Despite its more general title, Jeff Siegel's book deals primarily with contact varieties that have emerged in the Australia-Pacific region within the last 160 years (mainly Melanesian Pidgin English and Hawai‘i Creole English). Much of the data comes from previously published articles, but Siegel explains that he has revised his analysis and arguments in order to ‘form a more coherent picture of the emergence of pidgins and creoles’ (7). Siegel's conclusion that the formal simplicity typical of restricted pidgins is the same as that found in adult second language acquisition is not new. However, the book is aimed not solely at creolists, but also at researchers in the field of second language acquisition. Siegel's main argument is that a significant degree of morphological expansion took place in the pidgin predecessors of creoles. A large proportion of the expanded features were modeled on the grammatical morphology of the respective substrate languages through the process of transfer or reinforcement. Substrate transfer occurs in individuals as they expand their use of a pidgin, and substrate reinforcement occurs at the community level during the stabilization of the contact language (199).

Chapter 1, ‘Introduction’, sets the scene by defining some of the contact varieties to be discussed. Debates still rage within pidgin and creole studies about the precise nature of pidgins and creoles, and in particular whether creoles constitute a distinct class of languages on linguistic grounds. Siegel sidesteps this issue by saying that it will not be a focus of his book, and that the languages he refers to as creoles are those for which the label is not controversial. For instance, he considers Melanesian Pidgin English an expanded pidgin rather than a creole because the majority of its speakers still speak other languages and it is not the vernacular language of any distinct newly formed community (4). Nevertheless, there are still problems in drawing firm boundaries between pidgins and creoles as well as between them and other kinds of contact varieties such as the so-called New Englishes widely spoken in places such as India, Singapore and elsewhere. Other long-standing controversies that Siegel does take up include the questions of whether creoles generally have pidgin ancestors, whether a continuum of varieties normally emerges after a creole has developed or whether it was present from the beginning, why some substrate features appear in some creoles and not others, and why some creoles appear to have more substrate features than their pidgin predecessors (7).

Chapter 2, ‘Morphological simplicity in pidgins’, draws on data from Pidgin Fijian to illustrate some of the ways in which linguists have attempted to define the notion of simplicity. Siegel defines it mainly on the basis of surface structural features, and focuses on modular features rather than a global assessment of simplicity across entire linguistic systems. In practice, this boils down to the presence or absence of inflectional morphology as a means of expressing particular semantic distinctions such as tense or evidentiality. Chapter 3, ‘Morphological simplicity and expansion in creoles’, elaborates the idea that the expression of semantic distinctions with grammatical or purely functional morphemes rather than with lexical items is an indicator of greater morphological complexity (58). Siegel's metric of complexity relies on the notion of cline as used within grammaticalization studies, in which an item may be more or less grammatical or lexical, depending on its location along a continuum ranging from content item→grammatical word→clitic→inflectional suffix.

In chapter 4, ‘Sources of morphological expansion’, Siegel uses the grammaticalization of the temporal adverb baimbai (from by and by ‘afterwards/later’) as a marker of irrealis/future in Tok Pisin to illustrate the importance of internal development as a resource for morphological expansion. Ever since the general scenario was first proposed by Sankoff & Laberge (Reference Sankoff, Laberge, DeCamp and Hancock1974), this example has become paradigmatic for grammaticalization processes in creoles in much the same way that the Romance future tense has served for languages without a prior pidgin/creole history. A three-stage process was hypothesized in which the temporal adverb baimbai appears clause-initially (e.g. baimbai mi go ‘Later I go’), and then in phonologically reduced form is incorporated into the verb phrase in preverbal position (e.g. mi bai go ‘I will go’), via an intermediate stage in which reduced bai can be either clause-initial or preverbal. A similar scenario has taken place in Tok Pisin's sister varieties of Melanesian Pidgin English, namely Solomon Islands Pijin and Bislama (spoken in Vanuatu).

Nevertheless, there are numerous problems with this conventional three- stage model (Romaine Reference Romaine, Bybee and Fleischman1995). Phonological reduction and syntactic position are separate issues. The existence of examples such as (1), which goes all the way back to the earliest extant textual records and predates the beginning of the Pacific labor trade in 1863, indicates that the incorporation of the full form baimbai within the verb phrase probably existed as a grammatical option long before phonological reduction or creolization, or indeed, the existence of a community of fluent second language speakers.

1. (1) Me bye and by come back. (New South Wales 1844; Troy Reference Troy1985) ‘I'll come back.’

In addition, examples with preverbal baimbai can still be found (albeit infrequently) in current spoken and written varieties of Tok Pisin. The position of baimbai/bai is therefore not a straightforward reflection of diachronic stages. The existence of preverbal bai today may derive from in situ reduction of preverbal baimbai rather than by movement and, therefore, the hypothesized stages do not correspond to discrete points in time. This means that clause-initial bai need not represent an intermediate stage in the grammaticalization.

