DeGraff, Michel (ed.) (1999) Language Creation and Language Change: Creolization, diachrony, and development. The MIT Press: Cambridge, MA/London. pp.543 (+x; index).
Reviewed by Miriam Meyerhoff, University of Hawaii.
'Language Creation and Language Change' is a collection of fifteen papers by 21 authors. The first and last are substantial overviews provided by the editor DeGraff. Each of the other 13 chapters examines some aspect of language acquisition relating the findings to theories of creolization. The theory most authors engage with directly is some form of Bickerton's Language Bioprogram Hypothesis (LBH). All contributors subscribe to the existence of some form of Universal Grammar and are interested in exploring the extent to which this appears to constrain the probable outcome of first and second language acquisition (L1A and L2A, of creoles and non-creoles) and the development of creoles (this includes cross-linguistic comparison of the structure of creoles and their relationship to the lexifier). Discussion of transfer of features from substrate languages plays a minor role in the collection. In the review that follows, I simultaneously outline each chapter, and provide comment. Chapter 1 ('A prolegomenon', DeGraff) outlines a basic hypothesis of the collection, namely that creolization, language change and language acquisition reduce at the level of the individual to the same processes (38). Assuming this is so, it is not productive to separate an investigation into the role parameter settings might play in any one of these processes from the others. Thus the rather heterogeneous case studies that DeGraff has gathered together, in fact all relate to the process of creolization, which DeGraff defines as "the process by which [a] pidgin ... 'acquires' a relatively stable grammar with parameter values that are quite distinct from those of the ancestor language" (5-6). The chapter concludes with an invitation to incorporate social factors into further studies, extending the interdisciplinary connections this volume seeks to make. In the final chapter ('An epilogue'), DeGraff reviews the individual contributions and articulates more specifically what the key research questions emerging from the collection have been. These include: What role does imitation play in these related processes? Can everyone undertake the same strategies for dealing with variable input? Is all variable input dealt with equally? Is there any special relationship between processes of L2A and L1A when the output of the former constitutes input to the latter? What defines 'salience' of input-frequency, social factors, linguistic markedness? He reprises the overlaps between preceding contributions, places where results from one chapter feed assumptions or results of another, and points of conflict (a good deal of space is given to attempting to bridge the complementary positions on the role of children vs adults in the formation of creoles). This is somewhat repetitious for the reader who has just completed the volume, but will make it invaluable for those who need to read more selectively from the contributions. He concludes with his own goal, namely that future research might result in "a theoretical framework... in which morphological triggers (or absence thereof) are related... to a clustering of syntactic properties in pidgins and creoles and also in language change and in language acquisition" (523). DeGraff's chapters provide impressive coverage of historical and current research in creolization and language change; his discussion is not restricted to one theoretical paradigm (functionalists, formalists and social dialectologists are represented), nor is it restricted to North American research and theory. His own contributions are sufficiently substantial that one could imagine other creolists being tempted to market them as a slim monograph of their own. Chapters 2-5 deal most closely with the LBH. Bickerton ('How to acquire a language without positive evidence', ch.2) outlines his now well-known position that the input to creoles constitutes an extreme and extraordinary case of the degraded input all children are exposed to. Since the pidgin input lacks the syntactic structure (specifically the grammatical morphemes) of a natural language, children are thrown back on their own resources. As a result, UG/bioprogrammatic features of natural language surface as the default; these can be taken to be the fixed, unmarked structural properties of human language that a child (in extremis) can rely on. Lexical categories from, e.g., English are then bleached of their meaning and co-opted to fill the necessary grammatical functions. As evidence for the role of UG he points to the apparent lack of variability in the English creole in Hawai'i, despite the substantial variability between, e.g., Japanese, Portuguese and Hawaiian speakers' pidgin. This scenario highlights two important requisites for application of the LBH: speed with which the creole emerges from a pidgin (one generation exactly) and the degree of continued contact with substrate languages (none). Sankoff (1994), for instance, has shown that even children who are monolingual speakers of Tok Pisin (i.e. do not know a substrate language), may actively be grammaticizing features found in substrate languages. Bickerton has exempted the Melanesian creoles from the predictions of the LBH on the grounds that the LBH is only triggered when the input is especially morphosyntactically degraded, and the lengthy period of stabilisation for Tok Pisin, Bislama and Solomons Pijin has allowed a good deal of structure to be taken over from the substrates. One might like to know precisely when the sort of data Sankoff discusses does become notable, i.e. what degree of morphosyntactic complexity constitutes enough to prevent