3.1 Theoretical Framework and Corpus Data

3.1.1 Introduction

Until recently (cf. Jennings and Pfänder 2015), FGC counted among the French-based creoles that had not been systematically analysed for possible West African contributions. Neither early work by Saint-Jacques Fauquenoy (1972), the unpublished but highly detailed description by Peyraud (1983), the short essay by Corne (1971), nor the extensive analysis of the tense and aspect system by Pfänder (2000a, b) insisted on African (and more specifically Gbe) influence on the emergence of FGC. However, the sociohistorical analysis provided in Chap. 2 of this book concludes that Gbe languages were essential to the creation of the new creole in French Guiana. This chapter will test our conclusion from a linguistic viewpoint by building on the work of scholars who reconstruct the interaction of linguistic models in creole genesis. These include Siegel for various creoles, especially from the Pacific, Kriegel and Michaelis for French-based creoles in the Indian Ocean and Migge and Winford for mostly English-based Surinamese creoles. The main hypothesis of this chapter is that most FGC grammatical features can be modelled as one of Siegel’s four types of transfer from Gbe languages like Ewe , Fon and Gẽ. In cases where the transfer model does not work, we hypothesise that constraints such as the basic word order of French are responsible. We begin our investigation with a discussion of the dichotomy between inheritance and innovation evoked in the book’s title (Sect. 3.1.2), before moving on to the usage-based linguistics framework (Sect. 3.1.3), and the four types of transfer (Sect. 3.1.4). These conceptual approaches will then be applied to the analysis of features of the FGC noun phrase (Sect. 3.2), verb phrase (Sect. 3.3) and sentence structure (Sect. 3.4).

3.1.2 Inheritance and Innovation

Traditionally, linguistic continuity, often worded as ‘inheritance’, links the emergence of creole languages to only one linguistic ‘parent’, but this metaphor is tenuous given that languages are transmitted socially rather than genetically. If any, FGC has more than one ‘parent’ since Portuguese preceded French in the colony. In addition, the Arawak pidgin was a long-standing lingua franca , and Gbe languages such as Ewe, Fon and Gẽ were used persistently by slaves during at least the first fifty years of colonisation. The multilingualism of slaves and the presence of other African languages also need consideration. Even though most of FGC’s lexicon derives from French, other languages should not be ignored. As will be shown in this chapter, the creole’s grammar shows influences from both Romance languages (forms and word order) and Gbe languages (functions, strategies and categories). In such complex linguistic cases, it is difficult to apply conventional linear genealogy based on speciation, in which a language slowly diverges from its parent. Instead, inheritance from multiple sources driven and shaped by social factors seems a more likely approach to follow (Enfield 2005).

Linguistic innovation has been conceptualised in many ways. Matthews defines it as “[a]ny change by which a language diverges from others historically related to it” (2014: 196) and offers as an example the innovative word order of the modern English construction, clearly I have lost, which contrasts with the verb-second order found in Old English. Some scholars distinguish between “innovation” (an individual act by a speaker) and “change” (the incorporation into a language of innovations) (Milroy and Milroy 1985: 381). Others consider innovation to be “any element of usage (or grammar) which differs from previous usage (or grammars)” (Andersen 1989: 13). This is the view we subscribe to here because it is much in line with the usage-based framework we advocate in this chapter.

Such a usage-based approach aims to reconstruct the manner by which a specific speech community actually communicated—and changed its language use over time (cf. Cheshire 2006: 426). In this sense, innovation goes beyond individual choice as it has to survive a social process of selection within a community before it becomes part of the community’s language (Whitney 1979: 149–150; Patrick 2006: 578). This approach is very much in line with Labov’s early claim that “no change takes place in a social vacuum. Even the most systematic chain shift occurs with a specificity of time and place that demands an explanation” (1963: 274). Hence, the historical chapter has explored the “agents of change” that have triggered linguistic developments (Eckert 2008; Kerswill 1996, 2006).

Our study contributes to these considerations of inheritance and innovation in several empirically and theoretically relevant ways. First, our FGC corpus data shed light on the variety spoken in the rainforest (cf. Pfänder 2000a), as opposed to the coastal variety that has been the exclusive focus of publications so far (e.g. Schlupp 1997) despite being influenced by other creoles (notably from Martinique and Haiti ). Similarly, we study contact varieties of the principal lexifier by consulting a spoken corpus of French (www.ciel-f.org/vitrine) rather than standard French grammars.

We have chosen Ewe , Fon and Gẽ as widely spoken Gbe languages, with a particular emphasis on Ewe, due to the presence of slaves on the Rémire plantation in 1690 who came from areas where that language is now spoken. Although there may have been significant demographic changes since the seventeenth century, the three languages were likely familiar to most of French Guiana’s first slaves, and, for want of historical data and a spoken corpus of the language, we will test their possible impact through the analysis of modern Gbe grammars. In our study, the specific features of the language contact situation in French Guiana—which are substantially different than in other colonial contexts—play an important role. However, linguistic innovations cannot always be directly attributed to such language contact (“contact-induced innovations”); they also arise “internally” within a community (Cox 2013: 57–58). In a nutshell then, inheritance encompasses transmissions from various ancestor languages, and innovation subsumes both re-combining features and the making of a new system out of usage experiences.

3.1.3 Experience Counts: The Usage-Based Linguistics Framework

In terms of theory, we proceed from the central claim of usage-based linguistics , namely that linguistic structures emanate from usage events (Langacker 1987; Bybee 2006). The usage-based model assumes that “grammar is held responsible for a speaker’s knowledge, of the full range of linguistic conventions, regardless of whether these conventions can be subsumed under more general statements” (Langacker 1987: 494). In a very similar vein, Joan Bybee (2006) claims grammar “to be the cognitive organisation of one’s experience with language”:

A usage-based view takes grammar to be the cognitive organisation of one’s experience with language. Aspects of that experience, for instance, the frequency of use of certain constructions or particular instances of constructions, have an impact on representation that is evidenced in speaker knowledge of conventionalized phrases and in language variation and change. (Bybee 2006: 711)

In other words, our experience with both spoken and written language makes us develop probabilistic expectations. What we hear of any language or variety during our everyday interactions is what “counts” as our linguistic experience. Interestingly, this approach has not been used for language contact situations in which it would apply perfectly.

Beyond being a mere theorem, this experience-oriented approach to language has major implications for modelling the emergence of new linguistic structures or, as it were, of entirely new language systems, as in the case of creolisation .

Being in a contact situation thus means that when we try to perceive structures via similarity and difference, via pattern recognition and categorisation, we happen to be “building” language, and in some cases, to be building “a new language”—not because we lack consistent input or established competence, but because of our general cognitive abilities:

The proposal presented here is that the general cognitive capabilities of the human brain, which allow it to categorize and sort for identity, similarity, and difference, go to work on the language events a person encounters, categorizing and entering in memory these experiences. The result is a cognitive representation that can be called a grammar. This grammar, while it may be abstract, since all cognitive categories are, is strongly tied to the experience that a speaker has had with language. (Bybee 2006: 711)

Experience counts toward a better understanding of grammar. This chapter begins with this somewhat radical claim of usage-based linguistics , namely the idea that the perception of language use in contact situations impacts both on individuals’ processing and, in the end, on the emergence of entirely new linguistic structures in contact-induced language change and creolisation (for more details, see Siegel 2008b, 2010, 2012; Pfänder and Palacios 2014; Jennings and Pfänder 2015; Winford 2003).

Many fine-grained empirical analyses have demonstrated that theories of language acquisition and change cannot ignore usage-based explanations such as entrenchment , analogy, salience or frequency effects (Behrens 2009). It is time to examine these topics in creole emergence and contact-induced language change and to investigate whether similar processes are at work in those areas. The remainder of this introduction covers usage-based research and applies it to some of the core grammatical features of FGC.

To paraphrase Pfänder and Behrens (2016), we might say that language contact affects perception, and perception affects learning because reduced elements are less salient and therefore harder to learn. This, in turn, affects the language system. In contact language-learner varieties, for example, non-salient elements are often omitted. While first-language learners have enough exposure to ultimately turn their attention to fine morphological details, second-language learners in contact situations often ignore them because they are difficult to perceive and functionally redundant since the respective grammatical information is often also coded within other elements.

In spoken language, we—as hearers—have the task of segmenting the stream of sound we listen to. If certain items happen to occur together very frequently, our memory does not trace the items individually, but rather their co-occurrence. This is known in usage-based linguistics as entrenchment (cf. among others Blumenthal-Dramé 2012; Hilpert and Diessel 2016; Seton and Schmid 2016). Entrenchment facilitates the retrieval of more complex items because they are processed as a fused unit. If such units are fossilised, we speak of chunking, which is one of the basic prerequisites for many processes of lexicalisation and grammaticalisation . In grammaticalisation, more often than not, previously independent items are fused and the new gestalt is no longer a lexeme, but instead acquires a grammatical function. Hopper and Traugott (2003) have emphasised how useful the study of the usage frequency of entrenchment is for a better understanding of the building and dismantling of forms and constructions, or, in other words, the initial phase of processes of grammaticalisation. The forms gonna < going to and gotta < have got to and their emergence in overseas varieties of English are a good example here (Lorenz 2012, 2013).

Chunking may also lead to the phonetic reduction effect, which is the blurring and even loss of morphological boundaries between formerly separated items (cf. Bybee and Thompson 1997). In English, for example, as Krug (2003) has shown, palatalisation at the morphological boundaries between verbs and the pronoun you occurs more often in the frequently used don’t you than in the rarer good you (came). Similarly, the phrase let’s can thus become opaque in colloquial English and take extra pronouns, as in let’s you and him fight (cf. Hopper and Traugott 2003: 12–16). In French-based creoles, the usage-based approach explains forms of entrenchment like the agglutination of the definite article :

(1)

 

larie

 

street

‘street’ (Parépou 1885: 18)

However, there are many more types of entrenchment in FGC.

(2)

Yé

pa

trouvé

pyès

granmoun

3pl

neg

find

neg

adult

‘They didn’t find any adult.’ (Holder 1988: 11)

(3)

Nou

ja

maryé

to

ja

di

wi

1pl

already

marry

2sg

already

say

yes

aprézan

to

pri

    

now

2sg

take

    

‘We are already married. You already said yes; now, you are taken.’ (CIEL_FR_GF_RADIO)

(4)

I

gen

sa

osi

ki

pouvé

ekspliké

nou

poukisa

3sg

have

dem

also

rel

can

explain

1pl

why

‘There are also those who can explain the reason to us…’ (CIEL_FR_GF_RADIO)

The cases of reanalysis that lead to entrenchment usually concern two fused items with occasional cases of three, such as dlo, latowè, poukisa and tibonnomme. We find different types of entrenchment, as illustrated in the following section:

  1. a.

    Definite article agglutination : lari (la rue), larout (la route), lapwent (la pointe), lavi (la vie), lakaz (la case), lèr (l’heure), zarb (les arbres), lapli (la pluie), lanmò (la mort).

  2. b.

    Possessive pronouns: mopa (ma part).

  3. c.

    Partitive: diri (du riz), dlo (de l’eau), difé (du feu), dmaten (du matin), dilèt (du lait), dité (du thé), diven (du vin), dipen (du pain).

  4. d.

    Negation/loss of function: nenpòt (n’importe), pókó (pas encore).

  5. e.

    Interrogative pronouns: kitan (qui temps), poukisa (pour qui ça), aksa ((c’est) qui ça).

  6. f.

    Syntactic entrenchments: épi (et puis), konsa (comme ça), annan (en dedans), ka (qu’ a), ké (qu’aller), latowè (là tu vois), ya (eux là), jodla (jour là).

  7. g.

    Lexical entrenchments: bondyé (bon dieu), bonmaten (bon matin), paregsamp (par example), tifiy (petite fille), tigarson (petit garcon), tibonnomme (petit bonhomme), timoun (petit monde), tipoul (petite poule).

Experience does count and, crucially, goes beyond the mere perception of usage frequencies. The strength and nature of experience is also influenced by processing factors such as the context in which a unit occurs, its perceptual salience and memory-related factors such as recency . Constructions in contact depend on the timeframe in which they occur. A recent speech event has a greater influence on how a current speech event is processed than does an older one. This effect can be derived from the structure of memory and has been documented repeatedly in language processing (Szmrecsanyi 2006; Poplack and Tagliamonte 1996; Ellis 2012). Recency has been shown to be important for language change (Rosemeyer 2014) but is not considered here because of the lack of historical corpora of FGC. Salience, though, will be shown to play an important role in the reconstruction of creole genesis (cf. Sect. 3.5.).

In addition to recency and salience, other factors (including social context) play a crucial role in how speakers (and hearers) perceive linguistic structures (cf. Pfänder and Behrens 2016; Ambridge et al. 2015). Ambridge et al. (2015) demonstrate through a variety of studies how situational, social and individual aspects of interaction influence the intake of the input. Other factors such as the social setting or the attitude of the participants may be highly relevant, as well; these factors are rather difficult to put in direct causality with language change (but see Ennis 2008; Ennis and Pfänder 2013).

Similarity effects have been postulated to operate in cross-linguistic language processing (Cutler et al. 2003) and in language contact settings (Jarvis and Pavlenko 2008; Palacios and Pfänder 2014) based on the notion that speakers always build on previous linguistic knowledge and attempt to frame new material as identical or analogous to what they already know. As Cacciari has stated, “Similarity is thus a process of alignment and mapping of interconnected structures and systems of knowledge” (1995: 10).

Within a usage-based approach, the perception of similarity by the speakers of language A and language B, which is essential for triggering a convergence process (Baptista et al. 2014), is conceptualised in terms of “perceived similarity”, a notion developed in the realm of L2 acquisition by Jarvis and Pavlenko (2008). A speaker perceives categories and/or patterns in the L2 in a different manner than an L1 speaker of that language or a linguist would:

A perceived similarity is a conscious or unconscious judgment that a form, structure, meaning, function, or pattern that an L2 user has encountered in the input of the recipient language is similar to a corresponding feature of the source language. (Jarvis and Pavlenko 2008: 179)

Jarvis and Pavlenko’s definition of assumed similarity is in line with the assumption underlying non-contact-induced language change common in the recent cognitive literature, namely that speakers must be considered “pattern seekers”. This idea also applies to the concept developed in Kriegel et al. (2009, 2017). Its application to contact-induced language change is discussed in the following section.

3.1.4 How Experience Counts in Contact: Four Types of Linguistic Transfer

Based on Siegel (2012, 2015) and, in a similar vein, on Jennings and Pfänder (2015), we propose four types of linguistic transfer in creolisation :

  1. 1.

    Functional transfer Type 1 or ‘functionalisation’: A lexeme acquires a grammatical function.

  2. 2.

    Functional transfer Type 2 or ‘refunctionalisation’: The functional range of a grammatical item or construction is extended and/or reduced.

  3. 3.

    Morphosyntactic strategy transfer: Linguistic transfer promotes a morphosyntactic strategy that is originally only marginally present and/or a word order that is highly productive in the contact language.

  4. 4.

    Category transfer: Transfer leads to the stronger routinisation of a semantically defined category that is originally only marginally grammaticalised.

