Klein Creole Syllables 2003

Diversity and complexity in the typology of syllables in Creole languages Thomas B. Klein Georgia Southern University Department of Writing & Linguistics P.O. Box 8026 Statesboro, GA 30458 USA +1-912-681-0986 [email protected] www.georgiasouthern.edu/~tklein

Diversity and complexity in the typology of syllables in Creole languages Abstract This paper presents an analysis and systematic survey of types of syllables based on a typological sample of 23 Creole languages. Onsets, codas and the occurrence of the syllable types V, CV, CVC, VC, CCV, CCVC, CVCC, VCC and CCVCC are investigated based on data from Atlantic, African, Asian and Pacific Creoles with European and non-European lexifier languages. In this strictly synchronic approach, it is found that no Creole has exclusively CV syllables, that the syllable inventories of Creole languages are diverse and that they may exhibit considerable complexity. It is shown that the most common syllable template in Creole languages is (C) (C) V (C). These results have important implications for the understanding of Creole syllables because most views of syllable typology in Creole languages are grounded in the idea that simple CV syllables are dominant. Instead, Creole languages center on a medium degree of complexity in terms of the inventory of syllables they allow. Introduction Surveys of the structure of Creole languages have maintained that CV syllables, that is, open syllables consisting of a vowel preceded by a consonant, are dominant in the prosodic structure of these languages (e.g. Bartens 1995, Holm 2000, Kaye & Tosco 2001, Romaine 1988). It has even been claimed that CV syllables are the only type occurring in Creole languages. "Creoles have no initial or final consonant clusters. They have a simple syllable structure which consists of alternating consonants and vowels, e.g. CVCV." (Romaine 1988: 63) The alleged preference for CV syllables is taken to be the root cause for phonological restructuring in superstrate words. Processes of vowel insertion, consonant deletion or metathesis may indeed conspire to produce Creole forms that are more CV-like than the corresponding items in European superstrate languages. The restructuring seen in (1) results in Creole words with exclusively CV syllables. (1)

Creole form kupa sisa kini pilime sukú

Etymon ocupar sister knie plume oscuro, escuro

Language Príncipe Sranan Negerhollands Mauritian Papiamentu (Data from Holm 2000: 141 ff.)

Words with divergent syllable types, in particular involving complex onsets and codas, have generally been treated as late or decreolized additions to the lexicon of a given

1

Creole language (e.g. Alleyne 1980, Boretzky 1983, Mühlhäusler 1997). Recent research, however, has shown that even languages considered radical Creoles like Saramaccan and Sranan have syllables with complex onsets in the earliest stages of their formation. Examples shared by Early Saramaccan and Early Sranan are smoko ‘smoke’ and srepi ‘self’. In addition to CCV syllables, Early St. Kittian and Early Jamaican also routinely allow for syllable codas, as in lib ‘live’ and pass ‘past’. This evidence is important in the present context because it refutes the global idea that complex Creole syllables are necessarily the result of later developments or decreolization. In other words, there is strong evidence that the tendency towards CV word structure may have developed over time, at least in the Suriname Creoles (Aceto 1996, Plag & Schramm, forthcoming; cf. Mühlhäusler 1997: 189). A closer look at the evidence reveals that syllable types other than CV are attested even in items that have been subject to phonological restructuring and in Creole languages that are claimed to be influenced by the CV pattern. (2)

Non-CV syllable (a) a V (b) flo CCV gro CCV wro CCV

Creole form álima konofló groto wroko

(c) ret

sitiret

CVC

Etymon Language alma Príncipe knoflook Papiamentu groot Negerhollands work Sranan (Data from Holm 2000: 141 ff.) straight Tok Pisin (Example from Sebba 1997: 110)

The evidence in (2) shows that phonological restructuring in Creole languages does not uniformly result in CV syllables. Instead, restructured words may contain onsetless syllables, closed syllables or open syllables with complex onsets. This evidence suggests that syllable structures across Creole languages are significantly more varied than commonly assumed. This idea is supported by observations in the literature that certain Creole languages fail to exhibit any significant tendency towards CV structure. Atlantic French-lexified Creoles have been cited in this context (e.g. Holm 2000, McWhorter 2000, Parkvall 2000). Codas and consonant clusters may be found in other Creoles languages as well. For instance, Negerhollands routinely tolerates onset clusters and obstruent codas as in forfluk ‘sly’ (Sabino 1990, 1993, Stolz 1986; see also Parkvall 2000). Given that Creole syllable structure seems significantly more diverse than previously assumed, the question arises as to what the actual syllable types of Creole languages are. However, no systematic survey of the syllable typology of Creole languages is available to date to address it. Few descriptive works of individual Creole languages such as Carrington 1984 and Stolz 1986 provide detailed analyses of syllable types. Linguists have no general source to glean the syllable typology of Creole languages. This is a significant shortcoming given the importance of Creole syllable structure in discussions of substrate influence, diachronic development, decreolization, synchronic simplicity and uniformity, among other issues. Thus, it is important to develop and implement a survey and analysis of the synchronic syllable types of Creole languages based on objective criteria.

2

It has been argued recently that the synchronic grammars of Creole languages are in some significant sense simple in comparison to the grammars of non-Creole languages (McWhorter 2001). This Creole simplicity hypothesis has not been tested with data from syllable structure. It is not possible at this point to compare a sufficiently rich sample of the syllable structure of non-Creole languages with the present sample of Creole languages. No comprehensive typological database and analysis of non-Creole syllables comparable to, for instance, Maddieson’s (1984) work on consonants and vowels is available. Given that it appears that the syllable structure of Creole languages is richer than previously thought, it is possible to test the Creole simplicity hypothesis by looking at just the set of attested inventories of syllable types in Creole languages. If the phonology of Creole languages was indeed overwhelmingly simple, we would expect to find a significant majority of simple syllable type inventories, perhaps similar to the CV dominance claimed in earlier works. It will be shown in the course of this paper that this prediction of the Creole simplicity hypothesis is not confirmed. Instead, significant evidence is uncovered to support the idea that the synchronic syllable typology of Creole languages tends to occupy middle ground in terms of complexity. This paper takes seriously the idea of diversity in Creole syllable structure and presents results of a systematic survey of synchronic syllable types based on a typological sample of the world’s Creole languages. The remainder of the paper is structured as follows. I show next how the present sample is constructed. Following this, basics of syllable structure and syllable typology are laid out. The analysis of the database is followed by a discussion of some of the implications of the results before conclusions are offered. The appendix provides illustrative examples of the syllable types found in the languages in the present sample. Database of Creole languages The present sample is constructed to represent the range of Creole languages around the world. Care is taken to include Creoles from diverse locations and with European and non-European lexifiers. This is done to ensure that generalizations are based on Creole languages as a global phenomenon, not just a particular group of them. Following Maddieson’s (1984) work on the typology of phonemic systems, a quota sample is employed. The quota is to bring in two Creoles with the same or very similar lexifier from a given geographical area wherever possible. The aim is to provide a balanced sample, to enable sampling efficiency and to ensure maximal diversity of languages while keeping the size of the database manageable. Acrolectal Creoles, varieties or vocabulary are generally excluded. The reason is that acrolectal material tends to be closest to the syllable structure of the corresponding lexifier languages and is, hence, not indicative of Creole structures. The focus is on mesolectal and basilectal varieties to make sure that distinctly Creole linguistic forms make up as much of the database as possible. Recent additions to the lexicon of a given Creole such as loanwords from European lexifiers have been avoided wherever they are obvious or the sources identify them. For example, additions from Dutch to the lexicon of Papiamentu or loans from Guyanese Creole English in the lexicon of Berbice have been disregarded to focus on the core of these Creole languages rather than the periphery.

3

There is also a bias towards the quality of the available descriptions. Creole languages are chosen in part based on how detailed the information on the phonology and word structure is in the sources and how reliable this information seems. Most descriptions of individual Creole languages do not discuss syllable structure in great detail. The necessary data to establish the range of syllable types in a given Creole language must often be extracted from examples provided for other purposes or from texts, glossaries or dictionaries. As much as possible, only words that appear monomorphemic in the Creole languages are considered for the present analysis. Only languages with sources cumulatively containing enough data represented in some sort of phonetic representation could be considered for the present study. The sample contains 23 Creole languages. Table 1 lays out how they are classified according to geographic area and lexifier language. The sources for the data are identified in the appendix. Note that the sample includes varieties that have only recently become nativized as Creoles such as Tok Pisin and Sango. Area Atlantic

Indian Ocean/ Pacific

Africa Asia

IE-Lexifier Dutch

Creole Negerhollands Berbice English Ndyuka Saramaccan French Haitian St. Lucian Portuguese Angolar São Tomense Spanish Papiamentu Palenquero German Unserdeutsch English Bislama Tok Pisin French Mauritian Tayo Portuguese Sri Lankan Kristang Spanish Zamboangueño Ermitaño Non-IE-Lexifier Bantu Kituba Sango Arabic Nubi Malay Baba Malay Table 1. Creoles in sample

