All abstracts

in alphabetical order



Parth Bhatt & Emmanuel Nikiema, Toronto


Empty positions in Haitian Creole syllable structure


This paper argues that empty nuclei are an essential component of the phonological structure of Haitian Creole (henceforth HC). Evidence for the existence of empty nuclei in HC is provided by the behaviour of word-internal consonant clusters, and by the behaviour of the HC definite determiner. The analysis follows the framework of Government Phonology (Kaye, Lowenstamm and Vergnaud 1990 and Charette, 1991, among others).

The examples in (1) show that HC permits branching onsets in word-initial position (Cadely, 1988, 1994 and Anestin, 1987). The word-internal sequences in (2a) show similar consonant clusters as those in (1), and may also form branching onsets.


(1)   Word-initial sequences          

       flè        [fle]                        “flower”

       glas     [glas]                        “ice”

       tris       [tris]                         “sad”


(2)   Word-internal sequences

      a.      tranble [tRãble]      “to shake”       b. brasri    [bRasRi]         “brewery”

          mokri    [mOkRi] “mockery”          sovajri   [sovaZRi]    “savagery”

          potri      [pOtRi]   “pottery”            kozri      [kozRi]       “talk”


The examples in (2b), however, are problematic since they contain consonant clusters that are not permissible in word-initial position. One solution to this problem would be to propose that the examples in (2b) are in fact coda-onset sequences, thus explaining the absence of such sequences word-initial position. This hypothesis, however, leads to a violation of the Sonority Sequencing Principle (henceforth SSP, Selkirk, 1982), which states that the onset consonant should not have a higher degree of sonority than the preceding coda consonant. We propose therefore that the examples in (2b) in fact contain an empty nucleus separating the two consonants. Since the consonants are not adjacent in underlying representation, this accounts for the occurrence of the sequence despite the apparent violation of the SSP and the absence of these sequences in word-initial position.

Further evidence for the existence of empty nuclei in HC comes from the behaviour of the HC definite determiner. The HC definite determiner has five allomorphs, three consonant-initial “long” forms /la/, /na/ and /nã/ which combine with words that end in a consonant and two vowel-initial, empty onset “short” forms [a] and [ã] which combine with words that end in a vowel. To put this paradoxical distribution another way, words that end in a consonant combine with the consonant-initial allomorph to create a sequence of two consonants, while words that end in a vowel, combine with the vowel-initial allomorph to create a sequence of two vowels. The question thus arises as to why HC would create a sequence of two consonants when an empty onset allomorph is available and conversely create a sequence of two vowels when a consonant initial allomorph is available?

We propose that the underlying representation of the determiner /la/ consists of a bipositional syllable with an empty onset position (i.e. the initial consonant is floating). In forms such as [papaa], the empty onset position of the determiner is properly governed by the following vowel. In forms such as [Sat], which end in an empty nucleus, suffixation of /la/ creates a sequence of two empty positions that are subject to the ECP. The floating consonant is anchored in this context to avoid an ECP violation, producing the surface form [Satla] and not *[Sata]. In forms such as [papaa], we propose that the proper government relation is between the empty onset and the following nucleus is linearly adjacent, it holds at the segmental level. In [Satla], however, the government relation is between the two nuclei at the projection level.

Our analysis shows that proposing empty positions in underlying forms in HC provides a straightforward account for the behaviour of both non-transparent morpho-phonological processes and consonant clusters in word-internal position.



Anestin, Agnès (1987) Structure syllabique de l’haïtien et nasalisation, Mémoire de maîtrise, Université du Québec à Montréal.

Cadely, Jean-Robert (1988) “Représentations syllabiques et distribution des diphtongues en créole haïtien”, in Études créoles , Vol. XI (1):9-40.

Cadely, Jean-Robert (1994) Aspects de la phonologie du créole haïtien, Unpublished Ph. D. Dissertation, Université du Québec à Montréal.

Charette, Monik (1991) Conditions on Phonological Government, London: CUP.

Kaye, Jonathan, Jean Lowenstamm et Jean-Roger Vergnaud (1990) “Constituent Structure and Government in Phonology” Phonology  7:2, pp.193-231.

Selkirk, Elisabeth (1982) SyllablesIn H. van der Hulst and N. Smith (eds.) The structure of phonological representations, vol. 2, Dordrecht: Foris, pp.337-383.



Maria Braun, Siegen


How to create a word-formation system in two hundred years:

the case of Early Sranan


Recent work in the field of creole morphology (see Plag 2003) has helped us gain many insights into this terra incognito of creole studies. However, the question of  how creoles develop their word-formation systems and what influences this development remains almost unanswered in creole linguistics. Using the data from three early sources – Van Dyk’s Language Manual (c1765), Schumann’s Sranan Dictionary of 1783 and Focke’s Neger-Engelsch Woordenboek of 1855, the present paper investigates the development of word-formation patterns and devices of Early Sranan over the first 200 years of its existence with a special focus on the parallels between the initial stages of creolisation and untutored adult second language acquisition in the sphere of word-formation. 

First, it will be shown that already in the first one hundred years Sranan has developed a robust word-formation system with patterns and devices capable of expressing quite a variety of concepts. Second, it will be argued that the word-formation system of Early Sranan acquired greater complexity in the course of time: more word-formation patterns emerged, and the number of concatenated elements within a compound became greater. Applying Klein and Perdue’s (1997) concept of Basic Variety, I will demonstrate that this development may be argued to parallel the development of learner varieties in untutored second language acquisition: the creators of Early Sranan might have gone through similar stages while creating the word-formation system: the pre-basic variety, the basic variety and the post-basic variety.

Second, two hypotheses formulated by Broeder et. al. (1993) (published in Perdue 1993) for untutored second language acquisition as a result of a longitudinal study of the second language acquisition by learners of several European languages will be tested on the data from Early Sranan:


Hypothesis 1: the devices of composition precede the devices of derivation in untutored second language acquisition.

Hypothesis 2: learners make a creative and innovative use of word formation devices in semantic domains where complex concepts can be referred to by combining more elementary or basic concepts.


It will be shown that the first hypothesis is largely supported by our data. Thus, not even a single one of the English affixes survived in the process of creolisation of Sranan, and the derivational morphemes available developed from free morphemes. There is also a striking similarity between some word-formation devices used by the creators of Early Sranan and second language learners of Dutch, English and German in Broeder’s et. al. (1993) project: e.g. both use the same morpheme ‘man’ to create nouns denoting persons/agents. The second hypothesis also seems to hold for Early Sranan, since there is a great number of complex words in this creole language that can be regarded as innovations such that complex concepts are expressed by combining more basic ones.

One additional hypothesis will be advanced and discussed in the paper. I will argue that at the initial stages of creolisation, preference in word-formation will be given to nominal compounding, with N+N compounding as a dominant and most productive word-formation pattern. I will further show that the situation is similar in untutored adult second language acquisition where, according to Broeder et. al. (1993: 50), N+N compositions are the most frequent innovative word-formation device.



Ana Castro & Fernanda Pratas, Paris/Lisboa


Capeverdian DP-internal number agreement: additional arguments for a Distributed Morphology approach


1.      Introduction


The goal of this talk is to provide evidence from Capeverdian (CV) that confirms an analysis of DP-internal number agreement in terms of a Distributed Morphology approach.

In CV, when there is a determiner-like element in the DP, as 'a' and 'some', demonstratives and possessives, only these elements bear the plural marker -s.

