An encyclopedia of AUX: A study in crosslinguistic equivalence By Susan Steele et al. (review)

REVIEWS An encyclopedia of aux: A study in crosslinguistic equivalence. By Susan Steele et al. (Linguistic Inquiry, Monograph 5.) Cambridge, MA: MIT Press, 1981. Pp. viii, 328. Cloth $32.50, paper $15.00. Reviewed by Ellen M. Kaisse, University of Washington The admirable goal of this volume (hereafter EA) is 'to put the question of the similarity of categories across the grammars of particular languages on a substantive empirical basis' (2). Specifically, Steele and six co-authors attempt to specify a set of criteria which will identify a category aux (auxiliary), and to test the linguistic utility ofthe entities thus identified in the analysis of several languages. The criteria will be familiar to readers ofAkmajian, Steele & Wasow 1979 (ASW), since this monograph is an expansion and revision of the ideas set forth there. An aux is to be recognized across languages as follows: 'Given a set of language internal analyses, those constituents which may contain only a specified (i.e. fixed and small) set of elements marking tense and/or modality will be identified as non-distinct [and called aux].' (21, emphases supplied)1 If, the authors claim, one applies these criteria to several languages, and if other, non-definitional properties emerge as common to the category, then the linguistic significance of aux will have been supported. Some might disagree with the basic premises of this inquiry (for instance, the largely surface criteria which are used); however, my major reservations lie elsewhere. First, not all the analyses given are of sufficient depth and plausibility to convince me that any constituent at all has been identified. Because the sample presented in the body of the work is limited to four languages, this is a serious problem. Second, the majority of the non-definitional properties that emerge from the investigations have a more likely explanation than one based on the constituency or the Aux-hood of the formatives in question. Chap. 1, 'The theoretical framework', is a brief introduction to the aux controversy and to the orientation of the book. The central chapters of EA contain the language analyses (Chap. 2, ? definition and its instantiations') and a discussion of the non-definitional properties (Chap. 3, 'Equivalence'). Chap. 4, 'Further exploration', is a rather cryptic discussion of the semantics of aux. Two appendices, 'Aux in English', by Richard Oehrle, and 'The aux in German and the history of English', by Thomas Wasow and Adrian Akmajian , would have more profitably been integrated into Chap. 2—and their implications treated in Chap. 3—particularly since they deal with languages whose grammars are more thoroughly studied than those in the body of the book. Chap. 2 contains discussion of putative aux constituents in Luiseño (by 1 Steele persists in calling this list of criteria a definition—which, strictly speaking, it is not. 924 REVIEWS925 Steele), of Lummi (by Richard Demers), of Cairene Egyptian Arabic (by Eloise Jelinek), and of Japanese (by A. Akmajian and Chisato Kitagawa). By far the longest and most convincing treatment is accorded to Luisefto (Uto-Aztecan), which Steele has studied extensively and with insight. The worst criticism one might level at this section of the book is that it is too detailed for the present purpose; many of the facts recounted here do not bear crucially on the establishment of an aux, and would find a more congenial home in a monograph on Luiseño. Responding to criticism of ASW by Pullum 1981 and Kaisse 1981, S eschews arguments based on the phonological behavior and syntactic position of the particles that she claims form an aux. Instead, she argues that the particles have a semantic interdependence which would be accidental if they were not members of a constituent. Since that putative constituent has members conveying tense and modality (as well as some other notions, e.g. subject agreement), the terms of the definition are met. Arguments based on co-occurrence restrictions and semantic interdependence are not indisputable, but they are certainly suggestive; the reader is likely to concur with S's conclusions . However, Demers' analysis of Lummi (Salishan) suffers from the defects which both Pullum and I found in ASW. Lummi contains a set of elements, identified by D as particles, which...