A INFLUÊNCIA DO SN-SUJEITO NO PROCESSAMENTO DE ORAÇÕES RELATIVAS AMBÍGUAS NO PORTUGUÊS BRASILEIRO

Abstract

The present work aims at analyzing the influence of the internal structure of the NPsubject in the interpretation of ambiguous relative clauses of Brazilian Portuguese. The objective of the research was to investigate the performance of parser related to the preference of attachment of this type of clause, searching a possible relation with the structure, under the syntactic and prosodic points of view, which defined the combination of variables to be observed: simple and short NP; simple and long; composed and short; and composed and long. In order to compare silent and oral reading, collection of data instruments had been elaborated. The silent reading experiment was applied to a total of one hundred and ten informers; in turn, the oral reading experiment was applied to seven informers, all university students and monolingual speakers of Brazilian Portuguese. For the elaboration of the instrument, we took as a reference the type of construction used in the forth experiment proposed by Cuettos and Mitchell (1988). For the accomplishment of this study, we were based on the Implicit Prosody Hipotesis (Fodor, 2002), Prosodic Theory (Nespor and Vogel, 1986), studies about prosody-syntax interface (Inkelas and Zed, 2000), as well as studies refered to the processing of sentences in languages such as English, Spanish and Brazilian Portuguese (Bader, 1998; Fernández, 2002; Finger and Zimmer, 2005; Lourenço-Gomes, 2003; Maia et al, 2004, entre outros). We analyzed, in silent reading, the attachment preferences; and, in oral reading, such preferences had been associated to acoustic analysis, in order to find relations between the division of the structure in phnological phrases and the related choices. The results of the research had disclosed, for this type of structure, a significant preference for the early closure of the relative clause in both silent and oral reading. The results had converged in the direction to confirm the hypothesis that the more 20 natural prosodic contour of the language would influence the resolution of this type of ambiguity. However, the variable studied here revealed itself as irrelevant, which sent us to possible explanations related to the type and the length of the involved structures, as well as the proper collection of data instrument configuration. The results obtained in this analysis had allowed us to infer that the parser would be influenced by proper characteristics of each language, contributing, thus, for the deepening of the studies on language processing.