Hata, Keiko2019-12-292019-12-292019-12-28http://hdl.handle.net/10125/64998This paper assesses the predictive functions of two particles, WA and GA, and examines how these particles are interpreted by L1 and L2 speakers of Japanese. An approach of subject-ellipsis resolution using an audio-stimulated picture-selection task was adopted to compare between the two groups how these particles were respectively used to identify an elided subject (e.g. <em>Taro-WA/GA okotta kara kaetta</em> ‘Because Taro_i was angry ø_i/k went home’). The results of the L1 speakers showed that they, when given WA, instantly identified the elided subject as co-referencing with the WA-marked NP (e.g. ø = Taro) at an above-chance level, whereas there was individual variation in reactions to GA. These results suggest that WA has a predictive function in that L1 speakers attentively used the particle’s syntactic information and then associated the WA-marked NP as the antecedent of subsequent elided-subjects; on the other hand, GA is not a reliable source of syntactic information that induces a certain reading. The results of L2 speakers showed that WA and GA were generally not understood to serve different functions. These findings suggest that L2 speakers are confused with WA and GA, and that Japanese L2 instruction needs revising so that this confusion can be avoided.25en-USJapanese particlessubject-ellipsis resolutionparsingsecond language acquisitionWorking Paper