The title of this volume promises an overview of early child bilingualism and its relation to language contact phenomena in general. Given the title and my qualifications to assess an overview, I agreed to review this work. Instead, too late, I found that the volume is largely a case study of a small set of structures produced by six Cantonese–English bilingual children in Hong Kong. Because I have very limited knowledge of Cantonese and I am not a bilingual child language researcher, readers should interpret my assessments against these disclosures.

This book is an addition to the long publication record of Yip & Matthews (henceforth Y&M) on Cantonese and specifically on Cantonese–English child bilingualism. In brief, it details the production of a set of word-order and other changes, which are presented as instances of transfer – from the children's version of Cantonese to their version of English, and also from their English to their Cantonese. The issue raised is whether these word-order changes are indeed best called transfer. This is an important question, in light of the long-standing arguments and evidence in the literature that young bilingual children keep distinct at least the morphosyntax of their two languages. A related issue is the question as to what input the children may have received, even though a ‘one language–one parent’ regime was in effect, in a community where Cantonese prevails and it is unlikely that standard varieties of English are widely spoken. Y&M suggest that language dominance plays a major role in non-target production, which is a controversial topic in the literature, although space precludes full discussion of both sides of this issue.

The first three chapters of this volume are introductory. Chapter 1, ‘Introduction’, is wide-ranging but sketchy. Chapter 2, ‘Theoretical framework’, introduces relevant theoretical approaches and issues, and Chapter 3, ‘Methodology’, details methodological issues. In chapter 2, Y&M formulate a hypothesis about different rates of development for a given property× xacross languages: in the context of bilingual acquisition, if language A develops property× x first, this facilitates transfer of the property to language B. In effect, knowledge from one language is used to deal with another (44). This hypothesis is a main theoretical underpinning for the authors' interpretations of the data, along with the notion of dominance of one language and the recognition of ‘vulnerable domains’ created by cross-linguistic structural ambiguity.

The children's backgrounds are relevant to our interpretation of the data and the authors' conclusions. All six children studied are offspring of cross-cultural marriages, with parents who are native speakers of either standard varieties of Cantonese or English (five of the six mothers are native speakers of Cantonese). Yip, a native speaker of Hong Kong Cantonese, and Matthews, a speaker of British English, are the parents of the three principal children studied. All six children were exposed to both Cantonese and English from birth. Yip & Matthews acknowledge that quantity of input from outside the parents is an issue, given that the main language of Hong Kong is Cantonese, and they report that ‘at least up to age five, the balance of input increasingly favors Cantonese throughout the preschool years’ (65).

What about organized exposure outside the home? All of the Yip–Matthews children (Timmy, his 2-years younger sister Sophie, and seven-years younger Alicia) attended kindergartens where they heard both Cantonese and English. Timmy began attending a bilingual kindergarten from age 2;04 for three hours each day, and later attended a Cantonese-medium kindergarten in the morning and an English-medium one in the afternoon. The other three children came from similar family backgrounds but had different patterns of exposure. For example, Llywelyn, who, along with the three Yip–Matthews children, was Cantonese-dominant, had an older brother who played an important role in his language ecology. Another child, Kathryn, attended the Cantonese section of an international kindergarten from age 2;07, but her primary caregiver was her British English-speaking mother. Of the six children studied, she was considered most balanced in the two languages. Charlotte, the younger of two children, and the only one of the six children in Y&M's study judged to be stronger in English, was born in New Zealand, but moved to Hong Kong when 4 months old, where a Filipina domestic worker was her caregiver.

Y&M argue for a correlation between dominance and degree of transfer of the word-order features they studied. For example, they note that ‘extensive transfer from Cantonese is observed in Timmy and Llywelyn during periods of Cantonese dominance’ (80). Y&M see dominance as ‘related to underlying competence and not merely a measure of performance or language use’ (36). Mean Length of Utterance measured in words (MLU_w) is used to compare the children with each other regarding language dominance.

The authors indicate that frequent codeswitching characterizes adults' speech in middle-class Hong Kong families. While they note that the children engaged in some codeswitching, they do not consider its structure; for example, they do not identify the source of the abstract morphosyntactic frame of the bilingual clauses and therefore which language is the matrix language of these clauses (Myers-Scotton Reference Myers-Scotton2002).

The age span studied varies across the children from about age one to age three. Some studies begin at age two (two cases) or even three (one case); most end around age three, with one ending at age four. The children's everyday activities were audio-recorded and, in some cases, also video-recorded. The children were encouraged to speak Cantonese for half an hour and English for half an hour. Y&M kept diaries about their own children's language development, but used them only for qualitative analysis. The database contains 352 files in two languages.

Chapters 4, 5, and 6 are the heart of the volume; they contain the quantitative data that Y&M argue show transfers from the children's Cantonese to their English production. For comparison, similar data on the same structures from Singapore Colloquial English are also reported in some cases.

Chapter 4, ‘Wh-interrogatives: To move or not to move?’, deals with constituent questions. While wh-movement is required in English, it is not an option in Cantonese. Wh-in situ shows up in the English production of all six children, but it is most evident in the four Cantonese-dominant children (Timmy, Sophie, Alicia, and Llywelyn). For example, Timmy, at age 2;05, produces such utterances as It is for what? (87). For Sophie and Alicia, 100% out of their total of eleven (Sophie) and nine (Alicia) non-formulaic where-questions are in situ. Later, movement becomes optional and eventually where-questions are produced as in standard English varieties.

