In lieu of an abstract, here is a brief excerpt of the content:

STRUCTURE AND INTONATION Mark Steedman University of Pennsylvania Rules for assigning phrasal intonation to sentences are often assumed to require an autonomous level of 'intonational structure', distinct from what is usually thought of as surface syntactic structure. The present paper argues that the requisite notion of structure can be subsumed under the generalized notion of surface structure that emerges from the combinatory extension of Categorial Grammar. According to this theory, the syntactic structures and the intonational structures of English are one, and can be captured in a single unified grammar. The interpretations that the grammar provides for such constituents correspond to the entities and open propositions that are concerned in certain discourse-related aspects of intonational meaning that have variously been described as 'theme' and 'rheme'. 'given' and 'new', and 'presupposition' and 'focus'.* 1. The problem. It is well known that phrasal intonation organizes the perceived grouping of words in spoken utterances in ways which are, on occasion, inconsistent with traditional linguistic notions of syntactic constituency. For example, consider the following exchange: (1) a. I know that Alice likes velvet. But what does Mary prefer? b. (Mary prefers) (corduroy). One normal prosody for the answer (b) to the question (a) consists in not only marking the new information in the answer by the use of high pitch on the stressed first syllable of the word corduroy, but also in stressing the first syllable of Mary, using a high pitch-accent, and placing a final rise at the end ??prefers, with lower pitch interpolated in between. This intonation contour, which conveys the contrast between the previous topic concerning Alice and the new one concerning Mary, imposes the perceptual grouping indicated by the brackets (stress is indicated by small capitals).1 Such a grouping cuts across the traditional syntactic analysis of the sentence as a subject and a predicate VP. The presence of two apparently uncoupled levels of structure in natural language grammar appears to complicate the path from speech to interpretation unreasonably. Such a theory seems likely to be very difficult to apply in the form of computer programs for automatic speech synthesis or recognition. Despite its apparent independence from syntax, it is widely accepted that intonational structure is, nonetheless, strongly constrained by meaning, and in particular by distinctions of focus, information, and propositional attitude to- * 1 am grateful to Steven Bird, Dwight Bolinger, Elisabet Engdahl, Ellen Hays, Julia Hirschberg. Jack Hoeksema, Stephen Isard, Aravind Joshi, Ewan Klein, Bob Ladd, Mark Liberman, Mitch Marcus, Dick Oehrle, Donna Jo Napoli, Michael Niv, Janet Pierrehumbert, Henry Thompson, Bonnie Lynn Webber, and three anonymous referees for comments, advice, and moral support at various stages. The research was supported in part by NSF grant CISE IIP:CDA 88-22719, DARPA grant no. N00014-90-J-1863, and ARO grant no. DAAL03-89-C0031 to CIS, University of Pennsylvania . 1 The intuition of structure imposed by intonation is very compelling. A common initial problem in teaching formal syntax is to persuade students that this is not the notion of structure to which they are to attend. 260 STRUCTURE AND INTONATION261 wards concepts and entities in the discourse model. For example, the intonation contour in the above example seems to divide the utterance into what Prince 1986, following Wilson & Sperber 1979, calls an 'open proposition', and its complement corduroy. (A similar partition is embodied in Cresswell's [1973] and von Stechow's [1989] notion of a 'structured proposition'.) It will be convenient to refer to such partitions of the information in the proposition as the 'information structure' of an utterance. An open propositions are most simply exemplified as that which is introduced into the discourse context by a wH-question. Such an entity can be thought of as a proposition with a 'hole', which the discourse must 'fill in'. So for example the question in 1, What does Mary prefer?, introduces an open proposition which we might informally write as follows: (2)Mary prefers ... More formally, it is natural to think of open propositions as functional abstractions , as Jackendoff 1972 and Sag 1976 pointed out, and to write them using the notation of the ?-calculus, in which the place of '...' is taken by a variable, whose scope...

pdf

Share