(1)     uns/alguns/kes/nhas livru bunitu

         a- pl /some- pl /these/my-pl book beautiful


In determinerless DPs, the plural marker surfaces in the first element within the DP.


(2)     livrus bunitu

         book- pl beautiful

(3)     purmerus livru

first- pl book


Universal quantifier data show these two possibilities. If there is a determiner-like element (demonstrative or possessive), the plural marker -s surfaces in it.


(4)     tudu kes/nhas mininu

all this- pl/my- pl boy


If the universal quantifier is postnominal, the plural marker surfaces in the noun.


(5)     mininus tudu

boy all


Neverthless, the plural marker -s does not surface when the quantifier is initial.


(6)     tudu mininu

 all boy


2.      Assumptions


We will assume Costa & Figueiredo Silva (2003) analysis for Portuguese DP-internal number agreement, in the lines of the Distributed Morphology framework.

Their analysis accounts for the following differences between European Portuguese (EP) and Brazilian Portuguese (BP): in EP, there is full DP-internal agreement (7) while in BP number within the DP is marked either on the D head (definite and indefinite articles and demonstratives) or in all prenominal elements (8).


(7)     os primeiros filhos

the-pl first- pl son- pl

(8)     os primeiro(s) filho

the- pl first(-pl) son


Costa & Figueiredo Silva (2003) propose that the [plural] morpheme is realized in D, the head linking the DP to its LF interpretation - Enç (1991). The differences follow from the fact that in EP [plural] is a dissociated morpheme, while in BP it is realized as a singleton - Halle & Marantz (1993) and Embick & Noyer (2001). Therefore, in EP all elements able to bear plural morphology (determiners, adjectives, nouns) will carry the [plural] marker. In BP, only D, being the element anchoring number information, bears the [plural] marker.


3.      Analysis


We will show that Costa & Figueiredo Silva (2003) make the right predictions for a language as CV. Moreover, the facts from CV also give support to a framework that assumes Late Vocabulary Insertion.

We will propose that, as in BP, in CV, the plural marker is a singleton and it surfaces on the D head (as a suffix): if there is an overt element on D, it attaches to that element (1) and (4); if there is no overt element on D, it lowers to the next adjacent head (2-3) and (5). Lowering is a post-syntactic morphological operation that occurs under adjancency (see Bobalijk 1994 for English T to V+Asp). As for the example (6), we will show that this construction is not related with (5), so no [plural] marker should surface in it.

We will also raise the question (and try to give some clues for an answer) of whether this reduced agreement morphology within the DP has some relation to verbal agreement.



Stéphane Goyette, Ottawa


Haitian derivational morphology: borrowed or inherited?


A considerable lack of consensus exists as to creole genesis : whereas Hall (1966) saw Creole languages as being pidgins which had become the first language of a community, this theory has been losing ground : increasingly, Creoles are seen as the products of language contact which have never passed through a pidgin stage. Believers of the former view hold that pidgins, as languages which utterly lack bound morphemes, must either have created bound morphemes through grammaticization of originally free elements, or borrowed bound morphemes from some language spoken by the users of the pidgin. If this language was the lexifier of the pidgin, however, the facts are interpretable in a different fashion : theorists of creolization who deny the existence of a pidgin stage in the history of creole languages have argued that the bound morphemes found in creoles, far from having been borrowed, have in fact always been there.

This presentation will seek to contribute to solving this problem on the basis of an examination of Haitian Creole data. Believers of the theory holding that Creoles are nativized pidgins would hold these bound morphemes to have been borrowed at some point in the history of Haitian Creole (or its pidgin predecessor), whereas believers of the theory that creoles emerged through language contact would hold that these bound morphemes were always present in the transition from French to Haitian. In order to test which of these two theories is accurate, the behaviour of the affixes themselves will be examined : since some languages (including English) have borrowed French derivational affixes, a useful comparandum exists for the researcher : if the behavior of the affixes of Haitian is more reminescent of the behavior of French affixes borrowed in other languages than of French affixes in French itself, then the pidgin hypothesis would be reinforced. Preliminary results suggest that Haitian affixation is indeed closer to that of French affixes borrowed in other languages than to that of French affixes in French itself.



Magnus Huber, Regensburg


Phonological variation in Ghanaian Pidgin English

L1-influence vs. community norm


Although the number of mother tongue speakers of West African Pidgin English (WAPE) is quite low, WAPEs are morphologically and structurally stable and complex systems (as evidenced by the descriptions in e.g. Faraclas 1996 for Nigerian PE, Huber 1999 for Ghanaian PE, or Schneider 1966 for Cameroonian PE) and are thus best classified as Creoles rather than Pidgins. Yet, unlike the New World English Creoles, restructured varieties of English in West Africa continue to be in contact with African languages, since virtually all Pidgin English users have an African L1. One would therefore expect considerable L1 influence and variation in the area of WAPE phonology.

Interestingly, WAPEs deviate from the alleged prototypical Creole CV syllable structure by exhibiting rather complex consonant clusters, such as CCC onsets or even CCCVCC syllables (cf. NigPE styu, strayk; Faraclas 1996:264). However, one and the same speaker may use such complex forms alongside reduced ones, regularizing syllabic structure towards CV through paragoge, as well as through elision and epenthesis in consonant clusters.

Concentrating on Ghanaian PE, this paper addresses the question whether the phonological processes that operate on etymologically English forms – such as regularization towards CV, but also other phenomena potentially attributable to the influence of the speakers' L1, such as l ~ r alternation or n-dropping in coda with compensatory nasalization of the nucleus – are entirely due to L1-interference or if at least some of them represent a community norm, adopted even by those speakers whose African language would in fact allow forms closer to the English input or require to treat English etyma differently. To answer if such linguistic focussing (Le Page 1968) occurs in the Ghanaian PE speech community, the phonological processes observable in a spoken corpus will be compared to the phonology of speakers' respective L1.

The establishment of such community norms is important from a theoretical point of view, demonstrating that in phonology, just as in the areas of morphology and syntax, WAPEs are developing towards autonomous and stable systems even though they have almost no L1 speakers.



Faraclas, Nicholas G. 1996. Nigerian Pidgin. London, New York: Routledge.

Huber, Magnus. 1999. Ghanaian Pidgin English in its West African context. A sociohistorical and structural analysis. Amsterdam: Benjamins.

Le Page, Robert B. 1968. "Problems of description in multilingual communities". Transactions of the Philological Society: 189-212.

Schneider, Gilbert Donald. 1966. West African Pidgin-English. A Descriptive and Linguistic Analysis with Texts and Glossary from the Cameroon Area. Athens, Ohio: Hartford Seminary Foundation.



Valeri Khabirov, Ural U.


