Child language: The parametric approach (review)

William Snyder's new book fills two principal roles: first, it is a very useful graduate-level textbook for courses in formal approaches to child language acquisition; and second, it constitutes a well-developed argument that children's language, as measured by spontaneous production data, embodies the property of GRAMMATICAL CONSERVATISM, or, in S's words, that 'children do not begin making productive use of a new grammatical construction in their spontaneous speech until they have both determined that the construction is permitted in the adult language, and identified the adults' grammatical basis for it' (8). There are a number of things that should be said about the interesting property of grammatical conservatism and about spontaneous production data, but first I address the excellent course-companion dimension of the book.

After a brief introductory chapter, the second chapter introduces the idea of grammatical conservatism. Next, S draws a parallel between the way in which adult generative syntactic research in the principles-and-parameters framework has sought out syntactic macro-parameters crosslinguistically, and how child-language research can follow a similar approach by considering various stages of development to be 'comparable to a new language' (7). Considering children's developmental stages as new languages allows researchers to test predictions about the connections among dimensions of syntactic theory through their sometimes asynchronous development. This approach follows in the spirit of the CONTINUITY ASSUMPTION of Pinker (1984), Macnamara (1982), and others, which asserts that each stage of a child's grammar is an internally coherent system and may only be constituted of the same symbols and computations as adult systems. This assumption in nativist approaches to child language implies that, while linguistic theory may be used to characterize child language, child-language data may also be used to construct and evaluate adult linguistic theory.

The rest of the chapter is dedicated to a basic presentation of the mechanics of Noam Chomsky's minimalist program, which S elucidates, using the work of Longobardi (1994, 2001), Bobaljik and Thráinsson (1998), and Boškovic (2004) as examples of varying degrees of adherence to the minimalist idea of keeping the locus of syntactic variation outside of the computational component and in the lexicon. For the language-acquisition students with a background in graduate-level syntax who used this book in a course with me, this was an accessible introduction to minimalism. The appendix with the sample minimalist derivation was particularly helpful. The third chapter similarly serves as a brief introduction to government phonology and optimality theory.

Chs. 4 and 5 serve as a very effective guide to conducting formal linguistic child-language research with spontaneous production data, particularly those available from the CHILDES (child language data exchange system) database. In Ch. 4, S walks readers through a search for the onset of verb-particle constructions in the recording sessions that chronicle the language development of Sarah, from Brown's (1973) corpus. This includes instructions on how to use several of the CLAN programs, which, though the programs are well explained in the CHILDES documentation, are helpful when presented in the context of an actual study that S performed (Snyder & Stromswold 1997)—a technique S uses effectively throughout the book to illustrate its main points.

Ch. 5 is particularly important in its explanation of the binomial statistical test for determining the degree to which two constructions are acquired simultaneously or not in a child's linguistic development. This infrequently used calculation is useful for making more rigorous claims about the simultaneous onset of constructions that are hypothetically related in children's grammars. S illustrates the use of the test with his work on the development of compound words and verb-particle constructions (Snyder 2001).

Ch. 6 is a helpful discussion of statistical methods that can be used for cross-sectional studies, which include tests more familiar in behavioral science, including the Pearson correlation, the chi-square test, and the paired t-test. With further examples of how S's approach to child language...