Invited talk by Soichi Takahashi

In German, the wh-phrase `was für' (what for) can be split, as in (1).

 

(1)
a. [Was für Bücher] hast du gelesen?
what for books have you read

 

b. [Was] hast du [für Bücher] gelesen?
what have you for books read

 

`What (kind of) books did you read?'

 

However, if `was' survives ellipsis in sluicing and the für-phrase is included in the elided TP, unacceptability results. This fact appears to threaten the hypothesis that an abstract syntactic structure constitutes an ellipsis site. In this talk, we argue that it actually does not. To do so, we first discuss how the split is carried out. We adopt Abels' (2003) remnant movement analysis, according to which the für-phrase moves out of the wh-phrase, which is followed by remnant movement of the wh-phrase. We further show that remnant-creating movement of the für-phrase is scrambling. Capitalizing on Merchant's (2001) claim that ellipsis bleeds movement driven by the EPP property of a head, we now explain the impossibility of the split in sluicing: scrambling being EPP-driven movement, the necessary first operation for the split is inapplicable within a to-be-elided site. We also address an outstanding challenge to the remnant movement approach: the für-phrase cannot precede was. Together with the independently motivated internal structure of the wh-phrase, we propose that Fox and Pesetsky's (2005) theory of mapping syntactic structure onto linear order provides an answer to this restriction.

 

(This is joint work with Thomas Schökler)