The lexifier languages are assigned the central role in classifying the Creoles in the sample linguistically. The basic geographical division for Creoles with Indo-European lexifier languages is between Atlantic creoles and the parts of the world delineated by the

4

Indian Ocean and the Pacific. The five European lexifier languages Dutch, English, French, Portuguese and Spanish are easily matched with two Creoles each in the Atlantic area. This is less straightforward for the Pacific Rim/Indian Ocean Creoles because there is no Dutch-lexified Creole in this area. Rabaul Creole German (Unserdeutsch) takes the place of a Dutch Creole here given that the lexifier languages are very closely related. Creoles with non-Indo-European lexifiers are considered equally important for the comprehensive understanding of Creole languages. They are classified geographically according to the continent on which they are found. There is only one Arabic-lexified and Malay-lexified Creole represented because of a lack of descriptive resources to extract syllable structure information on other languages of this type. Syllable structure and typology For the purposes of this investigation, syllables are broadly defined descriptively as a surface sequence of consonant (C) and vowel (V) units dominated by the constituents onset, nucleus and coda. Only nuclei are obligatory, whereas all other units are optional, as shown by parentheses, where appropriate. As an illustration, the syllable structure of the Unserdeutsch word /traxt/ ‘dress’ is displayed in (3) (σ = syllable, O = onset, N = nucleus, Cd = coda). (3)

σ O

N

Cd

C

C

V

C

C

t

r

a

x

t

The sonority of the segments involved is a deciding factor in how they are arranged into syllables. Vowels are most sonorous; glides, sonorants, fricatives and plosives decrease in sonority, in that order. The nucleus is the sonority peak in a syllable, whereas segments preceding and following the nucleus have progressively less sonority. In the structure in (3), the peripheral plosive /t/ has the lowest sonority, the vowel peak /a/ has the highest, and the non-peripheral consonants /r/ and /x/ have intermediate degrees of sonority. A well-known complication in the relation between sonority and syllable structure concerns clusters of /s/ or // followed by obstruents, which can be found in many European languages. Against expectation, sonority does not progressively decrease towards the edge of the syllable in these clusters. There has been a diversity of approaches to resolve this issue in the phonological literature. Some researchers have argued that /s/ or // are prespecified in onset position (e.g. Giegerich 1992), whereas others have maintained that /s/ or // are adjoined to the onset node (see Kenstowicz 1994). For the purposes of this

5

paper, I assume that clusters of /s/ or // plus obstruent belong to the syllable onset wherever feasible. Among the optional constituents of the syllable, onsets are more important than codas. Following Kahn (1976), among others, onsets are maximal in that making legitimate onsets has priority over forming a legitimate coda. As one result, VCV sequences are syllabified as V.CV, where the dot indicates the syllable boundary. As another consequence, complex onsets are formed word-internally at the expense of codas if the complex onset can occur word-initially. However, complex onsets are not formed word-internally if they cannot occur word-initially. There is good reason to argue that the Arabic-lexified Creole Nubi presents an instance of the situation just given. This language exhibits a rich array of word-internal consonant sequences, but there are no corresponding word-initial clusters. For example, given words such as /libra/ ‘needle’ and /asma/ ‘name’, one would expect words beginning with br- and sm-. However, the only word-initial clusters found in Nubi are instances of consonant plus w in a few words of African or unknown origin as in /gwanda/ ‘cassava’ (see Heine 1982, Pasch & Thelwall 1987). If clusters such as br- and sm- are analyzed as hetero-syllabic, even though the segments would fit into a single onset according to their sonority profile, then the unexpected absence of word-initial clusters can be accounted for. They are not licensed because clusters not involving /w/ are not allowed anywhere in the language. Thus, I propose that Nubi has essentially a (C) V (C) syllable template like its lexifier Arabic (see Kenstowicz 1994 and references cited there) and that words such as the ones above are syllabified as /lib.ra/ and /as.ma/, respectively. Across a variety of theoretical persuasions, the CV-phonology model (Clements 1990; Clements & Keyser 1983) and the parameter-based approach to syllable typology in Blevins (1995) have been influential with researchers aiming to understand the typology of surface syllables in the languages of the world. The core syllable typology in the CV-phonology model in (4) may be seen as constrained variations of the prototypical CV syllable. Note that each C and V can represent a potential cluster in this model. (4)

Core syllable typology (Clements 1990, Clements & Keyser 1983) Type I: Type II: Type III: Type IV:

CV CV, V CV, CVC CV, V, CVC, VC

In type I, syllables must have onsets and must be open. Languages of type II prohibit codas, but onsets are optional. Closed syllables are allowed in type III, and onsets are obligatory. Type IV languages allow syllables without onsets as well as open and closed syllables. According to Blevins (1995), languages can be described in terms of a small set of binary-valued parameters defined over the sub-syllabic constituents onset and coda. The obligatory onset parameter determines whether an onset is required or not. The coda parameter decides whether a language allows codas or not. The complex onset and complex coda parameters determine whether a language does or does not allow complex

6

onsets or codas, respectively. Blevins (1995) also encodes edge effects, that is, the occurrence of certain syllable types in peripheral versus medial position in the word. Following Blevins’ distinction between simple and complex onsets and codas, I have investigated the occurrence of the following syllable types in Creole languages: V, CV, CVC, VC, CCV, CCVC, CVCC, VCC, and CCVCC. Based on the set of potential surface syllable types and the settings for the four syllable structure parameters, a language allowing only CV syllables shows the profile in (5). (5)

Typological profile for CV language (cf. Blevins 1995)

(a)

Syllable types V

(b)

CV yes

CVC

VC

CCV

CCVC CVCC

VCC

CCVCC

Parameter settings Oblig. Onset yes

Complex Onset no

Coda no

Complex Coda no

Naturally, a strict CV language would only allow that syllable type, as in (5) (a). Only the obligatory onset parameter would be set to ‘yes’, whereas all other parameters would be switched off, as in (5) (b). This is a simple inventory by any measure. Note that most, if not all, Creole languages should exhibit this profile, according to common wisdom in much of linguistics. As shown below, this turns out to be a myth when confronted with typological data from the world’s Creole languages. Onsets The investigation of Creole syllables may be approached by looking at onsets and codas as distinct constituents. As far as onsets are concerned, the present investigation has found that no Creole language in the sample requires them. In other words, Creoles allow syllables with or without onsets. This is expressed as the generalization in (6). (6)

No Creole language requires syllable onsets.

Table 2 provides data for onsetless syllables from all of the Creoles in the sample. More specifically, it provides examples for syllables consisting only of V in word-initial position. Many Creole languages also allow VC syllables in word-initial position; some even permit VCC syllables. Examples for such syllable types may be found in the appendix. In Table 2 and below, glosses have been translated if the original language of the source is not English. No gloss has been attempted if none could be found for a given item in the source. 7

Negerhollands Berbice Ndyuka Saramaccan Haitian St. Lucian Angolar São Tomense Papiamentu Palenquero Unserdeutsch Bislama Tok Pisin Mauritian Tayo Sri Lankan Kristang Zamboangueño Ermitaño Kituba Sango Nubi Baba Malay

/ale:/ ‘alone’ /atriti/ ‘reverse’ /agwado/ ‘stringed instrument’ /eside/ ‘yesterday’ /emab/ ‘amiable’ /epi/ ‘with, and’ /abi/ ‘April’ /ami/ ‘I’ /iglesia/ ‘church’ /abla/ ‘speak’ /abn/ ‘evening’ /akis/ ‘axe’ /abris/ ‘go past’ /ale/ ‘to go’ /ako/ ‘again’ /ra/ ‘miss’ /eli/ ‘he, she, it’ /amo/ ‘boss’ /ele/ no gloss given /awa/ ‘here’ /ape/ no gloss given /akulu/ ‘to eat’ /anak/ ‘baby’

/izu/ ‘iron’ /oboko/ ‘hen’ /okoo/ ‘okra’ /oto/ ‘other’ /ijen/ ‘hygiene’ /ofe/ ‘to offer’ /ome/‘human’ /ose/ ‘sky’ /uzu/ no gloss given /uto/ no gloss given /eli/ ‘early’ /epa/ ‘stingray’ /ena/ ‘there’ /ale/ ‘to go’ /æ:tik/ ‘tuberculosis’ /otro/ ‘other’ /ohos/ ‘eyes’ /i.nsi/ ‘land’ /ila/ ‘except’ /ula/ ‘snake’

