Contact-induced changes in Amerindian languages of French Guiana

Contact-induced changes in Amerindian languages of French Guiana Françoise Rose and Odile Renault-Lescure 1. Introduction French Guiana is an ideal place for the study of language contact, with its six Amerindian languages, various French based and English based Creoles, immigrant languages like Chinese or Hmong and, what is of interest for this volume, Romance languages like French, Portuguese and Spanish (Launey et al. 2003). This paper will focus on contact between on the one hand, two Amerindian languages, Kali’na and Emérillon, and on the other hand Romance languages: French, Portuguese and Spanish. This contact will regularly be compared with the contact of the same Amerindian languages with Creole languages. Theoretically speaking, this case study is particularly interesting in that it deals with on the one hand two typologically similar languages (the Amerindian languages Kali’na and Emérillon) and on the other hand languages that are typologically distant from each other (Romance languages and Creoles) and likewise distant from Kali’na and Emérillon. Section 2 will give as preliminaries a presentation of the Kali’na and Emérillon languages, with a short history of their contact situations. Section 3 and 4 will respectively deal with the phonological and morphosyntactic contact-induced changes in both languages. 2. Kali’na, Emérillon and their contact situations Kali’na1 is the Cariban language which covers the largest geographical area, spreading from the north-eastern savannahs of Venezuela to the north of the Brazilian state of Amapa, through all three Guyanas. In French Guiana, the Kali’na population (around 3500 people, but a lower number of speakers) is spread out into different villages to the west of Cayenne, as far as the Maroni river (Cf. map). Because of its coastal distribution, Kali’na has been in contact with a variety of other populations. 2 Erreur ! Style non défini. Emérillon is the northermost member of the geographically widespread Tupi-Guarani family, present in Guiana since the late 15th century. The 400 members of the group (whose autodenomination is Teko), are all fluent speakers of the language, and live exclusively in Guiana, in two areas of the rainforest, one next to the border of Surinam, and the other on the Brazilian border (Cf. Map 1). Because of its peripheral and more isolated situation, Emérillon has looser contacts with other non Amerindian populations. Map 1: Map of regional languages of French Guiana (Goury 2001) Erreur ! Style non défini. 3 These two languages display a different history of contacts. This contact history is basically constituted of three phases: (a) First, was the arrival of the Europeans and the merchandise trade. The Kali’na people, living on the Guianese coast, referred to contact goods with the words of the first Europeans they met, namely Spanish and Portuguese, and less commonly English, Dutch and French. The Kali’na lexicon took on a stock of borrowings which were then diffused all along the coast of the Guyanas, from the mouth of the Orinoco to the Approuague, in eastern Guiana. It was mainly diffused by means of a Carib based pidgin named “langue générale galibi” used between the different Amerindian populations, among them, the Emérillon population, who was meanwhile located in the hinterland, between the Inini, the Approuague and the Oyapock rivers. This first phase corresponds to direct but occasional contacts for Kali’na people (called “casual contacts” by Thomason 2001: 70), such as the use of interpreters with traders or missionaries. For the Emérillon people, those contacts were indirect, through the use of the Galibi Pidgin. (b) In a second phase, contacts with Romance languages decreased, being taken over by contacts with vehicular languages developed in the colonies with the development of slave trade. Those languages are: – Sranan Tongo, the Creole of the plantations of Surinam, born in the second half of the 17th century. Its lexicon is essentially based on English, with some contribution from Dutch, Portuguese and Kikongo. It has been hypothesized that a great number of its grammatical structures is based on those of the African Gbe languages. – The Guianese Creole, a French based Creole, with possibly some grammatical structures of the Fon language. A certain degree of bilingualism of the Kali’na speakers with Sranan lasted until the end of the 20th century, due to their history as refugees in the Dutch colony in the 17th and 19th century, and to strong commercial relations across the border. Meanwhile, the Emérillon population has just been in contact with the English based Creole Aluku in the late 18th century in the mid-Maroni region. The Kali’na population established intense social relations with the Guianese Creoles later on, as they went back to French Guiana and some of them very likely became bilingual. The Emérillon people also established some commercial contacts with Creole populations, but stayed isolated in the southern part of the colony. 4 Erreur ! Style non défini. (c) In a third phase, Guiana underwent “francization”, the unification of its administrative system as a French department, and the settlement of French institutions such as administration representations and schools (The first Kali’na children went to school in 1945, and the first Emérillon children in 1956). As a consequence, contact with French became more intense, especially for the Kali’na people, with French tending to substitute Creole as a vehicular language nowadays. Contacts with Creoles decrease, and speakers attitudes towards those languages change too. Mastering French is more or less seen as a key for social success (for work, studies, and implication in the political and administrative structures). Today, bilingualism with French is more widespread for Kali’na speakers than for Emérillon speakers. It is worthwhile to note that nowadays, those language contacts take place in a context of wider plurilingualism. Migrations of the late 20th century triggered a rise of Surinamese Creoles near the western border of the department and of Brazilian Portuguese near its eastern border. Table 1. Kali’na and Emérillon histories of contacts Historical times Main contact languages Type of contact for Kali’na (Kal.) Type of contact for Emérillon (Em.) 1. first contacts with Romance languages 2. rise of the Creoles Spanish (Sp.) Portuguese (Port.) direct and occasional indirect Creoles: Sranan Tongo (Sr.) Guianese Creole (Cr.) French (Fr.) some bilingualism little contact intense (widespread bilingualism) quite intense 3. “francization” To summarize, both Kali’na and Emérillon have successively had contacts with European languages, Creoles and finally French, each period being characterized by a stronger intensity of contacts. In each period, Emérillon contact situation is somewhat less intense than that of Kali’na. On the whole, contact-induced changes in Kali’na and Emérillon consist essentially of lexical borrowings, regardless of whether the source language is a Romance language or a Creole. A few syntactic changes will be presented further on. However, since lexical borrowing may gradually lead to phonological and structural changes in the borrowing language, we will Erreur ! Style non défini. 