Atlas of Pidgin and Creole Language Structures Online

1. Introduction ⇫ ¶

Berbice Dutch is the former creole vernacular of the plantation populations of the Berbice and Canje Rivers; these jointly constituted the privately-owned Berbice colony (now part of Guyana, South America). It is the last known Dutch-lexifier creole to have gone extinct, Skepi Dutch and Negerhollands (see van Sluijs) having gone before. Berbice Dutch was documented in its pre-language death form by Robertson (e.g. 1979) and Kouwenberg (e.g. 1994, 2007a).

2. Sociohistorical background ⇫ ¶

Berbice was established as a private colony in 1627 by Abraham van Pere, with permission of the Zeeland chamber of the Dutch West India Company. Van Pere dispatched a party consisting of sixty whites and six blacks to establish the colony (Robertson 1993: 298; Postma 1990: 13). Nearly a hundred years later, in 1720, ownership of the colony was transferred to a new joint stock company, the “Sociëteit van Berbice” (the Berbice Association). Smith (2000 [1962]) claims that, in 1732, there were ninety-three private plantations on the Berbice River and twenty in the Canje Creek. Almost 30 years later, in 1749, when the colony consisted of 111 plantations, the black slave population had not quite reached 1,500 (Kramer 1991: 60). According to Smith (2000 [1962]), by 1762, the population of the Berbice colony had reached 346 whites, 3,833 black slaves, and 244 Amerindian slaves. It is clear that the white population was outnumbered by blacks at least at the start of the eighteenth century, and possibly much earlier.¹

     The tax records which form the basis for these assessments also show that the Dutch distinguished between black slaves and Amerindian slaves. Tax records survive from around the middle of the 18th century.² By that time, Amerindian slaves constitute around 11% of the total slave population; that category of slaves disappears entirely from the records during the early 1760s. Although it is likely that Amerindian slaves constituted a larger proportion of the slave population earlier, it is not known whether they were ever a majority. However, it is important to note that a free population of Arawak Amerindians were a constant presence on and around the plantations throughout the history of the Berbice colony.

     Nothing is known from the historical record of the provenance of slaves in Berbice for the entire seventeenth century – the crucial period for the genesis of Berbice Dutch. Nonetheless, as first noted by Smith et al. (1987), the linguistic evidence points clearly to a significant presence of speakers of Eastern Ijo. Postma (1990: 106f) provides support for the idea that the Berbice colony received slaves from the Bight of Biafra, although he is unable to identify more precisely the source of these slaves. We will return to the relevance of the presumed Eastern Ijo presence below, in the discussion of linguistic features in §§4–9.

     Dutch control of the colony was punctuated by brief periods of French and British ownership. The colony was finally ceded to the British in 1814; with Demerara and Essequibo, Berbice became part of British Guiana, which became independent Guyana in 1966. In all likelihood, the demise of Berbice Dutch began during the eighteenth century, when plantation activity shifted downriver, and was consolidated during the nineteenth century, when plantation activity moved from the river banks to coastal locations, facilitated by a new drainage system which the Dutch were building along the coast. The redistribution of plantation locations, combined with the change to British ownership and the importation of new slaves from Barbados, facilitated the rapid development of an English-lexifier creole known as Creolese (or “Guyana Creole English”, see Devonish & Thompson). The population which remained in the former Dutch plantation area in upriver Berbice was largely of two kinds: a population of Arawak Amerindians who lived in ethnically and linguistically homogeneous communities, and a population of persons of mixed decent, known locally as “Bovianders” – a word thought to derive from Dutch bovenlanders, meaning “upriver dwellers” – who lived scattered along the river and its tributaries. It is this latter population which continued to speak Berbice Dutch for at least another century; individual or even widespread bilingualism between Arawak and Berbice Dutch is likely to have existed during this time. Eventually, Berbice Dutch speakers were convinced that they were better served speaking English-lexifier Creolese – a decision which was helped along by the efforts of missionaries who, from around the mid-nineteenth century, had a presence in Berbice and introduced English-language education there. Thus it is that the last living speakers of Berbice Dutch, located by Ian Robertson in the mid-1970s and by myself in the late 1980s, were bilingual in Berbice Dutch and Creolese if not clearly dominant in the latter. All modern work on Berbice Dutch documents this language in a pre-language death state, therefore. Unfortunately, only four short sentences and a wordlist of 44 words are known from early sources. (See Robertson (1994) for the wordlist, and Kouwenberg (1996a) for discussion).