Table 2. Onsetless syllables in Creoles The data in Table 2 show that syllables consisting only of a nucleus are allowed in all Creoles. In terms of the CV-phonology model, this immediately excludes Creole languages from Type I and Type III because these types presume obligatory onsets for all syllables. In terms of Blevins’ framework, the obligatory onset parameter is set to ‘no’ in all Creole languages. Furthermore, there appears to be a fairly strong edge effect, that is, onsetless syllables seem to be permitted only word-initially, for the most part. I have come across data and reports of vowel hiatus in several Creoles, however. It is attested in a few Creole languages in the sample, as in Baba Malay /a.os/ ‘thirsty’, Berbice /bu.in/ ‘hide’, Sango /to.a/ (no gloss given), St. Lucian /ãvi.e/ ‘curious’ and Zamboangueño /a.ora/ ‘now’. Unfortunately, the currently available data sources preclude a more detailed investigation of the occurrence of vowel hiatus across Creole languages. All Creole languages have CV syllables, that is, open syllables with a single onset consonant before the vocalic nucleus. This result is not surprising given that CV syllables are found in any natural language. Examples may be gleaned from the appendix and from Table 2, where most word-initial V-syllables are followed by a CV syllable. It has been claimed in some works that CV syllables are overall most frequent in Creole languages. Few researchers have presented any quantitative analysis of Creole syllable types, however. CV syllables are the most frequent type judging from Stolz’s (1986) numbers. In Carrington’s 1984 count of St. Lucian French Creole, CV syllables

8

occur in 61.33% of the total corpus analyzed. Whereas these results are suggestive, there has been no systematic investigation of syllable type frequencies across Creoles. The present investigation abstracts away from the question of syllable frequency. Instead, it focuses on determining which syllable types and constituents are licensed to occur in the core vocabulary of a given Creole. The goal is to uncover the possible, not the most frequent, syllable types of Creole languages and to make typological generalizations framed in the phonological models selected. Thus, it is tangential to the present concerns if CV syllables are the most frequently occurring type in Creole languages, even though this question is of interest to Creole studies and linguistics in general, of course. Syllables with complex onsets, that is, at least two non-vocalic segments before the vocalic nucleus, are found in nearly all Creole languages. It seems necessary to recognize two classes, however. A few Creoles allow only glides (G) to the immediate left of the nucleus, giving CGV structures, whereas a larger second class allows a fuller set of consonants in this position, in particular sonorants and glides. Saramaccan, Baba Malay and, with very few exceptions of Cw clusters, Nubi are the only languages in the sample allowing exclusively single onsets in citation forms. However, Pakir (1986) reports on a fast-speech rule of schwa-deletion in Baba Malay which creates word-initial CC clusters, e.g. /skali/  [skali], /prut/  [prot] (with vowel-lowering) (no glosses given). Thus, it appears that CC onsets may occur post-lexically in Baba Malay. Table 3 lays out Creole complex onsets in citation forms with examples for all languages in the sample. Additional initial cluster data may be found in the appendix. CG CC

Ndyuka /dyendee/ ‘splendid’ Kituba /kudya/ ‘to eat’ Negerhollands /bwa/ ‘preserve’ Berbice /plk/ ‘place’ Haitian /klu/ ‘nail’ St. Lucian /plãte/ ‘to plant’ Angolar /bwaru/ ‘good’ São Tomense /tlisa/ ‘jaundice’ Papiamentu /skapa/ ‘escape’ Palenquero /drumi/ ‘to sleep’ Unserdeutsch /fry/ ‘early’ Bislama /slo/ ‘slow’ Tok Pisin /tri/ ‘tree, three’ Mauritian /prekot/ ‘near’ Tayo /pli/ ‘plus’ Sri Lankan /dre:tu/ ‘correct’ Kristang /greza/ ‘church’ Zamboangueño /klase/ ‘class’ Ermitaño /klaa/ ‘club’ Sango /srango/ ‘doing’

/kwaka/‘cassava granules’ /kwenda/‘to go’ /groma/ ‘greedy’ /speki/ ‘pork’ /bra/ ‘arm’ /bizwe/ ‘to need’ /staka/ ‘picket’ /kwelo/ ‘rabbit’ /subla/ ‘blow’ /kumbla/ ‘to buy’ /kwel/ ‘source’ /smol/ ‘small’ /klos/ ‘clothes’ /lapli/ ‘rain’ /pukwa/ ‘why’ /kwæ:ntru/ ‘coriander’ /stiru/ ‘style’ /alegre/ ‘happy’ /entro/ ‘inside’ /skula/ ‘to wash’

Table 3. Complex onsets in Creoles

9

20 out of 23 Creole (= 87%) in the sample routinely have complex onsets of some kind in citation forms. 2 of these 20 Creoles allow only glides as C2 in C1C2V complex onset structures. All other Creoles (18 of 23 = 78%) allow sonorant consonants as C2. These results are the empirical backdrop for the generalization in (7). (7)

The great majority of Creole languages allows complex syllable onsets.

Recall that the CV-phonology typology does not distinguish between single and complex onsets. Thus, it cannot express the distinction between Creoles with only single onsets and those with complex onsets. As far as the parameter-based model is concerned, the complex onset parameter is set to ‘yes’ in the vast majority of Creoles, allowing complex onsets. A minority of Creole languages has this parameter set to ‘no’, allowing only single onsets. The data in (8) provide evidence concerning the question if there is an edge effect across Creoles with respect to complex onsets. A percentage of these examples is familiar from Tables 2 and 3. (8)

Word-internal complex onsets Negerhollands Berbice Ndyuka Haitian St. Lucian Angolar São Tomense Papiamentu Palenquero Unserdeutsch Bislama Tok Pisin Mauritian Tayo Sri Lankan Kristang Zamboangueño Ermitaño Kituba Sango

/forfluk/ /atriti/ /agwado/ /ijen/ /bizwe/ /uvwa/ /kopla/ /iglesia/ /abla/ /gmyk/ /katres/ /abris/ /lapli/ /pukwa/ /kwæ:ntru/ /ombru/ /otro/ /entro/ /uvwa/ /ndapre/

‘sly’ ‘reverse’ ‘stringed instrument’ ‘hygiene’ ‘to need’ ‘nine’ ‘to buy’ ‘church’ ‘speak’ ‘decorated’ ‘cartridge’ ‘go past’ ‘rain’ ‘why’ ‘coriander’ ‘shoulder’ ‘other’ ‘inside’ ‘nine’ no gloss given

The data in (8) reveal that all Creoles that allow complex onsets permit them wordinternally. Recall that Saramaccan, Nubi and Baba Malay do not tolerate complex onsets and, hence, do not appear in (8). Thus, there is no edge effect with regard to complex onsets found in any of the Creoles.

10

Codas The claim that CV syllables are dominant in Creole languages entails that the great majority of Creoles should allow only open syllables. This section presents results of the current investigation concerning codas in Creoles. Glides and nasals in potential coda position have been excluded from the count when these elements are the only ones found in coda position. Single codas of glides and nasals are often considered not to be ‘real’ codas, in part because of the vocoid properties of glides and the frequently observed alternations of oral vowels plus nasal codas with nasal vowels. Thus, only the occurrence of non-nasal sonorants or obstruents in the right periphery of the syllable establishes the presence of the coda constituent for the present survey. A handful of Creole languages do not allow codas by the measure used in this survey. They are Saramaccan, Ndyuka, Angolar, Kituba and Sango. This result is perhaps not surprising. Saramaccan, Ndyuka and Angolar are considered to be among the most radical and conservative Creoles in the world, whereas Kituba and Sango are lexified by Bantu languages, which generally do not allow closed syllables. The remainder of the Creole languages in the sample allows codas consisting of single consonants. An edge effect may be observed as well. Dots have been given in Table 4 to mark syllable boundaries that may not be immediately obvious from the outline of syllabification principles given earlier. No edge effect Negerhollands Berbice Haitian St. Lucian Papiamentu Unserdeutsch Bislama Tok Pisin Mauritian Tayo Sri Lankan Kristang Zamboangueño Ermitaño Nubi Baba Malay

Word-final /gobed/ ‘prayer’ /glof/ ‘believe’ /sik/ ‘sugar’ /dokte/ ‘physician’ /berdat/ ‘truth’ /gros/ ‘large’ /swit/ ‘sweet’ /fis/ ‘fish’ /latab/ ‘table’ /kat/ ‘four’ /dews/ ‘god’ /fasel/ ‘easy’ /abril/ ‘April’ /bos/ ‘voice’ /marid/ ‘fever’ /anak/ ‘baby’

Word-medial /forfluk/ ‘sly’ /fur.stan/ ‘understand’ /es.plãdid/ ‘splendid’ /uval/ ‘horse’ /falsu/ ‘mean’ /vl.h/ ‘which’ /has.ban/ ‘husband’ /alta/ ‘altar’ /palto/ ‘jacket’ /solda/ ‘soldier’ /korpu/ ‘body’ /uz.du/ ‘cuckold’ /kombersa/ ‘speak’ /pwelte/ no gloss given /wakti/ ‘time’ /basat/ ‘bed-bug’

Edge effect São Tomense Palenquero

/palma/ ‘palm tree’ /ah.ma/ no gloss given

/farkon/ ‘falcon’ /olbia/ no gloss given

Table 4. Single codas in Creoles

11

All of the Creoles in Table 4 show non-nasal sonorants or obstruents in coda position. This does not preclude nasals or glides from occurring in this position, of course. For example, the Bislama word /swit/ ‘sweet’ clearly establishes the coda by allowing the plosive /t/ at the end of the word. In the word /has.ban/ ‘husband’, the fricative /s/ is also a coda because it cannot form a complex onset with the following consonant /b/. Naturally, nasals as in the syllable /ban/ are found in coda position as well. A closer look at the data in Table 4 reveals the edge effect in the availability of codas. In São Tomense and Palenquero, codas are permitted in word-medial position, but no codas other than nasal consonants appear word-finally. In Palenquero, however, obstruent codas occur as the result of a post-lexical rule through elision of word-final unaccented vowels following consonants (e.g. [kus] - /kusa/ ‘thing’, [entons] - /entonse/ ‘then’ and [tampok] - /tampoko/ ‘either’ in Bickerton & Escalante 1970: 257). All other Creoles in Table 4 allow codas word-finally and word-medially in citation forms. 18 out of 23 Creole languages (= 78%) allow single non-nasal codas. This is an important result for the understanding of Creole languages because it is completely unexpected under the dominant CV idea. If open syllables were pervasive in the structure of Creole languages, we would expect most Creoles not to allow closed syllables. However, quite the opposite is the case, as expressed in the generalization in (9). (9)

The great majority of Creole languages allows closed syllables.