5 describe both how borrowings get adapted to the receiving systems, and how the systems adjust to the borrowings. Section 3 will present the integration of the borrowings at the phonological level, and section 4 at the morphosyntactic level. Our main interest will be to compare the integration of borrowings from different languages, in two different but comparable languages. Let us add three caveats. First, it is not always a simple task when studying a particular phenomenon of language interference to determinate whether code-switching or borrowing is concerned. As a consequence, this paper is based only on linguistic facts that are unambiguously borrowings (for a discussion of code-switching, Cf. Auer 1999). Second, the source language is not always easy to determine, especially within the following pairs of possible source languages: French and Guianese Creole, Guianese and French West Indies Creoles, and also Sranan Tongo and Aluku, especially when the word is quasi identical in both of the possible source languages. Third, although beyond the scope of this paper, it is important to keep in mind that interferences among Amerindian languages are also attested. For instance, Emérillon has borrowed a Cariban plural marker -kom, possibly from Kali’na. 3. Similarities and differences in the phonological integration of borrowings On the phonological level, besides phone substitution that we will not describe in this paper, many other processes of adaptation of the borrowed items can be found. In 3.1, we will focus on one very specific process: nasalization/denasalization in Emérillon. However, borrowings do not always completely adapt to the system1, and eventually it is sometimes the system itself that adjusts to the borrowed words and therefore undergoes remarkable changes. In 3.2, we will show how the Kali’na phonological system evolved in a substantial way on account of lexical borrowing. Eventually, and this point is particularly interesting from a theoretical standpoint, we will show in 3.3 how the same phonological constraint 1 The orthography of the source language was maintained for items that have not adapted at all to the phonological system of Kali'na, language for which there is a standardized orthography. 6 Erreur ! Style non défini. yields two different processes of integration for words of the same origin in Kali’na and Emérillon. 3.1. A specific integration process: (de)nasalization in Emérillon One of the peculiarities of Emérillon is its suprasegmental nasality. The /~/ feature is assigned at the lexical level and applies across a given morpheme to specify the [nasal] value of its phonemes. Only vowels and voiced consonants can be specified as [nasal]. Other phonemes are transparent and opaque to nasalization: they are not affected by nasalization and do not block its spreading. Table 2. Examples of oral and nasal morphemes in Emérillon Oral morphemes ba/e ~ mba/e thing tapˆdZ| house o-bo-aku he heats it 3-CAUS-hot Nasal morphemes mã/e) COMPL tãm¯ grandfather o-mõ-ãtã he hardens it 3-CAUS-hard Foreign items follow this constraint, and are therefore integrated as either oral or nasal morphemes, probably according to the nasal or oral value of the last phoneme, since nasality seems to apply from right to left. Accordingly, phenomena of denasalization and nasalization are observed in the borrowing process: Table 3. Examples of nasalisation in the borrowing process denasalization Source language MaCocotte [makokt] (Fr.) Mon Père (Fr.) [mõp] pommade (Fr.) ~ pomad (Cr.) [pomad] Emérillon bakokol [bakokt] bopeɾ [bopt] pobaɾ [pombat]2 Meaning pot priest gel Table 4. Examples of denasalisation in the borrowing process nasalization Source language dipen (Cr.) [dipe] farine (Fr.) ~ farin (Cr.) [fain] zoranj (Cr.) [zoa] Emérillon nĩpe panin Meaning bread flour zona orange (fruit) Erreur ! Style non défini. 3.2. 7 Contact-induced system-altering changes in Kali’na The first borrowings from the Romance languages into Kali’na at the time of colonization do not seem to have induced any structural change on Kali’na’s phonological system. This is not true, however, for the subsequent borrowings from Sranan Tongo. These borrowings induced tendencies towards certain phonological changes that were later reinforced by more recent borrowings from French and Guyanese French based Creole. Below are presented the introduction of a new phoneme and the transfer of a voice opposition. 3.2.1. Introduction of a new phoneme The Kali’na phonological system displays eleven consonants classified in Table 5 according to manner and place of articulation. Table 5. Kali’na consonant system Labial Apical p (f) t s m w n Palatal Velar Glottal k / h OBSTRUENTS stops fricatives SONORANTS nasals glides liquid y l In the first historical phase of contact, among other regular phoneme substitutions in borrowed words, /p/ regularly substituted for /f/ in borrowings, by virtue of being the only native obstruent at the same place of articulation (Renault-Lescure 1985). This is the case regardless of the source language. (1) (2) francês swafroe (Port.) (Sr. ) > palansi (Kal. ) > suwapulu (Kal.) ‘Frenchman’ ‘matches’ Later, however, variable realizations were tolerated, such as: (3) fensre (Sr.) > pesele ~ fensele (Kal.)’window’ After this period of instability, the regular replacement of /f/ by a /p/ stop in borrowings was brought to an end, the fricative sound being finally maintained in borrowings. 