This talk takes as its starting point the complementarity (or apparent complementarity) between the locality conditions that apply to head movement, and those that apply to phrasal movement (cf. Pesetsky & Torrego 2001:363). As a first approximation, head movement must be maximally local (the Head Movement Constraint; Travis 1984), while phrasal movement cannot be maximally local (Anti-Locality; Abels 2003).

As is often noted, however, feature-driven conceptions of minimality predict that there would be nothing wrong with head movement skipping a featurally-irrelevant head en route to a featurally-relevant one (e.g. Roberts 2010:179ff.). On the flip side, anti-locality predicts that maximally local head movement simply should not exist. If true, this would entail that what looks like maximally local head movement must be relegated to a non-syntactic component of the grammar (e.g. that it happens "at PF"). But this is problematic, since head movement has been argued to have interpretive effects; moreover, if we want to explain the (near-)complementarity between these two locality conditions, placing them in disparate computational modules seems like a rather poor starting point.

In this talk, I propose a unified way of deriving Anti-Locality along with the Head Movement Constraint in the usual case. Crucially, however, the account also allows for head movement to be less-than-maximally-local, under particular circumstances—namely, when a previous syntactic relation has already been established between the attractor and the maximal projection of the moving head. I argue that this is precisely what is going on in cases of clitic-doubling / cliticization—and potentially in other cases of long head movement, as well.

Finally, I show how this account solves a nagging problem regarding the cross- and intra-linguistic distribution of Person Case Constraint effects. It also provides further support for the view that there is generally no syntactic agreement where no morphophonological covariance can be observed (e.g. that there is no such thing as "null" agreement with objects in a language like English—contra Chomsky 2000 et seq.).

Data from Basque, Breton, Bulgarian, Hebrew, and English will be discussed.