Contact Linguistics. Chap

Exercise: Lefebvre (1996) argues that the TMA categories of Haitian Creole derive their semantics and functions directly from Fongbe via relexification and reanalysis. How similar is this explanation to that offered by Siegel for the emergence of TMA categories in Hawai‘i Creole English? How convincing are the respective arguments for substrate influence?


9.2.3. Transfer and creole syntax.

In cases where there is some degree of congruence between L1 and L2 structures, creole syntax reflects the common ground. Thus, SVO ordering in Atlantic English-lexicon creoles has parallels in both English and the relevant substrates. Again, such cases can be explained as “transfer to somewhere” (recall for example English learners’ placement of object pronouns after the verb in their L2 French). On the other hand, those substrate-derived aspects of creole syntax that have no counterparts in the superstrate (input), would presumably be treated as cases of “transfer to nowhere” and explained in terms of Anderson’s “relexification principle” (see Chapter 7, section II.5.2). Recall that, according to this, learners will use their L1 structure when they cannot perceive certain L2 structural patterns. That would explain the emergence of structures such as SVC’s in the Atlantic creoles. Anderson does not make it clear whether his conception of relexification is similar to that of Lumsden. Are abstract syntactic patterns “transferred” in toto, so to speak? Or do they result from the transfer or projection of substrate semantic and lexical representations onto superstrate-derived lexical forms? The latter approach would be closer to that of Lefebvre and Lumsden, though they view the process the other way around, that is, as calquing of superstrate-derived forms onto substrate categories.

9.2.4. Relexification or transfer?

Differences in the terminology used for describing L1 or substrate influence on learner versions of an L2 merely reflect differences in perspective. The terms themselves, “relexification” or “transfer”, in fact refer to the same objective psycholinguistic process of restructuring. Transfer views the effects of L1 influence from the perspective of (learner versions of) the TL, focussing on the ways in which input (intake) is changed under that influence. The relexification scenario views L1 more from the perspective of the L1 (input), focussing on how L2 items are incorporated into the learner system as labels for L1-derived semantic/functional categories. The dichotomy between the two approaches is only an illusion.

Both “transfer” and “relexification” are consistent with the notion that creole creators (or L2 learners) retain certain abstract categories or structures in terms of which they reanalyze or reinterpret substrate- or L2-derived forms. Such abstract continuities are in principle no different from the retention of overt L1 lexical (and sometimes grammatical) forms, which occurs in both L2A (see Chap. 7, Section I.2) and creole formation.
Whether learners “project” L1 categories onto L2 forms or “calque” the latter onto the former, the resulting reanalysis is the same. Moreover, the principles and constraints regulating this process are the same, whichever perspective we take.
Both relexification and transfer scenarios identify the same constraints on the reanalysis of L2-derived forms. Both appeal, for instance, to the notion of congruence leading to interlingual identifications between L1 and L2 forms, as we saw earlier. Siegel (1999) also argues that the selection of superstrate forms is determined by factors such as perceptual salience, transparency and frequency. As we noted in Chapter 7 (Section I.7.3), all of these notions can be subsumed under the concept of markedness. We return to this below.


9.3. Leveling in creole formation.

So far, we have been considering processes of creole formation primarily from the perspective of individual learners. As DeGraff (1999b:485) has noted, these individuals create I-languages (IL systems) on the basis of the linguistic data available to them. These I-languages provide the lexical and grammatical features that are adopted into the community language. Hence a theory of creole formation must also explain how a stable grammar emerges via a process of leveling, from among the competing variants introduced by different individuals or groups of speakers. Siegel (1997:126) describes the process of leveling as involving the reduction or attrition of variants in ways similar to koineization.