Given this generalization and the one in (6) above, only a minority of Creole languages in the CV-phonology typology model is of Type II, that is, allowing onsetless syllables, but no closed ones. Type IV must serve the majority of Creoles because it features onsetless and closed syllables. As far as the parameter-based model is concerned, the coda parameter is set to ‘yes’, therefore allowing closed syllables, in the great majority of Creole languages. A minority of Creoles has it set to ‘no’, therefore allowing only open syllables. In the present sample, complex codas are robustly attested in a few Creole languages: Haitian, Mauritian, Negerhollands and Unserdeutsch. Consider the data in (10). (10)

Complex codas (a) (b) (c) (d)

Haitian (Valdman1981) /kilt/ ‘cult’ /fiks/ /filt/ ‘filter’ /taks/ Mauritian (Baker & Hookoomsing 1987) /rekolt/ ‘harvest’ /fiks/ /filt/ ‘filter’ /taks/ Negerhollands /tomp/ ‘stomp’ (Sabino 1993) /kerk/ ‘church’ (Stolz 1986) Unserdeutsch (Volker 1982) /finf/ ‘five’ /as/

‘set, firm’ ‘tax’ ‘firm, stuck’ ‘tax’

‘afraid’

12

Valdman (1978) has stated that only the final consonant clusters /lt/ and /ks/ are found in French-lexified Creole languages. This is confirmed for Haitian and Mauritian, as seen in (10) (a) and (10) (b), respectively. The sources on St. Lucian and Tayo, however, do not provide enough information to investigate consonant clusters fully. We find /taks/ ‘tax’ in Mondesir (1992), but other candidate entries for the clusters are missing in this dictionary of St. Lucian. One instance of a complex coda in Tayo can also be found in Ehrhart (1993), namely, /lareserv/ ‘reserve’, but the glossary is too short to test the other candidates for complex codas. Negerhollands and Unserdeutsch are among the most complex Creoles known in terms of syllable structure. In addition to complex codas as shown in (10) (c) and (10) (d), respectively, Negerhollands allows clusters of three consonants in word-initial onset position (e.g. skrau ‘scratch’, Sabino 1993) and, according to Stolz (1986), in word-final codas, although he provides no example for this structure. Volker (1982) reports that in Unserdeutsch /s/ is added before // in clusters with /r/ and in word-final position, e.g. /spra/ ‘language’ and /mns/ ‘person’. The resulting clusters are unusually complex by any measure. Membership in the group of Creole languages lexified by Germanic languages other than English., however, does not predict the presence of complex codas, as the case of Berbice Dutch Creole shows. Berbice allows single codas, but complex codas are found only in loans from Guyanese Creole English (see the vocabulary presented in Kouwenberg 1994). The relative scarcity of complex codas in Creole languages is captured through the generalization in (11). (11)

Complex codas are uncommon in Creole languages.

The generalization in (11) holds over the core vocabulary of basilectal or mesolectal Creole varieties. Acrolectal lexicons, recent loans etc., in particular in Germanic-lexified Creoles, are well-known to show instances of complex codas with some frequency. The syllable typology in the CV-phonology model does not capture the distinction between Negerhollands, Haitian and Unserdeutsch on the one hand and the Creoles allowing only single consonant codas on the other. Both groups of Creoles fall into Type IV given that codas are allowed. The parameter-based model, however, readily distinguishes between the two groups. For Negerhollands, Haitian and Unserdeutsch, the complex coda parameter is set to ‘yes’, whereas it is set to ‘no’ for all other Creoles in the sample. There appears to be an edge effect, given that no complex codas could be found in word-medial position. In sum, this section has shown that single consonant codas are permitted in all but the most radical Creoles and the Bantu-lexified Creole languages. Complex codas, on the other hand, are scarce in basilectal and mesolectal varieties. The next section presents the results of the investigation of Creole syllable types. Syllable types Much of current thinking in linguistics would surmise that an extremely limited range of syllable types is observed across Creole languages. In particular, the idea of CV13

dominance would predict that there are no or very few syllable types other than CV in Creoles, notwithstanding the widely acknowledged occurrence of single codas in Atlantic French-lexified Creoles. The extent of the discussion of syllable types varies considerably in descriptions of Creole languages. Some works offer excellent and wide-ranging information on syllable structure (e.g. Carrington 1984, Smith 1977, Stolz 1986), whereas it is treated scantily or not at all in many other works. In the present article, I aim to tabulate the syllable types in Creole languages that can be extracted from the available sources and present this information so that meaningful comparisons across Creoles and with non-Creole languages are possible. Consider Table 5 in this context. V CV CVC VC CCV CCVC CVCC VCC CCVCC yes yes Saramaccan yes yes yes Angolar, Kituba, Ndyuka, Sango yes yes yes yes Baba Malay, Nubi yes yes yes yes yes São Tomense yes yes yes yes yes yes Berbice, Bislama, Ermitaño, Kristang, Palenquero, Papiamentu, Sri Lankan, St. Lucian, Tayo, Tok Pisin, Zamboangueño yes yes yes yes yes yes yes Haitian, Mauritian yes yes yes yes yes yes yes yes yes Negerhollands, Unserdeutsch Table 5. Attested syllable types across Creoles Table 5 shows that the Creole languages of the world exhibit a wide range of syllable types. The appendix presents structured data illustrating each syllable type. The data show that all nine syllable types under investigation are attested, albeit not in all Creoles. Creole languages exhibit a cline as far as the inventory of syllable types is concerned, licensing anywhere from two to nine syllable types. Despite the variation exhibited in Table 5, several generalizations are apparent. All Creoles have CV syllables, just like any other natural language. However, no Creole language is limited to CV syllables. All Creoles have at least one additional type, namely syllables consisting of just a vocalic nucleus. Furthermore, all Creoles, just like all other natural languages, exhibit the following property: if clusters of n Cs are possible syllableinitially, then clusters of n–1 Cs are also possible syllable-initially, and if clusters of n Cs are possible syllable-finally, then clusters of n–1 Cs are also possible syllable-finally (cf. Blevins 1995). Given Table 5, this means that the presence of clusters in onsets or codas entails the presence of single consonants in these positions, but not vice versa. For instance, a language like Baba Malay has single onsets and codas, but no complex ones. On the other hand, São Tomense has complex and single onsets.

14

Most of the inventories in Table 5 cannot be distinguished under the CVphonology typology. According to this model, Saramaccan, Angolar, Kituba, Ndyuka and Sango would all be Type II because they allow only open syllables, whereas all other Creoles would be Type IV, allowing codas. A more fine-grained model is needed to understand the variety of syllable types attested in Creole languages. The parameterbased model is better suited to account for the Creole syllable type inventories, but also cannot capture all of them. The first inventory allows V and CV syllables; it is easily accounted for in the parameter-based model. (12)

Inventory: V, CV (Saramaccan) Oblig. Onset no

Complex Onset no

Coda no

Complex Coda no

The syllable type inventory in (12) is the least marked that is possible in the parameterbased model, according to Blevins (1995). Given ‘no’ as the unmarked parameter setting and ‘yes’ as the marked one, the least marked inventory results when all parameters are set to ‘no’. This is interesting from the point of view of Creole linguistics given that is has been argued numerous times that Creoles grammars are in some significant sense unmarked or, conversely, lack marked patterns or inventories. By this token, we would expect a significant number of Creole languages with the parameter profile in (12) and the resulting syllable type inventory of V and CV. However, this is not the case by a wide margin. Most Creoles allow codas and complex onsets, showing that Creole syllable inventories are rarely entirely unmarked. Instead, the vast majority of Creole syllable inventories contains syllable types of at least some degree of markedness. Inventories allowing only open syllables, but also complex onsets are easily accounted for in the parameter-based model. Consider the parameter settings in (13). (13)

Inventory: V, CV, CCV (Angolar, Kituba, Ndyuka, Sango) Oblig. Onset no

Complex Onset yes

Coda no

Complex Coda no

The parameter settings in (13) allow for CCV syllables because the complex onset parameter is set to ‘yes’. The next profile restricts codas and onsets to single units. (14)

