Anda di halaman 1dari 8

Polish and English syllable structures.

How different are they?


Eugeniusz Cyran

0. Introduction
When confronted with such word-forms in Polish as kto who, rdest water pepper, mga
fog, tkn touch, pstry gaudy, or pastw country, gen.pl. and nastpstw consequences,
gen.pl., in which the initial and final clusters, which are traditionally assumed to correspond to
syllable onsets and codas respectively, may contain up to four consonants in the former case
and up to five in the latter, it appears that a sensible comparison between this language and
English is not possible. Even though English word-initial clusters may involve up to three
consonants, their melodic make-up is very restricted, in that they must have the fricative [s]
followed by the voiceless stop and a liquid or glide, for example, spr, spl, skr, str, skw, spj as in
spring, splash, skream, string, squash, spew. Two-consonant clusters are also constrained and
do not allow for any homorganicity (*tl, *pw), or insufficient sonority profiles as do Polish
initial clusters such as tk, rd, pt, gn and so on. A similar problem concerns the word-final
context. Polish may have up to five consonants, while English allows only up to two.
Additionally, in English the final cluster must be of falling sonority, e.g. lt, nd, ft as in belt,
hand, theft, while in Polish we also have rising sonority clusters, e.g. tr, kr, kl as in wiatr
wind, akr acre, cykl cycle.
In this paper, I will attempt to show that in some theories of phonological organisation the
enormous differences between such disparate systems as English and Polish are only apparent
and they can be reduced to a handful of parameters. This would be a welcome discovery since
the basic assumptions of universal linguistics suggest to us that linguistic systems cannot differ
too much in reality. In what follows, I will introduce the basics of a model of phonological
organisation called Government Phonology. The discussion will first concentrate on the
structure of onsets and the beginning of the word in the two languages and then some
indication is provided as to how the word-final clusters should be understood.

1. Syllabification in Government Phonology


Government Phonology (GP) is a theoretical model in which the main assumption is that
speech sounds organise themselves according to their internal properties and a few general
and very simple principles.1 Syllabification in GP proceeds from governing relations
contracted between consonants. Whether a consonant is a governor-(T) or a governee-(R) in
such relations is determined by their sonority differential. This means that a less sonorous
segment governs or controls the more sonorous one regardless of their linear order, as
illustrated below in (1a). In terms of the actual syllabic configurations (1b), the rightward
governing relation defines branching onsets and the leftward direction specifies a relation
between an onset and the preceding non-vocalic complement of a branching rhyme, that is, the
coda. It is obvious then, that government is the underlying principle of speech organisation
from which we can directly derive phonotactics, that is, melodic restrictions on possible
complex onsets, as well as on syllable contacts.
(1)

a.

b.
O

R
v

l d

) = direction of government, T = governor, R = governee

Let us disregard the exact substantive properties of g and l, which make them interact in this
particular fashion, and assume the traditional terminology that in a sequence of two consonants
T,

the governor, is less sonorous than R, that is, the governee. Though it is not impossible to

assign a fixed function to some segments as typical governees, for example, glides, or typical
governors, for example stops, we will assume that these functions are always worked out for
any given sequence. For example, f is likely to be a governor when adjacent to a liquid, as in
fling or alpha, or a governee when followed by a stop, as in hefty.

The literature on Government Phonology is quite sizeable. I recommend, e.g. Gussmann (2002), Harris
(1994), Kaye, Lowenstamm and Vergnaud (1990).

Government, however, should not be viewed as a mere theoretical rephrasing of contact


laws and sonority sequencing.2 One advantage of the model is that the nature of government
restricts possible syllabic types, because in any given direction only two positions the
adjacent ones may contract a governing relation. This, effectively, allows only for maximally
binary branching constituents: branching onsets, nuclei, and rhymes. This makes the model
highly constrained. On the other hand, there is nothing in the standard generative models or
Optimality Theory to constrain the size of syllables other than observation turned into language
specific constraints. All the possible syllabic constituents which are recognised in standard GP
are listed below.
(2)

a. onset

b. nucleus

c. rhyme

R
|
N
|
x

R
N
|
x

As for the simplex structures, it must be mentioned that a non-branching rhyme is in fact
identical to a non-branching nucleus and refers simply to a short vowel. Branching
constituents, on the other hand, may be defined as involving governing relations which are
from left to right. The only governing relation which goes in the opposite direction is that
between an onset and the preceding rhymal complement. Note that ternary structures would
either violate adjacency between governor and governee, or the directionality of governing
relations. Therefore, there are no ternary branching constituents. The model allows for a simple
definition of the syllable structure of a given system in that what is required is a statement
concerning the ability of particular constituents to branch, a statement which may be couched
in terms of parameters, for example. This is, in fact, the standard way of capturing syllable
typology in GP, which is illustrated below.
(3)
Branching

Polish

English

Onset

ON/OFF

ON

ON

Rhyme

ON/OFF

ON

ON

Nucleus

ON/OFF

OFF

ON

These refer to shape of sonority profiles on the border of two syllables and within a single syllable
respectively.