3. Sociolinguistic situation ⇫ ¶

Although the stigma which had come to be associated with Berbice Dutch meant that several of the small number of surviving speakers refused to be interviewed, I was able to collect sufficient material for a grammar to be compiled (Kouwenberg 1994). It should be noted that the persons whose speech was recorded did not form a community of speakers: They were individuals living in scattered locations along the Berbice River and its tributaries. Therefore, the late twentieth-century recordings cannot be taken to be representative of the state of the language during the time that it functioned as a community language. The sections which follow are largely based on Kouwenberg (1994) and subsequent publications.

4. Phonology ⇫ ¶

4.1 Vowels

The Berbice Dutch vowel system is essentially a simple 5-vowel system (i-e-a-o-u), with two complications: an e-ɛ opposition which looks like a remnant of an erstwhile harmony system, and significant allophony of /u/, which overlaps with /o/ to some extent (see Table 1).

The Eastern Ijo substrate uses an ATR harmony system, which pairs each [+ATR] vowel (i, e, u, o) with a [-ATR] counterpart (ị, ẹ, ụ, ọ, orthographically marked by underdot), with the exception of the low vowel (a). The word and any proclitics form a domain for vowel harmony; within that domain, all vowels are either +ATR or –ATR. This means, for instance, that clitic pronouns appear as either +ATR or –ATR, depending on the specification of the host.

     The Berbice Dutch vowel system does not reflect the harmonic system of its substrate. As seen in Table 1, vowels are not paired as they are in Eastern Ijo. Nevertheless, the distribution of the mid-front vowels suggests that a very limited form of vowel harmony may have survived. Specifically, no minimal pairs exist which contrast the mid-front vowels, and their distribution is largely predictable: We find /ɛ/ as final vowel, and in polysyllabic forms containing several occurrences of this vowel in successive syllables; we find /e/ elsewhere. Examples are provided in (1). The adaptations seen in a few forms of either English or Creolese origins, and the existence of alternate forms of some words (in 2), provides further evidence for the restrictions on the distribution of the mid-front vowels³:

(1)    Dutch-derived:               plɛkɛ ‘place’ vs. pleʃiri ‘enjoy oneself, have fun’

         Eastern Ijo-derived:      potɛ ‘old’ vs. duei ‘spirit (of the dead)’

         Arawak-derived:           kurɛmɛ ‘grass louse, mite’ vs. baren ‘all right’

         Creolese-derived:           dɛnɛ ‘then’ vs. redi ‘get ready’

(2)    Dutch-derived:               frufɛlɛ ~ frufeli ‘tire, get tired of’

         Eastern Ijo-derived:      bebia ~ biɛbiɛ ‘yellow, red, light-coloured’

However, exceptions do occur, mainly among Arawak- and Creolese-derived forms:

(3)    Eastern Ijo-derived:   jɛrma ‘woman’ (presumably via *jɛrɛma)

         mixed:                     jɛn(da) ‘be (there), exist’ (possibly < Eastern Ijo emi ‘exist’, Dutch daar ‘there’)

         Arawak-derived:    ite ‘palm tree (sp.)’, kuse ‘ant (sp.)’, etc.

         Creolese-derived:   stɛdi ‘steady’, ɛni ‘any’, etc.

Significant allophony is found in the realization of the high back vowel: /u/ is unstable and can surface as [u] or [o]. Thus, we encounter:

(4)    Dutch-derived:       bum ‘tree’: [bum], [bom]

         Eastern Ijo-derived:   nunu ‘pull’: [nunu], [nono]

         Arawak-derived:    turu ‘palm (sp.)’: [turu], [toro]

As a result, the distinction between /u/ and /o/ is sometimes lost. For instance, turu ‘palm (sp.)’ and toro ‘face, eye’ may both surface as [toro]. Note that quite a few occurrences of Berbice Dutch /u/ derive from a mid-back vowel in the etymological source.