8 Erreur ! Style non défini. (4) frigi (Sr.) > filiki (Kal.) ‘kite’ This tendency was reinforced afterwards through borrowings from French and French based Creole. (5) (6) (7) (8) lafinèt suflèt fil foto (Cr.) (Cr.) (Cr. ~ Fr. ) (Cr. ~ Fr.) > lafinet[ˆ] > suflet[ˆ] > fil[ˆ] > foto (Kal.) (Kal. ) (Kal.) (Kal.) ‘window’ ‘whistle’ ‘sewing thread’ ‘photo’ Our hypothesis, in keeping with Weinreich ([1953]1970: 18), is that the empty fricative slot in the labial consonant inventory of Kali’na was a structural factor favoring the introduction of a new phoneme filling a gap in the system. 3.2.2. Transfer of a voice opposition Although there is a the lack of opposition between p/b, t/d, and k/g in Kali’na, the stops /p/, /t/ and /k/ are sometimes realized as voiced stops [b], [d] and [g] when they are not word initial. This voicing of stops is difficult to explain. A study in progress (Renault-Lescure & Gomez 2005) shows a link between some of these realizations and the prosodic and syllabic structures, but does not yet explain all of these realizations. Among the hypotheses, a possible explanation relies on contact with languages in which the voicing opposition is relevant for stops, more specifically through lexical borrowings. With the introduction of loanwords maintaining a voiced stop word-initially, a new opposition is indeed emerging. The oldest loanwords retain the pattern of allophonic distribution, by which voiced stops are borrowed as voiceless word-initially. (9) bandera (10) barque (11) grasi (Sp.) (Fr.) (Sr.) > pantila [pandi’la] > paliki [paali’gi] > kalasi [kalaa’çi] (Kal.) ‘flag’ (Kal.) ‘bark’ (Kal.) ‘glass’ More recent loanwords, from the Creoles or from French, maintain a voiced realization word-initially, which has led to the introduction of a new opposition: (12) (13) pali panki ‘barrage’ ‘skirt’ ≠ bali ≠ banki ‘barrel’ ‘bank (seat)’ (< Sr.) ( < Sr.) Erreur ! Style non défini. 9 It is worth noting that Cariban languages do not usually display a voicing opposition, but that in certain of those languages, its emergence has been recorded and presented as a likely consequence of contact (Gildea 1998). 3.3. Differences in the processing of consonant clusters in borrowings Both Kali’na and Emérillon share a phonological constraint that restricts consonant clusters. Emérillon syllables are all open, with the exception of final syllables which may be closed by a single consonant. Consequently, no consonant cluster within the domain of the morpheme is allowed, and morphophonemic rules extend this domain to the word level. Table 6. Canonical syllabic pattern of the Emérillon word (C)V- …- (C)V - (C)V(C) In Kali’na, the syllabic structure is (C)V1(V2,C). Accordingly, consonant clusters are possible, but their number is restricted by the particular distribution of consonants. All eleven consonants, with the exception of the glottal stop, can occur in the onset position. The coda consonants are either nasals (wordinternally and word-finally) or the glottal stop (word-internally only). To summarize, the only sequences of consonants that are possible are word internal, C1 being necessarily a nasal or a glottal stop, and C2 being any consonant but the glottal stop. Now words borrowed from Romance languages often contain consonant clusters that are not allowed by the two recipient languages of our study. However, these clusters get integrated differently into the Kali’na system and into that of Emérillon. 3.3.1. In Kali’na Consonant clusters that violate the syllable constraints are readjusted by the insertion of a vowel between two consonants (in bold in Table 7). These processes are observed regardless of the source language. The second, third and fifth lines of Table 7 also show the insertion of a final vowel. Table 7. Examples of vowel insertion to break up a consonant cluster Consonant clusters pl åpVl fr åpVl Source language Kali’na Meaning plata (Sp.) francês (Port.) pˆlata palansi[si] money Frenchman 10 Erreur ! Style non défini. br åpVl sc åsVc sp åsVp kl åkVl brande-wijn (Dutch) biscuit (Fr.) spoen (Sr.) lakle (Cr.) palantuwini pisukuwi pipunu lakele rhum crackers spoon key The quality of epenthetic vowels is determined by progressive or regressive assimilation, or is by default a vowel prone to devoice (i, ˆ). These rules do not apply to the most recent borrowings (from Creole or French) that conserve consonant clusters. (14) garden (15) dilwil (16) taxi (Sr.) (Cr.) (Fr.) > > > kalden dilwil taxi [taksi] (Kal.) ‘mosquito-net’ (Kal.) ‘oil’ (Kal.) ‘taxi’ 3.3.2. In Emérillon Two processes occur with borrowed words to maintain the syllable constraints: either consonant deletion simplifies the cluster (as in Table 8) or vowel epenthesis breaks up the consonant cluster (as in Table 9). By and large, we can posit that deletion takes place when the first consonant of the cluster is a liquid3 (and possibly also when the cluster is in final position), and vowel epenthesis takes place between any other two consonants. Table 8. Simplification of consonant clusters Simplification of consonant clusters through deletion rm å m rt å t ld å d final position final position Source language Emérillon Meaning gendarme [ãdam] (Fr.) marteau [mato] (Fr.) ~ marto (Cr.) soldat [slda] (Fr.) ~ sòlda, soda (Cr.) (la) piste (Fr.) (la) table (Fr.) ~ tab (Cr.) ãdam bato policeman hammer soda soldier lapis latab airstrip table Table 9. Examples of vowel insertion to break up a cluster of consonants Erreur ! Style non défini. tr å ton Source language rt å let citron (Fr.) ~ sitron (Cr.) citrouille (Fr.) travail (Fr.) ~ travay (Cr.) biscuit (Fr.) ~ biskwi (Cr.) carta (Sp.) rk å lak arcabuz (Sp.) tr å tul tr å tal sk å sik Emérillon 11 Meaning sitono lemon situɾu⇔ taɾawa⇔ pumpkin work bisiku⇔ crackers kaɾeta paper, book, notebook gun aɾakapusa The quality of the epenthetic vowel is usually determined by assimilation to the quality of the following vowel. In certain cases (like kaɾeta and aɾakapusa), an epenthetic vowel is found in cases where the cluster starts with a liquid, but these words were borrowed indirectly from Spanish via Kali’na. In conclusion, the comparison between Kali’na and Emérillon shows that two languages with a similar constraint on consonant clusters may react differently to accommodate borrowed items. We can however hypothesize that Kali’na does not use the simplification process because it has a tendency to accept (and sometimes even favor) polysyllabic stems and words. Please note that as far as phonology is concerned, the processes induced by interference of Kali’na and Emérillon with other languages are similar regardless of the phonological system of the source language, whether it is a French or English based Creole or a Romance language. 