H) /'tuu 'karmhiinhEE/ tuu karm-hiin hae ('amja[]m) /'aamwaam/ aam-waam ('kanka[]n) /'kaankaan/ kaan-kaan ('kmka[a]) /'kaamkaaG/ kaam-kaa ('sa6[]sa) /'EEsaawEEsaa/ aesaa-waesaa
Complete reduplications always keep both stresses: ('laal 'laal) /'laal'laal/ laal-laal (d'HiiRe d'HiiRe) /d'hiireed'hiiree/ dhiiree-dhiiree
10.3.2.6. ere is a dierence between compounds and collocations (which
have independent sentence ictuses): ('laalpg&i;) /'laalpag#ii/ laal-pagii policeman ( red-turban) and ('laal 'pg-i, 'laal pg'ii) /'laal 'pag#ii/ laal pagii red
turban, ('kaala&pani) /'kaalaapaanii/ kaalaa-paanii penal colony ( black-water) and ('kaala 'pani) /'kaalaa 'paanii/ kaalaa paanii black water, ('miiha&te;l)
/'miihaateel/ mii=ha-teel sesame-oil ( sweet-oil) and ('miiha 'teel) /'miihaa
'teel/ mii=ha teel sweet oil, ('niil&ga;) /'niilgaaj/ niil-gaay (a species of) large antelope (blue-cow) and ('niil 'gaa) /'niil 'gaaj/ niil gaay blue cow.
For contrast, there is emphasis on the marked element (and attenuation on the
second occurrence of the unmarked element): ('HIndu"mndIR2 2 'jaa &HIndugHR3 3)
/'hinduu"mandir 'jaa 'hinduug"har./ hinduu-mandir, yaa hinduu-ghar? ('tmpReem2 2 'jaa &a;tmgjan3 3) /'atmp"reem 'jaa 'atmg"jaan./ atm-preem, yaa atm-gyaan
Intonation
10.3.4. 10.4 shows the preintonemes and intonemes of the neutral Hindi language. We will merely give some illustrative examples, to use for comparisons. Let
us observe that the posttonic syllables of the interrogative ((2 1)) and suspensive
((2 2)) intonemes, which are in the higher parts of the mid and high bands (as will
be clear from 10.4), are generally accompanied by falsetto (for all kind of voices), which is marked with () after the intonemes, in phonotonetic transcriptions.
In addition, paraphonically, there is a peculiar type of Indian voice, with murur
344
voice @ or, in the broadest accents (), with tense voice @:

/./: (&mHIndi boltaaHu3 3) /m'hindii bool'taahuu./ Mq hindii booltaa h.
/?/: (k'jaa6h pUs'tkpH R'Hii&H21) /k'jaawah pus'takpa#h ra'hiihEE?/ Kyaa
wah pustak pah rahii hae?
//: (b'm l'dii l'dii bol'taa&Hu2 2 ap'meeRi 'baat n'Hii &smCtee3 3)
/Gab'm Gal'dii Gal'dii bool'taahuu aap'meerii 'baat na'hii samaGh'tee./ ab
mq aldii-aldii booltaa h, aap meerii baat nahii samahtee.
10.4. Hindi preintonemes and intonemes.
/ / (2 2 2 2 2 2 2)
/./ (2 3 3)
/ / ( 2 2 2 2 2 2 2)
/?/ (2 ' 2 1)
/ / ( 2 2 2 2 2 2 2)
// (2 2 2)
/ / ( 2 2 2 2 2 2 2)
/,/ (2 ' 2)
Text
10.4.0. e story e North Wind and the Sun follows. According to the Phonetic method we start with the versions in Hindi pronunciation of (neutral British) English (the written text is given in 2.5.2.0). e Hindi translation follows,
in its neutral version.
At the end, as always, there is the version which gives the English pronunciation of Hindi, by neutral British speakers, uent in Hindi (after prolonged contact with native speakers, but with no help from the phonetic method), who have
and suprasegmental elements which are typical of neutral British English (for reference purposes, although, of course, a neutral accent is not so common). Obviously,
the same principle is valid for the foreign pronunciations of English, given rst.
Speakers of American English could prepare their own version both of the Hindi pronunciation of English and of their pronunciation of Hindi, as an excellent
exercise, by listening to native speakers, best of all after recording them. Of course,
speakers of other languages could do the same thing. e author would be happy
to receive their transcriptions and recordings, both in case of help should they
need it and to make their contribution known to others (possibly in our website
on canIPA Natural Phonetics 0.12).
10. hindi
345
Hindi pronunciation (of English)

10.4.1. (dnOR '6IN2 End'sn2 6RIspuuI 6IC 6sdeROgR3 3| 6En
e'RE6&lR2 keem 'lO2 REp Ine6Rm klook3 3| deeg'Rii2\ dEd6n HuuR
sk'siie2 ImmeekI d'RE6&lR2 eek hIsklook 'O2 2| &SU6iknsIeR eROgR
&dEnddR3 3||
'dEn2 dnOR '6IN2 6luu2 Es'haR2 EshikU3 3| &6dmOR Hi6'luu2 2| dmORk 'loos&li2 &Id'RE6&lR2\ ool HIsklook RuN&Im3 3| &ENE'la2\ dnOR '6IN2 gee6
p dEm3 3|| 'dEn2 dsn On u3 3 6Rmli3 3| &ENI'mii&li2\ dRE6lR Uk
O3 3 hIsklook3 3|| En'soo2 dnOR '6IN2 6zO6li ukM'Es2| &dEdsn3 3 6zdeROgR3 3 &O6duu3 3||
&Iu'lik21 de'oo&Ri2| &uu6ON uHiRI 'geen21|||)
Hindi text
10.4.2. Uttarii hawaa aor suura is baat par haga rahee thee ki ham doon
mqq zyaadaa balwaan kaon hae. Itnee mqq garam oogaa pahnee eek musaar udhar
aa niklaa. Hawaa aor suura doon is baat par razii hoo gayee ki doon mqq see
oo pahlee musaar kaa oogaa utarwaa deegaa wahii zyaadaa balwaan samhaa
aayeegaa.
Is par uttarii hawaa apnaa puuraa zoor lagaakar alnee lagii. Leekin wah aesee
aesee apnaa zoor bahaatii gayii waesee waesee musaar apnee badan par oogee koo
aor bhii zyaadaa kas kar lapee=taa gayaa. Ant mqq hawaa nee apnii kooi band kar
dii. Phir suura teezii kee saath niklaa aor musaar nee turant apnaa oogaa utaar
diyaa. Is liyee hawaa koo maannaa paaa ki un doon mqq suura hii zyaadaa balwaan hae.
Kyaa tumhqq yah kahaanii ahii lagii? Kyaa phir see sunaa?
10.4.3. (&UttRii H'6aa2 R suuR2 2| Is'baat2 &pRH'g R&Hethee2 2| ki'Hm
'don&me2| zjaada bl'6an2| k9nH3 3|| It'nee&me2| gxRm Coogaa &pHnee2 2|
'eek2 mUsaa&IR2 2| Ud'HR2\ aa nIklaa3 3|| H6aa2 2| R 'suu&R2| don2 2 Is'baat2
&pRRazii Hogxjee2 2|| kido&n mee&se2 2 &opHlee mU'saa&IR2| &kaCogaa3 3| UtR6a degaa2 2|| 6'Hii2| zjaada bl'6an2 sm'Haa aegaa3 3||
Is&pR2 2| &UttRii H'6aa2 pnaa puuRa 'zooR2 lgaakR Clnee lgii3 3|| 'lee&kIn2|
'6H2 se '&se2 &pnazooR2 2\ bHaati gxjii2 2| 6se '6&se2\ mU'saa&IR2|
pnee bdm 'pR2| Cooge 'koo2| R bHii zjaada 'ks&kR2| lpeeta gxjaa3 3||
nt 'mee2| H6ane pni koo&I2 2| bnd kRdii3 3|| p'hIR2| 'suu&R2 tezii kesaath
nIklaa3 3|| R mUsaaIR 'nee2| tURnt pna Cogaa3 3 UtaaR dIjaa3 3| IslI&e2 2| H6aa komanna p'aa2| &kiUndon mee2 2| suuR 'Hii2 zjaada bl6anH3 3||
kjaa tUm'Hee2| jh kx'Ha&ni2 C:hii l'gii21| kjaa phIRsee sU'naa&u21|||)
346
English pronunciation of Hindi

10.4.4. (&T5>I;i h'wA:2 5a 's>G32| s'bA;T2 &pA;G5gA: &>h'Th;I32| khIi5hm
'Dn&mIn2| zi5A;D b'wA:n2| 'khan&ha3 3|| T'nImIn2| g5>m c'gA: &ph'n;I32| 5Ik m'sA;f32| 'DA:2\ 5A: nk'lA:3 3|| h'wA:32| 5a 's>G2| 'Dnn32 s'bA;T2 &p>5zI;i &hg'j;I32|| khIi5Dnn 'mIn&sI32 &Gp5l;I m'sA;f2| &khc'gA:3 3| 5ThA;w DI'gA:32|| w'hI;i2| zi5A;D b'wA:n2 sm5GA: Ga'gA:3 3||
'spA;32| &T5>I;i h'wA:2 p5nA: 'ph> 'z:2 l5gA;k c'n;I l'gI;i3 3|| 'lIkn2|
'wA:2 5GasI 'GasI2 &pn'z:32\ bA;5hA;Ti g'jI;i32| 5wasI 'wasI2\ m'sA;f2| p5n;I b'Dm 'phA:2| 5chg 'kh;2| 5a 'bI;i zi'A;D 'khskA;2| l5phIT g'jA:3 3||
5nT 'm;In2| h5wA;nI 'pni 'khSS32| 5b kA;'DI;i3 3|| 'ph;2| 's>G2 TI5zI;i kI'sA;T
nk'lA:3 3|| 5a m'sA;f 'n;I2| 5Th>nT 'pnA; c'gA:3 3 5ThA: di'A:3 3| s'lIiI32| h5wA: k'mA;n p'>A:2| &khIin5Dnn 'm;In32| 5s>G 'hI;i2 zi5A;D b'wA:n&hI3 3||
5khjA: Tm'h;In2| 5jA: k'hA;ni2 5chI;i l'gI;i21| 5khjA: &ph's;I s'na;n21|||)
11. Chinese
11.0.1. We provide the modern neutral pronunciation of Mandarin Chinese
[ptohu /phu5thu6hwa/ (pu^thU,7wa), common language, based on Pekin{g}ese), by using the ocial transliteration [pinyin /5phin5jin/ (^pI~5jIn)), but
accompanied by a careful phonetic transcription (which is necessary to show and
acquire a good pronunciation) and by a phonemic transcription (established according to precise descriptive and teaching strategies).
Chinese has various (especially vocalic) taxophones and a syllabic structure with
denite phonotactic limitations. erefore, in the presentation of phonemes it is
always useful to deal with all the various possible syllables, especially when they
have peculiar characteristics.
We maintain the criterion that Chinese syllables have a consonantal initial before a final element, which in addition to a vocalic nucleus (/i, y, M, u e, X, o
a/) may have an intermediate consonantal element (/j, , w/ i u) a terminal one (which can be vocalic: /i, e u, o/; or consonantal: /n, , >/).
11.0.2. We will see that Chinese syllables always begin with a contoid, even
when they lack a real initial. In the case of close vowels, //i, y, u// (whereas /M/
only occurs after the initials /q[h], s/}, we actually have /ji, y, wu/ (recognized by
pinyin too: yi yu wu).
e other vowels, /e, X, o, a/, are preceded by () (or, but less advisably, by the
following voiced phones (y, H, r, ) respectively, velar semi-approximant, {lenis}
laryngeal approximant, uvular tap, and semi-provelar nasal).
11.0.3. Let us observe that the Chinese phonetic tradition still follows the
phonic fantasies of the Swedish sinologist Karlgren, who thought that Mandarin
had two retroex (or apical) vowels, rendered with q, Q. Actually, the former
is simply a back-central vocoid, (M) (where neither retroexion nor the lamina
are active); the latter, on the other hand, is precisely a postalveolar approximant
contoid found in the initial () />/ r changed into the corresponding intense
(syllabic) contoid, (). e same phonetic tradition also lists another rare
retroex (mid) vowel, rendered with & which in actual fact is just the sequence
(X) /X>/ er It seems surprising, to say the least, that the Journal of the International Phonetic Association has accepted and published (in 2003), as an illustration of
the IPA, a paper which still proposes q, Q although ipa-like masqueraded as ()
in both cases! by describing them as (apico-)laminal denti-alveolar (where no
groove is rightly mentioned) and apical post-alveolar approximants respectively.
e latter is correctly described ( can()), while the former is still not ( can());
there is a great (articulatory, auditory, and acoustic) dierence between () and
(M).
348
Continuing our review of the peculiarities of this kind of phonetics la chinoise (done in China) and based on a Karlgrenized pinyin, we will then see a
real phonetic transcription (whose appropriate symbols we will present here in
square brackets, ( )). In the West, some things can be rectied, but generally others are messed up erefore, Mandarin would have seven vowels: rst, i (i, I, )
/i/, but with two additional variants incorrectly placed among front vowels (because of the grapheme i] q (M) /M/, in zi (qM) /qM/, ci (qhM) /qhM/, si (sM)
/sM/, and Q () //, in i () //, >i (h) /h/, i () //, ri () />/; however, they are true phonemes, not variants. en, (y, Y) /y/ and u (u, U) /u/,
with the additional variant (y, Y) /y/, in yu ju qu xu where the only actual variant would be just the grapheme: u instead of after palatal consonantal graphemes, since as a phoneme /y/ is always /y/, irrespective of the way it is spelled.
11.0.4. Continuing, we have e (X, , , x, ) /X/, with its variant () /e/, in contact with /j, , i/, and () in /ae/. But it is decidedly safer to consider /X/ and /e/ as
two separate phonemes, given the great phonetic dierence, because of interphonemic reasons (even if, from an intraphonemic point of view, just within the
Chinese language, they could or should be considered as taxophones of the
same phoneme). In addition, o (, ) /o/, with its variant (U) /u/ which, on the
contrary, is a dierent phoneme. Instead, o in contact with labials [mo bo po
fo ou wo/Cuo] could be considered as a variant of /X/, as many Chinese phonologists do, by using (X, ). However, the occurrence of the syllables o yo lo (although rare) requires that a phoneme /o/ is introduced, which will obviously occur near labials as well. en we have a (a, , A) /a/, with its variants (jEn, n),
for ian an
nally, this phonetics la chinoise even has a zero (graphic) vowel, when
pinyin smartly economizes on vowels, by writing un for /wXn/ (wn), and iu for
/jou/ (jPU), ui for /wei/ (wI), where a vowel is denitely present, phonically! No
doubt, uen iou uei would have been much better. anks to the use of diaphonemes, then, more realistically we will also have the diphthongs /ae, ao/ (, A),
instead of a more abstract representation as /ai, au/: ai ao (pinyin stays on the
fence: -i -o)!
11.0.5. We will now present the ocial romanization. Starting from the (consonantal) initials, which are 23 (with a double occurrence of y with two dierent
values), we will not present them alphabetically, but by logical phonic groups, accompanied by their phonemic and simple phonetic transcriptions. e latter is
simple, because it shows only one variant (whereas, below, we will provide all
necessary details):
b /p/ (p), p /ph/ (ph), m /m/ (m), f /f/ (f)
d /t/ (t), t /th/ (th), n /n/ (n), z /q/ (q), c /qh/ (qh), s /s/ (s), l /l/ (l)
// (), > /h/ (h), // (), r />/ ()
j /q/ (), q /qh/ (h), x /s/ ()
g /k/ (k), k /kh/ (kh), h /h/ ()
w /w/ (w), y /j/ (j) // ().
11. chinese
349
11.0.6. e nals are 39 (with a triple occurrence of i with three dierent values). ey are presented according to their graphic syllabic nucleus. is time,
they are given in alphabetical order, even in the seven cases where the romanization has not been a happy one (as we will show in > ). Also the exclamations e
/e/ (), yo /jo/ (j) occur (with various tonemes, including zero). e three nals
which do not occur without an initial are marked with an asterisk (*):
a /a/ (a), ia ya /ja/ (ja), ua wa /wa/ (wa)

ai /ae/ (), uai wai /wae/ (wa) yai /jae/ (sole word, yi)
an /an/ (n), ian yan /jan/ (jEn), uan wan /wan/ (wn), an yuan (j/q/x +
uan) /an/ (n)
a /a/ (a,), ia ya /ja/ (ja,), ua wa /wa/ (wA,)
ao /ao/ (A), iao yao /jao/ (ja).
e /X/ (X, Xx|, x|), ie ye /je/ (j), e yue (j/q/x + ue) /e/ ()
ei* /ei/ (I) {(I)}, en /Xn/ (n), er /X>/ (X)
e* /X/ (x,), ue we /wX/ (w,) {(w,)}.
i /i, ji/ (i, ji), (s-/z-/c-) /M/ (M) > (with ], (-/-/>-) // () >
in yin /in/ (In, jIn) i yi /i/ (, j)
ui wei /wei/ (wI) {(wi)} > uei .
o /o/ (, x|), ou /ou/ (U) {(U)}, uo wo /wo/ (w, wx|) {(w, wx)}
o* /u/ (U,) > u , io yo /ju/ (jU,) {(U,)} > iu yu .
u wu /u, wu/ (u, wu) yu (j/q/x + u) /y, y/ (y, y)

un wen /wXn/ (wn) > uen ; n yun (j/q/x + un) /yn, yn/ ([]Yn) {([]In,
[j]Yn)}
iu you /jou/ (jPU) {(ju)} > iou .
Vowels
11.1.1. In our analysis, Chinese has eight vowel phonemes: /i, y, M, u e, X, o
a/ ( 11.1). Instead, 11.2.- show the thirteen taxophones ( contextual allophones grey markers), which are very important for a good neutral pronunciation of Mandarin Chinese.
On the other hand, 11.8 presents the most recommendable realizations for
an international pronunciation, which is slightly simplied, but not distorted.
Of course, this international pronunciation will not use the realizations shown
in 11.2-7, which however are very important, because the foreign speakers who
can recognize them will be greatly helped in better understanding spoken Chinese
(though they need not be able to use actively those shown in 11.6-7).
Let us start then, by systematically considering each phoneme, with all the taxo-
350
phones belonging to neutral pronunciation (indicating some possible phonetic

variant as well, again more to for recognition than for actual use, 11.4).
11.1.2. We rst nd the close front phoneme, /i/: /[j]i, [j]in, [j]i/ ([j]i, [j]In,
[j],) (C/y)i (C/y)in (C/y)i\ (5ji) /5ji/ yi ([~ii) /ni/ n` (4jIn) /jin/ yn (5In)
/5qin/ jin (7j) /6ji/ y (5,) /5qi/ ji
e close front-central rounded, /y/, has a fairly limited distribution, ju qu xu
yu n l (with or without -n), /[]y, []yn/ (y, Yn) [variants with -n (In, jYn)]:
(5y) /5qy/ ju ([yy) /qhy/ q (7y) /6sy/ x ([yy) /y/ y (5Yn, 5In)
/5qyn/ jun (4Yn, 4In) /qhyn/ qn (7Yc, 7Ic) /6syn/ xn (4Yn, 4In,
4jYn) /yn/ yn
More conveniently, pinyin could have used z c s y, by phonemically interpreting these syllables as we do: (y, hy, y) /qy, qhy, sy/. It could also have
avoided introducing an allograph without dieresis, by using y for ocial yu (y)
/y/ as well. Indeed, /y/ could even be /jy/; but since the phoneme // is present,
it is more adequate to use it. In this way, as it seems convenient, it could also be
used to show even phonemically the possible oscillations between (Yn, jYn,
In): /yn/, and /jyn, in/). erefore, it could have been possible to write
z/c/siu, instead of the ocial j/q/xio for (jU,, hjU,, jU,) /qju, qhju,
sju/ (thus also -u instead of the ocial -o for (0U,) /0u/).
11.1.3. en we nd the close back-central (without lip rounding!), /M/, which
only occurs in /qM, qhM, sM/ (qM, qhM, sM) zi ci si\ (7q) /6qM/ z (4qM)
/qhM/ c (5sM) /5sM/ si (but we could extend its use to a sequence like /M>/, in
place of //, for /, h, , >/ (, h, , ) i >i i ri\ ([) // `
(5) /5h/ >i (4) // (7) /6>/ r]
e last close phoneme is back rounded, /u/ (u, U), and we are obliged to use
two dierent graphemes, because of the nature of pinyinization: u o (instead of
a more logical u]\ ([wuu) /wu/ w (4lu) /lu/ l (5kU,) /5ku/ go (7U) /6u/
11.1. Mandarin vowels.

/[j]i/ ([j]i) (y)i
/[]y/ ([]y) (l/n) yu ( j/q/x + u]
/je/ (j) ie ye
/e/ () (l/n)e yue ( j/q/x + ue]
/M/ (M) (z/c/s)i

/[w]u/ ([w]u) (w)u
/X/ (X, Xx|) ()e /X>/ (X) ()er
/[w]o/ ([w], [w]x|) (u)o (w)o
/a/ (a) ()a /ja/ (ja) ia ya /wa/ (wa) ua wa
11.1.4. e inventory of the vowels of neutral Chinese continues with three

mid vowels. We always nd /e/ () e in the sequences /je, e/ (j, ) ye/Cie
yue/Ce\ (4j) /je/ y (5tj) /5tje/ die ([) /se/ xu (7) /6e/ yu.
en, /X/, with the following distribution /X, X>, [w]Xn, [w]X/ (X[x|], X, n,
11. chinese
351
wn, x,, w,) e er en un/wen e ue/we. e last one has the variant (w,)
(which, if necessary, could be indicated as /wo/): (7$[*|]) /6X/ (5n) /5Xn/ en
(4X) /X>/ r (7fc) /6fXn/ fn ([wc) /wXn/ wn (7f*) /6fX/ f (5w, 5w,)
/5wX/ we (7w 7w) /6wX/ w
nally, we have /o/ ([x|]) o which can occur in the syllables /o, jo/ ([x|], j[x|]),
in addition to normal /mo, po, pho, [0]wo/ ([[x|], ([x|], (h[x|], [0]w[x|]), mo
bo po wo/Cuo\ (5[[x|]) /5mo/ mo (4([x|]) /po/ b (7([*|]) /6pho/ p ([w,
[w*|) /wo/ w (5w) /5wo/ uo (A,5j) /ha5jo/ hyo (7[*|]) /6o/ ! en
nal, before a pause, /X, o/ normally have the accompaniment (x), as can be seen from
various examples. Instead of /mo, po, pho/, we could even pose /mwo, pwo, phwo/,
for ([[x|], ([x|], (h[x|]), while maintaining /mou, phou/ (mU, phU) (/pou/
does not occur), by dissimilation.
11.1.5. Lastly, we nd the open vowel, /a/, with its distributions and various
taxophones, /[0]a, jan, an, [w/0]an, [j]a, wa/ (a, jEn, n, [w/0]n, [j]a,, wA,):
(5a&a) /5aqa/ aza (7ta) /6tha/ t (4jEn) /jan/ yn ([tjEEc) /tjan/ din (7c)
/6an/ yun (5n) /5qhan/ quan ([c) /an/ n (4pn) /phan/ pn (4pa,)
/pha/ p ([wc) /wan/ wn (5ja,) /5ja/ ya ([wAA) /wa/ w
11.2.. Important taxophones.
/[j]in/ ([j]In) (y)in
/jan/ (jEn) ian yan

/an/ (n) yuan
( + j/q/x + un, lan]
/[]yn/ ([]Yn) {([]In), [j]Yn)} yun

( j/q/x + un, ln)
/Xn/ (n) ()en
/wXn/ (wn) un wen
/[w]an/ ([w]n) (u)an (w)an
11.2.. Further important taxophones.
/[j]i/ ([j],) (y)ing

/X/ (x,) eng /X|/ (x) e
/[j]a/ ([j]a,) (i)ang (y)ang
/u/ (U,) ong

/ju/ (jU) {(U,)} iong yong
{/[w]o/ ([w], [w]x|) (u)o (w)o]
/wX/ (w,) {(w,)} ueng weng
/wa/ (wA,) uang wang
Diphthongs
11.1.6. Moving to diphthongs ( 11.3), we have /[w]ei, [w]ae, [j]ao, [j]ou/ (I,
wI, , wa, ja, A, U, jPU) ei ui (u)ai (i)ao ou iu for /[w]ei, [j]ou/ there are
also the possible variants (shown in 11.4, which are not necessary for a good
pronunciation, and will be given only here), (I, wi, U, ju): (5fI 5fI) /5fei/ fei
(5twI 5twi) /5twei/ dui ([) /ae/ i (7wa) /6wae/ wi ([AA) /ao/ o ([AA)
352
/hao/ ho (4ja) /jao/ yo (5U 5U) /5ou/ Ou ([mU [mU) /mou/ mu

(5tjPU 5tju) /5tjou/ diu
11.3. Mandarin diphthongs.
/wei/ (wI) ui wei

/ei/ (I) ei
/ae/ () ()ai
/wae/ (wa) uai wai
/jou/ (jPU) iu you

/ou/ (U) ()ou
/jao/ (ja) iao yao
/ao/ (A) ()ao
11.4. Possible diphthong variants.

{/wei/ (wi) ui wei]
{/ei/ (I) ei]
{/jou/ (ju) iu you]

{/ou/ (U) ()ou]
Vowels with nal in />/ () -r

11.1.7. is characteristic is typical of Peking (Beijing). It enjoys prestige, but
is not completely accepted into neutral language, ptohu, although it is regularly indicated in dictionaries and grammars. Actually, the Pekinese themselves often
try to avoid it; whereas the other Chinese (whether Mandarin or non-Mandarin)
have not fully master of its use, uctuating considerably.
However, its true and genuine distribution shows very substantial neutralizations, since the segments that actually occur (besides //, preceded by /[h]-, -, >-/:
(, h, , ) i >i i ri, already seen in 11.1.3) are /X>, X>, a>, a>/ (X,
, , ), as shown in 11.5 (which includes (X) /X>/ er 1.4). ey are represented in pinyin either by e(n)r er a(n)r ar or by the original syllables with
the addition of -r although their pronunciation is reduced to the four already seen,
which can be preceded by the middle consonants /j, , w/, often deriving from
original /i, y, u/. Let us see at once the four types: (5kX) /5kX>/ ger ([p$$Y) /pX>/
bnr (4) /hX>/ hr ([nY) /na>/ nr (7pY) /6pa>/ bnr (5q) /5qa>/
zar (5w) /5hwa>/ huar (7mjY) /6mja>/ minr
We will now consider those with etymological writing, but which are always realized as /X>, X>, a>, a>/ (X, , , ), preceded or not by /j, , w/ (sometimes /, w/ correspond to and u/o]\ (4p) /pha>/ pir (7$Y) /6qhX>/ qr
(7w$Y) /6wX>/ r (7q$Y) /6qhX>/ cr ([q$$Y) /qX>/ z`r (5kw) /5kwa>/ guanr
(5kw) /5kwX>/ gor (5swn&~X, -~&~-) /5swXnnX>/ sunnr (4i&fwX) /sifwX>/
xfur (^4ljX) /5ljX>/ ilir (4jX) /qjX>/ jir (5a2b[tw$$Y) /5hapathwX>/
habatu`r (7tUj<Y) /6tuqjX>/ djir (7j$Y) /6jX>/ ynr ([s$$Y) /sX>/ s`r (5X)
11. chinese
353
/5hX>/ heir (5tj) /5thja>/ tianr (4mj) /mja>/ mior (7qw$Y) /6qhwX>/ cur
(4twX) /thwX>/ tur (^m5pwX) /5an5phwX>/ anpor
11.5. Vowels with />/ () -r.
/X>/ (X) -e(n)r

/X>/ () -engr
/a>/ () -a(n)r
/a>/ () -angr
11.1.8. Ocial orthography separates the forms which in Chinese characters

are normal plus r erefore, non-Mandarin people tend to pronounce the two
syllables separated and with their tones (as in a sort of international Mandarin).
In addition, many non-Pekinese Mandarin people join the two syllables into one,
but with less neutralizations. Hence, besides the four genuine syllables (with or
without middle consonants), we can also nd the following ones, excluding those
with front close vowels, /i, y/ (i, I, y, Y), that articulatorily do not adapt to (),
which is postalveolar. us, we mark them with an asterisk, to advise that, although it is possible to hear them pronounced in that way by Chinese people, they
do not fall within neutral pronunciation: *(7uY) /6wX>/ r *(5kU 5k)
/5kwX>/ gor *(4i&fu) /sifwX>/ xfur *(^4ljPU -4lju) /5ljX>/ ilir
*(4j 4j) /qjX>/ jir *(5ab[twY) /5hapathwX>/ habatu`r *(5 5)
/5hX>/ heir *(4mja) /mja>/ mior *(7qwY 7qwY) /6qhwX>/ cur *(4tU
4tU) /thwX>/ tur *(^m5p) /5an5phwX>/ anpor In any case, of course,
these pronunciations are not to be imitated.
Vowel reductions
11.1.9. In syllables with neuter/zero tonemes ( 11.3.3.4), some timbric reductions occur (together with shortenings), mainly in dierent positions from
word-nal ones, or in intonemes. As a matter of fact, in actual language, apart
from in slow and accurate speech, some neutralizations occur too. It is fundamental to know them, at least to understand spoken Chinese (well). ey need not be
used actively, although an appropriate use of them decidedly improves pronunciation. In order to adequately show the behavior of dierent unstressed syllables (for
vowel, consonant, and pitch reductions), sometimes we also had to use rare words,
rather than omitting something signicant.
11.6 shows the most important reductions, which actually occur in non-tonic
syllables ( in the real sense of deprived of any of the four basic tonemes,
11.3.3.1-3). e examples given are to be interpreted as in a sort of decreasing scale,
starting from a prepausal position, in an intoneme (for the rst variants), up to
modications produced within a sentence, in a preintoneme (for the others). As
354
can be seen, () applies to /i[n/], y[n]/: (7tii), (-3d) /6titi/ ddi (5kw~&i), (-3B)
/5kwansi/ guanxi (7jay), (-3i, -3) /6sjaqy/ xiqu and (), to /M[>], u[]/: (7ji), (-3z) /6jisM/ ysi (7), (-3) /6>Xn/ rni (5u&u), (-v) /5ufu/ ufu
11.6. Reductions in syllables with zero tones (in quick speech, not in an intoneme; whereas, in an intoneme, the distinct timbers of 11.1-3 are used).
/i, y/ ()
/Xn, e/ (), /[w]ei/ ([j])
/a[n]/ ([n]), /[w]ae/ ([j])
/M, u/ ()
/X[], o/ (X), /[j]ou/ ([]X)
/a/ (,), /[j]ao/ ([])
11.1.10. In addition, we have (), for /[j/]e, [w]Xn/, and () for /[w]ei/: ([j^), (-1, -1) /qjeqje/ jijie (7c), (-3c, -3c) /6ae>Xn/ iren (7mImI), (-3m) /6meimei/ mimei then (X), for /X, [w]X, [w]o/, and ([j]X), for
/[j]ou/: (5kX&X), (-2gX) /5kXkX/ gege (t^2X), (-1"X) /tiX/ dè (5jEN&x,),
(-2X,) /5sjanX/ xiane (7jm$), (-3m$) /6qjemo/ jimo (nw^j), (-1hjX)
/nwanhwo/ nunhuo (7UU), (-3d$) /6utou/ ytou (4px,&jU), (-2X)
/phXjou/ pyou
Lastly, we have: () in /a[n], [w]a, [w]ae/: (7paa), (-3b) /6papa/ bba (5ma&ma),
(-2m) /5mama/ mama (i^wan), (-1hjn) /sihwan/ x`huan (7ttha), (-3t)
/6thaethae/ titai and () in /a, [j]ao/: ([wâ,), (-1,) /wana/ wna
(5&A), (-2d) /5tao/ idao
11.7. Further reductions in syllables with zero tones (in even quicker speech, not in an
intoneme; then, in an intoneme, the realizations of 11.6 are also possible).
/i, y, M[>], u/ ()
/ae/ (), /ja/ (j)
/ou, uo, ao/ ()

