Descrição sintático-semântica de construções com o verbo-suporte ter e nomes humanos de relação para o processamento automático de linguagem natural

Abstract

This research presents a syntactic-semantic description of constructs with the support verb ter „have" and human nouns of relation, in structures of the type N0 ter Npred, observed in different contexts of use in the Portuguese Language, for Example, Ana tem uma mãe depressiva (Ana has a depressive mother); A princesa Daiana tem uma filha secreta ( Princess Daiana has a secret daughter); Paula tem um marido agressor (Paula has an abusive husband). The analyzes are based on the theoretical-methodological framework of Lexicon-Grammar, postulated by the French linguist Maurice Gross (1975), which defines formal syntactic criteria for lexical description with computational purposes. The criteria are applied in phrases constructed according to the linguistic and extralinguistic knowledge about the language.The research has as objectives a) to analyze the structures with the verb to have; B) describe the properties with the verb-support ter and human noun of relation; C) encode the properties of the support verb construction; D) present a table that represents a linguistic resource for Automatic Natural Language Processing (NLP), with the encoded structures. The verb to have like other verbs of the language can present different behaviors, depending on the syntactic context in which it is inserted. The description and coding of these constructions are inserted in a database that can favor Computational Linguistics, mainly in applications such as the automatic translation of texts from Portuguese to other languages. Therefore, this research about description and formalization of structures with the support verb ter represents a contribution to PLN. The advancement of technologies has required more rigorous and complete descriptive studies of the Portuguese lexicon. The most detailed and complete description, as well as the formalization of these structures, can improve the quality of PLN software that requires the lexical description.