In this connection, the “reinforcement principles” proposed by Siegel (1999), which we discussed in chapter 7 (Section II.6) are highly relevant. Recall that these principles determine which features will be retained and which discarded. A key principle seems to be that features shared across the individual IL’s will be retained. In case where (the majority of) the substrate languages are typologically similar, such features occur frequently as a result of similar kinds of L1 influence. They constitute part of the common ground that learners seek in communicating across language boundaries. In cases where competing variants arise, the more transparent or salient ones survive. In both instances, then, features that are less marked by virtue of similarity, frequency and transparency are most likely to be retained in the emerging community language. This would explain, for instance, why Melanesian Pidgin grammar retains a common core of features shared across its CEO substrates (Keesing 1988:123).
There is also reason to believe that children acquiring emergent creoles play a role in ironing out the variation due to influences from different source languages. Children themselves may introduce certain innovations into creoles, such as TMA markers and auxiliary combinations. We saw earlier that children bilingual in their ancestral language and HPE introduced several such innovations into HCE. It is very likely that children bilingual in West African languages and early forms of creole made a similar contribution to the latter.
It also seems reasonable to assume that children acquiring an established creole as a first language would have contributed toward systematizing and expanding the grammar. Studies by Newport (1999) and Kegl, Senghas & Coppola (1999) demonstrate that children acquiring American Sign Language and Nicaraguan Sign Language respectively as their first language imposed regularity and order on the disparate and variable input they received from adult signers.
As DeGraff (1999:494) suggests:

Via L2A, …. adults introduce innovative patterns into the linguistic ecology of language learners, whereas children, via L1A, play a key role in restructuring adults (and their own) innovations into stable grammars.

In this, children [and adults - DW] are guided by “UG acting as a sieve and guiding grammatical invention…” (op. cit.: 497).

10. Universal principles and creole formation.

The view that “universal principles” have priority over substrate influence in creole formation goes back, as we have noted, to scholars of the late 19^th century such as Coelho (1880). It was later given fuller expression by Bickerton (1981, 1984) who claimed that the bulk of creole grammar was due to Universal Grammar (UG), the innate language faculty. This approach treated creole formation as essentially a case of first language acquisition, in which adults had little role to play, except as suppliers of degenerate pidgin input. Some scholars still hold this view. For example, Veenstra (1996) argued that serial verb constructions in Saamaka were due to the workings of UG principles. In his view, absence of verbal morphology and consequently of verb movement to INFL triggered children’s creation of SVC’s in this creole (1996:176). He does not however rule out the possibility that adults (and hence substrate influence) also played a role, but at later stages of creole development (p. 194).

This strong position on the role of UG has given way lately to a more eclectic view. There is now broad consensus that the role of UG is to constrain the processes of restructuring by which superstrate and substrate inputs (intakes) are shaped into a viable grammar – one that conforms to universal principles of language design. Such principles play a role in all phases of creole formation – the early pidginization stage, the elaborating stages and the later developmental stages.
Absence of bound morphology, fixed word order, invariant preverbal negation and other instances of “reductive” simplification are presumably introduced into creoles via input from their pidginized antecedents. Such cases of simplification reflect the same processing and learning principles that apply to the early stages of SLA (Chapter 7, Section I.71) and pidgin formation (Chapter 8, Section 5.2). There is therefore no need to repeat them here. We are concerned only with the elaborative and later developmental stages of creole formation.


10.1. Universals and L1 influence.

The same principles that regulate the role of the L1 in the elaboration of early IL in SLA operate to constrain the interaction between substrate and superstrate input to creole formation. As we suggested in Chapter 7 (Section I.7.5), cognitive constraints based on processing principles conspire with structural constraints based on typological distance to regulate restructuring in IL development. As in SLA research, some theories of creole formation attempt to explain these constraints in terms of markedness principles. For instance, scholars like DeGraff (1999:509) and Muysken (1981b) see creole grammar as reflecting the least marked parameter settings triggered by the input. Other (not incompatible) approaches appeal to notions such as ease of perception and ease of learning – two notions central to markedness theory. Thus, Seuren & Wekker (1986:65-66) argue that the need for semantic transparency was a major factor in creole formation. This would explain general properties of creole grammar such as “uniformity” (isomorphism of form and function), “universality” (maximal use of rules that are least language specific) and “simplicity” (minimal processing required to get from semantic to surface structures and vice versa).

As Keesing (1988:110), following Kay & Sankoff (1974), notes: “the more speakers simplify and the deeper they dig, the more closely languages converge on basic minimally marked and maximally natural patterns.” To cast this in different terms, creole grammar might be seen as representing the unmarked core of language as distinct from the periphery of marked elements and constructions (DeGraff 1999:510; Macedo 1986). As we noted earlier, children may have played a role in creating these properties of creole grammar.