/X, -X- a, -a-/ ()
11.1.11. In 11.7 other attenuations are shown, or neutralizations too, which

are possible in even faster but still normal and genuine speech.
We nd () for /i, y, M, u >/ (z/c/s />/)i u\ (7tii), (-3d, -3d^) /6titi/ ddi
(5kw~&i), (-2B, -2B^) /5kwansi/ guanxi (7jay), (-3i, -3, -3^) /6sjaqy/ xiqu
(7ji), (-3z, -3z^) /6jisM/ ysi (7), (-3, -3^) /6>Xn/ rni (5u&u), (-2v, -2v^)
/5ufu/ ufu and () for /ou, uo, ao/: (nw^w), (-1hjX, -1h) /nwanhwo/ nunhuo (7UU), (-3d$, -3d) /6utou/ ytou (4px,&jU), (-2X, -2) /phXjou/
pyou (5&A), (-2d, -2d) /5tao/ idao
11. chinese
355
Besides, () in /ae, ja/ (, ): (7tth), (-3t, -3t) /6thaethae/ titai (4~&a), (-2, -2) />Xnqja/ rnjia and () for /X/ an for any other /a/ a: (5kX&X),
(-2g) /5kXkX/ gege (t^X), (-1) /tiX/ dè (7*c), (-3c) /6ae>Xn/ iren
(5jEN&x,), (-2,) /5sjanX/ xiane (7paa), (-3b, -3b) /6papa/ bba (5mama),
(-3m, -3m) /5mama/ mama (i^wan), (-1hjn, -1hjn, -1hn) /sihwan/ x`huan
(wâ,), (-1,, -1,) /wana/ wna
International Mandarin vowels
11.1.12. 11.8 shows the vowel realizations that are sucient for an international pronunciation of Chinese, without taxophones (thus, without either attenuations or neutralizations such as those given in 11.6-7; equally, we do not even
mark creaky voice, which is superuous in an international kind of pronunciation). So we have only eight monophthongs (and four diphthongs, obtained by
combining ve of the eight basic ones). ey are a satisfactory result, once (M, x)
have been learned.
erefore, we have: (5ji) /5ji/ yi ([~ii) /ni/ n` (jin) /jin/ yn (5in) /5qin/ jin
(7ji) /6ji/ y (5i) /5qi/ ji
en: (5y) /5qy/ ju ([hyy) /qhy/ q (7y) /6sy/ x ([yy) /y/ y (5yn)
/5qyn/ jun (hyn) /qhyn/ qn (7yn) /6syn/ xn (yn) /yn/ yn
Besides: ([wuu) /wu/ w (lu) /lu/ l (5ku) /5ku/ go (7u) /6u/
(j) /je/ y (5tj) /5tje/ die (7) /6e/ yu ([) /se/ xu and (7x) /6X/ (5xn)
/5Xn/ en (X) /X>/ r (7fxn) /6fXn/ fn ([wxxn) /wXn/ wn (7fx) /6fX/ f (5wx)
/5wX/ we (7wx) /6wX/ w
More: (5m) /5mo/ mo (p) /po/ b (7ph) /6pho/ p ([w) /wo/ w (5w)
/5wo/ uo (ha5j) /ha5jo/ hyo (7) /6o/ ! and (5a&qa) /5aqa/ aza (7tha)
/6tha/ t (jan) /jan/ yn ([tjaan) /tjan/ din (7an) /6an/ yun (5han)
/5qhan/ quan ([aan) /an/ n (phan) /phan/ pn (pha) /pha/ p ([waan)
/wan/ wn (5ja) /5ja/ ya ([waa) /wa/ w
11.1.13. And with />/: ([) // ` (5h) /5h/ >i () // (7) /6>/
r (5kx) /5kX>/ ger ([pxx) /pX>/ bnr (h) /hX>/ hr ([naa) /na>/ nr
(7pa) /6pa>/ bnr (5q) /5qa>/ zar (5hwa) /5hwa>/ huar (7mja) /6mja>/
minr and (pha) /pha>/ pir (7hx) /6qhX>/ qr (7wx) /6wX>/ r
(7qhx) /6qhX>/ cr ([qxx) /qX>/ z`r (5kwa) /5kwa>/ guanr (5kw) /5kwX>/
gor (5swxn&nx) /5swXnnX>/ sunnr (i&fwx) /sifwX>/ xfur (^ljx) /5ljX>/ ilir (jx) /qjX>/ jir (^ha2pa[thwxx) /5hapathwX>/ habatu`r (7tuj) /6tuqjX>/ djir (7jx) /6jX>/ ynr ([sxx) /sX>/ s`r (5hx) /5hX>/ heir
(5thja) /5thja>/ tianr (mja) /mja>/ mior (7qhwx) /6qhwX>/ cur (thwx)
/thwX>/ tur (âm5phwx, -n5p-) /5an5phwX>/ anpor
For diphthongs, we have: (5fi) /5fei/ fei (5twi) /5twei/ dui ([aa) /ae/ i
(7wa) /6wae/ wi ([aa) /ao/ o ([haa) /hao/ ho (ja) /jao/ yo (5u) /5ou/
Ou ([mu) /mou/ mu (5tju) /5tjou/ diu
356
11.1.14. In a type of international pronunciation, even vowel reductions are

not so necessary; thus (fairly safely) we have: (7ti3ti) /6titi/ ddi (5kwa~&i, -n&i)
/5kwansi/ guanxi (7jay) /6sjaqy/ xiqu and (7jisM) /6jisM/ ysi (7xN, -n)
/6>Xn/ rni (5u&fu) /5ufu/ ufu ([j^j) /qjeqje/ jijie (7axn) /6ae>Xn/
iren (7mimi) /6meimei/ mimei.
In addition: (5kx&kx) /5kXkX/ gege (ti^x) /tiX/ dè (5jaN&x, -n&x)
/5sjanX/ xiane (7jm) /6qjemo/ jimo (nwa^hw, nwan-) /nwanhwo/
nunhuo (7utu) /6utou/ ytou (phx&ju) /phXjou/ pyou also (7papa) /6papa/ bba (5ma&ma) /5mama/ mama (i^hwan) /sihwan/ x`huan (7thatha) /6thaethae/ titai and (waNâ, -nâ) /wana/ wna (5&ta) /5tao/ idao
11.8. International Mandarin Chinese vowels.
/i[n/]/ ([j]i[n/])
/y[n]/ ([]y, []yn)
/je/ (j), /e/ (), /[w]ei/ ([w]i)
/M/ (M)
/u/ ([w]u), /[j]u/ ([j]u)
/X/ (x), /[w]Xn/ ([w]xn), /[w]X/ ([w]x)
/[w]o/ ([w]), /[j]ou/ ([j]u)
/[j//w]a[n/], [w]ae, [j]ao/

([j//w]a[n/], [w]a, [j]a)
Consonants
11.2.0. e consonant system of (Mandarin) Chinese has some interesting peculiarities, as we will see, including some alternative possibilities of phonemic analysis. e most important one regards the choice on the phonemic status of (,
h, ); in fact, pinyin has chosen their ocial acknowledgement as j q x to
the detriment of vowels, which are rendered as i u after (, h, ), with the following values: /i, y j, /. On the contrary, it would have been sucient to systematically use (instead of uctuating between yu ju qu xu and n l] that is
always y j q x n l (all the more so because with n l both a dieresis and
tone-accents must be used). Besides, z c s (M, hM, M) would have been useful as well, by producing even greater coherence (which is always useful). Consequently, also > r ought to have been used (instead of the pinyin ocial
i >i i ri] with the value of (, h, , ).
At this point, having seen the vowel phonemic inventory (which we have established, 11.1.1-5), it is obvious that our analysis prefers to consider (, h, )
as the result of the assimilation of /q, qh, s/ in contact with /i, y j, /. All this, of
course, is irrespective of morphological considerations, which are lexically derived,
since they are not at all absolute (rather, they are quite capricious indeed). us,
even for native speakers, the ocial choices have only limited (or seeming) advantages; therefore, we are convinced that the possibility of considering (, h, ) as
realizations of /k, kh, h/ is much less preferable even though sequences like */ki,
khi, hi, ky, khy, hy/ do not occur, which would allow to speak of complementary
11. chinese
357
distribution. Phonology, in order to be useful, must be explicit and not at all esoteric (although this would not be wrong diachronically).
In the table of 11.9, which gives the consonantal articulations of Mandarin
Chinese, to save space, we show as /05/ (whereas here we use /0[h]/ = /0, 0h/)
the consonants which are distinct phonemically. In fact, simple consonants oppose
aspiration sequences ( Stops, 11.2.2, and Stopstrictives, 11.2.3).
Instead, 1.9-15 show the orograms, grouped by manners of articulation, of
all the contoids given in the chapters of this handbook for the 12 languages dealt
with, including secondary, occasional, or regional variants.
m
([) (M)
p5 (b) ((5 {)
(n) n
(N)
t5 (d)
q5 (Q)
5 (")
(~)*
(5 )
()
k5 (g)
laryngeal
uvular
velar rounded
velar
postpalatal rounded
palatal
bilabialized prepalatal (* prepalat.)
postalveolar
alveolar
dental
labiodental round.
labiodental
bilabial rounded
bilabial
11.9. Table of Chinese consonants.
f (v) (5 )
()
()
s (z)
()
( B)
j|() |(%) ()
l
// (,), />/ (), /h/ (, h, , )
w|(j) ()
Nasals
11.2.1.1. ere are three nasal phonemes, /m, n, / (m, [ M, n, N, ~, ,) m
n In syllable-nal position only /n, / occur; and the latter always maintains
its uvular articulation ((,)) even before any consonants. Instead, /n/ regularly assimilates (except in very slow and accurate pronunciation), even if assimilation is
only partial for /nj, n, nw/ (~j, ~, w) (instead of *(Nj, N/, w) which are
theoretical). However, let us notice the reduced forms of wmen and tamen, which
become monosyllabic, with /m/ ( 11.3.1.2). e tendency of /n/ to assimilation is so strong that we can nd cases such as: (j5jn2lX&ma31, -jl2lX-) /qje5hwXnlXma?/ Ji hun le ma? Mandarin has /n/ (~) + /i, y j, / (although Pekinese
presents instead (n), which can be used fairly safely). On the other hand, we need
not use a more specic symbol, (()), before dentals /t, th, q, qh, s/. On the contrary, it is useful to choose ([) in /mo/ (which could be rendered as /mwo/ ([[x]),
whereas /mou/ is (mU), by dissimilation). Syllable- and word-nal nasals always
have an inaudible release, even before a pause: ((n, ,)). (is is quite normal for
other oriental languages such as Cantonese, ai, Cambodian, cetnamese,
Tagalog, Malay, and Indonesian even for nal voiceless stops, which Mandarin
has lost.)
358
11.2.1.2. Here are some examples: (5mA) /5mao/ mao (5[) /5mo/ mo
([nAA) /nao/ no (4pn) /phXn/ pn (4px,) /phX/ p (c7$[*|]) /hXn6X/
hn (n[tii) /qhanthi/ qunt` (Ic[qAA) /qinqao/ j`nzo (0twn7ljEc) /6twan6ljan/ dunlin (m4mIn) />Xnmin/ rnmn (lwm5pn) /lwXn5pan/ lnban (0ka[faa) /6khanfa/ knf (^pjE~4i) /5pjanqi/ bianj (0I[)
/6qin/ jn` (0p4) /6phanqe/ pnju (pjE4y) /pjany/ biny
(k[c) /kan>an/ gnrn (pI[kwc) /5pinkwan/ bingun (0jE7wA)
/6qjan6wa/ jinw (ljE,5wn) /ljan5hwan/ linhuan
th //: (ji7lu p,5n) /ji6lu phi5an/ yl p'an (^fx,7mjEc) /5fX6mjan/
femin (kU7pu) /khu6pu/ kb (0la,7fI) /6la6fei/ lfi (^fx,[tii) /5fXti/ fed` (kU7i) /khu6si/ kx (p,4y) /phiy/ py (^x,7) /5X6>/ er (^fx,7c) /5fX6an/ fen (p,[kw, -[kw*|) /phikwo/ pgu
In an international-like pronunciation, it is sucient to use (), instead of (,),
and also (n) before /i, y j, / (and, possibly, before other consonants too]
Stops
11.2.2.1. We have three voiceless phonemes, /p, t, k/, which occur as either
plain, /0/, or in aspiration sequences, /0h/, with distinctive values. Before seeing some examples, we have to explain the nature and consistency of both voicelessness and aspiration.
In fact, except in slow (or very accurate) speech, /p, t, k/, are realized as (p,
t, k) only in fully stressed syllables (or after pauses), whereas in half-stressed syllables they are (, , ) or, in unstressed syllables, even (b, d, g). On the other hand
in a parallel way /ph, th, kh/, in fully stressed syllables (or after pauses), are
sequences of voiceless stops and a laryngeal constrictive (decidedly strong, (p,
t, k) (however, it is possible to nd a semi-constrictive as well, ((0)), which is
weaker). In half-stressed syllables, we have sequences with a laryngeal approximant,
(ph, th, kh); whereas, in unstressed syllables, we can even nd plain voiceless stops,
(p, t, k). We choose to adopt this kind of normalized transcription (although oscillations are frequent indeed, including sequences with a voiceless laryngeal semi-approximant, ((0h))).
A degree of distinctivity between aspiration and non-aspiration, however, is
guaranteed. In fact, we have: /0h/ ('=, &=h, =), /0/ ('=, &, ), with strong and
weak extremes meeting, as (=). Although they coincide phonetically, they remain
distinct functionally, from a phonemic point of view. Before /o/ (not /ou/, by dissimilation) we nd ((), which sounds rather dierent from the common (p) of
other languages.
11.2.2.2. Let us now see the following examples: (5pIn) /5pin/ bin (4() /po/
bo (7() /6pho/ p (4tj) /tje/ di (4kwI) /khwei/ ku (^p,5pa,) /5phi5pha/ pipa (0tjA[wuu) /6thjaowu/ tiow (5kx,) /5khX/ ke (7paa|,
7pa3b) /6papa/ bba (7tii|, 7ti3dI) /6titi/ ddi (5kX&X|, 5kX2g) /5kXkX/ gege (7-
11. chinese
359
ph|, 73p) /6phae/ pai (4pu&thA|, 4pu2t) /phuthao/ ptao (4pu2t[jU -2t-) /phuthaoqjou/ ptaoji (7qkh$|, -3k) /6qhMkhX/ cke
As we have seen ( 11.0.2 11.2.1), there is also a laryngeal stop, which is important though not phonemic (with some less advisable alternative variants, (y, H,
r, )). It is useful to adequately maintain the articulatory identity of nal consonants (/n, , >/) before initial vowels.
Stop-strictives
11.2.3. As far as this manner (of articulation) is concerned, in our analysis (with
the same normalized characteristics of voicing and aspiration as for Chinese
stops) we have /q, qh/ ('q, &, Q 'q, &qh, q) z c in addition to (', &, ',
&h, ) j q before /i, y j, / ( 11.2.0), and /, h/ (', &2, " ', &h, )
>
Here are some examples (expressly chosen with nal syllables without tonemes):
(5q) /5qae/ zai (7qwc) /6qhwan/ cun (5y) /5qy/ ju (4U, 4jU,)
/qu/ qi (7I) /6ei/ i (4) // > (4pi&M|, 4pi2Q) /piqM/ bzi
(4Yn&qh|, 4Yn2q 4jYn-) /ynqhae/ yncai (7tjEi|, 7tjE2) /6tjanqi/ dinji (5tjE~&hi|, 5tjE~) /5thjanqhi/ tianqi (4q&2|, 4q2") /qhae/ cii (4pI&hn|, 4pI2n) /pheihXn/ pi>en
Constrictives
11.2.4. ere are three constrictive phonemes. Again they are all phonemically
voiceless, but with the usual voicings which depend on syllable-strength attenuation. us we have: /f/ ('f, &, v) f, /s/ ('s, &, z) s and (', &, B) x, // (', &, ) \
(4fu) /fu/ f (4ma&n|, 4mavn) /mafan/ mfan (7swI 7swi) /6swei/ su (5fx,&u|,
5fx,2z) /5fXsu/ fesu (4&i|, 42B) /sesi/ xuxi (5u) /5u/ u (4lja,&|,
4lja,2) /lja/ lii Before /o/ (not /ou/) we nd (5) (as happens for other labial phonemes): (45{x}) /fo/ f erefore, we could even pose /fwo/ instead of /fo/
(but /fou/ (fU), by dissimilation).
Approximants
11.2.5.1. Chinese has ve approximant phonemes in our analysis. e rst is
postalveolar, />/ (): (4n) />Xn/ rn (4wc) />wXn/ rn (4X) /X>/ r ([) //
` (British English />/ (>) is laterally contracted and has a certain amount of lip
rounding, thus is dierent English ('>a;) /'<aO/ row dispute and Chinese
(7A) /'>ao/ ro coil.)
Besides, we have: /j/ (j, ), // (, %), /w/ (w, j), with attenuated variants in
weak syllables, that is unstressed, because with zero toneme (provided it is not
before a pause), respectively: semi-palatal (), semi-postpalatal rounded (%), and se-
360
mi-velar rounded (j): (5ja,) /5ja/ ya (4tj) /tje/ di (5j) /5qje/ jie (li^jEn|,
li1bEn, -1bn) /lipjan/ l`bian ([yy) /y/ y (7) /6qhe/ qu (7tiU|, 7ti3B%U) /6tisu/ dxio (4wA,) /wa/ w (5wI 5wi) /5hwei/ hui (tA^lwn|, tA1ljn) /thaolwXn/ tolun
nally, we nd /h/ (', &, h) (in addition to ('0, &0h, 0`), already seen for
stops and stop-constrictives): ([AA) /hao/ ho (5&A|, 5n2h) /5Xnhao/ enhao In theory, we could have two dierent phonemes in the inventory of consonants: /x, h/, with /x/ = (', &, h); but it is better to have only one, with
the taxophones already seen (which are dierent although auditorily quite similar, but with a phone of contact, (h)). Traditionally, the Chinese prefer to consider initial consonants in a unitary way, with /05/ dierent from /0/, intraphonemically; but this choice does not present any real advantage from an interphonemic point of view, which is interested in several languages of the world quite the
contrary!
International Mandarin consonants
11.2.5.2. In a kind of international pronunciation (as for vowels, 11.1.1214), it is sucient to always use the basic forms, independently of stress: (j, , w),
even for (, %, j), and (h) for /h/, without variations, also in /0h/ (0h) (against /0/
(0)), and again without creaky voice: (5ja) /5ja/ ya (li^pjan) /lipjan/ l`bian
([yy) /y/ y (7tiu) /6tisu/ dxio (4wa) /wa/ w (5hwi) /5hwei/
hui (tha^lwxn) /thaolwXn/ tolun However, it is important to be able to hear
dierent realizations, in order to understand spoken Chinese (better). Obviously,
to achieve a more genuine pronunciation, it would be decidedly better to appropriately acquire and use all the characteristics we describe.
Laterals
11.2.6. For the last manner (of articulation) necessary for Mandarin, we nd
just one phoneme, /l/ (l) (which remains unchanged even before /i, y j, /): (4l)
/lae/ li (4lU,) /lu/ l ([ljEEc) /ljan/ lin (7ly) /6ly/ l
Structures
11.3.0. We will deal, in particular, with the typical tones (tonemes as distinctive
elements), including a zero ton(em)e and their combinatory variants (taxotones),
in addition to degrees of stress, and nally intonation, which overlaps tone characteristics, by changing them further (as we will see).
A typical Mandarin Chinese pronunciation shows a particular kind of voice
with a paraphonic setting with raised larynx @, not only for wemen. Other Chinese languages, instead, use a normal (or modal) laryngeal setting, or else one with
11. chinese
361
a lowered larynx. ese paraphonic dierences can be a good cue in identifying

the origins of Chinese people, even when they speak English or other foreign languages.
Taxophonics
11.3.1.1. As regards possible reductions, especially in weak syllables, we have
already illustrated those of vowel and consonant timbres, in particular, in 11.1.911 and 11.6-7, and here and there in 11.2.0-6.
We would like to underline here that in fast speech, in a preintoneme, two other types of reduction are possible, which can make Chinese less easy to understand.
us it is important at least to know their mechanisms. Should foreign learners
want to apply them in an appropriate manner, naturalness would certainly be improved.
erefore, obviously not in an intoneme again, after dierent consonants from
sonants ( after /p[h], t[h], k[h] q[h], [h] f, s, , h/), and with tone 4 (above all
when it is complete, /6/ (7), see below), close vowels, /i, y, M, u/ (i, I, y, Y M u,
U), can be devoiced, (, , , , ). We will show this fact only here (and
just for partial devoicing): (0p4u) /6phi>u/ pr (0p[faa) /6pufa/ bf (0t5n)
/6thi5Xn/ ten (0t[kU) /6tukhou/ dku (04twn) /6qythwan/ jtun
(0k[aa) /6khuha/ k> (0k7k$) /6ku6khX/ gk (0q5i) /6qhM5qi/ cji
(0q[muu) /6qMmu/ zm (07fc) /6hu6fan/ >fn (0,7qw) /6qi6qwo/
jzu (0f7pu) /6fu6pu/ fb (0[jaa) /6sinja/ xny (0[yy) /6syqhy/
xq (04fu) /6ufu/ f (07A) /6hu6ao/ ho
Besides, in syllables with zero tones, but not in an intoneme, approximant realizations are possible: (, ), for /s/ (s, z , B), and () for // (, ), but also for
(q[h], Q [h], ) and ([h], "): (7kAz|, 7kA3, -3^) /6kaosu/ gosu (5pI&QM|,
5pI2, -2) /5peiqM/ beizi (4qU&QU|, 4qU2, -2) /qhouqhou/ cucou
(5tU,&Bi|, 5tU,2, -2) /5tusi/ doxi (jaa^,|, jaa1,, -1,) /jaqi/ yji
(5u&y|, 5u2, -2) /5huqhy/ >uqu (5&|, 52) /5/ ii (5pa,&"u|,
5pa,2, -2) /5pau/ bau (5,&u|, 5,2, -2) /5qhihu/ qi>u
11.3.1.2. As we have said, Chinese words are formed by initial and nal elements. However, their combinations have some limitations; in fact, among all theoretical possibilities, only 55% are actually used, for slightly over 400 typical syllables (which are regularly indicated in teaching textbooks, with some oscillations
for the less frequent ones). Here we will report only the most remarkable characteristics.
(X) /X>/ er occurs with no initial; /f/ has a fairly limited distribution: it does
not occur before /i, j y, w/ and some other phonemes; whereas, only in loanwords /v/ is found (which is thus a xenophoneme added to the (v) taxophone of
/f/). Besides, ([h], ) (with other variants already seen) are taxophones of /q[h],
s/, which only occur before /i, j y, /, and are spelled j q x Equally, /M/ only occurs after /q[h], s/; and // after /[h], , >/ /o/, only after /m, p[h], f, w/ (and in
362
some particular forms after /j, l/ or with a zero segment); and /y/, only after /j,
l/, /n/ (~), or after /q[h], s/ ([h], ).
nally, we would like to add that /m, p[h], f/ do not occur before /X, M, w y,
> [j]u/ or some other segments; neither do /[h], / occur before /i, j y, / (or
some others), nor /k[h], h/ before /i, j y, / or /o, wX/. ere are some other more
specic limitations; but this is the spirit of Mandarin phonotactics /n, / are the
only, very frequent, syllable-nal consonants (besides, />/, which is extremely rare).
ere are some remarkable reductions of words such as the following (which are
quite normal, except in intonemes or in formal speech): (4Xm, 4X, 4X2m, 4X2mX) /XnmX/ nme ([q$$C, q$1, q$1m, q$1mX) /qXnmX/ znme ([wC,
w, w1mn, -1mn) /womXn/ wmen (5tam, 5ta, 5tamn, -mn) /5thamXn/ tamen Contrary to some other Chinese languages, in Mandarin there is a
somewhat exceptional occurrence of /m/ in nal position, as indeed that of (),
which we nd in : (4) /Xm/ really?/what??, (7) /6Xm/ yes/I understand.
Stress
11.3.2.1. At the end of the previous section, as well as in 11.1.9-11, we saw a
few words with weak syllables too, without either a toneme or a strong stress
(at most they had a secondary stress on posttonic syllables). We will see below that
weak syllables may also occur before the last one, and that generally even grammemes (or grammatical particles) are weak.
On the contrary, in most Mandarin words all syllables bear one of the four
marked tonemes ( 11.3.3.1-3). Among these, two-syllable words have the following stress pattern (&I'I): (^4n) /5an/ iyun (ljE~7i) /ljan6si/ linx (Î[pc) /5heipan/ heibn (0jA5) /6qjao5/ jioi
Chinese writing has separate characters (ideograms), but without separation
between words, phrases, and sentences, except for punctuation (which also includes a mini-comma () that is used to separate words in a list, but it is used less
and less). In trisyllables, without any weak syllables, we have the pattern ('I&I'I):
(5jPU0[[kc) /5jou6mokan/ youmgn (fx,05i) /fX,6>Xn5qi/ frnji
(w^hX7c) /hwo5hX6an/ hu>en
en Chinese words reach four or ve syllables, lexical compounding coincides
with syntactical composition. In fact, the various components are simply juxtaposed. However, there is a way to manage to show lexicalization better, within sentences.
As a matter of fact, if for instance we separately have (0q[muu) /6qMmu/ zm letter, grapheme, (^pI~5jIn) /5phin5jin/ pinyin syllabication, when we put
the two words together, we obtain (5pI~^jIn 0[muu) /5phin5jin 6qMmu/ pinyin zm romanization, transliteration, alphabetic writing (to show pronunciation), instead of a simpler but more articial (^pI~5jIn 0[muu); consider
also (6ti0i0A7tA) /6ti6ti6tao6tao/ dddodo
11. chinese
363
11.3.2.2. In sentences, however (as will be seen from the text in 11.4.2.2 as
well), there are frequent oscillations with regard to actual stresses on dierent syllables, both for rhythmic reasons and prominence interplay, and even for communicative purposes, even when they are not distinctive. us, for instance, /5akanu6jiX/ agny can have two frequent realizations, (5a,u0ji[$$, 5a,2u7ji2$); equally, even forms such as the following often oscillate:
(0ji[tjEEc, 7jijEc) /6jitjan/ ydin (0(4fI, 7(&I) /6pofei/ pfi (^qM7wI, 5qM0wI) /5qM6wei/ ziwi (0I[fuu, 7Iu) /6sinfu/ xnf (În5tjEn, 5In^thjEn)
/5qin5thjan/ jntian (0[lii, 7li) /6aeli/ il` (^km5pI, 5kmÎ) /5kan5pei/
ganbei!
All this can happen both in intonemes and preintonemes. And, what is more,
even opposite cases are frequent, that is syllables with zero tone which bear a
strong stress, verging on mid pitch (again independently of any intoneme): (4px,&jPU, &px,'jPU) /phXjou/ pyou (7t, 0') /6tan/ dni (4&U, &'U) /hou/ hou In addition, even forms bearing a tone (even written) are
weakened, as happens to the grammemes and locatives in the following examples:
(5w0a, -&a,) /5wo 6a/ uo (6ti0ja, -&ja) /6ti 6sja/ d xi (5wuli, 5wu&li) /5wu li/ wu l` (4t,li, -&li) /thi li/ t l`
Tones
11.3.3.1. e most remarkable aspect of the various Chinese languages (and one
of the most serious problems for foreigners) resides in the use of tones. In fact, in
languages such as Mandarin, a dierent pitch on a syllable is able to make its meaning change completely. erefore, we have to speak of tonemes for distinctive elements of a phonemic system, and of tones for their actual realizations, including
taxotones, combinatory variants, as can be seen from the examples given below.
In order to draw attention to their importance, and not to confuse concepts and
words, let us choose the example of ([m) /mae/ mi to buy and (7m) /6mae/
mi to sell. We now add some classical examples (giving some of the most frequent meanings, among the possible ones, since transliterations may unify dierent words {as phonic transcriptions do too}, represented by dierent characters,
or ideograms): (5ma) /5ma/ ma mother, (4ma) /ma/ m hemp, ([maa) /ma/
m horse, (7ma) /6ma/ m scold, curse
As some transcriptions have already shown, tonemes 3 and 4 are realized with
creaky (or laryngealized) voice, when their low taxophones occur: ([), () and (7)
(with secondary stress, (0)), but not when the high ones occur: () and (6). Tonemes 1 and 2, which are high, (5), (4), always have normal voice quality.
A prudential warning is necessary about some diagrams which appeared in certain publications, not only in tourist-like books, since they are too often completely wrong so they are of no help at all to unfortunate readers
Mandarin is relatively simple, in comparison with other Chinese languages, also as far as taxotones are concerned. In fact, Mandarin has only four marked tonemes, traditionally indicated and listed as 1 (5) /5/ >1 o\ (5ji) /5ji/ yi 2 (4) // >Q \
364
(4wu) /wu/ w 3 ([) // >5 \ ([) /se/ xu 4 (7) /6/ >Z \ (7ja) /6qja/ qi
11.10 shows for the four tonemes the four basic tones, in addition to three
taxotones, which are as important phonetically (or rather tonetically): in addition to 3=2 ( 3+3 = 3+2) () // >5 ( a toneme 3 occurring before another
3 which becomes almost a tone 2 {although it is generally but falsely identied
with an actual tone 2}), we have a half 3, () // >5 ( a 3 before one of the other three), and a half 4, (6) /6/ >Z ( a 4 before another 4). As can be seen, from
both tonograms and phonetic transcriptions which will follow, the last two are actually realized as the rst half of their full forms.
In a kind of international pronunciation, toneme 2 can be realized as () (instead of (4) ((-1))), as the ocial taxotone of // followed by another //. e pitch
characteristics of the other ton(em)es ought to be very similar to the ocial ones,
even for the zero toneme, as we have done in our transcriptions (although for
the latter, somehow, a realization of mid pitch {and weak stress} could be sucient,
(2), again in this type of international pronunciation).
11.10. Mandarin tonemes and tones.
1 /5/ (5) >1 2 // (4) >Q
3 // ([) >5 3 // () >5 3 // () >5 4 /6/ (7) >Z 4 /6/ (6) >Z
11.3.3.2. Obviously, although at rst all this might seem to be a useless complication, in actual fact, it is a real facilitation for contextual realizations. As a matter of fact, it is a simplication, from a tonetic point of view, which is visible only thanks to accurate phonetic (or rather phonotonetic) transcriptions, whereas
phonemic ( phonotonemic) transcriptions, as well as graphemic renderings, do
not change at all, since their entity, their essence, remains the same.
As can be seen from the following examples, and above all from 11.11, the
allotones ( taxophones) of tonemes 3 and 4 present a simplication, by assimilation.
11.11. Fundamental taxotones.
/ 5/
( 5)
/ /
( 4)
/ /
( [)
/ 6/
( 7)
/6 6/
(0 7)
11. chinese
365
In fact, toneme 3, ([) = (), // >5 shortens (even as its segmental duration does),
when it occurs before tonemes which are dierent from itself: (w5X) /hwo5hX/ hu>e (t*4y) /tXy/ dy (kA7) /khao6/ ko on the other
hand, when it occurs before itself, / / >5 5 it has the taxotone (): (U[pjAA)
/oupjao/ ubio
Even in this case, it is right to keep its tonemic and graphemic representation
unchanged, since, should we modify them, they would lead us to think of dierent structures, not only from a phonic point of view, but also lexically and, then,
semantically.
Toneme 4 resorts to a (halved) taxotone before itself: (0fa,7ja) /6fa6qja/ fji Tonemes 1 and 2 do not have any variants; and, although the taxotone of 3 (+
3) is tonetically (more) similar to toneme 2, it nevertheless remains a taxotone of
toneme 3 (except in the international accent {which is a simplication}).
11.12. Taxotones of particular tonemic sequences.
/5 5/ =
/5 5 5/ (5 5)
/5 / =
/5 5 / (5 4)
/5 / =
/5 5 / (5 [)
/5 6/
/5 5 6/ (5 7)
11.13. Taxotones of other tonemic sequences.

