This book examines the acquisition of syntax in Seselwa and Morisyen children. Seselwa and Morisyen are historically related creoles spoken in the Seychelles and Mauritius, respectively. Working within Chomsky's (Reference Chomsky1981) framework, Adone focuses on the acquisition of (non-)reflexive pronouns, double object constructions (DOCs), passive and serial verb constructions (SVCs).

The book consists of nine chapters and four appendices. Chapter 1 provides information on these two creoles. Chapter 2 highlights issues in language acquisition. Chapter 3 outlines the four syntactic phenomena. Chapter 4 is on research methodology and participants in different experiments. Chapters 5-8 discuss these experiments and their results. Chapter 9 considers their implications for a theory of language acquisition.

The central thesis is that Seselwa and Morisyen children transcend their input since they have no access to a conventional language model. However, from the evidence presented in this book, it is difficult to see whether, and to what extent, this is really the case. As far as one can tell, these children's output appears to be modeled on, and constrained by, constructions available in abundance in adults' speech. This is particularly so with DOCs and SVCs. Furthermore, the claim that they lack a conventional language model is at odds with the fact that Seselwa and Morisyen became full-fledged languages in the late 1700s (Baker & Corne, Reference Baker and Corne1982).

The acquisition of reflexive (e.g. herself) and bare (non-reflexive) pronouns (e.g. her) in these children (chapter 5), Adone argues, demonstrates a pattern of development found in Chien & Wexler's (Reference Chien and Wexler1990) children (i.e. continued progress in the acquisition of principle A, but delay in the development of principle B). These two principles are central to Chomsky's (Reference Chomsky1981) Binding Theory. Informally, principle A requires that an anaphor (a reflexive or a reciprocal like ‘each other’) have a local antecedent (i.e. an antecedent in a clause which contains it, e.g. Mary likes herself), while principle B requires that a bare pronoun have a non-local antecedent (i.e. an antecedent outside the clause which contains it, e.g. Jane wants Mary to like her, where ‘her’ refers back to ‘Jane’ or another female person, not ‘Mary’). However, important differences are overlooked. First, Chien & Wexler's experiment shows a rather flat development of principle B, whereas Adone's shows a sharp U-shaped development. Second, children in Adone's study appear to perform just as badly on principle A as on principle B. These differences may be due not only to the fact that pronouns have non-local and local antecedents in adult Seselwa and Morisyen, but also to reflexives functioning as emphatic pronouns, which allows them to have non-local antecedents (see examples (3), (5) and (8) in the children's spontaneous data). Importantly, adult bare pronouns violate principle B since they can have local antecedents. This raises the question of whether these children's performance concerning principle B is measurable. Still, what is clear is that, contrary to what is claimed, these children are not re-creating their language but simply acquiring the adult input.

The acquisition of DOCs (chapter 6), it is argued, shows children using DOCs with verbs which adults only use as PDCs (Prepositional Dative Constructions), and the use of DOCs increasing with age as the use of PDCs falls. This reveals an interesting correlation, except that several of the verbs claimed to have only a PDC pattern in adult Seselwa also display a DOC pattern: examples (26)-(27), which are claimed to show children surpassing their input, are available in adult Morisyen. Likewise, (25) and (28) are possible in adult imperatives. Only (23)-(24) seem to support the above claim, although even these can be analysed as preposition-less PDCs. Note also that genuine DOCs ‘V-NP3-NP2’ and preposition-less PDCs ‘V-NP2-NP3’ are conflated as DOCs, which inevitably makes the PDC responses look lower than they are.

The acquisition of passive (chapter 7) is studied using van der Lely's (Reference van der Lely1996) comprehension test and Verrip's (Reference Verrip1996) production test. The crucial finding is that Seselwa and Morisyen children, like their Dutch counterparts, understand passive from an early age despite these constructions being rare in the input. This would be an interesting finding except it is far from clear whether Seselwa and Morisyen have passives. Without passive auxiliary (equivalent to English be or French être) and morphology, most basilectal creoles are said to lack short and long passive. Some (e.g. Seselwa and Morisyen) have a few English-like get-passives (Corne, Reference Corne1977; Papen, Reference Papen1978), as shown in the data on pages 123-25. Other passive-like constructions are in fact unaccusative constructions (e.g. (2), (5), (7) and (8)). Examples such as (44) (p. 47) are arguably topic constructions with a null subject rather than passive.

