Foundations of Language: Brain, Meaning, Grammar, Evolution

Article excerpt

Foundations of Language: Brain, Meaning, Grammar, Evolution by Ray Jackendoff (Oxford: Oxford University Press, 2002. 477 pp. $40.00. ISBN 0-19-827012)

MANY SCIENTISTS outside the field of linguistics have been frustrated by the lack of connection between linguistic theory and psycholinguistic models or investigations of the brain bases of language. Jackendoff attempts to bridge this divide by laying out a model of language that clearly relates to models of language processing and that can guide investigations of neural systems underlying language. Given the primary focus of the book, Jackendoff could have stopped with the main title, Foundations of Language, but the subtitle, Brain, Meaning, Grammar, Evolution, was needed to hook cognitive neuroscientists and psycholinguists. Attracting a broad readership is important because Jackendoff wants to convince cognitive neuroscientists and psycholinguists to give linguistics another chance and argues that his model provides an impetus for a new dialogue between linguistics and these related fields. To a large extent Jackendoff is successful-his proposals provide an accessible and useable foundation for cross-disciplinary research. In a similar vein he also urges linguists to adopt a more integrative approach to the study of language and argues that Chomsky-driven "syntactocentrism" must be abandoned.

Jackendoff reconfigures generative grammar into a parallel architecture containing a set of components (e.g., phonological, semantic, syntactic) and interface modules that link the structures within each component. The interfaces are not just between adjacent components (e.g., between syntax and semantics), but also between more distant ones, such as phonology and semantics. This latter interface accounts for certain intonational phenomena and lexical items such as ouch, which have a phonological and a semantic structure, but no syntax. In addition, the architecture allows for a connection between semantics (conceptual structure) and articulatory and perceptual systems-an interface that Jackendoff hypothesizes arises during language evolution but still remains and accounts for phenomena such as onomatopoeia in spoken languages. For sign languages, this interface might be used to account for the iconic mapping from the articulators to conceptual structures that Taub (2001) proposes in her dualmapping analysis of ASL metaphor. Rathmann and Mathur (2002) also use this interface in their account of the use of space in verb agreement and pronominal reference in signed languages. The parallel architecture JackcndofFproposes is extremely promising for studying properties of language that might be unique to signed languages.

Foundations of Language is divided into three parts. The first part (Psychological and Uiological Foundations) lays out the complexities of language that exist in even the simplest sentence-a complexity that nonlinguists often fail to recognize. This section ends with an important discussion of the theory of Universal Grammar (UG) that continues throughout the book. In the end, Jackendoff finds that "the dominant view of Universal Grammar as a highly complex specification of all possible grammars-whether in terms of parameters or ranked constraints-is untenable" (426). Feather, he argues that UG largely consists of architectural universals (i.e., what kinds of linguistic structures exist and how they interact) and some structural fragments ("attractors") that have evolved over the history of the species and that bias a child's acquisition of linguistic generalizations. Such an approach is much more amenable to cross-disciplinary approaches to understanding our innate linguistic endowment (after all, only human primates talk). It is my belief^ however, that the term Universal Grammar must be abandoned within such an approach because (a) it is too easily confused with "linguistic universals" (as Jackendoff notes), (b) it is already associated with the "grammar box" approach, and (c) the term currently engenders vehemently negative responses from many researchers outside the field of linguistics (even those who would favor Jackendoff's approach). …