The comparative method reviewed: Regularity and irregularity in language change Ed. by Mark Durie and Malcolm Ross (review)

BOOK NOTICES 893 movement of the volume from speech perception to language production. Although the editors acknowledge that important aspects of research m Spanish psycholinguistics have been omitted for lack of space, e.g. language acquisition and bilinguahsm, the book documents convincingly the impressive accomplishments in this field. [Glenn Frankenfield, University ofMaine at Farmington.] The comparative method reviewed: Regularity and irregularity in language change. Ed. by Mark Durie and Malcolm Ross. New York & Oxford: Oxford University Press, 1996. Pp. vi, 321. $75.00. This volume brings together ten chapters on the controversial subject of what is regular and what is irregular in language change and is concerned with the limitations and potential of the comparative method (CM). To this end, many of the chapters include applications of CM under challenging circumstances . Two are concerned with morphology and semantics, respectively, areas usually not considered to be within the purview of CM. In the first part of their erudite and provocative 'Introduction', Ross and Durie blend a summary of the contents of the volume with a discussion of the steps of CM and include a valuable discussion of what CM is not. Multilateral comparison, in a fundamental way, is not part of CM. The second part then deals with the ongoing paradigm shift in the way comparativists work—away from the quasi-Darwinian neogrammarian hypothesis which has heretofore underlain our conception of regularity in language change towards a speaker-oriented perspective which questions the difference between so-called regular change and change diffused lexically. The latter may introduce apparent irregularities owing to speakeroriented factors having to do with speech organs, communicative exigency, contact, etc. Johanna Nichols, 'The comparative method as heuristic' , demonstrates that paradigmacity is crucial in establishing linguistic relatedness and that two unrelated languages are unlikely to have more than a miniscule percentage of nonmarginal morphemes in common which match in form and meaning. Lexical comparison, by itself, is diagnostic of nothing. Lyle Campbell, 'On sound change and challenges to irregularity ', focusses upon linguistic and sociocultural factors such as sound symbolism, onomatopoeia, avoidance ofhomophony, etc., which may interfere with regularity but do not undermine the principle of regularity itself. Mark Durie, 'Early Germanic umlaut and variable rules', argues that some regular sound changes are not categorical, but probabilistic. George Grace, 'Regularity of change in what?', questions some usual assumptions about CM and concludes that its successful use depends upon congruence of linguistic communities and linguistically defined units. Malcolm Ross, 'Contactinduced change and the comparative method. Cases from Papua New Guinea', examines data involving typological change which is usually not treated within CM. Harold Koch, 'Reconstruction in morphology ', makes a distinction between morphological change and sound change in morphemes and presents a valuable typology of morphological change. David Wilkins, 'Natural tendencies of semantic change and the search for cognates', presents methodologies to assess semantic similarity and argues that there are natural tendencies of semantic change involving core vocabulary. This provocative volume has much to teach everyone working in comparative-historical linguistics. [Joseph F. Eska, Virginia Polytechnic Institute & State University.] The verbal system of Classical Hebrew in the Joseph story: An approach from discourse analysis. By Yoshinobu Endo. (Studia semítica neerlandica, 32.) Assen: Van Gorcum, 1996. Pp. xiv, 351. Classical or biblical Hebrew is perhaps the language that has received the largest amount of scholarly attention relative to the modest size of its attestation—the books known as the Hebrew Bible or Old Testament. Still no consensus has been reached with respect to its verbal system. The problem is whether the main inflectional categories are marked for tense or for aspect. In either case, there are both simple and compound forms with seemingly identical markings. Thus, e.g. (simple) qatal and (compound) wa-yiqtol may both be said to refer to the past, but it is not always clear what factors determine the choice for one form or the other. In the published version of his 1993 Bristol doctoral dissertation, Endo tackles the problem anew. His theoretical framework is discourse analysis, a school of thought which prefers not to regard the sentence as the highest level of grammatical description . Instead, it claims that sentences should always be seen in relation to each other...