4. Similarities and differences in the morphosyntactic integration of borrowings Once again, our data is particularly enlightening in that both Amerindian languages are typologically comparable, while they clearly differ in structure from French and Creoles, those latter being themselves typologically distant. Many linguistic constraints on contact-induced changes are indeed based on typological similarity and distance. Kali’na and Emérillon are typologically very similar: both tend to be agglutinating and polysynthetic. Accordingly, the predicate is necessarily 12 Erreur ! Style non défini. made up of a root and a person marker, but can also take numerous prefixes and suffixes, as well as clitics and an incorporated noun. Relations between the arguments and the predicate are marked on the predicate with a person index, following a hierarchical system (Renault-Lescure 2002, Rose 2002 & 2003a), and full NPs are facultative and actually rare in discourse. The syntactic importance of the predicate and the richness of its morphology contrast clearly contrast with their French or Creole counterparts. In this respect, Kali’na and Emérillon are clearly different from the fusional Romance languages and the isolating Creoles. French verbs are characterized by personal clitics and a fusional conjugation system. Creole verbs are very poor morphologically. On one side of the debate on linguistic constraints on interference (Cf. for instance Thomason and Kaufman 1988, chapter 2) are beliefs like Field’s Principles of System Compatibility, which asserts that the borrowing language’s morphological typology (as isolating, agglutinating or fusional) will constrain the possibility of borrowing from another language (Field 2002). In our case, Kali’na and Emérillon, being agglutinating languages, can borrow agglutinating but not fusional morphology. There are indeed no instances of borrowed morphology from Romance languages, which are fusional languages. As a consequence, only roots are borrowed. Another of Field’s assertions is that all languages can borrow instances of isolating morphology (Field 2002). The isolated morphemes of Creoles are indeed easily borrowed into Kali’na and Emérillon, whereas the Romance items that are borrowed are either simple roots, or roots with morphology (i.e NPs, or part of NPs) reanalyzed as simple roots. In 4.1, we will first describe the process through which borrowed items are ascribed to a category in both receiving languages, before showing how they integrate in the morphosyntax of their host systems in 4.2. Finally, section 4.3 will suggest a few syntactic changes induced by borrowings. 4.1. Class assignment In this section, we will focus on the attribution of a word category to borrowings. We will limit the discussion to the four predicative categories (nouns, verbs, adjectives and adverbs) Hengeveld (1992, chapter 4) uses to classify languages according to their parts of speech systems. Whereas the Romance languages distinguish clearly these four categories, respectively nouns, verbs, adjectives and adverbs, the Creole lan- Erreur ! Style non défini. 13 guages offer examples of multifunctionality (Bruyn 2002). Some words function as noun, verb or adjective without any change in word class being morphologically marked. This ability seems to be quite common in Sranan Tongo, especially with forms used as a verb and as a noun, and less common in Guianese Creole. Kali’na differentiates only three major word classes: nouns, verbs, and adverbs (playing the role of qualifiers once nominalized). All these words switch easily from one class to the other through derivational processes. Emérillon presents four classes of roots, verbs, nouns, adverbs, and a small class of roots with “descriptive/adjectival” meaning that cannot be analyzed as nouns, verbs, nor adjectives (Cf. Rose 2003b). 4.1.1. In Kali’na We shall now examine the various strategies found in Kali'na for assigning a word category to borrowings. Borrowings without category change are illustrated in Table 10. Table 10. Borrowings without category changes Category Source language in the source language Noun zapato (Sp.) kerki (Sr.) dilèt (Cr.) auto (Fr.) Noun perro (Sp.) poespoesi (Sr.) Adverb pannantan (Cr.) tijou (Cr.) exceptionnellement (Fr.) Category Kali’na in Kali’na noun (1) noun (2) adverb sapato keleke diletˆ oto pelo pusipusi pannantan tijou exceptionnellement Meaning shoes church milk car dog cat meanwhile always exceptionally Borrowed nouns, regardless of their source language and the period of borrowing, are integrated into the class of Kali’na nouns. The vast majority of them fall into the sub-class of alienable possessed nouns – category (1) of Table 10 – while only a few fall into the sub-class of non-possessed nouns which have a suppletive form in the possessive construction – category (2) of Table 10 : 14 Erreur ! Style non défini. O-lapal (17) paila ‘bow’, alakaposa ‘gun’ (< Sp.) (18) kulewako ‘parrot’ pelo ‘dog’ (< Sp.) y-ek (19) nimoku ‘hammock’ lit (< Cr.) ‘bed’ a-pat ‘my weapon, my bow, my gun’ ‘my pet, my parrot, my dog, my cat’ ‘your sleeping place, your hammock, your bed’ There are no instances of words being transferred into the sub-class of inalienably possessed nouns (kin terms, body parts, parts of a whole). Borrowed adverbs integrate into the category of Kali’na adverbs. Borrowings with category changes are now illustrated. Tables 11 to 13 show examples of items classified as nouns, verbs, or adjectives in the source languages and that are reanalyzed as nouns in Kali’na. These nouns fall into 3 sub-classes that are specific to borrowed items and show a defective behavior. Table 11. Borrowings of verb/nouns bases as nouns Category in Sranan Source language Tongo Verb/noun begi (Sr.) Category in Kali’na Kali’na Meaning noun (3) begi- prayer Table 11 shows that borrowings to the Sranan Tongo verb/nom category fall in a specific sub-class of nouns in the recipient language – nouns (3) – and are always treated as uninflected nominal stems to which a verbalizing suffix is attached to form transitive verbs, carrying a person prefix and a tense suffix : (20) tamusi si-begi-ma-e God 1A-prayer-VERB-PRES ‘I pray God’ on the model of a regular verbalizing process in Kali’na: (21) Kali’na (Courtz 1997) kasili s-aiku-ma-e beer 1A-liquid-VERB-PRES ‘I make cassava beer’ Erreur ! Style non défini. 