Proceedings of the Annual Meeting of the Cognitive Science Society

The advent of broad-coverage computational models of human sentence processing has made it possible to derive quantita-tive predictions for empirical phenomena of longstanding interest in psycholinguistics; one such case is the disambiguationdifficulty in temporarily ambiguous sentences (garden-path sentences). Adequate evaluation of the accuracy of such quan-titative predictions requires going beyond the classic binary distinction between ”hard” and ”easy” garden path sentencesand obtaining precise quantitative measurements of processing difficulty. We report on a self-paced reading study designedto estimate the magnitude of the disambiguation difficulty in two temporarily ambiguous sentence types (NP/Z and NP/Sambiguities). Disambiguation was more than twice as hard in NP/Z sentences as in NP/S sentences. This contrasts withthe predictions of surprisal estimates derived from current broad-coverage language models, which lead us to expect asmaller difference between the two.