Morphological changes in creolized Sango


The presentation deals with internal linguistic processes that led to certain types of changes in the words of a creolized language vis-a-vis its source language. These processes affect the lexicon of all languages, but the extent to which they are evidenced in creolized languages suggests that they are accelerated by restructuring. The process of creolization in the sango of Central Africa brought about the reduction of the lexicon (down to 300 words) and as a result  the reduction of the derivational morphemes existing in the source language - the ethnic sango-yakoma. The latter has in its lexicon verbs with general meaning and words with specialized meaning (process akin to semantic narrowing) produced by derivational morphemes which was lost during creolization. The following derivational suffixes can be found in the ethnic sango-yakoma: -rV (iterative meaning), -ngbì (meaning of insistence), -ngà (resultant action), -kà/-kè/-kò (iterative meaning), -sà (meaning of removal), -ndà (resultant action), -kò (resultant action), -ndò (meaning of accumulation) and some other suffixes whose meaning is not clear because of the limited number of glosses.  Though some of these suffixes can still be found in the creolized language, the verbs with such suffixes have the same meaning as the verbs devoid of them, for example: Wálī à yéngè (yèngèrè) fùkù  The woman has sifted flour. In the sango-yakoma the verb with the iterative suffix -rè will mean “ to sift with quick and often movements”. Words with the above mentioned suffixes of the ethnic language have become vestiges of the lexicon’s development. The forgotten words come to life again and the process seems to be akin to decreolization. The ethnic sango-yakoma monomorphemic “to make” with the iterative and process intensifying suffix –kε`  entered the lexicon of the creolized sango in the form of lέkε` losing its special meaning and having only the general meaning of “to make, to build, to arrange, to repair, to organize ”. At a later stage the derivative form lε`kε`rε` appeared with the meaning of “to repair several times and applying several operations”. Other suffixes from the above mentioned form new words whose meanings in creolized sango sometimes are not specialized but general, for example: “collect” – bóngbì  (var.búngbì) “to unite, get  together, join”; “to name” – díkò “to count, enumerate, verify, number, read, pray, implore”. An interesting case presents the so-called subject marker when it is agglutinated to the verb. The following conditional sentence “If he wants to come he will see him” is given by our informant the following way: 


Tōngānà                              áyèkè          báà    

                     COND   3s   want  to    come  3s     SM+FUT  see     3s


The subject marker á with the grammatically pertinent high tone is at the same time the agglutinated marker of the future tense. In creolized sango the subject marker has always the low tone and is not used after pronouns. In the absence of  the future tense marker á the future tense is expressed just by the predicative element yèkè in the creolized sango.  

     This analysis will provide further information on the changes in the creolized sango as compared with its lexifier. 



Alain Kihm, Paris


The phonological origin of language: Creole languages as a testing ground


Carstairs-McCarthy (1999) supports the hypothesis that human language results from an evolutionary accident that occurred perhaps as early as 200 000 years ago, viz. the descent of the larynx which first allowed our ancestors to distinguish between consonant-like and vowel-like sounds. Syllabic organization, i.e. the hierarchical contrast of a nucleus and a margin (the onset and the coda), followed. Syntax “as-it-is” – it could have been otherwise – rests on the basic distinction between noun phrases and sentences, according to the same author. Syntax is thus a reflection of and an evolution from syllabic structure, inasmuch as NPs correspond to margins and sentences are projections of the nuclear verb.

For his demonstration, Carstairs-McCarthy employs a standard, double-branching (CVC) model of the syllable including codas. Other authors (see, e.g., Lowenstamm 1996) convincingly argued, however, that syllables should be viewed as universally single-branching, i.e. CV, in their underlying representation, which might pose a problem to the “syllabic syntax” hypothesis. Yet, it can be shown that even a CV syllable preserves the necessary asymmetry, as every V segment that occupies the nucleus may also occupy the onset where it is realized as a glide, whereas C segments that occupy the onset cannot appear in the nucleus (i.e., [O/C I [N/V A]] is a possible syllable realised /ja/, and so is [O/C A [N/V I]], realised, e.g., //i/, but [O/C A [N/V k]] is impossible).

On the other hand, the CV model implies that the real basic contrast is between V elements (V = vowel or verb) and C/NP elements, and it makes the following prediction: V elements are able to occur where C/NP elements occur, but the reverse is not true. Syntactically, this means that nominalizations and sentences in argument positions are allowed, but NPs not in the dependence from a V element cannot be used as predicates.

The first part of the prediction is trivially borne out by language facts, while the second part seems to be equally trivially falsified by such common data as Russian Ivan – dobryj student ‘Ivan [is] a good student’. Two considerations have to be taken into account, however. First, such data may not be genuine counterexamples (cf. Ivan byl dobrym studentom ‘Ivan was a good student’). Secondly, “old” languages such as Russian may well not instantiate perfectly basic (bioprogrammatic) syntax as it emerged some 100 000 years ago, because of various historical accidents (compare the other Slavic languages, including Old Russian, which all show an overt copula in the present tense; also see Cohen 1984, Chapter 1).

Actually, Creole languages ought to constitute the best testing ground, under the assumption that, among present-day languages, they most faithfully realise bioprogrammatic syntax (see Bickerton 1998). Conversely, ascertaining the fact that Creole languages do bear out the prediction of “CV syntax” in a possibly more uniform way than other languages do would count as good support for the language bioprogram hypothesis, in its strict or relativized version (see Kihm 2000). It is very striking, in particular, that nearly all the Creoles use an overt copula in sentences like the Russian example above, especially since absence of a copula is what we expect as a result of the simplification and pidginization processes out of which they emerged, at least in part (see Ferguson 1971). The observation makes sense, however, if Creole languages represent the passage from protolanguage to real language in its more or less pristine form.

The whole field of Creole languages cannot of course be covered in a short talk. As a first sample I will use Upper African Portuguese Creoles, i.e. Guinea-Bissau Kriyol and Cape Verdean (see Kihm 1994; Baptista 2002). Both confirm the V-as-NP part of the prediction, as expected, but Kriyol is remarkable in the extent to which it uses morphological nominalizations and non-finite sentences as subject arguments. As for the no-NP-as-predicate part of the prediction, on the other hand, the comparison of Kriyol with Cape Verdean is especially revealing given the range of phenomena it shows. In Cape Verdean even adjective predicates require an overt copula e or sta (Baptista 2002: 101). In Kriyol, in contrast, more than one stage in the evolution of the language are still apparent. Some adjectives behave like verbs, other behave like nouns. Sentences like Jon i karpinteru ‘John is a carpenter’ are susceptible of three construals according to whether i is analysed as (i) the 3sg subject pronoun resuming the (obligatorily) dislocated subject DP Jon; (ii) a predicate marker; (iii) a copula similar and related to Cape Verdean e (cf. Portuguese é ‘is’). Under (i), and perhaps (ii), past tense is expressed by a postposed adverbial (Jon i karpinteru ba ‘John was a carpenter’); under (iii), it is expressed by an inflected form (Jon yera karpinteru ‘John was a carpenter’). My contention is that all three construals are legitimate and reflect the “naturalization” of the language as it evolved away from the original pidgin. (A comparison with 16th century língua de preto texts will also be made).


Baptista, Marlyse. 2002. The Syntax of Cape Verdean Creole: The Sotavento Varieties. Amsterdam: John Benjamins.

Bickerton, Derek. 1998. “Catastrophic evolution: the case for a single step from protolanguage to full human language”. In J.R. Hurford et al. (eds), Approaches to the Evolution of Language, 341-358. Cambridge: Cambridge University Press.

Carstairs-McCarthy, Andrew. 1999. The Origins of Complex Language: An Inquiry into the Evolutionary Beginnings of Sentences, Syllables, and Truth. Oxford: Oxford University Press.

Cohen, David. 1984. La phrase nominale et l’évolution du système verbal en sémitique. Paris: Peeters.

Ferguson, Charles. 1971. “Absence of copula and the notion of simplicity: a study of normal speech, baby talk, foreigner talk and pidgins”. In D. Hymes (ed), Pidginization and Creolization of Languages, 141-150. Cambridge: Cambridge University Press.

Kihm, Alain. 1994. Kriyol Syntax: The Portuguese-Based Creole Language of Guinea-Bissau. Amsterdam: John Benjamins.

