Séminaire de Recherche en Linguistique

Our talk examines the phenomenon of Free Relative clauses in languages like English that use wh-phrases in these constructions, as in “He ate what I cooked.” Our proposal is structurally conservative, following Groos & van Riemsdijk 1982 and many others in positing a null head, in our analysis called “D_FR”, with a relative-clause wh-CP in its projection (i.e. the “Comp” analysis). We provide evidence from complex wh-phrases in genitive constructions that the relation between this head and the wh-phrase in Spec CP is best characterized as an Agree relation: D_FR acts as a Probe in the sense of Chomsky 2000, and picks up its syntactic and semantic features by minimal search for a Goal bearing an operator feature with which it undergoes Agree. The notion of “Phase Edge” turns out to provide exactly the right locality relation for Agree; the “complexity” of the fronted whP is not a factor (as is often assumed). Our proposal naturally accommodates the wider distribution of Free Relatives bearing the morpheme –ever (“I’ll read which*(ever) book you read.”), and the peculiar phenomenon of Paucal Free Relatives (Andrews 1985). We also examine new data from Italian and English suggesting that some Free Relatives are actually ambiguous, and may also have a “Head” analysis (Bresnan & Grimshaw 1978) in which there is no null D_FR head: the wh-phrase is the head of the Free Relative. We briefly discuss the implications of these two structures for differences in patterns of Case-Matching within and between languages.

Finally, we present a modification of our analysis in light of Chomsky’s 2013 suggestion that the labels of syntactic objects are determined following the same principle of Minimal Search that determines the Goal for an Agree relation. This modification not only has the advantage of explaining the categorial flexibility of Free Relatives, but it unites labeling and Agree into a single process.