3.1.4.1 Functionalisation

The source of this Type 1 linguistic transfer process is a language B feature that has not yet or only weakly been grammatical(ised) but is perceived as being similar to a grammatical(ised) feature of language A (cf. Pfänder 2000; Ludwig and Pfänder 2003; Bruyn 1996,Footnote 1 2011 on the debate about grammaticalisation in creoles). The language B feature in question is the less-frequently used of the two possible variants of the language. Through transfer, the less-frequent variant is promoted. The other variant, lacking an equivalent in the contact language (language A), may undergo a decrease in frequency or even become lost:

Functionalisation is defined as applying the semantic/functional properties of a grammatical morpheme of a substrate language to a lexical morpheme of the lexifier or to a free grammatical morpheme that has completely different functions. When it occurs, the morphosyntactic strategy of the substrate language for expressing those functions is also adopted. This results in a change in morphological type and therefore the creation of a new grammatical morpheme that is not found in the lexifier. (Siegel 2015: 169)

Transfer through functionalisation can be found, for example, in the emergence of the Haitian Creole progressive marker ap ‘after’, which is said to be based on a structure that exists in north-western varieties of French ‘to be doing’. The more frequent variant ‘to be in the train of doing’, which has no equivalent in West African languages , is not picked up. Hence, ‘to be after doing’, the originally infrequent and geographically limited variant of language B develops into the regular progressive marker in some French-based creoles.

3.1.4.2 Refunctionalisation

The term refunctionalisation (Transfer Type 2) was first introduced in Jennings and Pfänder (2015) and has been taken up by Siegel (2015), who gives the following definition:

Refunctionalisation is defined here as applying the semantic/functional properties of a grammatical morpheme in a substrate language to an existing grammatical morpheme that has been acquired from the lexifier. In contrast to functionalisation, it follows the morphosyntactic strategy of the lexifier, not the substrate language(s), and does not result in a new grammatical morpheme, but rather an extension or contraction of the semantic/functional properties of an already existing morpheme from the lexifier. (Siegel 2015: 169)

The source of a Type 2 transfer process is thus a language B feature that is already grammatical(ised) and perceived as being similar to a grammatical(ised) feature of a language A. In contrast to Type 1, however, there is no variability between features. In Type 2 convergence tendency, a grammatical(ised) feature of language B is refunctionalised on the grounds of its similarity to a grammatical(ised) feature in the speaker’s language A. This type of convergence can be found in the refunctionalisation of the Spanish past perfect tense marking as an evidential or mirative marker in Andean Spanish. L1 speakers of Quechua have been argued to perceive a metonymic similarity between the temporal distancing function of the Spanish past perfect and the epistemic distancing function of Quechua evidential markers (Pfänder 2009). This metonymic similarity has thus been argued to provide the basis for the refunctionalisation.

3.1.4.3 Morphosyntactic Strategy Transfer

We define instances of morphosyntactic strategy as belonging to Type 3. Morphosyntactic—in creoles mostly syntactic—strategy refers to a process by which structures from a source language A are introduced to a typologically different language B. The strategy of using reduplication to mark an increase in intensity, for example, which is strongly grammaticalised in West African languages , is introduced to French, a language that is (a) typologically different and (b) does not use reduplication to mark an increase in intensity. Siegel elaborates on the type of transfer:

Morphosyntactic strategies include various morphological processes – for example, types of affixation or reduplication – that are used to express grammatical meaning. The most important substrate languages in the development of Bislama were the Eastern Oceanic language of central Vanuatu. Many of these languages use partial reduplication for various grammatical functions: with verbs, to show random or repeated action and with adjectives to indicate plural and distributive. (Siegel 2015: 162)

3.1.4.4 Category Transfer

We identify the transfer of semantically defined categories as belonging to Type 4. During the transfer, a new category is introduced and matched onto the form of an existing category, which in turn disappears. Siegel writes:

The linguistic transfer of grammatical categories follows on from conceptual transfer of the semantic categories on which they are based. One substrate language, Nunggubuyu, has a separate grammatical marker maker for six semantic roles; the others have collapsed some of the semantic roles together into one grammatical category, such as goal and location, which both have the same case marker. (Siegel 2015: 165f)

In contrast to Type 3 (morphosyntactic strategy), transfer processes that follow Type 4 (category transfer ) introduce a category that did not receive a grammaticalised expression in the target language prior to the transfer process. In French, for example, there is no grammatically expressed category like the spatial copula “to be at a certain place”, whereas in Portuguese , this category is expressed by ficar “to be located at/in”. In FGC, the locative copula fika can be considered the result of semantic category transfer .

3.1.5 Data and Sources for French Guianese Creole

The first textual evidence of FGC dates back to 1744, and other pieces of evidence come from 1797 to 1824. The principal nineteenth-century sources are an 1848 abolition proclamation (Sournia 1976: 3–8); a grammar accompanied by fables, songs and poems (Saint-Quentin 1872); and a fictional text written in a conversational and familiar style, Atipa (Parépou 1885). The first novel in FGC, it was written by Pierre Félix Athénodor Météran under the pseudonym of Alfred Parépou. The novel was first published in Paris in 1885 and tells the story of Atipa and his compatriots, revealing impressions of the colonial French Guianese language, culture and society. Principal twentieth-century sources are the studies and corpora by Tchang (1988), Contout (1973), Honorien (2010), Horth (1948), Jennings (1999), Peyraud (1983), Schlupp (1997), Pfänder (1996, 2000a, b, 2013) and Saint-Jacques Fauquenoy (1972, 1974). A dialect of FGC is described by Corne (1971). Despite obvious differences to French (and other lexically French-based creoles), descriptions of FGC are often influenced by French (and other creole) categories and translations.

In addition to the sources cited, we use a wide range of linguistic data, most importantly a sizeable quantity of recordings made in 1995 (from West to East, from North to South) in Mana, Organabo, Iracoubo, Corossony, Sinnamary, Kourou, Savanne Maya, Cayenne, Rémire, Roura, Ouanary, Saint-Georges, Maripasoula and Saül. We include a choice of excerpts of this corpus in the Appendix of this study; we deliberately choose transliterations that are appealing both linguistically and anthropologically.

In the pluriglossic language community of French Guiana, situations in which only one language is used are very rare. Among other factors, the speakers’ parents are an important issue here: it is difficult to find a speaker who does not have at least one parent from Martinique due to strong immigration since the eruption of Mount Pelée in 1902. Linguistic integration is problem-free due to the similarity of the two creole languages and sociopolitical contexts, both territories being French overseas departments.

With the help of the respective host family, Pfänder was able to make recordings in which individual family members or their friends talked with neighbours or relatives. Subject areas were agreed upon with one of the informants in the conversation to promote the thematic comparability of the material. The subject areas also included a well-considered order and associative “bridges” to remove social distance within the course of a longer conversation. The topics of conversation included education, work and magic.

The pluriglossic situation in the country is a subject worthy of linguistic research in and of itself. Some pointers regarding the geolinguistic situation, which is usually subject to much speculation, can be given. The creole spoken in Saül and Maripasoula, which are located in the rainforest, is particularly susceptible to the influence of the creole of St. Lucia because these two places developed as gold-digger towns, with the majority of the gold-diggers coming from St. Lucia. French Guiana continues to display a high rate of migration to this day, mainly for occupational reasons (cf. Léglise and Migge 2007). As a result, the rural-urban distinction can only be applied to some degree. The same circumstance may also be the reason for the fact that, at least at the morphosyntactical level and to a lesser degree at the lexical and phonetic levels, only minor differences can be found between the two places, which are located a great distance from one another and were often quite isolated. What does stand out, however, is the very specific prosodic structure of the creole spoken in Ouanary and St. Georges, places that are under the influence of both Portuguese and a language that belongs to the Arawak language family. FGC there has often been learned as an L2, which is rather rare for a creole language. In general, it can be said that the primeval forest communities were rather productive regarding the recordings of a creole that has had little contact with the French language .

Close to the coast, language contact situations are more or less restricted to European languages and Martinique Creole. It is interesting, though, that when looking for a creole that has only little contact with other languages, the so-called five-house settlements that can be found on the map are usually unproductive despite their isolated location since the younger population works in Cayenne or Kourou and commutes regularly. Many of the settlements have been deserted. People who speak FGC on a daily basis still can be found in little towns near the coast (Iracoubo, Sinnamary) and in very isolated settlements.

In addition to the 1995 recordings, we have material collected between 1995 and 2016 comprising audio recordings, school books, radio transmission, YouTube clips, Facebook posts and Twitter messages. This data set is the first in the study of FGC to consider both recordings from the rainforest hinterland and social media contributions. The Internet data allow us to draw some real-time comparisons between the 1995 fieldwork and today. The following excerpts provide a quick impression of the contemporary usage of FGC. Some of the following excerpts were taken from a program on ‘Radio RFO Guyane’ and transcribed for the corpus CIEL_F for the investigation of French. Here, we focus on the paragraphs with code-switching between French and Creole:

(5)

[French]

Son

mari,

 

il

dit:

 
 

poss

husband

 

3sg

say.3sg

 

[Creole]

‘Nou

ja

maryé,

ba

mo

ronflé.

 

3pl

already

married

give

me

snore

 

Menm

si

mo

pyé

ka

santi

 

even

if

poss.1sg

feet

prog

stink

[French]

c’est

pas

grave.’

   
 

be.3sg

neg

bad

   

‘[in French] her husband says: [in Creole] ‘We are married, (so) let me snore. Even if my feet stink, [in French] it doesn’t matter.’

Pioneering work in various other creole languages and contact varieties provides a model to be followed in French Guiana. Increasing contact between speakers of other contact languages on the Internet creates playful experiences in which these languages become intercomprehensible:

(6)

sé

moun-yan

bonjou^^

je suis camerounais et franchement j’aime graaaave le créole, surtout l’accent; Et je voulais vraiment élargir mon vocabulaire créol, pck je connais kelke petit mot.

‘Hello guys! (in FGC; he then goes on in French) I am from Cameroon and honestly, I reeeeeally like the Creole language, especially the accent; and I would actually like to enhance my vocabulary in Creole because I already know a few words.’

The Internet thus offers a global platform to make visible and audible what once was merely local (cf. Mair and Pfänder 2013).

It is not only the creole data that are unique for both fieldwork data and social media data, but also the use of newly available resources of French that we choose as an empirical ground for our analysis in the following sections. In studying French as the main lexifier, we rely upon the first large corpus of spoken French worldwide, CIEL-F (www.ciel-f.org/vitrine; Pfänder et al. 2012). This corpus covers twenty regions of the French-speaking world, offering spontaneous data in audio and video recordings (Mondada and Pfänder 2016).

To test the influence of West African Gbe languages on FGC is to enter new territory. We use as our principal focus the Gbe languages Ewe , Gê and Fon , which are widely spoken in and around the coastal area where many of French Guiana’s first slaves originated. Features listed in the Atlas of Pidgin and Creole Language Structures (Michaelis et al. 2013; Pfänder 2013) serve as our basis for comparison, and we ground our exploration of FGC’s grammar in Siegel’s (2007, 2012, 2015) typology of transfer . As opposed to his method of illustrating different categories with examples from various creole languages, we test all of the proposed types of transfer with the grammar of FGC. This task provides the structure of the following sections, which cover the noun phrase (Sect. 3.2), the verb phrase (Sect. 3.3) and sentence structure (Sect. 3.4).

3.2 Noun Phrase

3.2.1 Indefinite Article

FGC has both a definite and an indefinite article . The indefinite article oun (allegro form ‘n, emphatic form roun) precedes the noun or the noun phrase, just as in French:

(7)

kom

mo

té

gen

oun

kopen

when

1sg

pst

have

indf

friend

‘when I had a friend’a

  1. aThis and all other unattributed quotes of FGC are drawn from the FGC corpus (see Appendix).

(8)

To

ka

wè

a

pa

oun

lavie

2sg

prog

see

dem

neg

indf

life

‘You see, this is not a life.’ (CIEL_FR_GF_RADIO)

(9)

Bèt-ai

‘n

pé

 

rèd.

thing-sg.def

3sg

indf

little bit

difficult

‘This thing is a little bit difficult.’ (CIEL_FR_GF_RADIO)

(10)

Mo

ka

fè

‘n

mayouri.

1sg

prog

make

indf

mayouri

‘I am going to organise a mayouri (i.e., a cooperative work session).’

Unlike French, the indefinite article makes no gender distinction (oun wonm/oun fanm):

(11)

Paské

i

gen

oun

wonm

   

because

2sg

have

indf

man

   

té

ka

fè

oun

chanté

anlé

oun

wonm

pst

prog

make

indf

song

about

indf

man

‘Because there was a man who sang about another man.’

(12)

Yé

ka

tann

oun

fanm,

oun

madanm.

3pl

prog

hear

indf

woman

indf

lady

‘They heard a woman, a lady.’

(13)

To

ka

di

to

pran

oun

sèl

fanm.

2sg

prog

say

2sg

take

indf

single

woman

‘You say that you have only one woman.’ (CIEL_FR_GF_RADIO)

An interesting difference with French is that the indefinite article is not used for generic meanings in the FGC corpus:

(14)

Akisa?

A

kaz!

what-this

this

house!

‘What is this? This is a house.’

Ewe shows a similar omission of the indefinite article in this context:

(15)

Núka nyé ésia?

Ésia nyé xɔ!

What is this?

This is house!

‘What is this? This is a house.’ (Dzablu-Kumah 2015: 30)

As in Ewe, the indefinite article may also be omitted in generic contexts:

(16)

Wonm

ka

fè

so

zafè.

man

prog

make

poss

thing

‘A man does what he wants.’ (CIEL_FR_GF_RADIO)

In the modern media, however, we observe both the omission (example above) and the use (example below) of the generic indefinite article, the latter possibly following the French model:

(17)

Oun

fanm

ka

fè

so

zafè

indf.sg

woman

prog

make

poss.3sg

thing

‘A woman does what she wants.’ (CIEL_FR_GF_RADIO)

Unlike other French creoles in the Caribbean but very much like Ewe , FGC has a plural form of the indefinite article . In Ewe, the plural form is as follows:

(18)

Atí

áɖéwó

mu

tree

indf.pl

fallen

‘Some trees have fallen.’ (Dzablu-Kumah 2015: 31)

In FGC, speakers use the indefinite article dé:

(19)

I

té

gen

dé

ròt

kannon.

3sg

pst

have

indf.pl

other

boat

‘There were some other boats.’

However, the plural form tends to be omitted, as in the following example:

(20)

I

gen

moun

ki

fè

l’

hein

3sg

have

people

rel

do

it

you-know

‘There are people who do it, you know.’

FGC may have inherited the preposed indefinite article from French and Portuguese, in both form and functions. However, there is possible innovation because the form may be influenced by the Portuguese um, and the functions do not cover the same range as in French. The article is not used in FGC for presentative or generic contexts, which correspond to its use in Ewe. Ewe could thus have been a model for a refunctionalisation of the indefinite article in FGC. However, Ewe has no influence on the position. As in French and Portuguese but unlike Ewe, the indefinite article is preposed to the noun. If we take for granted the constraint established by Andersen (1983), there was no slot to which the postposed Ewe article could be transferred. In the following section, however, we see that Ewe speakers may have perceived a possible slot for the definite article .

