Communicative reception reports as hear-say: Evidence from indexical shift in Turkish

In the vast literature on the semantics of speech and attitude reports, communicative reception reports ('John heard/read that Mary's retiring') have been unjustly neglected. What makes these reports interesting is their hybrid nature: they can be like speech reports in semi-faithfully reporting another person's speech act (in which case they, unlike 'believe' or 'hope', can take direct quotation), but they can also behave more like belief reports in describing the subject as being the holder of a certain mental/information state (as in 'the doctor told John he has pneumonia but all he heard is that he's dying'). In Turkish, reception reports overtly express both a 'hear' and a 'say' component. We take this surface structure literally, analyzing 'x heard that p' roughly as 'x heard LOG saying that p', where LOG can pick up (i) the reported speaker, leading to a speech report interpretation, or (ii) the matrix subject, leading to an attitudinal interpretation (where 'say' is understood as a internal, mental speech/thought act). Assuming that 'say' can house a monstrous indexical shifter (Şener & Şener 2012), we now predict that on the speech report reading, embedded 1s can shift to the reported speaker, while on the attitudinal reading they can shift to the reported hearer. We provide novel Turkish data showing that first person indexicals under reception verbs in Turkish are indeed three-ways ambiguous (no shift, shift to subject, shift to reported speaker), and propose an analysis couched in an event-based, conjunctive analysis of attitude/speech reporting (cf. Kratzer 2006).