136Reviews Meaning-TextTheory:Linguistics,Lexicography,andImplications. 1990. James Steele, ed. Ottawa: University of Ottawa Press, ? + 489 pp. $35.00 U.S. Steele's book is a collection of thirteen papers on the nature and application of Meaning-Text Theory (MTT), the approach to linguistic description developed principally by Igor A. Mel'cuk and Alexander K. Zholkovsky some 20 years ago in the Soviet Union. MTT is a theory of the relationship between meaning and linguistic form; central to MTT is the lexicographically better known Explanatory Combinatorial Dictionary (ECD), the formal, theoretical (and, generally, non-commercial) lexicon of MTT. The book has three parts: 1. Linguistics: Theoretical Contexts (two chapters on the motivation and historical position of MTT); 2. Lexicography: The Explanatory Combinatorial Dictionary (six chapters on ECD entries) 3. Implications and Applications (five chapters on psychological and computational uses of MTT). There are also three appendices, an MTT glossary, an entry for an ECD of Chinese, and a bibliography of MTT writings in English and French. Alexander Nakhimovsky's "Word Meaning and Syntactic Structure: Some Comparative Notes" begins the volume with a survey of the lexicon and syntax in MTT and a comparison of MTT with current lexical theories, mainly Lexical Functional Grammar. Nakhimovsky rightly argues that Mil's lexical focus predates much of the current lexically based work, though it is odd to call this predating "anticipation" (as he does on p. 11), since MTT was developed pretty much outside the MIT/Stanford grammatical tradition that has produced lexical grammar. However, the proper historical and epistemológica! context of MTT is elucidated by Johanna Nichols in her "The Meeting of East and West: Confrontation and Convergence in Contemporary Linguistics" (reprint of 1979). Nichols contrasts the metatheories of MTT and generativism, laying out MTT's structuralist sympathies, e.g., its penchant for illustrated (versus argued ) truth and its distributional (versus deductive) claims. The paper is something of an apologia for MTT, a good piece to read for an understanding of why MTT proceeds as it does: e.g., why Government Binding theory tries to predict the appearance of certain kinds of languages, whereas MTT produces a detailed dictionary (like the ECD), combining meaning, syntax, and usage. The next six chapters, Part 2, comprise the bulk of the book. James Steele and Ingrid Meyer's "Lexical Functions in an Explanatory Combinatorial Dictionary: Kinds, Descriptions, and English Examples" has two goals. The first is to illustrate and explain in English the lexical functions of the ECD (the abstract operators that determine the syntagmatic and paradigmatic collocations of an entry word). The second is to suggest a general classification of lexical functions to facilitate their use and recall by the practitioner. Reviews137 The first goal is motivated by the fact that most lists of lexical functions —at least at the time Steele and Meyer wrote their paper—do not give English illustrations (though this depends to some extent on where and how you look: see, e.g., Evens 1980). Frankly, I originally thought this goal to be beside the point, since the existing illustrations are quite clear to begin with (see Apresjan, Mel'cuk, and Zholkovsky's (1969) English/Russian examples) and linguists and lexicographers generally are familiar with at least one ofthe languages (French, German, and Russian) in which the lexical functions have traditionally been explained. But Steele and Meyer explain ECD lexical functions with admirable clarity, and so in spite of my worries, I am convinced of this paper's utility and strength. I am less satisfied with their general classification of lexical functions, however. They broadly classify all lexical functions as either paradigmatic or syntagmatic and give a number of subcategories. However, I do not see the rationale for many of the classifications. For example, MANIF (manifestation ) and NOCER (harmful to) are classed with the "attendant features," but even though "attendant features" are described as "circumstantial," I do not understand what this classification means; MAGN (intensification) is with the "evaluative qualifiers," and GERM (the germ of) is with the "process features ," but I do not see why. My objections might bejust cavils if it were not that the very motivation for the labels is mnemonic ease. (A more useful taxonomy...