/ 5/ =
/ 5 5/ (4 5)
/ / =
/ 5 / (4 4)
/ / =
/ 5 / (4 [)
/ 6/ =
/ 5 6/ (4 63)
11.14. Further taxotones of tonemic sequences.

/5 / =
(5 [) =
(5 [)
/ / =
(4 [) =
(4 [)
/ / =
( [) =
( [)
366
11.3.3.3. As far as sequences of three tonemes are concerned, there are some
dierences (and they must be indicated and applied too) in comparison with tonemic (and graphemic) representations. 11.12 shows what happens to sequences
of /5 / followed by one of the other four tonemes: (5tU,^nM5fx,) /5tunan5fX/
donn-fe (5s~^~jE~4i) /5sannjanqi/ sannin-j (5jEN^~[jaa) /5sjan>Xnqja/ xianrn-ji (5iÛ,7) /5sihu6/ xih.
11.13 shows sequences of / / followed by one of the other four tonemes:
(4mI^laM5fa,) /meilan5fa/ Mi Lnfa (4^mI4l) /haemeilae/ hi mi
li (4w~^hn[tUU) /wanqhantu/ wnqun d (4Y~^ja,7jEc) /synja6qjan/ xny-jin
nally, 11.14 shows the behavior of the sequence / / preceded by one of the
rst three tonemes: (5s~^jEn[j<<Y) /5sanjanqjX>/ Sanyn Jir (4Nû[pjAA) /hanupjao/ hnbio (w^j[jU) /wojejou/ w y yu
11.3.3.4. In addition, there is a zero tone, which is realized in four dierent
ways (that are representable, however, with only three signs), according to the previous toneme. It is much simpler to refer to 11.15, which shows that a zero
tone has mid pitch after tonemes 1 and 2, although strictly speaking after toneme 1 pitch is [lower] mid, whereas it is [upper-central] mid after toneme 2: (5pi&(h) /5pipho/ bipo (4pi&hi) /piqhi/ bqi After a toneme 3, the unmarked toneme is [lower] high; whereas, it is [lower-central] low, after a toneme 4: (piâ,)
/pifa/ b`fa (7piwI) /6pihwei/ bhui
As we have said, three symbols are necessary (and sucient), since as we will
see shortly intonation slightly modies and compresses what we are explaining
here, which refers to an ideal pronunciation and is used in isolated forms, outside actual communicative contexts.
en two unmarked tonemes occur after a marked one, their pitch is as shown
in the tonograms of 11.16: (5k2ll$) /5khaelaelX/ kailaile (4na2ll$) /nalaelX/ nlaile (qU1l^lX) /qoulaelX/ zulaile (7sU3ll$) /6sulaelX/ slaile
Lastly, 11.17 shows how a zero toneme behaves when it occurs between
marked tonemes (by observing its tonetic logic as a gradual transition): (5tjE~1i7wI) /5thjanqi6wei/ Tianjiwi (4a,1"U7) /haou6/ Hou (li2hU,5a,) /lihu5a/ L`hoa (lja,2gX4n) /ljakX>Xn/ lige rn (6kn2dX7jEc) /6khantX6qjan/ kn-de-jin
11.15. Contextual pitches of a zero tone in bisyllables.
/5 / (5 &)
/ / (4 &)
/ / ( )
/6 / (7 )
11.16. Contextual pitches of two zero tones in trisyllables.
/5 / (5 2 )
/ / (4 2 )
/ / ( 1 )
/6 / (7 3 )
11. chinese
367
11.17. Contextual pitches of an internal zero tone in trisyllables.
/5 6/ (5 1 7)
/ 6/ (4 1 7)
/ 5/ ( 2 5)
/ / ( 2 4)
/ / ( 2 7)
11.3.3.5. To conclude about tones, we must indicate some syntactic-lexical variations, which have to be memorized as they are. We are talking about three numbers: (5ji) /5ji/ yi one, (5i) /5qhi/ qi seven, (5pa) /5pa/ ba eight, and the negation (7pu) /6pu/ b no, not. e three numbers are realized with a tone (6) before
one of the rst three tonemes (/5, , /): (6ji, 6i, 6pa), but with tone () before a 4
(/6/): (ji, i, pa) (but now even (5i) is acceptable). Also /6pu/, before a 4, becomes (): (pu). Besides, (5ji) /5ji/ yi becomes (4) //, even before a zero toneme,
which derives from (7) /6/, as in (4ji&X) /jikX/ yge.
Intonation
11.3.4.1. Although Chinese is a tone language, with relevant and distinctive
syllable pitch, it uses intonation too, as non-tone languages do. Obviously, in this
case, things are more complicated, since the basic tonemes with the taxotones we
have seen ( 11.3.3.1-5) are slightly modied by intonation, which superimposes on tones. In fact, it makes them change according to the typical movements of
the intonemes.
11.18 shows the preintonemes and intonemes of Mandarin, with their characteristics. e preintonemes force the pitch of every single tone to be canalized into the shapes indicated (which, in a more sophisticated than really necessary notation, could even be indicated with some small rings, as we will see shortly). A
normal preintoneme is compressed (/ / ( ), (( ))), an interrogative one is raised (/ /
( ), (( ))), whereas an imperative one is falling (/ / ( ), (( ))); nally, an emphatic
preintoneme is extended, non-compressed, (/ / ( ), (( ))).
11.18 also shows the modications undergone by the intonemes: the conclusive is falling (/./ (13)), the interrogative is rising (/?/ (31)), the continuative one is
compressed (/,/ (2)), whereas the suspensive is extended (/2/ (^)).
11.18. Mandarin preintonemes and intonemes.
/ / ( ) (( ))
/./ (13)
/ / ( ) (( ))
/?/ (31)
/ / ( ) (( ))
// (^)
/ /( ) (( ))
/,/ (2)
368
11.3.4.2. Besides isolated tones (typical of separate words {as examples are}
and of a suspensive intoneme, and of an emphatic preintoneme as well, which are
given in black), 11.19 also shows the taxotones (given in grey) of the four
marked tonemes, as they are modied when they are realized in conclusive (/./ (13)),
interrogative (/?/ (31)), and continuative (/,/ (2)) intonemes. It is obvious that the tonetic notation used is necessarily just an indication of the deformation undergone by actual tones. Of course, in non-tonal languages, the use of tonetic symbols is automatically much more precise, since it is free from any phonemic value applied to lexical pitch.
In addition, 11.20 shows in grey all the allotones occurring in the four preintonemes. e four marked in black, in the emphatic preintoneme, correspond to
the four isolated tones. ey are also useful for comparisons, which must be
done very carefully, in order to appropriately memorize and master them in actual sentences (at rst as a simple identication step, then as active drills as well).
Undoubtedly, they require a good deal of commitment and constancy. But any
eort will certainly be rewarded. Of course, paraphonics further modies things
depending on attititudes, feelings, .
11.3.4.3. Lastly, we will see three examples to start an accurate analysis of intonation which is always in an indissoluble relation with tones. Given the complexity of this enterprise, we will use the same basic sentence, with the same tonemes.
Of course, other tonal combinations are to be experimented as well, on the basis
of the examples provided:
/./: (5ta0 4ji2gX X7c2dX 5ku&~ja,13) /5tha jikX hX6antX 5kunja./ Ta yge
hn de gunia.
/?/: (5ta0 4ji2gX X7c2dX 5ku&~ja,2ma31) /5tha jikX hX6antX 5kunjama?/
Ta yge hn de gunia ma?
//: (5ta0 4ji2gX X7c2dX 5ku&~ja,^ 7t3 w1bu 7tha13) /5tha jikX hX6antX 5kunja 6tan wopu 6aetha./ Ta yge hn de gunia, dni w b i ta.
11.19. e four tonemes combined with the four intonemes.
the 4 tones with /./ (13)
the 4 tones with /?/ (31)
the 4 tones with /,/ (2)
the 4 tones with // (^)
11. chinese
369
11.20. e seven allotones combined with the four preintonemes.

//()
// ( )
// ( )
// ( )
Text
11.4.0. e story e North Wind and the Sun follows, given in four dierent
normalized versions. We start with the (neutral) Chinese pronunciation of (neutral British) English this is the rst step of the phonetic method (the written text
is given in 2.5.2.0). e Chinese translation follows, in its neutral and international versions.
Please, note the creaky vowels before pauses ( with creaky phonation type, or
laryngealization, (); whereas in Mandarin creaky voice is in relation with low
tones, also for the voiced consonants belonging to the same syllable coda). ese
creaky vowels confer great authenticity, together with a (paraphonic) postdorsal
setting, that is with a slight but permanent or frequent raising of the dorsum
towards the velum together with the raising of the larynx # @.
At the end, as always, there is the version which gives the English pronunciation of Chinese, by neutral British speakers, uent in Chinese (after prolonged
Speakers of American English could prepare their own version both of the Chinese pronunciation of English and of their pronunciation of Chinese, as an excellent exercise, by listening to native speakers, best of all after recording them. Of
370
Chinese pronunciation (of English)

11.4.1. #@ (2tX'ns 'jin2 2Xn2X'san2 2jX2is'phju2ti, 'ji 2jXs2Xs'thl2$13| 'jn
2X'thlV2lX2 'khIm 2X'l,2 'lp 2in2X'jm 'khlUk13| 2IX'li2\ 22X'jan 2u'fXs
2sXk'ii2 2in'mI2ki, 2X'thlV2lX2 'thIk 2is'khlUk 'f^| &u2i2kXn'i2X 2sX"thl2gX 2Xn2i'a2$13||
'n2 2X'ns 'jin2 'lu2 2Xs'aX2 2X2i'khu13| 2X2X'[ 2i'lu^| 2X'[ 'khlUs2li2
&i2X'thlV2l$2\ '5U 2is'khlUk X'lAn2iC13| 2Xn2Xt'las2\ 2X'ns 'jin2 'I 'Vap 2iX'thCp13|| 'n2 2X'san 'n 'At13 'jm2li13| 2Xn2i'mii2tli2\ 2X'thV2lX 'thuk 'f3 3
2is'khlUk13|| 2Xn'sU2 2X'ns 'jin2 2jXs2X'bl 2tX2kXM'fs2| 2X2X'san13 2jXs2Xs'thl2X13 2X2X'thu13||
&i2ju'lk31 2Xs'thli2| 2ju'jn 2tX'iX2lit 2X'c31|||)
Mandarin text
11.4.2. Yu y hu bife gen tiy zinar eln ide bni d,
uoe uoe lile yge zudorde, ena >uane yjin hu pozi. Tamen
li ji alia ho le uo, i n xian jio ge zudorde b tade pozi tuole xilai a, ji sun ide bni d.
Ho, bife ji `qi d jn li j`ngua j`ngua, k ta guade yu lhai, nige rn
b pozi gude yu j`n; do mlior bife mile fzi, `ho ji sunle. Yhu`r tiy ji >uli rrerde y i, ni zudorde m ji b pozi tuole xilai. Suy` bife b n b >rn dod` hii tiy b` ta bni d.
N` x`huan u rge ma? Wmen zi > y bin ma?
Neutral Mandarin Chinese pronunciation
11.4.3. #@ (jU7jiwI^| I5fx,2 2gn7t&ja,2\ 6x,&3nY ^2x,7lwc13|
4I2dX p1 7ta13|| 5w2"X^w2"X2| 4l2lX &i2gXU7tY3d$13|| 5N&a,2 5wN2"X2 4ji0jE 7U 4pA2Q13|| 5tam [ljaa2\ 0jU5a,2lja, A1lX 5wx2|| ^4I &nx,5jE~0jA2\ 7$3g$U7tY3d$2| pa5ta2dX 4pA2Q2| 5tw2lX 7ja3la2|| 0jPU7sw2 4I2dX p1 7ta13||
[AA^|| pI5fx,2 0jU1i 7ta2 7Ic&l2 I5kwa2 I5kwa13|| k$15ta^||
5kwa2dX 7 7li3h2 0nI3g$4n^|| [paa2| 4pA2Q2 kw1dX 0[IIc13|| 7tA 0[[ljY^||
pI5fx, 4mI2lX 4fa2Q13|| 4AA2 0jU7swc3l$13|| 0ji[w$$Y^|| 7tja, 0jU5u&l2|| 0$5X2dX ji713|| nU7tY3d$2|| ma7a2 0jU[paa 4pA2Q2 5tw2lX 7ja3l13|| swji2| pI5fx,2|| 0pu4nx,0u hx,7c^|| 7tAii2 4 7t&ja,2|
pi5ta2 p1 7ta13||
~ii^wN 7$U $Y5kX2ma31|| wC7q2 7a ji7pjEC3ma31|||)
11. chinese
371
International Mandarin Chinese pronunciation

11.4.4. (ju0ji4hwi^| pi5fx2 kxn0tha4ja2\ 0x7qa3na ^x7lwxn13| 4i2tx
pxN1 7ta13|| 5w2x 5w2x2| 4la2lx 4ji2kx qu7ta3tx13|| 5xN&a2 5hwaN2x2 ji7jan 7hu 4pha2qM13|| 5tham[xn] [ljaa2\ 0ju5a2lja ha1lx 5w2|| ^4i &nx5ja~
7ja2\ 7X3kX qu7ta3tx2| pa5tha2tx 4pha2qM2| 5thw2lx 7ja3laa2|| 0ju7swaN2
4i2tx pxN1 7ta13||
[haa^|| pi5fx2 0ju1i 7ta2 0in4la i5kwa2 i5kwa13|| khx15tha^||
5kwa2tx 7 7li3ha2 7ni3kx 4xn^|| [paa2| 4pha2qM2 kw1tx 0[iin13|| 7ta 0m[ljaa^||
pi5fx2 4mi2lx 4fa2qM13|| [haa2 0ju7swan3lx13|| 0ji[hwxx^|| 0tha4ja 7ju^hu4la2|| 0x5x2tx ji7a13|| niqu7ta3tx2|| ma7a2 0ju[paa 4pha2qM2 5thw2lx
7ja3la13|| sw[jii2| pi5fx2|| 7pu&nx7pu hx7xn^|| 0ta[tii2 4ha 0tha4ja2|
pi5tha2 pxN1 7ta13||
ni i5hwaN 0x[u x5kx2ma31|| w2m[xn]7qa2 7ha ji7pjam3ma31|||)
English pronunciation of Chinese
11.4.5. (j'jI;i&hwI32| bI'f:2 gn'Tha&j:2\ 5ch&tsan &Gl':n3 3| 5SID
'ph~Z 'ThA:3 3|| S'Gs&G2| 5lal &jIigts'ThA;D3 3|| 'S;n&S:2 c'A;~G2 5jIiGi& 'h; 'phats3 3|| 5ThA:m li'A:2\ G5Sli& 'hal S':2|| ^5S;I nSi'nGi&a2\
'c;g ts'ThA;D2| phA;5ThA;D 'phats2| Th5l Si'A;la2|| Gi&s'n2 'SID
'ph~S 'ThA:3 3||
'ha;32|| bI'f:2 Gi'S;Gi 'ThA:2 'chn&la2 &Gk'A: &Gk'A:3 3|| 5kh;Z 'ThA:32||
kh5A;D 'j;I 'lIi&ha2 &nIg'>:n32|| 'phA:2| 'phats2 'khwD jI'chn:3 3|| 5Tha; &mli'A:32|| bI5f 'mIl 'fA;ts3 3|| 'ch:&ha2 ci&'swnl3 3|| &jIi'hw:32|| 5Tha;&j Gi&'chula2|| >'>;D &jIi'Sa;3 3|| &nIts'thA;D2|| mA;'S:2 ci5phA: 'phats2 5Thw;l Si'A;la3 3|| 'sw;jIi2| bI5f:2|| &phu5n&phu ch'>:n32|| 'ThaDi2 5haZ 'Tha;&j2|
phIi'ThA:2 5ph;~Z 'ThA:3 3||
nIi5SIi&hwn 'ch;S 'kh;m21|| &wm'tsa;2 5ch &jIipi'nm21|||)
A short appendix
11.4.6. (6ta0M7pA) /6ta6qM6pao/ dzbo (5kU,&u) /5kufu/ gofu (kung
fu, 11.1.9 for (-2v)), (0i5kU,) /6qhi5ku/ qgo (7tA) /6tao/ do (4jhn7tA) /qjeqhan6tao/ jiqundo (jeet kune do), (5ti4n)
/6thaeqiqhan/ tijqun (5pa0wa[aa) /5pa6kwaa/ bagu
Besides: (Û,4kw[x]) /5ukwo/ ogu China (pI5,) /pei5qi/ Biji, (5tjEn^m4m wA[aa) /5thjan5anmXn kwaha/ Tian'anmn
Gu> (0s5wn) /6sM5hwan/ S>uan (t5wn) /thae5wan/ Ti-
372
wan (kwA5tU,) /kwa5tu/ Gudo Canton (0a[) /6ahae/ hi

More: (4mA^X5tU,) /maoqX5tu/ Mo Zdo (4N0N7A) /an6>Xn6ao/ Yun Rn o (traditionally: Yuen Ren ao ] (5U ^n4l) /5ou 5Xnlae/ ou Enli (7t* jA4p,) /6tX sjaophi/ D Xiop (lIm5pjA) /lin
5pjao/ Ln Biao (ljPU A4i) /ljou aoqhi/ Li oq
Lastly: (5snli4twn, -4twX) /5sanlithwXn, -twX>/ Sanl`tn(r) (0ku5kU,) /6ku5ku/ Ggo (a,4x,) /hahX/ > (4jiX4n) /jihXan/
Yhyun (5a, 0ji4[U) /5a 6jimou/ a Ymu (kU7li) /ku6li/ G L
(Û5U) /5ou5ou/ Ouou Europe, (lwn5twn) /lwXn5twXn/ Lndun London, (j,4kw[x]) /5jikwo/ Yigu England, (5suX4ln) /5sukXlan/ Sugln
Scotland, (6ji0a7li) /6ji6ta6li/ Ydl Italy, (l[maa) /loma/ Lm Rome, (5wI^~i5sM) /5weini5sM/ Weinsi Venice, (7pIw5fn) /6pei5two5fXn/ Biduofen
Beethoven.
12. Japanese
12.0.1. We provide the modern neutral pronunciation of Japanese, based on
that of Tokyo. Our transliteration avoids diacritics for vowels, by indicating long
vowels as ii ee aa oo uu (instead of i e a o u or ] As far as consonants are concerned, we prefer the most widely used system (by Japanese authors
too: the Hepburn system, with the exception methodologically considerable of
the choice to use n even before m p b instead of m]. us, the phonemic transcription has the task of indicating the systemic structuration; whereas the phonetic transcription, of course, aims at precision, without which everything would be
approximate and frankly useless.
erefore, we have: (qM) /tM/ _u (FM) /hM/ fu (i) /si/ i (j) /sja/ a
(j) /sjo/ o (jM) /sjM/ u (i) /ti/ >i () /tja/ >a () /tjo/ >o (M)
/tjM/ >u (i, Bi) /zi/ ji (j, Bj) /zja/ ja (j, Bj) /zjo/ jo (jM, BjM) /zjM/
ju (QM, zM) /zM/ zu But we prefer to use a more logical c> ([]) /tt[j]/
(which is less eurocentric than t>] and n (with n' + V or y] (P, , , , ) // (that
some systematically render with thus resolving in a graphonemic way the
slight problem of n' before V and y and also that of m before m p b] In addition, we have () //, that we render with g which can alternate with (g) /g/, as
we will see below, 12.2.1.1-2 (others use to compensate for the absence of
any transcriptions).
Length is distinctive both for vowels and consonants. We will mark it by doubling the phonemic and graphemic symbols /kappoo/ and kappoo respectively.
In our phonetic transcriptions, for contoid lengthening it is necessary to add (:):
(3kp'p:) (but, as can be seen and heard, after the second element).
12.0.2. In Japanese, even the pitch of the dierent morae is distinctive. ey
form syllables, words, phrases, and sentences. e example just seen shows that
pitch is not marked in spelling, that is in the transliteration, as well as in normal
hiragana writing (3i'm2n, 3i'm3n 3i'm2&n) /hiRaana -na (which is
added to characters, ideograms, that were taken from Chinese).
Normally, this does not happen in the other type of writing, katakana (3k't3kn, 3k't2kn) /katakana/, which is generally used in teaching and scientic
textbooks, in order to hint at the pronunciation of onomatopeic terms and recent loanwords (more recent than the Chinese ones, which adapted to Japanese),
or for stylistic reasons.
Instead, in phonotonemic transcriptions, we indicate with the point after
which the voice goes from a mid pitch to a low one. We call this akusento (
12.3.2.1), by using the term taken from English [accent] in order to indicate this
particular pitch phenomenon. As a matter of fact, it is not properly a stress element, since, in actual fact as we will see both pitch patterns and segmental con-
374
sistency or syllabic weight determine stress, 12.3.2.5-14. Indeed, these

points may even be more than just one, as can be seen from the two previous examples. However, these points have to be interpreted as possible variants, in a paradigmatic opposition, which is typical of phonemes, that manifest themselves on
the syntagmatic axis, as is typical of words in sentences. is means that one must
be chosen, excluding all the others.
On the contrary, phonotonetic transcriptions more concretely show the pitch
of every syllable in a word, or rhythm group, since there are precise patterns for
neutral Japanese (as we will see in detail later on, 12.3.2.2), even if with variants.
As a matter of fact, for the word hiragana we nd as many as three possibilities (although the last is less favorite and older), as we have just seen, and as many as two
for katakana ile, for a word like katana there is just one possible neutral
toneme: (3k't2n) /katana (/katana/ is only regional).
We prefer to use ( ) (instead of a more popular in Japan but less satisfactory (^ ) //), since the marked pitch is low, whereas the unmarked one is mid, not
high (as a more traditional notation would lead to think, being based more on a
tonemic rather than a tonetic criterion, which considers high what is non-low).
Vowels
12.1.1. Japanese has only ve vowels, which can be distinctively short or long
(or rather doubled, being realized almost as monotimbric diphthongs), and they can
combine into dierent kinds of sequences (as happens in a slow pronunciation of
Spanish in Saavedra or La Habana (aa'B;Ra, laa'Ba;na). However, too often
Japanese pronunciation is hastily described as having the vowels of Spanish and
the consonants of English. We will see that this is not the case at all.
12.1 shows the actual articulations of the ve vowels: (i, , , , M ii, , aa,
, MM) /i, e, a, o, M ii, ee, aa, oo, MM/ i e a o u ii ee aa oo uu Since long
vowels are phonemic sequences (and phonetic geminations), the marker of (short)
/a/ () is grey, seeing that it is articulated as a less open vocoid (even in stressed syllables). e most problematic Japanese vowel of course for non-native speakers
is (M) /M/ u which lacks the typical lip rounding of (u) so widespread in many
languages. Besides, it is articulated with the tongue dorsum in a fronter position
than (u) in most languages; in fact, it is back-central, not simply back.
Mostly, the younger generations articulate /M/ in a further front position, as
high central with partial or full rounding, () or (%), respectively but we always
notate (M), since those are youth characteristics which generally change into (M),
sooner or later. is peculiarity continues in relative time, without really changing the structure. Obviously, real non-neutral pronunciations are a dierent thing.
Let us now see some examples for each phoneme: (3i') /ie ie (i2k) /gikoo/ ginkoo (3s'6i3ii) /sabisii/ sabiii (i2P) /eie/ eien ('3) /eao/ egao
(3k'm2t) /kaRate/ karate (i) /ai/ ai (Q2:i) /zassi/ zasi (3m't) /mato/ mato ('n3k) /neko/ neko (3'i3mM) /okiRM/ okiru (3't2k) /otoko otoko (3sM'mi)
/sMmi sumi (2kM3m) /akMma/ akuma (3mM'i) /mMsi mui.
12. japanese
375
12.1.2. For monomorphemic ei and ou (but the latter is generally transliterated as oo] it is normal to have /ee, oo/ (, ): (3sPs) /sesee/ sensei (3sM'i)
/sMiee/ suiei (2) /eekjoo/ eikyoo (3jMM) /oosjMM/ oouu (2gP)
/oogo/ oogon (2P) /ooe/ ooen
Instead, for heteromorphemic ei and ou it is normal to have /ei, oM/ (i, M):
(ki2t) /keito/ keito (si) /sei/ sei (32msM) /aRasoM/ arasou (sM) /soM/ sou
12.1. Japanese vowels.
/i/ (i), /ii/ (ii)
/e/ (), /ee/ ()

/a/ ()
/M/ (M), /MM/ (MM)

/o/ (), /oo/ ()
/aa/ (aa)
12.1.3. Besides, neutral Japanese has the peculiarity of presenting vowel devoicing. Indeed, there are two partially dierent degrees.
e rst type is complete, and produces voiceless vocoids, that is with no vibration of the vocal folds, as also happens with consonants, such as (voiceless) (s) /s/
in comparison with (voiced) (z) /z/. More appropriately, these voiceless vowels are
lenis too (or lenited), only the arytenoids are open, as for (h). is concerns (i,
) between voiceless consonants, or between a voiceless consonant and a pause,
when in low-pitched syllables (but occasionally even in non-low-pitch ones, however never on accented morae, after which pitch becomes low), and never in interrogative post-intonemes (which shows a raising of the basic pitch): (3i'i) /kisi
kii (3k2i'6i2mM) /kMtibiRM/ ku>ibiru (3h'n2i) /hanasi hanai (3i't3q)
/hitotM/ hito_u.
e second type, or degree, of devoicing is half-voicing, which concerns the other three vowels, but acts less systematically. In fact, we nd (, ), especially in the
rst syllable of words, when followed by syllables containing the same vowel (more
rarely /e/ () is devoiced too): (3k't2n) /katana katana (3t'k2m) /tokoRo/ tokoro (3k's2ni) /kesanai/ kesanai
e second type again applies, instead, to all vowels before a pause, either short
or long, preceded by any consonant. Obviously, /i, M/ become (i, ), when they
are in the condition of total devoicing, in the context (=|). erefore, before a
pause, the last vocoid is (, , , , ) (even in diphthongs). It is important to state
that this type of devoicing is syntagmatic, which means that we have one vocoid
not two which begins as voiced and ends as voiceless. So the rst part is voiced,
whereas the second one is voiceless: (()) and this is hinted at by the pause context, even if the actual pause is short. Examples relating to this phenomenon can
be found in the passage of the Text section, 12.4. In simpler terms, we could say
that the auditory eect is almost that of a very short (h), ((h)) or a semiapproximant ((h)).
For emphasis, strictly speaking (or for other paraphonic implications connect-
376
ed with states of mind), often an actual (h|) sequence may be heard: (s3d3skh2) /soodesMka,/ Soo desu ka?!
Consonants
12.2.0. As we have already said, it is currently thought that Japanese consonants
are pronounced as those of English. We have to correct this false information,
which is further reinforced by the widespread use of > j f _ in transliteration
(as we do ourselves, for the sake of simplicity). In fact, > j stand for ( j,
Bj j), with no absorption at all of (j) /j/; besides, f is not (f), but (F). nally, there
is (), that as we will see is an intense (syllabic) provelar semi-nasal: (P) //.
e table of 12.2 gives the consonantal articulations of Japanese which are
necessary for an adequate pronunciation of this language.
Instead, 1.9-15 show the orograms grouped by manners of articulation of
all contoids treated in the chapters of this handbook (including secondary, occasional, and regional variants), which are needed for the 12 languages dealt with.
is exposition renders the necessary comparisons among dierent languages
more useful.
m
(n) n (~)
(N)
F p b
t d
( )
(q Q)
( )
(6)
s z
( B)
_
(F)
() j
/r/ m|()
laryngeal
velar
provelar
postpalatal
palatal
bilabialized
prepalatal
prepalatal
alveolar
dental
bilabial
12.2. Table of Japanese consonants.
k g
{(w)}
()
{()}
h {(H)}
Nasals
12.2.1.1. At the beginning of syllables, Japanese has three possible nasal phonemes: /m/ (m) m (bilabial), /n/ (n) n (alveolar; realized, however, as prepalatal,
(~), before /i, j/, by assimilation): (3~i'm2n) /nimono/ nimono and // () g (velar): (3'3k) /aakM/ agaku.
But we must add at once that, within words, // () g (which is sometimes transliterated as ) can be systematic only in the most neutral type of pronunciation,
after /, /. In fact, nowadays the oscillation between /g / (g ) is very
widespread, but with much uctuation among people and words. However, no
native speaker systematically has only (g) /g/. In (sentence, phrase, lexeme, or
grammeme) initial position, (g) /g/ occurs, even for speakers who possess () //;
12. japanese
377
for (enclitic) ga we normally nd () /a/; for ga (conjunction) we have (g), but