– 2000. “Are Creole languages ‘perfect’ languages?”. In J. McWhorter (ed), Language Change and Language Contact in Pidgins and Creoles, 163-199. Amsterdam: John Benjamins.

Lowenstamm, Jean. 1996. CV as the only syllable type. In J. Durand & B. Laks (eds), Current Trends in Phonology : Models and Methods, 419-442. Oxford : Oxford University Press.



Thomas B. Klein, Statesboro, GA


Creole phonology typology: Vowels and consonants


Synchronic work on the typology of creole phonologies is exceedingly rare. Furthermore, works that invoke phonological generalizations across creole languages exhibit one or more of the following limitations: (a) the generalizations are not from a language sample based on typological principles; (b) the geographical focus is either on Atlantic or Pacific creoles; (c) only creole languages with certain lexifiers, typically Indo-European ones, are included. The present work aims to provide a broader base for the comparative and typological study of creole phonologies. It ameliorates earlier methodologies by employing a typological quota sample (cf. Maddieson 1984; UPSID database) drawn from the entire range of creole languages around the globe with Indo-European and non-Indo-European lexifiers. In this way, a more comprehensive understanding of the phonology of creole languages may be achieved. Specifically, the present paper presents typological results from the segmental inventories of c. 20 creole languages and compares them to the UPSID languages, none of which are creoles. Particular attention is being paid to McWhorter’s (2001) hypothesis that the world’s simplest grammars are creole grammars.

The size of phoneme inventories in non-creole languages shows a great range, from 11 to 141 phonemes. However, the typical size (70%) is between 20 and 37 segments (Maddieson 1984). Creole languages show a much narrower range: A significantly higher majority of them (> 90%; p < 0.03) shows 20 to 37 phonemes. Concerning vowel quality inventories, the great majority of creole languages (> 80%) shows the 5 or 7 qualities also typical of non-creole languages (46%). Thus, significantly more creole than non-creole inventories fall within the typical range. Note also that no creole language shows the simplest non-creole inventory, that is, /i, u, a/. These results are corroborated in the consonant inventories. For example, creole languages prefer the ordinary two series of stops, plain voiceless and plain voiced (> 75%).  However, creole languages exhibit these stops series more than the UPSID languages.

The results from this study are surprising to McWhorter’s simplicity hypothesis because creole inventories occupy the range considered typical, not simple, for non-creole languages. But it seems that they may not be considered typologically complex either so that creole inventories emerge as the most typical ones of any language group.



Pamela MacDonald, Bangor, GB


Verb morphology in Indian Ocean Creoles

It is a fairly well-known fact about the Indian Ocean Creoles (Reunionnais aside) that a large class of verbs shows a morphological alternation between final zero and a suffix <e> (sometimes orthographically <en>), often referred to as 'short' and 'long' forms. This phenomenon is usually described in phonological terms as a rule of final vowel truncation. The contexts in which the truncation rule applies are traditionally expressed in rather diverse terms as a range of (mainly) syntactic environments, cf. Bollee (1981), Papen (1975b); but as a broad generalisation it may be said that (i) intransitive verbs do not normally truncate; (ii) truncation normally occurs in the presence of a complement phrase (eg. whether an object NP, PP or VP), but not a complement clause (CP), as illustrated by these examples from Seselwa :

(1)             Ou konn lavi (Short form konn)

'You know life'

(2)             Mon pa konnen kote i ale (Long form konnen)

'I don't know where he goes'

A further complexity is that the short (transitive) V-form does sometimes appear in intransitive contexts : it is then associated with a generic or property-type reading, while the long form correlates with a temporally anchored interpretation :

(3)             Lam van da labutik, pa da lafarmasi(Short form van)

'Razor blades sell in shops, not in the chemist's'

(4)             Lam vade da labutik (Long form vade)

'Razor blades are selling in the shops (right now)'

- Mauritian creole examples from Corne (1981).

Corne (1981) relates the truncation rule to Aktionsart and claims that truncation occurs whenever the Subject has the role of Agent. This proposal however runs into difficulties in the context of stative predicates such as kout/e 'cost' and pez/e 'weigh', which truncate. The question therefore remains why a creole grammar which otherwise lacks inflection should make this one morphological distinction.

In this paper I pursue Corne's proposal in spirit but along rather different lines. Recent work on the individual- vs stage-level predicate contrast, cf. Carlson (1977a) and Kratzer (1995), enables us to argue that the 'long' V-form is the morphological reflex of the spatio-temporal argument (the Davidsonian argument) which is syntactically present in stage-level predications. In transitive contexts in IOC this contrast is neutralised in much the same way as the contrast between habitual and particular sentences in English.


Susanne Michelis, Leipzig


Intonation and clause coordination in Seychelles Creole


The functions of intonation in multi-clause constructions have been studied very little. However, it seems that for spoken discourse intonation is one of the major devices for building text coherence. In this talk I will examine two intonational patterns in Seychelles Creole: (i) comma-intonation, (ii) integrative intonation. These two patterns are found in two different coordination constructions, which I call "individuating coordination" and "integrative coordination". The conjoined units are in both cases clause-like, and in both constructions we do not find any overt coordination marker.

In individuating coordination, after each clause we get a pause with a special non-final pitch contour. In Seychelles Creole every 3sg verb and some 3pl verbs need a verbal particle i  ("vp") (which is homophonous with the 3sg subject pronoun) if the subject is a full noun phrase or a proper name. As can be seen from example (1), this verbal particle is not repeated with the other verbs, so we do not get *i danse, i kriye, i sote. The semantic relationship between these syntactic entities is an AND-relation, a kind of enumeration where the single events are presented individually.


Individuating coordination (with comma-intonation)

(1)   bann  danm      i      reponn   danse   kriye    sote

        pl     woman  vp  answer dance shout  jump

        'The women answered, danced, shouted and jumped.'


In integrative coordination (example (2)), we find a special unifying intonation contour that extends over all constituent clauses. There is no pause between the linked elements. In this construction, all verbal information has to be repeated with every single verb. Thus, at the syntactic level we have fairly independent clauses, but intonation integrates these units into a hyper-construction at the discourse level. 


Integrative coordination (with integrative intonation)

(2) per     osi   i     pase     i       gete     i       ogarde i       dir   tinge  i     zoli

      priest       also         vp pass. by     3sg   look 3sg watch 3sg say tinge vp nice

      'The priest also came along, he looked (at the dancers), he watched    (them),       he said: tinge (a kind of dance) is nice.'


At the semantic level, these combined clauses typically describe successive subparts of a single overall event. Each clause sets the stage for the following one: The priest first has to come, before he can watch the dancers and again, before he can express his opinion about the dance. This is in clear constrast with the individuating construction, where no linear order needs to be respected and the single events could be easily reversed.

Several questions now arise: Are these intonation constructions part of the grammar or just pragmatically conditioned discourse phenomena? Do these constructions, especially the integrative coordination construction, reflect substrate influence, admitting that intonational patterns seem to be very likely to be kept in the new creole language? I will attempt to answer these questions in my talk.