3.2.2 Definite Article

The definite article occurs after the noun. There is allomorphic variation in this particle depending on the phonemes of the final syllable; the forms are -a (boug-a ‘the guy’) and -an when following /n/or /m/(fanm-an ‘the woman’):

(21)

Boug-a

ka

las

guy-def

prog

tired

‘This guy is tired.’ (CIEL_FR_GF_RADIO)

(22)

Fanm-an

té

ka

brè

woman-def

pst

prog

drink

‘The woman used to drink.’

(23)

Kaz-a

toujou

prop

house-def

always

clean

‘The house is always clean.’ (CIEL_FR_GF_RADIO)

(24)

Sa

genre

de

jeu-a …

dem

kind

of

game-def

‘This kind of game.’ (CIEL_FR_GF_RADIO)

(25)

Nou

tandé

l’

hein

nou

tandé

mo-a.

1pl

hear

3sg

right

1pl

heard

word-def

‘We heard it, you know…we heard the word.’ (CIEL_FR_GF_RADIO)

The syntactic make-up of the definite article exhibits features of both Transfer Type 2 (refunctionalisation ) and Transfer Type 3 (transfer of (morpho)syntactic strategies). In terms of syntactic strategy, the FGC structure closely follows the Ewe pattern of a postposed article in which the article is placed behind the noun phrase, as in (lit.) house the (first example) and house big the (cf. example 27). In FGC, we find:

(26)

kaz-a

 

house-DEF

 

‘the house’

(27)

kaz

gran

a(n)

house

big

def

‘the big house’

This is quite similar to what we find in Fon (cf. Akoha 2010: 87) or in Ewe :

(28)

xɔ lá

 

house def

 

‘the house’ (Dzablu-Kumah 2015: 4)

(29)

tó

lá

mountain

def

‘the mountain’ (Eklou 1987: 36)

(30)

xɔ

lolo

lá

house

big

def

‘the big house’ (Dzablu-Kumah 2015)

(31)

tó

kɔ́kɔ́

lá

mountain

high

def

‘the high mountain’ (Eklou 1987: 41)

Due to the omission of an explicit copula, placing the definite article in the final example (‘the big house’) directly after the noun would change the meaning completely by shifting it from an NP to a simple sentence, both in Ewe (first example) and in FGC (second example):

(32)

Xɔ

lá

lolo.

house

def

big

‘The house is big.’ (Dzablu-Kumah 2015: 4)

(33)

Kaz-a

gran.

house-def

big

‘The house is big.’

The word order here is reminiscent of the bipartite demonstrative pronoun structure in French, the house there. The postposed part of the demonstrative structure in French, there, is much more salient than the preposed part the. It is possible that early speakers of FGC only heard or perceived the phonetically salient postposed part of the French structure, là ‘there’, and consequently re-analysed the right bracket of the demonstrative structure as a definite article (for a more detailed study of a/la in FGC, cf. Wiesinger 2015).

The plural form of the definite article is –ya(n).

(34)

Tifi-ya

séparé

ké

tibonnonm-yan.

girl-pl.def

separate

com

boy-pl.def

‘The girls and the boys are separated.’ (Holder 1988: 7)

The plural form might be best described as a process that draws on both Transfer Type 2 (refunctionalisation ) and Transfer Type 3 (morphosyntactic strategy ). In the plural definite article, la is phonetically reduced to a and preceded by the third-person subject pronoun they. It is spelt ye and phonetically reduced to y. In other words, the plural form ya originated in the structure yé la, lit. ‘they there’. Some evidence for this comes from older texts in which the definite plural article is still separated:

(35)

A

nègue

place

yé

la,

qui

dévidé

Cayenne.

dem

people

place

3pl

def

rel

go

Cayenne

‘It is the people from the gold sites who go to Cayenne.’ (Parépou 1885: 18)

(36)

Nègue

Kourou

yé

la

wa

beau

vanté

yé

cassave.

people

Kourou

3pl

def

fut

in vain

praise

3pl

cassava

‘The people of Kourou will praise their cassava in vain.’ (Parépou 1885: 46)

While the phonetic material comes from French, the word-order strategy of FGC (second example below) follows the Ewe pattern exhibited in awó ‘they there’ (first example below):

(37)

xɔ

adé-a-wó

house

six sg.def-they

‘the six houses’ (Dzablu-Kumah 2015)

(38)

si kaz

yé-la

six house

3pl-def

‘the six houses’

In summary, plural marking in FGC exhibits features of both word-order transfer and refunctionalisation . It is crucial to note here that a pattern such as les arbres … eux … là (lit. ‘the trees they there’) can be found in corpora of informal spoken French. The structure is rare, though, and has a specific pragmatic function. When used, the speaker wants to make sure that the listener can identify an object that can be seen at some distance, along the lines of the trees you can see over there.

The origin of the form la is not easy to determine as it is virtually identical in French and Ewe. The fact that both FGC and Ewe have an allegro variant a (omitting the initial l-) further complicates matters. The similarity seems to be a case of mere coincidence. The transfer of la from Ewe is supported by the fact that Creole la can be reduced to a, which is possible for Ewe la but not for French là. The transfer of là from French is supported by the fact that FGC là does not have the rising tone of Ewe la and is thus closer to the phonic gestalt of French là. Overall, we might want to hypothesise that the convergence of similar forms enhances the possibility that they appear in the creole.

3.2.3 Demonstrative

Unlike in French, in FGC there is only one demonstrative , sa … a, for ‘this’ and ‘that’, where sa precedes the noun and -a follows it (cf. French ce …-là), as in:

(39)

Mo

kontan

sa

liv-a

1sg

like

dem

book-sg.dem

‘I like this book.’

In the plural, the demonstrative is sa …-ya, as in sa liv-ya (‘these books’).

(40)

sa

timoun-yan

dem

children-pl.dem

‘those children’ (CIEL_FR_GF_RADIO)

This syntactic bipartite strategy might be seen as an inheritance from both French and Ewe because French also shows a bracket-construction (41), while Ewe has postposition (42) and bracket-construction (43):

(41)

ces

enfants

là

dem.pl

children

dem

‘those children’

(42)

ɖeví

siawó

children

dem.pl

‘these children’ (Dzablu-Kumah 2015: 31)

(43)

to

ɖeví wó

 

dem

children pl

 

‘those children (over there)’ (Dzablu-Kumah 2015: 31)

The second Ewe example (43), being more deictic in face-to-face interaction in the early years of contact, is likely to be used. If this final assumption is correct, then FGC exhibits a morphosyntactic strategy transfer from Ewe insofar as the plural is marked within the right bracket of the construction, whereas in French, it is marked in the left bracket.

In modern media texts, sa …-ya can also be found in more complex structures, such as the following:

(44)

sa

genre

de

bèt-ya

dem

kind

of

thing-pl.def

‘these kinds of things’ (CIEL_FR_GF_RADIO)

It is important to note that sa …‑a (and in the same way, sa … -ya) can be used not only adnominally, but also pronominally, as in the following example:

(45)

Sa-a

movè

dem-sg.def

bad

‘This one is bad.’

In older texts, we find examples like the following:

(46)

Gadé

qué

oueil,

ça

la

Chinois

yé

wa

tombé.

Look

with

eye

dem

def

Chinese

3pl

fut

fall

‘Look around you, those Chinese people, they will fall.’ (Parépou 1885: 50)

It is important to note that ça is used for animate subjects in CIEL-F in some varieties in both Africa and Canada (Boutin and Gadet 2012). If we assume that this use was attested in the early years of contact, there is some inheritance from both French and Ewe here.

3.2.4 Personal Pronouns

There are three sets of personal pronouns—subject pronouns, object pronouns and adnominal possessive pronouns—that differ only in the third-person singular (see Table 3.1).

Table 3.1 Personal pronouns in FGC
Full size table

Both FGC and Ewe use the forms of subject (or object) pronouns to express possession, while French does not. Hence, mo can mean ‘I’, ‘me’ or ‘my’. While the lexical material comes from French, the strategy of using the same form for different types of pronouns seems to be due to transfer from Ewe. These types of syntactic strategy transfer are sometimes analysed as cases of simplification. However, FGC has only a few simplified subsystems, suggesting that the marking of possession is due to transfer from Ewe rather than to simplification of the French subsystem.

As in many other creoles, there is no gender distinction in the third-person singular in FGC: the forms i, li, l’ and so denote both male and female referents. However, some speakers use misyé and madanm as third-person singular pronouns; these forms seem to grammaticalise in the smaller townships but are rarely used in Cayenne or Kourou in our corpus.

3.2.5 Possessives

The possessive adjective in FGC precedes the noun, as illustrated in the following examples:

(47)

mo

liv

to

kaz

poss.1sg

book

poss.2sg

house

‘my book, your house’

(48)

So

kaz-a

toujou

prop

poss.3sg

house-def

always

 

‘His house is always clean.’ (CIEL_FR_GF_RADIO)

(49)

Oun

gran

larivyè

ki

ka

séparé

indf

big

river

rel

prog

separate

nou

péy

Lagwiyan

di

Sirinanm

1pl

country

French Guiana

from

Suriname

‘A big river separates French Guiana from Suriname.’ (Holder 1988: 7)

(50)

A

zòt

timoun

ki

la?

dem

poss.2pl

child

rel

there

‘Is that your child over there?’

In modern media, the French vòt (<votre) sometimes replaces zòt:

(51)

A

vòt

solusyon

dem

poss.2pl

solution

‘This is your solution.’ (CIEL_FR_GF_RADIO)

In both French and Ewe , possessive adjectives precede the noun. However, FGC possessive marking is more similar to Ewe. The forms of the possessive pronouns mine, yours, etc. developed in a lexicalisation process that fused two items, namely the possessive adjectives discussed above and a phonetically reduced form of the noun part:

(52)

mopa

topa

sopa

‘mine’

‘yours’

‘hers/his’

In older texts, we find the same construction; however, the fusion has not yet taken place and the elements are still written as two words with a meaning that could make the source of the construction transparent (>‘for my part’, ‘as for me’):

(53)

Mo

pa,

a

Oyapoc

mo

soti.

1sg

part

loc

Oyapock

1sg

come from

‘As for me, I’m coming back from the Oyapock.’ (Parépou 1885: 14)

This type of lexicalisation strongly resembles Ewe, where the possessive adjectives nyè, wò, etc. are fused with the noun to ‘part’. We would not expect a combined lexicalisation and grammaticalisation process in FGC, as the combination of a possessive adjective followed by the noun part would not be a highly frequent string in everyday French conversation. Hence, the forms of the possessive pronouns mine, yours, etc. are argued to be due to transfer from Ewe.

3.3 Verb Phrase

3.3.1 Introduction

In the previous section, we provided possible pathways of transfer for individual features of the noun phrase. The nature of the verb phrase in FGC, with its interrelated system of tense , mood and aspect (TMA) markers, requires a more holistic approach, and this section thus proposes a transfer hypothesis after carefully outlining the functional range of each marker. This approach also allows for a discussion of the full TMA system, which has been the cornerstone of debates on creole genesis since Bickerton’s controversial work (1981).

3.3.2 The Zero Marker: Perfective Aspect

In FGC, the zero marker (in the following, Ø) has to be analysed as a perfective aspect , irrespective of its temporal embedding; by contiguity, Ø often has a past-tense reading, but not necessarily. The following examples demonstrate how Ø can have past, present and future reference according to genre and context.

The perfective aspect marker Ø has a past time reference, especially in narrative sequences. This use is well known from what may be the first sentence in FGC for which we have a historical record (a wood carving):

(54)

Anglai

Ø

pran

Yapoc,

yé

Ø

mené

mon

English

pfv

take

Oyapock

3pl

pfv

lead

poss

père

alé

      

priest

go

      

‘The English have taken Oyapock, they have taken away the priest.’ (Saint-Quentin 1872: 195)

If we examine storytelling, we find long chains of zero-marked verbs. In the following example, we observe that a narrative is given only via bare verb forms:

(55)

mo Ø né isi,

‘I was born here’,

mo Ø levé isi

‘I was raised here’,

mo Ø maryé isi

‘I got married here’,

mo Ø fèt tout bagaj isi.

‘I did everything here.’ (i.e.: kindergarden, school, …)

apré mo mari Ø mori jèn

‘Then my husband died (although he was) young.’

mo mari Ø mori i té gen trantsiz an

‘My husband died at the age of 36.’

lò mo Ø wè sa mo té ganyen wit timoun

‘At that moment (lit.: = when I saw this happen), I had eight children.’

mo mari Ø lésé mo enceinte . gro vant

‘My husband left me (alone) pregnant.’ (first word in French, then in FGC)

mo Ø pati

‘I left (this village).’

mo Ø alé Cayenne sasé travay

‘I went to Cayenne (in order to) look for a job.’

The chain of actions justifies why the narrator, a woman aged 68 at the time of the interview, went to work in Cayenne, thereby leaving the village where she was born. It is important to note that the typical grounding of a chain of actions can be lò, ‘when’, and it is always zero-marked. Corpus examples include the following:

(56)

lò to Ø rivé

‘when you arrive(d)’

(57)

lò to Ø pati

‘when you leave (left)’

(58)

lò nou Ø wè sa

‘when we see (saw) that’

(59)

lò i Ø fè n promenad

‘when he goes (went) for a walk’

The second example of storytelling involves explaining field work. Here again, the actual activity in the field and going back home are marked as the salient actions.

(60)

Nou bati tou Ø proch tou Ø proch

‘Our field was not far away.’

Trwa jou nou ka alé serklé

‘We worked there for three days’

answit jiskan y Ø fini.

‘until everything was done.’

Épi nou ka désann

‘And then we went back (on the third day).’

Bèn lò to Ø rivé to Ø fè to ben-d-basen.

‘When we came back, we had a nice bath.’

To Ø lavé, to Ø alé pozé to kò.

‘We washed ourselves, we had a rest.’

To Ø ranjé to kabann,

‘We tidied up our belongings,’

épi sayè! Nou té ka viv, oh !

‘and then that was it, we lived (well), you see.’

The actions of cleaning up and having a rest are given as perfective actions, and it is thus not the event structure itself that is or is not objectively (im)perfective; rather, it is the speaker who may subjectively choose an (im)perfective perspective for different parts of her or his narrative reconstruction of the events. This matches perfectly well with the English etymon “aspect ” as well as with the original Russian term for aspect, “vid”, both of which mean “gaze / the action of looking at something”.

The perfective aspect marker Ø can have a present time reference or lack temporal reference all together, as in this greeting:

(61)

Sa

Ø

alé

byen

3sg

pfv

go

well

‘I am fine.’ (Pfänder 2000a: 90)

Here, alé, ‘to go’, is a verb not of movement (and thus does not mean ‘I went’), but of current state (‘I am fine’). Similarly, verbs like savoir and trouver have a present reading and do not refer to the dynamic action of learning or searching but instead refer to a state of mind that also expresses a stative meaning:

(62)

Oui,

mo

trouvé

ça,

réponne Atipa

Yes

1sg

find

dem

answer

Atipa

‘“Yes, I think so,” said Atipa.’ (Parépou 1885: 18)

(63)

To savé

2sg

know

‘You know.’ (Parépou 1885: 18)

(64)

Mo

krè

an

Bondyé

1sg

believe

in

God

‘I believe in God.’

