The Semantics of Evidentials in Questions.

This paper presents a novel cross-linguistic exploration of the phenomenon of Interrogative Flip at the semantics-pragmatics interfaces. Most previous studies describe an obligatory shift in the anchor of an evidential from the speaker to the addressee in interrogatives, across a diverse set of languages. In this work, we discuss a lesser-studied set of facts, which show that in many languages this shift does not take place. Modeling the contribution of evidentials with 'judge'-sensitivity in the semantics and with newly refined notions of commitment and sourcehood in an extended dynamic pragmatics framework, the presence or absence of Interrogative Flip is shown to lie in an evidential's ability to license a commitment update operator |$\uparrow $|⁠. All attested evidential systems are shown to fall in either the class of |$\uparrow $| licensors or not, with apparent exceptions explained across a heterogeneous array of data. A dynamic polar question operator is formulated and its interaction with |$\uparrow $| explored. Finally, a novel link between evidentiality and bias is established, by arguing that the lack of the Flip results in biased questions. [ABSTRACT FROM AUTHOR]