If the parameter on branching onsets is set in the

OFF

position, the system only has simplex

onsets. On the other hand, if the parameter is switched

ON,

the system possesses both

branching and simplex onsets. The parameter for branching rhymes in fact determines the
existence of internal codas, and, in a system which has long vowels, the possibility of having
super-heavy rhymes, e.g. bold, find, etc.
According to this model, the syllabic systems of Polish and English differ in terms of one
parameter (3): in Polish the parameter for branching nuclei is switched

OFF.

This effectively

deprives Polish of long vowels and super-heavy rhymes. Otherwise the systems may be said to
be similar; however, the complex initial and final clusters in Polish require an additional
explanation.

2. Complex word-initial clusters in Polish and English


Languages like Polish may begin their words with clusters exceeding the number of two
consonants. Because a governing relation in a given direction can be contracted only between
two consonants (adjacency), and consonants of the same governing properties cannot form a
relation, the only possible configuration in tkliwy touchy is one in which kl forms a branching
onset which is separated from the preceding consonant t by an empty nucleus. Likewise, in
krwi blood, gen.sg., a governing relation can be contracted between the first two consonants
which form a branching onset and must be separated from the following onset by an empty
nucleus. Similarly, if there are only two obstruents or sonorants, for example, tka weave and
lnu flax, gen.sg., the consonants also form separate onsets, that is /tPkatP/ and /lPnu/.3
Thus, the forms such as those in (4a) and (4b) are not treated as ternary branching onsets
but rather as a sequence of two well-formed onsets, which are separated by an empty nucleus.4
(4)
a. O N O
t P k
|
|
T
T
touchy

N
l i w y
|
R

b. O

N O N

k
r P w i
|
|
|
T
R
R
blood, gen.sg.

As seen in (4a), the only two consonants with a sufficient sonority differential are /k/ and /l/,
and they may contract a governing relation, while the first consonant must form a separate
3

Some support for the presence of the internal empty nucleus in krwi and lnu comes from the fact that it
alternates with [e]: krew / krwi [kref ~ krfi] blood / gen.sg., len / lnu [len ~ lnu] flax / gen.sg..
4
For a thorough analysis of these facts see Cyran and Gussmann (1999).

onset. In (4b) on the other hand, the first two consonants form a branching onset, while the
third consonant forms a simplex onset. It must be mentioned that the above analysis of the
three-consonantal clusters is very much in the spirit of Kuryowicz (1952), who also proposed
to treat such forms as sequences of onsets. The only difference is that in GP these onsets must
be separated by an empty nucleus.
Given that Polish seems to allow for the presence of an empty nucleus inside the wordinitial clusters, we predict that the biggest possible cluster in this language will consist of a
sequence of two branching onsets. This seems to be borne out by the form drgn shudder
which is represented below.
(5)
O
d

N
|
P

O
g

N
|

As in the case of krew / krwi blood / gen.sg., there are data in Polish which support the
assumption that drgn contains a vocalic site between two branching onsets. This site is
sometimes utilised by morphology to express, e.g. aspectual functions as in drygn shudder
once.
It seems then, that it is not the structure of onsets that is different between English and
Polish but a separate property of word structure which says that an empty nucleus can be
utilised inside word-initial consonant sequences. This single parameter accounts for the
apparently enormous difference between the word structures of the two languages.
Finally let us turn to a three-consonant cluster which is present both in English and in
Polish.
(6)

a.

b.
R

s
t
r y
|
|
|
R
T
R
[bstr] bystry clever

s
t
|
|
R
T
strong

r o
|
R

b y

In (6a), the stop governs in both directions. Note, however, that this cluster is not a constituent
as a whole, because only rightward government defines constituency. Leftward relations are of
an interconstituent nature. Thus, the fricative s finds itself in the coda position of the preceding
branching rhyme. Recall that governing relations depend on the internal properties of segments

which roughly correspond to sonority. Thus, we must assume that any instance of the
configuration s+stop+liquid will contract exactly the same governing relations, in any
language. For this reason, a word which phonetically begins with this sequence must also
exhibit the same syllabic configuration. This is illustrated in (6b). Since, the fricative is a coda,
that is, the complement of a branching rhyme, it is assumed that the constituent rhyme contains
a nucleus which is phonetically empty. Given the fact that related languages differ in terms of
allowing for initial s+stop or not, for example, the word for stadium is pronounced as
[stadio] in Italian and [estadio] in Spanish, it is assumed that also in this case we are dealing
with parametric variation. The term magic licensing is used to express the fact that the
phenomenon is not yet fully understood.5 However, as can be seen, Polish and English do not
differ with respect to this parameter.