(65)

Lò

blangue

Ø

baille

li

avance,

When

white

pfv

pay

3sg

advance

yé

pouvé

compté

là-sous

li

 

3pl

can

count

on

3sg

 

‘When the whites pay him in advance, they can rely on him.’

(Parépou 1885: 206)

In addition, the zero marker can contextually show a future time reference, especially if the previous sentence is an imperative:

(66)

Ba

mo

dé

jounen,

trwa

jou,

 

Give 1sg

two

days

three

days

 

mo

Ø

sabré

oun

ti

bati

la.

1sg

pfv

cut

det

small

clearing

det/loc

‘Give me two or three days, and I will cut a clearing.’

(Jean-Louis 1987: 295)

(67)

Konpè

Makak,

o,

mon

ami,

 

brother

Macaque

oh

poss

friend

 

ba

mo

mo

palto,

mo

Ø

alé.

give

1sg

poss.1sg

jacket

1sg

pfv

go

‘Brother Macaque, oh, my friend, give me my jacket, and I will go.’

(Jean-Louis 1987: 437; 444)

The common examples above provide good evidence to prevent misunderstanding zero marking as representing past reference; instead, it is perfective aspect marking, which—by default—may have past reference simply because an action that is presented as being ‘perfect’ tends to be complete. However, present or future readings are certainly possible, as we have seen, if the speakers want to talk about an action as an action without going into any process-related details.

3.3.3 Ka: Progressive Aspect

Since its earliest attestations, ka in FGC has mainly been used for progressive aspect .

(68)

Mo

ka

alé

1sg

prog

go

‘I am going.’

In the majority of examples, this marker refers to events that take place at the very moment of speech:

(69)

Coument,

to

ca

doumandé

mo,

What

2sg

prog

ask

1sg

si

ça

vrai,

réponne

Atipa.

cond

dem

true

reply

Atipa

‘“What, you’re asking me if it’s true?” replied Atipa.’ (Parépou 1885: 20)

This is still true today, as the following two examples illustrate. Whereas with to ka konprann, the speaker asks for some kind of processual understanding, the use of the zero marker ou ø konprann asks for a perfective understanding of what has been exposed so far:

(70)

Dé

timadanm

la

nou

k‘alé

bladyé

ké

yé

det

girl

loc

1pl

prog.go

have fun

com

3pl

anfen

voyé

blag!

so

send

joke

‘Girls are there; we have fun with them and make a lot of jokes.’

(71)

To

ka

konprann

2sg

prog

understand

‘You understand.’

(72)

Ou

komprann?

2pl

understand

‘Do you understand?’

In Martinican Creole (MAC), the imperfective marker ka may express progressivity, but mostly expresses habitual or iterative meaning (Damoiseau 1984: 23).

(73)

Tou

lé

dimanch

i

ka

manjé

pwason

every

Sunday

3sg

ipvf

eat

fish

‘He eats fish every Sunday.’ (Damoiseau 1984: 22)

MAC also differs from FGC in that it uses ka mainly with present reference and requires combination with the past marker té to form té ka to indicate past reference. The té can be dropped only once the speaker has established that the context is not in the present. FGC, on the other hand, can use ka with past reference without a past marker. Stative verbs in MAC rarely combine with ka and do so only to express inchoative (e.g. getting angry) as in (74) or habitual meaning (e.g. usually being angry). The core meaning of imperfective aspect , i.e. progressivity, is blocked by stative verbs in MAC:

(74)

Timanmay

la

ka

bèl

child

def

inch

beautiful

‘That child is starting to become beautiful.’ (Damoiseau 1979: 165)

In FGC, all verbs can be combined with the imperfective marker. If so-called stative verbs are combined with ka, the default meaning is progressivity. The main function of ka found in FGC data is to demonstrate the progressivity of an action, which is preceded by a verb of perception (typically visual), as in the first of the examples below:

(75)

Mo

manman!

La

i

Ø

gadé

moun.

poss

mother

here

3sg

pfv

look

people

‘Gosh! There, he spots the other group and sees people.’

Moun

ka

manjé

moun

ka

brè

 

People

prog

eat

people

prog

drink

 

‘People are eating; people are drinking.’

Mo

di:

a

mo

pa

ka

alé

1sg

say

to

1sg

neg

prog

go

an

mo

kaz

manjé.

   

in

poss

house

eat

   

‘I said, “I will not return to my house for lunch.”’

I

di:

ou

ka

rete

la.

  

3sg

say

2pl

prog

stay

here

  

‘She said, “You stay here.”’

Wonm,

mo

bat

ou

wonm.

   

Man

1sg

beat

2pl

man

   

ki

sa

ou

ka

kolè.

   

comp

be

2pl

prog

angry

   

‘Man, I beat you, mate. Why are you so angry?’

In the above example, the stative, denominal verb kolè (< la colère “anger”) is used with the imperfective particle. It is clearly not intended as an iteration or habituality here, but as progressivity. It is important to note that the sense here is not inchoative: The addressee is not about to be or becoming angry, but is currently angry.

In short, ka in FGC has progressive meaning and refers to an event that happens at the moment of speaking and before the speaker’s or listener’s eyes. As both dynamic and stative verbs are used in this context, we shall take a closer look at these different uses and compare expressions with dynamic and stative verbs.

3.3.4 Stative vs. Dynamic Verbs

One parallel between FGC and other lexically French creoles is that the same aspect marker may convey different meanings according to whether it marks stative or dynamic verbs. However, FGC behaves differently in that it is not limited to dynamic verbs, but is productive for all kinds of verbs or verboids (Ludwig 1996), as this section will show. This difference is significant because it indicates a greater convergence between FGC and languages of the Gbe family, which would be very much in line with the findings of both historical Chap. 2 and the findings in the previous (and following) linguistic sections of Chap. 3.

In FGC, verbs based on adjectives denoting the characteristics of objects (dry, full, white, fat, bloated) can be combined with the imperfective marker ka to express processual meanings, referring to things that are becoming dry, full, white, fat, bloated, etc. In contrast, zero marking would express “states”, referring to things that are dry, full, white, etc.

(76)

Toupannan

lenj-a

ka

sèk

asou

zèrb-a,

While

clothes-det

prog

dry

on

grass-det

timoun-yan

ka

tranpé

annan

dilo-a

‘children -det

prog

soak

in

water-det

‘While the clothes are drying on the grass, the children go into the water.’

(Schlupp 1997: 73)

In the next example, which refers to the preparation of manioc, blan (“to be white”) is combined with ka as a process:

(77)

To

ka

lavé l’

byen

prop

pis

I

ka

blan.

2sg

prog

wash.3sg

well

clean

more

3sg

prog

white

‘You wash it nice and clean because it becomes white [as you peel it].’

A similar example is the use of plen (‘to be full’) in one of Tchang’s transcribed fairy tales:

(78)

Mé

a

‘n

ti

bous…

  

Look it’s

det

small

purse

  

Toutan

i

Ø

plen

i

Ø

plen!

Always

3sg

pfv

full

3sg

pfv

full

‘Look, it’s a small purse … it’s always very full!’

(79)

Lò

wou

tire

tout

soumaké

i

 

when

2pl

draw

all

money

3sg

 

Ø

ganyen

annan,

kou

wou

  

pfv

have

inside

when

2pl

  

fin

tire

l‘

i

ka

plen

ankò

finish

draw

3sg

3sg

prog

fill

again

‘When you take out all the money that was inside, when you’ve taken it

all out, it fills back up again.’

(80)

Wou

tire

ou

dépasé

tout

soumaké

i

Ø

ganyen,

2pl

draw

or

spend

all

money

3sg

pfv

have

‘You take out all the money that’s inside and spend it.’

(81)

Lò

wou

fin

tire

dernyé-a

i

Then

2pl

finish

draw

last-det

3sg

Ø

rouplen

ankò!

   

pfv

refill

again

   

‘As soon as you’ve taken out the last coins, it’s already full again!’ (Guy Tchang 1988: 154.1)

Here, we have a purse that displays the rather nice characteristic (state) of always being full (see toutan i plen: “it is always full”). In the second line, the purse appears with ka , indicating the process of filling: “It fills back up again”. In the final line, it is described as full again (state).

The same logic holds for stative verbs that indicate a person’s mood as opposed to their character. While to ø fou would be used to address someone considered permantently insane, to ka fou refers to a temporary state as illustrated in the following excerpt:

(82)

Alò

to

ka

fou,

bagaj

ka

pasé

a

 

So

2sg

prog

mad

thing

prog

happen

in

 

to

tèt,

 

to

ka

ri

pou

bon

kyò

poss.2sg

head

 

2sg

prog

laugh

for

good

heart

‘So you are being crazy! Things are going around in your head, you are laughing without any motive or reason.’

In the extract below from a story about a parrot that befriends a lonely boy (Lohier 1960: 196/334), we find the stative verb tris (‘to be sad’) both with (87) and without ka (84). The narrator uses the zero-marked form to establish that the boy is characteristically sad. To show that the parrot views the sadness as a temporary illness that strikes each day, the narrator marks the same verb with ka:

(83)

Annan

lékol-a

gen

roun

gran

lakou

In

school-det

have

det

big

yard

koté

timoun-yan

Ø

kontan

anmizé

 

where

child-det

pfv

love

enjoy

 

‘At school, there is a large play area, where the children like to play.’

(84)

Li

gen

roun

timoun

so

tchò

toujou

Ø

tris

3sg

have

det

child

poss.3sg

heart

always

pfv

be sad

‘There is one child whose heart is always sad’

(85)

Pars

i

pa

gen

ti

boug

ki

because

3sg

neg

have

small

person

rel

lé

jwé

ké

li

   

want

play

with

3sg

   

‘because there are no children who want to play with him.’

(86)

Jako-a

rete

i

di:

“Non,

a

pou

Parrot-def

stay

3sg

say

no

3sg

for

mo

trapé

roun

rémèd

pou

mo

kolèg-a.

1sg

catch

det

cure

for

poss

friend-det

‘The parrot says, “No, I have to cure my friend.”’

(87)

I

ka

tris

tro

bokou.

3sg

prog

sad

too

much

‘He’s so sad.’

(88)

A

pou

mo

fè

l’kontré

 ké

roun

moun

3sg

for

1sg

make

3sg-meet

com

indf

person

ka

ari,

pou

vé

géri

li.

 

prog

laugh

for

want

cure

3sg

 

‘He has to meet someone who is happy and who is going to cure

him.’

A similar interpretation holds for the verb krè, “to believe”. In our corpora, we usually find this verb with Ø-marking. However, if believing is somehow momentary and linked to an ongoing action, it is marked by ka . Here, the speaker observes someone being saved from death, and in this very moment, he believes in God:

(89)

Mo

ka

we

sa,

mo

ka

krè

an

bondyé

1sg

prog

see

that,

1sg

prog

believe

in

God

‘I observe that I believe in God’

In FGC, even seemingly objective facts, such as a tree that is tall or a house that is far away, can be combined with the imperfective marker if the tree appears to grow taller upon climbing it or the house appears to become farther away upon walking towards it. The following examples should clearly be interpreted progressively:

(90)

Ti

boug-a

Ø

monté

i

Ø

monté,

Little

person-det

pfv

climb

3sg

pfv

climb

i

Ø

monté

    

3sg

pfv

climb

    

‘The guy climbed and climbed and climbed…’

(91)

Amizu

ti

boug-a

ka

monté,

More

little

person-det

prog

climb

amizu

pyébwa

ka

rot

 

more

tree

prog

high

 

‘…(but) the higher he climbed, the higher the tree grew.’

This is obviously more processual: the higher the hero climbs (amizu ti boug-a ka monté), the higher the tree appears to grow (ka rot ‘grows in height’; literally “it highs”) (Tchang 1988: 54.2; cf. Migge 2000: 216–217 for similar observations, comparing Ndyuka and Gbe).

(92)

Koté

Francine

i

Ø

lwen

side

Francine

3sg

pfv

be far

‘Francine lives far away.’

(93)

pou

alé

koté

Francine

i

ka

lwen

in order to

go

side

Francine

3sg

prog

be far

‘It’s a long way to go to Francine’s.’

While the distance in (92) is given without reference to the context, the action is emphasised in (93), which can be paraphrased as follows: ‘From the place you find yourself in at the moment of speaking, you have to walk for a long time in order to get to Francine’s.

In conclusion, our analysis confirms Saint-Jacques Fauquenoy’s (1972) observation that the distinction between stative and dynamic verbs may be crucial in MAC (and in other Caribbean creoles), but not in FGC. All verbs, including de-adjectival and denominal verbs (which are often called stative), can express progressive meanings, just as dynamic verbs do. This is very much in line with the findings presented in Migge (2000) and Migge and Winford (2007) on Surinamese Creoles.

Finally, in a usage-based linguistics framework, the aspectual perspectives of stative and dynamic verbs can be explained by experience: the default use of actions is in narratives, when they have happened; states are by default considered non-perfective.

3.3.5 Completive Aspect: Fin (and Kaba)

If a speaker wants to insist that an action has in fact been completed, he/she can use the marker fin or kaba .

(94)

Lò

yé

fin

koupé

yé

bati…

When

3pl

finish

cut

3pl

slash

‘When they had finished cutting down trees…’

(95)

Hm

mè

to

kontan

to

bati

fin

planté.

Hm

but

2sg

like

2sg

slash

finish

plant

‘yes, you were happy that your land had been planted.’

Fin has been said to derive from French finir de, which has similar functions in several varieties of French. However, in standard French, it tends more towards the lexical meaning of ‘to finish’, whereas in FGC, fin is clearly functionalised and thus part of core grammar (completive aspect ).

In the nineteenth-century novel Atipa, kaba, spelt caba, is best translated by ‘already’, as all the following examples illustrate (we skip the glosses in the following examples because only the global meaning of the often complex sentence is what interests us here):

(96)

Et to meinme a dé faim mo faim. Mo vente ca fait kioume caba.

‘As for me I am dying of hunger. My belly is already rumbling.’ (Parépou 1885: 38)

(97)

Mouché Chapy fait divin caba à Kourou et bon divin encò.

‘Mr. Chapy has already made some wine in Kourou wine of very good quality.’ (Parépou 1885: 66)

(98)

Sénato zòte ca doumandé la; blangue ca peyé oune moune caba qui pou ca fait engin ou crai yé wa wlé peyé dé pou bon kiò conça?

‘You’re calling for a new Senator! The whites are already paying for one who doesn’t do anything; do you think they will pay for two just like that for nothing?’ (Parépou 1885: 82)

(99)

Ça qui allé la danbois qué Apatou la a pas soumaqué li dépensé pou bon kiò? Ça li dit mouché Louprie palé di ça caba.

‘The man who went in to the forest with Apatou, didn’t he spend money in vain? Mr Le Prieur had already talked about what the man had said.’ (Parépou 1885: 174)

Twentieth-century uses of kaba show a growing tendency to express completive aspect . The contextual reading is often that of reassurance:

(100)

I

vini

kaba

3sg

come

compl

‘He did come.’ (‘He surely was there.’)

