, Amsterdam
Not to be reproduced in any form without written permission from the publisher.
Jack Sidnell
Northwestern University/University of Toronto
Introduction
1 Funding for the research reported in this paper was provided by the Social Sciences
and Humanities Research Council of Canada, the Wenner-Gren Foundation for Anthropological
Research, the Ontario Graduate Scholarship Program and the University of Toronto Alumni
Association. Many more people than I can list have commented on some version of this paper.
I’d like in particular to thank Gord Easson, Morris Goodman, Miriam Meyerhoff and Don
Winford as well as an anonymous reviewer for this journal. Finally, thanks go to the Indo-
Guyanese people who generously opened up their homes and lives to me between 1994–1996.
The author alone is responsible for any remaining inadequacies.
2 An exhaustive listing of works on tense-aspect in pidgin and creole languages (or even
those which are concerned with Caribbean English Creoles) would constitute its own indepen-
152 JACK SIDNELL
Creole in particular than it is for pidgin and creole languages in general (see
for example Allsopp, 1962; Bickerton, 1975; Edwards, 1975, 1991; Gibson,
1982, 1986, 1988; Jaganauth, 1987, 1994; Mufwene, 1984a, 1986; Rickford,
1987; Winford, 1993a). An important goal of this research has been to charac-
terize the sometimes extreme variation one finds in this area of the grammar
and to determine its bearing on issues of language change and contact. Bicker-
ton (1975) for example attempted to illustrate the types of diachronic change
which were putatively reflected in synchronic patterns of variation across
the Guyanese speech community.3 Much subsequent research has focused
on problems with Bickerton’s description and analysis of the tense-aspect
markers themselves. We now know that because Bickerton (1975) did not
begin with an entirely accurate description of the grammatical categories he
was investigating, his analysis of variation equated non-equivalent forms and
failed to take into account the range of possible contexts of occurrence.4 In
short, despite its importance for the growth and development of the field,
Bickerton’s (1975) description of variation in tense-aspect in GC fell short
of Labov’s (1982) principle of accountability.5 Jaganauth (1994) and Gibson
dent bibliographic project. A useful recent overview of the literature and some of the main
theoretical issues is to be found in Singler’s (1990) introduction to the volume which he edited
on the subject. More recent is Winford’s (1996b) column “Common Ground and Creole TMA,”
which also includes useful discussion of terminological issues. (See also Bickerton, 1975; Holm,
1989; Labov, 1990; Mufwene, 1984a.)
3 For discussion and critique of Bickerton’s approach to language variation see Sankoff
environments that are relevant to an analysis of a given variable. Labov (1982, p. 30) writes:
“for the section of speech being examined all occurrences of a given variant are noted, and
where it has been possible to define the variables as a closed set of variants, all non-occurrences
in the relevant environments.” Bickerton’s analysis did not account for non-occurrences (the
Ø form). More importantly, Bickerton did not include in his study a crucial environment for
the occurrence of aspectual markers — before statives. It is quite implausible to assume that
this omission reflects a gap in Bickerton’s corpus. My corpus of 55,000 words includes 57
examples of habitual statives. Assuming that variation is relatively uniform, Bickerton likely
had in the order of 200+ tokens. What happened to these examples, which are categorically
excluded in Bickerton’s model of Guyanese tense-aspect, is unknown.
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 153
The meanings subsumed under this one unified category of Imperfective are
alternately instantiated in a range of preverbal markers with more restricted
basic meanings: doz [+habitual], yuustu [+habitual, +past], Ø+V+ing [+pro-
gressive], and Ø. Examples of these alternatives follow:
(7) i doz kyerii meed gu wid am tuu. (Habitual, Non-Past)
‘He takes his maid with him when he goes (to America).’
(8) da taim abii-diiz doz go a skuul (Habitual, Past)
‘At that time, we used to go to school.’
(9) a foor a dem doz de tugeda (doz + LOCATIVE, Habitual)
‘Four of them are generally there together.’
(10) ii doz taioord (doz + PREDICATE ADJECTIVE, Habitual)
‘He is usually tired’
(11) evriidee yu doz ga waip am dong? (doz + MODAL, Habitual)
‘Do you have to wipe it off everyday?’
(12) somtaim ii doz bii jook (doz + BE, Habitual)
‘Sometimes it’s a joke.’
(13) yuu doz noo wa yu doz taak (doz + STATIVE, Habitual)
‘Do you understand what you say?’