3. Right edge of the word in Polish and English


In GP, surface word-final consonants are always followed by an empty nucleus phonologically.
Thus, there is no structural difference between forms with or without the inflectional vowel.
The only condition on the existence of word-final consonants or consonant clusters is whether
they can appear before an empty nucleus. Nuclei are licensers of onsets, so we may use the
formal term licensing, and differentiate between the types of licensers such as full vowels and
empty nuclei to determine what structures are possible in Polish and English.
(7)

a.

b.
O N

m a t a/P
mata/mat
mat,nom./gen.pl.

c.
R

O N

N
|
m a
r
t a/P
Marta/Mart
name,nom./gen.pl.

w i a t
r u/P
wiatru/wiatr
wind, gen./nom.

Thus, it seems that in Polish, the empty nucleus is able to license the same formal structures as
melodically filled vowels. However, there are some substantive differences. An empty nucleus,
by definition, is a weaker licensor than its melodically filled congener. For example, it is unable
to license voice in the word-final obstruents.

For detailed discussions of magic licensing see Kaye (1992) and Harris (1994).

English differs from Polish in one respect here. Namely, branching onsets cannot appear
word finally as they do in Polish (7c). Otherwise, there are single word-final consonants as in
pot, cat, etc., which structurally will correspond to (7a). There are two-consonant clusters of
falling sonority profile corresponding to (7b) as in belt, hand. But there will not be rising
sonority clusters.
The maximal well-formed cluster to be found word-finally in Polish may therefore contain
three consonants of which the first will be the rhymal complement preceding a branching onset
as illustrated below on the basis of the word sistr sister,gen.pl..
(8)
O

N
|
s
u
s
t
r
P
In English this structure is not possible for obvious reasons, namely, while st is allowed finally,
str is not because it would contain the ungrammatical branching onset word-finally.6
Earlier we saw that Polish uses sequences of onsets word-initially by allowing an empty
nucleus inside the clusters. It seems that the same tactics is also used word-finally. Below, I
propose the phonological structure for the biggest cluster found at the end of the word in the
Polish form nastpstw [nastempstf] consequence, gen.pl..7
(9)
O

N
|
a

N
|
e

N
|
P

Notice that this structure shows that the maximal binarity of syllabic constituents can indeed be
retained, which is good news for universal grammar, because it means that the typology of
syllabic structure is very simple and can be defined in terms of three parameters mentioned in
(3), while word-structure parameters are additional aspects of universal grammar although they
also can be boiled down to a handful of structural possibilities, for example, the utilisation, or
not, of empty vocalic sites in the representation of words. Only a restrictive model like the one
presented above can make comparisons between languages as different as English and Polish in
a meaningful way.
6

English does have surface sequences of more than two consonants, but they are always morphologically
complex and should not be mistaken as a true cluster, for example, sixths, belts, lengths, etc.
7
For a slightly more elaborate analysis of the right edge of words in Polish see Cyran (2003).

Conclusion
Formally, syllable structure is defined by three parameters on the possibility of the three
constituents; onset, nucleus, rhyme to branch. Polish differs from English only with respect to
one of these parameters, in that it does not allow its nuclei to branch. There are, of course,
some additional differences concerning the substantive aspect, that is, the melodic restrictions
on the shape of branching onsets, where Polish seems to be a bit more relaxed than English.
The most visible difference between Polish and English lies in the fact that Polish allows for
complex clusters both word-initially and word-finally. A strict application of GP principles
identifies their structure as a sequence of well-formed maximally branching constituents,
which, individually, are not much different from the English ones. Thus, the main distinction
lies in the presence or absence of cluster-internal empty nuclei. Polish allows for it, while
English does not. The same parameter accounts for word-final clusters in both languages.
To conclude, we are able to boil down the difference between English and Polish wordstructures to two parameters; the one on the presence of branching nuclei and the one on the
presence of internal empty nuclei. The slightly relaxed melodic restrictions in Polish require
further clarification.
References
Cyran, E. (2003) Complexity scales and licensing strength in phonology. Lublin:
Wydawnictwo KUL.
Cyran, E. and E. Gussmann (1999) Consonantal clusters and governing relations: Polish initial
consonant sequences. In H. van der Hulst and N. Ritter (eds.) The syllable. Views and
facts, 219-247. Berlin: Mouton de Gruyter.
Gussmann, E. (2002) Phonology: analysis and theory. Cambridge: Cambridge University
Press.
Harris, John (1994) English sound structure. Oxford: Blackwell.
Kaye, J. (1992) Do you believe in magic? The story of s+C sequences. SOAS Working Papers
in Linguistics and Phonetics 2, 293-313. Reprinted in H. Kardela and B. Szymanek
(eds.) A Festschrift for Edmund Gussmann from his friends and colleagues, 155-176.
Lublin: Redakcja Wydawnictw KUL.
Kaye, J., J. Lowenstamm and JR. Vergnaud (1990) Constituent structure and government in
phonology. Phonology 7, 193-231.
Kuryowicz, Jerzy (1952) Uwagi o polskich grupach spgoskowych. Biuletyn Polskiego
Towarzystwa Jzykoznawczego 11: 54-69.
8

Anda mungkin juga menyukai