(101)

I

té

vini

kaba

3sg

pst

come

compl

‘He had come.’ (‘for sure’)

Kaba is a postposed marker derived from Portuguese. According to Horth (1948), it was used with the meaning of “already” and/or “completed action” in vernacular or pidgin Portuguese on the Brazilian border with French Guiana.

The postposed aspect marker -kaba is also documented in Papiamento and classified as a perfect marker (Andersen 1990) or a pluperfect marker (Maurer 1998). Kaba has its possible sources in Portuguese:

(102)

Ele acaba de chegar em casa.

‘He has just arrived home.’

(103)

Os estudantes acabam de fazer o exame final.

‘The students have just written their final exam.’

Note that kaba is found with a similar or identical meaning in other creoles with a Portuguese lexical base, Negerhollands (which has mainly a Dutch lexical base), as well as in Sranan (which has mainly an English lexical base; cf. Stolz 1987: 300). Winford and Migge have argued in a very similar vein “Completive kaba in the Surinamese creoles is modeled after the VP-final marker of completion common to both Eastern and Western Gbe languages” (2007: 14).

3.3.6 Té, the Only Tense Marker in FGC

Having considered several options of aspect marking in FGC, we now present the only tense marker in this language, té . In FGC past tense is expressed by the marker té, as in the following examples:

(104)

Timoun-yan

pa

té

gen

tan

jwè

children-def.pl

neg

pst

have

time

play

‘The children had no time to play.’

(105)

Mo

té

fè

mo

granlajan

1sg

pst

make

poss

big-money

‘I have made a lot of money.’

(106)

Mo

té

gen

oun

kopen

1sg

pst

have

indf

friend

‘I had a friend.’

(107)

Gas

té

antré

gas

pst

get in

‘It was filled with air.’

(108)

Alò

nou

té

alé

a

so

kaz

so

we

pst

go

at

poss

home

‘So we went to his house.’

In our corpus (cf. Appendix), the story “Nearly Drowned” focuses on a boat trip home from a friend’s place. The narrator sets the scene by providing background information. To indicate that the journey to the friend’s place lies prior to the story’s main narrative, the speaker marks verbs with the past-tense marker té .

The results from the corpus data are quite intriguing because in most, if not all other French-based Caribbean creoles, the marker té is labelled by fellow researchers as indicative of past-before-past actions. To confirm that our corpus findings for FGC indicate past (and only rarely past-before-past), we ran an elicitation test, asking speakers both of FGC and MAC to translate sentences. While Martinican informants translated the French pluperfect with té, the French Guianese informants translated the pluperfect with Ø, as illustrated below.

(110)

Mo

frè

Ø

alé

lave

larivyè

ayè

bonma(n)ten.

my

brother

pfv

go

wash

river

yesterday

morning

‘Yesterday morning, my brother went swimming.’

(111)

Mè

li

Ø

manjé

anvan.

Alò

i

pa

té

But

3sg

pfv

eat

before

so

3sg

neg

pst

(pou )vé

najé.

       

can

swim

       

‘But he had eaten before, so he couldn’t swim.’

In the second line of the following example, té in pa té ge is necessary to disambiguate between a reading in which the speaker refers to a present situation and one in which the reference is to the past. Both readings would be possible if zero marking were used instead of té.

(112)

Mo

di

piti

tout

k’alé

Kayenn

aprézan.

1sg

say

youth

all

ipfv.go

Cayenne now

‘I said that the boys are all going to Cayenne now.’

 

I

pa

gen

sa

isi

  

3sg

neg

have

dem

here

  

‘There isn’t anything like that here.’

(113)

Pa

té

gen

vrè

aprézan

timoun-yan

viré

neg

té

have

real

now

children -detpl

return

mé

sa

épok

i

doumandé

  

but

dem

time

3sg

ask

  

‘There really wasn’t anything. Now, the children are coming back (to the village), but at that time, they weren’t sure.’

If ambiguity between a present- and a past-tense reading is excluded (for example, when the surrounding discourse explicitly recalls past events), té does not appear. Interestingly, most of the uses of té in narratives do not refer to events within the story itself, but are instead meant as a type of comment on the story as a whole. This is true for the example above and also holds true for the following example:

(114)

Mo

déja

té

wè

bagaj

wi

pou

mo

ti

laj

1sg

already

pst

see

thing

yes

for

poss.1sg

small

age

‘I had already seen many things, you know, in spite of my young age.’

Té does occur sometimes if the narrator wants to situate the narrated events in time, as in the following excerpts:

(115)

Nou

rive

I

té

senk

ér

d’maten

1pl

arrive

3sg

pst

five

o’clock

morning

‘(When) we arrived, it was 5 o’clock in the morning.’

(116)

Lè

 

nou

rivé

nou

alé

dronmi

when

 

1pl

arrive

1pl

go

sleep

i

té

senkèr/

 

zizèr

d’maten

3sg

pst

five o’clock

 

six o’clock

morning

‘When we arrived, we went to sleep. It was five/six o’clock in the morning.’

The same usage can be found in school books (e.g. Holder 1988) and on the radio, as in the following examples:

(117)

Tipolo

té

gen

dizan

Tipolo

pst

have

ten.year

‘Tipolo was ten years old.’ (Holder 1988: 9)

(118)

Mo

té

gen

patché

fanm

1sg

pst

have

a lot

woman

‘I had a lot of women.’ (CIEL_FR_GF_RADIO)

(119)

A

té

mo

lavi

dem

pst

poss

life

‘This was my life.’ (CIEL_FR_GF_RADIO)

TMA particles may be combined. For example, té + ka marks past progressive tense/aspect :

(120)

Mo

té

ka

pale

ké

yé

1sg

pst

prog

talk

com

3pl

‘I was talking to them.’

Adverbs may be inserted between the two particles:

(121)

Misyé

té

souvan

ka

vini

isi-a

He

pst

often

prog

come

here- def

‘He often came to our place.’

Our investigation of té finds that the particle is used mostly to mark a shift in narrative structure rather than in tense . In this sense, it does not behave like the tense-focused marker té in Antillean Creoles.

3.3.7 Tense and/or Mood: The Markers Ké and Wa

Modern FGC usually employs ka or ké to express future actions, although Ø is sometimes used. The case is similar in Martinique , although variants of ké also need to be examined (kèy, kay). In the early years of its existence, FGC had a marker wa that is no longer attested to today: Saint-Quentin’s 1872 grammar states that wa was a future marker (1872: 133), and Horth says that wa was the archaic form of ké (1948: 28). In nineteenth-century texts, the two markers coexisted, wa being more frequently used. Wa expresses an “uncertain” future (e.g. i wa manjé, ‘he will eat’) in contrast with the “certain” future (e.g. i ké manjé, ‘he is going to eat’). Preceded by té , wa expresses irrealis: i té wa manjé, ‘he would eat’ or ‘he would have eaten’.

(122)

To

ké

pren

quichose

ké

nous;

2sg

fut

take

something

com

1pl

apré,

nous

wa

soti.

  

afterwards

1pl

 

fut

go out

 

‘You’ll have something [to eat] with us; then we’ll go out.’ (Parépou 1885: 158)

(123)

Nous

wa

mangé

tout

lé

dé.

We

fut

eat

both

def

two

‘The two of us are going to eat (together).’ (Parépou 1885: 44)

Ké did not emerge until more than a century after the development of FGC. Its origin and evolution are clearly shown in nineteenth-century FGC texts: An 1848 proclamation uses kalle and ké (Sournia 1976), and Saint-Quentin (1872) uses ké, k’é and k’alé. These variations show the marker ka preceding ale “to go”. Ka must therefore have developed before ké .

(124)

Mo

k’alé

proméné.

1sg

prog-go

walk

‘I’m going out for a walk.’ (Saint-Quentin 1872: 31)

In FGC, ké expresses a so-called certain future while ka expresses a less certain future.

(125)

Épi

nou

ké

alé

enstalé

St. Louis

Then

2pl

fut

go

stay

St. Louis

‘Then we will stay in St. Louis.’ (Holder 1988: 7)

(126)

To

ké

byen

anmizé

to

ké

wè!

2sg

fut

well

enjoy

2sg

fut

see

‘You’ll have lots of fun; you will see!’ (Holder 1988: 6)

(127)

To

pé

ké

pran

kou

2sg

can

fut

take

stroke

‘You might suffer.’ (CIEL_FR_GF_RADIO)

(128)

A

mo

ki

ké

gen

to

   

dem

1sg

rel

fut

have

2sg

   

mo

ka

di

mo

ké

mannyen

to

mo

ké

1sg

prog

say

1sg

fut

thrash

2sg

1sg

fut

mannyen to

thrash 2sg

‘I am the one who will defeat you, I declare that I’ll thrash you.’

FGC uses ka to mark scepticism as in the following example:

(129)

Mo

Ø

krè

li

1sg

pfv

believe

3sg

‘I believe it.’

(130)

Mo

ka

krè

li

1sg

prog

believe

3sg

‘I sort of believe it.’

How can we explain this assertive meaning of perfective zero marking and of future ké vs. the assertive meaning of progressive ka? One of the main characteristics of the imperfective (progressive) aspect is that it does not provide any information about the beginning or the endpoint of an activity or event. It has been claimed that the lack of information, particularly regarding the endpoint of an activity, leads to an affinity for the imperfective aspect for non-assertive meanings because it is not possible to be sure that an activity or an event took place if we do not know if, when, or how it was completed. This may explain why ka in FGC is both progressive and non-assertive, whereas it can express both the present tense and an assertive future in MAC. In this case, activities that express present tense relevance are thought of as having already occurred and thus contribute to the expectation that they will indeed take place.

3.3.8 Aspect in FGC: A Case of Functional Transfer?

In conclusion, in the previous sections on the verb phrase, we found a strong emphasis on aspect in French Guianese Creole. Except for the marker té , which is most often used in combination with the imperfective aspect marker ka to signify the past progressive or past imperfective, all other markers express aspectual meanings only (Ø, kaba, and fin) or have a combined tense–mood –aspect range of functions ( wa and ké). Quantitative analysis supports these findings: more than 87% of all verb forms in the 60 hours of the rainforest recordings are marked as perfective or imperfective. The dominant category in FGC is clearly aspect, not tense .

French has provided most of the forms of the FGC verb phrase, but it cannot be the origin of the system’s aspectual focus. The history in Chapter 2 revealed the importance of Gbe languages in the creation of the creole, and this section has demonstrated clear links between the FGC verb phrase and that of the aspectually oriented Gbe systems (cf. e.g. Dzablu-Kumah 2015: 51ff. for Ewe and Akoha 2010: 100ff. for Fon ). We can thus speak of a category transfer from Ewe to the nascent FGC; aspect is a marginally grammaticalised category in French but has become a core category in FGC, being expressed obligatorily in almost all utterances. In addition, the markers used within constructions expressing aspect seem to be refunctionalisations of French forms (as seen in Sect. 3.5.).

The question of whether syntactic strategy transfer was also involved in the process is more complex and requires the examination of three characteristics. First, the syntactic strategy of combining unchangeable verb stems with monomorphemic separate markers and mono- or di-syllabic markers that can be combined is absolutely parallel between FGC (first and third examples) and Ewe (second and fourth examples):

(131)

Mo

alé

kaba

1sg

go

compl

‘I have gone.’

(132)

Mè –yì

vò

1sg-go

compl

‘I have gone away.’ (Pasch 1995: 46)

(133)

Mo

ké

alé

1sg

fut

go

‘I will go.’

(134)

M-

á–yì

1sg-fut-go

‘I will go.’ (Pasch 1995: 43)

Second, there is no morphologically marked verb class in FGC or Gbe languages like Ewe . It might be preferable to speak of verboids rather than of verbs in FGC (cf. Ludwig 1996 for Guadeloupe). FGC gran and Ewe k mean ‘tall’ (ADJ) or ‘to be tall’ (V), depending on the context.

Third, if the predicate in an utterance behaves more like a verb than an adjective, its zero form denotes the perfective aspect both in FGC and in Ewe:

(135)

Mo

alé

1sg

go

‘I went.’

Mè-yi

1sg-go

‘I went.’ (Dzablu-Kumah 2015: 39)

However, if the predicate behaves more like an adjective, the zero form denotes the imperfective aspect (Dzablu-Kumah 2015: 51):

(136)

Mo

gran

1sg

big

‘I am tall.’

Atíá

k

tree

big

‘The tree is tall.’ (Dzablu-Kumah 2015: 4)

These three characteristics provide evidence for a possible syntactic strategy transfer from Ewe , although word order raises a problematic point because unlike in FGC, most Ewe aspect markers are postverbal (Ameka 1991; Pasch 1995; Dzablu-Kumah 2015). However, there seems to be clear constraint not to distort French word order or, in Andersen’s (1989) words, to find some place to transfer things to (cf. Sect. 3.4). In addition, we might want to look more closely at what the (L1, L2, L3) speakers of Gbe languages in French Guiana could identify as the imperfective aspect in their own language (the perfective aspect being unmarked, as we saw earlier). We may consider the following two examples from Pasch:

(137)

Dàdi

lá

lè

tsìtsì-ḿ

cat

def

prog

grow-prog

‘The cat is making itself appear bigger right now.’ (Pasch 1995: 44 for Ewe)

K.

lè

tó

  

K.

prog

go-out

  

‘K. is on his way downtown [going out].’ (Pasch 1995: 44, cf. Bole-Richard 1983: 156 for Gẽ)

Here, imperfective (or progressive) aspect is not necessarily expressed only by a single morpheme (postposed -ḿ), but sometimes additionally by reduplication (tsìtsì, lit. ‘grow-grow’). Furthermore, not every Gbe language uses reduplication or a postposed marker, suggesting that multilingual speakers of Gbe languages are not unduly disturbed by the positional flexibility of markers. Finally, the preposed progressive marker lè, considered an auxiliary in some grammars, is always present, even if it is the only marker (cf. Bolé-Richard 1983: 156 for Gê), which means it is the most phonetically salient part of the construction. Lè is invariable and is translated by Dzablu-Kumah as “to be somewhere”, a meaning that is a typical source of progressive aspect markers in grammaticalisation chains (cf. Heine and Kuteva 2002: 97).

The characteristics examined above provide evidence of syntactic strategy transfer from Ewe to FGC. They also prove that when considering what speakers might perceive as a possible candidate for a construction, it is important to look at not just spoken varieties of Romance languages (French and Portuguese ), but also spoken varieties of the Gbe family (Ewe, Gê, Fon , etc.).

The forms used in FGC to express aspect and tense seem to all be of French (or Portuguese) origin. However, it is more difficult than in Sects. 3.2 (noun phrase) and 3.4 (sentence structure) to determine the exact source of the tense and aspect markers that have led to functionalisation or, more often, to refunctionalisation in FGC. While things are quite clear for fin (< finir de), ké (ka + aller, infinitive and/or participle form of ‘to go’) and wa (< va, third-person singular of ‘to go’), the origin of kaba , ka and té needs to be examined more closely in future research.