Emmanuel Nikiema & Parth Batt, Toronto


R diphthongs in French Lexifier Creoles


A cross linguistic examination of French Lexifier Creoles (FLC) shows that /R/ is always maintained in prevocalic position (1). Post-vocalic Rs have four possibilities: they may be maintained as in (2a) for Guyana Creole, reduced with a slight compensatory lengthening as shown in (2b) for Reunion Creole, assimilated to a vowel as in (2c) for Mauritius Creole which exhibits surface long vowels or completely deleted as in (2d) for Haitian and St. Lucian Creoles (Bernabé 1983, Carrington 1984, Chaudenson 1974, Hazaël-Massieux 1972, Valdman 1978, Cadely 1994, among others). In other words, the Onset position of prevocalic Rs is structurally stable (though the segment may weaken or assimilate), while post-vocalic Rs are unstable and subject to variability between languages (deletion and restructuring).


 (1) Prevocalic Rs are structurally stable (in all FLCs),

ravin          “creek”               bwa  “arm”                 mari “husband”

rev    “dream”             travaj         “work”               teras “terrace”

wut   “road”                grès  “grease”             fatra “garbage”


 (2) Post-vocalic Rs may be pronounced, reduced, assimilated or deleted

a. Guyana         b. Reunion                    c. Mauritius                    d. Haiti, St-Lucia

laport  “door”  [ma:r ]  “pond”             [ma:]  “pond”          “strong”     

artis  “artist”    [su: r s] “little creek”    [su: s]“little creek”          bab  “beard”

kuler “colour”  [fre: r ]  “brother”          [fre: ] “brother”     tòti  “turtle”


In order to account for this asymmetry, we suggest that the stability of Rs in Onset position is due to the fact that they are licensed by a vowel. The data in (3) show that when there is no vowel to license prevocalic R (as in word-final position for example), it is deleted. This explains the alternating patterns (deletion and reappearance) observed in morphologically related forms. The final R consonant in the underlying representation of milat is not pronounced because it is not licensed by a vowel, but reappears as soon as there is a following vowel. In other words, /R/ needs to be licensed by a (following) vowel when in Onset position.


(3) /R/ deletes after consonants (word finally)              

milat          milatrès    “mulattro”

labit           labitraz      “referee”                              

pov            apovri        “poor”               


(4)  a.          N               b.   N                            c.  x

 |                    / \                    / \

 x                    x    x                V   N

                  / \                  |    |

                 V    R               V   R


We also suggest that the instability of post-vocalic Rs is accounted for by the fact that they are in dependent position within the nucleus. More precisely, we contend that post-vocalic Rs when realized on surface are syllabified with the preceding vowel either as a light diphthong (non branching nucleus), or as a heavy diphthong (branching nucleus). The light diphthong structure suggested in (4a) is similar to the two-root node segmental structure proposed in Paradis & Prunet (2000) for nasal vowels (4c). If nasal vowels in FLC can have the representation in (4c) as suggested in Bhatt & Nikiema (2000), then our analysis predicts that R-diphthongs and nasality will be mutually exclusive because both R and the nasal element (N) occupy the same structural position. The light diphthong in (4a) accounts for the absence of regressive nasalization when the preceding nucleus contains an underlying diphthong as in the forms in (5). The heavy diphthong in (4b) accounts for the compensatory lengthening observed in Reunion and Mauritian Creoles.


(5) Blocking of (automatic) regressive nasalization (Hazaël-Massieux, 1972; Cadely 1994)

kòn          [k]n]    (*[k]n])       “horn”       as opposed to [zãmi]   “friend”

cham        [am] (*[ãm])     “charm”    as opposed to [ãm]    “room”

lame        [lame] (*[lãme])    “army”      as opposed to [kabãn] “bed”


We submit that /R/ is phonetically realized when linked to its own temporal position as in the branching nucleus structure in (4b).  As is the case in nasal diphthongs, the consonantal part of the segment is not realized when R is linked to the vocalic position. This analysis allows us to maintain as a true generalization that in all FLCs, word-final consonant clusters are prohibited. In other words, the final two segments in forms such laport “door” and form “shape” in Guyana Creole are not in consonantal positions: the first segment (/R/) is within the nucleus and the last consonant is the onset of an empty-headed syllable. Forms with liquids such as film and calm can be analyzed along the same line. The proposed diphthong structure accounts for why some vowels never undergo regressive nasalization and why word-final clusters are prohibited in all FLCs except when post-vocalic /R/ and /l/ are involved.



Ingo Plag & Mareile Schramm, Siegen


The emergence of creole syllable structure: a cross-linguistic survey


Apart from sweeping claims about the simplicity of syllable structure in creole languages, work on the creole syllable has been scarce and little attention has been devoted to the detailed description of the observable structures or to the question of which principles govern the development of syllable structure in creolization.

Based on the in-depth analysis of only one language, Sranan, Alber & Plag (2001) have recently argued that universal preference laws, transfer from the substrate languages, and superstratal influence are important in the creation of the creole, but each of them in a different and very specific way. The superstrate provides the segmental material which the emerging creole tries to preserve faithfully, but universal preference laws disturb faithful copying of the superstrate system. This is possible because the substrate exerts its influence imposing a particular grammar - high ranked structural constraints and low ranked faithfulness constraints - on the creole.

The present paper extends this research program to other creoles, i.e. Saramaccan, St Kitts and Jamaican Creole, in order to see how the observable cross-linguistic variation can be accounted for. Using the earliest available records of the respective languages, we will describe the syllable structure of each variety. It will be shown that each language has developed a constraint ranking of its own, with sometimes only small differences in the ranking creating significant structural differences between these varieties. We will argue that the differences in the constraint ranking can be attributed to the different substrate languages involved as a well as to the different socio-historical conditions pertaining at the time of contact.


Alber, Birgit, and Ingo Plag (2001): ‘Epenthesis, deletion and the emergence of the optimal syllable in creole’, Lingua 111, 811-840.



Paula Prescod, Paris


Stress assignment and functions of pitch in Vincentian Creole


This paper proposes an examination of how Vincentian Creole (VinC) uses suprasegmanetal features to convey meaning. Previous research on English-lexicon Creoles has underlined the importance of the notion of pitch polarity in understanding Creole lexicon (cf. Carter 1982 on Jamaican Creole). Devonish, (1989) accounts for the basic vocative and naming funtion of the second underlying high tone (UHT) as well as for the tonal nature of Guyanese Creole. Neither of these scholars have extended their research beyond the level of lexicon. This study goes beyond lexicon to give an instrumental analysis of connected utterances in VinC. We have observed that in the ten (10) pairs of phonetically identical utterances recorded, morphology and semantics are conditioned by fundamental frequency, intensity and duration. This accounts firstly for the demarcating role played by these three acoustic features but more so for their role in differentiating lexical innovations from grammatical morphemes and morphologically bound items from syntactically bound morphemes.



Marina Pucciarelli, Loreto

The allomorphs of Nigerian Pidgin third person singular subject pronoun


Sometimes Nigerian Pidgin English (henceforth NPE) shows allomorphic variation in its pronoun system depending on the variety it is taken into account. As for the third person singular subject pronoun, NPE studies do not usually provide with any explanation for its allomorphy except for Faraclas 1996 (pp. 98, 179) and Elugbe/Omamor 1991 (pp. 90, 95). According to the first one the alternation of ì and ìm is due to syntax, namely “ì tends to occur instead of ìm in relative clauses and (more rarely) in noun clauses” (p. 98). According to Elugbe/Omamor 1991, the occurrence of i and in depends on semantic-syntactical constraints: when the subject of the superordinate clause and the one of the subordinate clause are coreferential, “the main clause subject is i while the subject of the subordinate is in” (p. 90). In the language variety described by Faraclas 1996, the most frequent shape of the third person singular subject pronoun is “ì”, whereas in the one described by Elugbe/Omamor 1991 it is “in”. Four plays written almost entirely in NPE* have been examined in this paper on the basis of Elugbe/Omamor 1991’s rule because their NPE varieties conform to Elugbe/Omamor 1991’s description. The outcome is that a further syntactical constraint, which is not categorical, can be outlined. In fact, the occurrence of the literary allographs him and im instead of e/’e/he tends to be realised in the subordinate clause of the indirect reported speech in compliance with the coreferentiality constraint, whereas the literary allographs e,’e and he usually occur in all the other syntactical contexts. Thus, in the case of these four plays e/’e/he is the third person singular subject pronoun morpheme and him/im is its semantic-syntactically conditioned variant.



