Abstract
This is an excellent book for philosophers, and others concerned with natural language and cognition, who have not kept up with post-Aspects work in syntax, in particular with the Extended Standard Theory work on government and binding that relates to anaphora and quantification. It is a direct challenge to those who think that there must be a reasonably clearcut semantic level of description for sentences in natural language, one which is crucial for explaining how we learn, understand, and use natural language. Hornstein argues that the properties of quantification such as relative quantifier scope and pronoun binding in natural language are best explained in purely syntactic terms. "Not only are the explanations best construed syntactically but, more important, the distinctions and concepts needed to articulate them cut against the semantic distinctions one would naturally make". This syntactic theory will, of course, be supplemented in a full theory of language use by pragmatic theories that will assign meaning, force, reference, etc., in particular contexts, given all sorts of real world knowledge. But what we won't have will be a compositional, context insensitive level of semantic description in which true, denotation, object, model, etc., loom large. Though Hornstein is a philosopher, his arguments derive from a linguist's familiarity with recent syntactical work; however, the general line is consonant with Stich's From Folk Psychology to Cognitive Science: The Case Against Belief, in which Stich argues beliefs are so context and intrepretive-stance relative that they cannot be grist for a scientific psychology, which must rather limit itself to syntactical characterizations of sentences in cognition.