The University of Sussex

The representation of lexical semantic information

Ann Copestake

This thesis is an investigation of the representation of lexical semantic information from a computational linguistic perspective. An implemented representation language is described which is not specific to lexical semantics, but is based on the use of typed feature structures augmented with default operations. This language, which is formally specified, allows the lexical semantic representations to be tightly integrated with the syntactic component of the lexical sign, capturing generalisations by use of inheritance, while allowing for exceptions with the default mechanism. Default inheritance and default unification are discussed in detail. Grammar rules and lexical rules can be specified in the same formalism and thus the paradigmatic treatment of lexical semantics can be integrated with an account at the syntagmatic level. The use of the language is illustrated with some examples of the representation of verbs, the treatment of logical metonymy and of sense extension. This is followed by a more detailed account of individuation in nominals. The representation language is designed to allow the lexicon to be highly structured, with relationships between entries expressed in terms of (default) inheritance and lexical rules. The thesis discusses the way in which lexical entries structured by inheritance hierarchies may be derived semi-automatically by partial analysis of definitions in machine readable versions of conventional dictionaries (MRDs). The use of MRDs illustrates the practicality of constructing lexicons with detailed semantic information for use in natural language processing and provides a set of real examples with which to demonstrate the applicability of the representations described.

This paper is not available online