Abstract

To study Korean relative clause processing(specifically, 'gwan-hyeong' clause), one eye movement experiment was conducted. Garden-path model was accepted as a research framework for this study and temporarily ambiguous sentences were invented to observe reanalysis during reading. The Korean gwan-hyeong clauses followed by 'myeonseo' clause were constructed to lead to syntactic ambiguity, which would be disambiguated in the main verb phrase at the end of the sentence. It was assumed that if a certain syntactic analysis would be preferred in the first pass reading as the garden-path theory argues, it would lead to processing difficulty when encountered with the main verb phrase requiring a different interpretation at the end of the sentence. In this study, it was hypothesized that the initial syntactic analysis formed before reaching the disambiguation region, should follow the strategy requiring more syntactically minimal change. In the eye-tracking experiment, it was observed the strong garden-path effect in the non-minimal change condition, that is the rapid increase of the go-past time at the disambiguation area. This result was interpreted as supporting the minimal change hypothesis.