Mixed categories: Nominalizations in Quechua (review)

142LANGUAGE, VOLUME 66, NUMBER 1 (1990) I see a general need for better definitions for all the terms that are used in metrical discussions; ictus, stress, and rise and weight, strength, and prominence are overlapping concepts, but not necessarily identical, and it cannot be taken for granted that all practitioners use them with the same meanings. I do not have any better suggestions at the moment, but I recognize the need for some. Last but not least, I would like to congratulate the translator for the excellent translation. Department of Linguistics[Received 27 July 1989.] 204 Cunz Hall of Languages The Ohio State University 1841 Millikin Road Columbus, OH 43210 Mixed categories: Nominalizations in Quechua. By Claire Lefebvre and Pieter Muysken. Dordrecht: Kluwer, 1988. Pp. xvii, 304. $69.00. Reviewed by Mark Baker, McGHl University 1.Main topics. This book is a detailed study of a substantial portion of the syntax and related morphology of the Quechua language (the Cuzco dialect), based primarily on the authors' extensive fieldwork over a period of many years. The focus of the book is a class of nominalized clauses that constitute the principal means of complementation in the language, but Lefebvre and Muysken also discuss many other aspects of the language in connection with these. For each main point that arises, L&M present the data, state descriptive generalizations, and sketch a formal analysis within a modified GovernmentBinding framework based on Chomsky 1981 and related work. They then try to bring these analyses to bear on general theoretical issues, including such questions as what the fundamental difference between a noun and a verb is and why morphologically complex languages like Quechua are so different from English. Quechua nominalizations are interesting because they have both properties that are normally associated with clauses and properties that are normally associated with noun phrases (NPs). Semantically, they take obligatory subjects and refer to tensed propositions, just as clauses do. Syntactically, they bear Case morphology, just as noun phrases do. Most curious of all, the Case marking on the NPs inside the nominalization can follow the pattern characteristic of clauses (nominative -0 on subject, accusative - ta on object), or the pattern characteristic of NPs (genitive —q on subject, objective -0on object), or a mixture of the two (nominative -0on subject, objective -0on object). The central question is: how can such hybrid phrases be accounted for in a theory of syntax? 2.Main ideas. At the heart of L&M's approach to this problem is a particular system of syntactic features and the rules of projection that create phrases from those features. This system is laid out in Ch. 2. Following Chom- REVIEWS143 sky, L&M break up the syntactic categories into sets of binary features, with noun = [ + N, -V], verb = [-N, +V], and preposition = [-N, -V]. L&M then give formal expression to the fact that nominalizations have properties in common with both nouns and verbs by assigning them the feature complex [ + N, +V]. In Chomsky's original system, this set of features is associated with the category adjective, but L&M argue that adjectives are merely a special subclass of noun in Quechua. (They also claim that there is little content to the idea that adjectives share properties of both nouns and verbs in Chomsky's original scheme.) The +V feature allows nominalizations to assign Case to NPs within their phrases, just as ordinary verbs do. The +N feature, on the other hand, allows the Case assigned by some higher verb to be realized on the head of the nominalization, just as Case is realized on the head of an ordinary NP. L&M go on to show (in Ch. 4) that Case must be realized on all categories in Quechua—adverbs, adpositional phrases, and clauses, as well as NPs. From this they derive the fact that nominalization is the normal form of complementation in the language, because only in this way can the head of a clausal unit bear its Case. In addition to this modification of the feature system, L&M loosen slightly the principle of projection that relates the features of the head of a phrase to the features of the whole phrase. They suggest that in Quechua the features of the phrase can be the same as the features...