Elugbe, B.O / Omamor, A.P. 1991. Nigerian Pidgin: Background and Prospects. Ibadan: Heinemann Educational Books, Nigeria.

Faraclas, Nicholas. 1996. Nigerian Pidgin. London: Routledge.



Yolanda Rivera Castillo & Nicholas Faraclas, Puerto Rico


The emergence of systems of lexical and grammatical tone and

stress in Caribbean and West African Creoles


For thousands of years, there has been a dynamic and complex interaction between languages with very different pitch-related suprasegmental (i.e. tone, pitch-accent, stress) systems along the west coast of Africa.  Since the beginning of the European colonial expansion, the typologically different suprasegmental systems of Germanic and Romance languages have been brought into this rich linguistic mix, with predictably complex results in languages such as Nigerian Pidgin, where tone and stress (as shown at the 1rst Workshop on the Phonology and Morphology of Creole Languages) interact in more complicated ways that in any of its substrate or superstrate languages. This complexity is quantitative rather than qualitative, however, with tone and stress phenomena easily handled within the framework used by Africanists to account for lexical and grammatical/ morphological pitch related suprasegmentals in Niger-Congo languages.

In this paper, we show that the similarly complex contact situation that developed between and among speakers of Niger-Congo and Indo-European languages in the Caribbean during the era of colonial plantation slavery gave rise to Creoles with similarly complex pitch-related suprasegmental systems. Furthermore, we demonstrate that the interaction of tone and stress in languages such as Papiamentu, Ndyuka, and Saramaccan is best accounted for by utilizing the categories, principles, and parameters appropriate to the analysis of lexical and grammatical pitch-related suprasegmentals in Niger-Congo languages, rather than those normally used in the analysis of suprasegmentals universally or in the languages of South East Asia or Mesoamerica. Many of the complex interactions between stress and tone in African and Caribbean Creoles result from a reinterpretation of features from the superstrates in terms of those found in the pitch-related suprasegmental systems of the substrates. In Papiamentu, stress from lexifier languages has a fixed position in the word, just as it does in the suprasegmental systems of some Niger-Congo languages. On the other hand, a high tone in penultimate position attracts stress, in which case an Ijo-like pitch-accent pattern (including high tone spread over following ‘toneless’ syllables) incorporating features of both stress and tone results. For example, in the derivational morphology of Papiamentu, stress shifts its position to the penultimate syllable when a H tone affix follows a verb: ‘yuda + ábo --> yu’dábu (Römer, 1980:121; Agard,

1985: 239).

Finally, this paper challenges myths about simplification since Creoles exhibit intricate connections between their tone and stress systems and the constraints associated to these.



Agard, F. B. (1985). Papiamentu Grammars and the Structure of the Language. – In: V. Z. Acson, & R. L. Leed (Eds.), Oceanic Linguistics Special Publication No. 20 (pp.235-242). Honolulu: University of Hawaii P.

Römer, R. G. (1980) . Proclisis y enclisis en una lengua tonal. Diálogos hispánicos de Amsterdam 1, 113-123.


Nicole Rosen, Toronto


Stress assignment in Michif


The goal of this paper is twofold:  first, to present the first analysis of the stress system in Michif, a Cree-French contact language of the Midwestern United States and Canada; and second, to show that in this intense language contact situation, synchronic phonological structure has retained elements of both source languages, but displays a regularity such that no recourse to historical vocabulary source is necessary. 

         Previous work on Michif has claimed that there are two distinct phonological systems along the lines of the source languages: one for the historically French vocabulary, and another for the historically Cree vocabulary.  (Bakker 1997, Bakker & Papen 1997, Rhodes 1986) This paper examines a phonological process, rather than static inventorial facts in Michif to show that there is regularity in the system which does not rely on language source.

         In Michif words of 3 or fewer syllables, primary stress falls on the final syllable.  In words of 4 syllables or more, primary stress falls on the antepenult. Crucially, this system treats all vocabulary items identically, whether they be of French or Cree origin (see data in (1)). We further compare Michif stress assignment with that of French and Cree.  We find that Michif structure has parallels in both source language systems. In disyllabic words, primary stress is identical in all cases, in trisyllabic words, Michif stress is identical to French stress, and in words longer than three syllables, Michif stress is identical to Cree stress.  This paper formalizes these systems in terms of Hayes’ parameter settings for stress assignment (1996).  This comparison is shown in (1), where structural differences are marked in bold. 


(1)               2 syllable words            3 syllable words    4 syllable words           Parameters (simplified)

French                 Wd                                      Wd                                   Wd                   Foot Level:

                                  g                                         38                                    38                   -   binary feet, RH, QI, R-to-L,

                                 Fs                                    Fw      Fs                              Fw     Fs                   Iterative, no extrametricality

                                38                                   g      38                           38     38            Word level:

                            W      S                                 S   W   S                       W   S   W   S          -   RH, QI, R-to-L, Iterative,        

                            (x       x!)                              (xÙ)  (x   xÚ)                      (x   xÙ)   (x   xÚ)            no extrametricality.



Cree                      Wd                                      Wd                                   Wd                   Foot Level:

                                  g                                         38                                    38                   -   binary feet, RH, QI, R-to-L,

                                 Fs                                    Fw      Fs                              Fw       Fs                 Iterative, no extrametricality

                                38                                   g      38                           38     38            Word level:

                            W      S                                 S   W   S                       W   S   W   S          - RH, QI, R-to-L, Iterative, R-    

                            (x       x!)                              (xÚ)  (x   xÙ)                      (x   xÚ)  (x   xÙ)             most foot is extrametrical


Michif                  Wd                                      Wd                                   Wd                   Foot Level:

                                  g                                         38                                     38                  -   binary feet, RH, QI, R-to-L,

                                 Fs                                    Fw      Fs                              Fw       Fs                 Iterative, no extrametricality

                                38                                   g      38                           38     38            Word level:

                            W      S                                 S   W    S                      W   S    W   S         - RH, QS, R-to-L, Iterative, R-

                            (x       x!)                              (xÙ)  (x    xÚ)                     (x   xÚ)   (x   xÙ)            most foot is extrametrical




Michif               (mi tSo!-w)                         (mi tSo!- wa!k)               (nI- mi)(tSO- n-aÙn)       (Cree as source language)

Examples        ‘He is eating’                  ‘They are eating’        ‘We are eating’

                            (ka rO!t)                               (SOÙ) (kO laÚ)                    (Oto!) (mO bIÙl)                (French as source language)

                            ‘carrot’                              ‘brown’                         ‘car’


In conclusion, we show that stress is systematic in Michif, and that both source languages have contributed elements to present-day phonology. Further, we show that it is worthwhile to look at Michif as a single system in order to find emergent patterns which might be missed if treated as two distinct languages.



Bakker, P. (1997) A language of our own. Oxford University Press.

