Studies in language origins, vol. 1 Edited by Jan Wind, Edwin G. Pulleyblank, Eric de Grolier, and Bernard Bichakjian (review)

BOOK NOTICES 425 strained unification formalisms in order to determine 'their suitability (not just their adequacy) for describing natural language structures ' (40). The discussion section is long and interesting, with contributions from Fernando Pereira, Mitch Marcus, and Martin Kay in addition to the authors of the position papers. Marcus presented a thoughtful critique of unificationist approaches, arguing that these are too general to be likely to produce interesting results. His remarks were replied to by, among others, Gazdar , who distinguished two approaches: the development of PATR grammars as grammarwriting tools, and linguistic theories expressed in the PATR formalism, such as GPSG, LFG, and HPSG. Despite the undertone of frustration, this is an exciting volume. Work on syntax in computational linguistics has been rejuvenated, work on semantics is progressing on all sorts of fronts, connectionism is continuing to attract interest, and in general the level of sophistication and extent of coverage of NLP work have increased greatly. [Chet A. Creider, University of Western Ontario.] Studies in language origins, vol. 1. Ed. by Jan Wind, Edwin G. Pulleyblank, Eric de Grolier, and Bernard Bichakjian. Amsterdam & Philadelphia: Benjamins, 1989. Pp. xxii, 331. This volume is a selection of papers from the 1985 Language Origins Society meeting at Oxford , the 1986 meeting at Vanderbilt, and other papers sent in to the editors. The intended audience is 'the general scholar' (back cover). The collection is good but not very exciting. One finds a tendency to repeat oneself and deal with peripheral matters in this field, which is both too narrow and too wide. We cannot completely discount the reasons for the Pans Society's decision to impose the famous ban on languageorigin papers in 1866. The present papers are 16 in number, not including Bichakjian's Introduction and comments on each contribution in the book and the Contents and detailed Notes on Contributors. The papers may be divided into five categories : general, gesture, 'phenomenon X and language (L)', primarily linguistic, and biological. Unfortunately. Jeffrey Laitman's important Vanderbilt paper on the flexion of the basicranium as evidence for the changing hominid vocal tract is not included. In limited space, I can comment on only a few papers. I find the large category of 'X and L' to be not very successful. Fire and supernaturalism (discussed by Johan Goudsblom and Gordon Hewes, respectively) are undoubtedly tied up in feedback relations with the development of language in the crucial periods of about 300,000 B.P. (significant biological changes in hominids) and 35,000 B.P. (cultural expansion), but they are more part of the effect than the cause. Of the four biological contributions, two are in the technical area of brain anatomy and largely beyond the grasp of the intended audience . In the most accessible of these papers, Jan Wind gives a thorough survey of the evolution of the human speech organs beginning three billion years ago ( !). with a very brief summary of the work of Lieberman and Laitman (187). It seems to me that he goes too far in emphasizing the priority of the nervous system to the 'peripheral system' of vocal language. For example, he concludes that the only advantage of the human vocal tract is in encoding/decoding speed. This is probably as true and as trivial as saying that we can do all our communication with the simplest code if we are willing to do it fantastically slowly; it is like doing all mathematics with a Turing machine. Bernard Bichakjian raises some serious problems for Chomskyan innateness (parameters exist which need to be set by exposure to given languages) and Bickertonian innateness (parameters are already set). I cannot believe that anyone really believes that phrase-structure rules are innate, so this is a straw man (226). and as for the chasm between nominative and ergative languages (218-19), I think this can be bridged by a sufficiently abstract grammar. The appropriate level of abstractness for Universal Grammar (UG) may be that of such conditions as 'c-command', not that of 'island constraints', and certainly not PS-Rules. Are not Bichakjian 's idea of a human potential for 'grammatism' and an 'evolving innate grammar' compatible with both Chomsky and Bickerton? Robin Allot proposes...