Contact Linguistics. Chap

8.2.3. The Relexification Hypothesis

The most comprehensive attempt to identify substrate influence on a creole is the research done on Haitian Creole origins by LeFebvre, Lumsden and their associates (Lefebvre & Lumsden 1994, Lumsden 1999). They developed a particularly strong version of the substratist theory of creole formation, known as the Relexification Hypothesis (RH). The process of relexification will be discussed more fully below, in Section 9.2.1. According to this theory, most of HC grammar is derived more or less directly from that of Fongbe, its principal substrate language.

However, this account of HC origins has been strongly challenged by several scholars, including Chaudenson (1996), DeGraff (1999) and Singler (1996). In the first place, the RH rests on a somewhat limited account of the sociohistorical background to HC formation, neglecting the role of substrate languages other than Fongbe. Second, the RH has little to say about the precise sources of the superstrate input to HC formation. As Chaudenson (1992) and others have shown, it was provincial French dialects of the 17^th and 18^th centuries that provided this input. Moreover, many aspects of HC grammar can be traced to these sources.
Finally, the RH seems to present only a partial picture of the mechanisms and constraints involved in creole formation. Its critics have questioned whether relexification constitutes the only mechanism of substratal influence or indeed the only mechanism of change involved in creole formation. Singler (1996:218) points out that substrate influence (and by extension the RH) cannot account for certain features of creoles which are due to universals of acquisition or internal motivation. Lumsden (1999:230, fn 7) acknowledges this, pointing out that “the RH is NOT a claim that relexification is the one and only process involved in creole genesis, nor is it argued that relexification can account for all the properties of creole languages.” Such an account must address both superstrate and substrate inputs, as well as the principles that determined the selection of particular source language features and not others. In addition, it must account for innovations in creole grammar that arose independently of such input. We will discuss the various mechanisms and principles involved in all of these aspects of creole formation below, in sections 9 and 10.


8.2.4. Restructuring and internal developments.

The elaboration of early creole grammar, like that of developing interlanguage and expanding pidgins, involves innovations driven by tendencies already present in the developing system. Kouwenberg (1996), while acknowledging Kalabari (Eastern Ijo) as the source of several aspects of Berbice Dutch grammar, also points to several others that cannot be explained in this way. Some of the latter can be attributed to Dutch influence, for instance features like adjectives, prepositions and postnominal relatives. However, this still leaves several features that can be explained in terms of neither superstrate nor substrate influence. They include invariant SVO order, preverbal auxiliaries and negative marker, predicate cleft, and a serial verb construction in which a verb “say” introduces complement clauses.

Features like these appear to be due to processes of internal restructuring similar to those found in developing IL and in the elaboration of pidgins. In all cases, speakers exploit intake from both L1 and L2 sources to create a maximally simple grammar, and this can lead to innovations peculiar to the developing language. In the case of Berbice Dutch, this creativity may explain the features listed above, though this explanation doesn’t necessarily hold for similar features in other creoles. For instance, in English-lexicon Atlantic creoles, features such as SVO and preverbal TMA and negation have parallels in the substrate languages.
More generally, internally-driven innovations in creole grammar seem to arise from the restructuring of superstrate and substrate input, regulated by universal principles of acquisition (the need for economy and simplicity). A case in point is the development of the copula systems of Sranan and Ndjuka. In these creoles, the form na/da is employed as the focus marker in presentative and contrastive focus constructions and in equative (identificational) constructions lacking TMA marking. The following Paamaka examples from Migge, to appear:) illustrate:

(21) a. Mi na Sa D. (Migge, page 19)

1sg COP Ms D.
“I am Ms D.”

b. na a udu ya, a fu mi (Migge, page 22)

COP DEM wood here, 3sg for me
“It’s THIS WOOD that’s mine (not that one)”

On the other hand, the copula de is employed in locative constructions and in equative constructions containing negation or TMA marking, as in the following Paamaka examples (op. cit:33):

(22) a. A o de wan laulau sani

it FUT COP one badbad thing
“It’ll be an unimportant thing”

b. Mi an de a fesiman

1sg NEG COP ART leader
“I’m not the leader.”