the child from having to rely on the LBH, and what degree of consistency across and within speakers is required for the input to be interpreted by the child as good enough. (These questions are addressed in other chapters, e.g. Sprouse and Vance, Lightfoot, Roberts.) One criticism that has often been levelled at the LBH is the fact that, notwithstanding Bickerton's claim that the English creole spoken in Hawai'i "resembles... other creoles formed under similar circumstances" (54), virtually every other creole brought out to test the LBH against has been placed beyond the pale, usually because the input at the crucial time is not deemed sufficiently degraded. Assuming the LBH generates unmarked structure, Adone & Vainikka ('Acquisition of wh- questions in Mauritian Creole', ch.3) ask, is it the case then that children acquiring the ability to form questions in a creole go straight to the adult norm? The results were mixed. Mauritian Creole allows both long distance and partial wh- movement from embedded clauses; children seem to favour partial wh-movement in the tests, and don't get the long distance movement until they are older (5-8yrs). A&V suggest that partial wh- movement might be the UG default (it also occurs in English-speaking children's responses). It is also possible, given the length of time since MC creolized (18thC), there has been some language change from the LBH. Mufwene ('On the LBH: Hints from Tazie', ch.4) considers the extent to which features emerging in his daughter Tazie's speech between 20-30 mths parallel the LBH. Obviously, the nature of the input is rather different, however Mufwene finds overlaps (Tazie, for instance, initially lacks an overt complementiser, and negates predicate adjectives and verbs as if there were distinct classes). He suggests that L1A and creolization are similar insofar as the core elements of each are less marked (here 'markedness' sometimes seems to be a measure of frequency; sometimes a measure of mutual compatibility between systems). The absence of some typical creole features in Tazie's speech however prompts Mufwene to suggest that adults play a non-trivial role in the formation of creole grammars. Lumsden ('Language acquisition and creolization'. ch.5), too, argues that adults play a major in creolization through relexification of their L1, and draws on parallels between creoles and L2A errors/interlanguage forms. Lumsden suggests that pidgins lack a pronounceable form of the functional categories found in the speakers' L1(s) (cf. Bickerton who sees these categories as being wholly absent), and Lumsden proposes that relexification is motivated by the need to assign phonological form to these categories. As to the question of why the functional categories of the lexifier do not get carried over in some form to the creole, he proposes a principle that only elements with some denotational semantics are available for relexification. He argues that the lexical items speakers reanalyse to satisfy the need to express the functional categories of their L1 are selected primarily on the basis of semantic criteria, perhaps also phonological, but that syntactic factors are "irrelevant" (152). (De`prez' discussion of negation in French and French creoles (ch.12) provides a good example of how restricted the transfer of aspects of a word's semantics may be.) However, the discussion of the Fongbe substrate for Haitian Creole prepositions left me rather unclear of what Lumsden considers semantics, what syntax and what might be best considered preferences for information structure. In fact, it is not clear how little a role Lumsden believes that syntax plays in the process of relexification. The account of the transfer of Fongbe NP structure to Haitian Creole relies on speakers having both a fairly sophisticated knowledge of not just Fongbe, but also French syntax. This is somewhat reminiscent of Myers-Scotton's (1997) recent proposals that speakers actually have some competence (if not proficiency) with the syntax of the lexifier. Chapters 6 and 7 are important contributions to the literature on creolization. Newport ('Reduced input in the acquisition of signed languages') and Kegl, Senghas & Coppola ('Creation through contact: Sign language emergence and... change in Nicaragua') review diachronic studies of the acquisition of signed languages by an individual (Newport) and by groups (KS&C). In both cases, the input is characterised by irregularity, idiosyncrasy and features that are not typical of natural languages. Newport shows that although Simon bases his sign system on what he is exposed to through his parents (i.e. his restructuring of the irregular input is not as extreme as the LBH might predict), he does not simply match the patterns found in his (non-native) parents' signing. Instead he produces a "cleaner, more rule-governed system" (168), forcing the tendencies in the input "to be internally consistent" (173). Newport asks the interesting question of why Simon so carefully regularises the variation he was exposed, whereas in the normal course of events children produce variable output when their caregivers provide variable input, e.g. Roberts ). She suggests several reasons for this: the degree of internal variation for each parent; the variation between his parents; the absence of systematic conditioning factors associated with the variants. KS&C's chapter is a weighty contribution reviewing a number of studies conducted in Nicaragua with signers of various lects, and evaluating the results from these studies as evidence for creolization vs L1A. They show that a major influence on the kinds of features acquired (.e.g. person and spatial agreement marking, object classifiers, anaphora) is the age at which the signer came into contact with Nicaraguan signing (contact before c.7 years is required