Inventory: V, CV, CVC, VC (Baba Malay, Nubi) Oblig. Onset no

Complex Onset no

Coda yes

Complex Coda no

The parameter settings for this inventory seem straightforward: the complex onset parameter and the complex coda parameter are both set to ‘no’. 15

The profile in (15) accounts for most Creole languages. (15)

Inventory: V, CV, CVC, VC, CCV, CCVC (Berbice, Bislama, Ermitaño, Kristang, Palenquero, Papiamentu, Sri Lankan, St. Lucian, Tayo, Tok Pisin, Zamboangueño) Oblig. Onset no

Complex Onset yes

Coda yes

Complex Coda no

The settings in (15) are clearly marked, given that two parameters are set to ‘yes’, including the complex onset parameter. The robust occurrence of this profile is unexpected under the traditional understanding of the syllable structure of Creole languages. However, Creole languages across the spectrum of European lexifier languages allow single codas and complex onsets resulting in six different syllable types. The most complex inventory is also straightforwardly accounted for in the parameter-based model. (16)

Inventory: V, CV, CVC, VC, CCV, CCVC, CVCC, VCC, CCVCC (Negerhollands, Unserdeutsch) Oblig. Onset no

Complex Onset yes

Coda yes

Complex Coda yes

This is the richest inventory in the sample, allowing nine syllable types. Complex codas are found together with all possibilities for the onset position. All parameters except the obligatory onset parameter are set to ‘yes’ to accomplish this. The parameter-based syllable typology model has been able to account for most of the Creole syllable type inventories straightforwardly. However, three inventories cannot be captured by this model. The ‘yes’ setting of the coda parameter and the complex onset parameter would have to be chosen to arrive at V, CV, CVC, VC, CCV in São Tomense. However, this predicts CCVC syllables as well, but there is no evidence for them in this language. Furthermore, the complex onset parameter and the complex coda parameter have to be switched on to generate CCV, CCVC and CVCC syllables in Mauritian and Haitian. However, these settings also incorrectly predict robust VCC and CCVCC syllables in these languages. In short, the problem with the parameter-based model is that the occurrence of complex onsets and codas in certain syllable types should license them across the board, whereas this is not observed in several languages. Given this shortcoming, future research needs to develop a theoretical model that can account for all attested syllable inventory types in the Creole languages of the world. It is beyond the scope of this paper to address this issue.

16

Diversity and complexity in Creole syllable structure The key concept in previous thinking about Creole syllable structure has been uniformity. It has been held that Creole languages do not allow many syllable types beyond CV, if any. Few Creole languages such as Atlantic French-lexified Creoles and Negerhollands have been recognized to license syllable codas routinely. These languages have been treated as special cases, however, rather than the norm. The present survey has produced quite the opposite result. The Creole languages of the world show considerable diversity in terms of their syllable structure, particularly as far as the inventories of syllable types are concerned. Only one parameter was found to be set uniformly: All Creole languages allow onsetless syllables. Naturally, Creole languages display universals of syllable structure that are found to be true in all natural languages: All Creole syllables must have a nucleus, all Creoles have CV syllables, and complex onsets and codas presuppose simpler ones. This should not deter from the fact that Creole syllables vary along several dimensions including edge effect, whether or not codas are allowed and whether or not complex onsets or complex codas are permitted. The wide range of syllable types employed in Creole languages may be illustrated through the list of syllable templates in (17). Syllable templates are understood in the present context as a descriptive short-hand device encoding the full set of syllable types in a given language. For example, the syllable template (C) (C) V (C) encapsulates the types V, CV, CVC, VC, CCV and CCVC. (17)

Syllable templates in Creoles (a) (b) (c) (d)

(C) V: (C) (C) V: (C) V (C): (C) (C) V (C):

(e)

(C) (C) V (C) (C):

Saramaccan Angolar, Kituba, Ndyuka, Sango Baba Malay, Nubi Berbice, Bislama, Ermitaño, Kristang, Palenquero, Papiamentu, Sri Lankan, St. Lucian, Tayo, Tok Pisin, Zamboangueño Negerhollands, Unserdeutsch

The syllable template notation enables a compact picture of the diversity of Creole syllable types. However, it is not able to capture all occurring Creole inventories correctly in this way. Complex onsets and single codas occur in São Tomense, but using the template (C) (C) V (C) would incorrectly imply that CCVC syllables occur in this language. Analogously, Mauritian and Haitian have complex onsets and complex codas, but the template (C) (C) V (C) (C) would incorrectly imply that VCC or CCVCC syllables are attested in these languages. The case for diversity, however, is not affected by this issue. Sizable syllable type inventories are found across lexifier languages. Codas and complex onsets may be found in Creoles lexified by any of the European languages, as the lists in (17) (d) and (17) (e) show. Recall that recent work by McWhorter (2001) has expounded the idea that Creole languages are in some significant sense typologically simple. The Creole simplicity hypothesis is that Creole languages are consistently simpler in structure than non-Creole languages and that Creole grammars are the world’s simplest grammars overall. The

17

preceding discussion has made clear that Creole languages occupy a wide range of syllable types similar to non-Creole languages. About the only inventory that was not found for Creole languages is an exclusive CV language such as Hua (see Blevins 1995). Given the results of the present study, we may test the simplicity hypothesis Creole-internally. If it is the case that Creoles are simple overall, then Creole grammars should cluster significantly around simple structures. In other words, there should not be sizeable groups of Creole languages attesting to structures of any notable degree of complexity. Most linguists would agree to consider (C) V, (C) (C) V and (C) V (C) syllable templates to be quite simple. In such languages, only open syllables are allowed or no clusters are permitted. Under this measure, Saramaccan, Angolar, Kituba, Ndyuka, Sango, Baba Malay and Nubi would be considered to have simple inventories of no more than four syllable types among the Creole languages in the present sample. Note, however, that these Creoles are a minority (7/23 = 30%). On the other hand, inventories allowing the full or nearly the full set of complex onsets and codas may be considered complex. Haitian, Mauritian, Negerhollands and Unserdeutsch fall into this category because they allow complex codas in addition to complex onsets, even though only Negerhollands and Unserdeutsch display the full set. This group is a small minority (4/23 = 17%). A middle group of languages allows complex onsets or complex codas, but not both. This comprises a slight majority of Creole languages, ranging from São Tomense to Zamboangueño in Table 5 (12/23 = 52%). I conclude that Creole syllable structure is not particularly simple, not is it highly complex in the majority of cases. Instead, it displays what may be called medium complexity, with the majority of Creole languages showing five, six or seven syllable types. It is unexpected the Creole simplicity hypothesis that the great majority of Creoles languages shows syllable type inventories of medium or higher complexity (5 to 9 syllable types) (16/23 = 70%). This outcome is not a great surprise, however, in light of recent detailed investigations into the syllabic and segmental structure of Creole languages. Plag & Schramm (forthcoming) have concluded that syllables in early Suriname and Caribbean Creoles are of medium complexity, allowing codas or complex onsets. Furthermore, Klein (forthcoming) has found that segmental inventories of Creole languages occupy the medium range identified for non-Creole languages in Maddieson’s (1984) typological study. Thus, the evidence from Creole languages converges on the idea that their phonologies exhibit a typical or medium degree of complexity. In terms of the parameter-based model, it is surprising to the Creole simplicity hypothesis that unmarked parameter settings are attested in only a minority of Creole languages. We may consider those parameter settings as unmarked in which no more than one parameter is set to ‘yes’. These Creoles show no more than four syllable types. If Creole languages in general tended towards unmarked, simple syllable inventories in their synchronic phonology, one would expect a majority of unmarked parameter settings. Given the diversity and complexity found in Creole syllables, it seems that syllable typology is not a defining linguistic characteristic of Creole languages. We may test this idea by investigating the syllable structure of Creole languages that have been argued to be prototypical and identifiable as a typological class: Ndyuka, Tok Pisin, Saramaccan, Haitian, St. Lucian, Mauritian, Fa d’Ambu, and Negerhollands (McWhorter 1998). All of these are in the present sample, with the exception of Fa d’Ambu. Judging

18

from Post (1995), however, the syllable structure of Fa d’Ambu is similar, if not identical to its neighboring relative Angolar. Thus, in the present context, Angolar takes the place of Fa d’Ambu in exploring if there is a correlate in the distribution of syllable types to arguably prototypical Creoles. (18)

