Michael Tomasello has an impressive opus of publications in both child language development and primate research. This book represents a synthesis of that work as well as the exposition of an account of language development, which provides a thought-provoking challenge to the generative enterprise. As such, it is an important contribution to the literature of first language acquisition, and by extrapolation, of SLA, because it offers an account of how learners can extract language from the input without reference to linguistic-specific predispositions. It succeeds in many respects and fails in others, as this review will try to explain.

The basic argument is that children become symbolic as a result of their capacity to (a) “read” the communicative intentions of others, (b) acquire words and holophrases through repeated exposure to them in the input, (c) create generalizations through schematization and analogy out of the concrete pieces of language they have heard, constrained by processes of entrenchment and competition, and (d) carry out a functionally based distributional analysis to arrive at paradigmatic categories of linguistic constituents. The claim is that children do not learn rules or representations and that only general cognition is needed to account for the gradual abstraction of language from concrete exemplars experienced over the course of early and middle childhood. Tomasello leaves the reader in no doubt as to how he views alternative proposals when he says:

As this quotation makes clear, the book is an attempt to provide evidence not only for the usage-based account but also against the generative account.

Having described usage-based linguistics in chapter 1, the author considers in chapter 2 the emergence of symbolic capacity and contrasts children with animals in this respect. Chapter 3 considers the acquisition of first words, and chapter 4 examines the acquisition of early constructions, which are argued to be lexically restricted and not evidence for genuine (lexically neutral) syntactic representations. Chapter 5 argues explicitly against both the generative and the connectionist enterprises—the latter for being too form based and not including communicative intention in the modeling—and in favor of the three processes of (a) analogy, which allows children to create new utterances on the basis of ones they have heard; (b) entrenchment of constructions through repeated exposure; and (c) preemption, which prevents unattested possible generalizations winning out over attested ones. This chapter is the central statement of the model, and the remaining ones simply flesh it out in relation to more complex constructions.

Chapter 6 covers nominal and clausal constructions as means of reference and predication and discusses the acquisition of tense, aspect, negation, and so forth. Chapter 7 covers complex sentences and discourse, with the final two chapters laying out the philosophy of the model under the titles of “Biological, Cultural, and Ontogenetic Processes” (for chapter 8) and “Toward a Psychology of Language Acquisition” (for chapter 9).

I found the early chapters of this book exciting and convincing. The evidence for the beginnings of a theory of mind underlying the capacity to be symbolic and of interpersonal understanding steering the child's attention to early words and holophrases depending on perspective (I would say it was Relevance driven in the sense of Sperber & Wilson, 1995) accords well with the work of people such as Hobson (2002) and is a convincing account of early communication. As Tomasello's account progresses through early childhood, however, it becomes, to my way of thinking, more and more strained. Although it gets further than many earlier pattern-extracting accounts (and includes critiques of these; early Slobin, for example), it still does not fully appreciate the complexity that the generative account is trying to explain.

Some arguments simply will not touch Tomasello. For example, to complain that he offers no account of the distributional and interpretive facts about conjoined clauses or “do so” serves no purpose. For Tomasello, language is as it has evolved phylogenetically to be. Back in the mists of time, everything in language came into existence to meet functional communicative needs, but changing needs, processing constraints, and inefficient transfer from one generation to the next have, as it were, messed things up, leaving modern children with an imperfect system peppered with vestiges of historical change that they pick up because they are there. More to the point, however, are long-distance dependencies such as “Which ghost did John imagine Fred saw last night?” in which “which ghost” must be interpreted as the object of “saw” and not “imagine” or (to borrow an old example from Otsu, 1981) “What is the little girl drawing a picture of a monkey drinking milk with?” This asks about a picture of a girl drawing a monkey drinking milk with a straw, where the answer must be something like “a crayon” and cannot be “a straw.” These are the kinds of complexities that generative theory has been built on and that do not figure into Tomasello's account at all. There can be no doubt that these are real facts about native speaker knowledge, so why are they not addressed? Part of the answer seems to be that usage-based theoreticians have not yet got around to tackling these kinds of facts. Tomasello admits as much when he says, after a brief discussion of coreference in utterances such as “Beside him, Bill saw a snake” that “no developmental research has been done on issues of coreference from a cognitive-functional (usage-based) point of view” (p. 207). It might have been wise to wait until such an account could be developed before trying to go head-to-head with generativists with all the aggression that the quotation at the beginning of this review suggests.

In fact, there are not a few promissory notes in this volume and even though (on p. 304) Tomasello acknowledges that his account does not fully succeed in demonstrating that a cognitive-functional account of acquisition can work, I could not help wondering whether this book was not perhaps a bit premature. To give a flavor of what is still to be done, on page 231, he says, “no one has yet proposed a comprehensive account from a usage-based or construction grammar perspective of how children integrate TAM [tense-aspect-modality] marking with their utterance-level constructions”; on page 300, he says that “we still know very little about the specifics of how all of this [referring to entrenchment and preemption] works.” On page 318, he says, “As their utterance-level constructions become gradually stronger and more abstract, children may begin to find links among them in various ways. We are not in a position to say much about this at the moment, as there is very little relevant research.” Certainly, those who favor a cognitive-functional or usage-based approach can find plenty of research topics in this volume. Those who prefer a generative account remain frustrated by the criticisms.

The readers of SSLA will no doubt be wondering why, if first language acquisition can be accounted for with an approach such as this, SLA seems to be so difficult. Surely, adult learners have not lost their general cognitive capacities. There are a few references to SLA in this book, and together they suggest that, according to Tomasello, adults do not get enough simple constructions as input to work with and the first language “interferes” (p. 313) with the acquisition of the second. No doubt readers will have their own responses to these as causes of incomplete SLA, depending on their theoretical stance. However, the reason I agreed to review this volume for SSLA was that I thought an explicit usage-based account might provide food for thought for those who would appreciate discussion of a well-articulated set of mechanisms that do not make reference to Universal Grammar. Those who are in this camp will, I think, find such a discussion, even though I think the model as a whole is not entirely successful.

Finally, I must draw attention to the fact that Tomasello's whipping boys are too frequently just Chomsky (mostly from work published in the early 1980s) and Pinker (1994). One wonders why he does not put his account up against more recent generative accounts and particularly why he does not compare his work with that of Jackendoff. Jackendoff (2002) not only has several points of contact with Tomasello, particularly in the importance and role of the lexicon, but he actually addresses some of Tomasello's published criticisms of his own work and would have made for a much more satisfying debate partner.