(14) ee, huu lukin afta mi? (V+ing, Progressive)
Hey, Who is looking after me?
(15) evriibadii Ø taak neem mi na gu see mi Ø duu it hoom aloon.
(Ø, Habitual)
‘Everybody gossips, I’m not saying that I do it at home alone.’
It is the variation between these options which is investigated in the follow-
ing report. I argue that these latter markers of aspect (with the exception
of Ø which is not in fact a marker, see Sidnell, 2000b), have entered the
conservative creole (whose speakers primarily use preverbal a to convey all
Imperfective meanings) through borrowing. Their source is an intermediate
or so called “mesolectal” creole similar to, and almost certainly genetically
related to, the variety now spoken in Barbados (see Edwards, 1975, 1984;
Winford, 1997; and below). Contemporary patterns of variation reflect in part
156 JACK SIDNELL
6 The details of the contact between these two varieties is the subject of papers by Edwards
While Jaganauth provides texts which seem largely supportive of her argu-
ment, based on the data examined here several objections may be raised.
First, Jaganauth restricts attention to longer stretches of talk, extracted from
158 JACK SIDNELL
7 The problem here merges with that of relying strictly on interview narratives. In conver-
sation we can look at the recipient’s behavior to see the ways in which participants themselves
break discourse into thematic bundles and similarly the ways in which topical coherence is
produced and recognized as an ongoing accomplishment.
8 It may be true however that these forms mark cohesion∼discontinuity as Jaganauth
(1994) suggests for some speakers specifically those for whom doz occurs infrequently. This
is to say that the marking of discourse relations (including thematic cohesion but also prob-
ably including emphasis, foreground and background relations etc.) emerges as a particular
discourse-use of the markers for those speakers who use them in the proportions presupposed
by the model. However this is certainly not true for all speakers.
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 159
9 Members of the working group are agricultural laborers and their families. These indi-
viduals spend the vast majority of their time in the village. Their social networks are highly
focused and local. The “non-working group” includes people who live in the village who are not
160 JACK SIDNELL
Fig. 1
agricultural laborers, rather they are shopkeepers, small business owners and school teachers.
These occupations bring members of the non-working group into contact with a large number of
people from outside the village. All the speakers included in this group have friends, relatives
or business associates who live in Georgetown, Guyana’s capitol. As such, members of the
non-working group are in fairly regular contact with people who speak an urban, intermediate
or mesolectal variety of Guyanese creole.
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 161
Fig. 2
The full evidence, in this respect, cannot be reviewed here for reasons of
space. However, mention can be made of another study (Sidnell, 2000a)
which analyzed constraint rankings for a number of factor groups (e.g., gram-
matical person, discourse sequencing). Although some aspects of the system
were convergent across working and non-working groups, others showed
quite divergent weightings. For instance, the presence of a temporal adverb
shows a reversed effect for the two groups. For the working-group a temporal
adverb disfavors the use of doz, for the non-working group the adverb favors
162 JACK SIDNELL
the use of doz. This may indicate that for the working-group, doz is itself
interpreted as a temporal adverb and is less than completely grammatical-
ized into the verbal complex. Supporting evidence for such a conclusion is
found in the fact that among working-group speakers, doz is unaffected by
the morphophonemic processes described by Rickford (1974, 1980; see also
Satyanath, 1991). Thus for the working group, doz occurs as [doz] or [das]
but never as [oz] or [z] et cetera. Such morphophonemic reduction process
are often associated with later stages of grammaticalization (cliticization) and
their absence in the speech of the working group speakers suggests that this
process is less fully advanced for them as compared with the non-working
group.
For these reasons and others (summarized in Winford, 1997), it is rea-
sonable to speak in terms of relatively distinct varieties, intermediate and
conservative. While the referents of these terms are essentially the same as
those for mesolect and basilect, their use does not imply that contemporary
patterns of variation resulted from decreolization.
The data for this study consists of a total of 1001 environments for the
possible occurrence of Imperfective a.10 These were coded for subsequent
analysis with VARBRUL.11 The totals for each variant form are indicated
in Table 4.
The task of isolating variants raised a number of methodological issues.
In the first place a decision had to be made regarding phonologically null
10 The corpus used for this study was collected in a rural Indo-Guyanese village of about
700 inhabitants and from a nearby settlement about three miles inland.