Syllable type inventories in prototypical Creoles V, CV

V, CV, CCV

V, CV, CVC, VC, CCV, CCVC Saramaccan Angolar, Tok Pisin, Ndyuka St. Lucian

V, CV, CVC, VC, CCV, CCVC, CVCC Haitian, Mauritian

V, CV, CVC, VC, CCV, CCVC, CVCC, VCC, CCVCC Negerhollands

The display in (18) shows that the languages of the Creole prototype do not correspond to a uniform syllable structure. Instead, we find five inventories with considerable diversity of syllable types across these languages. Furthermore, only two of the five inventories may be considered simple. Whereas Saramaccan, Angolar and Ndyuka have only open syllables, the other languages have closed syllables with single or complex codas. Thus, it appears that there is no correlation between potentially prototypical Creole languages and uniformity or simplicity of syllable structure. Conclusion The results of the present investigation make necessary a fundamental rethinking of the syllable structure of Creole languages. Far from being uniform, Creole languages exhibit anywhere from two to nine syllable types in their inventories. Strong evidence has been mounted against the idea that Creole languages are limited to CV syllables in any significant sense. No Creole language is limited to CV syllables, whereas all Creoles have them as part of their inventory, just like any other natural language. There is a fair range of syllable types, with the most frequent template being close to or more complex than (C) (C) V (C). The simplest template is (C) V; the most complex template is (C) (C) V (C) (C). Given that acrolectal varieties and recent loans have been excluded from consideration in this investigation as much as possible, it seems clear that the more complex inventories are not due to decreolization. Instead, it appears that the phonology of the core vocabulary of basilectal and mesolectal Creole varieties encompasses a good degree of diversity and complexity in terms of syllable structure. However, decreolization and post-creole loans may result in even more complex syllable types if the language exerting the contact influence has them. The extensive literature on variable word-final consonant cluster simplification in English-lexified Creoles provides evidence for this because it shows that word-final complex codas are attested, at least in a sizeable minority of cases. Thus, the syllable structure of certain Creole languages is probably more complex than laid out in this investigation when acrolectal varieties and recent loans are considered, in particular where Germanic lexifier languages are involved. The two models of syllable typology employed in this investigation have been able to provide important guideposts for the analysis. However, it has become apparent

19

that the structure of Creole syllables provides challenges to both of them. The CVphonology model is not fine-grained enough to distinguish between the diverse Creole syllable inventories, in part because it relies on the presence versus absence of obligatory onsets as a defining characteristic. The present analysis has shown that this parameter is vacuous for Creole languages, given that none of them requires onsets. The parameterbased model has done a good job in accounting for most of the Creole syllable inventories, but does not capture all of them. The issue for future research is to devise a model where parameter settings need not necessarily apply across the board. The results of the present investigation provide strong evidence against the Creole simplicity hypothesis. The syllable types attested in Creoles may be classed as simple in only a minority of Creole languages. A distinct majority of Creoles, however, displays syllable structure of medium or higher complexity. This result corroborates recent evidence from other phonological investigations strongly supporting the hypothesis that the phonology of Creole languages is of medium or typical complexity. Given the present analysis, a number of tasks for future research have become apparent. The cause of the phonological restructuring of superstrate words needs to be revisited, now that it seems clear that CV syllables are significantly less dominant within Creole languages than previously thought. Furthermore, a theoretical explanation for the considerable diversity and complexity of syllable structure in the Creole languages of the world needs to be found and a viable model accounting for all Creole syllable inventories needs to be developed. In addition, the phonotactics of Creole languages, that is, the constraints on permissible sequences of sounds in Creole words and syllables need to be investigated in more detail. Furthermore, the hypothesis that Creole phonologies exhibit medium or typical complexity needs to be examined further, particularly in comparison to its competitor, the Creole simplicity hypothesis. Finally, it is hoped that this paper has contributed to showing that the phonology of Creole languages in general is an exciting and rewarding area of linguistic research.

20

Appendix The purpose of the appendix is to identify the sources used and to provide structured data exemplifying each of the syllable types that are attested in a given Creole language in the sample. More data are given in total than would have been possible in the main text. The Creole languages are presented in the order in which they appear in Table1. Glosses from sources written in languages other than English have been translated wherever possible. Where no glosses were given in the source, an effort has been made to retrieve the meanings of the items in question. This is not always been successful as indicated by the annotation ‘no gloss given’. Periods are provided to indicate syllable boundaries only where this is not immediately obvious. Negerhollands Sources: Sabino (1990, 1993) (Sa 90, 93), Stolz 1986 All data are from Stolz (1986), unless indicated otherwise. V /izu/

‘iron’

/ale:/

‘alone’

CV /fekete/

‘fight’

/bini/

‘inside’

CVC /pik/

‘pluck’

/gobed/

‘prayer’

VC /ef/

‘if’

/alma/

‘all’

CCV /bwa/

‘preserve’ (Sa 93)

/groma/

‘greedy’

CCVC /glof/ /frag/

‘believe’ (Sa90, 93) ‘ask’ (Sa 90)

/forfluk/

‘sly’

CVCC /tomp/ /mant/

‘stomp’ (Sa 93) ‘month’ (Sa 90)

/kerk/ /wort/

‘church’ ‘word’ (Sa 90)

VCC /aks/

‘axe’ (Sa 90)

/akt/

‘eight’ (Sa 90)

CCVCC /an.stons/

‘instantly’ (Sa 90)

/skerp/

‘sharpen’ (Sa 90)

21

Berbice Source: Kouwenberg (1994) V /atriti/ CV /toko/

‘reverse’

/oboko/

‘hen’

‘child’

/pili/

‘arrow’

CVC /purkaru/ /timbabos/

‘to fight’ /talma/ ‘type of worksong’

‘serve out’

VC /bu.in/

‘hide’

/ondro/

‘under’

CCV /speki/ /plk/

‘pork’ ‘place’

/waskolo/

‘why’

CCVC /furstan/ /glof/

‘understand’ /krombu/ ‘believe’ /klup/

‘bend’ ‘hit’

Ndyuka Source: Huttar & Huttar (1994) V /okoo/

‘okra’

/uma/

‘woman’

CV /nosu/

‘nose’

/sido/

‘to sit’

CCV /dyendee/ /kwaka/

‘splendid’ ‘cassava granules’

/alandya/ /agwado/

‘type of citrus fruit’ ‘stringed instrument’

Saramaccan Sources: Aceto (1996) (A), Johnson (1974) (J), Rountree (1972) (R) V /oto/

‘other’ (A)

/ahalala/

‘centipede’ (R) 22

/eside/

‘yesterday’ (J)

CV /si.nki.ni/ /koti/

‘skin’ (A) ‘cut’ (R)

/gba.mba/

‘meat’ (R)

Haitian Sources: Cadely (2002) (C), Nikiema (2000) (N), Valdman (1978, 1981) All data are from Valdman’s works, unless noted otherwise. V /emab/

‘amiable’

/ijen/

‘hygiene’

CV /tete/

‘bosom’ (C)

/pare/

‘prepare’

CVC /sik/

‘sugar’

/ps/

‘plaque’ (C)

VC /aksidan/

‘accident’ (N) /egzod/

‘exodus’ (N)

CCV /klu/ /megri/

‘nail’ /bra/ ‘to become thin’ (N)

‘arm’

CCVC /pwop/

‘clean’

/klas/

‘class’ (N)

/CVCC/ /kilt/ /filt/

‘cult’ ‘filter’

/fiks/ ‘set, firm’ /taks/ ‘tax’ St. Lucian

Source: Carrington (1984) V /ofe/

‘to offer’

/epi/

‘with, and’

CV /batõ/

‘stick’

/hote/

‘height’

23

CVC /dokte/ VC /i/

‘physician’

‘offspring; child’

/uval/

‘horse’

/ãvi.e/

‘curious’

CCV /plãte/ /ublie/

‘to plant’ ‘to oblige’

/bizwe/

‘to need’

CCVC /blag/

‘joke’

/vwel/

‘sail’

Angolar Sources: Lorenzino 1998, Maurer 1995 All data are from Maurer (1995). V /abi/

‘April’

/ome/ ‘human’

CV /biri/

‘to open’

/kega/ ‘to carry’

CCV /bwaru/ /uvwa/

‘good’ ‘nine’

/fyoko/ /vevya/

‘to hurry’ ‘to grow’

São Tomense Sources: Lorenzino (1998) (L), Valkhoff (1966) All data are from Valkhoff (1966), unless noted otherwise. V /ose/

‘sky’

/ami/

‘I’

CV /galufu/

‘fork’

/supeto/

‘clever’

CVC /palma/ /di.glasa/

‘palm tree’ /farkon/ ‘misfortune’ (L)

‘falcon’ (L)

24

/VC/ /alvuli/

‘tree’

CCV /bligasõ/ /kwelo/

‘obligation’ ‘rabbit’

/tlisa/ /kopla/

‘jaundice’ ‘to buy’

Papiamentu Main sources: Kouwenberg & Murray (1994; K & M), Munteanu (1996) Auxiliary source: Maurer (1998) Data are from Munteanu (1996) unless noted otherwise. V /iglesia/

‘church’

/uzu/

no gloss given

CV /suku/

‘sugar’ (K&M)

/zoya/

‘swing’ (K&M)

CVC /falsu/ /harta/ /berdat/

‘mean’ (K&M) ‘to glut’ ‘truth’

/nos/ /sigel/

‘our’ ‘age’

VC /altar/

‘altar’

/ermen/

no gloss given

CCV /skapa/

‘escape’ (K&M)

/subla/

‘blow’ (K&M)

CCVC /tres/

‘three’

/skirbi/

‘to correspond’

Palenquero Sources: Bickerton & Escalante (1970) (B&E), Lewis (1970) All data are from Lewis (1970), unless noted otherwise. V /abla/

‘speak’(B&E)

