Antonio Zampolli

From Monolingual to Bilingual Automated Lexicons: Is There a Continuum?

Lexical Data Bases (LDB) are investigated as essential tools in the storing and processing of lexical data for their capability of providing direct and immediate access to sets of words sharing specific properties or features. LDBs are considered as multifunctional in many respects: with respect to the data, to the applications, and also to different linguistic theories. The Italian LDB and its future developments are seen as an example of implementation of a LDB according to the lines described. The aim of the project is to create a large repository of lexical data, where access is provided at the various levels of lexical units, properties, and relations. We are tending towards an automated dictionary which represents lexical information by relations from words to words, or from words to metalinguistic codes, using existing dictionaries as one of the sources for the raw data. One of the purposes of implementing these relations is to transform a particular natural language text, i.e. definitions, into a knowledge base, and to relate natural language words to an underlying and probably interlinguistic taxonomy or network of concepts. A suitable structure makes it possible to obtain wide access to lexical information both in breadth and in depth, and for a number of foreseeable applications. In the perspective of realizing a neutral syntactic component, we are investigating the possibility of representing in the lexicon the linguistic information used in parsers and generators in such a way that it can be reutilized in many different theoretical frameworks. A project is also described which uses bilingual machine readable dictionaries as a bridge connecting two otherwise independent monolingual lexical data bases. One of the objectives is to integrate the different types of information traditionally contained in monolingual and bilingual dictionaries, so as to expand the informational content of the single components in the new integrated bilingual system. In the new bilingual system, there would no longer be a source language and a target language, since the look-up and access procedures are independent and neutral with respect to direction. Bidirectional cross-references will also be automatically generated for the information contained at each sense level as semantic indicators (i.e. synonyms, hyperonyms) or contextual indicators.