() is possible too, even after a pause.
Again, there is (g) /g/ above all in loanwords, in onomatopeic expressions, in
Chinese reduplicate words, and after certain proclitics (which are obviously heterorganic): ('i3) /gia/ giga (i3) /gia/ ginga; (3mi'i) /migi/ migi (3m'gM2m) /magMRo/ maguro (3mgP) /moge/ mongen (3m'g) /mago mago
12.2.1.2. An even bigger phonic problem regarding nasals arises from a fourth
Japanese nasal phoneme, moraic n, // (sometimes transliterated as ] which
always occurs in a syllable coda and has a prevailing articulation as semi-provelar
( provelar semi-nasal, with no actual contact between the dorsum and the velum). It is phonetically more energetic, intense (P).
It occurs in three positions: (1) before continuous consonants ( those produced
with an incomplete occlusion of the oral cavity), that is /s, z j, h/ (s , z B
j, h F ), (2) before a vowel (which is heterosyllabic, of course), and (3) in
nal position before a pause (or, again, before continuous consonants, or vowels)\
(3P'sP) /esa/ ensan (h2j) /hasja/ hana (MN2jM) /MjM/ un'yu (kP2) /kaa/ kanwa (3ssP3i) /seesahi/ seisanhi (sP3j3k) /sehjakM/ senhyaku (3g'sP3F) /gosehM/ gosenfu (tP3i) /tai/ tan'i
However, by assimilation, // has other realizations too. In fact, it is articulated
as a homorganic nasal (to a following consonant and is always intense): (1) (,
, ), before the correspondent stops /p, b t, d/ (p, b t, d) and /k, g/ (k, g, ):
(s3p) /sapo/ sanpo (i2bMP) /sibM/ inbun (M2tP) /Mte/ unten
(tN3i) ((-3i)) /teki/ tenki (t3k) /taka/ tanka and (2) (, ) before dental
or bilabialized prepalatal stopstrictive taxophones of /t, z/ (q, Q) and (, ):
(3bQi) /bazai/ banzai (2) /etjoo/ en>oo (~i2iP) /nizi/ ninjin and, naturally, () before the frequent stop(semi){con}strictive variant, (w,
), of /ka/ (k) ka\ (t3w, -3) /taka/ tanka.
In addition, we have (3) (, , , ) before nasals, /m, n, / (m n g) and again
() before r /R/ (), : (M3m) /Mmee/ unmei (3ni) /anai/ annai (~i2~i2k) /ninikM/ ninniku (2i, -3i) /aai, a-/ angai and (3'2k[]3i)
/aRakMsi/ anrakui
Stops
12.2.2.1. Japanese has three (voicing) diphonic pairs for stops: the voiceless are
/p, t, k/ (p, t, k) (as already seen, often /k/ is (w, ) + /a{a}/; occasionally /t/ becomes (th) + /a{a}, o{o}/; and often /p/ is (ph), in the context between // and /i{i},
a{a}/): (3k'pi, 3w-, 3-, -'phi) /kapai/ kanpai ('t3k 'th-) /tako/ tako (3ip'p:M) /kippM/ kippu (3't -'th) /ata wata (k3p) /kepoo/ kenpoo
(word-initial /p/ only occurs in loanwords and onomatopeic forms). In emphatic
speech, there is more aspiration.
e greatest oddity regards /t/ (t) which, by assimilation, before /i, j/ (i, j) is
realized as a bilabialized prepalatal stopstrictive () (transliterated as > even if /j/
378
(j) remains; but it is important to note that lip rounding is reduced, by assimilation to /i, j/, but the phone remains dierent from ()): (3i'i) /titi >i>i
(2t) /tjato/ >anto In addition and even more strangely /t/ (t), before
/M/ (M) is realized as a dental stopstrictive (q) (transliterated _): (3qM'Ni) ((-i))
/tMi/ _ugi
Let us observe, once and for all, that the palatal realizations of /k, g, /, (,
, N), are instead postpalatal, or retracted palatal, and can be represented better
with special symbols, ((, 8, )). ey may safely be used (although, more often,
(, , N) are used), after stating that they are realized in the rear part of their articulatory space, we dene postpalatal, which in any case remains distinct from the
prevelar articulation, (, , ), typical of most languages before front vowels and
(j), as in English (('hI;i, '>jl)) {Am. Engl. (('<jl))} (('s)) /'kIi, '<Egjl, 'sII/ key regular singing.
us, we have: ((3ip'p:M, 3mi'8i, 3qM'i)) /kippM, migi, tMi/ kippu migi _ugi
or (3ip'p:M, 3mi'i, 3qM'Ni), provided (Ni) remains dierent from (~i) (therefore,
/ni/ (~i) should not be rendered as if it were (Ni), as we nd in quite a few publications).
12.2.2.2. For /b/ (b) b the variant (6) is more common ( a voiced bilabial constrictive), which occurs after vowels, especially in non-slow and non-formal pronunciation; less frequently the corresponding approximant, (B), is used: (3sM'63mM) /sMbeRM/ suberu (bi2ki) /baikai/ baikai (jM3bi) /zjMbi/ junbi
e phoneme /d/ (d) d poses no problems, except that genuine Japanese words
never have *di, *dyV and *du substituted by ji /zi/ (i, Bi) jV /zj/ (j, Bj)
and zu /zM/ (QM, zM): (di3BiP -3iP) /daizi/ daijin (3jM'zM -'QM) /zjMzM
juzu Its normal distribution, non-emphatic and non-slow, is (B[j] |[j], [j]).
Also /g/ (g) g poses no problems, apart from a complementary (or alternative)
distribution with // () [g which sometimes is transliterated as as already said):
('g3g3k) /gogakM/ gogaku
Word-initial vowels, both at the beginning or in the middle of phrases and sentences, are generally preceded by (), especially for emphasis or to separate vowels
of adjoining words.
is fact will be indicated prevailingly in connected transcriptions, such as those
in 12.4. In addition, especially (but not only) in women's pronunciation, short
utterance-nal vowels, mostly with a suspensive intoneme, can be followed by ([])
( with or without {an audible} release) as an alternative pronunciation instead of
a possible more normal partial devoicing of the last vocoid: ('[]3k|, '[]3k[]|)
/oke/ oke (3[]'sM|, 3[]'sM[]|) /asM asu In the passage in 12.4.2.3, a couple of
cases are indicated.
Constrictives
12.2.3. Japanese has a pair of grooved dental constrictives /s, z/ (s, z) s z e
voiceless one, /s/ (s) s is realized as prepalatal round () (with reduced rounding,
12. japanese
379
by coarticulation) before i /i/ (i) (transliterated as ) and before yV /j/ (j) (transliterated as V although /j/ (j) does not disappear at all): (3j'iP) /sjasi/ ain
e corresponding voiced sound, /z/ z is (z) (slow careful: (Q)) and (|Q,
Q). Hence, it is realized as a dental constrictive, between vowels, either in words
or sentences, either in normal or fast speech. However, after a pause or //, it is realized as a stopstrictive: (3k'z) /kaze/ kaze (slow careful: (3k'Q)), (3QM'P)
/zMa/ zuan (s3Q) /sezo/ senzo
In addition, we nd /z/, before i /i/ (i) (transliterated as j instead of z] and before yV /j/ (j) (transliterated as jV although /j/ (j) remains), which is realized as
prepalatal rounded (again, with reduced rounding) (Bi, Bj) (slow careful:
(i, j)) and (|i, |j i, j): ('FM3Bi) /hMzi/ fuji (3i'tP) /zite/
jiten (k2i) /kazi/ kanji
Approximants
12.2.4.1. Japanese has three approximants. e rst, /j/ (j) y is (voiced) palatal:
(3s2j'n3m, -2m, 3s2j'n3m) /sajonaRa, -Ra, sajoonaRa/ sayonara (-yoo-) (3j2i'i,
3j:'i) /jasiki yaiki (3mi'2k) /mijako/ miyako (j2MM) /jookjMM/ yookyuu (jMM3i) /sjMMsi/ uui
As can be seen from the examples, it remains unchanged in word-initial position /j/ (j), whereas it is realized as a phonetic zero, (`), when preceded by
i /i/: /ij/ = (i). In non-slow speech, the same is possible for /ej/ = (), or ();
thus we will mark () (by using a palatal semi-approximant symbol): (3h') /heja heya.
After the voiceless stop phonemes /pj, tj, kj/, /j/ is devoiced, (): (p, , );
but it remains (j) after other consonants (even if voiceless, /s, h/ (, ) s h] and
without being absorbed by /t, s, z, h/: (, j, Bj, j, j) (in spite of transliterations such as >, , j). Sequences such as *yi, *ye do not occur.
12.2.4.2. e second Japanese approximant, // () w (which occurs in the
syllable wa), is (voiced) provelar diering from (w), which is velar rounded; it has
the same relationship with /M/ (M) u as happens in English between /w/ and /Uu,
U, u/ (w), (u/Uu, , ) (('wn:, 'khwk, 'Th;u/-U;u, 'Thk, Th'IiT) win, quick, two,
took, to eat): (3't2i) /atasi/ watai (3k') /kaa/ kawa (3M'2s) /Masa/ uwasa (dP2) /dea/ denwa
e third approximant, /h/ (h) h is voiceless laryngeal; however, in fast pronunciation (h) can become voiced, (H), after vowels. But the most remarkable fact is
that, by assimilation, /hM/ is (FM) (a voiceless bilabial approximant, transliterated as fu] and that in /hi, hj/ (i, j) we have a voiceless palatal approximant.
In addition, a voiceless velar approximant, (), is very frequent for /ha{a}/: (3h'hP, 3'HP) /haho/ hahon (3h'hP, 3'HP) /hahe/ hahen (3Fk) /hMkoo/ fukoo (3i') /hie/ hige (3j'kM) /hjakM hyaku
380
Trills
12.2.5. Japanese has one phoneme of the trill type, which is similar to Spanish
r /R/ (R), as in interpretar /inteRpRe'taR/ (in&teRpRe'taR). A realization like this could
be sucient for a fairly good pronunciation of Japanese, all the more so because
that is indeed one of the possible realizations. However, it is better to learn the
two most typical articulations given shortly (which are alveolar again), and use
them instead of (R).
e rst taxophone of /R/ r is (m), lateral ap (or lateralized ap 10.13 in
NPT/HPh, besides 1.14.2-3 in this handbook), which occurs after vowels (even
within sentences): (3k'2m) /kaaRa/ kawara (3'mi) /eRi eri (the same phone also occurs in American English before (), ('bm) /'bE/ better] A non-lateralized
ap is possible too, ([): (3k'2[) /kaaRa/ kawara (3'[i) /eRi eri (which is the
main American phone, in all contexts without (), ('b[i) /'bEi/ Betty] In Japanese it is better to use (m), although ([, R) are possible too.
e second taxophone of /R/ r is (), lateral tap (or tapped lateral, 1.15.3),
which occurs after // (even within sentences) or after pauses: (b2i) /beRi/ benri (3M'i2Bi, -2i) /RMizi/ ruiji (32i'i, -:'i) /Rekisi/ rekii (3i) /RoRi/ ronri
For the sake of simplicity, we could say that the dierence between the two realizations consists in dierent degrees of lateralization. In fact, (m) is less lateralized,
as lateralization is an added, or secondary, component (let us say 1); whereas, for
() lateralization is prevailing, or primary (let us say 2).
Occasionally, lateral realizations can be heard as well, (l) (which is still alveolar), and postalveolar ones (of various manners of articulation respectively: ap,
tap, stop, apped lateral, lateral: (, e, , , $). Obviously, these phones need not
be actively acquired it is sucient to be simply able to recognize them.
Palatalization
12.2.6.1. Before /i, j/, the phonemes /n, t, k, g s z h/ have peculiar but necessary realizations: (~i, Ni) /ni, i/ ni, gi (i, i, i) /ti, ki, gi/ >i, ki gi (i Bi, |i,
i i) /si zi hi/ i ji hi\ (~i2jMM) /nizjMM/ ninjuu (3k'Ni2mM) /kaiRM/
kagiru (3i') /tie/ >ie (3i'2mM) /kieRM/ kieru (3i'2k) /gikM/ giwaku (3i'm) /sim ima (3'Bi) /azi/ aji (3i'mi) /zimi jimi ('m3i) /mahi/ mahi
In addition, we nd: (~j, Nj) /nj, j/ ny, gy (, , j) /tj, kj, gj/ >, ky gy (j
Bj, |j, j j) /sj zj hj/ j hy\ (~jMM22k) /njMMakM/ nyuugaku (N2Nj) /aja/ angya (MM3) /tjMMtjo/ >uu>o (3MMMM3j) /kjMMkjMMsja/ kyuukyuua (jMM2~jMM) /gjMMnjMM/ gyuunyuu (3i~jMM3s)
/sinjMMsee/ innyuusei (kN2jMM) /kajMM/ kan'yuu (3id3j) /zidoosja/ jidooa (3j'kM) /hjakM hyaku
Again, we must remember that, for /k, g, / + /i, j/, the actual articulation is
postpalatal ((, 8, )) (rather than fully palatal, (, , N)), and that (j) remains.
12.2.6.2. All other consonants have no palatalization (although certain linguists and phonologists state the contrary, because they carry theorism to an ex-
12. japanese
381
cess). So, we regularly have: (mi, |i, i) /Ri/ ri (mi, pi) /mi, pi/ mi, pi (bi, 6i)
/bi/ bi. us: (326i'2mM) /obieRM/ obieru (3mi'n2mi) /minoRi/ minori (3i's2k) /RisokM/ risoku (i3i) /siRi/ inri (2pi2q) /epitM/ enpi_u
In addition, /0j/ (0j) CyV, ChV\ (mj, |j, j) /Rj/ ry (mj, p) /mj, pj/ my py
(bj, 6j) /bj/ by: (3j) /eRjo/ enryo (jMM) /RjMM/ ryuu (3p'p:2k) /RoppjakM/ roppyaku (bj) /bjoo/ byoo (bM2mj2k) /bMmjakM/ bunmyaku
(3Qimj) /zaiRjoo/ zairyoo
Gairaigo loanwords
12.2.7. As in any language, even in Japanese, loanwords (of which about 10,000
are of English origin) require some adaptation to the syllable structure (which is
based on morae in Japanese) and new phonemic combinations for new sounds.
For these typical adaptations, let us consider two examples: (2kMm36M) /kMRabM/ club (3s[]tmi3[i]) /sMtoRaiki/ (workers') strike (3s[]tmi3k[]) /sMtoRaikM/ (baseball) strike. It is clear how the Japanese syllabic structure changes
original monosyllables, with consonant clusters, into actual polysyllables.
Among new combinations, in the traditional phonotactics (which is quite rigid
and with a fairly limited number of possibilities), the most common are: (paa3ti)
/paat-i/ party (3di'm3k[]3taa) /d-iRekMtaa/ director ([]3i) /tjezi/
change ([j]t3t:) /djetto/ jet ('[j]3F[]) /sjehM/ chef (3FM'i2mM2mM,
-3mM3mM, 'i-, 'Fi-, 'Fi-) /hMiRMmM, h[M]iRMmM/ lm (i3F3~ii) /sihMonii/ symphony (3kq3n) /katMoone/ canzone.
Some examples clearly show that, besides placing some phones into new combinations, certain sequences are slightly denipponized (as the possible dropping,
in these words but not in genuine ones, of (j) after prepalatal {rounded} articulations), becoming slightly more international.
Structures
12.3.0. In particular, we will deal with Japanese akusento, which is determined
by pitch (even if it does not lack a stress component), and with intonation, which
is superimposed to akusento, slightly changing it.
A typical Japanese pronunciation shows a particular kind of voice with a paraphonic setting with lowered larynx @, especially for men.
Taxophonics
12.3.1.1. e basic things have already been said. In fact, we have seen devoiced
vocoids ( 12.1.3), and the few taxophonic characteristics regarding Japanese
consonants.
382
Gemination
12.3.1.2. We must state that a mora coincides with a light syllable, as the one
formed by a short vowel (/i, e, a, o, M/), or by //, or else by the rst element of a
geminate consonant (/-0-0/ (0-0:)).
A half-heavy syllable corresponds to a geminate vowel (/ii, ee, aa, oo, MM/) or a
diphthong, or to a short vowel + // (//) or + the rst element of a geminate consonant (whose second element belongs to the following syllable, together with its
vowel).
Instead, a heavy syllable presents a geminate vowel (or diphthong) + // (//)
or + the rst element of a geminate consonant: /0[-0]/.
As we have seen from various previous examples, in Japanese, vowel length is distinctive, short and long or rather geminate (or doubled) vowels: (s) /so, so
so (one mora), (s) /soo, soo/ soo (two morae); ('t3i) /toki/ toki, (t3i) /tooki/ tooki. Even consonant length is distinctively present, as in (3i't) /kite/ (from kiru /kiRM/ to wear) and /kite kite (from kuru /kMRM/ to come), both with two
morae; (3it't:) /kitte[]/ (stamp) and ('it3t:, 3it't:) /kitte kitte (from kiru /kiRM/ to cut), both with three morae: /ki-t-te/; (3g'k) /gaka/ gaka artist, (3gk'k:) /gakka/ gakka lesson.
From a phonetic point of view, a doubled consonant consists of two morae: the
rst one coincides with the rst element of the gemination (even if it is actually
shorter, (t)), whereas the second one (which is decidedly longer, since actually
lengthened, (t:)) constitutes another mora together with the vowel that follows it:
(it-t:) (although here we have omitted the pitch features given above). In fact,
/sotto/ is (3st't:) sotto softly (while a similar Italian word sotto /'sotto/ under
is ('sot:to) in an intoneme, or ('sotto) in a preintoneme). Let us observe well and
listen even more carefully to the dierence between (0:0) and (00:). Both for
Japanese /sotto/ or for Italian /'sotto/, however, we always have two phono-syllables, even if Japanese /sotto/ has three morae.
Moraic n is always postvocalic, but it can also be followed by a vowel (and it
is transliterated as Vn' V in order to make people realize we are dealing with //
(P-), not with VnV /n/ (-n), and the same goes for Vn'yV /j/ (P-j),
which is dierent from VnyV /nj/ (-~j)): (t3d) /tedoo/ tendoo (, a similar Italian word, tendo /'tEndo/, or ('tEn:do) in an intoneme).
However, in Japanese, in addition to // (P-) and /n/ (-n), as in (P3i)
/ai/ an'i ('3~i) /ani/ ani, we can also have // (-) (which is the combination of // and //): (N2Ni) /ei/ engi and /n/ (-n) too (combination
of // and /n/): (2n) /ana/ anna (3~i) /ani/ anni ( Italian: /'anna, 'anni/ ('an{:}na, 'an{:}ni) Anna, anni] Let us add this not useless example: (hP3j)
/ho-ja/ hon'ya
erefore, the structure of // (, , , N, , P) (one mora) is dierent from the
geminate one /0-0/ (two morae, or three including the rst vowel: /-0-0/).
However, there is no dierence for the counting of morae.
12. japanese
383
Japanese accent akusento

12.3.2.1. Both phonetic and phonemic transcriptions, as we have seen in the
previous section too, indicate that in Japanese akusento is actually a pitch accent.
We are not faced with real tones (and tonemes), as in Chinese or cetnamese
(languages where even gliding or compound movements on each syllable are prevailing). On the contrary, in Japanese a pattern stretches over whole words, or
whole rhythm groups formed by one or more words and by their (strictly connected) enclitic functional syllables (grammatical particles).
e term akusento (23k[]s3t) /akMseto/, indicates the point, the mora, after which pitch is lowered, that is the change from mid to low pitch. In our
transcriptions, this is marked by writing after the mora in question. Any other
morae before the akusento have mid pitch, except for the very rst one, which is
low. If a word or rhythm group has no akusento, the rst mora is low, whereas all
the successive ones are mid; thus without going back to low pitch, according to
the pattern that follows shortly (which is limited here to four morae).
12.3.2.2. Only in the following table (which relates to 12.3), we will show a
tonemic () and a tonetic pattern (, where (3) indicates a low-pitched syllable,
with the vowel timbre of /o/; whereas (2) indicates mid pitch). We consider them
to be more useful and convenient, in order to describe and learn/teach. We add
pattern () that is most recommendable in transliterations which do not ignore akusento, when no transcriptions are used.
According to the general principles of not explicitly writing unmarked prosodic elements, in transcriptions, the notation (2) for mid pitch could or should be
left out. However, it is certainly more useful to show it, all the more so because in
actual examples it is much less obtrusive than in the table. In any case, in pattern
() it has been left out.
For useful comparisons, we will add the most widespread patterns used in transliterations (, ), and the one used in katakana moraic transcription () with typically oriental graphic complexities. We also show a phonotonetic pattern turned
into a more orthodox one (), originating from pattern (). To indicate any mora, here we use (), /o/, o $ (the last one to indicate katakana, in ):
/oooo/
(3222)
oooo
oooo
oôoo
$999
()
(3)
/oooo
(3222)
ooo
oooo
oôoo
$990
()
(3)
/oooo/
(3223)
ooo
oooo
oôoo
$90$
()
(33)
/oooo/
(3233)
ooo
oooo
oôoo
$0$$
()
(333)
/oooo/
(2333)
ooo
oooo
ôooo
0$$$
()
(333)
384
12.3. Pitch-accent patterns.
/oooo/
(3)
/oooo
(3)
/oooo/
(33)
/oooo/
(333)
/oooo/
(333)
12.3.2.3. In type- tonetic transcriptions, the rst instance (/oooo/) diers from
the second (/oooo), because for the latter we also indicate the succeeding lowering (which for obvious reasons is absent in type- transcriptions). Indeed, it is
not objectively present in actual reality, when no words follow (as we will see shortly). In type- transliterations, the accent could even be acute [] but the grave one
is to be preferred since it can show actual movements better in fact, the pitch
falls from mid to low.
Type- and type- transliterations reect the rst ones in a more abstract way:
tonemic and tonetic (, ); even the katakana transliteration or transcription
() shows the same characteristic, but in a more abstract way in comparison with
real transcriptions (, , ).
We do not use transcriptions of the type /oooo/ (^), which some use
though (on the wake of type and type- transliterations), because if syllables/morae were really pronounced in a high pitch, instead of the mid one, the result
would not be at all convincing. Let us add that in certain textbooks it is possible
to nd both types and with katakana, and type with transcriptions.
As far as the indication or pitch variants is concerned, as we have done in
12.0.2, for hiragana ((3i'm2n, 3i'm3n 3i'm2&n) /hiRaana -na) and katakana ((3k't3kn, 3k't2kn) /katakana/), the most important thing is to indicate them (unless there are usage dierences). Our phonotonetic transcription
also shows their (wider or narrower) diusion and recommendability (which is to
be interpreted gradually from the rst onwards).
On the other hand, in a pronouncing dictionary, which would be worth publishing (using a transliteration, followed by the ocial writing and, of course, by an
IPA transcription, certainly a phonemic one), preferences could and should be
shown, by indicating /hiRaana, hiRaana, hiRaana and /katakana, katakana/,
obviously in a shortened form: /hiRaana, -Ra- -na and /katakana, -ka-/. Of
course, in an introduction, the precise phonetic and tonetic characteristics would
be fully treated, with accurate transcriptions, and with all the necessary vocograms, orograms, and tonograms, .
12.3.2.4. A short mono-moraic syllable may have two pitch possibilities: /ne/
ne sound, tone (absence of akusento), /ne ne root (presence of akusento), but
tonetically they are both non-low ( said in a mid pitch: ('n); the same goes for
/ki/ ki spirit, /ki ki tree, ('i), and /ha/ ha leave, /ha ha tooth, ('h).
In the case of two morae, we can have (s) /soo/ soo monk, (s) /soo/ soo
villa, inn, which are monosyllables realized right as (s), with half-low pitch
but slightly falling or rising, respectively, since they combine mid and low, or low
and mid, pitch ( 12.4).
12. japanese
385
In unstressed bimoraic syllables, we nd (3) (which is low, but raised up to the

border with the mid band, as can be seen from the gure, in comparison with the
low pitch of ()), as in: (3sPs) /sesee/ sensei (k3ii) /koohii/ koohii.
For two syllables formed by adding to a monomoraic monosyllable particles
such as (, ) /a, a/ ga wa which are akusento-less (since their pitch depends
on what precedes, even if {when pronounced metalinguistically in isolation} they
are obviously (', ') /a, a/), we obtain respectively: (3n') /nea/ ne ga and
('n3) /nea/ ne ga, (3i') /kia/ ki ga and ('i3) /kia/ ne ga, (3h') /haa/ ha
ga and ('h3) /haa/ ha wa.
It is fundamental not to believe that Japanese has two tonemes low and mid.
As a matter of fact, akusento is not at all a kind of actual pitch height (as it is not
stress either). On the contrary, it is a pitch fall. It is a sort of catatonic point, after which the pitch falls, passing to the low band, as the examples clearly show.
But above all, akusento is either present or absent. In English, (the position of)
stress is phonemic; whereas it is not so in Japanese. Besides, in English, pitch depends only on intonation (and paraphonics); whereas, in Japanese, pitch is phonemic and fundamental.
is language has mid pitch until an akusento comes, after which the pitch becomes low. If no akusento occurs, the pitch remains mid. On the contrary, stress
in Japanese is not phonemic and depends on a complex interplay of various factors, such as the presence or absence of akusento, where it is placed, and the syllable structures of the rhythm group.
Naturally, the tonograms suciently highlight that, in a rhythm group (or in
an isolated word), the rst mora is low and contrasts with the second one, which
is mid; provided that (as we have already seen) the rst mora is not followed by an
akusento, in which case it is mid and what follows is low.
Again with two morae, we also have ('h3n) /hana/ hana edge, (3h'n) /hana
hana ower, (3h'n) /hana/ hana nose (all bisyllabic), as for the monosyllabic
examples with a particle seen above.
As soon as we add a particle, the eect of akusento is immediately clear: (3n')
/nea/ ne wa sound, tone, ('n3) /nea/ ne wa root, ('h3n3) /hanaa/
hana ga edge, (3h'n3) /hanaa/ hana ga ower, (3h'n2) /hanaa/ hana ga
nose; and so on.
12.4. Movements in syllables with morae of dierent pitch.
/, / =
/, / =
(, P),
(3, 3P)
386
Stress in Japanese
12.3.2.5. Although stress is not distinctive in Japanese, nevertheless it has a fundamental phonetic function. On the other hand, when acculturated native speakers talk about Japanese accent, they surely mean pitch accent akusento which is
distinctive. However, in an automatic way, even non-acculturated natives inevitably use dierent degrees of stress for the various syllables which form sentences.
Since stress is not distinctive, it can oscillate and shift in sentences, phrases, and
rhythm groups. is can also depend on communicative, pragmatic, paraphonic,
and emotional factors. It can even change according to which monosyllables are
added.
However, we will give some indications about the phenomenon of stress, since
we believe it is impossible to continue ignoring it; although this is exactly what
still happens. Let us now proceed in order and start from monosyllables, by reecting on the fact that bimoraic words, as the following, are actually monosyllables
(in spite of contrary confused indications): (ii) /ii/ ii (M) /aM/ au (bi) /bai/
bai (k) /koe/ koe () /kjoo/ kyoo (bMP) /bM/ bun (i) /ooi/ ooi
(baai) /baai/ baai (bj2iP) /bjooi/ byooin (this last example has four morae,
but not four syllables, rather only one! English ('g, 'g) /'gOUI/ going).
In these examples, a stressed syllable is always half-low, but it is slightly falling
(since it derives from the combination of mid and low pitch, within the same syllable), except in the last two examples, where it is slightly rising instead (since it derives from the combination of low and mid pitch, tautosyllabically 12.4).
12.3.2.6. True problems begin with bisyllables, though. In fact, there are dierences between (3'm) /ame/ ame candy, ('3m) /ame/ ame rain, and (3n'mM)
/neRM/ neru to sleep, ('n3mM) /neRM/ neru annel. Bisyllables of two morae,
that is with two light syllables, are stressed on the second syllable, unless akusento
follows the rst one, which is then stressed: (3k'k) /koko/ koko (3'Bi) /azi/ aji
(3M') /Me/ ue (3i'M) /iM/ iu (3'i) /oi/ oi and (3't) /oto oto (3'i) /asi ai
(3mM'm) /mMRa mura (3qM'Ni) /tMi _ugi (3i') /sio io (3i') /ie ie However, we have: ('d3m) /doRe/ dore ('3i) /aki/ aki ('qM3m) /tMma/ _uma .
ree-mora bisyllables are stressed on the heaviest syllables ( with more morae
than others), although there are some oscillations that we will indicate. It is important to accurately observe akusento dierences (which are pitch dierences), in
phonemic transcriptions, since sometimes they are the only actual dierences (but
12.4 must be carefully considered): (3'mi) /omoi/ omoi (3mi) /omoi/ omoi
(3k's) /kasoo/ kasoo (3ks) /kasoo/ kasoo (3i'mi) /iRai/ irai (2imi) /iRai/ irai
(3'i) /aoi/ aoi (3i) /aoi/ aoi (3i'i) /siai/ iai (3i'n) /kinoo/ kinoo (3j't)
/jotee/ yotei (2ii) /hiai/ higai (but ('i3Nj) /kijoo/ kigyoo] (2kMm) /kMRoo/ kuroo (2bm) /baRee/ baree
More: (3iki) /kikai/ kikai (3ikP) /sike/ iken (2ikP) /ike/ iken (3i'kP)
/ike/ iken (2gzP) /goze/ gozen (2QM6P) /zMbo/ zubon (k2Bi) /koozi/
kooji (k3Bi) /koozi/ kooji (N2Ni) /ei/ engi (i2d) /aida/ aida (hi3mM)
/haiRM/ hairu (3t) /kjooto/ Kyooto (ki2Ni) /kaii kaigi (di3k) /dai-
12. japanese
387
kM/ daiku (b3i) /beRi/ benri (mi2n) /mina/ minna (3miq'q:M) /miqqM
mit_u (3k'k:) /akka/ akka (but: ('k3k:) /akka/ akka ('3:j) /Ressja/ resa
where pitch prevails on other factors).
Generally, in four-mora bisyllables, stress falls on the rst syllables, unless it is a
light one ( with just one mora) or there is an akusento after the second one (or if
the rst is only half-heavy, checked by (==:)): (jMM26j) /zjMMbjoo/ juubyoo (jMM36j) /zjMMbjoo/ juubyoo (h2k) /hookoo/ hookoo (jMM3d) /zjMMdoo/ juudoo (k2Bj) /koozjoo/ koojoo (3kBj) /koozjoo/ koojoo (3tiFMM) /taihMM/ taifuu
More: (k2t) /ketoo/ kentoo (3kt) /ketoo/ kentoo (s2t) /setoo/ sentoo (s3t) /setoo/ sentoo (3sPs) /sesee/ sensei (iP2P) /kie/
kin'en (jMM26iP) /jMMbi/ yuubin (3gk'k:) /gakkoo/ gakkoo (3t':MM) /tokkjMM/ tokkyuu (but: (b3:P) /bottja/ boc>an] (3'iiP) /ekii/ ekiin (3ip'p:i, 3i-) /sippai/ ippai (t2) /tookjoo/ Tookyoo
12.3.2.7. ree-mora trisyllables are stressed on their second syllable, unless it
contains a devoiced vowel, which makes stress shift forwards (if an akusento is
there) or backwards: (3i'm2n) /kimono/ kimono (3jM'6i2) /jMbia/ yubiwa
(3i'k2m) /tikaRa >ikara (3F't2mi) /hMtaRi/ futari (2m3i) /aRasi/ arai
Besides: (3k'2mi) /kaeRi kaeri (3k'2mM) /kaeRM/ kaeru (3k'3mM) /kaeRM/ kaeru
(but (k3mM) /kaeRM/ kaeru), (2t3mM) /taoRM/ taoru (3M'2i) /Meki/ ueki (3i'k2kM) /sikakM ikaku (2kz3k) /kazokM/ kazoku (with this pitch pattern,
and especially with non-devoiced /i, M/ in the last but one syllable, it is also possible to have: ('d3i3m) /dotiRa/ do>ira ('d3M3i) /deMti/ degu>i] ('k3i3i)
/kesiki/ keiki ('i3k3qM) /ikMtM/ iku_u ('p3s3t) /posMto/ posuto (32F'kM) /ahMkM/ wafuku (32i't) /asita aita (3j2k'j) /jakMsjo/ yakuo
Four-mora trisyllables are stressed on the rst syllable, unless it is a light one:
(ki2m2n) /kaimono/ kaimono (2s2k) /oosaka/ Oosaka (saa36i3s) /saabisM/ saabisu (j2Bi3i, -2Bi2i) /sjooziki oojiki (t2pM2m) /tepMRa/ tenpura (Q3k3k) /zekokM/ zenkoku (2im3t) /siRooto/ irooto (3bM'qM2z)
/bMtMzoo/ butuzoo (3s'i2~iP) /sekini/ sekinin (2k3P) /akatja/ aka>an
(3m'M2mi) /maeMRi/ maeuri (3'i2i) /oiai/ oiwai (3M'2i) /Mekija/ uekiya
More: (3sM'i2sP) /sMise/ suisen (but: (3'3m) /ooame/ ooame (3'm3Bi)
/Roomazi/ roomaji (3kP'3F) /kaohM/ kangofu (3's3k, -'s2kM) /RoosokM roosoku (3jMM'2qM) /zjMMatM juuga_u where the akusento prevails).
Regularly, also: (3~i'hP2) /nihoo/ Nihongo (3p'i2k) /patiko/ pa>inko
In compounds, usually we nd a seemingly irregular pattern, such as: (3b'm2i)
/bamesi/ banmei (3gi'k2kM) /gaikokM/ gaikoku (3jMM'd2i) /jMMdati/ yuuda>i (3hi'z2m) /haizaRa/ haizara
12.3.2.8. Four-mora quadrisyllables tend to be stressed on the second syllable
from the beginning: (3k'mi2d&m) /kamidama/ kamidama (3n'2M2q, -&qM)
/naaMtM/ nagagu_u (3n'mi2m&n) /noRimono/ norimono (3bM't2~i&kM, -ik)
/bMtanikM/ butaniku (3s'k2n&j) /sakanaja/ sakanaya (3i'&s) /siaase/ iawase (3'm2mi&k) /ameRika/ Amerika (3M'mi2i&m) /MRikiRe/ urikire (3t'm2d&i)
388
/tomodati/ tomoda>i (3M'k2q2k, -2qM&k) /MketMke/ uke_uke (3i'k3i3qM,

