Polissemia e homonímia: uma avaliação da correlação entre acesso lexical e contexto

Abstract

The present PhD dissertation presents an analysis of eye motion behavior during readings of polysemic and homonymous words, using eye-tracking. The design of such analysis is composed by two experiments: in the first one, polysemic and homonymous words were preceded by contexts that either restricted interpretation to one specific sense or were neutral; the second one brought a tendentious context in which we incorporated a polysemic or homonymous word or a word of the same semantic domain, but one that violated the expectations created by the preceding context. Authors such as Swinney (1979) and Tanenhaus et. al. (1979) defend the position known as Multiple Access Hypothesis according to which context affects lexical access of ambiguous words only tardily. Tabossi e Zardon (1993) and Simpson (1994), on the other hand, defend the Selective Access Hypothesis, which states that only the contextually specified sense of an ambiguous word is accessed during reading. The results of this experiment indicated that polysemic and homonymous words are read more rapidly in contexts that tend to subordinate or dominant senses than in neutral contexts. More, in some of the tested conditions, reading time in cases that involved dominant senses was shorter than when subordinate senses were involved. As for the second experiment, linguists like Klein and Murphy (2001, 2002) oppose the idea that there is some difference in reading time between homonymous and polysemic words. On the other hand, Klepousniotou and Baum (2007) or Beretta, Fiorentino and Poeppel (2005) defend the advantage of the related senses effect or polysemy effect. According to advocates of such view, the reading time of polysemic words is shorter than the reading time of homonymous words in constrained contexts, what would suggest that the way these words are organized in our mental lexicon is fairly distinct, with a temporal advantage for words whose senses are shared by a semantic core. Our results corroborate this second theoretical position, since they show that polysemic words are read more rapidly than homonymous words in contexts that are strongly restrictive to one of the senses of such words. The findings of the first experiment suggest that readers do predictions and that contexts play an important role in the anticipatory processes; data from both the first and the second experiments present evidence that supports the related sense advantage or polysemic advantage hypothesis, since they show that there is a difference between polysemic words and homonymous words regarding linguistic processing in reading.