In any case, despite his discussion of this example, Siegel does not regard grammaticalization as one of the most important processes in morphological expansion in creoles on the grounds that grammaticalization normally takes place gradually over centuries rather than within one or two generations, as typically happens in expanded pidgins and creoles, and that changes due to grammaticalization are usually considered strictly internal and therefore distinct from contact-induced language change (65–66). Despite these objections, however, interest in grammaticalization in creoles has been increasing since at least the mid-1990s. Indeed, Hopper & Traugott (Reference Hopper and Traugott2003: 212) raise the question of ‘whether studies of language contact situations raise special issues regarding grammaticalization’. They go on to say that ‘pidgins and creoles illustrate in rather extreme form a number of theoretical insights into how the progression of grammaticalization across languages as well as time can be conceptualized’ (Hopper & Traugott Reference Hopper and Traugott2003: 213). Heine & Kuteva (Reference Heine and Kuteva2005: 242) contend that contact-induced grammaticalization is ‘a major driving force in the grammatical development of pidgins and creoles’. More specifically, they use the term ‘replication’ to describe the process whereby speakers create a new use, pattern or category in their language based on the model available in some other language. The outcome is not an exact copy of the model language, but is shaped by what is available in their own language, by universal constraints on conceptualization, by what speakers deem to be pragmatically appropriate in the situation in which contact takes place, and by the length and intensity of contact (Heine & Kuteva Reference Heine and Kuteva2005: 7). Interestingly, Heine & Kuteva (Reference Heine and Kuteva2005: 103) observe that ‘among all grammatical categories it is future tense that appears to be the most likely to be replicated in situations of language contact’.

This remark is interesting in the light of Bickerton's (Reference Bickerton1981) claims about the universality of a creole TMA (tense–mood–aspect) prototype, consisting of a three-member inventory of preverbal particles marking the same semantic functions. Creole TMA markers have assumed a morphosyntactic function they did not have originally and moved down the cline of grammaticality to become more grammatical. Siegel dismisses Bickerton's so-called ‘bioprogram hypothesis’, i.e. the idea that creoles draw on our universal innate capacity for language; and when Siegel notes that his demonstration of substrate influence in the morphological expansion of Melanesian Pidgin English would count as evidence for Bickerton, this is presumably an error (91). Bickerton does not consider Melanesian Pidgin English relevant to the bioprogram.

Significantly, these new TMA distinctions can all be thought of as new and previously unexpressed categories, something which Heine & Kuteva (Reference Heine and Kuteva2005: 83) believe is common in contact-induced grammaticalization but less often the case in non-pidgins/creoles. Nevertheless, they stress that pidgins and creoles do not behave any differently with respect to grammatical replication in general and contact-induced grammaticalization in particular; that is, they work in the same way regardless of whether the language involved is a pidgin or a creole (Heine & Kuteva Reference Heine and Kuteva2005: 242–243). As a rule, however, replicated grammatical categories are less grammaticalized than the corresponding model categories (Heine & Kuteva Reference Heine and Kuteva2005: 117). Heine & Kuteva add that some scholars appear to be ‘overly concerned’ about the distinction between contact-induced change deriving from substrate influence and language-internal or universally-induced change. Thus, substrate languages may provide the models that creoles replicate, but rather than replicating a grammatical concept or category, creoles replicate a grammaticalization process (Heine & Kuteva Reference Heine and Kuteva2005: 539).

In chapter 5, ‘Transfer’, however, Siegel argues that what Heine & Kuteva call contact-induced grammaticalization is ‘neither grammaticalization in its usual sense nor a process distinguishable from functional transfer’ (141). Substrate influence is manifested by dint of transfer of forms from the lexifier language along with properties of grammatical morphemes from the substrate languages. For Siegel, it is the use of a pidgin for wider functions that triggers morphological expansion: ‘the ultimate extension of use of a pidgin L2 occurs when speakers shift to it as their primary language, which they pass on to their children’ (129). This explains why some creoles appear to have more substrate features than their pidgin predecessors. In order to compensate for shortcomings in their L2 (second language), speakers draw on their L1 (first language) as a communicative strategy to fill in gaps arising from the expanding functions of the pidgin. This essentially endorses Sankoff's (Reference Sankoff1994: 314) argument that maximal substrate influence occurs not at the point of initial contact, but at the point of language shift.

In chapter 6, ‘Constraints on substrate influence’, Siegel identifies congruence and perceptual salience as the most important constraints on substrate influence (169). As far as the lexifier is concerned, the most important factor appears to be its degree of availability during the expansion phase of the pidgin when speakers are shifting to it as their primary language (170). Chapter 7, ‘Substrate reinforcement’, argues, among other things, that differential patterns of substrate reinforcement led to differences between the three varieties of Melanesian Pidgin English. The issue of grammaticalization is relevant here too, but is not discussed. Chapter 8, ‘Predicting substrate influence’, attempts to apply this framework for predicting the nature of substrate influence to the TMA system in Tayo (spoken in New Caledonia) and to the verb phrase in Roper Kriol (spoken in Australia). Siegel concludes that although his approach is ‘nowhere near perfect’ (234), it offers more than other substrate analyses (such as, for example, the essentially unconstrained relexification hypothesis).

This book will be very useful to creolists as well as to students of language acquisition.