Vortrag von Erik Zyman (Universität Chicago)

In the context of syntactic inquiry that skews more “representational” than “algorithmic” (Müller 2015), a major and controversial research topic has been the internal structure of the verb phrase. On the widely adopted Larsonian approach (Larson 1988, 1990, a.m.o.), a series of internal arguments following the verb in languages such as English (as in Katie gave A BOOK TO MIKE) form a constituent—specifically, a VP whose head V has moved out of it. But there is something somewhat unsettling, though by no means empirically fatal, about that analysis: in English, the putative headless VPs it posits seem to pass no constituency tests other than coordination (Jackendoff 1990). For example, they are normally uncleftable (*It was A BOOK TO MIKE that Katie gave). This talk, however—building and expanding on observations from Nakajima 1994 and Takano 2002, a.o.—shows that sentences with that general structure are in fact possible with the right distribution of focal stresses: No, no. (?)It was a BOOK from FRANCE to a GIRL from GERMANY that Mike gave for Christmas. The empirical profile of such sentences (“Nakajima-clefts”) is investigated, and it is argued that they involve clefting of a headless VP, not multiple clefting of individual arguments (cf. Müller 2018 on German complex prefields), an analysis that receives independent support from certain types of fragment answers. Headless VPs, then, are real, and much less ghostly than they initially appear: they do pass (non-coordination-based) constituency tests, even in English, as long as we look in the right places. The talk concludes with some speculations about types of clauses that do not launch Nakajima-clefting, and (on a more algorithmic than representational note) what these may tell us about the elementary operations that build and externalize the verb phrase.