-iqM) /tikasitM/ >ikai_u (3'2k3ki, -2kMki) /oakMkai/ ongakukai
More: (3kM'd3mn) /kMdamono/ kudamono (3t'6M3kMm) /tabMkMRo/ tabukuro (3k'n2zMi) /kanazMti/ kanazuti (3i'm2i&m) /hiRosima/ Hiroima
(3h'n2i2t, -i&t) /hanasite hanaite (3t'n2i&mi) /tanosimi/ tanoimi (3k'n2m&i) /kanemoti kanemo>i (3i'n2mM&mi) /inemMRi inemuri (2i'tP2j&j)
/zitesjaja/ jitenaya (3hP's2i3i) /hosekiti/ honseki>i
However, stress generally falls on the last but one syllable, when it is a heavy
one, or is followed by an akusento, or when the second one contains a devoiced
vowel. Mainly, this also happens with nal -_u -ri or in obvious compounds: (2s'n36) /asaneboo/ asaneboo (gi2k'kM3BiP, -3iP) /gaikokMzi/ gaikokujin
(2mM6i3t) /aRMbaito/ arubaito (g2qMj36i) /getMjoobi/ ge_uyoobi (m2kMj36i) /mokMjoobi/ mokuyoobi (&bi3mMdi3M) /biRMdiM/ birudingu
(~i2'kaa3m) /niakaame/ niwakaame (2km3t) /tjokoReeto/ >okoreeto (MM2'kM2) /tjMMokMo/ >uugokugo (but (3gi'k2kM&) /gaikokMo/ gaikokugo]
In addition: (32F'kM2m, 2FM-) /ohMkMRo/ ofukuro (3j2k's2kM, j2kM-)
/jakMsokM/ yakusoku (3k2i'k2t, k2i-, -3t) /kakikata, -ta/ kakikata (3j2q'k2d, j2qM-) /jotMkado/ yo_ukado (t2i'2mi) /tosijoRi toiyori (k2mi'n2mi, -3mi)
/kaminaRi kaminari (s2mi'2qM, -2q) /saRaietM/ saraige_u (h2i'2qM)
/hatiatM hatiga_u (i2i'2qM, i-) /sitiatM itiga_u (jMM2~i'2qM)
/zjMMniatM juuniga_u (2s'm2i) /asamesi/ asamei (i2mM'm2i) /iRMmesi/
irumei (2nn3k) /onanoko/ onnanoko (dN2igi3j, -i-) /dekigaisja, -ai-/ denkigaia (i2i'd2i, 3i-) /hikidasi/ hikidai (k2i'd2i) /kasidasi/
kaidai (h2t'mi2i) /hatamiti/ hatami>i
12.3.2.9. In ve-syllable words, which are generally not simple words, stress decidedly tends to fall on the last but one syllable, except for particular compounding or devoicing: (3k&t2zM'k3mM) /katazMkeRM/ katazukeru (3i&n26ik3mi) /inabikaRi/ inabikari (3&t2kn3k) /otokonoko/ otokonoko (3n&Bi2m3i) /nezimaasi/ nejimawai (3j&m2n63mi) /jamanoboRi/ yamanobori (3&kM2mi'm2n) /okMRimono/ okurimono (3&2kM's2m, 22k-) /okjakMsama/ okyakusama (3&t3qMdi3sP) /otetMdaisa/ ote_udaisan (3&m2mi'k3BiP, -3iP) /ameRika zi/ amerikajin
More: (3s&t2kM'm2n) /setakMmono/ sentakumono (3i&i2~i'i2BjMM) /itinitizjMM/ i>ini>ijuu (dN2i2s't36M) /dekisMtoobM/ denki-sutoobu
(MM22kMmj3mi) /tjMMokMRjooRi/ >uugokuryoori (k2djMM3t3k) /koodazjMMtakM/ koodan-juutaku (d2b's3qM36i, d2b 's-) /daboosetMbi/ danboo-se_ubi (2m'2mi&sP) /omaaRisa/ omawarisan (3~jMM&2kM2ikP) /njMMakMsike/ nyuugaku-iken (m2ii2i3qM, -i3q) /matiaisitM/ ma>iaii_u
12.3.2.10. In verbs in -ru generally stress is on the preceding syllable and akusento immediately after it: (26'3mM) /oboeRM/ oboeru (k2'3mM) /kaaeRM/
kangaeru (2qM'm3mM) /atMmaRM/ a_umaru (i2m'63mM) /siRabeRM/ iraberu
12. japanese
389
(k2i2k'k3mM) /kosikakeRM/ koikakeru but (3'm23s) /aRaasM/ arawasu

(3i'k2z3k) /tikazakM/ >ikazaku
Usually, adjectives are stressed on the last, or last but one, syllable: (3&t2tki)
/atatakai/ atatakai (mM2iqMi) /mMsiatMi/ muia_ui (3j&k2mii) /jakamasii/ yakamaii (3'6M2ni) /abMnai/ abunai (i2ii) /oisii/ oiii (ii2mi)
/kiiRoi/ kiiroi (3'k2mMi) /akaRMi/ akarui (3qM'm2ti) /tMmetai/ _umetai but
there are even cases like: (3mM'zM2k&ii) /mMzMkasii/ muzukaii (3't2mii) /ataRasii/ ataraii (3'm2imi) /omosiRoi/ omoiroi (3mM'i2qMi) /mMsiatMi/
muia_ui (3m'd2ksi) /medookMsai/ mendookusai
In iterated forms stress and akusento belong to the rst syllable: ('mM3zMmM3zM) /mMzMmMzM/ muzumuzu ('3z3z) /azaaza/ wazawaza nally,
here are some akusento-less forms: (3s'RM, 3qM'RM, M'kRM) /saaRM, tMaRM, MkeRM/ sawaru _ugaru ukeru
12.3.2.11. Since there are no real rules for akusento and stress in compounds,
especially for less long ones, we just give some examples, to rouse reection (also
on the paramount importance that a real pronouncing dictionary would have {by
using IPA} for akusento stress, segments, devoicing, ): (3FMjM'3[i]3i) /hMjMesiki/ huyugeiki from (3FM'jM) /hMjM huyu ('k3[i]3i) /kesiki/ keiki
Other examples: (3jM&~jMM2kM'd3m3n) /jMnjMMkMdamono/ yunyuukudamono from (3jM'~jMM) /jMnjMM/ yunyuu (3kM'd3m3n) /kMdamono/ kudamono and more: (jMM2j2k'z3m) /jMMjakezoRa/ yuuyakezora from (jMM2j2k)
/jMMjake/ yuuyake ('s3m) /soRa/ sora (3s&t2k3m) /satookoRo/ satogokoro
from (3s't) /sato/ sato (3k'k3m) /kokoRo/ kokoro.
Furthermore: (3~i'M3mM3m) /niMRMma/ niguruma from ('~i) /ni ni (3kM'mM2m) /kMRMma/ kuruma (3's3k3z) /asakaze/ asakase from ('3s) /asa/ asa
(3k'z) /kaze/ kaze (3iP2s[]&t2tk3ii) /isMtatokoohii/ insutantokoohi
from (3iP2s[]t3t) /isMtato/ insutanto (k3ii) /koohii/ koohi (gi2k'kM3BiP, -3iP) /gaikokMzi/ gaikokujin from (gi2k2k) /gaikokM/ gaikoku
(iP) /zi/ jin (gi2k&kM2Bit3m3k, -2i-) /gaikokMzitooRokM/ gaikokujin-tooroku from (gi2k'kM3BiP, -3iP) /gaikokMzi/ gaikokujin (t2m2k)
/tooRokM/ tooroku
12.3.2.12. In forming Japanese sentences, some modications are introduced
regarding akusento (and somehow for rhythm-group stress too). Let us consider
the following examples, in order to see how they work, by carefully observing the
pitch of the syllables in the second rhythm group. In fact, normally, the rhythm
groups that are not separated by pauses, after a rhythm group with akusento, remain mid-pitched as the previous one. Here are some sentences: (3't2k2i&n
2n'm) /atakMsino namae/ watakui no namae (in (3n'm) /namae/, na- becomes (2n), by pitch assimilation to the preceeding syllable), ('b3kMn 3sPs)
/bokMno sesee/ boku no sensei (in (3sPs) /sesee/, -see becomes low), (3~i'h
'i3zM) /niho tizM/ Nihon >izu (without modications).
More sentences: (2s36P) /asa ba/ asa ban (from ('3s) /asa/ and (bP) /ba/,
by unifying everything into one rhythm group and lowering ban] (3j's2ii hP)
390
/jasasii ho/ yasaii hon (without modications), (ii 3Bi6i3i, 3i-) /ii zibiki/ ii
jibiki (from (3i'6i2i) /zibiki, with lowering of -biki] (3sM'zM3ii 3h) /sMzMii
heja suzuii heya (in (3h') /heja, -ya is lowered), (3sPsd3s, -s:, -s) /sesee
desM/ sensei desu (with stress reduction on ('d3s, -s:, -s) /desM/ and lowering of
de-]
More: (3t't2m 2'm2imi hP) /totemo omosiRoi ho/ totemo omoiroi
hon (from (3'm2imi), with raising of o- and lowering of hon] (3i'Bj2~i 2F&kM2z'qM2n 'm2di) /hizjooni hMkMzatMna modai/ hijoo ni fukuza_una
mondai (with raising of fu- and mon-] ('mt3t: 3jsMi hP) /motto jasMi ho/
motto yasui hon (with lowering of -sui and hon] (3'h2j 2g'zi2m3s, -s:, -s) /ohajoo gozaimasM/ ohayoo gozaimasu (with raising of go-]
Some others: (3'mi33t 3gzi3s, -s:, -s) /aRiatoo gozaimasM/ arigatoo gozaimasu (with lowering of -zaima-\ everything is low, except -ri-] (d3m 3mi33t
3gzi3m3s, -s:, -s) /doomo aRiatoo gozaimasM/ doomo arigatoo gozaimasu
(with lowering of -ri- and -zaima-\ everything is low, except doo- which is half-low).
In addition: (M2i 'k2m '3i m3d| 3&mM2i'm3s, -s:, -s) (in the case of fast
speech with no breaks: (M2i 'k2m '3i m3d 3mM3im3s, -s:, -s); but, separately, word by word: (3M'i, 3k'm, '3i, 'm3d, 3&mMi'm3s, -s:, -s)} /Mti kaRa eki made aRMki masM/ u>i kara eki made arukimasu (please, note the stress change in the
rst rhythm groups, mainly due to /i/ devoicing).
Also: ('~i3Bi&k3m\ s3i m3d| 3i'mM3j3~i\ 3im3s, -s:, -s) possibly with
fewer breaks, in less slow speech: ('~i3Bik3m s3i m3d| 3i'mM3j3~i 3im3s, -s:, -s). On the other hand, in separate rhythm groups, we have: ('~i3Bi, 3k'm,
s3i, 'm3d| 3i'mM3j3~i, 3i'm3s, -s:, -s) (with modications on kara] /nizi
kaRa sazi made zimMsjoni imasM/ niji kara sanji made jimuo ni imasu ('j3Bi
3md3~i\ 3k'k2~i 2i't 2kM2dsi) (slower: ('j3Bi 3md3~i\ 3k'k2~i\ 2i't 2kM2dsi); on the other hand, in separate rhythm groups, we have: ('j3Bi, 3md3~i, 3k'k~i, 3i't, 3kM2dsi), with stress modications on made ni in comparison with
made] /jozi madeni kokoni kite kMdasai/ yoji made ni koko ni kite kudasai
12.3.2.13. Here are further examples illustrating stress modications, in forming rhythm groups: (2md3~i) /madenia/ made ni wa (3'i2n2) /okinaa/ Okinawa but (3&i2n'2) /okinaaa/ Okinawa wa Obviously, it is
not always easy to readily distinguish between the eect of pitch prominence and
stress prominence. On the contrary, when mid pitch coincides with stress, prominence is quite clear.
If all this combines with a fairly heavy syllable, prominence is even more evident. In any case, if several nearby syllables share the same characteristics, it becomes less easy to distinguish clearly.
However, pitch remains the most important element, being the distinctive one,
though undoubtedly stress has a considerable role. It is important to nd an appropriate balance between the elements, although oscillations are quite possible and
normal, indeed.
In an example like (t26'm3i3tk31) /tabemasitaka?/ tabemaita ka? the low
pitch and secondary stress on the syllable ta- and the mid pitch and weak stress on
12. japanese
391
-be- may give a similar prominence eect. But, of course, it is inferior to that of the
syllable -ma- and decidedly inferior to that of ta and even less so than in the syllable -i- with devoicing up to the loss of syllabicity: (3i, 3:, 3).
In the case of loanwords, the interplay of pitch and stress (as well as of syllabic
weight and akusento placing) often undergoes some hierarchic reversal, by moving closer to the original (stress) pattern: (k3ii) /koohii/ koohii (coee), (2ht3mM) /hoteRM/ hoteru (hotel).
12.3.2.14. Obviously, in current speech, some reduction phenomena occur in
Japanese too. Here we will consider some of the most normal ones. e particles
no and ni in particular, are often reduced to //: ('kM3mM 'n3m) /kMRMno naRa/
kuru no nara (i2miP 2M'i) /kimino Mti kimi no u>i (g2i 'n3mM) /gekini naRM/ genki ni naru In the negative, forms with -r-V-nai change /R/ into //:
(3i'ni) /siRanai/ iranai (32kM'ni) /okMRenai/ okurenai More frequently,
there can even be contractions such as: (2miM) /mite simaM/ mite imau
(2jjM) /jode simaM/ yonde imau
Intonation
12.3.3.1. en actual Japanese sentences are pronounced, the pitch of their
single rhythm groups is partially modied by intonation (too), which is added
with its characteristics, according to the four types of intoneme.
12.5 shows the Japanese preintonemes and intonemes, with their characteristics. Every preintoneme forces the pitch of its individual tones, by canalizing them
into the indicated shapes (which, in more sophisticated notations than necessary,
could even be marked with small rings, as we will see shortly). A normal preintoneme is compressed (/ / ( ), (( ))), the interrogative is raised (/ / ( ), (( ))), whereas the imperative one is falling (/ / ( ), (( ))); nally, an emphatic preintoneme is
non-compressed, (/ / ( ), (( ))).
Again in 12.5 it is possible to see the modications of the intonemes: the conclusive is falling (/./ (13)), the interrogative is rising (/?/ (31)), the suspensive is extended (// (^)), whereas the continuative intoneme is compressed (/,/ (2)).
12.6 shows the modications undergone by light and heavy syllables (on two
morae of dierent pitch { 12.4}), when the four Japanese intonemes superim 12.5. Japanese preintonemes and intonemes.
/ / ( ) (( ))
/./ (13)
/ / ( ) (( ))
/?/ (31)
/ / ( ) (( ))
// (^)
/ / ( ) (( ))
/,/ (2)
392
pose on them. Mainly the last syllable of an intoneme undergoes these changes,
whether it be stressed (as in these examples) or unstressed. As can be seen, a conclusive intoneme makes the last syllable lower and slightly falling; an interrogative
one makes it raise by giving it a clear rising movement. A suspensive intoneme
practically does not modify anything; whereas, a continuative one slightly compresses it towards the mid band.
Before moving to the illustrastive examples, let us observe that, in Japanese,
questions are made by adding the particle ka (k) /ka/ at the end, and using an interrogative intoneme with total questions, but a conclusive (or a continuative) one
with partial questions. is is the most recommedable and common pattern. However, since ka is very recognizable, as an interrogative element, a simple conclusive
intoneme can be used, even with total questions, or an interrogative one even with
partial questions. Lastly, above all colloquially, ka need not necessarily be used, in
which case then an interrogative intoneme is necessary with total questions.
Lastly, here are three examples to illustrate the most recommendable use of
marked intonemes:
/./: (3&k2mi2msd3i3t13) /akaRimase desita./ Wakarimasen desita.
/?/: (3&k2mi'm3i3t3k31) /akaRimasitaka?/ Wakarimasita ka?
//: (3FM'jM dt3t: 3km3d^ 3i2i'm3i3t13) /hMjM datta keRedo ikimasita./ Fuyu datta keredo, ikimasita.
Typically, Japanese has a peculiar paraphonic setting, with lowered larynx @,
mostly for men.
12.6. Pitch movements for light or heavy Japanese syllables, with the four intonemes.
// =
(13)
(31)
(^)
(2)
/'/ =
('13)
('31)
('^)
('2)
// =
(13)
(31)
(^)
(2)
/'/ =
(13)
(31)
(^)
(13)
Text
12.4.0. e story e North Wind and the Sun follows, given in four dierent
normalized versions. We start with the (neutral) Japanese pronunciation of (neutral British) English this is the rst step of the phonetic method (the written text
is given in 2.5.2.0). e Japanese translation follows, in its neutral phonotonetic and phonotonemic versions, since it is important in this language to see its akusento and the actual phonotonetic rendering, to make useful reections.
At the end, as always, there is the version which gives the English pronunciation of Japanese, by neutral British speakers, uent in Japanese (after prolonged
12. japanese
393
Speakers of American English could prepare their own version both of the Japanese pronunciation of English and of their pronunciation of Japanese, as an excellent exercise, by listening to native speakers, best of all after recording them. Of
Japanese pronunciation (of English)
12.4.1. (3Q'n2sM 'i2d2 32Q'sP2 32i2s'pMM2iP 'i2i 32z2s2t'm2g13| 'P 32t'm26M2m2 'k2mM 'mP2 'm2p2t 3i2n'2mM 3k'm2k13|
3Q22M'mii2d2\ 3Q2t2z'P 3FM'h2mM2s2t 3s2k'ii2i2d2 3i'm2iP 3z2t'm26M2m2 t2i 2i2zM2k'm2k '2F^| 'M2d 36i2kP'i2d2d 3s2t'm2g
3z2i'2z13||
Q2 3Q'n2sM 'i2d2 36M'mMM2 32zM'haa2d2 32Bi'kM2d13| 6t2t: 3z'm
3i26M'mM^| 3Q'm 3k'm2sM2mi2 i2d2z2t'm26M2m2\ h2mM2d 3i2zM2k'm2k 3'mM2i2m13| 32d2t'maa2s2t2\ 3Q'n2sM 'i2d2 g '6p2p:
2Bi't2p2t13|| Q2 3Q'sP 'P 'M2t13 2mM2m13| 32i'mii2j2t2m2\
3Q2t'm26M2m qM2k '2F13 3i2zM2k'm2k13|| 3's2 3z'n2sM 'i2d2
32z26M'mi2Bi2d 3qM2kP'h2s2| 3Q2t's13 32z2s2t'm2g13 3262z'qM13||
i2BjM'mi2i31 3z2s't2m2| 3jM' 3t'i2mi 3qM'P31|||)
Japanese text
12.4.2. Arutoki Kitakaze to Taiyoo ga >ikara-kurabe o imaita. Tabibito no
gaitoo o nugaseta hoo ga ka>i to yuu koto ni kimete, mazu Kitakaze kara hajimemaita. Kitakaze wa, Nani, hitomakuri ni ite miseyoo, to, hageiku fukitatemaita.
Suru to tabibito wa, Kitakaze ga fukeba fuku hodo gaitoo o ikkari to karada ni
kut_ukemaita. Kondo wa Taiyoo no ban ni narimaita. Taiyoo wa kumo no aida kara yasaii kao daite, atatakana hikari o okurimaita. Tabibito wa dandan
yoi kokoromo>i ni natte, imai ni wa gaitoo o nugimaita. Soko de Kitakaze no
make ni narimaita.
Kono hanai omoirokatta? Moo i>ido yomu?
394
Phonotonemic transcription
12.4.3. /aRMtoki,| kitakazeto taijooa, tikaRakMRabeo simasita.|| tabibitono
gaitooo,\ nMaseta hooa,\ kati tojMM, kotoni kimete.|| mazM,| kitakaze kaRa,
hazimemasita.|| kitakazea| ^nani,| hitomakMRini site, misejoo,| to| haesikM,
hMkitatemasita.||
sMRMto,\ tabibitoa| kitakazea hMkeba,\ hMkMhodo| gaitooo,\ sikkaRito,\
kaRadani kMttMkemasita.|| kodoa, taijoono bani, naRimasita.|| taijooa,\
kMmono aida kaRa. jasasii\ kaoo dasite,| atatakana hikaRio, okMRimasita.|| tabibitoa,| dada "joi,\ kokoRomotini natte.|| simainia,\ gaitooo nMimasita.||
sokode\ kitakazeno,\ makeni naRimasita.||
kono hanasi, omosiRokatta?| moo itido, jomM?|||/
Phonotonetic transcription
12.4.4. (2mM3t3i2| 3i&t2k'z2t 2ti32 3i&k2m'kM3m36 3i'm3i3t13|| 3t'6i26i&t2n gi2t;2\ 3nM'3s3t h32\ 3k2it3jMM2 3k't3~i 3i'm2t2|| 'm3z2|
3i't2k&z 2k'm2 3h&Bi2m'm3i3t13|| 3i&t2k'z2^| ^'n3~2| 3i&t2mkM3mi3~i3i3t2 3mi's32| 't^|| 3h'3i3kM2 3F&i2t2t'm3i3t13||
3sM'mM2t2\ 3t'6i26i&t2^| 3i&t2k'z2 2F'k362\ 3F'kM3hd^| gi2t;2\ 3ik2k:mi3t2\ 3k'm2d&~i 3kq&q:M2k'm2i3t13|| k2d22 ti33n b3~i2 3n2mi'm3i3t13| 2ti32\ 2kMm3n i2d 2k'm13 3j's2i^\ 3k'2 'd3i3t2| 3&t2tk3n 3i'k2mi2 3&kM2mi'm3i3t13|| 3t'6i26i&t22| d2dP j2\ 3k&k2m2m'i2~i 'nt3t:13||
3i'mi2~i32\ gi2t; 3nM2Ni'm3i3t13|| 3s2k'd^\ 3i&t2k2z'n2\ 3m'k2~i n2mi'm3i3t13||
3k'n 2h'n2i2 3&m2i2mkt3t:31| m 3i2i'd2 'j3m31|||)
English pronunciation of Japanese
12.4.5. ('>uT&kIi1| &khIiTk5zIT ta'jg2 cIi&k>5kh>bI&j S'mSt3 3|| Th5bIib&Tn 'gaT2\ n5gsT 'hg2\ &khc'Thju2 k5Thni k'mTI2|| 'mzu2| k5ThkzI k'>A: h&Zm'mST3 3|| kh&Tk'zw32| ^'nA;ni2| h&Tm'kh>nS&TI2 m'sI&j2| 'Th;32|| h'gSku2 fu&kTT'mST3 3||
s'>uT2\ Th'bIib&T32| kh&Tk5zg f'khb2\ f'khuh&D32| 'gaT2\ Sk'>IiT2\ kh5>A;D&nIi k&tsuk'mST3 3|| 'khnD2 5Than 'bn-i2 &nA;>'mST3 3| Tha'j2\ k5mn 'aD k'>A:3 3 j'sA;SIi32\ khA;';2 'DSTI2| &TT5khA;n h'kh>i&2 &k>'mST3 3|| Th'bIib&T2| 5DnD 'j;2\ kh&k;>m5chini 'nTI3 3|| S'man&wA;2\ 5gaT &ng'mST3 3|| &sk'D;I32\ kh&TkzI'n;2\ mA;5khni
&n>'mST3 3||
kh'n; h'nSi2 &mS>'khT21| 5m; c'D;2 'jmu21|||)
13. Esperanto
13.0.1. Esperanto is a social fact and a living language. It is an easily accessible
language that can be mastered fairly well in a short time. It can be quite useful
both in cultural and working environments, including commerce and tourism.
In particular, Esperanto is an ideal language for science. For a real world circulation, to know and allow all people in all nations to grasp any subject, avoiding the
problems and costs of having good translations done. Furthermore, it is extremely appropriate as a general basis for learning and teaching foreign languages with
all their historical complications, and even to reect on the structures and functioning of one's own national native tongue.
Perhaps, the day is still a long way o when peoples and governments can realize that an international auxiliary language ( a second language for all mankind,
certainly not as a substitute for the present various languages) can have many advantages. All the more so because it is not merely a (rough) vehicular language,
but can also meet every-day needs in addition to technical and scientic demands
as well as general cultural ones. Its most evident advantage is that it is nobody's
mother tongue, but everybody's second language. Everyone has to study it, and
no-one need not do it only because all the others accept to learn and use one particular native language. th a language which is a second one for all, nobody has
any more linguistic privileges that allow them to take advantage of others
Esperanto is not so complicated to learn as other languages are. In fact, its grammar is simple, logical, and regular; with no capricious exceptions so typical of
natural languages. Even lexically, there are no surprises such as child, do, say
/'caEd, 'dUu, 'sEI/ and children, does, said /'cId<n, 'dz, 'sEd/; nor has it semantic
ambiguities caused by polysemy. For this reason Esperanto would be incredibly
appropriate for the Internet. Its only drawback resides in its original orthography,
which uses six letters with particular diacritics that produce dierent combinations from those of other languages: c g (c, G), s j (S, Z), h (x) (w, u). Although
these letters allow to identify Esperanto immediately, they are nevertheless one of
the major limitations towards its circulation through the press and electronic media.
13.0.2. However, an important use of Esperanto for phonetic purposes consists
in utilizing it as an active drill to apply the phonetic method to a reality that presents fewer obstacles than any other language. As a matter of fact, even its pronunciation is regular, without exceptions, and fairly natural, as we will see (although
it is an articial language, a planned language). Of course, this holds for its
spelling too; the real problem is constituted by certain specic graphemes (that
may have dierent values in other languages) and by its particular phones (since
the world's languages have their own phonic systems, with their own rules and par-
396
ticular realizations, which are not necessarily like those of Esperanto).

erefore, a serious study of Esperanto neutral pronunciation is an excellent
training in order to then learn the pronunciation of natural or ethnic languages too.
However, the scanty attention devoted by school and society to the importance
of pronunciation leads even Esperantists to considerably mistreat the pronunciation of Esperanto itself (and that of their mother tongues as well). Actually, people pronounce it their own way, depending on their personal pronunciation of
their own language. erefore, in spite of the simple rules of Esperanto, they start
from their actual pronunciations, by rendering it with the phones and intonation
of their personal (and more or less regional) pronunciation, although they do not
realize it at all. However, this does not take anything away from the scientic value of our phonetic experiment. Contrary to what we have done for the other languages (unless we had to deal with minimal pairs which were useful to our explanation), we will add the meaning of the examples provided, all the more so because
ambiguities are practically almost excluded. Instead, this is what can regularly happen with ethnic languages, for many words (for which it is always problematic and
risky to give glosses).
Vowels
13.1 Esperanto ( E-o] presents only the ve most normal and natural vowel
phonemes (which are the most widespread in the various languages of the world):
(i, , a, , u) /i, e, a, o, u/ ( 13.1). e ve E-o vowels are realized very much
like the ve (stressed) vowels of Spanish, as in: ('iR, 'tRs, 'maR, 'ds, 'tu) /'iR, 'tRes,
'maR, 'ds, 'tu/ ir tres mar dos t. In the most international pronunciation of Esperanto, e o always have the timbres (, ), both in stressed and unstressed syllables. Here are some examples: ('tRiki, 'b-l, 'ka;Ra, 'd;m, 'u;nu) /'trinki, 'eble, 'kara, 'domo, 'unu/ trinki eble kara domo unu to drink, perhaps, dear, house, one.
e ocial Esperanto diphthongs are of the (i, u) /i, u/ type, and are represented with Vj V, even though, of course, other vowel sequences occur, which
from a phonetic point of view are real diphthongs as well ( 3.1.2-3, about Italian
diphthongs, and NPT/HPh: 1.16 2.10-1), as, for instance, /eo, io/ in (&ggRa'fi) /geogra'fio/ geografio geography. Given the nature of Esperanto orthography, a more logical solution Vi, Vu was not available, because of its word-stress
assignment rule (as we will see in 13.3.4).
/i/ (i)
/u/ (u)
/e/ ()
/o/ ()
13.1.
Esperanto
vowels.
/a/ (a)
13. esperanto
397
Consonants
h (H)
()
k g
S Z*
R|(r)
l
velar
palatal
(~)
cG
s z
(l)
(n)
prepalatal
alveolar
laryngeal
(n)
t d
q
velar rounded
m (M)
F pb
f v
_
dental
labiodental
bilabial
13.2. Table of Esperanto consonants.
postalveo-palatal
protruded
13.2.0. 13.2 shows the consonants of original, or traditional, Esperanto (including the rarest two, (Z, x), which are marked with a postponed *. It would be
better to remove them in order to render the language more modern, more functional, even simpler and more easily diused ( 13.4.3-4). In fact, in natural languages a phonemic opposition between (Z, G) or (x, h) is certainly not frequent
nor preferred, unless they belong to complete and related series, or unless /x/ is
made decidedly more energetic (than a simple velar constrictive), or even pronounced as a trill).
In Arabic and in Tuscan, for instance, (Z, G) do not oppose. In German, (x, h)
are not free from problems or regional and social variations; however, they could
be considered to have complementary distribution
1.9-15 show the orograms, grouped by manners of articulation, of all the contoids given in the chapters of this volume (even as secondary, occasional, or regional variants) for the 12 languages treated. is exposition makes the necessary comparisons between dierent languages more immediate.
x*
j
(L)
Nasals
13.2.1. Esperanto has two nasal phonemes, /m, n/ (m, n): ('m;n, 'fndi) /'mono, 'fendi/ mono fendi money, to split. ile /m/ never assimilates: (m'fa;z,
&mmkn'sRv) /em'fazo, memkon'servo/ emfazo memkonservo emphasis, selfconservation, there are various possible assimilations for /n/ (M, ~, ), which render the pronunciation more uent and natural: (iM'f;Ra, 'ma~Gi, si~'j;R, 'SRak) /in'fera, 'manGi, sin'joro, 'Sranko/ infera mangi sinjoro sranko infernal, to
eat, gentleman, cupboard (including (n, ~) ((, n)): ('sndi, 'ma~Gi) /'sendi,
'manGi/ sendi mangi to send, to eat).
However, between lexemes (including prexes) no assimilation takes place (on
the contrary, we also nd secondary stresses, even on syllables contiguous to a
stressed one): (&kun'm;ti, &kun'v;ni, &sn'pa;ga) /kun'meti, kun'veni, sen'paga/ kun-
398
meti kunveni senpaga to combine, to meet, free/for nothing.