If we assume a similar development for the obsolete FGC future marker wa (> French va 3SG), we might want to hypothesise that the French 3SG était, ‘was’, could have been a source for the FGC marker té . We might also take French 1/2SG and 3PL (étais, ‘was’, and étaient, ‘were’) and maybe even the past participle été (‘been’) into account as convergent sources since they may have sounded similar in the varieties of French spoken during the early years of FGC’s development. We would then have another (slightly more complex) case of refunctionalisation via the extension of one or several forms of a morphosyntactic paradigm into the only marker of past-tense reference (for all verbs, all persons, singular and plural). The source of the imperfective marker ka might be the most hypothetical case, and we would assume an origin of the kind attested to today, for example, on the Caribbean island of Saint-Barthélemy, namely il n’est qu’à faire, ‘s/he is doing’; however, this hypothesis clearly needs further research.

In conclusion, we have found strong emphasis on aspect in FGC in the VP section of this study. This emphasis is not explained by French, even if we concede that spoken French may have more aspect marking than written French, but it is explained by Ewe both historically (Chap. 2) and grammatically (Sect. 3.3). Our explanation of how the system evolved from Ewe to FGC is shown in the successful testing of the hypothesis of transfer (here, category and strategy transfers).

3.3.9 Modality: Can and Must

In FGC, deontic modality (i.e. obligation) is expressed by fo (> il faut (faire), an impersonal form of obligation, ‘it has to be (done)’) and bezwen (> French avoir besoin, ‘to need’).

(138)

Aswè-a

nèg

fo

to

ba

l’

tonight-def

man

have to

2sg

give

3sg

oun

bon

ti

bagay…

  

indf

good

little

thing

  

‘Tonight, my friend, you will have to give him something…’

(139)

Nou

bezwen

jité

bwa

we

need

cut

wood

‘We have to cut the trees.’

(140)

Démen

yé

ké

bezwen

yé…

tomorrow

3sg

fut

need

3sg

‘Tomorrow… they will be the ones who need you.’

In FGC (though not in other French creoles, such as Martinican Creole ), there is yet another deontic marker: pou (‘for’).

(141)

Pou

mo

bout

for

1sg

finish

‘I have to finish.’

It is important to note that in FGC, deontic pou is always at the head of the phrase:

(142)

Pou

mo

alé

koté

doktèr

for

1sg

go

by

doctor

‘I should go to the doctor.’

(143)

Pou

to

fé

topa

bati

for

2sg

make

poss

farm

‘You should run your own farm.’

(144)

Pou

li

komprann

for

3sg

understand

‘He should understand.’

In several French creoles, pou carries future or modal meaning that can be metonymically close to deontic meanings. This use is also attested in older FGC texts:

(145)

Zòt

pou

fè

2pl

for

do

‘You should do it.’ (Sournia 1976: 7)

However, this final example is different from modern use in its placement of pou. As indicated above, current FGC places pou before the subject pronoun:

(146)

Pou

zòt

fè

for

2pl

do

‘You should do it.’

A concept such as reanalysis may explain the use of pou, “must”, as a marker of deontic modality . As Lang and Neumann-Holzschuh (1999: 7) state, they use the term reanalysis “when a listener who has access to the structural order of the utterance gives the linear order a structural interpretation different from that implied by the speaker”.Footnote 2 We should add that reanalysis can lead to “rebracketing” (Haspelmath 1998: 330) as well as to “category relabelling” (Harris and Campbell 1995: 63; for detailed discussions with explicit reference to creole languages, cf. Neumann-Holzschuh and Schneider 2000; Kriegel 2003; Detges 2003; Neumann-Holzschuh 2003). A possible candidate for a reanalysis is the construction “in order to”. To explain the emergence of this peculiarity of FGC, we may assume that the following reanalyses have occurred (cf. Pfänder 2003):

(147)

[I

nésésèr

pou

mo]

alé

koté

dokteur

3sg

necessary

for

1sg

go

side

doctor

‘It is necessary for me to go to the doctor.’

(148)

I

nésésèr

[pou

mo

alé

koté

dokteur]

 

3sg

necessary

for

1sg

go

side

doctor

 

‘It is necessary that I go to the doctor.’

From the domain of necessity, we now move on to possibility. In FGC, pouvé (and allegro form vé) > fr. ‘pouvoir’ is used both for root (‘be able to’), deontic (‘be allowed to’) and epistemic modality (‘it might well be that…’). Thus, depending on the context,

(149)

i

pouvé

alé

3sg

can

go

can mean ‘he is able to walk’, ‘he is allowed to go’ or ‘maybe he can go’. Sometimes the context makes it clear that we are dealing with root modality meaning “be able to”:

(150)

Mè

nou

pa

pouvé

alé

but

1pl

neg

be-able

go

piske

 

pa

gen

kannon

because

 

neg

have

boat

‘We were not able to go, because there weren’t any boats.’

(151)

I

ka

bas,

la

la

a

louvri

pou

louvri

3sg

prog

low

these

def

def

open

for

open

ou

janmb

pou

pouvé

tire

mannyòk

   

poss

legs

for

be-able

pull

manioc

   

‘It grows low; therefore, you have to move your legs out of the way to be able to pull up the manioc.’

If, on the other hand, the context displays markers of uncertainty, an epistemic reading is preferred by most speakers, as in the following example:

(152)

Dimen

si

bondyé

lé

pitèt

i

pouvé

alé

tomorrow

cond

God

will

maybe

3sg

can

go,

m’

pa

save

     

1sg

neg

know.’

     

‘Tomorrow, if God wills, maybe he can go; I don’t know.’

Syntactic reorganisation (the position of the aspectual or temporal particle) enables semantic differentiation between epistemic and non-epistemic meaning. Therefore, in the following examples:

(153)

I

divèt

té

ka

dronmi

3sg

must

pst

prog

sleep

‘Maybe he was sleeping.’

(154)

I

divèt

ka

travay

3sg

must

prog

work

‘Maybe he is working.’

only an epistemic interpretation would be a natural choice in FGC (Peyraud 1983: 135).

3.3.10 Imperative

Imperative forms tend to be marked by Ø in FGC (first example below), their negative form by ka (second example):

(155)

Fromen

zòt

bouch

close

poss

mouth

‘Close your mouth.’

(156)

Pa

ka

ouvri

zòt

bouch

neg

prog

open

poss

mouth

‘Do not open your mouth.’

As far as we know, this has not been reported for Martinican Creole .

Different morphological markers on imperatives and negated imperatives are quite common. Marking the negative imperative with the less assertive form is in this respect similar to languages like Spanish , in which the negated imperative is marked by the subjunctive mood :

(157)

Dime

‘Tell me.’

(158)

No me digas

‘Do not tell me.’

Furthermore, FGC (like other creoles) makes good use of mitigated, so-called hortative imperatives:

(159)

Annou

pati

imp

leave

‘Let’s go.’

In Ewe , we find a similar structure:

(160)

Na

mí

dzo

give.2sg

1pl

leave

‘Let’s go.’ (Ameka 1991: 52)

3.4 Sentence Structure

3.4.1 Copula Constructions

Very much like English , French has one copula, whereas FGC resembles Ewe, with three different copula constructions.Footnote 3

First of all, in simple sentences with a predicate adjective, FGC uses a zero-copula strategy, just like Ewe, as illustrated in the following examples:

(161)

FGC:

I

gran

 

3sg

tall

 

‘He is tall.’

 

(162)

Ewe:

É

kɔ̌

 

3sg

tall

 

‘He is tall.’ (Dzablu-Kumah 2015: 4)

Second, if the predicate involves a noun, the copula sa is used in present tense contexts:

(163)

I

sa

gran-grèk

3sg

cop

professor/researcher

‘She is a professor at university.’

while the copula té is used in past-tense contexts:

(164)

Misyè

té

doktér

avan

He

pst

doctor

before

‘He was a doctor before.’

This again is similar to Ewe , in which different forms of the copulas lè and nò are used in similar present- (lè) and past-tense (nò) contexts.

Third, there is a separate copula for constructions that indicate the particular location of a person or an entity, as illustrated here:

(165)

I

fika

Kayèn

3sg

cop

Cayenne

‘He is living in Cayenne.’

This type of spatial copula also exists in Portuguese. While all other copula forms, including sa and té , come from French, the locative copula fika is transferred from Portuguese , in which it seems to have the same functions as in FGC:

(166)

Porto Allegre fica no sul do Brasil.

‘Porto Allegre is located in the South of Brazil .’

(167)

O professor Pereira fica mais dois anos na Universidade de São Paulo.

‘Professor Pereria will stay at the University of São Paulo for two more years.’

(168)

Hoje Anderson fica em casa.

‘Anderson is staying home today.’

(169)

A Juliane vai ficar dois meses em Berlim para estudar alemão

‘Juliane will stay in Berlin for two months to learn German.’

However, there is one function that has not been transferred to FGC:

(170)

Ele ficou surpreso.

‘He was surprised.’

We thus analyse the use of FGC fika as a slight refunctionalisation (reduction of functional range) of a Portuguese morpheme—a reduction that possibly took place due to Ewe influence.

3.4.2 Negation

In FGC, the negation marker is pa, ‘not’. The double negative construction from French has not been transferred to FGC. Another important divergence between FGC and French can be found in the positioning of the negation marker pa. Negation is expressed by the particle pa, which precedes all tense–aspect–mood markers but follows the subject:

(171)

Pa

gadé

déyè

neg

look

back

‘Do not look back!’

(172)

I

pa

gen

frè

3sg

neg

have

brother

‘He has no brother.’ (Holder 1988: 6)

(173)

Si

to

pa

ka

fè

cond

2sg

neg

prog

make

‘if you don’t do it.’ (CIEL_FR_GF_RADIO)

(174)

Mo

pa

savé

1sg

neg

know

‘I don’t know.’ (CIEL_FR_GF_RADIO)

(175)

Nou

pyès

pa

wè.

1pl

feet

neg

see

‘We don’t see.’ (CIEL_FR_GF_RADIO)

(176)

To

pa

aksepté

hein

to

ka

toléré.

2sg

neg

accept

right

2sg

prog

tolerate

‘You don’t accept (it), no; you tolerate it.’ (CIEL_FR_GF_RADIO)

(177)

Sa

yé

pa

ka

wè

pa

ka

fè

ou

mal

dem

eye

neg

ipfv

see

neg

ipfv

make

2sg

bad

‘What the eye does not see, the heart does not grieve over. [What they don’t see, doesn’t hurt them]’ (CIEL_FR_GF_RADIO)

(178)

To

pé

pa

fè

sa

2sg

can

neg

make

dem

‘You can’t do that.’ (CIEL_FR_GF_RADIO)

Unlike in French or English , all tense and aspect markers in FGC are placed between the negation marker and the main verb, as illustrated in examples below:

(179)

Mo

pa

gadé

déyè

I

neg

look

back

‘I do not look back.’

(180)

Mo

pa

ka

gadé

déyè

I

neg

prog

look

back

‘I am not looking back.’

(181)

Mo

pa

té

gadé

déyè

I

neg

pst

look

back

‘I had not looked back.’

(182)

Mo

pa

té

ka

gadé

déyè

I

neg

pst

prog

look

back

‘I was not looking back.’

In most Gbe languages, the negation marker is combined with tense and aspect markers, which are also placed in between the negation marker and the main verb. Thus, at first glance, negation in simple sentences may appear to be a contact-induced innovation in FGC. However, if we take informal spoken French into consideration, the particle pas also mostly occurs preverbally due to periphrastic constructions:

(183)

FRENCH

      

FGC

   

Je

(ne)

parle

pas.

   

Mo

pa

ka

palé

I

neg

speak

neg

   

1sg

neg

prog

speak

(184)

Je

(ne)

vais

pas

parler

  

Mo

pa

ké

palé

I

neg

go

neg

speak

  

1sg

neg

prog

speak

(185)

Je

(n‘)

ai

pas

parlé

  

Mo

pa

Ø

palé

I

neg

have

neg

spoken

  

1sg

neg

pfv

speak

What is nevertheless truly innovative is the possibility for tense and aspect markers to appear between the negation marker and the predicate in FGC.

Negative indefinite pronouns occur with the negation particle:

(186)

Mo

pa

wè

pesòn

1sg

neg

see

nobody

‘I did not see anybody.’

In the respective Standard French phrase Il ne parle pas, we observe that the etymologic NEG ne precedes the full verb, and the lexeme ‘step’ pas is lexicalised as the right negation bracket. Again, the oral input shows otherwise: the NEG ne /n/is frequently omitted, and in the cases in which it is realised, it is far less salient, meaning that /n/is more difficult to identify in the sound chain than pas /pa/. Apart from the phonetic argument, there is also a syntactical explanation. In French, pas is placed after the inflected verb. If there is an auxiliary plus a full verb, pas appears after the finite auxiliary and before the non-finite full verb in its form as an infinitive or participle: Il n’a pas parlé or Il ne va pas parler. This means that in terms of perception, pas must have been audible before the verb in the input of early Cayenne society.

Word-order transfer from Ewe can be found to be successful in those cases for which the word order in question exists as a pragmatically heavily constrained structure in spoken French. A postposed intensifier, as in He is crazy, like, completely, for example, enables transfer from Ewe, where intensifiers occur in postposed position. Although the underlying structure in French involves reconstruction and is thus different from the underlying structure in Ewe, it can be perceived as a suitable target structure. Surface resemblance to Ewe wins out against frequency. Even if the French target structure is pragmatically marked and used infrequently, perceived similarity to an Ewe structure encourages word-order transfer. In other words, the idea is that speakers intend to use French word-order patterns but involuntarily gravitate towards those patterns that are similar to Ewe structures, even if they are infrequent and pragmatically constrained.

3.4.3 Predicate Doubling

Many structures of FGC have a repeated element that can work as a focus strategy and an intensifier:

(187)

A

pati

Georges

 

ka

pati

dem

leave

Georges

 

prog

left

‘Georges left as if he were running for his life.’

(188)

A

chanté

yé

té

ka

chanté

dem

sing

3pl

pst

prog

sing

‘They were singing out loud.’ [literally: ‘It is singing; they were singing.’]

While there is a very productive process of verb doubling embedded in a presentative or cleft construction with the particle a, ‘this is’ (from an Arawak etymon a, ‘this/this is’; see Pfänder 2000a), the underlying origin is quite likely to be Ewe , in which more than 8% of words occur in a reduplicated form (Ameka 1999: 78). Repeated words in focus strategies are common:

(189)

Sí

Kofí

sí

Escape

Kofi

escape

‘Kofi ran away as fast as he could.’ (Ameka 1991: 45)

A common superlative construction in FGC involves repetition of the relevant noun phrase and a clause-final fèt, apparently derived from fait, the past participle of the French verb faire, ‘to make’. Note that the final consonant of fait was pronounced in many patois and regional varieties of French, so fèt does not necessarily trace its origin from the feminine form.

(190)

To

pè

pasé

pè

fèt

2sg

fear

more

fear

made

‘We [one] had never been so scared.’

(191)

Madanm

troumantan

pasé

troumantan

fèt

Woman

troublemaker

more

troublemaker

made.

‘She is the biggest troublemaker (I have ever met).’