     Finally, it is worth noting that the front rounded vowels of Dutch correspond either to an unrounded counterpart in Berbice Dutch, or to a back rounded vowel:

(5)    Dutch-derived:       ʃiri ‘sour’ (Dutch zuur /zyːr/)

                                          spuku ‘vomit’ (Dutch spugen /spyxə/)

                                          loi ‘lazy’ (Dutch lui /løy/)

                                          kuiti ‘calf (of leg)’ (Dutch kuit /køyt/)

4.2 Consonants

The Berbice Dutch consonant system is unremarkable, consisting of plosives, fricatives, nasals, liquids and approximants at the expected places (see Table 2).

Complex consonants in the lexifiers failed to survive in Berbice Dutch (Smith 2008). Thus, Eastern Ijo implosives (orthographically marked by underdots) and coarticulated stops correspond to plosives of the same place and voice in Berbice Dutch; the examples provided here are drawn from the Kalaḅarị variety of Eastern Ijo. Similarly, Dutch velar fricatives correspond to velar plosives of the same voice, and voiced fricatives correspond to their voiceless counterparts:

(6)    a.  Eastern Ijo-derived: pɛmbɛ ‘hatch (of egg)’ (Kalabari kpẹmgbẹ)

                                                  baku ‘beat’ (Kalabari ḅákụ́)

                                                  duei ‘spirit (of deceased)’ (Kalabari ḍụ́ẹ́ị́n)

         b.  Dutch-derived:         skapu ‘sheep’ (Dutch schaap /sxap/)

                                                  gutu ‘thing’ (Dutch goed /ɣut/)

                                                  flutu ‘flood’ (Dutch vloed /vlut/)

/s/ before /i/ is realized as [ʃ] without exception, as seen in (7a), but not all occurrences of [ʃ] are restricted to this environment; hence, phonemic status is assigned to it where it occurs in other environments, as illustrated in (7b).

(7)  a.    Eastern Ijo-derived:           ʃima ‘move’

              Dutch-derived:       meʃi ‘knife’

              Arawak-derived:    ʃiberu ‘frog (sp.)’

       b.    Dutch-derived:       ʃepu ‘soap’

              Arawak-derived:    ʃukuli [affectionate term of address used with children]

4.3 Syllable structure and stress

Although many Dutch-derived forms display final added vowels, Berbice Dutch is not an open-syllable language: Word-internal syllables are, more often than not, closed syllables (e.g. bok.tu ‘bend (in river)’). Moreover, onset clusters of up to three consonants are possible. Onset clusters of two consonants take the form obstruent + liquid/semivowel (e.g. froto ‘rot’) and s + any consonant except voiced obstruents (e.g. slem ‘clever, smart’); of 3 consonants s + voiceless plosive + liquid/semivowel (e.g. skrifu ‘write’).

     Most simplex words with final syllables of the form CVN have final stress (e.g. sna.ˈpan ‘gun’); where the final syllable is open, the default placement of main stress in simplex words is penultimate (e.g. hul.ˈwa.tri ‘wave’). A number of trisyllabic words display irregular initial stress (e.g. ˈa.la.la ‘tongue’). Suffixation, which has the effect of adding a syllable, is sometimes accompanied by reanalysis resulting in stress shift to the “new” penultimate syllable (e.g. o.ˈbo.ka.pu ~ o.bo.ˈka.pu < oboko-apu [chicken-pl]).

