A Minimalist Parsing Account of Attachment Ambiguity in English and Korean

This paper investigates whether attachment ambiguity preference attested in psycholinguistic literature can be correctly predicted, utilizing memory based metrics developed by Graf et al. (2017). Two different types of attachment ambiguity in Korean, the dative argument attachment ambiguity and the relative clause attachment ambiguity, are mainly tested. Furthermore, in order to examine whether the Minimalist Grammar Parsing Model can predict the broad human sentence processing effects, the work in the previous literature in Graf et al. (2017) were tested together. While none of the single metrics made the correct predictions, the combination of metrics based on tenure and size could predict the preferences of all different kinds of constructions reported in psycholinguistic studies. This suggests that human sentence processing effect cannot be understood by a single element. The well-selected metrics in this paper can tell us what kind of structural complexity are related to processing sentences.