Bare quotatives as embedded speech acts

In Japanese and Korean, embedded clauses headed by -to/-ko can appear without selecting predicates. We review a wide range of empirical phenomena surrounding this structure, which we call Bare Quotatives, and argue that the Bare Quotative structure represents an embedded speech act. While the interpretation of the missing predicate is either say or think, the difference is neither lexical nor categorical. It is a consequence of the referent of the addressee’s argument of the embedded speech act. When it is distinct from the speaker, the embedded quotative is an actual utterance. When it is coreferential to the speaker, on the other hand, the embedded speech act is a monologue by the speaker, which can lead to the think-interpretation. Our analysis also correctly predicts that the BQ structure is incompatible with the purely epistemic interpretation of the complement of think.