15 This strategy involves a creative adaptation process and conforms fully to Kali’na patterns of derivation. A similar result of adaptation is mentioned for Japanese by Loveday 1996 (cited in Winford 2003: 50). Thus Kali’na borrowed the Sranan verb/noun forms as nouns, but conserved their verbal meaning by using them in a verbalizing construction. Why is the Sranan verb, unspecified for valency, always used in a transitive construction? All the authors converge in the idea that morphological adaptation and class assignment may be hindered when the recipient language has complex rules. It is the case with the verbal system of Kali’na that displays a sub-system of split intransitivity. One may think that this strategy permits to avoid the assignment of the borrowed verb to one of the verbal intransitive sub-classes, “active” or “stative”. Table 12 illustrates items that were verbs in the source language (the invariable verbal form from Guyanese Creole and the infinitive form from French) and that, once borrowed in Kali’na, are reanalysed as nouns. The form a new sub-class of nouns – nouns (4), which are characterized by the absence of flexion and their necessary integration into a postpositional group, where the postposition is always poko ‘busy with’. Table 12. Borrowings of verbs as nouns Category in the source language Verb Source language Category in Kali’na Kali’na pentiré (Cr.) pentiré nétwayé (Cr.) ~ noun (4) nétwayé ~ nettoyer (Fr.) nettoyer comprendre (Fr.) comprendre Meaning painting cleaning understanding The postpositional group functions in a single-participant copular construction (22) or in a two-participant construction with the verb l ‘to put’ (23). (22) pentiré poko man paint busy.with 3S.COP.PRES ‘He is painting.’ (23) woto nettoyer poko s--ya fish cleaning busy.with 1A-put-PRES. ‘I am cleaning the fish.’ (lit. ‘I am putting it out for cleaning.’) Table 13 shows adjectives from Guyanese Creole and French, or multifunctional forms from Sranan, that are reanalyzed as nominal roots, constituting sub-class (5). Table 13. Borrowings of adjectives or verb/nouns as nouns 16 Erreur ! Style non défini. Category in the source language Verb/noun Adjective Source language pina (Sr.) pur (Fr.) Category in Kali’na noun (5) Kali’na Meaning pina pur (be) miserable pure person These nominal roots are invariable and take a predicative suffix to be used as a predicate in a very productive structure with a copula. (24) pur-me pure-PRED ‘He is pure.’ man 3.COP.PRES On the Kali’na model : (25) pitani-me child-PRED He’s a child. man 3.COP.PRES To summarize, multifunctional forms, verbs and adjectives undergo a category change when borrowed into Kali’na. They are systematically reanalyzed as nouns and therefore, in order to be used as predicate, need to be embedded in predicative structures. It is noteworthy that those nouns form sub-classes characterized by a defective behavior. In line with Wichmann and Wohlgemuth’s typology of loan verb embedding patterns, two strategies can be identified: the strategy of indirect insertion with affixation of a verbalizer as in example (20), and the light verb strategy as in (22), (23) and (24). This propension to favor the borrowing of multifunctional items supports Thomason’s claim that “less tightly structures features are easier to borrow than features that fit into tightly integrated closed structures” (Thomason 2001: 69). 4.1.2. In Emérillon In contrast, most of the words borrowed from Romance languages into Emérillon are assigned to the equivalent word class. The great majority of those borrowings are nouns, with only one verb and one adverb borrowed as such. Table 14. Examples of borrowing with no category change Erreur ! Style non défini. Word category Noun Source language camisa (Sp.) Verb/noun travailler (Fr.) ~ travay (Cr.) vite (Fr.) Adverb Emérillon kamita 17 taɾawa⇔ Meaning fabric, traditional skirt (piece of fabric tied to the waist) to work/work wiɾ [wit] quickly The verb taɾawaʤ ‘to work’ originates either in the French verb (or noun) root or in the Creole verb. In Emérillon, it is used with the usual verb morphology (person prefix, tense…). (26) si-taɾawa⇔-taɾ. 1INCL-work-FUT We will work. On the other hand, some borrowings are the result of a reanalysis of several morphemes as one root. The most obvious process is the fusion of the French article (or part of it) with the noun it determines. Most of such cases may be indirect borrowings through Creole, and then the fusion probably took place in the borrowing process from French into Creole. Table 15. Fusion of several morphemes into one Emérillon root article+noun part of the article +noun French l’école [lekol] orange (PL: des oranges [dezoã]) Creole lekol zoranj Emérillon lekol zona Meaning school orange (fruit) Although rare, the most interesting case concerns the few roots that are borrowed in a different category from the one they belonged to in the source language. Field (2002: 44) defines the Principle of Reanalysis. As far as equivalence of word class is concerned, semantic characteristics appear to be more relevant than word classes themselves. Thus it is possible that a form belonging to one semantic sub-type in a language may need to be reanalyzed as belonging more properly to a corresponding semantic subtype that belongs to an entirely different word class in the recipient language. This is the case for Romance adjectives borrowed in Emérillon. Emérillon possesses two classes of words with “adjectival meaning”. One expresses qualities concerning human beings, such as physical sensations or mental phenomena, and is morphosyntactically a sub-class of nouns. The other expresses properties usually applied to objects, like size, color, value, 18 Erreur ! Style non défini. and consists of descriptive roots that can be analyzed neither as nouns, verbs, nor adjectives. Our data show five cases of borrowing of French adjectives. Interestingly, they seem to fall into the two categories described above (i.e. the subclass of nouns expressing human qualities, and the class of descriptive roots expressing object qualities) on semantic grounds. Four of them are reanalyzed as nouns (i ‘richness’, en ‘youth’, alu ‘jealousy’ and tadisjonel ‘tradition’) and refer mainly to human qualities, only one enters the descriptive root category, and is definitely an object quality (sale ‘salted’). As Field proposed, this reanalysis seems to be due to the semantic sub-type of the items (Field 2002). Therefore, examples of category change include four examples of French adjectives borrowed as nouns and also one example of a