10.2. Universals and internal developments.

Those features of creole grammar that arise via internal processes and have no counterparts in either substrate or superstrate sources are of particular interest to universalist theories of creole origins. Such innovations may be due either to creativity in the process of creole formation itself, or to subsequent internally motivated change. In both cases, children may have played a vital role.

Some developments in the TMA system of Sranan will serve to illustrate these types of innovation. For example, the emergence of go as Future marker and been as Past marker seem to have been driven more by internal motivation than by either superstrate or substrate influence. This may explain why these forms were reinterpreted in similar ways in various Atlantic EC’s, as well as in Hawai‘i Creole English. In the pidgin from which the latter developed, both forms already displayed the potential to develop into TMA markers (Siegel 2000). We can assume the same to be true in early Sranan and CEC generally. In both HCE and Atlantic EC’s, the grammaticization of Future go was further encouraged by the existence of substrate and superstrate models in which a form derived from “go” indicated futurity. But in addition to this, the fact that go and other verbs of motion tend to develop into Future markers crosslinguistically (Bybee et al. 1994) suggests that universals were also at work. In the case of Past been in Sranan, however, no substrate model appears to have existed. Hence this is a likely candidate for an explanation in terms of universal principles of grammaticization (see Detgers 2000).
The emergence of auxiliary combinations in creoles is another internal development that seems to owe its origin partly to universal principles. Again, however, substrate influence played some role in the Atlantic creoles. For instance, the classic TMA ordering found in the Surinamese creoles has models in Gbe. Jondoh (1980:52) informs us that this is the order of auxiliary elements in Gengbe. The following examples illustrate.

(30) Sranan En dan a man ben musu e breiti

And then the man PAST MUST IMPFV happy
"So the man should have been happy”
(31) Gengbe é lá téü no du gàlí (Gengbe: Jondoh, p. 52)
he FUT can PROG eat gali.
"He will be able to eat gali"

Similarly, Siegel (2000:221) suggests that auxiliary order in HCE has a model in Portuguese.

In addition, however, the Surinamese and other Atlantic EC’s have far more complicated patterns of auxiliary ordering, all of which seem to have developed internally. The following examples from Sranan and Belize creole illustrate.

(32) a. A ben e musu e taki nanga unu. (Sranan. Elicited)

T A M A
"S/he usually had to be talking with us."

b. A ben sa e musu e wani go na foto. (Elicited)

T M A M A
"S/he would have had to be wanting to go to Paramaribo."
(33) a. Jan don mi wã di iit if i mi kom in taim. (Belize Creole)
A T T A
"John would have already been eating if he had come in time."

b.. Jan mos mi wa kom ya laas nait if i mi de da Belize.

M T T
"John would certainly have come here last night if he was in Belize."

Again, without models in either the substrates or the superstrates, these complex tenses must be the result primarily of internal developments.

10.3. Grammaticalization in creoles.

The emergence of grammatical categories in earlier creole has sometimes been attributed to a process of “grammaticalization.” This term has been defined in various ways (Joseph 2001:165), but in general it refers to the language-internal process(es) by which grammatical morphemes develop gradually out of lexical morphemes (Bybee et al. 1994:4-5). As Joseph (2001:166) points out, the term “grammaticalization” is really a cover term that describes the results of “other recognized mechanisms of change, especially sound change, analogy, or reanalysis.” One example is the gradual reanalysis of the Old English verb willan “to want” as the Future tense auxiliary will. A similar example is the emergence of the modern Greek Future prefix Qa from Medieval Greek structures like the following Joseph 2001:181):

(34) thélo: hina grápho:

want-1sg that write-1sg
“I want to write” (literally: “I-want that I-write”

Later, through processes of phonological reduction, thélo: and hina became /Qe/ and /na/ respectively. Thus we find future constructions like the following (ibid.):

(35) Qé na Vráfo

FUT write-1sg.
“I will write.”

Eventually, Qé + na were further condensed to the modern Future form Qa.

In short, grammaticalization involves gradual change that is internally motivated. Bruyn (1996:30) has pointed out that the term, in its usual sense, should not be applied without qualification to the reanalysis of superstrate-derived items as creole grammatical categories. In the first place, this reanalysis is not gradual, but the result of a relatively rapid change. Second, it is promoted by external (usually substrate) influence, as opposed to purely language-internal motivation. This means that the creation of creole grammatical categories is due to direct calquing, unlike the processes of analogy, metaphorical transfer and phological reduction found in “ordinary” grammaticalization. The former type of innovation might instead be referred to as “grammaticization”, to distinguish it from the more gradual historical process.