11 VARBRUL (or Goldvarb) is a computer application used for analyzing the statistical
significance of co-occurring factors conditioning linguistic variation. See Rand and Sankoff
(1988).
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 163
contexts. After the preverbal negator na (17) and the relative pronoun wa
(18), Imperfective a is not distinguishable.
(17) wen yu Ø jringk yu na bada mii?
“Don’t you bother me when you are drinking?”
(18) di wan wa de a gyadin gu kos
“The one that is in the garden will curse.”
Since these contexts almost categorically result in a phonologically null re-
alization of imperfective a they were excluded from the analysis. As such
it is not possible to describe the effects of negation on the choice of aspec-
tual marker. Other issues were raised by the interpretation of zeros. While
in some cases it is clear that the unmarked verb carries habitual meaning, in
the absence of adverbial specification a few instances were less than com-
pletely transparent. For example, the following could be interpreted either as
referring to a past/perfective instance of talking or a habitual situation.
(19) i se shi dadii Ø taak nais wid am
“He said that her father talks/talked nicely to him.”
Cases such as this which were ambiguous were excluded from the quan-
titative analysis. There were also some cases of a which were ambiguous
between habitual and progressive meaning. These too were excluded from
quantitative analysis. A final consideration involves the use of gu. Both Rick-
ford (1987) and Jaganauth (1994) suggest that gu can carry habitual meaning
in an appropriate context.12 However, the corpus used for the current study
included no such uses of gu. One speaker whom I recorded does use gu to
convey habitual meaning as illustrated in (20).
(20) R: hou laang shi na ga waan neks wok shi na gu lef
“As long as doesn’t have another job, she won’t leave this one.”
S: we shi gon gu?
“Where will she go?”
R: eniiwee eniiwee shi get waan, yu noo?
“Anywhere, anywhere she gets another job, you know?
mii an aal na laik di nait wok
“I don’t like night work
transcribed by hand in the field by the author and a field assistant who is a native speaker of
Guyanese Creole and member of the community in which it was recorded. At present only a
small section has been prepared as a searchable electronic document.
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 165
Semantic Type
Imperfective markers typically have a range of discourse uses and sec-
ondary meanings (Bybee & Dahl, 1989; Dahl, 1985; Comrie, 1976). A basic
distinction has been made in the following between habitual and progressive
meaning. Bybee et al., (1994) note that the progressive “views an action as
ongoing at reference time. [. . . ] it applies typically to dynamic predicates
and not to stative ones. Thus the progressive is typically used for actions
that require a constant input of energy to be sustained” (see also Mufwene,
1984b). Habitual situations on the other hand “are customarily repeated on
different occasions” (Bybee et al., 1994, p. 127). Comrie (1976, pp. 27–28)
168 JACK SIDNELL
D: wich chrok a gu
“Which truck is going?”
S: abii wan
“Ours”
D: aiyu a kyeri abii
“Will you (pl) take us with you?”
These examples illustrate the secondary use of imperfective a to convey a
sense of future intention. Notice that in (31) the first speaker characterizes
the future situation using a and, in response, his interlocutor refers to the
same, as yet unrealized, action using prospective future a gu. In examples
(32) and (33) the fact that these uses of a convey a sense of future intention
is confirmed by their co-occurrence with time adverbials tudee “today” and
tuunait “tonight.”
Pastness
Imperfective a is neutral with regard to time-specification. Earlier ex-
amples have illustrated the use of this marker to characterize both past and
non-past situations (e.g., past in examples (3) and (5)). All tokens were coded
for a parameter of pastness to test whether this affected the choice of aspect
marker. It should be noted that since time specification is not always overtly
marked the analysis which follows does not test the effect of tense and aspect
marker co-occurrence.15
Clause Type
As noted above, Bickerton (1975, p. 31) argued that the “main stative
rule” precluded the possibility of a occurring in temporal and conditional
clauses. However Rickford (1987) includes two apparent exceptions to this
rule in his collection of texts. These are reproduced below.
(34) wen mi a kom bak a kichen, dis naiz meek ageen
“As I was coming back into the kitchen, I heard this noise again.”
(line 1127 in Rickford, 1987, p. 229)
15 On the issue of past marking see Patrick (1999); Rickford (1987); Sankoff (1990). I
thank Peter Patrick for helpful discussion on the problem of past-marking and its relation to
the habitual markers discussed in this report.
170 JACK SIDNELL
(35) wel, wen dem a tek ot brekfos, agin dis nais mek.