French verb borrowed as an Emérillon noun (Table 16). The French verb toucher ‘to touch’ is used as an Emérillon noun to refer to “social allowances”, on account of the collocation of both items in the French phrase toucher les allocations’ ‘to receive social allowances’. Table 16. Examples of borrowings with a category change Form in French toucher [tue] Category in French verb traditionnel [tadisjonl] adjective Meaning in French to touch (to receive, for allowances) traditional jeune [œn] adjective young Form in Emérillon zo-tue Category in Emérillon possessed noun nõdetadisionl i-en possessed noun possessed noun Meaning the allowances our traditions his-youth ~ he is young Manifestly, whereas French or Creole nouns are always borrowed as facultatively possessed nouns, French verbs and adjectives seem always to be borrowed as obligatorily possessed nouns. We hypothesize that this fact is tightly linked with the high predicatibility of possessed nouns in Emérillon. In fact, every noun with a personal prefix can constitute a possessive predicate, as illustrated in (27). Nouns expressing qualities are most often used with that function (28). (27) e-men 1SG-husband ‘I have a husband.’ Erreur ! Style non défini. 19 (28) e-kaneõ 1SG-fatigue ‘I am tired.’ As a consequence, French adjectives borrowed as obligatorily possessed nouns maintain the possibility to be used as predicates, one of their main functions. (29) ie-i-en 4 RED -3-youth ‘They are young.’ Both Emérillon and Kali’na data seem to confirm Field’s Hierarchies of Borrowability according to which nouns are more easily borrowed than adjectives and verbs (Field 2002). This is observable both in terms of the quantity of items borrowed in each category and in terms of the need for a morphological reanalysis to allow the integration of the borrowed item. Now focusing more specifically on verb borrowing, Weinreich ([1953] 1970) asserted that verbs are hard to borrow. A usual explanation lies in the fact that inflected forms are harder to borrow. Our data illustrate perfectly this issue, since verbs of both Amerindian languages and French take obligatory flexion. The precise nature of verbal borrowings is very enlightening: from Sranan, Kali’na borrowed verb/nouns, from Creole, verbs that are anyway uninflected in the source language, and from French, some noninflected infinitive verbal forms. We have already seen that all of them are turned into nouns in Kali’na. Now the only French (or Creole) verb that Emérillon borrowed as a verb is actually also a noun in both possible source languages. It is in fact also borrowed as a noun in the recipient language: travail ‘work’ / travailler ‘to work’ gave rise to taɾawa⇔ ‘work, to work’. (30) taɾawa⇔-am za-bae-okaɾ work-TRANSF INDET-make-CAUS ‘It would provide our sons some work.’ zo-aɾ-a-pe INDET-son-a-for In brief, most verbs that are borrowed in Kali’na and Emerillon display some nominal properties, and all of them are actually borrowed under a noun form (with taɾawa⇔ also displaying a verbal form). 20 Erreur ! Style non défini. 4.2. Full grammatical integration Now that we have described how borrowings get ascribed to a word category, let us show how, in Kali’na (Renault-Lescure, 2005) and Emérillon, borrowed words become fully integrated into the morphosyntactic system of the language. On the one hand, they become compatible with the very rich morphology of their receiving language, and on the other hand, they assume syntactic roles just like those of the other members of the categories they now belong to, with the exception of sub-class (4) and (5) of Kali'na nouns, the defective behavior of which has been described in 4.1.1. The series of examples (31) to (36) show how borrowed nouns and verbs in Emérillon are compatible with the complete morphological apparatus of the language : possessive affixes (31), plural (31), demonstrative (32), derivative suffix (33), second position discourse particles (34), and also with the reduplication process (35), person indexes and tense markers on verbs (36). (31) o-iu o-sapato-kom 3COREF-shoe-PL 3SG-put ‘He puts on his shoes.’ (32) dati a magas COP DEM store ‘This store wasn't here.’ o-mõduɾ-o (33) aɾakapusa-uhu gun-big 3-send-PL Big guns were sending bullets. (34) mama-ne eɾe-⇔ ika Mom-CONTRAST 2SG-kill ‘It is Mom that you killed!’ (35) ie-i-en RED-3-youth ‘They are young.’ (36) si-taɾawa⇔-taɾ 1INCL-work-FUT ‘We will work.’ < zapato (Sp.) < magasin (Fr.) ~ magazen (Cr.) baɾ bullets < arcabuz (Sp. via Kal.) <maman (Fr.) < jeune (Fr.) ~ jenn (Cr.) < travailler (Fr.) ~ travay (Cr.) The following examples show how borrowed nouns are integrated in the Emérillon syntax. They can be modified by a numeral (37) or a descriptive root (38). Erreur ! Style non défini. (37) aipo bap ɾ ãdam now three policeman ‘Now there are three policemen.’ (38) s big besin-a-mae o-iɾuɾ basin-a-REL 21 < gendarme (Fr.) ~jandanm (Cr.) < bésin (Cr.) 3-bring ‘He brought back a big basin.’ A borrowed noun can be subject (39), object (40), object of postposition (41), genitive modifier (42), or possessive predicate (43). (39) toti o-ze-mim-o < toti (Cr.) turtoise 3-REFL-hide-CONT ‘The turtoise is hiding.’ (40) ɾadio o-indu-o radio 3-listen-CONT ‘He is listening to the radio.’ (41) wt pita-kom maso-got o-ho far.away child-PL nun-to 3-go ‘They were going away to the nuns.’ (42) lekol-a-zaɾ school-a-master ‘school teacher’ (43) d-e-sapato-⇔i NEG-1-shoe-NEG ‘I don’t have shoes.’ < radio (Fr.) ~ radyo (Cr.) < maso (Cr.) < l’école (Fr.) ~lekol (Cr.) < zapato (Sp.) Although the indigenous languages Kali’na and Emérillon are structurally distant from the Romance languages, no major morphosyntactic feature seems to have been transferred. Kali’na and Emérillon have borrowed mainly lexical items from the Romance languages, and most loanwords are fully integrated into the morphosyntactic systems of the recipient languages. However, those borrowings may sometimes induce indirect syntactic change. 