Note also that grammatical elements in some creoles appear to derive from superstrate grammatical morphemes that are simplified and reanalyzed. For example, Detgers (2000:150ff) argues convincingly that the Past marker te of French-lexicon creoles derives from the French auxiliary était, used in constructions of the type il était à écrire “He was writing.” This is somewhat different from the change of lexemes into grammatical formatives referred to above.
Some creole grammatical categories seem to emerge via calquing followed by internally driven processes of change. One such case is the development of the earlier Sranan Progressive marker de into an Imperfective. The progressive function of copula de was apparently due to substrate influence. Gbe, like other West African languages, uses the same form as both a locative copula and a progressive marker. The following Gengbe examples from Jondoh (1980:46) illustrate:
(36) e! le~ ekplO$-a òi
it be table-the on.
"It's on the table"

(37) e! le~ nu! d¡u~d¡u& kO~

he be thing eating at
"He's eating."

The progressive function seems to have been transferred to copula de in earlier Sranan, as in other CEC’s such as Belizean and some varieties of JC. In contemporary Sranan, however, (d)e functions as a true Imperfective marker, conveying notions such as ‘progressive’ and ‘habitual’ and used with predicates of all types, stative and non-stative. This development has no models in Gbe or other substrate languages, which employ distinct Habitual and Progressive categories. Nor did it take place in Belize or Jamaica. Hence the broadening of de’s functions seems to be an internally motivated development in Suriname. The change of Progressives into Imperfectives is of course common crosslinguistically (Bybee et al 1994:82) and seems to be promoted by similar principles of semantic extension.

We also find innovations in creole morphosyntax that are due to grammaticalization in the traditional sense. These represent another instance where universal principles of change shape the grammar of creoles. A case in point is the development of the modal auxiliary man in Sranan (van den Berg 2000, van den Berg & Arends 2001). In the contemporary language, man conveys the sense of ability constrained by physical law or other forces beyond the agent’s control (Winford 2000c:77). In the following example, a woman complains that poverty prevents her from being able to afford a peanut butter sandwich.

(38) A diri k’falek, yu no man bai en. Mi no man.

It expensive terrible, you NEG can buy it. I NEG can.
“It’s terribly expensive. You can’t buy it. I can’t.”
In early Sranan, however, man was clearly a noun, meaning ‘male human being.’ It was uses in structures like the following:
(39) a. Mingo, yu no man (SN text 1707; van den Berg 2000:14)
Mingo you NEG man
“Mingo, you’re not man [enough] (You don’t dare)”

b. Mi no man va hoppo dati (Schumann 1783)

I NEG man for lift that
“I’m not man [strong] enough to lift that”

Over time, its meaning must have extended from “being man enough” to “being strong enough,” to “being physically able to,” to the more general sense of ability it now conveys. The fact that a woman utters the example in (37) underscores the change.

Similar paths of internal change have occurred in all creoles, sometimes leading to drastic differences between the earlier and later stages of these languages. DeGraff (1999:497) points to various cases of “grammatical invention” in Haitian Creole that are due to such internal processes. These have caused contemporary Haitian to diverge significantly from earlier HC (see also Alleyne 2000).


11. Externally motivated change in later creole development.

In addition to the kinds of internally-motivated changes discussed so far, creole continue to change under external influence. Sranan, for example, borrowed the modal auxiliary mag from Dutch, the official language of Suriname. The modal expresses only permissibility, as in the following example:

(40) Yu mag go prei baka te yu kaba wasi den doti sani

you may go play after when you finish wash the-pl dirty thing.
“You may go and play after you finish washing the dirty dishes.”

There is also evidence of Dutch influence on certain areas of Sranan syntax, such as attributive (property) predication and comparative structures. With regard to the former, conservative varieties of Sranan emply structures in which the predicative property item is verbal, as in the following (Winford 1997b:257f):

(41) a. A liba sa bradi
The river POT broad.
“The river may be wide”

b. A pikin musu weri

“The child must be weary.”

c. A watra faya tumsi

The water hot too-much
“The water is too hot.”

