Preferências sintáticas e semânticas no processamento da correferência anafórica: evidências de movimentação ocular

Abstract

In this dissertation, our main objective is to contribute for the development and understanding of psycholinguistics studies that attempt to experimentally demonstrate relevant theoretical conjectures about anaphoric processing. Under the conceptual frameworks of the Theory of Accessibility (ARIEL, 1991, 2001), the Theory of Centering (GROSZ; JOSHI; WEINSTEIN, 1995), the studies on the typicality of the antecedent term (GARROD; SANFORD, 1977; VAN GOMPEL; LIVERSEDGE; PEARSON, 2004), the Informational Load Hypothesis (ALMOR, 1999), and the Position of Antecedent Hypothesis (CARMINATI, 2002), we propose that the prominence of the syntactic position in complex sentences plays a major role on the anaphoric resolution in the Portuguese language. Adopting a psycholinguistic methodology based on on-line (tracking of eye movements) as well as off-line observations, we searched for evidence to support our hypothesis from the results of the following set of studies: (i) an experiment to evaluate the comprehension of complex sentences due to coordination, in which both the position of the antecedent and the type of semantic relationship between antecedent and anaphora are manipulated; (ii) an experiment to evaluate the comprehension of complex sentences due to subordination, in which both the type of anaphoric coreference, in the form of a plain or null pronoun, and the position of the coreference, anaphoric or cataphoric, are manipulated; (iii) an experiment for generation of complex sentences, using plain or null pronouns as coreferentials; and (iv) a reading experiment of non-manipulated texts to establish a comparative standard for reading flux in Brazilian Portuguese. Our on-line experiments were performed with an eye-tracker of 120 Hz, which allowed eye movements to be recorded at each 8 milliseconds. The following dependent variables related with the eye movement have been analyzed: (i) the number of fixations; (ii) the duration time of the first fixation; (iii) the average duration of the fixations; and (iv) the total time of fixation. The overall analysis of our results, based on the investigation of complex sentences, suggests that the resolution of the coreferential anaphora is a function of the prominence of the subject position. Moreover, the information load of anaphoric expressions with semantic content seems to increase the cost of the anaphoric processing of a highly accessible antecedent.