5. Noun phrase ⇫ ¶

5.1 The nominal projection

Articles are prenominal (8a), whereas demonstratives are postnominal and require the presence of the definite article (8b); all are Dutch-derived. Use of the definite article di implies that the referent can be identified in the preceding discourse. The indefinite article occurs only with count nouns, as can be expected from the fact that it derives from the numeral en ‘one’; it is actually ambiguous in many of its occurrences. Generic-reference nouns usually occur without an article (8c). Adjectival modifiers are prenominal (8d), relative clauses postnominal (see §8). The noun is bold-printed in each case:

(8)    a.      en                 buku      – di         buku    –               biaka        – di biaka

                  one/INDF    book            DEF    book         corn                DEF corn

                  ‘one book/a book’           ‘the book’            ‘(some) corn’    ‘the corn’

         b.      di         tafn          di       –      di         titi     dida

                  DEF    afternoon DEM         DEF    time     DEM

                  ‘this afternoon’                     ‘that time’

         c.      o       no-ko           mja      kɛnɛ  fi       boro    danga  ka

                  3SG  NEG-RES   make   person         for       pass    there   NEG

‘He (= God) did not intend for people to pass there’ [‘there’ refers to a dangerous place]

         d.      di      andri   weki      –  en       boko        jɛrma

                  DEF other   week            INDF Arawak  woman

                  ‘the next week’                ‘an Arawak woman’

The overt definite article di is used to signal that unique reference can be established in the discourse. Additionally, a null definite determiner exists. Its function is to signal that unique reference can be established outside the discourse, through knowledge of the world, as in (9a), knowledge of the community or immediate situation which establishes pragmatic uniqueness, as in (9b), or knowledge of the virtual situation set up by the discourse, as in (9c) (Kouwenberg 2007b). Texts are replete with definite-reference bare nouns of these types.

Possession is marked either by juxtaposition of possessor and possessed, or by a dislocated possessive construction, which uses the possessive pronoun ʃi:

(10)           di          jɛrma     papa             di         potɛ-man   ʃi            toro

                  DEF     woman  father           DEF    old-man    3POSS   eye

                  ‘the woman’s father’            ‘the father’s eye’

5.2 Personal pronouns

Pronouns are invariant for case, with the exception of the 3SG possessive form ʃi. The variation in 1PL and 3PL forms is geographic: Berbice River speakers use the first, Wiruni Creek speakers the second form. The restricted distribution of the short 3SG pronoun o, which is unacceptable in dislocated, focused and conjoined use, further suggests that it is a clitic, whereas the long form ori, which occurs in all those uses, is an independent pronoun. Short o also cannot be modified, in contrast with all other pronouns, which allow for reflexive use (pronoun + selfu ‘self’), and which accept the nominalizing enclitic jɛ or the emphatic marker di; I should note though that it is not always clear what these markers contribute to the interpretation of the pronoun.

Examples follow (emphatic use in (11a), reflexive use in (11b), dislocated use in (11c), nominalization in (11d)); see Kouwenberg (1994: 172ff) for more complete discussion. Note that the context of the utterance (11a) does not suggest an emphatic ‘my wife’; nor does the presence of jɛ in (11d) appear to make any difference.

The 3rd person forms o, ori and eni/ini are derived from Eastern Ijo pronouns, whereas the 1st and 2nd person forms and the possessive pronoun have Dutch sources. The nominalizer jɛ and emphatic marker di are of Eastern Ijo and Dutch source, respectively. The Eastern Ijo-source forms have gone through some amount of semantic bleaching: o and ori both derive from male forms, but are ungendered in Berbice Dutch; nominalizer jɛ derives from an Eastern Ijo form which has singular nonhuman reference, whereas it is a generic nominalizer in Berbice Dutch. It is worth noting that the Berbice Dutch plural suffix -apu (§9) derives from an Eastern Ijo human plural nominalizer, and has lost both its human reference, and its independent status.

6. Verb phrase ⇫ ¶

Tense-mood-aspect is marked as shown in Table 4.

While the Perfective and Imperfective Aspect markers have the same form and range of functions as their Eastern Ijo etyma, other Eastern Ijo markers were not inherited. Nor does the Berbice Dutch TMA system reflect the overall system of distinctions and restrictions of Eastern Ijo. Thus, where Berbice Dutch has a Past Tense marker, Eastern Ijo marks only Aspectual and Modal distinctions and does not mark Tense in any way. It may be noted, furthermore, that Berbice Dutch allows combinations of TMA markers, whereas Eastern Ijo does not. Finally, only a special form of the Mood marker is admissible in negative contexts in Eastern Ijo, whereas Berbice Dutch excludes only the perfective suffix from negative contexts. (See Kouwenberg 2009 for fuller discussion.)

