Revisiting unselected embedded questions in the light of Classical Greek wh-clauses
Abstract This paper addresses the issue of the questions embedded under predicates normally selecting for propositions like know (unselected embedded questions UEQ). This problem was handled in Adger and Quer (Adger, David & Josep Quer. 2001. The syntax and semantics of unselected embedded questions. Language 77. 107–133) and Öhl (Öhl, Peter. 2007. Unselected embedded interrogatives in German and English : S-selection as dependency formation. Linguistische Berichte 212. 403–438). Both articles notice a difference between yes/no- and wh-questions. The distribution of the latter seems to be less restricted. However data from Classical Greek shed new light on the matter. Classical Greek uses two sets of wh-items in what looks like embedded questions (h- and t-). It is shown that h-clauses do not denote questions but propositions. The selection mismatch arises with t-clauses. They denote questions and have the same distribution as yes/no-questions. Moreover Classical Greek provides new evidence in favor of 1) the sensitivity of the UEQ to the polarity of the environment, building on Giannakidou’s (Giannakidou, Anastasia. 1998. Polarity sensitivity as (non) veridical dependency. Amsterdam & Philadelphia: John Benjamins) definition of nonveridicality; 2) the presence of a determiner on top of the UEQ as proposed in Adger and Quer (2001). The article argues nevertheless that the sensitivity is not due to the determiner, but to an operator going along with UEQs. It is shown that the determiner is a type-shifter turning the question into a proposition and thus repairing the apparent selection mismatch.