This dataset consists of two parts. The first part contains eye-movement data reflecting the reading times gathered when native-Chinese speakers processed coreferential noun phrases which (i) either did or did not contain a grammatical determiner, and (ii) either did or did not contain a semantically incongruent word. The second part consists of the transcripts from think-aloud interviews which were conducted subsequent to the eye-tracking experiment to investigate how readers at different levels of proficiency processed the coreferential noun phrases.