The domain of transfer

Ke argues that Transfer applies to whole phases, edge included, once they are complete, rather than to the complement of a phase head, and that a phase may wait to be transferred until the next phase is complete, by which time what it contains with unvalued features has moved out. Full-phase transfer at completion would trap a phrase at the edge of vP (edge_trapped_at_completion), but with the delay it does not (transfer_after_next_phase). The edge effects of successive-cyclic movement then come not from Transfer, which a phrase skipping the edge of vP obeys (skip_obeys_transfer), but from Fox and Pesetsky's preservation of spell-out order, which only movement through every edge satisfies (ex13), and only when whole phases are spelled out (complement_spellout_fails).

Implementation notes

• The phase heads are v and C. The clause of (13) is built without T-to-C movement, and nominals and the wh-phrase are single heads.
• Spell-out order preservation is the consistency of the derivation's spell-outs (Minimalist.Derivation.Linearizes). A whole phase is spelled out when it is complete (Minimalist.Derivation.phaseSchedule), a complement when its phase head merges.
• Transfer is read off the chains of the derived object, a link failing when its lower copy lies in a domain transferred by the stage it lands at. A phase is its head's projection under HeadFunction.raising, so a phase head's specifiers are in its phase.

TODO

• The feature criterion (8), under which a transferred phase crashes if it contains an unvalued feature other than on a lower copy, and which makes the delay the only route for a successful derivation.
• Head movement across a phase boundary (14)–(15), and the problems of Table 35.1 beyond the domain and timing of transfer and the edge effects.

References

• Ke 2026, The domain of transfer
• Fox & Pesetsky 2005, Cyclic Linearization of Syntactic Structure formalized
• Chomsky 2000, Minimalist Inquiries: The Framework
• Chomsky 2001, Derivation by Phase