The superlative structure is not from French, however. Ewe uses a similar repeated form for superlatives, in which “the two NPs are linked by the possessive connective fé”:

(192)

Wó-be-na

dé

wó-fé

xɔ-wó

me

le

3pl-hide-hab

all

pl-poss

room-pl

containing.region.of

at

gogloefé fé

gogloefé-wó

    

deep.part poss

deep.part-pl

    

‘They hide in their rooms in the most hidden part of all places.’ (Ameka 1999: 89)

 

Repetition also occurs very frequently in narratives, as is apparent in many of the stories transcribed in Appendix:

(193)

Nou

pè

nou

pè

nou

pè

3pl

fear

3pl

fear

3pl

fear

‘We were very afraid.’

(194)

Mo

ajounou

 

mo

ajounou

1sg

kneel

 

1sg

kneel

‘I knelt down, I knelt down.’

The many instances of repetition might show transfer from Ewe , since they are not attested, at least not all of them, in French or another Romance language.

3.4.4 Comparative Marking

In French, there is variation between two comparative patterns. This is similar to English , where we find variation between comparative marking with more, comparative marking with –er, and suppletion: This event is more scandalous than the last one. /This event is cheaper than the last one. /This event is better than the last one. In French, variation can be found between comparative marking with more + positive adjective and suppletion. There is no morphologically marked comparative in French, and none of these syntactic strategies is attested to in the Gbe language family. Comparative marking in Gbe is illustrated in the example below:

(195)

Atíglínyi

lolo

wú

to

elephant

big

compar

buffalo

‘The elephant is bigger than the buffalo.’ (Dzablu-Kumah 2015: 25)

This structure in Ewe may have inspired the structure in FGC that is illustrated in the example below:

(196)

Zarb

gran

pasé

kaz

tree

big

surpass

house

‘The tree is bigger than the house.’

In the rainforest recordings, the verb pasé , ‘surpass’, is selected by default in order to mark the standard:

(197)

Georges

bel

pasé

Stéphane

George

handsome

surpass

Steven

‘George is more handsome than Steven.’

(198)

Mo

ké

mo

fè

oun

chanté

pou

to

1sg

fut

1sg

make

indf

sing

for

2sg

Pasé

 

mo

jouré

to

   

Surpass

 

1sg

insult

2sg

   

‘I will create a song about you. (This is) better than insulting you.’

(199)

Est-ce li gain langue, qui millò, pou ça, passé criole?

‘Is there any language that is more beautiful than Creole?’ (Parépou 1885: 12)

Comparative marking in FGC exhibits features of both Transfer Type 1 (functionalisation) and Transfer Type 3 (syntactic strategy). The lexeme surpass now functions as a grammatical marker of comparison within the complex construction A adjective surpass B, which was transferred from Ewe .

3.4.5 Comitative and Additive

In French, the comitative avec (‘with’) and the additive et (‘and’) are different categories expressed through different markers, but in FGC, these categories are fused and both are expressed via comitative ké (‘and, with’ < French avec ‘with’):

(200)

Mo

Ø

vini

ké

mo

fanmiy.

I

pfv

come

with

poss.1sg

family.

‘I came with my family.’

This extension of the functional range of ké (‘with’) might well be inspired by the Ewe form klpé which also combines comitative with additive (Dzablu-Kumah 2015: 14).

(202)

Kofí

kplé

Ama (Ameka 1991: 59)

Kofí

com

Ama

‘Kofi and Ama’

3.4.6 Nominal Coreference

Coreference between the subject and object can be expressed via object omission for body care and grooming verbs:

(203)

I

lavé

3sg

wash

‘S/he washes.’ or ‘S/he washes her/himself.’

Nominal coreference can also be established via the body (so kò, literally ‘his/her body’):

(204)

Kannon-a

anni

tounen

anlè

so

kò

boat-def

just

turn

on

poss

body

‘The boat just turned over.’

(205)

Tipolo

ka

santi

so-kò

tou

tris.

Tipolo

prog

feel

himself

totally

sad

‘Tipolo always felt sad.’ (Holder 1988: 6)

(206)

Tipolo

sere

so

kò

tou

piti

 

Tipolo

make

poss.3sg

body

very

small

 

annan

ti

kwen

    

in

little

corner

    

‘Tipolo is hiding away in a corner.’ (Holder 1988: 6)

or via body parts (e.g. so tchò, literally ‘his heart’):

(207)

Mo

té

di

mo

tchò

I

said

to

poss.1sg

heart

‘I said to myself’

(208)

A

oun

moman

mo

té

di

mo

tchò.

at

indf

moment

1sg

pst

from

poss

heart

‘At some point, I will say to myself [lit. ‘to my heart’] /I told myself.’ (CIEL_FR_GF_RADIO)

(209)

Sa

ka

fè

to

chò

mal.

dem

prog

make

poss.2sg

heart

sick

‘This hurts you [lit. This is making your heart sick].’ (CIEL_FR_GF_RADIO)

In FGC, these expressions are no longer lexical, and we tend to speak of incipient grammaticalisation . The fact that they are written as one or two words and translated by native speakers as ‘myself’, ‘my body’ or ‘my heart’ provides solid evidence for this conjecture; however, further research is clearly needed here. In Ewe , we have a self-based construction, as in English (cf. Dzablu-Kumah 2015: 143), so there is no functional transfer from Ewe here, but rather innovation or transfer from other contact languages.

Reciprocal relationships are expressed with the reciprocal pronoun kompagnen, ‘each other’ (originally ‘friend’, < French compagnon):

(210)

Wonm-yań

tchoué

yé

kompannyen

Men-def

kill

3pl

recp

‘They killed each other.’

This syntactic strategy may be a transfer from the very similar Ewe strategy:

(211)

Ŋútsuáwó

wu

wó

nɔewó.

Men-def

kill

3pl

recp

‘The men killed each other.’ (Dzablu-Kumah 2015: 146)

Again, we can speak of weak or incipient grammaticalisation of reciprocity here.

3.4.7 Wh-Questions

The interrogative phrase in FGC is fronted in content questions:

(212)

kimoun

to

wè?

Who

2sg

see

‘Whom did you see?’

In French, each wh-question is introduced by a specific lexical item (where, who, what). FGC did not inherit those interrogative pronouns, although they should have been part of everyday conversation. Transfer (morphosyntactic strategy , and in this case, also syntactic transparency via analogy) thus trumps frequency. In FGC, innovative bipartite interrogative pronouns are used. They are based on a lexicalisation process that involves a noun, such as side, site, person or thing, combined with an unmarked interrogative pronoun, ki.

(213)

ki-koté

which-side

‘where’

(214)

ki-moun

which-person

‘who’

(215)

ki-bagay

Which-thing

‘what’

This seems to be due to transfer from Ewe , where we can find the same strategy that combines a noun with an interrogative pronoun.

(216)

afí-ka

here-which

‘where’ (Pasch 1995: 80)

(217)

ame-ka

person-which

‘whose’ (Ameka 1991: 176)

(218)

nú-kà

thing-which

‘what’ (Pasch 1995: 79)

Nevertheless, in some cases, the interrogative pronoun ki is omitted, and the noun itself is used as the interrogative pronoun.

(219)

“Mé

to,

frè!”

dit

Atipa.

“Côté

to

soti?“

but

2sg

brother

say

Atipa

side

2sg

go out

‘“There you are, brother,” said Atipa. “Where did you come from?”’ (Parépou 1885: 14)

(220)

Koté

sa

timoun-yan

k’alé?

where

dem

child.indf.pl

go-prog

‘Where are all those children going?’ (Holder 1988: 5)

(221)

Koté

zot

ka

alé…

annan

chan

lougarou-a?

where

2pl

prog

go

into

field

werewolf.def

‘Where are you going? Into the “werewolf field”?’

In these examples, koté is used as an interrogative pronoun meaning where. In our corpus, koté is also used to indicate a place where someone lives (koté docteur, ‘at the doctor’s; koté koumèr, ‘at the neighbour’s) and as a locative adverb (toutkoté = ‘everywhere’). From a usage-based perspective, the fusion process still seems to be underway. One piece of evidence can be seen in the written depiction of these items as one or two words.

Thus far, we have discussed a syntactic strategy transfer for the interrogative pronouns; however, there is another syntactic strategy transfer to report, namely the emergence of a morphosyntactically realised question marker:

(222)

Kisa

I’

fè

a

what

3sg

do

q

‘What did he do?’ (CIEL_FR_GF_RADIO)

The function (and form) is quite similar to what has been noted for Gbe languages, for example, Ewe and Fon :

(223)

Kofí

dzó

a?

Kofí

leave

q

‘Has Kofí left?’ (Ameka 1991: 54)

(224)

É ká

ná

sìxú

kpé

dī

à?

It mod fut can be-enough now q

‘Could that be enough now?’ (Fon, cf. Akoha 2010: 105)

Whether this is a clear case of functional transfer or whether the whole form–function unit is transferred here would be a worthy subject of later research. Even if the forms appear similar, they could also be functionalised allegro forms of the independently existing pragmatic markers, which have been discussed for the French language by Wiesmath (2003) and Ludwig and Pfänder (2003) and for Ewe by Lafage (1985) and Ameka (1991).

A third and again very clear case of transfer of the mophosyntactic strategy in the realm of question marking can be found in the construction (literally) that who/those who, often best translated by ‘whoever’, which we find in both FGC (225) and Ewe (226):

(225)

Sa

 

ki

douvan,

rantré

dem

 

rel

in front,

turn

‘Whoever is in front of me, turn around.’

(226)

ame-si

person-who

‘whoever’ (Dzablu-Kumah 2015: 132)

3.4.8 Relative Clauses

In FGC, the relative clause is marked either by preposed relative pronoun ki, as in the following example

(227)

Mo

savé

to

gen

oun

frè

ki

 

1sg

know

2sg

have

indf

brother

rel

 

pa

té

annan

sa

    

neg

pst

into

dem

    

‘I know that you have a brother who wouldn’t do that.’ (CIEL_FR_GF_RADIO)

or by ki and additionally a postposed right bracket a, as in this example:

(228)

Mèr-a

ki

té

la

a

té

roun

mayor-def

rel

pst

there

rel

pst

indf

natif

Saül

     

born

Saül

     

‘The mayor who was there had been born in the small town of Saül.’

This use of a as a right bracket of a relative clause might be analysed as the functionalisation or refunctionalisation of the pragmatic tendency found in many varieties of French (and used to emphasise the definite article le/la) and as restrictive relative clauses with a right bracket là (cf. Ludwig and Pfänder 2003). In some cases in French, là functions as a right bracket for a larger noun phrase or a relative clause. In many varieties of French on the African Continent (as well as in America and Europe), we find a right bracket of larger syntagmatic unit, often a relative clause, as in the following examples:

(229)

lui qui vient là (Ploog 2000: 1423/217)

‘(the guy) who came LA’

(230)

(les) jeunes là qui venaient là (Martel and Beauchemin 1973: 22)

‘the young people LA who came LA’

(231)

celle-là qu’a fait le canal-là (Stäbler 1995: 73)

‘the one who built the canal LA’

(232)

le truc que tu m’as passé là (Ludwig 1996: 319)

‘the thing you gave me LA’

(233)

les petites affaires rouges t’as mangées là (Canada, Wiesmath 2000)

‘the little red things that you ate LA’

It is important to note that while in French, these right-bracket constructions are only a tendency—a routine form of structuration from a pragmatic point of view—their creole equivalents are part of creole core grammar.

In order for this to happen, transfer from Ewe cannot be excluded. Ameka (1991: 267) reports strikingly similar uses of “information structure” as he has it, in Ewe:

(234)

ga

si

ne`

ɖó ́

ɖe ́-

m

lá ́

 

money

rel

2sg

send

to

1sg

tp

 

’the money you sent me’ (Ameka 1991: 267)

Lafage (1985: 399; 406–407) reports a postposed marker –là with falling intonation that has a function of “modalité de precision” (a “precision marker”) and a final particle –lá with rising intonation (note the diacritic), which the author labels “particule finale de subordonnée” (a “final subordinating particle”) (1985: 399).

3.5 Results and Discussion

In this chapter, we have analysed important features of FGC core grammar in the light of four transfer types : functionalisation, refunctionalisation , morphosyntactic strategy and (semantic) category. In the following sections, we examine each strategy, provide a list of all the features we identify as belonging to each type and discuss the implications of these findings for the dichotomy of innovation and inheritance in the emergence of FGC.

3.5.1 Revisiting the Four Types of Transfer

3.5.1.1 Functionalisation

During functionalisation, or functional transfer Type 1 (see Sect. 3.1.4), a lexeme acquires grammatical function. While the form is clearly inherited, the function is innovative (from lexical to grammatical). In FGC, functionalisation is maybe best illustrated using the verb surpass as a comparative marker, as in Georges gran pasé Steven, literally: Georges tall surpass Steven, ‘Georges is taller than Steven’. Another functionalisation process can be found in the usage of misyé and madanm, which serve as personal pronouns (3SG) in FGC. We find a similar process in Brazilian Portuguese , where the lexeme a gente (‘the people’) is used as the personal pronoun we. A further example of functionalisation has been described in the fusion of the personal pronoun mo (mon ‘my’) with the lexeme pa (part ‘part’). When it comes to verbal phrases, we again find instances of functionalisation. The emergence of a reflexive pronoun from nouns referring to the body or body parts, for example, is a very common functionalisation process that can be considered an areal characteristic of West African languages even if Ewe is an exception that uses a postposed word meaning ‘oneself’. The French reflexive pronoun is preposed so speakers of West African languages in French Guiana may have grammaticalised the expression ‘son corps’ (‘his/her body’) to fill the gap where they expected to find the pronoun. In FGC, the combination of the personal pronoun so and the lexeme kò has taken on the grammatical function of a reflexive pronoun in phrases such as i ka tchoué sokò (‘he killed himself’). Other body parts can be used in the same way, as we can see in so tchò in FGC.

Our analysis suggests that functional transfer Type 1 is crucial for an in-depth understanding of the creation of a new language such as FGC. When compared with the structure of standard French, these structures appear highly innovative. From a more general perspective, however, these structures are not necessarily surprising as they follow well-established pathways of grammaticalisation (cf. Heine and Kuteva 2002). Siegel (2012), though, argues that we should not consider them as such because grammaticalisation processes take a long time. If at all, grammaticalisation processes should then bridge the gap between two languages—that is, they may begin in one language and continue in the contact language (cf. Ludwig and Pfänder 2003). This issue needs to be addressed in further research.

3.5.1.2 Refunctionalisation

Refunctionalisation has been shown to be a critical strategy in contact-induced dialect change, such as in Andean Spanish and in varieties of English in the Pacific. In FGC, we have found refunctionalisation from French to be less productive than functionalisation, which is in line with Siegel’s (2015) findings. Refunctionalisation is not as productive in creolisation as in contact-induced dialect change (e.g. in Andean Spanish ) mostly because input from French is less complete and less intense. In the case of incomplete input, the lexicon is more readily accessible than are possible (grammatical) candidates for refunctionalisation. Our analyses in Sects. 3.2, 3.3 and 3.4 and of the FGC corpus in the Appendix have found the following cases of refunctionalisation in FGC:

The third-person singular form ‘va’ of the verb ‘to go’ has become generalised, without undergoing significant semantic change, as the future marker ‘wa ’.

(235)

Nous

wa

mangé

tout

lé

dé.”

