a ‘a '4 ..J'.;= PHILOSOPHICAL PSYCHOLOGY, VOL. 2, NO. 1, I989 55 Philosophical Speculation and Cognitive Science GEORGE LAKOFF Linguistics Department, University of California, Berkeley, CA 94720, U.S.A. Lycan’s book Logical Form in Natural Language raises important general issues, issues that transcend the specific issues discussed in the book, which are now in many cases 20 years out of date. For me, the most important issue is the proper relationship between philosophy and cognitive science. How does a priori philoso- phical speculation about language and mind affect research in cognitive science? And should results from cognitive science change the way philosophers go about their business? Discussing Lycan’s book is a useful a way of addressing such questions. I will spend most of this discussion on the book itself, and return to the general issues at the end of the paper. Let me say at the outset that there is a major thing I like and admire about Lycan’s book: Lycan has actually read a lot of the relevant linguistics and thought about it carefully. It is something one cannot say about many other philosophers who write about language. The critical comments I have about Lycan’s book are made from the standpoint of a sincere appreciation for the author’s efforts. There are a myriad of small problems with Lycan’s book, but they are arise from one major one: Lycan simply never addresses the concerns of the papers by the linguists that he is criticizing. Let me explain. I and other linguists within the generative semantics tradition (which Lycan mostly discusses) have abided by a criterion of adequacy that we take as essential to the study of linguistics as a scientific endeavor, namely: The Generalization Criterion: linguistics is primarily concerned with the statement of general principles governing linguistic elements and struc- tures at all levels. The Generalization Criterion comes with characterizations of syntax, semantics, and pragmatics. —Syntax concerns the statement of general principles governing the occur- rence of grammatical morphemes (like -ing), grammatical categories (like preposi- tion), and grammatical constructions (like the imperative construction). —Semantics concerns generalizations governing semantic elements and struc-