22 4.3. Erreur ! Style non défini. Syntactic changes induced by borrowings in Kali’na This section presents three syntactic changes in which contact may have triggered grammatical evolution. The following change in Kali’na syntax seems to be indirectly induced by contact: the role of the copula construction is reinforced by the very frequent use of the copular construction and the parallel construction with the verb l ‘to put’ in order to integrate nouns that were originally verbs in their source language. Noteworthy is the disappearance of the borrowing process of indirect insertion via a verbalizer as in begi-ma ‘to pray’ due to the stop of borrowing from Sranan. The copula construction is now the exclusive way to integrate verbs from whatever language source into present-day Kali’na, due to the recent extensive borrowing of French verbs. Examples with Brazilian Portuguese as the source language have also been recently collected in Brazil: (44) misa ta reza poko wai mass in prayer busy.with 1.COP.PERF ‘I prayed at the mass’ < reza(r) (Port.) This type of change where an originally more marginal native construction has been enhanced as a consequence of contact is called “change of enhancement” by Campbell (1987: 271). The second change discussed here is induced by the borrowing of a frequent functional word, the coordinative conjunction nanga5 (< Sr.) inserted between the two NPs as in (45). In the traditional Kali’na construction, a comitative postposition malo ‘with’ is postposed to the second NP as illustrated in (46). (45) wayamaka nanga Iguana and ‘Iguana and Caïman’ akale Caïman (46) wayamaka akale malo Iguana Caïman with ‘Iguana with Caïman’ The coexistence of the two constructions create a significant typological change. This evolution has been reported, especially as a result of grammaticalization processes, as a shift from a “with-language” (with a postposition) to an “and-language” (with both a comitative postposition and a coordinator) in Stassen’s work (2000). Erreur ! Style non défini. 23 Furthermore this conjunction is observed both as a coordinator between two phrases or two sentences. Two remarks can be drawn from this case, and will also apply to the next contact-induced syntactic change to be described. As was pointed out in Stolz (2001), conjunctions and particles are the most frequent grammatical borrowing from Spanish into Amerindian languages. The idea that utterance modifiers, regulating linguistic-mental processing activities, are the most vulnerable items to contact-related linguistic change, due to cognitive pressure, is largely developed in Matras’ work (1998). The second remark is also taken over from Stolz’ article. This author argues against the gap hypothesis, assuming there is no reason to posit a gap when facts show coexistence of autochthonous items and loan items, often with a stylistic difference.6 The third change to be discussed is linked to the borrowing of conjunctions in Kali’na. Kali’na is starting to replace its nonfinite subordination construction introduced by traditional postposed conjunctions such as (47) with a finite subordinate constructions introduced by a borrowed preposed conjunction (48), very different from the inherited model, thus opening up the possibility of a typological change in subordination. (47) [na’na ’wa maina pom-l yako] 1EXCL. by garden plant-NLZ when lo ko’ki kini-kolo’-nen exactly my.little.boy 3-burn-PAS ‘It was just when we were planting the garden that my little boy burned himself’ (48) awu ko lupota-l s-amo-ya I but 1.be tired-NLZ 1A-begin-PRES [pase awosinpe awu wa] because heavy I 3.COP.PRES kn-ka-no kaitusi mo’ko wala ’wa 3-dire-PRES jaguar DEM ibis to ‘ “But I begin to be tired, because I’m heavy”, says the jaguar to the ibis’ A debate could be open to argue whether the conjunction alone is borrowed so that the typological change would be indirect, or whether the whole construction is borrowed (“conjunction + finite clause”) and the typological change would be direct. On one side, the literature asserts that grammatical words are always borrowed along with the rules regulating their linear ordering (Moravsik 1978), arguing for a combined borrowing of morpheme 24 Erreur ! Style non défini. and structure and therefore a direct change. On the other side, examples are given of the borrowing of conjunctions as a trigger for another change: Thomason (2001: 62) notes that Siberian Yupik, after borrowing conjunctions from Chukchi, replaced its inherited nonfinite subordinate constructions with constructions consisting of “conjunction + finite clause”, therefore creating an indirectly contact-induced change. In Kali’na too, the change of structure actually seems to be posterior to the borrowing of conjunctions. A few examples constitute an intermediary stage where conjunctions borrowed from Sranan are followed by a non finite clause. (49) an-ukut-’pa man-ton NEG-know-NEG 3.COP.PRES-PL [efi tabene molo ot wai-l] if expensive DEM thing be-NLZ ‘They didn’t know that gold had a great value’ The complete change to a “preposed conjunction + finite clause” is therefore indirect, but not total, being restricted to those subordinate constructions with borrowed conjunctions. As a conclusion, Kali’na data do show some contact-induced changes, but for now with a limited impact on the typological characteristics of the language. This supports the idea that grammatical borrowings are not facilitated between typologically distant languages. In contrast, borrowing of grammatical morphemes is attested between typologically similar languages: for instance, Emérillon borrowed its plural morpheme -kom from Kali’na. 4. Conclusion In the course of this paper, we noted several times that our data is confirming various assertions concerning language contact. In section 2, it is shown that borrowing of lexical items is easier than borrowing of grammatical categories. In 3.2, data illustrated the gap hypothesis. Section 4.1 illustrated the Principle of Reanalysis and the Hierarchy of Borrowability. However, our main point here is to take advantage of the typological characteristics of the languages involved to weigh their role in the contact situation. The recipient languages Kali’na and Emérillon are quite similar, while they strongly contrast with the typological characteristics of the Erreur ! Style non défini. 