The long-distance taziji revisited: experimental evidence for inherent non-local bindings

Chinese reflexives provide a unique test for evaluating both theoretical and psycholinguistic models. A prevailingly claim is that the reflexive *taziji* allows only local bindings while the reflexive ziji can take both local and non-local antecedents. Some long-distance taziji cases have been documented but only in contexts lacking a legitimate local antecedent. We report experimental evidence against this claim showing that long-distance binding for taziji is inherently available, i.e., the non-local binding is acceptable even when a local binding is also possible. In a series of formal judgment studies, we first tested how native speakers of Mandarin Chinese interpreted the pronoun ta and the reflexive taziji embedded in bi-clausal structures. By manipulating the gender of the potential antecedents such that they matched or mismatched the anaphor’s gender, we show that both syntactic and non-syntactic constraints influence anaphor resolution (Experiment 1) and that the reflexive taziji can take a long-distance antecedent even when a local option is available (Experiment 2). In Experiments 3 & 4, we refuted the hypothesis that the long-distance taziji is a logophor or a by-product of the independent influence of a logophoric interpretation. The results confirm our hypothesis that the long-distance binding for taziji is indeed inherently available. We also discuss the implications of this study from both theoretical and methodological perspectives.