/uto/

no gloss given

CV /kusa/

‘thing’ (B&E)

/tampoko/

‘either’ (B&E)

25

CVC /tumba/ /tad.de/

no gloss given no gloss given

/lihto/

no gloss given

VC /entonse/ /olbia/

‘then’ (B&E) no gloss given

/ahma/

no gloss given

CCV /drumi/ /flako/

‘to sleep’ no gloss given

/negro/ /kumbla/

‘black’ ‘to buy’

CCVC /blanko/

no gloss given

/pwehko/

no gloss given

Unserdeutsch Source: Volker (1982) V /abn/

‘evening’

CV /ve/ /dasu/

‘way’ ‘to that’

/son/ /gsa/

‘sun’ ‘said’

CVC /ber/ /vlh/

‘mountain’ ‘which’

/bu/

‘book’

VC /is/

‘ist’

CCV /fry/

‘early’

/kwel/

‘source’

CCVC /gmyk/ /flansu/

‘decorated’ ‘plantation’

/gros/

‘large’

CVCC /finf/

‘five’

/hols/

‘wood’

26

VCC /as/

‘afraid’

/unt/

‘and’

CCVCC /traxt/

‘dress’

/sekraft/

‘ability to see’

Bislama Sources: Lynch (1975) (L), Tryon (1987), Meyerhoff (1993) (M) All data are from Tryon (1987), unless noted otherwise V /akis/

‘axe’

/eli/

‘early’

CV /holem/

‘to hold’

/tede/

‘today’

CVC /bokis/ /pipol/ /hasban/

‘box’ ‘people’ ‘husband’

/sotfala/ /naf/

‘short’ ‘enough’

VC /olketa/ /emti/

‘all’ (L) ‘empty’

/is/

‘east’

CCV /kopra/

‘copra’

/slo/

‘slow’ (M)

CCVC /swit/ /katres/ /smol/

‘sweet’ ‘cartridge’ ‘small’ (M)

/gris/ /brok/

‘fat’ ‘broken (adj.)’ (M)

Tok Pisin Sources: Smith (2002), Verhaar (1995) (V) All data are from Smith (2002), unless noted otherwise. V /abris/

‘go past’

/epa/

‘stingray’ (V)

27

CV /redi/

‘ready’

/pikinini/

‘child’

CVC /fis/ /aninit/

‘fish’ ‘underneath’

/abus/

‘animal food’

VC /ensin/ /alta/

‘engine’ ‘altar’ (V)

/asde/

‘yesterday’

CCV /tri/

‘tree, three’

/bekri/

‘bakery’ (V)

CCVC /fren/

‘friend’

/klos/

‘clothes’

Mauritian Sources: Baker (1972), Baker & Hookoomsing (1987) (B&H) All data are from Baker (1972), unless noted otherwise. V /ale/

‘to go’

/ena/

‘there’

CV /vini/

‘to come’

/fizi/

‘rifle’

CVC /palto/

‘jacket’

/latab/

‘table’

VC /alzeb/

‘algebra’ (B&H)

/apsoli/

‘absolut’ (B&H)

CCV /prekot/

‘near’

/lapli/

‘rain’

CCVC /deswit/ /pret/

‘immediately’ ‘priest’ (B&H)

/blag/

‘joke’ (B&H)

CVCC /rekolt/ /filt/

‘harvest’ (B&H) ‘filter’ (B&H)

/fiks/ ‘firm, stuck’ (B&H) /taks/ ‘tax’ (B&H)

28

Tayo Source: Ehrhart (1993) V /ale/

‘to go’

/ako/

‘again’

CV /pu/

‘for’

/tape/

‘to beat’

CVC /kat/

‘four’

/solda/

‘soldier’

VC /er/

‘hour’

/o:t/

‘disgrace’

CCV /pli/

‘plus’

/pukwa/

‘why’

CCVC /pret/

‘priest’

/labrus/

‘brousaille’

Sri Lankan Source: Smith (1977) V /ra/

‘miss’

/æ:tik/

‘tuberculosis’

CV /læ:ñ/

‘firewood’

/kupa/

‘stare’

CVC /korpu/

‘body’

/dews/

‘god’

VC /is.kamu/ /altu/

‘fish scale’ ‘tall, high’

/:n.d/

‘wave’

CCV /dre:tu/

‘correct’

/obrigadu/

‘thanks’

CCVC /kwæ:n.tru/

‘coriander’

/friz.mu/

‘dear’

29

Kristang Sources: Baxter (1988), Thurgood & Thurgood (1996) (T&T) All data are from Baxter (1988), unless noted otherwise. V /eli/

‘he, she, it’

/ati/

‘until’ (T&T)

CV /kaza/

‘house’

/dagu/

‘jaw’

CVC /forsa/ /mas/ /aros/

‘strength’ /fasel/ ‘more’ /muler/ ‘rice’ (T&T)

‘easy’ ‘wife’ (T&T)

VC /ombru/ /albi/

‘shoulder’ /uzdu/ ‘tree’ (T&T)

‘cuckold’

CCV /stiru/ /otru/

‘style’ /greza/ ‘other’ (T&T) /alegri/

‘church’ ‘happy’ (T&T)

CCVC /drentu/ /tras/

‘inside’ /tres/ ‘behind’ (T&T)

‘three’ (T&T)

Zamboangueño Sources: Forman (1972), Whinnom (1956) (W) All data are from Forman (1972), unless noted otherwise V /amo/ /a.ora/

‘boss’ ‘now’ (W)

/otro/

‘other’

CV /gana/

‘win’

/barato/

‘cheap’

CVC /soltero/

‘brother’

/kombersa/

‘speak’

VC /anda/

‘go’

/algun/

‘certain’

30

CCV /klase/

‘class’

/alegre/

‘happy’

CCVC /abril/

‘April’

/tyempo/

‘time’

Ermitaño Source: Whinnom (1956) V /ele/

no gloss given /ohos/

‘eyes’

CV /sumi/

no gloss given /boka/

‘mouth’

CVC /pulseras/

‘bracelets’

/bos/

‘voice’

VC /el/

‘the’

/urta/

no gloss given

CCV /klaa/

‘club’

/entro/

‘inside’

CCVC /pwelte/

no gloss given /etras/

‘after’

Kituba Sources: Fehderau (1962) (F), Mufwene (1997) (M), Swift & Zola (1963) (S&Z) V /i.nsi/

‘country, land’ (S&Z) /awa/

‘here’ (F)

CV /munoko/

‘mouth’ (M)

/dikulu/

‘leg, foot’ (M)

CGV /ku-vwanda/ /uvwa/

‘to sit down’ (S&Z) ‘nine’ (F)

/ku-dya/

‘to eat’ (S&Z)

31

Sango Sources: Pasch (1997) (P), Walker & Samarin (1997) All data are from Walker & Samarin (1997), unless noted otherwise V /ape/

no gloss given /to.a/

no gloss given

CV /mere.nge/ /sukula/

‘child’ /men/ ‘to wash’ (P)

‘certain’

CCV /nda.pre/

no gloss

‘doing’

/srango/ Nubi

Main sources: Heine (1982), Pasch & Thelwall (1987) Auxiliary source: Owens 1997 All data are from Pasch & Thelwall (1987). V /akulu/ /ila/

‘to eat’ ‘except’

/abat/

‘armpit’

CV /kidima/ /bara/

‘work’ ‘outside’

/mutSele/

‘rice’

CVC /marid/ /asab/ /wakti/

‘fever’ ‘vein, root’ ‘time’

/kuwes/ /sas/ /lib.ra/

‘good, nice’ ‘frying pan’ ‘needle’

VC /abdegi/ /alan/ /ag.rab/

‘butterfly’ ‘because’ ‘scorpion’

/agder/ /as.ma/

‘to be able’ ‘name’

Baba Malay Sources: Lim (1981) (L), Pakir (1986) (P)

32

V /ula/

‘snake’ (L)

/anak/

‘baby’ (L)

CV /budak/

‘child’ (L)

/tahu/

‘to know’ (P)

CVC /tingal/ /basat/

‘to stay’ (L) /nasi/ ‘bed-bug’ (P) /cakap/

‘rice’ (P) no gloss given (L)

VC /erti/ /a.os/

no gloss given (L) /a-pau/ ‘thirsty’ (P)

no gloss given (L)