for full acquisition of the system). An interesting aspect of both this study and Newport's is that the input varieties cannot be analysed as substrate languages. So the analysis of their results is a relatively clean test of the LBH. The emergence of a same-switch reference system governing the distribution of overt and null anaphora in Nicaraguan Sign must, for instance, be independent of the same/switch reference systems constraining the distribution of null anaphora in varieties of Spanish (discussed in Cameron 1992) due to the social segregation of the signing and Spanish-speaking communities studied. In addition, the highly variable input to acquisition in these cases makes them qualitatively different from L1A. Henry and Tangney ('Functional categories and parameter setting in L2A Irish', ch 8) looks at what children in Irish immersion programmes in Belfast do with the non-native, but fluent, input of their teachers. Here the input is 'restricted' largely in terms of the amount of time that the learners are exposed to Irish. H&T find several characteristics of the learners' Irish that differs from L1 (maturational) errors. The children do not go through a period of using SVO word order On the other hand the children do show signs of simplifying the copula system, eliminating the form ('is') which occurs in a more restricted set of tense-mood domains. Even though c.10% of their teachers' copula sentences occur with this copula, this does not seem to meet the threshold required for the learners to pay attention to it. H&T suggest that the learners have created a more 'simple' grammar, where simple means a grammar in which there is "a unique specification of [V- or N-features] for each functional head" (239). In the accelerated learning context of immersion classes, they propose that such UG considerations override the inclination to match the input of the teachers. It is also possible, of course, that 10% of copula sentences is simply not frequent enough to prompt pattern-matching. Sprouse and Vance ('An explanation for the decline of null pronouns') is a fine example of the study of language variation. S&V carefully define the 'envelope of variation' for the alternation between phonetically null and overt pronouns in Romance and Germanic languages (some previous accounts for French are rejected on the grounds that they generalised from non-equivalent data). They conclude (following Kroch 1989) that variation between syntactic forms reflects not a mixed grammar but competition between two (or more) underlying grammars. The discussion of the alternation between overt and null pronouns in Old Icelandic contributes an important insight to this model, because this case shows that two anaphora may come into competition due not to differences in their licensing but to differences in their referential properties (and hence the binding conditions they are subject to). Ian Roberts' chapter ('Verb movement and markedness', ch.10), like H&T, attempts to specify markedness in terms of principles of UG. Here, Roberts proposes that 'weak' values for a parameter are less marked than 'strong' ones, and he embeds his analysis within Clark's (1992) learnability theory and subsequent developments of this. Roberts states clearly his position that creoles have no "privileged relation to UG" (303), like any other language, the extent to which they depart (or not) from unmarked parameter settings is a direct function of the nature of the triggers the learner is exposed to. One consequence of Roberts' analysis of verb movement in Haitian Creole (and English and French) is that he tentatively concludes UG may set movement parameters differently for lexical verbs and auxiliaries. This would, in H&T's terms, create a less 'simple' grammar, and this raises the possibility of further research and dialogue between contributors' various positions. Adrienne Bruyn, Pieter Muysken & Maaike Verrips ('Double-object constructions in the creole languages', ch.11) is a splendid example of the quantitative study of language variation. The documentation for their conclusions is especially rich (and for this reason perhaps somewhat less clear-cut than in others, but for all that more compelling). This chapter seems most closely related to those of H&T and Roberts. BM&V ask why double object constructions (DOC) not only occur in almost all creoles, but why they are also the preferred means of expressing the arguments of ditransitive verbs. They review L1A data showing that DOC emerges very early and in some cases before prepositional dative constructions (PDC); they suggest the DOC is less marked than the PDC because it requires setting only one parameter while the PDC requires the setting of two. The actual definition of 'unmarked' that is proposed is not actually based on parameters, it specifies an unmarked setting as the one children will choose "regardless of the input they receive" (363). One of the other merits of BM&V's contribution is that they consider data that falls outside of a P&P model of syntax. They note pragmatic (dare one say, functional?) properties associated with the distribution of indirect and direct objects and unstressed pronouns in DOCs/PDCs and show how these effects are remarkably consistent in the historical records of Sranan and Negerhollands. They argue that such factors play(ed) a significant role in the development of synchronic syntactic patterns. Viviane De`prez ('The roots of negative concord', ch12) looks at French and French-lexified creoles (especially Haitian). The chapter investigates whether negative concord is the most unmarked expression of negation (the notion of gradience is De`prez'). In the end, it is unclear what her conclusions are re. markedness. Her proposal is that negative concord is a consequence of semantic (not syntactic) properties