“Well, while they were putting out the breakfast, we heard this noise
again.” (line 1152 in Rickford, 1987, p. 230)
Rickford notes that in both examples a conveys progressive rather than ha-
bitual meaning, and thus they do not support Edwards (1983) claim that
habitual markers (including doz) can occur in temporal clauses. Rickford
(1987, p. 227) continues,
[. . . ], these wen clauses do not have the sense “whenever,” “on any occasion
which,” that Bickerton specifies as necessary for the deletion or nongenera-
tion of the aspect marker; instead, they are equivalent to “while,” describing
processes in progress when the event in the main clause takes place.
In order to test the effect of clause type on the choice (or non-generation) of
aspectual marker, all tokens were coded according to the most local clausal
context.
Four clause types were distinguished for the present study — main,
subordinate, temporal and conditional. The category of subordinate includes
both typically subordinate clauses (for example clausal complements to pred-
icates such as se “say,” taak “talk,” biiliiv “believe,” ondastan “understand”
etc.) and relative clauses.16 The following examples illustrate each of these
categories.
Main Clause
(36) shi a kom rong bak
“She’s coming round back.”
Subordinate (clausal complement)
(37) daadii i a kos se aiyu diiz a taak out shi.
“Daddy, she’s cussing saying that you (pl) are talking about her.”
Subordinate (relative clause)
(38) iir en ga bai a kot eer
“This place doesn’t have a boy who cuts hair.”
16 It should be noted that no distinction was made between truly subordinate clauses
(e.g., indirect speech) and those which exhibit no morphological indication of a grammatically
subordinate status (e.g., direct speech).
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 171
effect on the choice of variant. The future intention use of progressive fa-
vors the use of V+ing as opposed to a. This category of secondary uses,
future intention, is rather distinctive. While both imperfective and straight
progressive characterize events as unbounded and ongoing at reference time,
when used to convey a sense of future intention, “a pragmatic interpretation
brought on by implicatures in the discourse context” (Winford, 2000a, p. 7),
the sense of unboundedness is neutralized. Interestingly, although there is
significant structural and semantic mismatch between conservative and inter-
mediate grammars, both a and V+ing can be used in similar ways to convey
future intention. We might say then that these secondary uses exhibit a degree
of functional congruence between the two varieties.
We find a greater range of variation in the choice of habitual marker
as well as a more pronounced use of the non-a variants. The results are
summarized in Table 9.
Looking first at clause-type, we see that the effects for this constraint are
again largely categorical rather than variable. Thus, excepting past contexts,
conditional and temporal clauses categorically occur with Ø marked habitual
predicates (as suggested by Bickerton, 1975; and Rickford, 1987). Some vari-
ation is also found between main and subordinate clause types. However, a
stepping-up-stepping-down run with VARBRUL eliminated this factor group,
showing it to be statistically insignificant (apart from the categorical effects
exerted by conditional and temporal contexts).18
In contrast, the effect of predicate-type on the occurrence of variant
habitual markers is robust. Generally, stative predicates favor the use of doz.
The locative verb de is an exception.19 Given that stative predicates, with
the exception of de, appear to condition the selection of an aspectual marker
in essentially the same way they have been grouped together for the purpose
of analysis. Table 10 shows the distribution of habitual markers with stative
versus dynamic verbs.
locative expression. It may then be that (d)a retains some of its locative-semantics and is thus
favored in constructions expressing locative meaning.
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 175
Table 10. Distribution of habitual markers with stative and non-stative verbs
a doz Ø yuustu Total N %
Predicate-type
Dynamic N 326 98 101 11 536 90
% 61 18 19 2
Stative N 21 30 4 2 57 10
% 37 53 7 4
Total N 347 128 105 13 593
% 59 22 18 2
Fig. 3
Table 11. Distribution of doz and other habitual markers by predicate type
doz other (a, zero, yuustu) Total N %
Predicate-type
Dynamic N 98 438 536 90
% 18 82
Stative N 30 27 57 10
% 53 47
Total N 128 465 593
% 22 78
Table 12. Probabilities for the occurrence of doz with stative and non-stative predicates
Group Factor Weight App/Total Input & Weight
Predicate-type Dynamic 0.462 0.18 0.18
Stative 0.810 0.53 0.53
ers in the intermediate creole. There are a number of other differences be-
tween the two systems. Thus although both doz and a are essentially neutral
with regard to time specification for most speakers — all occur in both past
and non-past contexts — they differ in terms of combinatory possibilities.
