Abstract

The purpose of this study is to investigate whether the inference about the emotional states of characters is generated in text comprehension. And the minimalists' and the constructionists' proposition ate compared in inference process. In this connection, the form of emotional representation is identified. For this purpose, as the on-line measurements, in Experiment 1 and 2, each of the word recognition task and the lexical decision task was given to subjects. The results showed that both semantic association words and emotional state words were inferenced in text comprehension. Another result implied that emotional state words were encoded as much as semantic association words. These two experiments demonstrated that the minimalists' proposition was limited. In Experiment 3, although not the on-line measurement approach, the context verification task was used to investigate whether the target word was activated from text's context. The result showed that the response time of relevant semantic association words was longer than that of relevant emotional state words. These experiments suggested that although emotional state words were activated in text comprehension, measuring the representation of the concrete words could not reflect the degree of activation of emotional state words, because they could be expressed many different kinds of emotional state words within the same sub-group.