of the negative words in a language. She suggests negative words are indefinites and they may lack intrinsic quantificational force (the case in Haitian) or they may behave more like a numeral ('zero', the case in French). (DeGraff in ch.15 raises some difficulties with this analysis.) De`prez argues that both types of N word may license concord but that the nature of the concord will be qualitatively different in the two systems. The bulk of the paper is given over to demonstrating that in Haitian, negative concord involves unselective binding of N words by an operator, while this is not the case in French. Like Baptista's (1997) work on Capeverdean negation, De`prez shows that even though an element may carry over phonological form and order in the clause from the lexifier, this may mask some major underlying differences between element in both languages. These sorts of studies of the outcomes of language contact and creolization inject a welcome degree of sophistication to analysis of creole structure. They provide a challenge to those who continue to propound strong theories of substrate transfer or relexification. Although exponents of this approach increasingly talk in terms of 'adult SLA', it is clear that this is just relexification dusted off and given a new dress for the millenium. Notwithstanding the problems with any particular analysis (Rizzi and DeGraff both have reservations about De`prez' analysis), I welcome work that understands languages (and therefore language contact) involves something more that surface word order facts and phonological form. The collection concludes with review papers by David Lightfoot ('Creoles and cues', ch.13) and Luigi Rizzi ('Broadening the empirical basis of UG models', ch. 14) and DeGraff's synthesis paper (discussed above). Rizzi's chapter is committed to stressing fundamental similarities between the structural features that emerge in creoles and in other kinds of L1A, arguing (contra Bickerton and KS&C) not to privilege creoles on account of the circumstances in which they evolve (cf. Roberts) and the speed with which change occurs in them. Lightfoot engages primarily with the papers discussing verb movement in the preceding collection, and outlines his own learnability theory (more sensitive to quantitative aspects of the input than, e.g., Roberts'), and he evaluates historical data on basic word order from English and synchronic data from Berbice Dutch. Like BM&B, Lightfoot's contribution tries to determine with some precision how two grammars in contact (here, Dutch and Ijo) might synthesise and emerge with novel forms. On the question of verb movement, he suggests Roberts has overgeneralised the extent to which verb movement is proscribed in creoles, and notes (citing Sankoff on Tok Pisin) that some English-based creoles do allow the main verb to move to the head of the clause for emphasis. This may be true, but we should also be realistic about including it as a possible 'cue' for the child learning Tok Pisin. I cannot speak for Tok Pisin specifically but in conversational Bislama verb fronting is extremely rare. In my corpus, I have subjects, objects and adverbs being fronted for emphasis but no (that's as in De`prez' 'zero') verbs. In conclusion, the volume is a substantial contribution to the study of language contact and language change. It places creolization (I believe, rightly) in the larger context of these fields, and generally de-emphasises any supposed uniqueness of the structure of creoles. It assumes familiarity with the principles and parameters model of syntax (some ability to code-switch between P&P and minimalist terminology) and is probably best suited to reasonably advanced graduate readers or subject specialists. For those readers, it provides a stimulating perspective on L1A, L2A and historical evidence of language change, building bridges between different subfields of linguistics in a way that is ultimately most rewarding. ( A note on production: There are a few substantive typographical errors in the text (e.g. p438, l.2 . 'lack' should be 'have'; p382, l.6 should surely read "pas must NOT be a head"; and p532, fn.28 should read "sociolinguistic variability in LANGUAGES is transmitted across generation"). DeGraff et al. were somewhat also let down by the editorial staff at MIT Press who let through a number of other less significant typos.)
References Baptista, Marlyse 1997. On the syntax of negation in Capeverdean. Paper presented at meeting of the SPCL, Chicago. Cameron, Richard 1992. Pronominal and null subject variation in Spanish: Constraints, dialects and functional compensation. PhD dissertation, University of Pennsylvania. Clark, Robin 1992. The selection of syntactic knowledge. Language Acquisition. 2. 83-149. Kroch, Anthony S. 1989. Reflexes of grammar in patterns of language change. Language Variation and Change. 1, 3. 199-244. Myers-Scotton, Carol 1997. 'Matrix language recognition' and 'morpheme sorting as possible structural strategies in pidgin/creole formation. In Arthur K Spears & Donald Winford (eds) The Structure and Status of Pidgins and Creoles. Amsterdam/Phila: Benjamins. 151-174. Roberts, Julie 1997. Hitting a moving target: Acquisition of sound change in progress by Philadelphia children. Language Variation and Change. 9, 2. 249-266. Sankoff, Gillian 1994. An historical and evolutionary approach to variation in the Tok Pisin verb phrase. In CLS 30, vol. 2 (Parasession on variation in linguistic theory). 293-320.
About the reviewer: Miriam Meyerhoff is Assistant Professor in Linguistics at the University of Hawai'i at Manoa. She works on language contact, language variation, and language and gender, in Bislama and English.
Miriam Meyerhoff Asst. Prof., Department of Linguistics 1890 East-West Rd University of Hawai'i at Manoa Honolulu, HI 96822