Nevertheless, we have: (km'pR;ni, &siamman't;ln) (with a rhythmic secondary stress) /kom'preni, siamman'telon/ kompreni sian mantelon to understand,
one's own mantle (accus.) ( (si'a;mam man't;ln) /si'aman man'telon/ siaman
mantelon (a) Siamese mantel (accus.)).
In normal, current, relaxed pronunciation, the unstressed grammeme /-n/ >-n
(of the accusative case) regularly assimilates (whereas only in formal and solemn
pronunciation can we nd (si'a;man man't;ln) /si'aman man'telon/).
In spite of this, there is a general tendency to keep words separated, both under
the inuence of ethnic languages, and because E-o is a dierent language, not
yet adequately internalized (or free from any interference). thout such rules, inevitably everyone would use one's own most familiar structures since, subjectively, they are thought to be quite natural.
Stops
13.2.2. Esperanto has three diphonic pairs of stops: /p b, t d, k g/ (p b, t d, k
g). Of course, /t, d/ are dental (t, d), not alveolar as in English, since most languages have dental articulations. Besides, g is velar, with no exception, even before
i and e (where it becomes prevelar (()), and /k/ as well (()) as in: ('gfT, 'gT) /'gIft,
'gEt/ gift get}: (gi'gant, &ggRa'fi) /gi'ganto, geogra'fio/ giganto geograo giant,
geography; ('pnsi, ba'bi;li, 'tr;ti, d'qi;di, 'kRa;ki, 'ga;g) /'pensi, ba'bili, 'treti,
de'qidi, 'kraki, 'gago/ pensi babili treti decidi kraki gago to think, to chat, to
tread, to decide, to clap/crack, gag.
Stopstrictives
13.2.3. In addition, there are three stopstrictives: /q c, G/ (the last two are a
diphonic pair): (q) c (as in German: ('hInq) /'hInq/ Hintz, not as in English:
('hnts) /'hInts/ hints); then, there are (c) c and its voiced counterpart, (G) g (as in
English: ('fc, 'G;) /'fEc, 'GOE/ fetch joy).
ey always maintain the values we have seen, before any vowel or consonant:
('qnt, 'pa;q) /'qent, 'paqo/ cent paco hundred, peace, (c'va;l, f'li;ca) /ce'valo, fe'lica/ cevalo felica horse, happy, (GaR'd;n, 'pa;G) /Gar'deno, 'paGo/ gardeno pago garden, page (of book). e spelling dz represents a (rare) sequence,
(dz), not the voiced counterpart of c (which would be (Q), as in Italian ('QE:Ro)
/'QEro/ zero zero]\ ('dz) /'edzo/ edzo husband.
Constrictives
13.2.4. ere are three diphonic pairs of constrictives and a rare voiceless velar
one (which should be withdrawn from modern and future Esperanto, 13.4.3-
13. esperanto
399
4): /f, v s, z S, Z x/ (f,v s, z S, Z x). It is sucient to recall that s is always (s)

(voiceless, as in English: ('sns) /'sIns/ since] whereas z is always (z) (voiced, as in:
('zp) /'zIp/ zip]\ ('mu;z, 'mu;s, 'sli;p) /'muzo, 'muso, 'slipo/ muzo muso slipo
Muse, mouse, slip (of paper).
Besides, (S, Z) /S, Z/ s, j are as in English ('Sp, 'vZn) /'SIp, 'vIZn/ ship vision: ('fi;S, ZuR'na;l) /'fiSo, Zur'nalo/ so jurnalo sh, newspaper.
e last Esperanto constrictive is the rarest (and practically useless, so it could
protably be withdrawn, by merging into k or h as has already happened for quite
a few forms). It is h (x), voiceless velar, as in (Austrian) German ('ax) Bach: (m'na;x) /mo'naxo/ monaho monk.
Approximants
13.2.5.1. Among the Esperanto approximants (which have a freer passage of
phono-expiratory air in comparison with constrictives, and a fairly reduced friction noise, which on the contrary is very strong in constrictives), we nd the laryngeal /h/ (h) h. It can eectively be pronounced as a voiced (H), instead of voiceless
(h), in order to become more dierent from the inconvenient and disadvantageous
(x) (but this requires particular phonic considerations). e most important thing,
especially for Romance-language speakers, is not to completely neglect the phoneme /h/; in fact, ('h;R) /'horo/ horo hour is quite dierent from (';R) /'oro/
oro gold. And it is also dierent from ('x;R) /'xoro/ horo chorus (by now very
often wisely substituted by koruso (k'Ru;s)), and from ('k;R) /'koro/ koro
heart, as well.
13.2.5.2. e other two approximants, /j, w/ (j, w), are represented by j and
and correspond to English ('js, 'wn:) /'jEs, 'wIn/ yes win\ ('js, 'v;j, si~'j;R,
'wst) /'jes, 'vojo, sin'joro, 'westo/ jes vojo sinjoro esto yes, road, gentleman,
west.
However, when j and are not followed by any vowels in the same word, they
are realized as the corresponding vowels /i, u/ (i, u) (and it is to be avoided for
spelling to lead people to think that, in such positions, they have to pronounce
what are actually normal and real vowels as consonants): ('tui, 'Raiti, 'nau, 'laudi)
/'tui, 'raitoi, 'nau, 'laudi/ tuj rajtoj na ladi immediately, rights, nine, praise.
Apart from this last case of greater phonetic naturalness, i and u always need to be
distinguished from j and before vowels: (mi';l) /mi'elo/ mielo honey, ('mj;l)
/'mjelo/ mjelo spinal cord, (fi'ask, &mani';R) /fi'asko, mani'ero/ asko maniero
asco, manner.
Let us also consider the following cases, for which (as will be seen in 13.2.5.2)
j is forced to have two dierent values owing to the stress-assignment rule: ('plid)
/'pleido/ plejdo plaid, but (pl'ja;do) /ple'jado/ Plejado Pleiad.
In lexical derivation, when is followed by a vowel, we have two possibilities,
/w, u/: ('na;wa, 'naua) /'nawa, 'naua/) naa ninth; the same goes for (bal'da;wa,
-aua) /bal'dawa, -aua/) baldaa prompt, from ('baldau) /'baldau/) balda soon.
400
However, in compounding, we only have /u/: (lau'i;gi) /lau'igi/) laigi to adapt.

All this must not be unduly modied by the dierent analogical extensions of the
various ethnic languages.
13.2.5.3. In addition, even for stressing, let us consider cases (that we present
here in advance) such as: ('js, 'jam) /'jes, 'jam/ jes jam yes, already and ('is, 'iam)
/'ies, 'iam/ ies iam someone's, once, and (s'i;fi, 'sil, ba'lau, 'baldau) /so'ifi, 'soilo ba'lau, 'baldau/ soi sojlo balau! balda to be thirsty, threshold, sweep!,
soon. To write has always been a severe problem both for the press and typewriting (and today for computers, too, unless some special fonts have been installed).
e rst ocial Esperanto publication by Zamenhof appeared in 1887. e International Phonetic Association was founded in 1886 and the rst version of the International Phonetic Alphabet [IPA] was produced in 1888. ese coincidences explain why they were not aware of each other's existence. Today, however, the parallelism between (j, w) and j w is quite evident and it spontaneously leads to the
substitution of with w which would certainly be to the advantage of E-o itself.
It is neither weakness nor betrayal at all It would just be common sense!
Trills
13.2.6. ere is an alveolar tap, (R) (which, for emphasis, can oscillate with a
true trill (r), alveolar as well): ('Ra;n, 'kRp, 'tR;) /'rano, 'korpo, 'tre/ rano korpo
tre frog, body, very (much). Any other pronunciation of /r/ (although frequent,
because of the dierent national languages used by Esperanto speakers) is not neutral.
Laterals
13.2.7. Lastly, we nd a pure alveolar lateral, (l), with no particular nuances (so
it is dierent from the English {non-prevocalic} l sound as in ('w:{}) well(s)]\
('la;n, mul't;ga) /'lano, mul'tega/ lano multega wool, very very. Not even a
slight palatalization ((), perhaps before /i/) would be acceptable: ('li;gi) /'ligi/ ligi to bind e only normal assimilation occurs before dentals, (l) ((l)), and before postalveo-palatal protruded consonants, within lexemes or in unstressed grammemes, () ((L)): ('alta, 'faci) /'alta, 'falci/ alta falci high, mow.
Structures
13.3.0. In addition to the realization of its segments ( vowels and consonants,
as in the previous sections), an international language must be as free as possible
from microstructural peculiarities ( syllabication, assimilation, stress, rhythm,
and intonation) and without typical characteristics of any particular languages.
13. esperanto
401
erefore, it must have strict rules, which are coherent and systematic, yet clear
and simple, without concessions to any language, or to groups of languages.
Most of all, any peculiarities of one's mother tongue should be avoided. In fact,
if this is not done, in the end all Esperantists will speak their own dialect of Esperanto. As a matter of fact, this is what happens when people use E-o without considering its phonic aspect. us what they obtain is quite close to the linguistic
Babel which E-o seeks to solve.
In neutral Esperanto pronunciation stressed vocoids are realized as half-long in
(either word-internal or word-nal) unchecked syllables, when they occur in prominent positions, in intonemes, as usually at the end of sentences: ('sa;na, 'a;) /'sana, 'a/ sana A healthy, (the letter) A. In checked stressed syllables (either in word-internal or word-nal position), Esperanto vocoids are always short: ('paRt) /'parto/ parto part. Besides, Esperanto diphthongs are always short, as in German
(while in English they are long when not followed by voiceless consonants or by
unstressed syllables, as in ('w;I{}, 'w;I 'wIT, 'wID/-D) /'wEI{z}, 'wEId 'wEIt,
'wEId/ way(s) wade wait waider]\ ('uR) /'euro/ ero Euro, ('ti) /'tie/ tie
there.
Consonant clusters
13.3.1. Consonant sequences neither undergo voicing assimilation (except for
the place of articulation with n in lexeme-internal position or in unstressed grammemes), nor do they simplify. Everything must be pronounced according to the
values of each element: (&sn'n;ma, &mal'lga, &huf'f;R, &dis'slv, &lip'ha;Ri, 'gli;ti, 'digna, ds'qndi, 'kna;b, 'kvin, 'ligv) sennoma mallonga huero dissolvo
lipharoj gliti digna descendi knabo kvin lingvo anonymous, short, horseshoe
(hoof-iron), disintegration, moustache (lip-hair), to slide, dignied, to go down,
boy, ve, language. E-o would have a more modern and international aspect, by
introducing the variants ('kwin, 'ligw) (in place of (kv, gv)}, with a corresponding spelling: kin lingo or, better still, at last: kwin lingwo (and (pRs'va;di)
/pers'vadi/ persvadi to persuade = (pR'swa;di) /per'swadi/ persadi = perswadi).
e consonant assimilation of place of articulation, for /n/ + /0/, occurs in two
cases only: rstly, in lexeme-internal position: ('siki) /'sinki/ sinki to sink, as
(km'pR;ni) /kom'preni/ kompreni to understand implicitly shown, unlike (&kun'pR;ni) /kun'preni/ kunpreni to bring, (&kun'm;ti) /kun'meti/ kunmeti to combine, (&sn'pa;ga) /sen'paga/ senpaga free/for nothing; secondly, in pragmatically unstressed grammemes: (miam'plakn) /mian'plankon/ mian plankon my
oor (accus.). In (&siamman't;ln) /sianman'telon/ sian mantelon one's own
mantle (accus.), there is a rhythmic secondary stress. On the other hand, for emphasis, we have: (&mian'plakn) (where the secondary stress is not rhythmic), or
even ('mian 'plakn) /'mian 'plankon/ mian plankon.
Here we show that voiced and voiceless consonants do not inuence each other in the least in neutral pronunciation (unlike in many ethnic languages): (k'zisti, &abs'lu;t, &naz'tu;k, &kdk'du;) /ek'zisti, abso'lute, naz'tuko, okdek'du/ ekzisti
402
absolute naztuko okdek du to exist, absolutely, handkerchief (nose-kerchief),

eighty-two.
13.3.2. Some Esperantists strive to follow to the letter the indications that Zamenhof (1962, the inventor of E-o) used to give enthusiastic people in answer to
their questions and doubts. However, it is to be understood that, when he stressed
the point that each letter must be pronounced clearly separated from the neighboring ones, his sole aim was to make people avoid overly marked ethnic pronunciations. Surely, he had no clear intention to lead them to reect on the Esperanto
phonic structure, which is certainly neither based on avant-garde theories nor expressed in strict phonetic terms (least of all phonemic ones!).
Here are some examples of dierent current ethnic pronunciations (even by expert and uent Esperantists): (u'R;p) Eropo *('pP, j'>p) ('laudi) ladi *('l_i, 'l;DIi, 'lAdi, 'laodi, 'Oudi) ('pa~-j) panjo *('pa;No, 'phni) ('lg)
longe *('l, 'lO, 'lAI) ('vRt) vorto *('vOto, 'v;TU, 'f<to, 'vtP) ('sta;Ri) stari
*('Sta:i, 'sT>i, 'stORi) (la'tmp) la tempo *('tEm:b, l'thEmp{h}P) (in'tnsa)
intensa *(in'dEn:Qa, 't:sa, in'tns) (la'pa;q) la paco *(la'Faq:qP, la'baqqso, l'pto) (mi'G;jas) mi gojas *(miG'GOj:jasse, mi'Z:jasse) (b'l;q) ebleco *(eb'b]q:q, ji'b;q) (la'ku;b) la kubo *(la'guubbo, la'h:BP) (li'b;lo) libelo *(ib'bE;, li'B;lo) ('kRu;c) kruco *('kru:So, k'Rju;Cx) ('pa;Si) pasi *('paS:Si, 'paai) (mi'p;tas) mi petas *(mi'bE;Edas:, mi'phe:t{h}as) ('js) yes *(', 'jE).
Paying excessive attention to spelling can lead people to uselessly force themselves to always unnaturally realize the grapheme n as (n), in every possible context, only because the way it is written seems to require such an articulation. On
the contrary, other even more evident dierences which elude control are unconsciously produced, like those we have just seen.
However, the fact of writing n before all consonants except p b certainly does
not mean that (n) has illogically to be maintained, against (mp, mb). It is quite
the contrary, even if it has been expressed in a rather rudimentary way. Indeed, all
languages having a homorganic // to a following consonant ( with (mp, Mf,
~c, N, k) ) show this fact by using the only other unitary grapheme which is
a phoneme as well: m /m/. Otherwise, they would even write np and nb, as is done
by serious orthographies of heterorganic dialects (like, in general, those of northern Italy).
Ending with some Esperanto examples, we have: ('kmbi, km'pR;ni) /'kombi,
kom'preni/ kombi kompreni to comb, to understand, (kM'v;ni, kM'f;si)
/kon'veni, kon'fesi/ konveni konfesi to be suitable, to confess, (kn'du;ki, 'kntRau) /kon'duki, 'kontrau/ konduki kontra to drive, against, (k~'cRt) /kon'certo/ koncerto concerto, (k'gR;s, k'ka;va) /kon'greso, kon'kava/ kongreso
konkava congress, concave.
But (&kunm'tb-la, &kun&pRduk'ta;d, &kun&bata'lant) /kunme'tebla, kunproduk'tado, kunbata'lanto/ kunmetebla kunproduktado, kunbatalanto combinable,
coproduction, comrade in arms; in fact, they belong to dierent lexemes, because
they are compounds.
13. esperanto
403
Syllabication
13.3.3. e neutral Esperanto phonic syllabication, within a word, regularly
occurs between two consonants: ('Rs-ti, 'lib-R, 'b-l, 'sig-n, si~'j;R, d'zi;n)
/'resti, 'libro, 'eble, 'signo, sin'joro, ed'zino/ resti libro eble signo sinjoro edzino
to remain, book, perhaps, sign, gentleman, wife.
Instead, at lexeme-boundaries, the sequences are maintained intact: (mi'bl;vis)
/mi'blovis/ mi blovis I blew, (mi'sqias) /mi'sqias/ mi scias I know, (lian&tawn'i;Ris, -aun'-) /liantawen'iris, -auen'-/ li antaeniris he came along, (&Sisa'lu;tis23 'antau l&fR'i;Ri23) /Sisa'lutis. 'antau olfor'iri./ si salutis anta ol foriri she greeted before leaving. en the formation of words is respected, the meaning is more easily understood. In other words, if something is correctly understood, this means
that its formation has been analyzed well. For instance: (&vRt'a;R) /vort'aro/ vortaro dictionary = word [vort-) collection [ar-].
As can be seen, a stressed syllable maintains the lexemes (or stems) separated,
whereas the grammemes do not. As a matter of fact, we do not have *(&vRt'aR-)
*/vort'ar-o/. It should be obvious, however, that the transcription we have shown
does not allude at all to excessive pronunciations such as *(&vRt'aR). Naturally,
as a more Teutonic way of pronouncing it like *(&vRt'a;R) would not respect the
internationality of this language, neither would a more Romance pronunciation
like *(&vR'ta;R). e correct solution, for all nations, is to follow a middle course,
thus: (&vRt'a;R) (otherwise the structures of some particular languages prevail,
since all speakers are led to believe that the most familiar solutions for them are
the most correct ones, as happens for regional or foreign accents).
th clusters of several consonants, words are syllabied according to the criteria of natural phonetics: ('ks-tR, ks'qi;ti, s'tRa;do) /'ekster, eks'qiti, es'trado/ ekster eksciti estrado outside, to excite, platform.
In lexical compounding, neutral pronunciation separates the lexemes, but not
the desinential grammemes (which are then resyllabied into more natural structures, even by modifying word boundaries): fervojisto malantae bankroti (&fRv'jist, &mal-an'ta;w, -au, &bak'R;ti) /fervo'jisto, malan'tawe, -aue, bank'roti/
railwayman, backwards, to go bankrupt (without introducing here too sharp
separations as the laryngeal stop, (): *(&fR-v'jist, -voi'is- &malan'taw, -au)}.
Word stress
13.3.4. In Esperanto, there is no stress exception, contrary to most ethnic languages: it invariably falls on the last vowel but one of each (non-monosyllabic)
word, even if this may cause seemingly strange or curious dierences for similar forms in various languages: fralo ('fRaul) bachelor, praulo (&pRa'u;l) ancestor, balda ('baldau) soon; balai (ba'lai) to sweep, soi (so'i;fi) to be thirsty,
sojlo ('sil) threshold; masino (ma'Si;n) machine, muziko (mu'zi;ko) music,
tragedio (&tRag'di) tragedy, sukero (su'k;R) sugar, logika (l'gi;ka) logic, emfazo (m'fa;z) emphasis, stacio (sta'qi) station, jam ('jam) already, iam
404
('iam) once, sabato (sa'ba;t) Saturday, oceano (&q'a;n) ocean.

Instead, talking about the last syllable, for word-stress assignment, is extremely
inaccurate. In fact, completely dierent and contrasting criteria are still followed
upon the actual judgment and nature of syllables. ese are too often considered
only from a graphic, grammatical, and metrical point of view, which is incredibly
subjective and variable depending on languages, when not totally absurd.
Sentence stress
13.3.5. Let us now give some thought to the stressing of utterances. As it would
be absurd in ethnic languages to stress every single word appearing in a sentence,
such is the case even in E-o, whose grammemes ( grammatical words, lacking a
real semantic value) are completely unstressed, unless they are voluntarily emphasized for some particular reasons.
A sentence like Mi estas la amiko de via frato I'm your brother's friend would
certainly not be *('mi; 'stas 'la; a'mi;k 'd; 'via 'fRa;t), which would rightly
make people hate E-o as something unbearable! (On the other hand, even for foreign languages, it is not rare to hear such things, but this depends on incomplete
learning, as well as on widespread ignorance of phonetics and its advantages.) A
more appropriate rendition of the example given would be: (mi&stasl{a}a'mi;k
&dvia'fRa;t).
As can be seen from previous transcriptions, even in compounds, we nd dierent stress-degrees in their components, depending on their semantic relevance.
However, the main stress of compounds falls on the stem vowel of the last element.
Quite evidently, a slow and strained delivery, with too many stresses on grammemes as well, in addition to the inevitably high frequency of the (sole) auxiliary
verb esti can before long become a real nightmare if estas estis estos are always
stressed in a rather mechanical way.
erefore, accurate speakers carefully avoid falling into this nasty habit. In fact,
they will destress, even completely, the forms of esti by introducing the stylistic
renement of using instead the allomorphs with 'st-: (&mistas&ci'ti, &listsf'li;ca)
/mistasci'tie, listosfe'lica/ Mi 'stas ci tie Li 'stos felica I'm here, he'll be happy
(either only when pronouncing or when writing as well by providing a considerable visual-mnemonic help). Even the phrase (ti'stas) /tio'estas/ tio estas that is,
when it is not emphasized, surely gains by becoming (&tistas) /'tiostas/ tio 'stas.
Obviously, it is quite the contrary for: (&cu'sti2| &au"n; 'sti23) /cu'esti,| au"ne 'esti./ Cu esti, a ne esti? To be, or not to be?.
Unfortunately, this use is not generally followed, since the backward-looking
ideas that characterize most language teaching (starting from one's own language)
do not manage to clearly separate the graphic level (which is secondary) from the
phonic one (which is primary). All this leads people to believe that reduced forms
are a signal of corrupted and slovenly language. is is the case of (&aSD'ThDm, --), which is quite normal for I should have told them while on the contrary something like *(5a; S&hv'Thd&m, --) would be quite unusual in-
13. esperanto
405
deed. As a matter of fact, (&aSD'ThDm, --) is exactly what is uttered by

competent people, when they do not speak too slowly and are free from any spelling blackmail; since it is very important to be able not to confuse real language
with current writing, because real language is, rst of all, pronunciation (which is
above all not embalmed).
13.3.6. Returning to our rst example, we will have the following realization:
(&mistasla'mi;k d&via'fRa;t), which is possible to write as: Mi 'stas l'amiko de via
frato. e elision of la is ocial, although it is often erroneously limited to poetry.
Likewise, personal pronouns, which frequently appear at the end of sentences,
do not need to be always and mechanically stressed. On the contrary, only for emphasis or contrast may they have a strong stress, otherwise they are destressed and
enclitic (although written as separated forms, as in English, but not in many other languages); other monosyllabic particles behave in the same way: (Si'vi;dis&lin) /Si'vidislin/ si vidis lin she saw him, ('daknal&vi) /'dankonalvi/ dankon al vi
thank you, ('ti&ci) /'tioci/ tio ci this, ('cu;&n21) /'cune?/ cu ne? isn't it?, aren't
you?, doesn't he?, haven't they?
It is worthwhile comparing the following cases: (Si'vi;dis 'lin) /Si'vidis 'lin/ si vidis
lin she saw him, ('dakn al'vi;) /'dankon al'vi/ dankon al vi thank you!, (&ti'ci;) /tio'ci/ tio ci this here, (&cu'n;21) /cu'ne?/ cu ne? isn't that so?.
e case of ju (mal)pli, des (mal)pli is very interesting for stressing: ('ju; &plivipa'R;las2 2 'ds &malpli&mikm'pR;nas23) /'ju plivipa'rolas 'des malplimikom'prenas./ ju pli vi parolas, des malpli mi komprenas the more you speak, the less I understand, ('ju; &pli'mult2 2 'ds &pli'b;n23) /'ju pli'multe 'des pli'bone/ ju pli multe
des pli bone the more, the better.
Intonation
13.3.7. 13.3 shows the preintonemes and intonemes of neutral E-o (free from
any inuence from ethnic languages), so we will see only three fundamental examples:
/./: (&ili'ft kn'sultas &l{a}sp'RantaM &vRt'a;Rn23) /ili'ofte kon'sultas l{a}espe 13.3. Esperanto preintonemes and intonemes.
/ / (2 2 ' 2 2 ' 2 2 ' 2)
/./ (2 ' 2 3)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/?/ (2 ' 2 1)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
// (2 ' 2 2)
/ / ( 2 2 ' 2 2 ' 2 2 ' 2)
/,/ (2 ' 2)
406
'rantan vort'aron./ Ili ofte konsultas la esperantan vortaron ey often consult the
Esperanto dictionary.
/?/: (&cu&ili'ft kn'sultas &l{a}sp'RantaM &vRt'a;Rn21) /cuili'ofte kon'sultas
l{a}espe'rantan vort'aron./ Cu ili ofte konsultas la esperantan vortaron? Do they often consult the Esperanto dictionary?.
//: (&ili'ft kn'sultas &l{a}sp'RantaM &vRt'a;Rn2 2| &sd'tu;t &sn'fRukt23) /ili'ofte
kon'sultas l{a}espe'rantan vort'aron| sed'tute sen'frukte./ Ili ofte konsultas la esperantan vortaron, sed tute senfrukte ey often consult the Esperanto dictionary,
but that's of no use.
13.3.8. Besides being used in total questions, cu occurs in indirect questions
too, with a subordinate function: (&min'sqias &culi'v;ns) /mine'sqias culi'venos/
mi ne scias, cu li venos I don't know whether he will come. Cu is also used in disjunctive clauses with a coordinate function: cu li, cu si either he or she. In these
cases where, by the way, no nal question mark is written it is obvious that no
(rising) interrogative intonation has to be used. us, the choice of an intoneme
becomes important, decisive, and not at all redundant or useless. On the other
hand, in colloquial and expressive usage sometimes it is possible to omit an interrogative cu: (li'dRmas21) /li'dormas?/ li dormas? (is he) sleeping?, ('v;R21) /'vere?/ vere? really?. erefore, it is fundamental to use an interrogative intonation.
(In cases such as these, in all ethnic languages of the world, colloquially, it is possible to omit initial or nal particles.)
In partial questions, instead (with\ ('kiu, 'kiui, 'ki, 'kia, 'kis, 'kial, 'kiam, 'ki,
'kil, 'kim) /'kiu, kiui, 'kio, 'kia, 'kies, 'kial, 'kiam, 'kie, 'kiel, 'kiom/ kiu, and kiuj,
who (and a plural who), kio what, kia what kind (of), kies whose, kial
why, kiam when, kie where, kiel how, kiom how much), the most natural intoneme to use is conclusive (falling, (2 ' 23)), but combined with a regular interrogative preintoneme (tendentially rising): ('kial &ili'ft kn'sultas &l{a}sp'RantaM &vR'ta;Rn23) /'kial ili'ofte kon'sultas l{a}espe'rantan vort'aron./ Kial ili ofte konsultas la esperantan vortaron? y do they often consult the Esperanto dictionary?.
In the case of commands or orders, an imperative preintoneme is used, where
the falling movement of the conclusive intoneme is brought forward, although in
a rather limited way (as happens for the interrogative one, which however moves
in an opposite direction): ('ft kn'sultu &l{a}sp'RantaM &vR'ta;Rn23) /'ofte kon'sultu l{a}espe'rantan vort'aron./ Ofte konsultu la esperantan vortaron! Consult the
Esperanto dictionary often!.
Spelling and internationality
13.4.1. A consideration about E-o spelling, in the third millennium, is inevitable with regards to the six letters with a hat {(ca&pl'i;tai li't;Ri) /capel'itai li'teroi/ capelitaj literoj]. In fact, the time is ripe for denitely admitting that they are
the major obstacle to the printing and diusion of Esperanto publications, as we
13. esperanto
407
have already said ( 13.0.1). Besides, before Unicode, computers needed special
fonts to be installed, or particular programs to create new characters or to modify existing ones in order to use them, and they are still necessary for the many
things which are not yet part of Unicode.
But even in this way, these six letters are still a problem for orthographic correction or alphabetical arrangement. Besides, few normal typewriters could combine
4, and only for small letters; to say nothing of T, for , too often replaced by 5, or Z,
or Q, or 4: , (unless one studies hard to become a magician).
e problem could be overcome by simply modifying not the language (!), nor
its phonemic system (!), but merely its spelling, although for many people this
seems to mean instead to alter the very essence of E-o itself (whereas this is
normal for natural languages)! us some evolution and structural simplication
would be quite logical, as in natural languages. e way has already been indicated by Zamenhof himself, although it should be followed with rigorous criteria.
On the other hand, we nd an appealing and reassuring parallelism with the International Phonetic Alphabet, which consists in the use (already suggested by various scholars) of w x instead of h Even for c g j s, Zamenhof himself was
aware of the anti-modernism inherent in these signs, and suggested that if there
were typographical diculties (which is not merely hypothetical) they could be
substituted with ch gh jh sh hh u (for h ].
As usual, it is hard to manage to separate the prevailing graphic level from the
underrated phonic level. But it is even more dicult to separate personal
emotionality from linguistic objectivity.
13.4.2. Apart from the obvious inappropriateness of confusing and u (especially seeing that w was clearly available!), even in the other cases since in E-o compounding is the rule it ought to have been quite clear that forms like the following were not just a possibility: (&mus'ha;R, &paq'hu;l) /mus'haro, paq'hulo/ musharo pachulo a hair of a mouse, a hull of peace, which must not be confused with
(&muS'a;R, pa'cu;l) /mu'Saro, pa'culo/ musaro paculo a horde of ies, patchouli.
On the other hand, the forced use of digraphs (with a second Roman letter)
should not be with h as we have seen, but with another letter having no value in
E-o, one which is not already used, but is common and normal.
erefore, given that these sounds have a strong palatal component, added to
their main articulation (which is lamino-postalveolar), the letter that will function
as a diacritic has to be y\ agyo ajyo kacyo kasyo. In fact, by itself it has no value,
except possibly when writing foreign family names, in addition to other characteristics which may lead to an appropriate identication of the language.
It is nothing but a diacritic, without ambiguities; certainly it is no longer above
the letters, but immediately after them; and this is their advantage-disadvantage
of having no more technological limitations for the press and international communications by any possible means based on the Roman alphabet, with no further
complications. On the Internet, often the diacritic-letter is x seemingly more
by a visual-mnemonic inuence (in fact, the lower part of x can recall 4 and the
higher one, T; all the more so because x is quite rare in many languages), rather
408
than for useful phonic and graphic considerations.

