Lexical semantics without thematic roles By Yael Ravin (review)

238 LANGUAGE, VOLUME 68, NUMBER 1 (1992) to see that his common-sense evaluations of family trees, waves, dialects, and diasystems, still topical in today's continually renewed examination of historical processes and relationships , lead offthe collection. These assessments are followed by five papers on both specific and general topics of segmental phonology (diphthongs , sound laws, IE phonology, Greek phi in Romance, and French schwa). Most of the eight papers on prosody also have a Romance flavor, not surprising given P's primary areas of interest and expertise. Students of French, in particular, will be happy to re-establish contact with P's views on the behavior of schwa and the French 'prosodies' (pause, syllable structure , stress, and pitch) and, in the next section, with his six analyses of certain morphological and (to a lesser extent) syntactic matters. Of particular note in the morphological domain, especially in light of the current resurgence of typological studies, are the papers dealing with verbal paradigms and the functions of tenses. In the section 'On languages, future', P indulges in comments on topics that the less daring might hesitate to address. After first disposing of the notion of progress in language (280), he ventures various predictions concerning future developments of the French vowel system and verbal morphology, English derivatives in -wise, and Italian ciavere, each conditioned , at least in part, by pressures from colloquial spoken varieties. And finally, in a partially autobiographical, partially exhortatory epilogue, we find P's observations about his more than forty years as a student of language. One clear conclusion is that a sceptical eclecticism and a concern for profound mastery of the intricacies of individual languages can lead to great insight into language itself, a claim for which On Language and On Languages provide ample testimony. [Douglas C. Walker, University of Calgary.] Lexical semantics without thematic roles. By Yael Ravin. London & New York: Oxford University Press. 1990. Pp. viii, 248. Cloth $49.95. Theories ofmeaning (semantic representation or logical form) can be broadly classified as either autonomous or restrictive: the autonomous view, championed by Jerrold Katz, maintains that there is no general correspondence between semantic structures and syntactic ones (only a partial dependency); the restrictive view, held by Chomsky, among others, views thematic roles as essentially syntactic primitives and limits lexical semantics to thematically-defined argument structure. In this book, a revision of her 1987 City University of New York dissertation, Ravin argues forcefully against the restrictive view and in favor of the position that semantics is decompositional, nonthematic, and autonomous of syntax. The first halfofthe book begins with chapters entitled The relationship between syntax and semantics' and ? restrictive versus a non-restrictive approach', in which R sets out the general issues she wishes to explore. She then goes on to discuss Charles Fillmore's Case Theory, Government and Binding Theta Theory, Ray Jackendoffs Conceptual Semantics, and the MIT Lexicon Project. R's exposition involves both summarizing these approaches and pointing out problems that arise. Discussing GB, for example, she argues that thematic roles cannot be justified in terms of the theta criterion, or in terms of any independent requirements on Logical Form, or in terms of the goal of subsuming subcategorization information to thematic-role assignment. When she discusses the MIT Lexicon Project (R refers here to work by Beth Levin & Malka Rappaport and by Kenneth Hale & S. Jay Keyser), she argues that semantic classes of verbs cannot serve as a basis for thematic -role structure and that roles cannot be directly correlated with syntactic relations. R is more sympathetic toward Jackendoffs Conceptual Semantic theory, which is autonomous rather than restrictive and which is not strictly based on thematic roles: her main criticisms concern certain apparent complications in handling types ofcausation and change and in Jackendoff s commitment to a motional model of semantics (see now Jackendoffs Semantic structures, Cambridge, MA: MIT Press, 1990, for the most recent version of his Conceptual Semantics). The second half of R's book contains two chapters entitled ? theory of semantic decomposition ' (I and II) and two entitled 'An analysis of some event concepts' (I and II), along with a concluding chapter on The relationship between syntax and semantics revisited' . This part of the book...