     Of note is the fact that the discourse status of utterances (partly) determines Tense marking: Background is Tense-marked, foreground is not. An example, drawn from the Berbice Dutch version of “Jack and the beanstalk”, follows:

(12)  ʤak     wa       da        en        loi     toko,    en     toko     alen     di       jɛrma    habu,    

         Jack    PST    COP    INDF  lazy  child,   one   child    alone   DEF  woman have

         wɛl,      ʃi            papa       wa       da        riki   man, [...]

         well,    3POSS   father     PST   COP    rich  man  […]

         ʃi            mama        tiri-tɛ         o          nau     fu        furkopu 

         3POSS   mother      send-PFV 3SG     now     for       sell

         di         laʃti     kui       wati        eni       habu,  […]

         DEF    last      cow     COMP   3PL     have    […]

         toko     bi      wɛl    a          moi,     [...]   pringi-tɛ    di      kui    ʃi            atri      bofu,

         child    say   well  3SG     good    […]  jump-PFV         DEF cow        3POSS back on

         wɛl       o          rei-a          di         kui       mu-a         makit      [...]

         well     3SG     ride-IPFV DEF    cow     go-IPFV   market   […]

‘Jack was a lazy child; the woman had just one child. Well, his father had been a rich man. [It is clear from the context that the father is not alive anymore, and that Jack and his mother are impoverished.] His mother then sent him to sell the last cow that they had. (The) child said “well, all right”. He jumped on the cow’s back. He was riding the cow to the market ...’

     Standard negation is marked clause-finally, as in the examples below. Negative verbs kanti ‘cannot’ and furi ‘not be there, not exist’ normally appear with the standard negator; this is true also of the preverbal negative resultative marker noko (no-ko [NEG-RES]), which appears in (14). Note that negation scopes both over the main clause and the embedded clause there. Kouwenberg (2000) provides an overview of negation in Berbice Dutch, while Kouwenberg (2009) discusses its relation to negation in the Eastern Ijo substrate.

7. Simple sentences ⇫ ¶

Basic word order is subject – verb – object (SVO) – different from that of both contributing languages: Eastern Ijo is predominantly SOV, whereas Dutch combines SOV order in subordinate clauses with verb-second order in matrix clauses, resulting in main clause word orders such as SAuxOV, XVSO, etc. But rigid SVO order does not necessarily translate to consistent left-headedness: Berbice Dutch uses postpositions as well as some prepositions (see §9), and has clause-final negation (§6).

     Non-verbal predicates are treated as follows: Nominal predicates are introduced by copula da (see 15a), adverbial predicates by jɛn(da) ‘be there, exist’ (see 15b) or its negative counterpart furi ~ furda (da is a dummy predicate < Dutch daar ‘there’), and adjectival predicates appear without a copula (see 15c). (See Kouwenberg 1996b for arguments that such forms are, indeed, adjectives.)

Constituent focus moves an argument or adjunct to initial position, yielding a contrastive effect. The focused phrase is optionally introduced by the equative copula da; its right edge is optionally marked for focus by so~sa (< Eastern Ijo). The following example shows that a focused constituent may be preceded by a topic:

Sentential focus, which asserts the truth of the proposition, is marked by copula da. Because da necessarily precedes a subject, contextual knowledge is required to disambiguate between subject focus and sentential focus:

Verb focus or predicate cleft similarly makes optional use of da and/or so~sa; the focused verb appears with both these markers in (18); note that blɛndɛ is the verb ‘to blind, to become blind’. A nominalized copy of the verb, again optionally accompanied by markers, may be used; it appears without markers in (19). Temporal clefts employ a nominalized verb heading a relative clause, as in (20).

