The origins of grammar: Evidence from early language comprehension by Kathy Hirsh-Pasek and Roberta Michnick Golinkoff (review)

BOOK NOTICES 679 elsewhere. The focus throughout is on reconstruction of phonological and morphological systems rather than on individual roots or inflections. InCh. 1, 'Relationship and reconstruction' (1-18), realism in reconstruction is argued to include consistency with implicational universals, following the uniformitarian assumption that languages have not changed in kind within the time span accessible to reconstruction. This section includes a brief critique of long-range comparison, although no Nostratic reconstructions are considered in the book. Ch. 2, 'Principles of typology' (19-31), reviews the relevant typological background from a diachronic perspective. Ch. 3, 'Typology and reconstruction ' (32-43), addresses the relative value of absolute and statistical universals. Ch. 4, 'Indo-European phonology' (44-64) discusses the PlE vowel system and the laryngeal and glottalic theories, the typological support for which is clearly articulated. Jakobson's critique of traditional reconstructions of the ?G? consonant series may be seen as having launched the field under discussion, and with the development of the glottalic theory, this case remains the clearestapplication oftypology to reconstruction. This seems likely to continue with Nostratic/Eurasiatic , as similar arguments have been used by Allan Bomhard and others in evaluating alternative visions of Nostratic. Ch. 5 on morphology (65-73) discusses the applications of theories of grammaticalization and markedness to the murkier ground of PIE grammar; in Chs. 6 and 7 the approach is applied to nominal and verbal morphology respectively. The focus of Ch. 6 (74-88) is the case system, applying markedness and the case hierarchy to such topics as syncretism in IE case paradigms and the alignment of the PIE case system. Reconstructions of the IE nominative in -s as a former ergative marker and the -m accusative inflection as marking the object of an antipassive construction are found plausible but inconclusive , while the hypothesis of PIE as a stativeactive language appears more speculative. Ch. 7 (89-109) reconsiders the tense-aspect-mood and voice systems, connected as these are by links such as that between passive and stative perfect. The Hittite evidence suggests an alternative proto-system rather simpler than the Sanskrit/Greek-inspired model of earlier reconstructions. Not surprisingly, the discussion of grammatical features proves much less conclusive than that of phonological ones. One reason given is that application of typology in this domain is largely probabilistic , and the statistical strength of many (especially diachronic) universals of grammar has yet to be established . Such problems should not, however, deter historical linguists from pursuing S's line of argument further with respect to ?G? or other proto-languages . [Stephen Matthews, University of Hong Kong.] The origins of grammar: Evidence from early language comprehension. By Kathy Hirsh-Pasek and Roberta Michnick Golinkoff. Cambridge, MA: MIT Press, 1996. Pp. x, 230. The authors have packaged their research material with a comprehensive introductory overview and speculative conclusion. The inclusion of the nonresearch material may provide students with an understanding of the intellectual context in which research is conducted, and allows the book to reach a larger audience of non-linguistically-trained professionals in related fields of child development and psychology . Ch. 1 (1-10) provides a broad overview of the common questions and assumptions about first language acquisition. Ch. 2(11-52) reviews current language acquisition theories. Ch. 3 (53-72) briefly reviews problems in the methodology of previous studies and introduces the procedures which Hirsh-Pasek and Golinkoff use to alleviate the problems discussed. They address the widespread intuition that comprehension precedes production and the likelihood that experiments which claim to prove otherwise have some procedural problems , such as requiring the child to produce language in order to demonstrate comprehension. Some researchers have tried to solve this difficulty by asking children to perform a nonlinguistic action, but this introduces the problem ofdistinguishing noncomprehension from noncompliance. Drawing on research which shows that children are more likely to look at a scene which matches a given linguistic stimulus, H & M use an ingenious arrangement of videotapes and audiotapes which requires only the child's eye movement to signal comprehension, thus allowing them to investigate the comprehension of children who may not yet be speaking at all. The next three chapters present the details of their experiments. Ch. 4 (73...