The polysynthesis parameter By Mark C. Baker (review)

REVIEW ARTICLE The polysynthesis parameter. By Mark C. Baker. New York and Oxford: Oxford University Press, 1996. Pp. xix, 556. Reviewed by Jean-Pierre Koenig and Karin Michelson, State University of New York at Buffalo.* The Polysynthesis Parameter is an investigation into the nature of polysynthesis within a formal principles-and-parameters framework. The languages defined as polysynthetic by the author (hereafter B) have certain distinctive properties, which he classifies into two broad categories : nonconfigurational phenomena and noun incorporation. Nonconfigurational properties include (after Hale 1983) relative freedom of word order, pervasive dropping of noun phrase arguments, and the existence of discontinuous constituents. Noun incorporation must be productive , and nouns must regularly occur independently as an unincorporated nominal. The focus is on Mohawk (Northern Iroquoian), based largely on B's own field research at Kahnawake (near Montreal), but data is cited from six other languages (Ainu, Chukchee, Mayali, Náhuatl, Southern Tiwa, and Wichita), whenever it is available in published sources or via personal communication. The goals of the book are ambitious: B aims to provide a comprehensive description of morphosyntactic phenomena of a single polysynthetic language (in Baker's sense), Mohawk, and of principles which he suggests are at the heart of the typological class of languages to which Mohawk belongs. B presents theoretical principles with characteristic ease and clarity. The attempt to extend theoretical hypotheses of a principles-and-parameters approach to a language outside of ones to which such a framework has been widely applied presents interesting challenges to the formal principles of the theory, to which B persistently seeks solutions. At the same time, B provides detailed and often subtle observations about a large class of constructions in Mohawk, and consistently cites published analyses that contradict his principles. Furthermore, B provides a goodly number of examples for each issue that he addresses. Does the book live up to its ambitious goals? We think the answer to such a question depends, to some extent, on one's perspective on a number of issues we review below. We provide a basic summary of the claims of each chapter and offer certain criticisms after each. General criticisms, as well as comments pertaining to more than one chapter, are addressed at the end of the review. Although our review ultimately raises more negative points than positive ones, we applaud the serious undertaking this book represents. The question B sets out to answer is What makes a polysynthetic language polysynthetic? The answer, he argues, is a macroparameter, a parameter general enough to affect the entire grammar of a language. The particular macroparameter that B posits is simple enough: whether arguments must be recorded via a morpheme on the predicator they depend on in order to be 'visible'. Mohawk and other polysynthetic languages are set positively for the parameter. The two ways to satisfy the parameters are via agreement morphology and via noun incorporation. B's hypothesis about the nature of polysynthesis is clearly innovative from a principles-andparameters approach (although it draws from Jelinek's earlier work). But, in some sense, there is nothing earthshattering in this claim, since it basically amounts to saying that Mohawk is a consistently head-marking language. Otherwise put, B's claim simply recasts the very traditional idea thatpolysynthesis characterizes languages in which words can be sentences, so that predicateargument relations, which are at the core of the structural make-up of sentences, are defined and satisfied within the word. What makes the book theoretically innovative is the attempt to derive so many properties of polysynthetic languages from this single fact. It is important to note that * The reviewers would like to thank Mark Baker, Holger Diessel, Matthew Dryer, and Marianne Mithun for comments on earlier drafts of this review. The order of reviewers' names is alphabetical since the effort was collaborative. 129 130 LANGUAGE, VOLUME 74, NUMBER 1 (1998) B's use of polysynthetic does not refer to an independently defined class of languages. In fact, for B polysynthesis is defined by the polysynthesis parameter. The question, then, is whether indeed the salient and distinctive properties of certain languages follow from a positive setting for this one parameter. In B's analysis of Mohawk and other...