Pergamon Lqquoxe Sriences. Vol. 15, No. 4, PP. 357-376. 1993 Copyright 0 1994 Elsevier Science Ltd Printed in Great Britain. All rights reserved 0388-OilOl/93 $6.00+0.00 LANGUAGE UNIVERSALS, DISCOURSE PRAGMATICS, AND SEMANTICS PAUL J. HOPPER and SANDRA A. THOMPSON Within ‘functional linguistics’, semantic explanations have often been offered for cross-linguistic gram- matical generalizations. These explanations have been based on such semantic properties as animacy, volitionality, referentiality, and Fillmorean case roles. Dixon has expressed the relationship by proposing that ‘grammar is frozen semantics’. Intriguingly, examination of a range of cross-linguistic generalizations leads increasingly to the view that grammar is primarily shaped by the entire range of cognitive, social, and interactional factors involved in the actual use of language. In this paper we discuss a number of specific grammatical phenomena which support the view that grammatical regularities arise because of certain strategies people habitually use in negotiating what they have to say with their hearers, in terms of what the hearer is likely to know or be able to identify, what needs to be highlighted or presented as newsworthy, what makes a good story, and so forth. This process is known as ‘grammaticization’. and suggests that grammar is best thought of, not as ‘frozen semantics’, but as something more like ‘sedimented conversational practices’. Introduction Within ‘functional linguistics’ two general types of explanations have been offered for cross-linguistic grammatical generalizations. Many explanations have been based on such semantic properties as animacy, volitionality, referentiality and case roles. Explanations from discourse have concentrated essentially on such discourse factors as information flow, i.e. the way in which speakers package information according to their assessment of what their hearers know, and discourse manipulability of refer- ents, foregrounding vs backgrounding, assumed degree of challengeability, mechanisms of the turn-taking system, and speaker stance toward propositional content. The examination of a range of cross-linguistic generalizations has led us increas- ingly to the view that grammar is primarily shaped by the entire range of cognitive, social, and communicative factors involved in the actual use of language. A number of researchers have suggested that what grammars code are semantic properties (Dixon 1982, 1984, 1989; Keenan 1984; Wierzbicka 1978, 1981, 1986, 1988; Langacker 1987). Dixon (1984) has expressed the relationship by proposing that ‘grammar is frozen semantics’. We are sympathetic with the assumptions underlying this approach; grammar cannot be autonomous, and a natural direction in which to seek motivation for grammatical regularities is in the area of meaning. In this paper, however, we seek to motivate an integrated approach to the explanation of linguistic Correspondence relating to this paper should be addressed to: Dr S. A. Thompson, Department of Linguistics, University of California at Santa Barbara, Santa Barbara, CA 93106, U.S.A. or Professor P. J. Hopper, Department of English, Carnegie Mellon University, Pittsburgh, PA 15213, U.S.A. 357