Polissemia: efeitos contextuais no acesso lexical

Abstract

Polysemy and homonymy are, on the one hand, phenomena that have posed difficulties to lexicographers, as well as theoretical, methodological and descriptive issues to semanticists. On the other hand, these phenomena seem to be unproblematic to speakers, who usually have no restrictions in acquiring and processing language in their daily lives. Current studies have focused on both polysemy and homonymy, not only because of the difficulties in classifying words into one phenomenon or the other (which can for instance be observed in dictionaries), but also because it is still unknown whether such classification has practical meaning and whether the differentiation between these two phenomena does exist in the storing process of individuals. mental lexicon. Against this background, this MA thesis aims at analyzing how polysemy and homonymy are accessed by readers/listeners and also finding out whether there is any difference when it comes to deciding the appropriate meaning while accessing close or totally unrelated meanings, which would eventually mean that one task is more effort-consuming than the other one. As the objective was also to understand the roll of context in the access phase, the study involved a cross-modal priming experiment, in which homonymous and polysemous words were embedded in sentence contexts that either clearly tended to one of the associated meanings (e.g. João chupou a manga/Flávia costurou a manga, in which manga means mango in the former and sleeve in the latter) or were ambiguous (e.g. João cortou a manga, in which the verb cortar [cut] can be associated both with mango and sleeve). Subjects were asked to listen to a spoken sentence stimulus containing either the polysemous or the homonymous word (prime), and 100ms after the prime exposure, they were introduced to a visual input (semantically related to the dominant or the subordinate meaning of the prime, or else unrelated to any of these meanings). In addition, subjects performed a naming task with the visual input: if there were a context-based influence during the access phase, subjects should react faster to inputs related to the contextually determined primes than to the contextually ambiguous primes or also the unrelated inputs. Our findings show that there is no difference between the access of dominant and subordinate targets in any of the contexts (subordinate, dominant or ambiguous), although both subordinate and dominant targets were read more quickly than the neural ones. This result supports the multiple access hypothesis. Regarding the relationship between polysemy and homonymy, the reaction time to polysemous words did not differ from the reaction time to homonymous words. This finding provides evidence for the hypothesis that both homonymous and polysemous words are accessed in the same way, irrespective of other meanings diachronically associated with the polysemies.