Mental representations: The interface between language and reality Edited by Ruth M. Kempson (review)

408 LANGUAGE, VOLUME 67, NUMBER 2 (1991) The prefixes ko- and /- then have their own entries elsewhere in the dictionary. K includes the Proto-Athabaskan source and reconstructed morpheme for each entry for which this information is known. There is also a brief but interesting discussion of the relation of Ahtna to PA in the introduction. K has conducted his research in collaboration with a large number of Ahtna speakers. Ahtna has four regional dialects, which, according to K, differ only in lexical and phonological details . Each entry in this dictionary indicates the status of the lexical item in each of the dialects. In addition, the introduction includes a thorough overview of the major phonological differences among the dialects. The overall organization of this dictionary is based on K's theory of Ahtna word formation, and this foundation makes the work particularly accessible. K's model, which he has developed in more detail elsewhere, is lucidly summarized in the Introduction. The model distinguishes levels of word formation according to the abstract categories Lexicon, Derivation. Inflection, and Post-Inflection, rather than according to the surface order of the morphemes . Generalizations about lexical relatedness are thus captured in a perspicuous way, and the data are made transparent to the linguist who is interested in theoretical issues. The appendices provide (a) lists of loanwords, numerals, and kinship terms; (b) a summary of topics in verb phonology; (c) a discussion of directional words; and (d) paradigms of theme formation, aspectual derivation, nonaspectual derivation, and complete inflected verbs. For its target audiences, this dictionary is a superb achievement, and it is likely that lexicographers of non-Athabaskan languages will find that it provides a valuable model. [Margaret Speas, University of Massachusetts . Amherst.) Mental representations: The interface between language and reality. Ed. by Ruth M. Kempson. Cambridge: Cambridge University Press, 1988. Pp. viii, 229. Cloth $39.50. This book is a collection of papers originally presented at a conference on 'Mental Representations : Properties of Logical Form", held in England and organized by Ruth Kempson. The principal impetus behind the conference and the book was an interest in understanding to what degree some of the major approaches to the interface between language and cognition might be compatible, and especially to what degree these might inform research in Sperber & Wilson 's 'Relevance Theory'. The volume contains some excellent representative papers from researchers in Government & Binding Theory (GB; James Higginbotham, Robert May, and Michael Brody & M. Rita Manzini). Situation Semantics (Robin Cooper and Elisabet Engdahl), and Relevance Theory (Deirdre Wilson & Dan Sperber, Diane Blakemore, and Robyn Carston). The result is a very useful and interesting book that anyone with an interest in semantics, pragmatics, or syntax will enjoy reading. However. 1 myself was not persuaded that any synthesis of views is likely to be pieced together. In fact, these models still seem too far apart on even their object ofinquiry to consider a potential synthesis, although there are certainly areas of overlap, as Kempson points out in her introduction and concluding article. Kempson's thought-provoking and stimulating introductory article provides an excellent overview of the various approaches in the book. The articles by Higginbotham, Cooper, and Engdahl are concerned with the nature of linguistic semantics. Higginbotham argues for a view that finds relevant for natural-language semantics only those features of meaning which belong to the linguistic system, that is, those attributes which are used by speakers and referred to by grammatical rules. Purely technical issues, e.g. analyticity and syntheticity, fall outside the scope of linguistic semantics. He discusses the well-known problem of disquotational statements and how these can be used to shed light on interesting semantic problems. Cooper and Engdahl argue for ways of incorporating the notion of situation into linguistic semantics. Cooper proposes a situation model of negation, conjunction, and disjunction and then considers whether his findings should be interpreted in a mentalistic or platonistic fashion (more specifically whether invariants of linguistic interpretation are to be interpreted in terms of a language of mental representation or a general cognitive filter or part of the nonmental world). He largely suspends judgment on the matter, claiming (58) that ? do not think that the study of semantics alone will...