Bakker, P. & R. Papen (1997) "Michif: A mixed language based on Cree and French", in S. Thomason, ed., Contact Languages.

Hayes, B. (1996) Metrical stress theory : principles and case studies.  University of Chicago Press : Chicago.

Rhodes, R. (1986) "Métchif:  A Second Look", Proceedings of the 16th Congress of Algonquianists.



Jean-Louis Rougé & Emmanuel Schang, Orléans


The status of the liquid consonant in Saotomense.


In this study we analyze a corpus of 1767 saotomense words (containing neither compounds nor derived forms) in order to investigate the phonological status of the liquid [l] in the main Gulf of Guinea Portuguese Creole. We gathered 828 occurrences of [l] in this corpus (the second phoneme for the number of occurrence: |a| 1477, |i| 635, |u| 609, |k| 456…). Such a high frequency contrasts with the very low rate of occurrence of the liquid in the other related Creoles (Angolar and Principense). A first explanation for this fact could be found in the fusion of Portuguese /l/, /r/ and /R/ into /l/. However, we show that the phenomenon is more complex. Relevant factors to be considered are:

-         Frequent deletion of the liquid of the Portuguese etymon when it is /r/ or /R/ (118 deletions vs. 36 deletions of /l/)

-         The genesis of non etymological /l/ in words of Portuguese and African (mainly Kikongo and Edo) origin in at least 30 attested words

-         The formation of consonant clusters with liquid, resulting from various phenomena (metathesis, epenthesis, vocalic deletion, etc.)

This raises the question of the formation of Saotomense in regard to other Creoles of the area: Is there a peculiar phonological phenomenon in Saotomense that explains the distribution of /l/? Is it possible to derive a creole word from its etymological counterpart (be it Portuguese or African)?

We show with quantified examples that a phonological explanation is not relevant here.



Eric Russel Webb, Western Michigan University


The naturalness of voice: examples from Negerhollands and Afrikaans


The topic of voice neutralization and assimilation has enjoyed a great deal of interest in the case of ABN (Algemeen Beschaafd Nederlands, Standard Dutch) and in European Netherlandic dialects.  Instances of word-final devoicing and cluster voice homogeneity in extra-territorial derivatives of Netherlandic have not, however, received significant attention within the phonological community.  This paper constitutes a first look at phenomena involving voicing in Afrikaans and Negerhollands, offering a survey of the relevant linguistic forms, a comparison to Dutch data (both standard and dialectal) and an analysis of those phonological comportments from an Optimality-Theoretic (OT) perspective.  Secondary questions addressed in the paper include the role of markedness and faithfulness in Creole phonology and the proposition and formalization of phonological constraints.

The first section provides a brief overview of the historical and sociolinguistic development of Negerhollands and Afrikaans, acknowledging the questionable Creole status of the latter language.  A second section describes voice quality alternations in each of these and contrasts these languages to Dutch and its dialects, highlighting both the similarities and differences evidenced in data.  Surface parallels, for example the question of word-final devoicing, are also discussed in light of contrastive underlying representations.  From this section it emerges that lexicalization has played a vastly different role in shaping the phonologies of each language; whereas the phonology of Dutch and, to a lesser extent, Afrikaans promotes word-final devoicing, lexical and pre-phonological constraints on word formation would seem to obtain word-final obstruents in Negerhollands.  Similar observations are made for intervocalic voicing and voice assimilation in clusters.  A third section examines the causative issues involved in active and passive voicing; here, focus is turned to the biomechanical principles involved in voicing, whether this is accomplished by design or by default.  Effort reduction and avoidance principles are subsequently integrated into a phonetically driven phonological grammar using OT, wherein universal constraints promoting input-to-output faithfulness (the correspondence of an output form to its underlying representation) are countered by markedness principles (statements about which linguistic behaviors and operations are more or less difficult and produce a better or worse perceptual result).  Crucially, the analysis of voice alternations in Negerhollands and Afrikaans highlights a common tendency in Creoles, i.e. a great deal of obedience to markedness considerations, relatively unconstrained by faithfulness concerns.  These observations are partially borne out in the lexicon of Afrikaans and Negerhollands, in the latter case to a much greater extent.  A final section synthesizes the analysis, addresses questions of falsifiability and suggests paths of future research.



Shobha Satyanath; Delhi


Morphology within and across generations of speakers: the case Guyanese Creole English


Various theoretical paradigms in linguistics have for long systematically prevented linguists from exploring alternative ways of analysing creole languages. Undue emphasis on abrupt creolization despite evidence to the contrary in many cases, comparison of creoles with their modern standard superstrate and substrate counterparts have all added to the inadequacies of such languages and in turn difficulties to the understanding of the systems of grammar operating in such languages. Though some of the orthodoxy has been shaken now, many of the assumptions continue such as those that still believe in superiority of the inflectional systems of morphology over other systems of morphology and so on.

As far as morphology in Guyanese is concerned, what has intrigued scholars is the seemingly unstable relationship between the presence of inflection like categories associated with verbs and nouns and the meanings they supposedly express as expected in modern standard English. Such variability in turn has been widely but erroneously analyzed in terms of decreolization. This in turn has contributed to a gross misunderstanding of morpho-syntactic structures of Guyanese and in particular, the TMA system of Guyanese, which involves morpholgy, syntax and semantics (see Satyanath 1991, 1994, 1998).  In my view, the problem is not just the apparent variability of the role of inflection and its alleged functions but more important the stability of the observed patterns.

The present study intends to reanalyse the morphological structure of Guyanese. Its specific objectives are: (i) the nature of morphological structure(s) in Guyanese (ii) development of morphology over the last hundred years in the speech of the East Indians (born during 1900-1990) combining data from acquisition, development and change, and finally (iii) what inferences can be drawn from (i) and (ii) regarding the emergence of system(s) of morphology in Guyanese creole. The period chosen represents a crucial phase involving social-political-economic and linguistic transformation in the history of Guyana.



Norval Smith, Amsterdam


A historical explanation for the difference between English-lexifier creoles with Surinam-type vowel-systems and Jamaican-type vowel systems


In Smith (1999) a distinction is made between  Surinam-type vowel systems and Jamaican-type vowel systems in Atlantic area English-lexifier creoles. Surinam-type vowel systems are found in the following: the Surinam creoles: Sranan, Saramaccan, Nduka, etc. (older) Jamaican Maroon Creole (also known as MSL) Krio while Jamaican-type vowel systems include:


1.                 Jamaican Creole and systems derived from it

2.                 Barbadian, Guyanese, Trinidadian

3.                 Providence, Miskito Coast, Belizean


The essential difference between these two types is that in originally closed syllables no length opposition is retained. The Early Modern English long vowels and diphthongs are reflected by short vowels, cf. Sranan feti and Jamaican fait, both from English fight, or Sranan futu and Jamaican fuut, both from English foot.

We will depart from Smith’s hypothesis that all CEC’s had Surinam-type vowel systems in 1739 (Smith 1999), and that some greater centralization in the administration of the various British colonies might have been responsible for the change to Jamaican-type vowel systems in those creoles that have them now. The creoles that had this type of vowel system at the end of the 18th century were all spoken in areas under direct British control at the time.

Instead we will present an alternative proposal where we associate the more English-like Jamaican-type vowel systems with the transportation into slavery of hundreds of prisoners from the failed Monmouth rebellion of 1685. These prisoners were transported as slaves to Jamaica, Barbados and the Leeward Islands. It is in these areas, and other areas colonized from them that Jamaican-style vowel systems are encountered.



