The Null Subject Parameter Edited by Osvaldo Jaeggli and Kenneth J. Safir. (review)

Abstract

REVIEWS367 Kiparsky, Paul. 1968. How abstract is phonology? Bloomington: Indiana University Linguistics Club. Sagey, Elisabeth. 1986. The representation of features and relations in non-linear phonology . Cambridge, MA: MIT dissertation. Department of Linguistics[Received 5 September 1990.] 204 Cunz Hall Ohio State University Columbus, OH 43210 The Null Subject Parameter. Edited by Osvaldo Jaegglit and Kenneth J. Safir. (Studies in natural language and linguistic theory, 15.) Dordrecht: Kluwer, 1989. Pp. 320. Cloth $89.00. Reviewed by Raffaella Zanuttini, University of Pennsylvania The subject-predicate relation has long been assumed by grammarians to be a basic structural feature of all languages. From the point of view of comparative syntax, however, it is a striking fact that in some languages every sentence must have a subject, while in others it is systematically possible for subjects not to occur. Moreover, this dichotomy is not the result of genetic developments , as closely-related languages differ on this point (e.g. Bani-Hassan Arabic and Levantine Arabic), while members of clearly unrelated families (e.g. Italian and Chinese) behave alike with respect to the obligatoriness of subjects. Given the linguistic importance of subjects, their possible absence is interesting on many levels, and one might expect a variety of perspectives to be relevant, perhaps including pragmatics, discourse processing or parsing concerns , phonological deletion, and even free variation. The present volume represents a body of work in the syntactic framework of Government and Binding (GB) theory which convincingly employs a strictly syntactic approach to determine what sort of grammar allows, or does not allow, null subjects. The problem of licensing null subjects within a Chomskyan framework is at least as old as GB itself, and is still current in the more recent parametric approach; both theories have developed and changed considerably since the early 1980s. The Null Subject Parameter (NSP) is the canonical example of a parameter, and its development—as presented in this book—reflects to a large extent the development of both parametric theory and GB. The original idea of parameters was simple, strong and attractive: (i) the apparent complexity of syntactic variation across languages can be reduced to a necessarily limited set of universal parameters, each of which has (at least) two options, or settings (e.g. 'subjects obligatory' versus 'subjects optional'), (ii) The consequences of a particular setting are a cluster of related properties; it is thus no accident, according to parametric theory, that Italian allows both null subjects and postverbal subjects, while English allows neither, (iii) The simplicity and universality of the parameters facilitates learning: positive data from a single 368LANGUAGE. VOLUME 67. NUMBER 2 (1991) construction trigger a parameter setting; the choice of alternatives in related constructions (i.e. those relevant to the parameter's cluster of properties) then follows automatically. (For further discussion of parametric theory, see Jaeggli & Safir's introductory discussion.) Because it distinguishes minimal differences between grammars and proceeds by identifying properties that correlate both within and across languages, parametric theory has encouraged a comparative approach to syntax, and has begun to bring together dialectologists and theoretical syntacticians. The contributions collected in this volume reflect this positive change, considering seriously a wide range of data from different dialects, languages, and language families, among them Trentino, Walloon, Chamorro, Chinese, Germanic, and Semitic. Most solutions to the problem rely on an old intuition correlating the richness of inflectional morphology (primarily subject-verb agreement) with the possibility of null subjects, which appears to be confirmed—and, in fact, made more precise—for a number of languages. It is clear, though, that agreement inflection cannot be the only factor that accounts for the possibility of null subjects across languages; this possibility is variously sensitive to the thematic properties of the subject, the form of the verb, and the type of clause, among other factors. The majority of contributors to this volume implement some version of the agreement solution, but the partial nature of this solution is also demonstrated , and most authors appeal to other factors as well. The editors themselves offer a comprehensive and ambitious synthesis which attempts both to unite the various analyses and to reconcile the data. While one may legitimately conclude that there is no single straightforward account for the...