Lastly, c (q), which has a dierent IPA value, would remain such, as in many
Slavic languages.
Esperanto 26
13.4.3. Certainly, the most modern solution and most useful to E-o itself
would be to move on to Esperanto 26, (&sp'Rant &dudk'ss), which uses only
the twenty-six letters of the Roman alphabet, by renouncing with more gains
than losses the two rarest and less useful sounds: (x, Z) /x, Z/ h j. us the phonemic inventory will be simplied, by removing the source of some main problems,
when we consider the dierent phonemic systems of the various languages of the
world: it is not always easy to distinguish (h, x G, Z) (as few languages systematically do).
Obviously, it is not sucient to delete. It is necessary to merge the forms containing these two signs (and sounds) with words having more similar sounds. is
means: (x) = (k h), (Z) = (G j). It is already a common practice to eect the former substitution: (x) = (k), as for kaoso kemio jakto chaos, chemistry, yacht; also the change (Z) = (G) has already been recorded: jazo = gazo jazz. is same application needs only to be generalized. en similar words containing (k, G) already exist, (h, j) will (have to) be used; otherwise, the shape of the new words
which would coincide will be slightly changed; or else they will have to be substituted, if necessary.
In the case of horo chorus, since we already have horo hour and koro heart,
the new form koruso has been added in order to nally replace it. For holero cholera, since kolero anger is already there, the solutions can be: holero or kolro
(26). In a series of oppositions like justa just, fair and gusta right, exact, we
may insert juro law, juro = guro vow, jako jacket, jako = gako jack as well.
As far as jeto jet, geto jetty, jeto throw are concerned, if this last one becomes
('G;t) yeto (26), the second example can safely become (G't) yeteo (26),
approaching the related form (m'l) moleo mole.
A really concrete case is ajo thing (and the corresponding sux aj with an
analogous meaning, which is very widely used) and ago age. Here, so as not to
risk overlapping with the exclamation aj! ow!, ouch!, from which in E-o it is certainly possible to derive ajo a cry of pain, and without thinking of changing a
whole series of words already formed with aj (and some more in the future), it
would be convenient and unproblematic to change ('a;Z) into ('a;G) [ajo =
ago = ayo 26), and ('a;G) into ('aiG) [ago = ajgo = ajyo 26) or probably, better still, into ('iG) [ago = ejgo = ejyo 26).
13.4.4. Once the number of (distinctive) sounds has appropriately decreased to
twenty-six, we have to decide how to pair the sounds with the new letters, as we
have already shown in brackets. e advantage of not having to worry about the
diacritics will certainly help to overcome some initial perplexities (leaving aside
13. esperanto
409
any consideration about the perversion of the language, since they are not at all
natural, nor necessary).
Obviously, as seen above, would become w (w), ('na;wa, 'naua) [naa = nawa
26), ninth; however g would become y (G), without causing much ado, given
that even in Spanish, in many contexts and variants, for y we nd a very similar
pronunciation, ('pa;G) [pago = payo 26) page (of book). In addition, c would
become q (c), and here again we nd similar pronunciations in Chinese and Albanian, (c'va;l) [cevalo = qevalo 26) horse. (On the other hand, does Esperanto not want to be the second language for all peoples?) nally, x (having no
further need to indicate (x) or to function as a diacritic) would be used for (S),
which has a similar sound in Portuguese, Catalan, Basque, Chinese, Maltese, and
Sicilian (besides in Old Spanish), ('fi;S) [so = xo 26) sh.
It would be a matter of becoming accustomed to the value of the new letters;
after all only for Slavists is c for (q) actually familiar, ('q;l) celo aim. For instance, is not h a real nothing for most Romance-language speakers: ('ha;vi) havi
to have (and consequently a real problem)?
ose who defend the preservation of (Z, x), because they are useful to render
some personal and place names of certain languages (with similar but not identical sounds), should also worry about all the other sounds that E-o does not manage to render (which are many more) e exchange of j and y (with the value of
j /G/ and y /j/), although more in tune with a widespread English-like graphemic
tradition, would push E-o away from both IPA usage and from the orthography of
many languages.
Text
13.5.0. e story e North Wind and the Sun follows, given in a number of
normalized accents, mostly as useful introductory examples of the phonetic method of natural phonetics. As always, rst there is the version which gives the typical
Esperanto pronunciation of English. en follows the actual Esperanto text, showing neutral E-o pronunciation.
Esperanto pronunciation (of English)
13.5.1. (z'nRs 'wind2 andz'san2 wRdis'pju;ti 'wic wzz'stRgR23| 'wn 'tRavlR2 'km a'l2 'Rapt in'wm 'kluk23| za'gRid2\ zatz'wan hu'fRst sak'si;did2 im'm;ki z'tRavlR2 'tk iz'kluk 'f2 2| &Subbikn'si;dRd "stRgR &zanzi'a;z23||
'zn2 z'nRs 'wind2 'blu;2 az'haRd2 azhi'kud23| &batz'mR hi'blu;2 2| z'mR 'klusli2
&didz'tRavlR2\ 'fld hiz'kluk a'Randhim23| &andat'last2\ z'nRs 'wind2 'gv 'ap zia'tmpt23|| 'zn2 z'san 'Sn 'at23 'wRmli23| &andi'mi;diatli2\ z'tRavlR 'tuk 'f23 hiz'kluk23|| and'su2 z'nRs 'wind2 wz'blaG{d} tukM'fs2| &zatz'san23 wzz'stRgR23 &vz'tu;23||
&didju'lak21 z'st;Ri2| dju'wnt tu'hi;Rit a'gn21|||)
410
Esperanto text
13.5.2. Iam, la norda vento kaj la suno disputis, car ciu el la du asertis esti pli forta ol la alia. En tiu momento, ili ekvidis vojaganton, kiu antaeniris volvite en sia
mantelo. La du disputantoj, do, decidis, ke konsideratos pli forta tiu el la du kiu sukcesos igi la vojaganton forpreni sian mantelon.
La norda vento ekblovis tre forte; sed, ju pli gi blovis, des pli la vojaganto mallozigis
sian mantelon, kaj, nne, la kompatinda vento devis rezigni. Tiam, la suno montrigis en la cielo, kaj, mallonge poste, la vojaganto, al kiu estis varme, forprenis la mantelon. Tiel, la norda vento devis agnoski, ke la suno estas pli forta el la du.
Cu vi satis la historion? Cu ni ripetu gin?
Esperanto pronunciation
13.5.3. ('iam2 la'nRda 'vnt2 &kaila'su;n2| dis'pu;tis23 'caR2 ciu &lla'du;2 a'sRtis23
&sti&pli'fRta2 &ll[a]a'lia23| n&tium'mnt2 2| &ili&k'vi;dis23 &vja'Gantn23 &kiuan&tawn'i;Ris23 vl'vit n&siaman't;l23|| la'du &dispu'tanti2 'd;2 d'qi;dis23 &kkn&sid'Ra;ts2
&pli'fRta2 2|| &tiulla'du;2| &kiusuk'q;ss2 &igila&vja'Gantn2 2| &fR'pRni &siamman't;ln23||
la'nRda 'vnt2 2| &k'blvis &tR'fRt23| 'sd2 'ju; &pliGi'bl;vis2 2| 'ds2 &plila&vja'Gant2 &mall'zigis &siamman't;ln23| &kaifiM'fi;n2| la&kmpa'tinda 'vnt2| &dvisR'zigni23|| 'tiam2| la'sun mn'tRi;Gis2 &nlaci';l23| &kaimal'lg 'pst2 2| la&vja'Gant2 al&kiu[]stis'vaRm2 2 &fR'pR;nis23 &laman't;ln23|| 'til2| la'nRda 'vnt2 &dvis-ag'nski23 &kla'su;n2 &stas&pli'fRta23 &lla'du;23||
&cuvi'Sa;tis21 la&hist'Rin2| &cuniRi'p;tu&Gin21|||)
Foreign pronunciations of Esperanto
13.5.4. We add 13 foreign accents ( {British and American} English, Italian,
French, German, Spanish, {Brazilian and Lusitanian} Portuguese, Russian, Arabic,
Hindi, Chinese, and Japanese. In the Italian version of this book 11 regional accents are also given, which are omitted in this English adaptation ( Turinese,
Genoese, Milanese, Venetian, Bolognese, orentine, Roman, Neapolitan, Barese,
Palermitan, Cagliaritan).
Speakers of other languages could prepare their own versions. e author would
be happy to receive their transcriptions and recordings, both in case of help
should they need it and to make their contribution known to others (possibly
in our website on canIPA Natural Phonetics 0.12).
13. esperanto
411
British pronunciation
13.5.5. ('Iim2 l5n;D 'vnT2 &khal'sun2| D'spuTs3 3 'chA:2 5chIi l'D;u2 'sTs3 3 &sTi&phlIi'f;T2 &;l'lIi3 3| n&ThIim'mnT32| &Iilik'vIiDs23 &v'GA;nT3 3 5khIi n&Twn'>s3 3 v5vIiTI n&sIimmn'Thl3 3|| l5Du &Dsp'ThA;nT2 'D:2 D'tsIiDs3 3 khI&khnsD'>A;Ts2 &phlIi'f;T32|| &ThIil'D;u2| &khIisk'tsss2 5Iigi &lv'GA;nTn32| f;5ph<ni &sIimmn'Thln3 3||
l5n;D 'vnT32| k5blvs &h>I'f;TI3 3| 's;2 'j;u &phlIiGIi'blvs32| 'Ds2
&phlIilv'GA;nT2 &mA;l5zIigs &sIimmn'Thln3 3| &khafM'fIinI2| l&khmp'ThIinD 'vEnT2| &Dvs>'zIigni3 3|| 'ThIim2| l5s;un mN'h>IiGs2 &nlci'l3 3|
&kham5lgI 'phsTI32| l&v'GA;nT2 &khIi5sTs 'vA;mI32 f;'ph>ns3 3 &lmn'Thln3 3|| 'ThIi2| l5n;D 'vnT2 &Dvsg'nski3 3 &khIl'sun2 &sTs&phlIi'f;T3 3 &l'D;u3 3||
&chuvi'SA;Ts21 l&hsT'>Iin2| ch&nIi>'phT&GIin21|||)
American pronunciation
13.5.6. ('Iim2 l'n;<D 'vn[[]2 &khal'sUun2| D'spUu[s23 'chA:<2 'chIi l'DU;u2 's<Ts23 &sTi&phlIi'f;<T2 &O;l'lIi23| n&ThIim6mn[[]2 2| &Iilik'vIiDs23 &v'GA;n[[]23 'khIi n&Twn'I;<s23 vO;'vIi[I n&sIimn'Thl23|| l'DUu &Dsp'ThA;n[[]2 'DO:2 D'tsIiDs23 khI&khO;nsD'<A;Ts2 &phlIi6f;<T2 2|| &ThIil'DU;u2| &khIisk'tsss2 'Iigi &lv6GA;n[[]n2 2| f;<'ph<ni &sIimmn'Thln23||
l'n;<D 6vn[[]2 2| k'blvs &Th<I'f;<TI23| 's;2 'ju &phlIiGIi6blvs2 2| 'Ds2
&phlIilv'GA;n[[]2 &mA;l'zIigs &sIimmn'Thln23| &khafM'fIinI2| l&khO;mp'ThIinD 'vn[[]2| &Dvs<'zIigni23|| 'ThIim2| l'sUun mO;n'Th<IiGs2 &nlci'l23| &kham'lO;gI 6phO;sTI2 2| l&v'GA;n[[]2 &khIi'sTs 6vA;<mI2 2
f;<'ph<ns23 &lmn'Thln23|| 'ThIi2| l'n;<D 'vn[[]2 &Dvsg'nO;ski23 &khIl'sUun2 &sTs&phlIi'f;<T23 &l'DU;u23||
&chUuviSA;[s21 l&hsT'<Iin2| ch&nIi<ph[&GIin21|||)
Italian pronunciation
13.5.7. ('i;am2 la'nOrda 'vEn:to2 &kaila'su:n2| dis'pu:tis23 'ca:r2 'ciu ella'du2 a'sEr:tis23
&stipli[f]'fOr:ta2 &llaa'li;a23| en&tiumo5men:t12| &iliek'vi:diz23 &voja'Gan:to23 kiu&antawe'ni:Riz23 vol'vi:te en&siaman'tE:lo23|| la'du[d] dispu'tantoi2 'dO[d]2 deq'qi:dis23 &kekkon&side'ra:tos2 pli[f]5fOr:ta12|| &tiuella'du2| &kiusuk'qE:sos2 'i;gi la&voja5Gan:ton12| foR'prE;ni
&siamman'tE:lon23||
la'nOrda 5vEn:to12| ek'blO;vis tR'fOr:te23| 'sEd:2 'ju[p] pli[G]Gi5blO:vis12| 'dEs:2 pli[l]la&vo-
412
ja'Gan:to2 &mallo'zi;gis &siamman'tE:lon23| &kaifiM'fi:n2| la&kompa'tinda 'vEn:to2| &dvizRe'zig:ni23|| 'ti;am2| la'su;no mon'tri:Gis2 &enla'cE:lo23| &kaimal'lOge 5pOs:te12| la&voja'Gan:to2 al&kiu'Estis 5var:me12 foR'prE:niz23 &laman'tE:lon23| 'til2| la'nOrda 'vEn:to2 &dvi-sag'nOs:ki23 &ke[l]la'su:n2 &staspli[f]'fOr:ta23 &ella'du23||
&cu[v]viS'Sa:tis21 la&isto'ri;n2| cu[n]&niRi'pE:tu21 'Gin:2|||)
French pronunciation
13.5.8. ('im2 l'nd 'vntP2 &il'snP2| _isp+is3 3 [T]'S:2 [T]'Si ll'd2 sE+is3 3 &s+ipi'ft2 &ll;li3 3| 3n&+im5mntP1 1| &ilivi_is3 3 &v,Zntn3 3 &inteniis3 3 vl'vite 3n&simntElP3 3|| l'd &_isp'tntPi2 'dP2 d[t]si_is3 3 &ekn&side'tPs2 &pi5ft1 1|| &+ill'd2| &isk'sEsPs2 &iilv,5Zntn1 1| f'peni &simntEln3 3||
l'nd 5vntP1 1| k'blvis tfte3 3| 'sEd2 ', &pi[D]Zi5blvis1 1| 'dEs2 &pil&v,'ZntP2 &mllO'ziis &simntEln3 3| &ifin'fine2| l&kp+ind 'vntP2| &devisezigni3 3||
'+im2| l'snP mn'ti[D]Zis2 &3nl[T]SelP3 3| &iml'lNe 5pste1 1| l&v,ZntP2 &i'Es+is 5vme1 1 fpenis3 3 &lmntEln3 3|| '+il2| l'nd 'vntP2 &devi-sgnsi3 3 &el'snP2 &sts&pift3 3 &ll'd3 3||
&[T]SviS+is1 1 l&ist'in2| &[T]S&niipet1 1 '[D]Zin2|||)
German pronunciation
13.5.9. ('i:am2 la'nOda 'fEnto2 &khaela'zu:no2| I'spu:tIs3 3 'cha:2 'chi:u &lla'du:2
a'zEtIs3 3 &sti&phli'fOta2 &llaa'li:a3 3| n&thi:umo5mEnto1 1| &ilik'i:dIs3 3 &fo,at'antn3 3 &khi;uan&taon'i:Is3 3 fl'vi:t n&ziaman'the:lo3 3|| la'du: &dIspu'thantOY2 'do:2
de'qhi:dIs3 3 &khekn&zid'a:ts2 &phli;5fOta1 1|| &thi;ulla'du:2| &khi;uzUk'qhe:zs2 'i:gi
la&fo,at5antn1 1| f'phe:ni &ziamman'the:ln3 3||
la'nOda 5fEnto1 1| k'lo:vIs &the;'fOt3 3| 'zEt2 ',u: &phli;ti5blo:vIs1 1| 's2 &phli;la&fo,at'anto2 &mallo'zi:gIs &iamman'the:ln3 3| &khaefIM'fi:n2| la&khmpa'thInda 'fEnto2|
&evIse'zIgni3 3|| 'thi:am2| la'zu:no mn'thi:tIs2 &nlaci'e:lo3 3| &khaemal'lOg 5phOst1 1| la&fo,at'anto2 al&khi;u'EstIs 5am1 1 f'phe:nIs3 3 &laman'the:ln3 3|| 'thi:l2| la'nOda 'fEnto2 &devIsak'nOski3 3 &khela'zu:no2 &stas&phli;'fOta3 3 &lla'du:3 3||
&chu;vi'Sa:tIs1 1 la&hIsto'i:n2| chu&ni;i'phe:tu&tIn1 1|||)
Spanish pronunciation
13.5.10. ('ian2 la'nRa 'Bnto2 &kaila'u;no2| di'pu;ti3 3 'CaR2 'Ciu e{l}la'u2 a'Rti3 3
&ti&pli'fRta2 &{l}la'lia3 3| en&tiumo'mnto1 1| &ilie'Bi;i3 3 &Bo,a',anton3 3 &kiuantawe'ni;-
13. esperanto
413
Ri3 3 Bol'Bi;te en&iaman't;lo3 3|| la'u &ipu'tantoi2 '2 e[t]'i;i3 3 &kekon&ie'Ra;to2

&pli'fRta1 1|| &tiue{l}la'u2| &kiuu';o2 &iila&Bo,a',anton1 1| fR'pR;ni &iaNman't;lon3 3||
la'nRa 'Bnto1 1| e'Bl;Bi &tR'fRte3 3| '2 'u &pli,i'Bl;Bi1 1| 'd2 &plila&Bo,a',anto2
&ma{l}lo'i;i{}2 &ia{N}man't;lon3 3| &kaifiM'fi;ne2| la&kompa'tinda 'Bnto2| &dBiRr:e'ini3 3|| 'tian2| la'uno mon'tRi;,i2 &enla'Cj;lo3 3| &kaimal'lge 'pte1 1| la&Bo,a',anto2 al&kiu'ti'BaRme1 1 foR'pR;ni3 3 &laman't;lon3 3|| 'tiel2| la'nRa 'Bnto2 &Bia'nki3 3 &kela'u;no2 &ta&pli'fRta3 3 &e{l}la'u3 3||
&CuBia;ti21 la&ito'Rion2| Cu&niRr:ip;tu21 ',in2|||)
Brazilian pronunciation
13.5.11. ('iAn2 ]AnOd 'vntu2 &kai]'su;nu2| ispu;Cis3 3 'a2 iu e{}]'du2 asECis3 3 &sCi&p]i'fOt2 &o{}]]i3 3| e~&Ciummnt2 2| &i]ieivi;is3 3 &voaAntn3 3
&iuAn&tajeni;Ris3 3 voviCi e&siAmAnte;]3 3|| ]du &ispu'tAntji2 'dO2 isi;is3 3
&ek&side'Ra;tus2 &p]ifOt2 2|| &Ciue{}]'du2| &iusui'sE;sus2 'ii ]a&voaAnt3 3| fopR;ni &siAmAnte;]3 3||
]AnOd vnt2 2| &eib]Ovis &tRfOCi3 3| 'sEi2 'u &p]iib]O;vis2 2| 'dEs2 &p]i]a&voaAntu2 &mA{}]ozii{s} &siAmAnte;]3 3| &kaifi'fi;ni2| ]&kmpaCind 'vnt2| &devi{s}ezigni3 3|| 'CiA2| ]sunu mn'tRi;is2 &en]je;]u3 3| &kaimA{}]9Ni pOsCi2 2| ]&voaAntu2 A&iuEsCiz vami2 2 fopR;nis3 3 &]AmAnte;]3 3|| 'Ci2| ]AnOd
'vntu2 &dvizainOsi3 3 &e]'su;nu2 &sts&p]ifOt3 3 &e{}]du3 3||
&uvia;Cis12 &]isto'Ri2| u&niipe;tu12 'i2|||)
Lusitanian pronunciation
13.5.12. ('i4n2 ]'nOR 'vnt[u]2 &kai]'su;nu2| dSpu;t?S3 3 'Sa2 'Siu e{}]'u2 sEtNs3 3
&StN&p]i'fOta2 &o{}]a]i3 3| en&tiumu'mntu2 2| &i]ikvi;S3 3 &voZAntn3 3 &kiuAntjni;RS3 3 vo'vit e&si4m4nte;]3 3|| ]'u &Spu'tAntji2 'O2 si;S3 3 kNk&si'Ra;tuS2 &p]i'fOt2 2|| &tiue{}]'u2| &kiusuk'sE;sus2 'ii ]&voZ4nt3 3| fu'pRE;ni &si4m4nte;]3 3||
]'nOR 'vntu2 2| ek'b]OvS &tRfOt3 3| 'sE2 'u &p]iZ'b]O;vS2 2| 'dES2 &p]i]&voZAntu2
&mA{}]u'zi{S} &si4m4nte;]3 3| &kaifi'fi;n[]2| ]&kp'tind 'vnt[u]2| &dev{Z}zini3 3|| 'ti4| ]'sunu mn'tRi;ZS2 &en]Sje;]u3 3| &kaimA{}']9g 'pOSt[]2 2| ]&voZAnt[u]2
A&kiu'EStZ vaRm2 2 fupRE;nS3 3 &]m4nte;]3 3|| 'ti2| ]'nOR 'vnt[u]2 &viznOSki3 3 &ki]'su;nu2 &StS&p]i fOt3 3 &e{}]u3 3||
&Suvi'Sa;tS21 &]aiStu'Ri2| Su&nii'pe;tu21 'Zi2|||)
414
Russian pronunciation
13.5.13. ('ixm2 njRdx 'vntx2 &kaI'sju;nx2| IspjuIs3 3 'CaR2 Ciu'dju;2 Is3 3 &i&pI'fjtx2 &jix3 3| I8&iummntx2 2| &iiivi;Is3 3 &vjantxn3 3 &IuxntxF~i;Is3 3 vvi;i i8&ixmn;x3 3|| dju &Ispu'tFantxi2
'dj;2 IqF;is3 3 &Ikx8&iI'RFa;txs2 &pIfjtx2 2|| &iu'dju;2| &Iusuk'qF;sxs2 i;I
x&vjantxn2 2| fpe;~i &ixmn';xn3 3||
njRdx vntx2 2| kbj;vIs &tRfjI3 3| 't2 'ju; &piIbj;vIs2 2| 's2 &pIx&vjantx2 &mazi;i &ixmmn';xn3 3| &kaifIm'fi;~I2| x&kxpindx 'vntx2|
&evIsi=ig~I3 3|| 'ixm2| sju;nx mxn'ti;is2 &InC;x3 3| &kimjNI pjsI2 2| x&vjantx2 &IusIs vFaRmI2 2 fpe;~Is3 3 &xmn';xn3 3|| 'i2| njRdx 'vntx2 &evIsgnjsi3 3 &I'sju;nx2 &stxs&pIfjtx3 3 &'dju;3 3||
&Cvi5a;Is12 x&Istx'ixn2| C&~iI5p;tu&In12|||)
Arabic pronunciation
13.5.14. ('i;m2 l'n5d 'vntU2 &kail'su:nU2| dIs'pu:tIs23 'Sa52 'SiUIll'du:2 's5tIs23
&stI&blI'f5t2 &ll'li;23| In&tiUmU'mntU2 2| &ili&Ik'vi:dIs23 &vj'ZantUn23 &kiUntwI'ni:RIs23 vUl'vi:tI In&simn'tlU23|| l'du &dIspU'tantUI2 'd2 dIt'si:dIs23 &kIkUn&sIdI'Ra:tUs2
&pli'f5t2 2|| &tiUIll'du:2| &kiUsUk'ssUs2 'i:gI l&vj'ZantUn2 2| fUR'b5ni &simmn'tlU23||
l'n5d 'vntU2 2| Ik'blvIs &tR'f5tI3 3| 'sd2 'ju; &bliZIb'lvIs2 2| 'ds2 &plil&vj'ZantU2
&mallU'zi:gIs &simmn'tlUn23| &kaifIM'fi:nI2| l&kUmb'tInd 'vntU2| &dvIsRIgnI23||
'ti;m2| l'su:nU mUn't5i:gIs2 &Inl'SjlU23| &kaiml'lgI 'bstI2 2| l&vj'ZantU2 l&kiU'stIs
'va5mI2 2 &fUR'b5nIs23 &lmn'tlUn23|| 'til2| l'n5d 'vntU2 &dvIs-g'nskI23 &kIl'su;nU2 &sts&bli'f5t2 &Ill'du:23||
&SuvI'Sa:tIs21 l&hIstU'5iUn2| SU&nIRI'ptU&ZIn21|||)
Hindi pronunciation
13.5.15. ('iiAm2 ln '6N&o2 &kil'suu&no2| IspuuIs3 3 'C2 Ciiu &ell'uu2
sIs3 3 &esi&plI'&2 &ollalii3 3| eN&iiumomN&o2 2| &iliek6iiIs3 3 &6oNon3 3 &kiiuAN6eniiRIs3 3 6ol6iite en&siimNlo3 3|| luu &Ispu'N&oi2 '2
etsiis3 3 &kekon&sie'Raa&os2 &pli&2 2|| &iiuell'uu2| &kiiusUk's&sos2 iigi l&6oN&on2 2| opReeni &siiAmmNlon3 3||
ln 6N&o2 2| egbl6Is &Ree3 3| 'sd2 'juu &pliibl&6Is2 2| 'ees2
&plil&6o'N&o2 &mllosiigIs &siiAmmNlon3 3| &kiIM'ii&ne2| l&kompIN
'6N&o2| &e6IsResIgni3 3|| 'iiAm2| lsuuno moN'Rii&Is2 &enlCjeelo3 3| &kimll9-
13. esperanto
415
ge p&e2 2| l&6o'N&o2 l&kiiuei 6aa&me2 2 opReenIs3 3 &lmNlon3 3|| 'iiel2| ln '6N&o2 &e6Is-gnski3 3 &kel'suu&no2 &es&pli3 3
&elluu3 3||
&Cuu6i'aa&Is21 l&HIo'Riin2| Cu&niRi'pee&u21 'In2|||)
Chinese pronunciation
13.5.16. ('iam2 2la'n;2da 'vn22 &ka2la'su2n2| 2dIs'phu2Is13 'hja;2 'hiu &;2la'tu2 2a's;2Is13 &s2ti&pli'5;2a2 &;2la'lia13| 2n&tiu2['mn2^| &i2li&2kM'vi2dIs13 &2ja'jan2U,13 &kiu2n2a2wn'i2lIs13 2;'vi2 2~&ia2man'th2l13|| 2la'tu &Is2u'than2I2
't2 2d'hi2dIs13 &k2U,&i2d'la2s2 &pli'5;2a^|| &tiu&;2la'tu2| &kiu2su2kM'q2ss2 'i2gi 2la&2ja'jan2U^| 25;'phl2~i &ian2man'th2lU13||
2la'n;2da 'vn2^| &2kM'pl2vIs &tl'5;213| 's2M 'ju &pli2i'pl2vIs^| 'ts2 &pli2la&2ja'jan22 &ma;2l'i2gIs &ian2man'th2lU13| &ka2fIn'fi2n2| 2la&km2pa'thIn2a 'vn22|
&2vIs2l'Ig2ni13|| 'thiaC2| 2la'su2n 2[n'thli2is2 &n2la2hj2l13| &ka2ma;'l2g
'(hs2^| 2la&ja'jan22 a;&kiu's2tIs 'va;2m^ 5;'phl2~Is13 &la2man'th2lU13||
'thi2| 2la'n;2da 'vn22 &2vIs2ag'ns2ki13 &k2la'su2n2 &s2tas&pli'5;2a13 &;2la'tu2||
&ju2vi'ja2Is31 2la&hIs2t'liU2| 2ju&~i2li'ph2u&Ic31|||)
Japanese pronunciation
13.5.17. ('ia2m2 2ma'n2mM2da '62t2 &kai2ma'sM2n2| 3i2s'pM2i2s 'a2mM2
'M &2m2ma'QM2 2a's2mM2i2s13 &2s2i2p&mi'F2m2ta2 &2m2ma'mia13| 3&iM2m'm2t^| i2mi&2i'6i2i2s13 &62ja'ja2tP13 &iM2a2ta2P'i:2mi2s13 262mM'6i2t 2P&ia2ma't2m13 2a'dM 2i2s2pM'ta2ti2 'd2 2de'qi2i2s13 &k2kP&i2d'ma2t2s2 2pM&mi'F2m2ta13|| iM22m2ma'QM2| &iM2s2k'q2s2s2 &i2i 2ma&62ja'ja2tP^| F2mM'p2m2~i &ia2ma't2mP13||
3a'n2mM2da '62t^| 2k2bM'm26i2s &t2m'F2m2t13| s2Q2 'jM: 2p&mi2Bi2b'm26i2s^| d2s2 2p&mi2ma&62ja'ja2t2 &ma2mM2m'Bi2i2s &ia2ma't2mP13| &kai2iP'i2n2| 3a&k2pa'i2da '62t2| &d26i2s2m'BiP2~i13|| ia2m2| 3a'sM2n &m2t'mi2i2 &2a'2m13| &kai2ma2m'm2g 'p2s2t^| 3a&62ja'ja2t2 &a2m&iM'2s2i2s '6a2m2m^ &F2mM2p'm2~i2s13 &ma2ma't2mP13|| i2mM2| 3a'n2mM2da '62t2
&d26i2s2a2'n2s2i13 &k2ma'sM2n2 &2s2ta2sM2p&mi'F2m2ta13 &2m2ma'QM13||
M26i'ja2i2s31 2ma&i2s2t'miP2| 3M&~i2mi'p2qM&BiP31|||)
From L. Canepari (2005) A Handbook of Pronunciation, Lincom Europa.
Utilizable bibliography
e utilizable bibliography is presented in separate chapters.
Prelude ( 1)
bouquiaux, l. cloarec-heiss, f. thomas, j. m. c. (1976) Initiation la phontique Paris:
puf/orstom (a vinyl record with recorded text, to be used alone or in reference to thomas
bouquiaux cloarec-heiss { 3}; expanded IPA).
canepari, l. (1983) Phonetic Notation _ La notazione fonetica Venezia: Cafoscarina (with 2 enclosed audiocassettes; almost canIPA).
(1985) L'intonazione. Linguistica e paralinguistica Napoli: Liguori (almost canIPA).
(2000) Introduzione alla fonetica Torino: Einaudi (expanded IPA).
(2004 {amended modied reprint} 2003) Manuale di fonetica Fonetica "naturale&: articolatoria, uditiva, funzionale. Mnchen: Lincom Europa (canIPA).
(2005) A Handbook of Phonetics: Natutal phonetics articulatory, auditory, and functional Mnchen: Lincom Europa (English version of the MaF, canIPA).
(2006) Avviamento alla fonetica Torino: Einaudi (canIPA).
catford, j. c. (1977) Fundamental Problems in Phonetics Edinburgh: E. Univ. Press (IPA).
(1988) A Practical Introduction to Phonetics Oxford: Clarendon Press (guided drills to develop phonetic kinesthesia, to be performed accurately, step by step; however, the 2001 edition should be avoided because of too many technical problems during its unsuccesful
updating; IPA).
chapman, w. h. et alii (1988) Introduction to Practical Phonetics Horsleys Green: Summer
Institute of Linguistics (substantially IPA).
costamagna, l. (2000) Insegnare e imparare la fonetica. Torino: Paravia (with an enclosed audiocassette; canIPA]
delattre, p. et alii (1951) vwljl stetik l d@ frmt e vwljl klrdinll in Le Matre Phontique, 96:30-5 as all MPh contributions, fully transcribed in IPA.
haudricourt, a. g. thomas, j. m. c. (1976) La notation des langues. Phontique et phonologie Paris: Inst. Gographique National (with 2 enclosed vinyl records; IPA).
hyman, l. m. (1975) Phonology: theory and Analysis New York: Holt, Rinehart nston
(non-IPA).
jones, d. (1956) Cardinal Vowels London: Linguaphone Inst. 2 {78 rpm} records with booklet; now downloadable; IPA.
(1967) e Phoneme: its Nature and Use Cambridge: Heer (IPA).
ladefoged, p. maddieson, i. (1996) e Sounds of the World's Languages Oxford: Blackwell
with stated instrumental rigor, but uses also other poeple's works, by accepting them too
peacefully, or drawing dierent interpretations; 16 Macintosh diskettes for HyperCard were
in partial correspondence with parts of this book; IPA.
laver, j. (1980) e Phonetic Description of Voice Quality Cambridge: C. Univ. Press (with a
non-enclosed audiocassette; IPA).
(1994) Principles of Phonetics Cambridge: C. Univ. Press (one may be doubtful whether to
indicate it or not, because it tries to use the few symbols and various awkward diacritics of
414
ocial IPA, in transcriptions which are always prosodically decient, whereas in two pages
{arranged in three: 558-60} it tries to give accurate descriptions, which only reveal the
method's heaviness and complexity, with results that cannot be suggested and languages
which are unrecognizable even to native speakers, when one tries to reproduce them as represented; oIPA).
malmberg, b. (1974) Manuel de phontique gnrale Paris: Picard (IPA).
o'connor, j. d. (1973) Phonetics Harmondsworth: Penguin (IPA).
schubiger, m. (1977) Einfhrung in die Phonetik Berlin: De Gruyter (IPA).
smalley, w. a. (1964) Manual of Articulatory Phonetics Terrytown (~): Practical
Anthropology (with 33 non-enclosed {18 cm, 19 cm/s} reels, lasting 32 hours; non-IPA).
thomas, j. m. c. bouquiaux, l. cloarec-heiss, f. (1976) Initiation la phontique Paris:
puf (completed by the vinyl record of bouquiaux cloarec-heiss thomas; expanded
IPA).
trubeckoj, n. s. (1939) Grundzge der Phonologie Gttingen: Vandenhoeck Ruprecht
(1969) Principles of Phonology Berkeley: Univ. of California Press (non-IPA and sometimes
mixed, so that some symbols can have {very} dierent values).
westermann, d. ward, i. c. (1990 {1930/31}) Practical Phonetics for Students of African
Languages London/New York: Kegan Paul International in association with the International African Institute (IPA).
English ( 2)
American English Pronunciation Program (1999) New York: Living Language/Random House
(graphemic booklet, but with 6 audiocassettes).
Australian Learners Dictionary (1997) Sydney: National Centre for English Language Teaching
and Research (gives the distributions of Australian pronunciation; IPA).
BBC Pronouncing Dictionary of British Names (1983, 1971) Oxford: O. Univ. Press (British
pronunciation; also IPA).
canepari, l. (in progr.) English Pronunciation*. Mnchen: Lincom Europa (with 2 neutral and
2 mediatic pronunciations { American and British}, in addition to the international one
and 5 further neutral pronunciations, together with several native regional accents from all
over the world and several broad foreign accents, too; canIPA]
dauer, r. m. (1993) Accurate English Englewoods Clis: Regents/Prentice Hall (American pronunciation; with 4 audiocassettes; almost IPA).
Gage Canadian Dictionary (2000) Gage Educational Publishing Company: Vancouver (gives
Canadian pronunciation; IPA).
gimson, a. c. cruttenden, a. (2001) An Introduction to the Pronunciation of English
London: Arnold (now: cruttenden, a. <Gimson's Pronunciation of English (!); still the
most recommendable for British pronunciation, although it has not reached the declared
and hoped for updating of symbols and conceptions, and in spite of some electronic little bungles in the latest two editions; IPA).
gimson, a. c. ramsaran, s.m. (1982) An English Pronunciation Companion Oxford: O.
Univ. Press (British English; with an audiocassette; IPA).
jones, d. (1960, 1918) An Outline of English Phonetics Cambridge: Heer (British pronunciation; meanwhile, pronunciation has changed and symbols have been improved, but it is
still worthwhile reading, rather than many more recent books; IPA).
jones, d. (2003, 1917) English Pronouncing Dictionary Cambridge: C. Univ. Press (as the
previous edition, the present one, edited by P. Roach J. Hartman J. Setter, in addition
to British pronunciation, gives the American one; but it has lost much of the original spir-
utilizable bibliography
415
it, by standardizing the transcriptions { Jones Gimson Ramsaran}; there is also a version with a , which allows one to listen to the pronunciations and to search for homophones by typing in a transcription; IPA). (e former version of the Cambridge Advanced
Learner's Dictionary for American pronunciation {assigned to Hartman and to a clever
editorial sta} had monstrosities like blaster with /s/, that is /s/!)
jones, d. gimson, a. c. ramsaran, s. (1988[) English Pronouncing Dictionary London:
Dent (British; now pronunciation has changed a bit, but it is still worthwhile consulting
evenly, to have a taste of all the nuances indicated and to be able to enter the spirit of the
language; IPA).
kenyon, j. s. (1950, 1924) American Pronunciation George Wahr: Ann Arbor (American
English; meanwhile, pronunciation has changed, but it is still worthwhile seeing; there is
also an augmented edition by others, but with no real additions, 1994; IPA).
knott, t. a. (1953) A Pronouncing Dictionary of American English Springeld, .:
Merriam (American pronunciation; meanwhile, certain pronunciations have changed, but
it is still worthwhile consulting; IPA {while, incredibly in the 3rd millennium mostly in
America, non-IPA dictionaries are still being published!}).
Macmillan English Dictionary for Advanced Learners (2002) Oxford: Macmillan Education
(strangely, this dictionary created in Britain and the usa only gives British pronunciation
and with only tiny detail; however, it is useful for the British stressing of lexical collocations; IPA).
Macquarie Dictionary, e (1997) Sydney: e Macquarie Library (encyclopedic, gives
Australian pronunciation; IPA).
Merriam-Webster's Collegiate Dictionary (2003) Springeld, Mass.: Merriam-Webster
(American pronunciation; interesting for its frequent phonetic variants, although it shows
absurd secondary stresses, since they are marked for almost every non-attenuated V unfortunately non-IPA).
Oxford Advanced Learner's Dictionary of Current English (2000) Oxford: O. Univ. Press
(British pronunciation with only major American dierences; it shows cases of marked
stresses for certain lexical collocations; IPA).
Oxford Advanced Learner's Dictionary of Current English Encyclopedic Edition (1992) Oxford:
O. Univ. Press (British pronunciation with only major American dierences; it shows cases of marked stresses for certain lexical collocations; IPA).
quaggiato, m. e. (1997/98) Fonetica inglese neutra e del Sud. Univ. of Venice, graduation thesis guided by L. Canepari (canIPA]
ragazzini, g. (1995) Dizionario inglese {italiano e italianoinglese}. Bologna: Zanichelli
(British pronunciation with only major American dierences; its transcriptions have been
revised by L. Canepari and A. Venturi but reduced by the editorial sta, however it presents the innovation of /E/ instead of /e/ and very many reduced forms, more than in pronunciation dictionaries, although less than in this HPr IPA).
Random House Dictionary of the English Language (e) Unabridged (1987) New York:
Random House (American; reliable for secondary stresses; but non-IPA).
Random House Webster's Collegiate Dictionary (1997) New York: Random House (American;
reliable for secondary stresses; but non-IPA).
roach, p. (2000) English Phonetics and Phonology. Cambridge: C. Univ. Press (British pronunciation; with 2 audiocassettes; IPA).
santipolo, m. (1997/98) A Socio-phonetic Description of Some Varieties of South-eastern British
English. Univ. of Venice, graduation thesis guided by L. Canepari (canIPA]
turrin, n. (1996/97) e Accents of Northern England. Univ. of Venice, graduation thesis guided by L. Canepari (canIPA]
upton, c. kretzschmar, w. a. jr. konopka, r. (2001) e Oxford Dictionary of Pronunci-
416
ation for Current English Oxford: O. Univ. Press (although it is the latest to be projected,
it lags behind its predecessors. In addition, it is exactly the opposite to the very convenient
diaphonemic and interphonemic transcriptions as a matter of fact, besides uselessly repeating also identical forms, in addition to those with slightly dierent phonetic renderings,
for every variant given it repeats the whole transcription {and not continuing the line, but
starting a new paragraph}, instead of indicating more clearly the sole dierences {failing
then to give the fundamental immediacy}. Besides, it uses dierent criteria and symbols for
the two accents { British and American}, leading the reader to think there are dierences
even where actually there are none, as for instance also for secondary-stress markings after
primary stresses, which are indicated in the American but not in the British pronunciation,
for forms which are instead absolutely identical such as teacake erefore, considering the
great amount of blank space and its many useless transcriptions, it uses twice the number
of pages actually needed: for instance, for the article a it uses 10 lines {ten!} to give in substance a /'EI, /; besides, it mixes up neutral pronunciations and others which are not (yet)
neutral, for ex. dune with British variant identical to June, /dZu:n/, or latter shown, in the
American pronunciation, exactly like ladder, /'ldr/, not even as a variant; mixed IPA]
Webster's New Encyclopedic Dictionary (1993) New York: Black Dog Leventhal (American
pronunciation; it marks destressable monosyllables; non-IPA).
wells, j. c. (1982) Accents of English Cambridge: CUP (3 vols.; extensive survey however
mainly based on available written sources; IPA).
wells, j. c. (2000) Longman Pronunciation Dictionary Harlow: Longman (British American pronunciations; together with Jones it provides a reliable survey, especially for British English; for American English, /:/ has been removed while it was present in the rst edition {1990}, although it could be more useful than /r/, which is still there, whereas /o:r/
has become /oUr/, which is more economical, but far less appropriate; IPA).
Italian ( 3)
bonazzi, i. (1989) Dizione e qualit espressiva della voce Torino: Centro Maier (with 6 very
short audiocassettes; traditional pronunciation; hypodiacritical spelling).
camilli, a. fiorelli, p. (1965) Pronuncia e graa dell'italiano. renze: Sansoni (simplied
).
canepari, l. (1970) tljn prnnsexn in Le Matre Phontique, 133:6-8 (as all MPh contributions, fully transcribed in IPA].
(1986) Italiano standard e pronunce regionali Padova: (with 2 enclosed audiocassettes,
the second one is about regional pronunciation; almost canIPA]
(2000 {amended modied reprint} 1999) Dizionario di pronuncia italiana. Bologna:
Zanichelli (60,000 forms with transcription and pronunciation variants, which correspond
at least to 180,000 actual words; pronunciations given: modern neutral, traditional neutral,
acceptable, tolerated, slovenly, intentional and lofty; canIPA]
(2004 {amended modied reprint} 1999) Manuale di pronuncia italiana Bologna: Zanichelli (with 2 enclosed audiocassettes; it introduces modern neutral pronunciation, in addition to the traditional one, besides other types, including 22 regional koins; canIPA]
costamagna, l. (1996) Pronunciare l'italiano Manuale di pronuncia italiana per stranieri. Perugia: Guerra (with 4 audiocassettes and a ; canIPA]
fiorelli, p. (1965) Crso di pronnzia italiana Padova: Radar (with 14 vinyl records; traditional pronunciation; hypodiacritical spelling]
migliorini, b. tagliavini, c. fiorelli, p. (1981) Dizionario d'Ortograa e di Pronunzia.
Torino: (still with I and J mixed together; it also contains family and place names; 1st
ed. 1969 with a vinyl record; non-IPA).
417
muljaCi, Z. (1972) Fonologia della lingua italiana Bologna: Il Mulino (traditional pronunciation; IPA]
tagliavini, c. (1965) La corretta pronuncia italiana Bologna: Capitol (with 26 vinyl records;
traditional pronunciation; hypodiacritical spelling]
Unfortunately, in the Journal of the International Phonetic Association (2004, 117-21), an implausible description of Italian was published (showing incredibly mixed and contrasting
features for a native speaker, independently from some queer transcriptions), which simultaneously presents both northern, central southern Italian characteristics and even a number of foreign ones. e text wording itself of e North Wind and the Sun is a new translation from English, for instance with il vento del nord the North Wind instead of la tramontana. An answer was given that four Italian scholars had approved it for publication.
e only recording used for the transcription is partially better (as one can immediately
hear), being somehow less possessed than the transcription provided in the paper.
French ( 4)
armstrong, l. e. (1932) e Phonetics of French London: Bell (reprinted many times; in spite
of its years, although pronunciation has slightly changed, it still remains good; IPA).
boch, r. (1995) Dizionario francese {italiano e italianofrancese}. Bologna: Zanichelli (its
transcriptions have been revised by L. Canepari but reduced by the editorial sta, however it presents the innovation of neutral /, / instead of the old and outdated /, /; IPA]
canepari, l. (in prog.) French Pronunciation* (with 4 national neutral pronunciations, besides
a mediatic and an international one, together with several native regional accents from
Europe and Canada, and some broad foreign accents, too; canIPA Perhaps also as Les prononciations du franais).
capelle, g. (1968) Le rythme et l'intonation de la phrase franaise Paris: Didier (a vinyl record
and booklet, with no notation).
carton, f. (1979) Introduction la phontique du franais Paris: Bordas (almost IPA).
coveney, a. (2001) e Sounds of Contemporary French Exeter: Elm Bank Publications (not
too reliable for things dierent from French, reported from controversial written sources;
IPA).
gleissner, h. (1964) Franzsische Aussprache Leipzig: Verlag Enzyklopdie (a vinyl record and
booklet; IPA]
kaneman-pougatch, m. pedoya-guimbretire, e. (1989) Plaisir des sons Paris: Alliance
Franaise/Hatier/Didier (with 4 audiocassettes; with no explicit intonation; IPA]
lerond, a. (1980) Dictionnaire de la prononciation Paris: Larousse (inconveniently separates
common and proper names into two sections; IPA with some additional symbols).
malmberg, b. (1976[) Phontique franaise Malm: Lromedel (IPA]
tranel, b. (1987) e Sounds of French Cambridge: C. Univ. Press (with an audiocassette;
IPA]
walker, d. c. (1984) e Pronunciation of Canadian French Ottawa: Univ. of O. Press; without intonation; almost IPA]
warnant, l. (1987, 1962/66) Dictionnaire de la prononciation franaise Paris/Gembloux:
Duculot (previous editions had an enclosed vinyl record and a less blank-page format; inconveniently separates common and proper names into two sections; IPA]
(1996) Orthographe et prononciation en franais Gembloux: Duculot (derived from the preceding title, gives the 12,000 words which are not pronounced according to their orthography; but does not give any proper names, which are often more problematical; IPA]
418
German ( 5)
Aussprachewrterbuch (2000[, 1962) Mannheim: Bibliographisches Institut F. A. Brockhaus
AG (the duden; also gives Chistian, family, and place names for various languages with
their original pronunciation; IPA with /a, a:/, but /r/, however, now at last accepts /r/-vocalization even after short V although it continues to transcribe only /r/).
barberis, p. sarnow, a. (2000) Phonetik lehren und lernen Beiheft Italien Torino: Paravia
(with 3 audiocassettes; IPA]
bithell, j. (1952) German Pronunciation and Phonology London: Methuen (IPA]
canepari, l. (in prog.) German Pronunciation*. Mnchen: Lincom Europa (with 4 national
neutral pronunciations, together with several native regional accents and some broad foreign accents, too; canIPA Perhaps also as Die Aussprachen des Deutschen)
Dizionario Garzanti di tedesco} (1994) Milano: Garzanti (the transcriptions have been planned
by L. Canepari, prepared by P. Paschke, but reduced by the editorial sta; however, it gives
many reduced forms (on p. 568), although less than those given in this HPr; anyway, it is
the rst dictionary which unies /:, , 0/ with // just for the sake of uniformity
with other publications instead of giving, in a more complex way, /:, K, / for /:,
, 0/ or, as here, /:K, K, 0/ (:, , 0); IPA).
Groes Wrterbuch der deutschen Aussprache (1982) Leipzig: Bibliographisches Institut (the
gwda; IPA but with /a, A:, r/, and /Y/ instead of /Y/, in the style of Soviet fonts).
kohler, k. j. (1977, 1995) Einfhrung in die Phonetik des Deutschen Berlin: Schmidt (IPA
but with strange choices for symbols, mitigated in the latest edition).
(1994) Glottal Stops and Glottalization in German in Phonetica, , 38-51 (IPA]
sterreichisches Wrterbuch (1998) en: bv Pdagogischer Verlag (non-IPA but now at
least it gives a table of correspondences).
Siebs Deutsche Aussprache (1969) Berlin: De Gruyter (a handbook and a small pronouncing
dictionary; IPA but with /A, A:/ and /r/).
sttzer, u. (1975) Deutsche Aussprache Leipzig: Verlag Enzyklopdie (with 2 vinyl records;
IPA but with ('a, a, 'A:, A, r), and /Y/ instead of /Y/, in the style of Soviet fonts, and also
//(:), (:), :, a:// for /y:, y, Y :, , E:, a:/).
Spanish ( 6)
Unfortunately, no Spanish pronouncing dictionary exists yet! Even Antonio Quilis' meritorious
project of preparing one, with two neutral accents: Iberian and American (also with a version,
with the possibility of recording one's voice and comparing it with the model provided) could not
be fullled
canepari, l. (in prog.) Spanish Pronunciation*. Mnchen: Lincom Europa (with 6 or 7 national neutral pronunciations and an international one, together with several native regional
accents from Europe and America and some broad foreign accents, too; canIPA Perhaps also as Las pronunciaciones del espaol).
dalbor, j. b. (1980) Spanish Pronunciation Fort Worth: Holt, Rinehart and nston (American with Iberian indications; numerical intonation with pitch levels; non-IPA]
Gran Diccionario de la lengua espaola (1985) Madrid: (Iberian; the only dictionary with
generally reliable transcriptions for Iberian Spanish, in spite of many attempts to make them
not to appear so; almost IPA]
miotti, r. (1995-96) Descripcin fono-tontica de las variedades regionales del espaol de Amrica
y Espaa Univ. of Venice, graduation thesis guided by L. Canepari (Iberian and American,
419
with 12 regional koins; canIPA]