Tonjes Veenstra, Berlin


Head ordering in synthetic compounds: acquisition processes and grammatical theory


Synthetic compounds in the Surinamese Creoles are of particular interest to creole studies, since neither the superstrate language (English, NV) nor the major substrate language (Gbe, NV) seems to have been the model for the head-modifier order in these constructions. Thus, in both these languages the order of the noun-verb combination is the mirror image of the order found in Saramaccan (VN). The issue is to which factor in the creolization process this ordering difference is due: second or first language acquisition?

Although synthetic compounds is a relatively unexplored topic in SLA, all the studies (e.g. Lardiere 1994, 1995) report that “errors” were made in affixation, head-modifier order, and semantic interpretation, due to (UG-constrained) L1 influence. If the head-modifier order in interlanguages is due to L1 influence in SLA, we would expect that the order of Gbe (NV) would have survived in the Saramaccan, contrary to fact.

Studies on the L1 acquisition of synthetic compounds (e.g. Clark 1993) show that initially children have problems in determining the correct position of the affix as well as the head-modifier order. The latter is particularly unstable in the first few years (in production as well as comprehension). Three stages are distinguished. By the first stage, children are able to produce agentive V+MAN compounds. At this point, modifier and head appear in compound order. Clark suggests that children have made a generalization about head position: the rightmost noun designates the semantic category. At the second stage, compounds are formed with canonical predicate order. Clark suggests that another generalization is being made: “… what children at this stage appear to do is nominalise the verb phrases in the descriptions they hear.” The third stage of verbal compound acquisition is merely the realization that canonical predicate order does not apply to verbal compounds. Thus, the head-modifier order in Saramaccan corresponds to stage 2. We argue that the creators of Saramaccan sticked to this order due to the structurally-underdetermined input they encountered. The conclusion, therefore, is that the head-modifier order in synthetic compounds is due to FLA.

The position of the affix, however, is not so readily explained in a FLA scenario, since in stage 2 the affix is on V, N or both, the V-ER+N pattern being most frequent. There are two possible analyses: (i) -ma was at first not an affix in Saramaccan; (ii) -ma is an affix from the start and its final position is due to the Head Ordering Principle of Hawkins & Cutler (1988). We argue for the latter, and give an account in terms of psycholinguistic processing preferences.

We give a DM-analysis of synthetic compounds as nominalized VPs which involves zero-affixation.



Josje Verhagen, Eva van Lier, Suzanne Dikker, Hugo Cardoso, Jacques Arends, Amsterdam NL


Morphological richness and formative context of four Romance-based creoles


This paper addresses the ongoing debate on the status of morphology in creole languages and the most recent views on the subject put forward by such scholars as John McWhorter (1998) and Michel DeGraff (2001). McWhorter (1998) claims that, prototypically, Creoles have little or no inflectional morphology and only semantically regular derivational morphology. In reply, DeGraff (2001) disputes this observation; with particular reference to data drawn from Haitian Creole, he argues that this view is simplistic and therefore inadequate.

Departing from a survey of the morphology extant in Angolar, Seychellois, Palenquero and Papiamentu, our paper aims at evaluating the two conflicting views described above. A close comparative analysis of the data indicates that Palenquero and Papiamentu surpass Angolar and Seychellois in terms of morphological richness, especially in the domain of verbal morphology. While Palenquero uses morphological means to express or produce past tense, and both Palenquero and Papiamentu have adjectivally used past participles, and gerunds, these morphological features are largely absent in Seychellois and Angolar.

Relating these facts to the debate on the status of morphology in creole languages, we will try to account for them in socio-historical terms. Regarding the latter issue, a superstrate explanation is not very plausible since, with the exception of Seychellois, all four creoles are lexically based on 16th-17th-century Ibero-Romance. Similarly, all four languages are characterized by a strong substrate influence from Bantu, West Bantu in the case of the Iberian Creoles and East Bantu in the case of Seychellois. This suggests that an explanation should be sought elsewhere, e.g. in the differential social contexts. Although synchronically both Angolar and Palenquero may be seen as Maroon creoles, they are distinguished by the fact that Angolar, but not Palenquero is an off-shoot of a plantation creole. With regard to Papiamentu it is important to note that, although that language is sometimes classified as a plantation creole, the history of Curaçao suggests that the formative context of Papiamentu was very different from that of the classical plantation creoles.

If this is correct, the four creoles discussed in this study could be categorized in two groups: (former) plantation creoles (Angolar and Seychellois), and non-plantation creoles (Palenquero and Papiamentu). To the extent that this categorization is justified, it would correlate with the difference in morphological richness referred to above. Arguably then, this correlation could be of relevance to the debate on morphology in creole languages, in the sense that the morphological design of a creole is at least partly determined by the external circumstances in which it arose.



Bao Zhiming, Singapore


Word Accentuation in Singapore English


Singapore English, an English-based contact language, has been described as having staccato or syllable-timed rhythm, illustrated below (acute, primary stress; grave, secondary stress).


(1)   a.  phònòlógy/*phònológy      b.  mònòpóly/*monopoly


In the examples, primary stress falls on the penult, and secondary stress on the pre-tonic syllables, in apparent violation of clash avoidance (*phònológy, *mònopóly). Yet in words such as abacus and calendar, the stress pattern obeys clash avoidance:


(2)   a.  abácus/*àbácus        b.      caléndar/*càléndar


There are, of course, words with identical or similar metrical structure. Topic has the same stress pattern, and generate is gèneráte in Singapore English, and géneràte in English. The foot structure is otherwise identical. In Singapore English, primary stress is on the final foot.

     In this talk, I show that this peculiar word accentuation pattern is due to the emergence of the new foot typology (H)/(sL) in Singapore English, which interacts with the English-derived typology (H)s/(sL)s (Burzio 1994). The staccato impression reflects the metrical structure enriched by the (H)s/(sL)s and (H)/(sL) footings.

     Given a string of two syllables s1s2, the two typologies produce three distinct metrical footings, enumerated in (3) ([, English foot structure; (, Singapore English foot structure).


(3)   a.         [s1s2 ® [(s1(s2 Þ s1(s2         (ábacus vs. abácus)

       b.        s1[s2 ® (s1[s2                        (phonólogy vs. phònòlógy)

       c.         [s1s2 ® [(s1s2                        (tópic vs. tópic)


(3c) is the matching type, words such as topic, generate belong here. (3a,b) are mismatches, where the two foot typologies produce different footings. Abácus and caléndar belong to (3a), and phònòlógy and mòrphòlógy to (3b). What is puzzling is that the clash between s1 and s2 is tolerated in (3b) (-no- in phònòlógy), but not in (3a) (a- in abácus). The puzzle can be explained if we take Avoid Clash (Halle and Idsardi 1995) as an active constraint on each metrification—one based on (H)s/(sL)s, one on (H)/(sL)—that bars illicit metrical structure from being created. The structure [(s1(s2 violates Avoid Clash on Singapore English metrification, motivating the step marked by Þ. In (3b,c), Avoid Clash is obeyed in each of the two metrifications, even though the surface (s1[s2 configuration constitutes stress clash. Clash avoidance is a constraint on derivation, not on representation.

     The (H)/(sL) typology is a novel foot typology that has emerged in Singapore English. It interacts with the English-derived (H)s/(sL)s typology, creating the rich metrical structure, modulo the effect of Avoid Clash, that gives the casual observer the staccato impression.