     The formation of polar questions is marked only by rising intonation. Information questions are formed by movement of a wh-phrase to an initial position; like other focused constituents, the wh-phrase is optionally marked by da and/or so~sa:

8. Complex sentences ⇫ ¶

Finite complement clauses are optionally introduced by dati; such complements are selected by propositional verbs, including psychological verbs, perception verbs when followed by an indirect perception complement, and utterance verbs. Most of these can, instead, appear as first verb in a complementizer serial verb construction, where bi(fi) ‘say’ introduces the complement clause. Occasional use of dati following bi(fi) is also attested, as in (22). Direct perception complements take the form of bare clauses, as in (23).

     Headed relative clauses are most commonly introduced by wati, as in (24), where long-distance relativization has applied. Although wati is identical to the wh-word wati ‘what’, its semantic and distributional behaviour suggest that relativizer wati is a complementizer rather than a relative pronoun (Kouwenberg 1994: 370ff). Deletion of wati is possible only where a subject or direct object is relativized, as in (25).

     Serial verb constructions (SVC) of different kinds may be distinguished: Argument-introducing deki ‘take’ appears as first verb in an instrumental SVC, pi ‘give’ as last verb in a dative SVC (cf. (26)), twa ‘put’ as last verb in a locative SVC, and bi(fi) ‘say’ as last verb in a complementizer SVC (see (22) above). Additionally, modifying constructions may be distinguished: Fama ‘finish’ appears as last verb in a completive SVC, mu ‘go’ and kumu ‘come’ as last verb in a directional SVC (cf. (27)), and sequences of verbs which denote an event and a result of that event appear in the action-result SVC (cf. (28)).

As may be seen in (27) and (28), it is possible for both verbs in the SVC to appear with an aspectual suffix, here Perfective. This is impossible only in the completive SVC and in the complementizer SVC, where fama and bi(fi), respectively, do not accept inflection.

9. Other topics ⇫ ¶

9.1 Morphology

Here, I will only illustrate plural formation, nominalization, and reduplication. For a more complete discussion of Berbice Dutch morphology, including compounding and category conversion, see Kouwenberg (1994, 2003).

     The plural marker -apu is suffixed on the noun, as illustrated here for the bare noun bitapu (29), and in the preceding for definite di bitapu in (23). (30) illustrates the use of the enclitic nominalizer -jɛ.

     Reduplication is mainly used to mark iconic interpretations, although frequently with secondary connotations: iteration in verbs denoting events, often implying aimlessness, sometimes distribution; plural number in nouns denoting objects, always implying distributed occurrences of the objects; intensive in adjectives denoting gradable attributes, emphasis in adjectives denoting nongradable attributes. Thus:

9.2 Prepositions and postpositions

One of the striking characteristics of Berbice Dutch is its use of (mainly Dutch-derived) forms as both locational nouns and locative or directional postpositions. For instance:

Dutch has a fair number of prepositions, and makes limited use of directional postpositions; Eastern Ijo, on the other hand, uses nominal postpositions throughout. Although the Berbice Dutch use of postpositions resembles that of Eastern Ijo, it should be noted that the only Eastern Ijo-derived postposition in Berbice Dutch is anga (a generic locative/directional postposition), which is not a postposition in the source language.

     Several Berbice Dutch prepositions introduce clauses: Fan ‘from’, foro and foroteki ‘before’ introduce temporal adverbials; sondro ‘without’ may also introduce propositions; compare:

Extraction of a nominal complement, resulting in preposition stranding, is allowed by most prepositions as an alternative to pied-piping. This is not true for postpositions, which resemble possessive structures in many ways. More detailed discussion can be found in Kouwenberg (1994).

9.3 Passive formation

Passive formation in Berbice Dutch involves the suppression of an Agent and the promotion of a Patient to subject position without any overt marking: no passive auxiliary, passive morphology, or by-phrase. Passives are infrequent in spontaneous speech, and of those few that I attested, most were used in generic statements, denoting customary activity, as in (36). A few more denoted resultatives, as in (37). See Kouwenberg (1994) for details.