Multiple constraints and the resolution of Korean null subject anaphor

This study investigates how Korean adults interpret and identify the referent of a null subject in a narrative text, given different types of topic continuity and person features. We have found that the first-person feature was most accessible in the weak topicality condition in resolving the null subjects, and that the target sentences ending with the first-person modal suffix ‘-lay’ were read and responded to faster, and interpreted more correctly than other types of stimuli involving a third-person modal (‘-tay’) and a person-neutral modal (‘-e’). Furthermore, of the two first-person-specific featured types, the null subjects in the topically weak contexts were processed significantly better than those in the topically strong conditions. It was argued that anaphoric dependency would be formed more discursively than morpho-syntactically in the strong discourse continuity contexts involving no extra processing load due to the shift among multiple eligible candidates. It was also argued that, in the absence of discourse topic assigned strongly to more than one eligible referent in advance, morpho-syntactic cues involved in verb modality are likely to become prominent in the mind of the processor. It is concluded that these main findings support a constraint-based approach, but not the Centering-inspired work.