25 source languages, among which Romance languages on one side and Creoles on the other side are also typologically distant. Following this, certain aspects of our data lead to further questions: – Why should two typologically comparable languages treat differently the borrowed items? We refer for example to the fact that Kali’na and Emerillon both share a constraint on consonant sequences but deal differently with the borrowed items, or to the fact that verb borrowings are frequent in Kali’na, not in Emerillon. – Why should a language treat differently borrowings from typologically comparable languages? We refer here for example to the fact that Kali’na treated differently the verbs borrowed from Sranan, and those borrowed from Guyanese Creole. – Why should items borrowed to two typologically distant languages be treated in a similar way in the same receiving language? We refer here for example to the same integration process for verbs borrowed from French and Guyanese Creole into Kali’na. We have looked for explanations in the system of the source languages, and in the system of the recipient languages. Now the sociolinguistic environment should be investigated, namely the specific type of contact between each pair of languages. For example, comparing contact-induced changes in Emérillon and Kali’na is always biased, since as we have shown in section 2, their histories of contact are always out of line. For example, using Thomason and Kaufman borrowing scale (1998: 74–76), Kali’na would be in step 3 “more intense contact” with French, while Emérillon would rather be between step 1 “casual contact” and step 2 “slightly more intense contact”. Emérillon borrowings to French are almost limited to lexical items (although code-switching may sometimes introduce French conjunctions into Emérillon speech). More explanations could probably be put forward by referring to social factors, such as degree of bilingualism, degree of education and literacy, size of the community. For instance, our data showed that verb borrowing is attested only in contexts displaying a certain degree of bilingualism, whereas noun borrowing is attested in every contact situation. A finer analysis of the contact situations is now necessary, since the sociolinguistic contexts have not been described before. Looking now towards the future, the borrowings we have been dealing with do not constitute a closed set. While the same processes are still in progress, several elements also show that, as the socio-cultural pressure and multilingualism are getting stronger, the languages enter a new phase in 26 Erreur ! Style non défini. their contact history: examples of nonce-borrowings and code-switching are numerous, and let us foresee other interesting types of contact-induced changes. Notes 1. 2. 3. 4. 5. 6. This autodenomination [kaiña] corresponds to what is called “Carib” in the literature, or “Galibi” that applies more specifically to the oriental dialect spoken in French Guiana. The phonetic nasality of /o/ is the result of regressive assimilation, /b/ being realized phonetically as a prenasalized [mb], but still counting structurally as an “oral” phoneme. It is important to note that liquids are usually not deleted in other positions, French or Creole // being regularly substituted by /ɾ/. One of the functions of reduplication of the predicate is to indicate the plurality of participants (Rose 2005). More rarely, the conjunction et (p[u]is) (< Fr.) is used. Some data illustrates this very nicely, displaying a combination of both malo and nanga in utterances of the bilingual speech of children at school (Alby 2001). References Alby, Sophie 2001 Contacts de langues en Guyane française: une description du parler bilingue kali'na-français. PhD Dissertation, Université Lyon II, France. Auer, Peter 1999 From codeswitching via language mixing to fused lects: Toward a dynamic typology of bilingual speech. The International Journal of Bilingualism 3.4: 309–332. Bruyn, Adrienne 2002 The structure of the Surinamese Creoles. In Atlas of the Languages of Suriname, Eithne B. Carlin, and Jacques Arends (eds.), 153–182, Leiden: KITLV Press. Campbell, Lyle 1987 Syntactic change in Pipil. International Journal of American Linguistics 53.3: 253–280. Courtz, Henk Erreur ! Style non défini. 27 Karaïbs Nederlands woordenboek. Paramaribo: Instituut voor Taalwetenschap. Field, Fredric W. 2002 Linguistic Borrowing in Bilingual Contexts. (Studies in Language companion Serie.) Amsterdam/Philadelphia: John Benjamins Publishing Company. Gildea, Spike 1998 On Reconstructing Grammar, Comparative Cariban Morphosyntax. New York/Oxford: Oxford University Press. Goury, Laurence (ed.) 2001 Langues de Guyane. (Amerindia 26/27.) Paris: AEA. Hengeveld, Kees 1992 Non-verbal predication. Berlin/New York: Mouton de Gruyter. Launey, Michel et al. 2003 Langues de Guyane. In Les langues de France, Bernard Cerquiglini (ed.), 269–302. Paris: PUF. Matras, Yaron 1998 Utterance modifiers and universals of language borrowing. Linguistics 36.2: 281–331. Moravcsik, Edith 1978 Language Contact. In Universals of Human Language, Vol. 1: Method & Theory, Joseph Greenberg (ed.), 93–122. Stanford: Stanford University Press. Renault-Lescure, Odile 1985 Evolution lexicale du galibi, langue caribe de Guyane. Paris: ORSTOM, TDM F16. 2002 Dynamique des relations actancielles en kali'na de Guyane française (ou galibi). In Langues de Guyane (Amerindia 26.), Laurence Goury (ed.), Paris: AEA 2005 Intégration grammaticale des emprunts et changements linguistiques dans la langue kali’na de Guyane française (famille caribe). In Dinámica lingüística de las lenguas en contacto, Claudine Chamoreau, and Yolanda Lastra (eds.), Universidad de Sonora: 103-120. Renault-Lescure, Odile, and Elsa Gomez-Imbert 2005 Etude phonologique du kali’na. Rapport d’activité final, MOM, ms. Rose, Françoise 2002 Cross-referencing in Emérillon (Tupi-Guarani): a hierarchical agreement system. Santa Barbara Papers in Linguistics 11: 71-84. 2003a Le marquage des personnes en émérillon (tupi-guarani) : un système d'accord hiérarchique. Faits de Langues 21: 107-120. 2003b Morphosyntaxe de l'Emérillon. Une langue tupi-guarani de Guyane française. PhD Dissertation, Université Lumière Lyon II. 1997 28 Erreur ! Style non défini. 2005 Reduplication in Tupi-Guarani languages: going into opposite directions. In Studies on Reduplication, Bernhard Hurch (ed.), 351–368, Berlin/New York: Mouton de Gruyter,. Stassen, Leon 2000 AND-languages and WITH-languages. Linguistic Typology 4: 1–54. Stolz, Thomas 2001 Les hispanismes grammaticaux dans les langues amérindiennes. Faits de Langues 20: 163–173. Thomason, Sarah Grey, and Terence Kaufman 1988 Language Contact, Creolization and Genetic Linguistics. Berkeley: University of California Press. 2001 Language Contact. Washington: Georgetown University Press. Weinreich, Uriel 1970 [1953] Languages in Contact. The Hague : Mouton. Wichmann, Sören, and Jan Wohlegemuth (this volume) Loan verbs in a typological perspective Winford, Donald 2. 2003 An Introduction to Contact Linguistics. Oxford : Blackwell Publishing. Erreur ! Style non défini. 29 30 Erreur ! Style non défini.