A Lexical Database for the Analysis of Portuguese MWEs

Abstract

Abstract. We present an Access database that will help to analyze and classify Portuguese MWEs according to their linguistic properties. We used a subset of an existing lexicon of Portuguese MWEs that was extracted from a corpus using Mutual Information as a statistical measure, followed by manual validation. The lexicon is organized in a three-level structure: the main lemmas, the group lemmas and the variants of those groups. This subset was imported to the database together with linguistic information about the MWEs and their variants (morphosyntactic structure, syntactic category, frequency, MI, grammatical function, discursive function, etc.). A semantic and syntactic fine-grained typology was established, and, by selecting a particular MWE, we can classify it with respect to its degree of semantic decomposability and syntactic transformation. The database is highly customizable and enables the addition/deletion of semantic or syntactic categories considered important throughout the analysis. In the end, all the information of the database will be exported to a XML format, resulting in a lexicon enriched with linguistic information.