Constructive case: Evidence from Australian languages By Rachel Nordlinger (review)

BOOK NOTICES 213 the grammar (Chs. 1 and 2). Ch. 2 also contains an overview of evidentials in Quechua Wanka and their possible diachronic origins. The main semantic parameters of evidentiality in Quechua are the source of information and the 'degree of responsibility' taken by the speaker. Ch. 3 considers the connection between evidentials and deixis. The next three chapters are the 'core' of the book. The semantic description of evidentials is carried out in terms of prototypes and extensions; in each chapter the author presents the facts about evidentials in Quechua Wanka, the ways in which they are used in texts of different genres, and compares them with the data from other Quechua languages so that the reader gets a substantial idea of what evidentials are like across the Quechua family. Ch. 4 presents an analysis of the direct evidential. Its use goes together with 'direct responsibility' of the speaker—for instance, it is preferred when one is talking about one's own feelings and physical states (90; note that in quite a few other South American languages one cannot use firsthand evidentials for this purpose). It is also employed in describing one's own dreams and in referring to well-known facts of which the speaker is sure (103). Ch. 5 discusses the 'conjectural' evidential employed to describe events for which the speaker has no firsthand evidence and for possible or probable events. It is also used to refer to someone else's psychological and physical states (similar to 'subjective' predicates in Japanese) and to mark ironic statements . Ch. 6 considers the usage of the reported evidential . Besides marking 'secondhand' information, it is employed in folktales and riddles. A typologically unusual property of the reported evidential is its use as a marker of mirativity, that is, of new and unusual information. Ch. 7 presents general conclusions as to the epistemic nature of evidentiality. It contains interesting data as to existing correlations between the marking of evidentiality in different tenses and between the occunences of evidentials with different persons (246-47): over 53% of all the occurrences of the direct evidentials are with first person, and over 84% ofthe reported are with third person. The list of references is impressive. However, the lack of a subject index makes this book less user-friendly than it could be. The author—who has spent many years with Quechua speakers—collected a large amount of data and has an excellent knowledge of the language, which allows him to make insights into the pragmatic and semantic intricacies of evidentials. This book is a greatassetto any typologistworking on evidentiality, modality, and related verbal categories, in South American languages and elsewhere. [Alexandra Y. Aikhenvald, Australian National University/La Trobe University.] Constructive case: Evidence from Australian languages. By Rachel Nordlinger . Stanford, CA: CSLI Publications , 1998. Pp. 194. Constructive case is a slightly revised version of Rachel Nordlinger's Stanford University PhD thesis, an investigation into case marking and nonconfigurationality in Australian languages. N begins with an overview of traditional analyses of nonconfigurational languages such as Warlpiri and Mohawk, the criteria for the identification of a language as 'nonconfigurational ', and the problems with the usual treatment of nonconfigurational languages in a number of generative theories. Nonconfigurational languages are usually highly head-marking. Some analyses of nonconfigurationality within a generative framework have relied on the treatment of cross referenced noun phrases as adjuncts. N shows that for a number of Australian languages this analysis is unsatisfactory since tests of scrambling and omission show that these arguments do not behave in the same way as true adjuncts. N proceeds from the basis that in nonconfigurational languages, the case morphology is the (usually sole) indicator of grammatical function; one way of defining the distinction between configurational and nonconfigurational languages is thus that the former define grammatical relations in the syntax whereas in the latter languages other parts of the grammar, particularly the morphology, are responsible for the identification of grammatical relations. Much of the remainder of the book is thus concerned with the possibility ofmodelling the assignment ofgrammatical relations by morphology. N makes use of the theory of LFG to present her model of nonconfigurationality. Since in nonconfigurational languages one...