(1998) Descrizione fono-tonetica delle variet regionali dello spagnolo d'America e di Spagna
in Annali di Ca' Foscari 393-440 (derived from the thesis, but laid out according to phonemic elements; Iberian and American; canIPA]
navarro toms, t. (1932[, 1918) Manual de pronunciacin espaola Madrid: Publicaciones
de la Revista de lologa Espaola (Iberian; still being reprinted; it remains the only reliable work for Iberian Spanish, in spite of an excessive subphonemic and biased hyperdierentiation and a non-IPA alphabet]
rohwedder, e. (1969) Spanische Aussprache Leipzig: Verlag Enzyklopdie (American; a vinyl
record and a booklet; non-IPA]
saussol, j. m. (1983) Fonologa y fontica del espaol para italfonos Padova: Liviana (Iberian;
with an audiocassette; semi-IPA]
Portuguese ( 7)
Unfortunately, no Portuguese pronouncing dictionary exists yet (and with metaphony).
canepari, l. (in prog.) Portuguese Pronunciation*. Mnchen: Lincom Europa (with {Brazilian
and Lusitanian} neutral pronunciations and an international one, together with several native regional accents from Europe, America and Africa and some broad foreign accents, too;
canIPA Perhaps also as As pronncias do portugus).
cristfaro silva, t. (2001) Fontica e fonologia do portugus So Paulo: Contexto (Brazilian;
with a ; IPA]
cuesta, p. vzquez da luz, m. a. mendes (1971) Gramtica portuguesa Madrid: Gredos
(Lusitanian with Brazilian indications; non-IPA]
Dicionrio da lngua portuguesa (1981) Porto: Porto Editora (Lusitanian; does not transcribe,
but indicates with italic letters in brackets occurrences of /'e, 'o/ [ ] cases of unstressed
/E, a, O/ [ , ] which is one of the uses of our /, A, / and consonant clusters including x).
Dicionrio da lngua portuguesa contempornea (2001), Lisboa: Editorial Verbo (Lusitanian;
too much proudly presented as the publishing enterprise of the beginning of the 21st century and of the 3rd millennium, which began before the French Revolution and nally led
to this two-volume dictionary; it does indicate metaphony, however, giving only plurals
with dierent vowel timbres, and hiding them at the end of the entry, not showing them
immediately after the headword, together with the other transcriptions, which include feminine forms; in addition, it puzzles a lot because in the 3rd millennium it displays such
things as /, g, $, R, ], l/, instead of /, g, S, K, , L/, and even /, , , Z/ instead of /E, ,
O, Z/; semi-IPA).
grtner, e. thomaz jayme, j. (1984) Portugiesische Aussprache Leipzig: Verlag Enzyklopdie
(Brazilian; a vinyl record and booklet; almost IPA]
giangola, j. p. (2001) e Pronunciation of Brazilian Portuguese Mnchen: Lincom Europa
(without intonation; hybrid IPA with (y, w) for (j,i w, u))
irmen, f. cortes kollert, a. m. (1995) Taschenwrterbuch _ PortugiesischDeutsch {Deutsch
Portugiesisch} Berlin: Langenscheidt (Lusitanian; indicates metaphonetic variations, but
not systematically; presents /e/ + /S, Z, N, L, j, i/, for (), but (Ai), for (3I, I) and for (Ai,
4i); IPA]
mea, g. (2003) Dicionrio de PortugusItaliano Porto: Porto Editora (Lusitanian; in the present reprint the transcriptions have been added {whereas earlier they were only present in
the reverse part}, which is phonemic even if between ( ), with (b, d, g) in every context; it
420
distinguishes (l), but uses (!) in place of (); in addition, it gives (j, w) for (i, u),
and () for (), instead of //; it indicates unstressed V which do not undergo reduction, but not always in a reliable way; metaphony is indicated if new separated entries are
present; IPA)
Pequeno dicionrio {italianoportugus} portugusitaliano (1993) So Paulo: Melhoramentos
(Paulista {pron. of So Paulo}; indicates some metaphonetic variations; IPA]
Pocket {EnglishPortuguese} PortugueseEnglish Dictionary (2001) Glasgow: Collins (Carioca
{pron. of Rio de Janeiro}; indicates some metaphonetic variations; IPA]
tagliavini, c. (1938) Grammatica elementare della lingua portoghese Heidelberg: Groos (Lusitanian; cana old-style IPA).
viana, a. r. gonalves (1903) Portugais. Phontique et phonologie _ Morphologie _ Textes Leipzig: Teubner (Lusitanian; still the most reliable work, in spite of several inevitable misprints;
archaic IPA]
(1912) Vocabulrio Ortogrco e remissivo da lngua portuguesa Paris/Lisboa: Aillaud/Alves
(still the constant reference work of various authors and lexicographers for the even limited phonetic cues; current spelling with some additional diacritics. For the Portuguese orthography, it bases itself on a rather rational structuring, followed by several semi-reforms,
with qualities and failings {for foreigners, the latter would not be a problem any longer if
somebody were determined to prepare a Portuguese pronouncing dictionary, by taking account of the problems of metaphony and of Lusitanian peculiarities, which are completely
ignored by present-day ocial spelling, whereas that of the intermediate time was denitely more careful}).
Russian ( 8)
Unfortunately, no IPA Russian pronouncing dictionary exists yet, with full transcriptions of the
dierent words (also avoiding archaic-Soviet morphonological or orthographical symbols,
mainly based on the Cyrillic alphabet).
ageenko, f. l. zarva, m. v. (1993) Clobap ydapen$= pycckogo rvka. Mockba: Pycck$=
Rvk (hypodiacritical spelling for stress and major exceptions, which indicates neither neutralizations nor all assimilations, inconveniently separates common and proper names into
two sections, separately signed {in this edition}).
borunova, s. n. voroncova, v. l. es'kova, n. a. (1999) Opfo<$ueck$= clobap pycckogo rvka. Mockba: Pycck$= rvk (hypodiacritical spelling for stress and major exceptions, which indicates neither neutralizations nor all assimilations, gives only common
words; many pages, but with too much blank space in the two columns).
boyanus, s. c. (1955) Russian Pronunciation $ Russian Phonetic Reader London: Lund Humphries [IPA]
comrie, b. stone, g. polinsky, m. (1996) e Russian Language in the Twentieth Century
Oxford: Clarendon Press (archaic-Soviet IPA).
husler, f. (1974) Russische Aussprache Leipzig: Verlag Enzyklopdie (a vinyl record and booklet; non-IPA]
jones, d. ward, d. (1969) e Phonetics of Russian Cambridge: C. Univ. Press [IPA; it remains the best treatment on the subject, extended and up-to-date, in spite of its archaic-Soviet phonological choices).
kalenCuk, m. l. kasatkina, r. f. (1997) Clobap tpydnocte= pycckogo po$nowen$r.
Mockba: Pycck$= rvk (hypodiacritical spelling for stress and major exceptions, which indicates neither neutralizations nor all assimilations).
kasatkin, l. krysin, l. Zivov, v. (1995) Il russo renze: La Nuova Italia (non-IPA a kind
421
of phono-transliteration of archaic-Soviet derivation, which mixes and confuses the phonic

and graphic levels).
shapiro, m. (1968) Russian Phonetic Variants and Phonostylistics Berkeley/Los Angeles: Univ.
of California Press [archaic-Sovietized IPA)
trofimov, m. v. jones, d. (1923) e Pronunciation of Russian Cambridge: C. Univ. Press
[IPA; the only one with a realistic phonic {pre-Sovietic} structuring, in spite of a pre-revolutionary orthography and traditional pronunciation)
Arabic ( 9)
Unfortunately, no Arabic pronouncing dictionary exists yet, with full IPA transcriptions of the
dierent words.
al-ani, s. h. (1970) Arabic Phonology e Hague: Mouton (almost IPA).
canepari, l. (1983) Fonetica e tonetica araba in Scritti linguistici in onore di Giovan Battista
Pellegrini Pisa: Pacini, 1105-21 [sacriced IPA for typographical limitations, and with a
photographed consonant table, but cleaned up by an extremely clever editorial sta,
who cancelled ' and , used for and H!).
cantineau, j. (1960) Cours de phontique arabe Paris: Klincksieck (non-IPA).
gairdner, w. h. t. (1925) e Phonetics of Arabic Oxford: O. Univ. Press (without intonation;
IPA]
kstner, r. t. (1981) Phonetik und Phonologie des modernen Hocharabisch Leipzig: Verlag Enzyklopdie (without intonation; IPA]
krotkoff, g. (1976) Taschenwrterbuch _ ArabischDeutsch Berlin: Langenscheidt (in spite of
its gaps and naveties; almost IPA]
mitchell, t. f. (1972) Arabic Pronunciation London: (a vinyl record and booklet; transliteration with upper-case letters for emphatic consonants and with for /H/; non-IPA).
(1990) Pronouncing Arabic. 1. Oxford: Clarendon Press (transliteration with upper-case letters for emphatic consonants and with for /H/, ? for //, @ for (a, A, , ), in addition to inelegantly barred letters both horizontally and diagonally for constrictives; non-IPA,
neither , nor).
nasr, r. t. (1967) e Structure of Arabic Beirut: Librairie du Liban (with some numerical
intonation; semi-IPA]
Hindi ( 10)
Unfortunately, no [IPA] Hindi pronouncing dictionary exists yet
chaturvedi, m. tivari, b. n. (1975 D) A Practical Hindi-English Dictionary Delhi: National Publishing House (with transliteration using a: = /aa/, a = /A/, and CC also for /0/ + /w,
j, r, l/).
jones, w. e. (1971) Syllables and word-stress in Hindi in Journal of the International Phonetic
Association 1:74-78 (IPA).
kachru, y. (1990) Hindi-Urdu in b. comrie (D) e Major Languages of South Asia, the Middle East and Africa London: Routledge, 53-72 (mixed IPA]
matthews, w. k. (1954) fnetiks nd fnoldi v hindi:, in Le Matre Phontique 102:18-22
(as all MPh contributions, fully transcribed in IPA].
mcgregor, r. s. (1977) Outline of Hindi Grammar Delhi: Oxford Univ. Press (traditional diacritical transliteration].
422
mehrotra, r. c. (1965) Stress in Hindi in Indian Linguistics 26:96-105 (semi-IPA].

pandey, p. k. (1989) Word Accentuation in Hindi in Lingua 77:37-73 (semi-IPA transliteration].
ray, p. s. (1966) Hindi-Urdu Stress in Indian Linguistics 27:95-101.
rosini, s. (1979-80) Indagine sulla struttura fonetica e tonetica della lingua hindi Univ. of Venice, graduation thesis guided by L. Canepari (traditional diacritical transliteration with IPA
indications).
shukla, s. (1999) Hindi Phonology Mnchen: Lincom Europa (with semi-IPA transliteration,
without intonation).
(Mandarin) Chinese ( 11)
Unfortunately, no [IPA] Chinese pronouncing dictionary exists yet
chao, y. r. (1948) Mandarin Primer Cambridge (.): Harvard Univ. Press (with vinyl records; non-IPA, but with tone-letters {now ocialized but trivialized by the reform of the
IPA}).
cheng, c-c. (1973) A Synchronic Phonology of Mandarin Chinese e Hague: Mouton (almost
IPA]
dow, f. d. m. (1972) An Introduction to the Pronunciation of Chinese Edinburgh: E. Univ. Press
[IPA, but with not too reliable transcriptions).
(1972) An Outline of Mandarin Phonetics Canberra: Australian National Univ. Press [IPA,
but with not too reliable transcriptions).
kratochvl, p. (1968) e Chinese Language Today London: Hutchinson (Chinese/Karlgrenized IPA: with q for (M), that is with a vocoid still considered as if it were the intense contoid (); on the other hand, with the vowel Q for the true intense contoid ()]
norman, j. (1988) Chinese Cambridge: C. Univ. Press [Sino-like IPA: with , for (M, )]
Practical Chinese Reader (1985) Beijing: e Commercial Press (with audiocassettes; Chinese/Karlgrenized IPA: with q for (M), that is with a vocoid still considered as if it were
the intense contoid (); on the other hand, it has the vowel Q for the true intense contoid
()]
qin, z. (1980) On Chinese Phonetics Beijing: e Commercial Press [Chinese/Karlgrenized
IPA: with q for (M), that is with a vocoid still consideres as if it were the intense contoid
(); on the other hand, it has the vowel Q for the true intense contoid ()]
shan pao li, j. (1975) Introduction to the Chinese Language Taipei: Ke Ji Publishing Co. [moderately Sino-like IPA]
speSnev, n. a. (1980) Fonet$ka k$ta=ckogo rvka. Len$ngpad: *datelctbo Len$ngpadckogo Yn$bepc$teta (a very hybrid IPA]
Unfortunately, in the Journal of the International Phonetic Association itself Standrad Chinese (Beijing) has appeared {2003, 109-112}, which is semi-Karlgrenized IPA, with both (M,
) rendered as (), but described as () and (), including many more transcriptional queer
inaccuracies.
423
Japanese ( 12)
Unfortunately, no IPA Japanese pronouncing dictionary exists yet
akamatsu, ts. (1997) Japanese Phonetics Mnchen: Lincom Europa [IPA]
fujito, y. nakano, e. seton, c. (1979) Japanese Pronunciation Guide Tokyo: Bonjinsha
(with audiocassettes; transliteration, not transcription).
jorden, e. h. (1963) Beginning Japanese New Haven: Yale Univ. Press (with tapes or audiocassettes; transliteration, not transcription).
kindaichi, h. akinaga, k. (2001) Shinmeikai Nihongo Akusento Jiten Tokyo: Sanseido (non-IPA\ tonetic and diacritic katakana, with variants).
martin, s. e. (1990) Martin's Pocket Dictionary Tuttle: Rutland (transliteration, not transcription; indicates tonemic variants and /i, M/ devoicing).
New Japanese-English Dictionary (1974[) Tokyo: Kenkyusha (non-IPA tonetic transliteration).
Nihongo Hatsuon Akusento Jiten (1966) Tokyo: Nihon Hoso Kyokai (non-IPA\ tonetic katakana).
tsujimura, n. (1996) An Introduction to Japanese Linguistics Oxford: Blackwell (semi-IPA]
vance, t. j. (1987) An Introduction to Japanese Phonology Albany: State Univ. of New York
Press (semi-IPA]
Esperanto ( 13)
canepari, l. (1998) Non c' proprio niente da dire sulla pronuncia e la graa dell'esperanto? in
La linguistica, le lingue pianicate e l'Esperanto a special issue of L'Esperanto 44-55
[canIPA]
cresswell, j. hartley, j. sullivan, l. h. (1987) Teach Yourself Esperanto Sevenoaks:
Hodder Stoughton (with an audiocassette of Anglo-Esperanto; spelling).
kalocsay, k. waringhien, g. (1980[) Plena analiza gramatiko de esperanto Rotterdam:
(spelling).
wells, j. c. (1978) Lingvistikaj aspektoj de esperanto Rotterdam: / (spelling).
(s. a.) Esperanto Pronunciation Disc London: (a vinyl record and leaet; spelling).
zamenhof, l. l. (1962) Lingvaj respondoj Marmande: (spelling).

Handbook of Pronunciation Many Languages - Luciano Caneparia

Diunggah oleh

Informasi Dokumen

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Handbook of Pronunciation Many Languages - Luciano Caneparia

Diunggah oleh

Hak Cipta:

Format Tersedia

From: L.

Canepari (2005) A Handbook of Pronunciation Lincom Europa

2. English American British (neutral other accents)

4. French (neutral other accents)

5. German (neutral other accents)

6. Spanish Iberian American

7. Portuguese Brazilian Lusitanian

11. Chinese Mandarin

From: L. Canepari (2005) A Handbook of Pronunciation Lincom Europa

headship ({Amer./Brit.} Eng.)

('kT, 'T) ('wks, 'ws)

wicks (mediatic Brit.)

('sud:H, -d:) ('Es:th, -t)

('pe) ('p#) ('pi) ('p) ('ppa) ('pis)) ('tX)

alto (Lus. Port.)

present or not, depending on vocal-fold vibrations during the productions of phones.

1.5. Vocoid classication (with labiograms of high vocoids).

1.6. Vocoid orograms.

nuances and dierences better.

1.8. Labiograms for the dierent vocoids.

1.9.2. Semi-nasal contoids.

1.10. Stop contoids.

1.11.1.. Grooved stop-strictive contoids.

1.11.2. (Grooved and at) stop-semi-strictive contoids.

1.12.1.. (at) constrictive contoids.

1.12.1.. Grooved constrictive contoids.

1.12.2. (Grooved and at) semi-constrictive contoids.

1.13.1. Approximant contoids.

1.13.2. Semi-approximant contoids.

1.13.3. Lateralized approximant contoids.

1.14.1. Trill contoids.

1.14.2. Tap contoids.

1.15.1. (Bi)lateral contoids (and two semilateral (, ))).

1.15.2. Unilateral contoids.

1.15.3. Lateral tap contoids.

1.16.1. (Two perspectives of) consonant labiograms.

1.16.2. Further consonant labiograms (only front view).

1.16.3. Consonant orograms with dierent labial positions.

1.17. Consonant dorsograms and palatograms.

(appr.) unilateral constr. unilat.

see you on Sat

don't] see you

don't] see you

1.22. e four neutral intonemes of British English.

1.23. Two neutral preintonemes of British English.

low parenthesis: ( ) / / mid parenthesis: ( ) / / quotation: (^ ) /^ /

1.26. e four basic ton(em)es of (Mandarin) Chinese.

1 /5/ (5) >1

4 /6/ (63) >Z

with all the imperfections that this inevitably implies).

the inconsistency of these rules.

on a vocogram (as well as on an orogram or on a tonogram).

unstressed /u/ (%)

1.4.4. But let us now proceed to the conventions concerning diphthongs

wide, with rounded

1.4.6. Moreover, according to the direction of their movement, diphthongs can

1.31. Dierent diphthongs.

1.32. Slit and grooved constrictives: () and (s).

1.35. Groupings of the fundamental manners of articulation.

reduced V (in duration; = )

e ocial IPA chart

1.36. Ocial IPA Chart (1996).

Bilabial Labiodent. Dental

Alveolar Postalveol. Retroex