Using Subcategorization to Resolve Verb Class Ambiguity
43391Authors
nan
Abstract
Levin's (1993) taxonomy of verbs and their classes is a widely used resource for lexical semantics. In her framework, some verbs, such as give exhibit no class ambiguity. But other verbs, such as write, can inhabit more than one class. In some of these am- biguous cases the appropriate class for a particular token of a verb is immediately obvious from inspec- tion of the surrounding context. In others it is not, and an application which wants to recover this infor- mation will be forced to rely on some more or less elaborate process of inference. We present a simple statistical model of verb class ambiguity and show how it can be used to carry out such inference. 1
Keywords
Full Content
Using Subcategorization to Resolve Verb Class Ambiguity Maria Lapata School of Cognitive Science Division of Informatics University of Edinburgh 2 Buccleuch Place Edinburgh EH8 9LW, UK mlap@cogsci.ed.ac.uk Chris Brew HCRC Language Technology Group Division of Informatics University of Edinburgh 2 Buccleuch Place Edinburgh EH8 9LW, UK chrisbr@cogsci.ed.ac.uk Abstract Levin's (1993) taxonomy of verbs and their classes is a widely used resource for lexical semantics. In her framework, some verbs, such as give exhibit no class ambiguity. But other verbs, such as write, can inhabit more than one class. In some of these am- biguous cases the appropriate class for a particular token of a verb is immediately obvious from inspec- tion of the surrounding context. In others it is not, and an application which wants to recover this infor- mation will be forced to rely on some more or less elaborate process of inference. We present a simple statistical model of verb class ambiguity and show how it can be used to carry out such inference. 1 Introduction The relation between the syntactic realization of a verb's arguments and its meaning has been exten- sively studied in Levin (1993). Levin's work re- lies on the hypothesis that "the behavior of a verb, particularly with respect to the expression and in- terpretation of its arguments, is to a large extent determined by its meaning" (Levin, 1993, p. 1). Verbs which display the same diathesis alterna- tions-alternations in the realization of their argu- ment structure-are assumed to share certain mean- ing components and are organized into a semanti- cally coherent class. As an example consider sentences (1)-(3) taken from Levin. Example (1) illustrates the causative/inchoative alternation. Verbs undergoing this alternation can be manifested either as transi- tive with a causative reading (cf. (la)) or as intransi- tive with an inchoative reading (cf. (lb)). Examples (2) and (3) illustrate the dative and benefactive al- ternations respectively. Verbs which license the for- mer alternate between the prepositional frame NP- V-NP-PPto (cf. (2a)) and the double object frame V-NP-NP (cf. (2b)), whereas verbs which undergo the latter alternate between the double object frame (cf. (3a)) and the prepositional frame NP-V-NP- PPJ~r (cf. (3b)). (1) a. Janet broke the cup. b. The cup broke. (2) a. Bill sold a car to Tom. b. Bill sold Tom a car. (3) a. Martha carved the baby a toy. b. Martha carved a toy for the baby. Verbs like crack and chip pattern with break in li- censing the causative/inchoative alternation and are associated with the semantic class of BREAK verbs. Verbs make and build behave similar to carve in licensing the benefactive alternation and are mem- bers of the class of BUILD verbs, whereas sell and give undergo the dative alternation and participate in the GIVE class. By grouping together verbs which pattern together with respect to diathesis alterna- tions Levin defines approximately 200 verb classes, which she argues reflect important semantic regu- larities. 2 Motivation Levin provides an index of 3,024 verbs for which she lists the …
Content truncated to 500 words
Labeled Text
Levin's (1993) taxonomy of verbs and their classes is a widely used resource for lexical semantics. In her framework, some verbs, such as give exhibit no class ambiguity. But other verbs, such as write, can inhabit more than one class. In some of these am- biguous cases the appropriate class for a particular token of a verb is immediately obvious from inspec- tion of the surrounding context. In others it is not, and an application which wants to recover this infor- mation will be forced to rely on some more or less elaborate process of inference. We present a simple statistical model of verb class ambiguity and show how it can be used to carry out such inference. 1
Publication Details
- Paper ID:
- 43391
- Year:
- 1999
- Publication:
- TIPSTER
- Publication Type:
- conf
- DOI:
- nan