The comprehension experiment (a picture pointing experiment, where children point to one of four pictures on hearing a test sentence), it is claimed, shows children preferring the more complex verbal/actional passive to adjectival/stative passive, particularly when presented with an ambiguous passive. Verbal passive, unlike adjectival passive, is complex because it involves movement of the object (Borer & Wexler, Reference Borer, Wexler, Roeper and Williams1987). This leads Adone to conclude that these children have a verbal representation of passive from an early age (p. 134). This contrasts with the performance of English children (Borer & Wexler, Reference Borer, Wexler, Roeper and Williams1987; van der Lely, Reference van der Lely1996). However, the preference for actional interpretation in Seselwa and Morisyen children may have another explanation: constructions such as sa tediber fin/pe kud ‘the teddy bear is/is being mended’, without passive auxiliary and morphology, can be interpreted as active topic constructions with a null subject. Additionally, the use of progressive pe in the ambiguous short passive sa tediber pe kud also favours actional rather than stative interpretations. The high reversal responses (subject of passive interpreted as agent) in such constructions are also expected since the verb kud can be used intransitively, meaning ‘the teddy bear is sewing' rather than ‘the teddy bear is being sewn’.

The production experiment (a sentence completion exercise), on the other hand, is said to show children displaying knowledge of implicit arguments, as suggested by their preference for instrumental phrases. This in turn is said to suggest a syntactic representation of passive in these children. However, the evidence – particularly from the responses under the copula condition – appears problematic. Under this condition, children are required to complete a test sentence such as Pyer pe penn. Sa pu en madam ek … ‘Peter is painting. This will be a woman with …’ either with a comitative, e.g. en lisyen ‘a dog’ or instrumental, e.g. en peinso ‘a paintbrush’. The prediction is that they will produce fewer instrumental than comitative responses under this condition than under a causative or anti-causative passive one. In fact the opposite is the case, and the instrumental preference under this condition is attributed to non-linguistic factors. However, one possibility, which is not considered, is the strong association between the verb penn ‘to paint’ and the instrument en peinso ‘a paintbrush’ (e.g. 34). It may be this rather than knowledge of implicit arguments that favours the instrumental response. Additionally, there is an instrumental/implicit argument bias in the choice of ganny ‘get’ as in Get en sat fin ganny rule ek en baton ‘Here a cat is rolled over with a stick’. A further confounding factor is ek, which functions as both preposition and conjunction (e.g. 31). On the conjunction reading, the choice of zoranz ‘orange’ in (Gete, en pom ape ganny manze ek … (en zoranz ‘orange’/en furset ‘fork’) ‘Here is an apple being eaten and/with ... (an orange/a fork)’) is blocked, as it means illicitly extracting a conjunct out of a coordinated phrase.

The finding on the acquisition of SVCs (chapter 8) is interesting but again, contrary to what is asserted, most of the SVCs in the spontaneous data (p. 146-9) are very common in adult Seselwa and Morisyen (see Bickerton, Reference Bickerton1989; Syea, Reference Syea2013). It is no great surprise then that children use them productively.

Adone finally claims that the existence of Seselwa and Morisyen alongside English and French, lack of literacy and corrective input, and optionality of certain functional categories (p. 28) lead to children being exposed to variable and ambiguous input, but it is difficult to see how or why. The input which Seselwa and Morisyen children receive is neither qualitatively nor quantitatively different from that which children acquiring other languages receive. Having observed pre-school children learn their native creole in Mauritius on several occasions, I find Adone's claim concerning access to a conventional language model very surprising indeed. Comparison with the acquisition of Sign Languages or with cases of language deprivation seems unwarranted, and instances of over-regularisation (e.g. in the acquisition of DOCs or SVCs) found in these children are not different from those in children acquiring other languages.

The book is well organised, and it contributes to the database and literature on child language. Clearly, since little exists on the acquisition of creoles, this is a much welcome addition. However, more care should have been taken with the empirical data. Likewise, a few of the test sentences on passive (chapter 7) should have been more carefully selected to avoid any bias in interpretation. Presentation, interpretation and discussion of the quantitative data, particularly in chapters 6 and 7, should have been clearer and more informative. Inclusion of pictures alluded to in the reflexive and pronoun experiments and additional information on the experiments on DOCs and SVCs would have helped the reader. Above all, claims concerning adult Seselwa and Morisyen constructions should have been more carefully checked. Overall presentation is frequently marred by poor editing and typos in both text and figures (see figure 7.1–7.3 in particular).