next up previous contents
Next: Standardisation by consensus: Requirements Up: The content of the Previous: Levels of description covered

Formats - Representation

Work on lexical description typically deals with two aspects of the lexicon:

  1. The linguistic description of lexical items (`content');
  2. The (formal or formalisable) representation of lexical descriptions (`container').

As far as standardisation proposals are concerned, the CLWG has concentrated on the first issue, leaving the work towards proposals for the second to the Computational Linguistic Formalisms Working Group. This does however not mean that the CLWG disregards the evident need for a formal expression of the contents-oriented proposals put forward: on the contrary, the ELM documents series (ELM stands for Eagles/Lexicon/Morphosyntax Subgroup), as well as the subcategorisation report (EAGLES, 1996e) have been produced to satisfy the requirement of creating formal specifications which can be used in an implementation.

The work of the CLWG uses different representation formalisms to express its proposals, at different levels. In the morphosyntax area, proposals have been represented in different formats, ranging from a tabular presentation of the first proposals in EAGLES (1996g) over a set of hierarchical tables inspired by a typed feature logic representation (in the ELM reports series) to a specification implemented in Prolog (in the tagset mapping exercise documented in EAGLES (1996h). The results of the work on syntactic subcategorisation are represented in a feature structure format inspired by HPSG, and a Document Type Definition (DTD) has been constructed which can accommodate the syntactic lexicon entries proposed. The different parts of EAGLES (1996e) distinguish the different representations.