언어의 생산과 이해, 그리고 뇌의 예측성

Abstract

The purpose of this paper is to argue that the theories of language production and comprehension adopt the predictive process of the brain. Predictive processes can be found in many domains of language: discourse, syntactic, semantic, and phonological domains. In the discourse domain, the predictive process is found in the conversations that take turns in about 200 milliseconds. Syntactically anomalous structures, e.g., garden-path sentences, elicit a large positive wave (P600), indicating the unexpectedness of the structure. Semantic prediction is exhibited in an evoked response (N400), whose amplitude is modulated when a given word is not congruous in a context. A phonological rule of a mother tongue also creates predictions about phonologically legal forms in pseudowords. The predictive mechanism of the brain is supported by neurological studies: Bar et al. (2007) show the default mode or baseline of the brain areas (Raichle et al. 2001) overlap with the regions activated by tasks that recruit associative processing. This means associative activation equals the process of the brain's baseline state. Bar et al. (2007) propose that continuous generation of predictions is derived from associative processes and analogical mapping of the brain. This study proposes that language production and comprehension theories employ this predictive process of the brain in that prediction seems to affect speech errors as well as implicature recovery.