However, in contemporary Sranan, such structures are giving way to alternatives in which the copula de is used, suggesting that the property items are adjectives. The following illustrate:

(42) a. A liba sa de bradi

the river POT COP broad
“The river may be wide.”

b. A pikin musu de weri

“The child must be weary.”

c. A watra de tumsi faya.

“The water is too hot.”

This tendency is reinforced by the continuing borrowing of Dutch adjectives that require copula support, as in the following examples (Winford 1997b:283)

(43) a. A man de ernstig
“The man is serious.”

b. Mi de enthoesiast.

‘I’m enthusiastic.”

With regard to comparatives, conservative Sranan also preserves structures like the following, in which a verb moro (< more) “surpass” conveys the comparison.

(44) Hertoch bigi moro Ronald

Hertoch big surpass Ronald
“Hertoch’s bigger than Ronald.”

Under Dutch influence, alternative comparative structures like the following have emerged.

(45) Hertoch de moro bigi leki Ronald.

Hertoch COP more big than Ronald
“Hertoch is bigger than Ronald.”

Here moro is an adverbial modifier qualifying the adjective bigi, and the comparative structure on the whole is closely modeled on that of Dutch. Many native speakers of conservative Sranan are aware of these changes, and comment openly on them (Winford 1997b:278).

Varieties of Sranan spoken by native speakers of Dutch or Dutch-dominant have long been subject to these kinds of external influence (Eersel 1971). However, there is an increasing tendency for these changes to spread into native Sranan as well, especially among the young (Arends 1989:51). This is not surprising, given that SN speakers are generally bilingual in Dutch.
We could adduce many further examples of more recent contact-induced changes in creoles. The most widely discussed cases of this are the Atlantic English-lexicon creoles outside of Suriname. These have been subject to continuing influence from English, and have converged toward the latter in a process known as “decreolization.” Such creoles have always been part of a continuum ranging from more acrolectal (English-like) to mesolectal (intermediate) to basilectal (most divergent). Hence it is not always easy to identify the line separating the creole varieties from each other, and from the acrolect. This makes it difficult to determine which features of the more basilectal varieties are innovations due to external influence, as opposed to variation long inherent in the grammar. For example, basilectal Guyanese Creole employs modal auxiliaries like wuda “would” (< would have), kuda “could” (< could have), shuda “should” (< should have) etc., which compete with auxiliary combinations like bin go/sa, bin kan, bin mos etc. respectively. The following are examples:

(46) a. Jan bin kyan kom yesidee.

John PAST can come yesterday.
“John could have come yesterday.”

b. Jan bin go du am.

John PAST FUT do it.
“John would have done it.”

(47) a. Jan kuda (bin) kom yesidee.

“John could have come yesterday.”

b. Jan wuda (bin) du am.

“John would have done it.”
Are the auxiliary combinations giving way to the single forms, or have they always been in competition? A comparison with the Surinamese creoles, which have only the auxiliary combinations, would suggest the former. If this is the case, did this process of change also occur in creoles like Jamaican, which disallows the combination “Past + Future” (ben + wi) but allows Past + Prospective (ben a go)? Did the contemporary JC Future auxiliary wi (< will) replace an earlier go future similar to that found in the eastern CEC’s?
Questions like this are almost impossible to answer, given the dearth of early textual evidence from these creoles. Yet some researchers (e.g. Alleyne 1980) believe that creoles like JC and GC were once much closer to the Surinamese creoles in many areas of their grammar. Further research will no doubt clarify whether this is true.
12. Summary.

We’ve seen in this chapter that the languages called “creoles” include a diverse set of contact vernaculars that arose in European colonies in the 15^th through 19^th centuries. There are no absolute criteria, either sociolinguistic or structural, that define creoles as a type. In general, however, we can distinguish among them in terms of their degree of divergence from the European (superstrate) languages from which they derived most of (the phonological shapes of) their lexical and grammatical items. Creoles range from second language varieties that are close approximations to the superstrates, to “radical” outcomes that depart significantly from the latter. Between these two extremes there is a continuum of outcomes, with “intermediate” creoles like Bajan closer to the superstrate and “basilectal” creoles like rural Guyanese and Jamaican closer to the radical end. To a large extent, these differences are due to varying degrees of substratal influence from the L1’s of their creators, as well as to internal developments within the creoles themselves.

