Evidência neuronal de processamento morfossintático via parsing: um estudo de MMN

Abstract

This dissertation aimed to answer the question of how we represent morphosyntactic knowledge and how we process it. To this end, we conducted an experiment of event-related brain potentials extraction in which participants simultaneously watched a muted series and listened to phrases that varied in grammaticality and could occur as either a “standard” stimulus (more frequent repetition) or a “deviant” stimulus (rare repetition). This paradigm, known as the multifeature paradigm, is recognized for eliciting the electrophysiological response called mismatch negativity (MMN), which is known to be automatic and relatively independent of attention. The MMN occurs when a deviation from a regularity formed in memory about repeated stimuli is detected. When modulated by syntactic factors, this effect is known as syntactic MMN. The phrases in this experiment fit into the following conditions: “grammatical” (“I adopt.1p.sg”), “1 violation” (“I *adopt.2p.sg”, a person agreement violation), or “2 violations” (“I *adopt.2p.pl”, a violation of both number and person agreement). The hypothesis was that, if processing occurred via parsing of the phrases, access to morphosyntactic features and their combination, then the MMN would be larger in the “2 violations” condition than in the “1 violation” condition, reflecting a higher processing cost. The findings point to a negative amplitude effect that is a direct, cumulative, sustained and automatic effect of morphosyntactic processing, even though an MMN effect cannot be entirely ruled out. Thus, we contribute to the debate on the nature of language with respect to its representation and processing by the brain by indicating that we possess abstract morphosyntactic knowledge, employed in the course of computation, and that comprehension processing involves parsing and combinatorial mechanisms.