This distribution of copulas has no exact parallel in the substrates, nor indeed in other Atlantic creoles. First, unlike the creoles in question, the substrates employ contrastive focus markers distinct from their equative/presentative copula (Migge, to appear: 23). In addition, the focus markers follow the focussed constituent in the substrates, but precede it in the creoles (recall that Saamaka follows the Gbe pattern in this respect). Moreover, the generalization of copula de beyond locative environments to equative environments that are specified for TMA and negation is not found in the substrates.¹⁹ The latter employ an equative copula distinct from the locative copula in all these cases. The creoles therefore have restructured their copula domain to create a simplified and more transparent system distinct from those of the substrates. How do we explain these developments?

Arends (1986, 1989) argues convincingly that the equative copula da/na derives from the demonstrative da (

(23) a. Da mi, Filida (Van Dyk, n.d: 73, cited in Arends 1986:108).

That ø me, Filida
“It’s me, Filida.”

b. ‘adjossi’, da Bakkratongo (Schumann 1783:46, cited in Arends 1986:113)

‘adjossi’, that ø bakra tongue
“’Adjossi is Europeans’ Sranan.’

Sentences like these appear to represent the kind of simplified English that formed the early input to the Surinamese plantation creole. This would explain the initial position of da/na as focus markers, as well as the fact that in earlier SN, the negative marker followed the equative copula (a pattern still found today in competition with structures containing no followed by de).

(24) Hertoch a no yu mati.

Hertoch COP NEG your friend

Copula de, on the other hand, appears to have arisen via reinterpretation of adverbial de (< there) in early SN structures like the following:

(25) a. Mastra soopie de (Van Dyk n.d., cited in Arends & Perl 1995:170)

master drink there
“Master, there is the drink”

b. Mastra, wini no de na battra (Arends & Perl 1995:180)

master wine NEG there? LOC bottle
“Master, there’s no more wine in the bottle”

Migge (to appear: 53) argues that the resemblance of such strings to existential and locative structures in Gbe and Kikongo triggered an interlingual identification between de and the substrate copulas. Compare the following sentences from Xwelagbe (Migge to appear: 53):

(26) a. makik?kwe lé (Migge to appear: 53)

bananas COP
“There are bananas.”

b. ixhe mO ∂é dO mE kliya (Migge to appear: 34)

fish PAST COP net in IDEO
“Fish were in the net in great quantity.”

Substrate influence, then, can partly explain the emergence of de as an existential/locative copula in the creoles. But it can’t explain the generalization of de to other copula constructions. The latter must be due to internal developments in Sranan and Ndjuka. The fact that the locative copula de found in other Atlantic creoles has not been generalized in this way, would lend support to this suggestion.

The conclusion to be drawn from this overview of creole formation is that this process was a complex one, involving a variety of linguistic inputs and strategies of restructuring. The theories of creole origins that have been proposed (“superstratist”, “substratist” and “universalist”) tend to focus only on one or another aspect of this complex process.²⁰ The challenge to contemporary scholarship is how to integrate these sometimes conflicting accounts into a unified explanation of creole creation.


9. Mechanisms, constraints and principles in creole formation.

Creolists generally agree that, in the emergence of creole grammar, superstrate-derived lexical items (or their phonetic shapes) are selected to express the morphological and syntactic categories of the creole. The processes by which these forms are integrated into creole grammar, and the principles and constraints governing their selection, remain subjects of debate. This is compounded by disagreement over the degree to which superstrate lexicon and grammar, as opposed to substratal input and universal principles, contribute to the formation of creoles.

Several questions arise here. First, what constrains the degree of superstratal as opposed to substratal input? Second, what principles regulate the selection of specific elements, lexical as well as structural, from each of the inputs? Third, what mechanisms and constraints are involved in the interaction between superstratal and substratal input? Finally, what role do universal principles play in this and other aspects of creole development?