These differences among creoles are due primarily to differences in the social contexts in which they were created. The relevant factors include the demographic make-up of the groups in contact, the types of community settings in which they interacted, and the social codes regulating that interaction. In general, creoles arose in plantation colonies where transplanted slaves or subjugated indigenes were subjected to control by smaller numbers of Europeans.
Intermediate creoles like Bajan and Reunionnais arose in small farm or homestead settings where there were higher or at least equal ratios of Europeans to slaves. These settings allowed for closer interaction between the groups over several decades, promoting the consolidation of second language varieties of the European languages. The intermediate varieties display characteristics due to simplification and substrate influence, but these are far less pronounced than the continuities from their superstrate sources. In effect, then, these creoles are akin to colonial dialects of the European languages. Similar second language varieties arose even in classic plantation societies, within those sectors of the slave and free non-European population that had more privilege and hence greater access to the superstrate models.
In other colonies to which the plantation system was introduced early, the société d’habitation (homestead) phase of contact was neither long nor intense enough to override the effects of the massive demographic and sociopolitical changes that accompanied the rise of large plantations. In these cases, more radical creoles emerged among the majority of the slave population, particularly the field slaves. These outcomes displayed significant influence from substrate languages and from internally-motivated innovations, resulting in greater divergence from their superstrates.
The effects of substratum influence on creoles can be found in many aspects of their phonology, lexicon and (morpho-)syntax. At the same time, there are continuities from their superstrate sources, though that input was subjected to varying degrees of simplification and reanalysis. This accounts for characteristic features of creoles such as the loss of bound morphology and the reduction of case, gender and other distinctions.
More radical creole formation involves a variety of complex interrelated processes by which learners restructure input (intake) from the source languages into new grammars. The various theories that attempt to account for this kind of language creation tend to focus on one or the other aspect of the process. Some argue for the primacy of the superstratal contribution, while others assign more significance to substrate influence. Still others maintain that universal principles of language acquisition account for much of creole grammar. These disagreements are compounded by a lack of consensus on the precise nature of the superstrate input to creole formation, particularly whether it consisted of a previous pidgin or second language varieties acquired by earlier slaves.
Despite these differences of opinion, creolists generally accept that creole formation was primarily a process of second language acquisition in rather unusual circumstances. Moreover, children may have played a role in regularizing the developing grammar. The processes of restructuring that creates creoles are in principle the same as those found in SLA and to a lesser extent, L1A, modulo differences in the inputs and the social constraints on access to the target. In general, like SLA, creole formation involves the selective adaptation of both superstrate and substrate input, and is guided by universal principles of language acquisition and design. There is also agreement that radical creoles result from the continuing re-learning of previously acquired contact varieties by successive waves of newly arrived slaves over several decades.
The restructuring of available materials into a creole grammar also involves processes familiar in SLA. First, processes of simplification operate on the available input to produce a maximally transparent grammar. Second, processes of elaboration come into play, reshaping previously acquired grammars by drawing more on L1 sources as well as internally driven change. This interaction between L1 knowledge and learners’ intake from superstrate-derived input operates within the developing IL system itself – or more accurately, within the minds of individual learners creating IL systems or I-languages.
Various mechanisms and principles similar to those that operate in SLA guide this process of restructuring. Simplificatory processes are regulated by principles of economy and transparency. L1 influence, whether described as “relexification” or “transfer”, leads to reanalysis of superstrate-derived lexical (and sometimes grammatical) forms in terms of substrate semantic and functional categories. Factors such as congruence and markedness determine which forms are selected and which features from the input will be retained. Processes of internally driven elaboration result in innovations peculiar to the creole. Such developments appear to be motivated by universal principles of acquisition and change.
Competition among features and innovations introduced into individual IL’s is resolved in a process of leveling, via which a common core of features is retained and conventionalized in the new community language. Over time, other changes take place in creole grammar under both internal and external motivation. Processes of grammaticalization lead to the former, which continuing co-existence with an official language (usually but not always the lexifier) leads to contact-induced change.
The course of creole development and its relation to language evolution in general are neatly summed up by Roberts (1999:317):

Creoles are particularly interesting because they represent an extreme of language change, but it is the mechanisms of language change, which are ubiquitous in the history of every language and every language family, that have made creoles what they are.