Imperfective a combines freely with the relative past marker bin. Doz, on
the other hand, cannot combine with bin.
(45) *Parsad bin doz wok wid dem
(46) *Parsad doz bin wok wid dem
So unlike the imperfective marker a, doz cannot enter into productive com-
binations with other tense-aspect markers. Furthermore while many speakers
use doz to convey both past and non-past habituality, for some speakers doz
carries a feature [+non-past] and contrasts with juustu which carries the fea-
ture [+past]. This makes doz (and juustu) quite unlike imperfective a and
other tense-aspect markers in CGC which do not fuse aspect and tense into
a single morpheme (e.g., a = imperfective, don = completive, bin = relative
past).
These differences, along with differences in semantics, are indicative
of structural and categorial mismatch. This said, there are some areas of
congruence between the two systems. The combination of preverbal a with
178 JACK SIDNELL
Conclusion
But as Bickerton himself noted (1975) habitual doz apparently derives not
from Modern English do-support (i.e., the superstrate), but rather from the use
of periphrastic-do in expressions of the simple present — a usage which was
waning in 18th century England, even if it survived in some regional dialects
(e.g., the Southwestern ones, see Elworthy, 1877). Clearly doz in Caribbean
English Creole predates emancipation by at least a hundred years. So if we
accept the idea of decreolization giving rise to mesolects, we must admit
that the transition from a to doz (the latter playing a particularly important
role on the definition of “mesolect” — see Rickford, 1974, 1980) does not
fit the traditional model. In the midst of his analysis of the continuum as the
result of decreolization Bickerton (1975, p. 62) notes that, “The most likely
180 JACK SIDNELL
possibility is that doz entered Guyana through Barbadian immigrants. Doz has
long been established in Barbados (Collymore, 1965) and was presumably
spread from there to Trinidad, where its use is widespread (Solomon, 1966).”
Quite apart from these problems of source and chronology are ques-
tions raised by the phrase “in the direction of the approved variety.” Can
one in fact say that an aspectual system which grammaticalizes a category
Habitual is anymore English-like than one which grammaticalizes a cate-
gory Imperfective? Such an assumption seems to be based on the fact that
the Imperfective category is typologically associated with “radical” creoles
such as those of Suriname (for Sranan see Winford, 2000a; for Ndyuka see
Huttar & Huttar, 1994) whereas the habitual category is associated not only
with the intermediate creoles but also dialects of English such as “AAVE
(see Rickford, 1986) and Hiberno-English (see Harris, 1986). But despite
these typological associations, neither category is, in itself, closer than the
other to the Modern Standard English pattern of using simple present to
convey habituality. Such considerations suggest that decreolization does not
offer an adequate explanation for the variation between a and doz in Guyana
or elsewhere.
In a comparative study of tense-aspect in several Caribbean English
Creoles, Winford (in press, p. 13) notes:
The semantic space covered by the label ‘imperfective’ represents one of the
areas that are grammaticalized quite differently across Caribbean creoles. The
two primary notions subsumed under imperfectivity, ‘progressive’ and ‘habit-
ual,’ are subsumed under a single Imperfective category in Sranan and (appar-
ently) the other Surinamese creoles, as well as in conservative GC [. . . ] None
of the other Caribbean English Creoles surveyed here has an Imperfective cat-
egory. Rather Progressive and Past Habitual are grammaticalized as distinct
categories, while present habituality is conveyed by various means. In JC, for
instance, Progressive is conveyed by a, alternating with de in some dialects,
while in Belizean Progressive is conveyed by di.
Although the Caribbean English Creoles display a great many structural and
semantic similarities, as Winford discusses above, the semantics of “imper-
fective” aspect are grammaticalized quite differently in these languages. This
is all the more remarkable when compared with the uniformity across the
same languages in terms of other TMA categories such as Past (all use a rel-
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 181
ative past marker derived either from English “been” or “did”). Some sense
of the differences is illustrated in Table 13.20
Bajan differs in significant respects from the other varieties represented
in Table 13. Although it now seems that Cassidy was correct in positing the
existence of a now archaic (and in some areas extinct) conservative creole
in Barbados, (see Cassidy, 1980, 1986; Hancock, 1980; Fields, 1992; Rick-
ford, 1992; Rickford & Handler, 1994; Roy, 1986), recent accounts stress
the structural similarities between contemporary vernacular Bajan and the di-
alects of South West English which played a particularly important role in its
formation (Winford, 2000b). Relevant to the current discussion are the forms
and functions of doz and did for which periphrastic, unstressed, non-emphatic
do appears to be the etonym. Periphrastic do appeared in constructions con-
veying simple present in the South West English dialects and this form was
apparently reanalyzed — possibly under substrate influence — as a marker
of habitual aspect in Bajan.21
While the use of periphrastic do appears to have continued as a restricted
regional variant until the modern period, it has not been widespread since
the 17th or perhaps early 18th century (see Rickford, 1986; Winford, 2000b).
