Word Formation and the Validation of Lexical Resources

Pius ten Hacken (Universit.t Basel WWZ — Abt. Geisteswissenschaftliche Informatik Petersgraben 51, 4051 Basel, Switzerland)

WP2: Lexicons

In the framework of Word Manager (WM), morphological dictionaries are produced by the classification of lexemes in terms of a rule database. The intricate structure of the resulting lexical resources, conceived primarily for flexible use, also offers novel opportunities for the validation of the lexical specification. Many of the inconsistencies and errors encountered in lexical specification in a text file are excluded in WM, because the lexicographerÕs interface supports decisions by the exploitation of the procedural nature of inflection and word formation rules. There remains a set of lexicographic decisions, based on facts of the language and on the theoretical analysis of these facts, which cannot be supported in this formal way. They include the contents of the lexicographic guidelines. For the validation of these decisions, two types of browser are provided, the tree browser which gives access to partitionings of the set of lexemes, and the lexeme browser which concentrates on information for a single lexeme and on its links to other lexemes. The possibilities available because of the structure in the database constitute a challenge for the generality of the approach to validation described by Underwood & Navarretta (1997), which requires the reduction of lexical databases to text files.

Lexical resources

67.pdf