Usage-Based Models of Language (review)

Abstract

Reviewed by: Usage-based models of language ed. by Michael Barlow, Suzanne Kemmer Jean-Christophe Verstraete Usage-based models of language. Ed. by Michael Barlow and Suzanne Kemmer. Stanford: CSLI Publications, 2000. Pp. xxviii, 356. ISBN 1575862204. $22.95. Usage-based models of language provides an overview of recent work in linguistics that takes usage data seriously in modeling and theory formation, regarding language use not just as an instantiation of more abstract structures, but as a driving force that shapes and changes these structures. The editors’ introduction to the volume is a useful discussion of [End Page 875] some basic theoretical and methodological issues in usage-based approaches to language, such as the importance of patterns of frequency and variation in language change, the role of cognitive mechanisms as functional constraints on language structure, and the importance of contextual and cultural factors. The other papers in the volume elaborate specific aspects of usage-based approaches, ranging from highly programmatic papers that discuss what a linguistic model should look like if it takes usage data seriously, over psycholinguistic and computational experiments relating to the cognitive aspects of a usage-based approach, to data-intensive papers that show what it actually means to implement such an approach in linguistic description. The papers by Ronald Langacker and Sydney Lamb are the most programmatic ones in the volume: Langacker discusses why and how his model of ‘cognitive grammar’ is usage-based, and Lamb examines the theoretical and methodological prerequisites for network models of language. The articles by Brian MacWhinney and by Connie Dickinson and Talmy Givón both report on experiments that investigate the neural and cognitive aspects of usage-based models. MacWhinney provides an interesting overview of his work on connectionist modeling of language learning and acquisition, and Dickinson and Givón investigate the relationship between conversational interaction and the storage of information in episodic memory. The five other papers illustrate the implementation of a usage-based approach with data from different linguistic domains. Joan Bybee looks at phonetic variation relating to t/d deletion in Chicano English, showing that it is not distributed evenly over the lexicon and that it varies with the frequency of lexical items, and investigating the implications of these observations for phonological theory. Mira Ariel contrasts NP detachment and accessibility theory as two alternative explanations for the development of agreement markers from pronouns, and shows how synchronic patterns of usage in a corpus of Modern Hebrew can be used as evidence for the explanatory adequacy of accessibility theory. Arie Verhagen investigates the diachrony of causative constructions in Dutch, and argues that changes in the relative frequency of the two basic causative verbs laten and doen can be related to changes in cultural conceptions of authority. Douglas Biber illustrates how corpora can be used to study patterns of association between various linguistic features on different levels, and between linguistic and nonlinguistic features. Michael Barlow, finally, studies grammatical variation in collocations on the basis of corpora and provides an explanation in terms of blending theory. Taken together, the articles in this volume form a welcome introduction to usage-based models of language, with convincing arguments for the importance of taking usage data seriously, and interesting implementations of usage-based approaches on different linguistic levels. Jean-Christophe Verstraete University of Leuven Copyright © 2004 Linguistic Society of America