Análise comparativa de construções com verbos leves nas línguas portuguesa e inglesa: um estudo baseado no Léxico Gerativo (LG)

Abstract

The proposal of the Generative Lexicon (GL) coined by Pustejovsky (1991a, 1995) is presented as one of the most modern component approaches for lexical-semantic studies. The premise of this theoretical model is the study of aspects which have already been assumed as problematic by Lexical Semantics theories, such as the polymorphic nature of language. In this sense, this doctoral research aims to analyze the semantic behavior of some light verb constructions (LVC) seen as prototypical in Portuguese: dar, fazer, ter and tomar. As it is a comparative study, some constructions with the counterparts of these verbs in the English language (to give, to do/to make, to have and to take) are also analyzed, following the same protocol. Hence, the LVC are discussed under the perspective of the GL, considering, as the theoretical apparatus, the studies by Pustejovsky (1991a, 1991b, 1995, 2013, 2014a, 2014b), building a connection with the studies about light verbs (LV), based on Jespersen (1954), Kearns (1988), Brugman (2001), Butt (2003), Perini (2017), among others. Assuming that it would be very hard for Lexical Semantics (and many linguists) to agree with the assumption made by traditional grammars and language-in-use grammars (Portuguese and English), that there is a semantic void of these verbs when they function as light verbs (LV), we seek to answer the following methodological questions: (a) are LVC simply combinations of “verb + complement” kind, as the hegemony of the studies on this verbal class states, or is there a change in the meaning of these verbs due to this complement? and (b) if there are registers of emptiness of meaning, is there any Lexical Semantics theory able to explain the behavior of LVC? From the search of data in corpora of language in use, nine constructions with the mentioned verbs, which are representative of what is intended to be analyzed and verified, are defined as a corpus of analysis. Based on the hypothesis that not all light verbs would be meaningless, we aim, through data analysis, to show that a new meaning could be attributed to these verbs, considering that each of them could be on a different spectrum of the “lightness gradation”. The results show that the proposed model for the analyzes is really positive and make it possible for us to answer the research questions. We conclude that there are restrictions when applying this definition of semantic void to LV, since each verb (when compared to each other) may present a different ‘lightness gradation’, depending on the internal argument presented. This said, we prove that, in both languages, we can find some “lighter” verbs than others, that is, there are verbs which present a more accentuated semantic bleaching than others, even if it means to contradict the Lexical Semantic studies. Therefore, we suggest that it should be added observations in the lexical entries of dictionaries in order to draw the attention of the users of each language about the semantic behavior of these verbs.