Y&M conclude that language dominance ‘plays a clear role here’ (127) because the Cantonese-dominant children contrast with the English-dominant Charlotte and the balanced child Kathryn in their rates of producing wh-in situ in their English. However, the authors also state that input ambiguity may be a factor along with dominance; that is, in addition to hearing Cantonese structures, the children may hear echo questions in English and apply their word order more generally.

Chapter 5, ‘Null objects: Dual input and learnability’, investigates the use of null objects of transitive verbs, which are understood, but not pronounced, in the English production of the Hong Kong children. For example, Alicia, at 2;08, utters I want to put, while carrying a jar of face cream (133). Of course, English has null objects, but it contrasts with Cantonese where null objects are a basic feature, highly frequent and acquired early. Generally, the four Cantonese-dominant children produce more null objects with five transitive English verbs in comparison with the other two children. Again, language dominance and input ambiguity are invoked as explanations.

Chapter 6 deals with ‘Relative clauses: Transfer and universals’. The children studied by Y&M produce prenominal relative clauses in their English; that is, the relative clause precedes the head noun, as it does in Cantonese, rather than following it as it does in standard English varieties. Thus, Timmy, at 2;10, uses You buy that tape is English? to express the question ‘Is the video tape that you bought in English?’ (155). The children's prenominal relatives mostly show object relativization and are primarily of a type that is characteristic of spoken Cantonese. Moreover, the children's relative clauses share the canonical word order of main clauses. However, Y&M do not provide many examples in the book, and because the ones cited contain a number of null elements, they are difficult for the reader to interpret.

Chapter 7, ‘Vulnerable domains in Cantonese and the directionality of transfer’, shifts the focus to transfer from English to Cantonese. The authors argue that certain ‘vulnerable domains’, where possible word orders overlap between the two languages, are the focus of these transfers. The chapter considers placement of prepositional phrases in the children's Cantonese, specifically ones with hai2 ‘at’. In the children's speech, the prepositional phrases appear in postverbal position within the verb phrase ([V PP]), notwithstanding the fact that the target order in Cantonese is [PP V]. Of course, [V PP] is the main English order. Moreover, dative constructions with Cantonese bei2 ‘give’ appear in the children's Cantonese with verb–recipient–theme order, whereas verb–theme–recipient order is expected in Cantonese. The proportion of non-target forms is nearly twice that of target forms. Finally, non-target order in verb particle constructions is found in the children's Cantonese. In the target Cantonese structure, a pronoun is not allowed between the verb and its particle; yet, such examples occur in the children's Cantonese.

Because of space limitations and the focus of this review on the volume's central topic, I have little to say about chapter 8, which deals with ‘Bilingual development and contact-induced grammaticalization’. In this chapter, the authors suggest that not all grammaticalization is internally motivated. In intense contact situations, structural similarities across languages can promote grammaticalization across languages. They cite transfer in bilingual child language acquisition as a possible mechanism behind some contact-induced grammaticalizations.

The final chapter, ‘Conclusions and implications’, recapitulates Y&M's central claim about the importance of language dominance; the authors ‘see pervasive influence of the dominant language on the weaker language’ (258). But the notion of input ambiguity is similarly important, and also in an asymmetrical way: the language where there is no ambiguity of evidence (i.e. where word order is uniform) will influence the language with ambiguity (259f.).

Several important unanswered questions remain. First, assuming that Y&M's data reflect transfer, how central to the grammar are these phenomena? The main aspects of word order that are affected are wh-interrogatives and relative clauses. In some ways, these and null objects do not affect basic argument structure, but rather the (non-)mapping of arguments onto co-indexed and recoverable null elements. Further, most of the word orders that Y&M argue are transferred show some variability in the supposed recipient language. Still, even if one accepts that there is cross-linguistic transfer in these data, I stress the point made above that these are peripheral features of the grammar.

The many researchers who have reported on bilingual acquisition involving Indo-European languages do not report such cross-linguistic patterns of transfer. What they do report supports the notion of a separate development of the morphosyntax of early bilingualism (cf. De Houwer Reference De Houwer, Kroll and DeGroot2005 for an overview). Y&M affirm ‘the assumption that the bilingual child has two separate and differentiated systems’ (257), but they do not comment on the relation between this assumption and the transfer that they detail.

Second, the question of input remains, especially concerning the children's exposure to English. Could what is called Hong Kong English have been an important part of their input? In Hong Kong, at least 95% of residents are first-language speakers of a variety of Cantonese. English is a second official language, but standard varieties of English are not widely spoken. Another variety of English is spoken, however widely, called Hong Kong English, which is divergent enough from standard English varieties to be included in the set of varieties researchers call ‘World Englishes’. Moreover, what about the English input from outside the family? I do not question the quantity of input here but its quality. All the homes seem to have had Filipina domestic help who did not speak Cantonese, and it is questionable whether they in fact spoke standard English. Similarly, it is not clear whether the teachers/caregivers and other children at the kindergartens that the children attended all spoke standard English. Y&M acknowledge that most of the children had more input in Cantonese than in standard English, but they are silent on a possible role for Hong Kong English.

In sum, this book deserves attention for its extensive data base. In addition, it is an important book because of the analyses that it gives for the word-order patterns in the data, and because of the issues that the data and their interpretations raise. It is a ‘must-read’ for researchers studying language acquisition. Moreover, since it deals with the permeability of grammars, even if at the peripheries, its relevance to linguistic theory in general is obvious.