Conference presentation: Deducing improper movement from phase based C-to-T phi transfer

Miki Obata and Sam Epstein presented their talk, Deducing improper movement from phase based C-to-T phi transfer, at the 27th annual meeting of the West Coast Conference on Formal Linguistics, held at UCLA May 16-18.

Abstract

We seek to reveal and address empirical and theoretical consequences stemming from
Chomsky’s (2005:OP) C-to-T phi-feature inheritance (hereon CTI) analysis. We claim: [1] Simultaneousvattraction as a consequence of CTI “splits” the features on a wh-phrase: [Case]/[Phi] moves to Spec-T andv[Q] to Spec-C, [2] [1] makes it possible to rule out improper movement (IM) without appealing to the activity condition (see Nevins 2005 for arguments against activity and for a different approach to IM).

We argue that contra standard assumptions, IM is not a unified phenomenon but is
correctly classified into two distinct types: Case on a moving element is valued after A’-movement in (6) and before A’-movement in (7). We demonstrate that both types of IM are explanatorily excluded under the phase-based approach and it is especially the latter case that empirically supports the feature split system presented above. The derivation of (7) is shown in (8). When C and T each attract “who1″ simultaneously (→8b), the features on “who1″ are split: [Q] goes to Spec-C and [phi]/[Case] goes to Spec-T In the matrix clause (→8c), assuming PIC, only (the edge) “who3″ in the embedded Spec-C is visible to matrix C-T probing. Notice that “who3″ has only [Q], not [Phi] by virtue of feature split, so that “who3″ is not an appropriate (matching) goal for the probing matrix T precisely because “who3” lacks [phi]. In contrast, the matrix C can attract “who3″ but [uPhi] on matrix T is never valued, causing crash. This is a
direct result of the CTI, coupled with OP under which C and T separately attract different featural subsets, simultaneously from the ‘same’ launch site. The absence of [phi] on “who3″ makes it impossible for “who3″ to improperly move (“back”) into an A-position, (Spec-T). The other type of IM (6) is also excluded straightforwardly: when the derivation reaches the embedded CP in (6), the transferred TP includes unvalued [uCase] on “who1/2″ causing crash.