1pl

fut

eat

all

def.pl

two

‘The two of us are going to eat (together).’

(Parépou 1885: 44)

Another marker in FGC is a generalised form of a French verb. Té , derived from various parts of ‘être’ (‘to be’), has acquired many different functions (see Sect. 3.3.6).

(236)

I

té

gen

dé

ròt

kannon

3sg

pst

have

indf.pl

other

boats

‘since there were other boats’

(237)

Mo

té

fè

mo

granlajan

1sg

pst

make

poss

big-money

‘I made a lot of money.’

A similar case arose with the modal marker divèt, derived from forms of the verb ‘devoir’, which has become generalised and is used as a marker of epistemic modality in FGC.

(238)

I

divèt

té

ka

dronmi

3sg

must

pst

prog

sleep

“Maybe he was sleeping.” (Peyraud 1983: 135)

The interrogative and relative pronoun ‘qui’ is now very productive and has been refunctionalised across a wide range, from a generalised relative pronoun (qui/que/dont) to acomplementiser (‘I think that you …’) as well as contributing to new lexicalisations, such as kimoun, kitan and kikoté.

Instead of using the highly frequent inherited interrogative pronouns ‘who’, ‘where’, ‘when’ and ‘what’, FGC makes use of a refunctionalised marker ki, fused with nouns like ‘people’, ‘side’, ‘time’ and ‘thing’.

(239)

Kimoun

vini

who

come

‘Who came?’

In the construction a pou (‘you must’), pou expresses deontic modality instead of finality:

(240)

I

 di:

“Manman

a

pou

mo

mennen

ou

3sg

say

mum

dem

for

1sg

take

2sg

koté

oun

espisyalis!”

     

side

indf

specialist

     

‘She [my daughter] said, “Mum, we have to take you to a specialist!”’

In FGC, the functions of commitative (‘with his wife’), instrumental (‘with a spoon’) and additive (‘a man and a woman’) are all expressed with ké (‘with’), thus we observe a refunctionalisation in the form of an expansion of functions.

(241)

Wonm

ké

famn

vini

Man

com

women

come

‘a man and a woman came’

There exists at least one case of reduction of functional range of the definite article oun (‘a’) which is omitted in general expressions like:

(242)

fè

to

wonm

make

2sg

man

‘You have to be a real man.’

The right bracket ‘-là’ of the French demonstrative pronoun ‘ce N-là’ has been refunctionalised as the FGC postponed singular definite article -a. Since the definite article is less obligatory than in French, -a is closer to a demonstrative.

(243)

kaz-a

house-def

‘the house’

Note that two other sources of -a are possible, including a Type 1 functional transfer; the locative adverb là is a lexeme that already in French can have the pragmatic function of identifying both deictically (‘the person there’) and/or anaphorically (‘the person mentioned earlier’). The second source is a direct transfer (borrowing) since the Ewe morpheme -a has the same form and function. Ewe has both -la and -a as allomorphs, but seems to prefer -a. Although direct transfer seems the most obvious explanation, it does not account for why none of the other 30 or so tranfers of Type 1 to 4 are directly borrowed from Ewe (though this has been attested for other contact languages, cf. Pfänder 2009). It should be remembered that these various pathways are not competing but convergent. Speakers may subconsciously use different pathways to arrive at the same result.

The French deictic expression ‘eux-là’ (‘those’) has acquired a new function as a plural marker (ya/n) instead of pragmatic precision/deixis.

(244)

tifi-ya;

timoun-yan

girls-def.pl

boys-def.pl

the girls

the boys

Note that an alternate pathway resulting from a morphosyntactic strategy transfer and functionalisation Type 1 is possible. Ya was first attested as yé la, and Ewe has a postposed personal pronoun 3PL that follows the definite article . However, the word order in Ewe is different and this transfer should have yielded la yé instead of yé la. Ewe speakers would have found eux là to be salient because it was in the right place for an Ewe plural marker. It would also have been frequent in the spoken French given as orders to slaves. Again, as with the singular -a, converging pathways increased the chance that a particular feature would be transferred from Ewe to FGC.

Finally, FGC has right-bracket marking relative clauses that can be analysed as a refunctionalisation of the definite article :

(245)

Boug

ki

vini

a

Man

rel

come

rel

‘the man who came’

3.5.1.3 Syntactic Strategy Transfer

Our study has found many cases of syntactic strategy transfer , the most salient of which are detailed here.

While there is no syntactic strategy in French that includes a sentence-final interrogative marker, Ewe has a grammaticalised question marker:

(246)

Kofí

dzó

a?

Kofí

left

q

‘Has Kofí left?’ (Ameka 1991: 54)

We find the same marker with the same function in FGC:

(247)

Georges

pati

a?

Georges

left

q

‘Has Georges left?’

Instead of making use of the highly frequent French interrogative pronouns qui, ‘who’; où, ‘where’; que, ‘what’; and quand, ‘when’, FGC developed a completely new paradigm of wh-pronouns (kimoun, kikoté, kisa, kitan; literally which people, which side, etc.) that imitate the syntactic strategy of Ewe (afi-ka).

In French, the intensifier is marked preverbally with the adverb très, ‘very’. In FGC, however, the intensifier uses the postposed lexeme beaucoup, ‘much’:

(248)

Zarb-a

gran

bokou

Tree-def

big

much

‘The tree is very big.’

Again, this seems to be a syntactic strategy transfer from Ewe :

(249)

Atí

lá

kɔ́

ŋútɔ́

Tree

def

big

much

‘The tree is very big.’ (Dzablu-Kumah 2015: 7,11)

Ewe makes no use of a preposition if things are contextually clear:

(250)

Kwami

mé

fɔ́

kábá

yi

agble

o

Kwami

neg

wake up

quickly

go

farm

neg

‘Kwami did not get up early and go to the farm.’ (Ameka 1991: 65)

We find the same strategy in FGC:

(251)

I

lékol

He

school

‘He is at school.’

Another, often-disputed topic is that of serial verbs . Space does not permit a detailed study, but it is noteworthy that both Ewe and FGC display syntactic strategies that have been classified as serial verb constructions:

(252)

É

fɔ́

do go

3sg

arise

go out

‘He got out.’ (Ameka 1991: 58)

(253)

I soti alé

I alé désann

‘She left.’

One type of serial verb construction is classified as a focus strategy. As we do not find this kind of syntactic strategy in French, we propose classifying this strategy as a transfer from Ewe :

(254)

Sí

Kofí

sí

Escape

Kofi

escape

‘Kofi ran away as fast as he could.’ (Ameka 1991: 45)

(255)

A

pati

Georges

ka

pati

it.is

leave

Georges

prog

leave

‘Georges left as if he were running for his life.’

As the example above suggests, there is some use of serial verbs by the Guianese, especially for verbs of movement:

(256)

Georges

ké

Marie

kouri

alé

Georges

com

Mary

run

go

‘Georges and Mary went away quickly.’

Unmarked verbs have a present reading for statives (cf. first example below) but may have a past reading for dynamic verbs (second example below):

(257)

Kofí

kɔ́

Kofi

tall

‘Kofi is tall.’ (Ameka 1991: 80)

(258)

Kofí

vǎ

Kofi

come

‘Kofi came.’ (Dzablu-Kumah 2015)

We find the exact same distinction in FGC, as discussed in Sect. 3.3:

(259)

Georges

gran

Georges

big

‘Georges is big.’

(260)

Georges vini

Georges come

‘Georges came.’

A very clear example of syntactic transfer from Ewe is the comparative construction, which in FGC uses ‘pasé ’ (‘to surpass’) after the adjective:

(261)

I gran pasé ròt

he big surpass other

‘He is taller than the other (person).’

Ewe does exactly the same, using the verb ‘wú’ (‘to surpass’) after the adjective:

(262)

xɔsia

lolo

wú

éma

Room-this

big

surpass

it-that

‘This room is bigger than that one.’ (Warburton et al. 1968: 182.)

In FGC, word order is the same, regardless of whether the object is pronominalised or not (as in English ):

(263)

Yé

ka

vandé

kaz-a

3pl

prog

sell

house-def

‘They sell the house.’

(264)

Yé

ka

vandé

l’

3pl

prog

sell

3sg

‘They are selling it.’

(265)

Ils vendent la maison.

‘They are selling the house.’

(266)

Ils la vendent

‘They are selling it.’

Here, FGC clearly is in line with Ewe , but contrasts with French; in French, the object is normally postverbal, but the pronoun is pre-verbal.

3.5.1.4 Category Transfer

Comparison of FGC with French reveals at least two semantically defined categories expressed in different ways. Whereas both FGC and French make use of copulas, in French, there is only one copula expressed with conjugated forms of être (‘to be’), but in FGC, there are three types of copulas: zero, sa and fika . Fika is used as a locative copula (I fika Cayenne ‘he is in Cayenne’), sa is used as an existential copula (I sa dantis ‘he is a dentist’) and zero is used for all other purposes (I gran ‘he is tall’). Maybe the most striking difference between French and FGC is the overwhelming tendency in FGC to express aspect and only rarely tense , whereas French strongly emphasises tense. These divergences between FGC and French can be classified as (semantic) category transfers from Gbe languages. Ewe works the same way as FGC for copulas (Dzablu-Kumah 2015: 46) and aspect (2015: 51).

3.5.2 A Usage-Based Perspective on Creole Grammar

Our analysis of some thirty features of FGC grammar is inspired by usage-based linguistics , a theoretical framework that aims to explain language variation and change from the perspective of the listener and her or his individual experience of language. We therefore followed two basic premises in our research:

  1. (a)

    Perception outweighs formal classification. A participant might believe that a particular segment or element of a second language is relevant even if a grammarian would say otherwise. Listeners might perceive French lexical compositions such as ‘est-ce’ / ,vous autres’ or ‘de l’eau’ as a single word (es /zòt /dlo).

  2. (b)

    Ecological conditions are key to language perception. This premise means that if the slaves were able to agree on one African lingua franca like Ewe , there was no need for them to create a new language within their community. The relative numerical parity of slaves and slaveholders and the presence in effect of only two languages in the colony (as presented in Chap. 2) led us to expect that the variety of French that the slaves used to communicate with slaveholders would have been more Ewe-like than other French-lexicon creoles.

Drawing from these considerations, we report the following concerning FGC grammar:

  • Good evidence for one or several types of transfer from Gbe languages for all 30 features of FGC analysed.

  • Confirmation of Siegel’s claim that the four types of linguistic transfer are sufficient to model all of the investigated features. Three of the four types of transfer were especially successful, namely the transfer of morphosyntactic strategies and the two types of functional transfer.

  • FGC displays many instances of frequency effects , salience-induced changes, reanalyses and entrenchment . The small amount of input instead leads to the hypothesis that salience outplays frequency in FGC (‘le’ as an article is highly frequent in French, but the FGC marker ‘-la’ derives from the more salient demonstrative ‘ce mec-là’; ‘où’ is frequent, but the question marker is ‘kikoté’).

  • Experience is the decisive factor in the creation of a new language. Listeners simply analyse what they hear in an L2, but each person perceives language potentially differently because there is no written input or formal teaching by native speakers.

Interaction studies combined with cognitive studies may in future provide important insights to help identify which kinds of features were frequent and salient in the colonial language contact situation and which combination of features is cognitively plausible and follows well-known grammaticalisation paths. To sum up, we could thus formulate three basic constraints to transfer.

  1. 1.

    Phonetic salience . In the French example ce type-là, ‘that guy’ (lit. the guy there), the first part, ce, is phonetically much less salient than the final part, là, and does thus not qualify for being transferred.

  2. 2.

    Syntactic slots. In most West African languages , the article is postposed. This makes the transfer of the postposed item, là, more likely than the transfer of the preposed item, ce.

  3. 3.

    A semantic constraint. Interestingly, most of the functionalisation and refunctionalisation processes follow well-known paths of grammaticalisation and thus generally involve metonymic changes.

All three constraints can be easily explained from a usage-based perspective. We therefore close this chapter by reiterating a strong plea for a usage-based approach to language contact and creolisation .

In language acquisition, variation and change, a clash of two of the main principles, namely frequency and salience, is possible, especially within a usage-based paradigm. The concept of frequency assumes that the user will acquire or transfer the most commonly heard of competing variants. On the other hand, the most salient version might win out even if it is not heard as much as other versions. In FGC, salience seems to dominate, perhaps due to the extreme social context of creolisation that restricts the learner’s access to the language. Examples in which salience predominates include the demonstrative marker la instead of highly frequent le and the interrogative pronoun ki (< ,qui’) instead of ke (< ,que‘).

By assuming that speakers had no choice but to work with the heterogeneous language situation in which they found themselves, our aim in this study was to demonstrate how fruitful it is to combine both archive work and corpus analysis (Ennis and Pfänder 2010, 2013). Among other things, the archive work in Chap. 2 revealed that the early decades of settlement in French Guiana showed no record of a significant number of Bantu speakers. This finding fits perfectly with the linguistic observations in this chapter, namely that the divergences between FGC and MAC can be linked to Bantu influence on creole formation in Martinique (where a Bantu population was attested) (Crane et al. 2011). Among these divergences:

  • In FGC, possessive pronouns are preposed, as in many (but not all) Gbe languages, but postposed in MAC, as in Bantu languages.

  • FGC and Gbe languages like Ewe share the feature of an overt question marker; this is not the case for MAC, and not as far as we have found for Bantu languages.

  • Aspect is important for all French-based creoles, but more important than tense in both FGC and Ewe ; this clear preponderance of aspect over tense has not been reported for MAC or for Bantu languages.

These initial findings suggest that further research into comparative Bantu influence on creoles is needed.

This study has also made a number of findings about inheritance and innovation . Regarding inheritance, it might be conducive to regard a rhizome structure as the basic assumption instead of a one-track root. In a complex multilingual setting such as that of early French Guiana, we might want to use the metaphor of multiple birth (cf. Aitchison 1995) and might show continuity in every feature in more than one language. There are further traditional assumptions of language that need to be challenged: a language is not a comprehensive dictionary that can simply be transferred, and its grammar is not an integral system. After all, we can easily conceive of inheritance as being based on people’s daily life experiences, on the constructions and routines they hear (and want to hear). In this sense, we wish to stress the materiality of inheritance (as in the social routine of greeting gestures) as well as people’s drive to seek patterns and to access and acquire routines and structures. This directly relates to questions of power within the specific sociohistorical context, namely a slave-based plantation society in which the slaves’ relation to language is framed by two distinct hierarchies as they communicate with those in power and with fellow slaves.

Regarding innovation , the findings of both the history and the linguistics chapter encourage us to think that creole genesis should not be misunderstood as a process that begins from square one. The emerging language might well represent a gestalt-like whole and a genuinely new entity, but if we look closely at each grammatical feature, we see that there is much continuity in material inherited from parent languages.

Siegel writes that “the development of pidgins and creoles shows language to be a multifaceted, ever-changing system of communication that is highly adaptive to the requirements of its use and to the environment it is used in” (Siegel 2008: 279). This study of FGC encourages us to conceptualise language creation not as an innovation from scratch, but rather as a combination of inherited elements from various languages that are perceived and identified by the speakers as material for creating a new language.