33

References Aceto, M. (1996). Early Saramaccan syllable structure: An analysis of complex onsets from Schumann’s 1778 manuscript. Journal of Pidgin and Creole Languages 11 (1), 23-44. Alleyne, M. (1980). Comparative Afro-American. Ann Arbor: Karoma. Baker, P. (1972). Kreol. A description of Mauritian Creole. London: Hurst. Baker, P. & V. Hookoomsing (1987). Dictionary of Mauritian Creole. Morisyen – English – Français. Paris: L’Harmattan. Bartens, A. (1995): Die iberoromanisch-basierten Kreolsprachen. Frankfurt a. M.: Peter Lang. Baxter, A. (1988). A grammar of Kristang (Malacca Creole Portuguese). Pacific Linguistics (B-95). Canberra: Australian National University. Bickerton, D. & A. Escalante (1970). Palenquero: A Spanish-based Creole of Northern Colombia. Lingua 24, 254-267. Blevins, J. (1995): The syllable in phonological theory. In J. Goldsmith (Ed.), The handbook of phonological theory (pp. 206-244). Oxford: Blackwell. Boretzky, N. (1983). Kreolsprachen, Substrate und Sprachwandel. Wiesbaden: Harrassowitz. Cadely, J.-R. (2002). Le statut des voyelles nasales en Créole haïtien. Lingua 112, 435464. Carrington, L. (1984). St. Lucian Creole. Hamburg: Buske. Clements, G. N. (1990). The role of the sonority cycle in core syllabification. In J. Kingston & M. Beckman (Eds.), Papers in laboratory phonology I. Between the grammar and physics of speech (pp. 283-333). Cambridge: Cambridge University Press. Clements, G. N. & S. J. Keyser (1983). CV phonology: A generative phonology of the syllable. Cambridge, MA: MIT Press. Ehrhart, S. (1993). Le créole français de St-Louis (le tayo) en Nouvelle-Calédonie. Hamburg: Buske. Fehderau, H. (1962). Descriptive grammar of the Kituba language. A dialectal survey. Léopoldville: American Mennonite Brethren Board of Missions. Forman, M. (1972). Zamboangueño texts with grammatical analysis. A study of Phillipine Creole Spanish. Unpublished doctoral dissertation, Cornell University. Giegerich, H. (1992). English phonology. An introduction. Cambridge: Cambridge University Press. Heine, B. (1982). The Nubi language of Kibera – An Arabic Creole. Berlin: Reimer. Holm, J. (2000). An introduction to pidgins and creoles. Cambridge: Cambridge University Press. Huttar, G. & M. Huttar (1994). Ndyuka. London: Routledge. Johnson, M. (1974). Two morpheme structure rules in an English proto-creole. In D. DeCamp & I. Hancock (Eds.), Pidgins and Creoles: Current trends and prospects (pp. 118-129). Washington, D.C.: Georgetown University Press. Kahn, D. (1976). Syllable-based generalizations in English phonology. Unpublished doctoral disstertaion, MIT. Distributed by Indiana University Linguistics Club. Kaye, A. & M. Tosco (2001). Pidgin and Creole languages. München: Lincom Europa.

34

Kenstowicz, M. (1994). Phonology in generative grammar. Oxford: Blackwell. Klein, T. B., forthcoming. Creole phonology typology and complexity: Evidence from phoneme inventory size, vowel quality and stop consonants. In P. Bhatt & I. Plag (Eds.), The structure of Creole words: Segmental, syllabic and morphological aspects. Tübingen: Niemeyer. Kouwenberg, S. (1994). A grammar of Berbice Dutch Creole. Berlin: Mouton de Gruyter. Kouwenberg, S. & E. Murray (1994). Papiamentu. München: Lincom Europa. Lewis, A. (1970). A descriptive analysis of the Palenquero dialect. Unpublished M.A. dissertation. University of the West Indies, Mona. Lim, S. (1981). Baba Malay: The language of the ‘straits-born’ Chinese. Unpublished B.A. (Hons.) dissertation. Monash University. Lorenzino, G. (1998). The Angolar Creole Portuguese of São Tomé: Its grammar and sociolinguistic history. München: Lincom Europa. Lynch, J. (1975). Bislama phonology and grammar: A review article. Kivung 8 (2), 186204. Maddieson, I. (1984). Patterns of sounds. Cambridge: Cambridge University Press. Maurer, P. (1995). L’angolar: Un créole afro-portugais parlé à São Tomé. Hamburg: Buske. Maurer, P. (1998). El papiamentu de Curazao. In M. Perl & A. Schwegler (Eds.), América negra. Panorámica actual de los estudios lingüísticos sobre variedades hispanicas, portuguesas y criollas (pp. 139-218). Frankfurt a. M.: Vervuert. McWhorter, J. (1998). Identifying the creole prototype: Vindicating a typological class. Language 74 (4), 788-818. McWhorter, J. (2000). The missing Spanish Creoles. Recovering the birth of plantation contact languages. Berkeley: University of California Press. McWhorter, J. (2001). The world's simplest grammars are creole grammars. Linguistic Typology 5, 125-166. Meyerhoff, M. (2003). Reduplication in Bislama. An overview of phonological and semantic factors. In S. Kouwenberg (Ed.), Twice as meaningful. Reduplication in Pidgins, Creoles and other contact languages. (pp. 231-238). London: Battlebridge. Mondesir, J. (1992). Dictionary of St. Lucian Creole. Berlin: Mouton de Gruyter. Mufwene, S. (1997). Kitúba. In S. Thomason (Ed.), Contact languages. A wider perspective (pp. 173-208). Amsterdam: Benjamins. Mühlhäusler, P. (1997): Pidgin and Creole linguistics (2nd ed.). London: University of Westminster Press. Munteanu, D. (1996). El papiamento, lengua criolla hispánica. Madrid: Gredos. Nikiema, E. (2000). Lexical and epenthetic initial vowels in Haitian Creole. Journal of Pidgin and Creole Languages 15 (1), 171-177. Owens, J. (1997). Arabic-based pidgins and creoles. In S. Thomason (Ed.), Contact languages. A wider perspective (pp. 125-172). Amsterdam: Benjamins. Pakir, A. (1986). A linguistic investigation of Baba Malay. Unpublished Ph.D. dissertation. University of Hawai’i. Pasch, H. (1997). Sango. In S. Thomason (Ed.), Contact languages. A wider perspective (pp. 209-270). Amsterdam: Benjamins.

35

Pasch, H. & R. Thelwall (1987). Losses and innovations in Nubi. In P. Maurer & T. Stolz (Eds.), Varia creolica. Bochum: Brockmeyer. Parkvall, M. (2000). Out of Africa. African influences in Atlantic Creoles. London: Battlebridge. Plag, I. & M. Schramm, forthcoming. Early Creole syllable structure: A cross-linguistic survey of the earliest attested varieties of Saramaccan, Sranan, St. Kitts and Jamaican. In P. Bhatt & I. Plag (Eds.), The structure of Creole words: Segmental, syllabic and morphological aspects. Tübingen: Niemeyer. Post, M. (1995). Fa d’Ambu. In J. Arends, P. Muysken & N. Smith (Eds.), Pidgins and Creoles. An introduction (pp. 191-204). Amsterdam: Benjamins. Romaine, S. (1988). Pidgin and Creole languages. London: Longman. Rountree, S. C. (1972). The phonological structure of stems in Saramaccan. In J. E. Grimes (Ed.), Languages of the Guianas (pp. 22-27). Norman, Oklahoma: Summer Institute of Linguistics. Sabino, R. (1990). Towards a phonology of Negerhollands: An analysis of phonological variation. Unpublished doctoral dissertation, University of Pennsylvania. Sabino, R. (1993). On onsets: Explaining Negerhollands initial clusters. In F. Byrne & J. Holm (Eds.), Atlantic meets Pacific. A global view of pidginization and creolization (pp. 37-44). Amsterdam: Benjamins. Sebba, M. (1997): Contact languages. Pidgins and Creoles. London: Macmillan. Smith, G. (2002). Growing up with Tok Pisin. Contact, creolization and change in Papua New Guinea’s national language. London: Battlebridge. Smith, I. R. (1977). Sri Lanka Creole Portuguese phonology. Unpublished doctoral dissertation, Cornell University. Stolz, T. (1986). Gibt es das kreolische Sprachwandelmodell? Vergleichende Grammatik des Negerholländischen. Frankfurt a. M.: Peter Lang. Swift, L. B. & E. W. A. Zola (1963). Kituba basic course. Washington, D.C.: Foreign Service Institute. Thurgood, E., & G. Thurgood (1996). Aspect, tense, or Aktionsart? The particle ja in Kristang (Malaacca Creole Portuguese). Journal of Pidgin and Creole Languages 11(1), 45-70. Tryon, D. (1987). Bislama. In introduction to the national language of Vanuatu. Pacific Linguistics (D-72). Canberra: Australian National University. Valdman, A. (1978). Le créole: structure, statut et origine. Paris: Klinksieck. Valdman. A. (1981). Haitian Creole – English –French dictionary. Bloomington, IN: Indiana University, Creole Institue. Valkhoff, M. (1966). Studies in Portuguese and Creole. Johannesburg: Witwatersrand University Press. Verhaar, J. (1995): Toward a reference grammar of Tok Pisin: An experiment in corpus linguistics. Honolulu: University of Hawai’i Press. Volker, C. (1982): An introduction to Rabaul Creole German (Unserdeutsch). Unpublished M.A. dissertation. University of Queensland. Walker, J. & W. Samarin (1997). Sango phonology. In A. Kaye (Ed.), Phonologies of Asia and Africa (including the Caucasus) vol. 2 (pp. 861-880). Winona Lake, Indiana: Eisenbrauns.

36

Whinnom, K. (1956). Spanish contact vernaculars in the Philippine islands. Hong Kong University Press.

37