통사적 정보와 의미적 정보가 한국어 어휘 처리에 미치는 영향: ERP 연구

Abstract

An ERP experiment was conducted to investigate the semantic and syntactic processes as well as their interplay in Korean lexical integration using the primed lexical decision task paradigm. In this study, participants were presented with Korean noun phrases consisting of the prime Eojeols and the target words in four conditions: either congruent, syntactically incongruent, semantically incongruent, or both syntactically and semantically incongruent between the prime and the target. As a result, the ERP patterns of the congruent and incongruent conditions, recorded during the lexical decision on the target words, were compared. The syntactic violation, which consisted of adverb-noun combinations, elicited early left anterior negativity (ELAN) but nor P600 neither LAN (left anterior negativity). The semantic violation, which was created by semantically incongruent adjective-noun combinations, evoked early starting N400. More importantly, it was noted that the syntactic violation combined with the semantic violation did not cause greater N400 comparing to the semantic violation alone, indicating no interaction of the semantic violation with the syntactic one at N400. These findings suggest syntactic and semantic information may be processed sequentially and independently at the phrase level, and a well-formed phrase structure may not be necessarily a prerequisite for the semantic integration.