Head movement: cyclicity and labels in a derivational model
Balazs Suranyi
November 2, 2005, 11:00 AM
Abstract:
Head movement (HM) is known to have proven problematic within the minimalist program, especially for a checking theory (c-command, Extension Conditon/cyclicity, strict locality: HMC, Uniformity Condition, a series of empty inflectional heads, inconsistency w.r.t. to the 'strong/weak' property of certain functional heads, explaining the Mirror Generalization etc). In this paper I put forward an alternative to current proposals to treat HM (in terms of remnant XP-movement, PF-movement), which are often pointed out to create descriptive difficulties, as well as introduce extra machinery (cf. Benedicto 1997, Müller 2001, Zwart 2001, Lechner 2005 etc), and even at that price, some of the existing problems are still left unresolved.
My proposal is that the null hypothesis (in a derivational model) holds:
HM and XP-movement are symmetric in that both HM and XP-movement re-merge
an element to the current root (cf. Suranyi 2002, Fanselow 2003). The
difference between them is that in HM it is the moved element that projects,
which is due to the Labelling Generalization (the element that projects
its label is the one whose label has selector/probe features to be 'checked',
cf. Chomsky (1995, 2000)): in HM it is the moved head that has a feature
to be 'checked'. I combine this conceptualization of HM with (i) the Locus
Principle (Collins (2002): the locus being checked stays the same until
it is fully saturated by checking, cf. Featural Cyclicity), (ii) the null
hypothesis w.r.t. the size of phases, namely that each XP is a phase (Müller
2003, cf. also Takahashi 1994, Epstein & Seely 2002), and (iii) the
assumption that reprojection of labels (cf. Hornstein & Uriagereka
2002) (in this case: within a moved head) is possible if it results in
the checking of some feature (by Last Resort). This treatment of HM is
shown to emerge as a
strong competitor of current approaches.
Last Modified: October 22, 2005