Inanimate antecedents of the Japanese reflexive zibun: Experimental and corpus evidence

This research sets out to challenge a conventional wisdom in Japanese linguistics, that the reflexive pronoun zibun is unable to take an inanimate antecedent. Through careful presentation of the data, including corpus sources, it is unequivocally demonstrated that the reflexive use of zibun can indeed overcome the animacy constraint and be anteceded by an inanimate antecedent, without any personification present. This has specific theoretical consequences in the sense of providing a theoretical simplification behind reflexivity modeling in Japanese. This is followed by a psycholinguistic experiment investigating how native Japanese speakers judge and process sentences with inanimate zibun. The key factors tested are the animacy of the antecedent, and also if the type of sentence, episodic versus generic, will affect the acceptability of inanimate zibun. Results from the experiment show that native speakers indeed do find inanimate zibun acceptable, and appear to process it slightly slower than the animate counterpart. The episodic versus generic distinction does not play a role in either the judgment or processing. Combining the corpus and experimental data anchors the phenomenon firmly. Finally, our attention turns to how to account for inanimate zibun theoretically, where we draw information from how zibun can already take inanimate antecedents if they are construed as agentive, or if zibun is used as an adverbial reflexive, showing that in fact, inanimate zibun does not require additional theoretical treatment - leading to a reformulation of Kuno's animacy constraint.