As Bickerton and others noted, this suggests that the intermediate varieties
formed early in the history of the Caribbean colonies (and are not the result
of decreolization see Bickerton, 1988; Winford, 1997).
In the case of Sranan, CGC, Jamaican and Belizean the marker of habit-
uality and/or progressive appears not to be a reflex of English do. A number
of creolists have suggested rather that these conservative aspectual markers
derive from expressions of location (Mufwene, 1986; Pochard & Devonish,
1986; Rickford, 1987; Sidnell, 2000b; Winford, 1999). As Bybee et al. (1994)
have noted there is a relatively strong cross-linguistic tendency for progres-
sive markers to develop from locative expressions (such appears to be the
case for English).22 It is therefore perhaps not surprising to find formal par-
allels in CGC and other creoles between constructions expressing duration
with those which convey location.
(47) Jan de a iit
John ASP ASP eat
“John is eating”
(48) Jan de a gyadin
John LOC PREP garden
“John is in the garden.”
In GC the de+a construction expressing “durative aspect” is well attested (see
Winford, 1993a; Rickford, 1987). It therefore seems reasonable to suggest that
Imperfective a is the fully grammaticalized and phonologically reduced reflex
of de+a. This suggestion is supported by a look through some earlier texts,
such as those of McTurk (circa 1895), in which we find that Imperfective a
consistently alternates with da (as it does in contemporary Kittitian Creole,
see Cooper, 1978). Similarly, in the Early Suriname texts collected by Arends
(1995) we find Imperfective conveyed by a marker formally equivalent to the
locative copula de. The eighteenth century use of locative de is illustrated in
Nepveu’s “Annotations” (1770, cited in Arends, 1995, p. 35).
(49) a de nami heddi
it LOC PREP-my head
‘I won’t forget.” (cited in Arends, 1995, p. 35)
The use of de as progressive (expressing future intention) is attested also in
Nepveu. Example (50) illustrates.
22 Thus forms such as “a-courting” derive historically from “at (on, in) courting.”
HABITUAL AND IMPERFECTIVE IN GUYANESE CREOLE 183
(50) adekom
“he/she’s coming’ (cited in Arends, 1995, p. 82)
Finally Nepveu includes examples of de used to convey the sense of habitu-
ality as illustrated in example (51).
(51) a de wakka langa him
“She/he walks with him/her.” (cited in Arends, 1995: 81)
Although modern Sranan distinguishes between locative copula de and Im-
perfective e, these 18th century examples suggest that they derive from a
common source.
Finally, the dialectal alternation between progressive a and de in Ja-
maican (mentioned by several authors, see Mufwene, 1986; and Pochard
& Devonish, 1986, for example) further supports the idea that these forms
developed out of a single locative expression. Such a path of grammatical-
ization may have been facilitated by the existence of similar, although not,
in all cases, identical, substrate patterns (see Winford, 1999).
The aspectual function of de would then have been extended to marking
imperfective meaning in general in some but not all creoles. Such internally
motivated development of progressive to imperfective is discussed by Bybee
and Dahl (1989, p. 82). They note that:
While the most restricted and more prototypical use of the progressive requires
that the activity actually be in progress at event time, it has often been pointed
out [. . . ] that the English Progressive is more general in its use, since it may
be used to describe activities that are not actually in progress at the reference
time, but that are characteristic of a certain time frame which includes the
reference time. [. . . ] The extended usage in English is also compatible with
temporal phases indicating repeated or habitual activities, [. . . ] Such usage,
then, represents a generalization of the earlier progressive meaning. Compara-
tive evidence suggests that in Igbo, Yoruba, Scots Gaelic [. . . ] and Kuwaa [. . . ]
a progressive construction (in all cases derived from a locative construction)
has evolved into a general imperfective. [. . . ]
References