9.1. Constraints on the linguistic inputs.

Siegel (1997:137) has proposed a number of “availability” constraints - that is, external and internal factors that determine the input to the earlier stages of creole development. Essentially, the input consists of the data available to learners as a target or model for acquisition. A crucial factor here is whether learners have relatively full as opposed to simplified input from the relevant L2. The external factors that influence this include those discussed earlier, in section 3, for example, the demographics of the groups in contact and the social contexts of the contact. Both external and internal (linguistic) factors influence the nature of learners’ intake from the superstrate, that is, the actual L2 materials they acquire and employ in the construction of their interlanguage. The internal factors are constraints on learners’ ability to process the L2 input, which lead to strategies of reduction and simplification of that input. Such strategies are common to all early attempts to achieve communication across language boundaries, including pidgin formation and early IL, as we have seen in earlier chapters. On the other hand, we can assume that the full resources of their L1’s were available to creole creators when they needed to compensate for limited mastery of the L2 (whatever variety of it they were exposed to). This raises the question of what types of L1 features creole creators actually use, and what constraints there are on this compensatory strategy.

9.2. Substrate influence: mechanisms and constraints.

As we saw earlier, In addition to overt retention of L1 forms, L1 influence in creole formation takes two general forms: use of abstract syntactic patterns into which superstrate phonetic shapes are incorporated, and reinterpretation of these superstrate forms in terms of substrate morphosyntactic and lexico-semantic categories. The same is true of extended pidgins, as discussed in the previous chapter. Hence the discussion of mechanisms and constraints here applies equally to them.

Two rather different explanations have been offered for these kinds of L1 influence. First, there are approaches such as Lefebvre’s and Lumsden’s (mentioned earlier) which appeal to notions like “relexification” and “reanalysis.” Then there are approaches like Siegel’s (1999, 2000) which appeal to “transfer”, a central concept in SLA (see Chapter 7). Let us examine each of these approaches in turn.
9.2.1. Relexification and substrate influence.

Lumsden (1999:225) defines relexification as “a mental process that allows a language learner to create a new vocabulary of lexical categories (i. e., nouns, verbs, adjectives, prepositions and adverbs) by linking new phonological forms with syntactic and semantic information that is already established in the lexicon of his native language.” The classic example of this process is of course Media Lengua, which we discussed in Chapter 6. For example, the Media Lengua verbal root bi- “see” derives its phonological shape from Spanish ver “to see”, but it displays the semantics and derivational properties of Quechua riku- “see” (Chapter 6, section 3.2). Relexification in this case explains the re-interpretation of L2 forms in terms of L1 lexico-semantic categories.

The RH also explains the similarity between abstract syntactic patterns in creoles and their substrates by arguing that the phonetic shapes of superstrate verbs assume the syntactic and semantic representations of substrate counterparts. For example, Lumsden (1999:241) argues that, in Haitian Creole, French-derived forms like voye “send”, vann “sell”, etc., assume the argument structure of Fongbe verbs like sE! “send”, sà “sell”, etc. Thus we find syntactic similarities like the following (Lumsden 1999:239):

(27) M voye rad la pou Mari (Haitian)

I send clothing DET for Mari
“I sent the clothing for the benefit of Mari.”

(28) Báyí sE! àwù i dó Ajwá (Fongbe)

Bayi send clothing DET for Ajwa.
“Bayi sent the clothing for the benefit of Ajwa.”

Lefebvre (1996) also appealed to the RH to explain the emergence of creole morphosyntactic categories such as TMA. Again, superstrate lexical forms are reinterpreted as labels of substrate categories. Lumsden (1999) revises this somewhat, suggesting that relexification accounts only for the entry of the superstrate lexical shapes into the early creole. These forms were subsequently reanalyzed as the signals of substrate grammatical categories through a process Lumsden (p. 250) refers to as “grammaticalization” (see further discussion below).