I REALLY enjoyed Stefan Keine and Rajesh Bhatt’s paper, ‘Interpreting verb clusters’, in NLLT. It’s a tremendous piece of work, the motivation for predicate composition triggered by head movement is very compelling and interesting, and the facts about non-moving constituents taking widest scope w/r to the matrix verb in long passives are mind-boggling. Three thoughts:

a) The objection to the straw-man analysis proposed for a Wurmbrand-type approach to some of the scope facts doesn’t hold up in more recent developments in the vP; once you split syntactic argument-introducing up from semantic argument-introducing, e.g. the phasal, external-argument saturating, QR-targetting VoiceP and the nonphasal, external-argument-variable-introducing vP (cf Pylkkänen’s ‘bundling’ hypothesis), you can have your cake and eat it too, and the argument that the verb+object combo should also be of type <st> and hence a potential target for QR should go away. That’s a bit of a distraction though because…

b) the real argument against that account is the difference in matrix scope behavior for embedded quantificational elements with respect to the matrix verb compared to with respect to other matrix quantificational elements. Embedded quantificational elements should take wide scope w/r to the verb but behave normally w/r to other quantificational elements, on K&B’s account, but should behave as if they’re in the matrix clause on the alternative Wurmbrand-type approach. They sort of showed this but didn’t drive it as hard as possible.

c) There may be one important problem for their account though. For me, the key data point for distinguishing the head-movement account from the intermediate-adjunction-side, phasal, Wurmbrand-style account is the absence of restitutive scope for ‘again’ in long passives with change-of-state embedded predicates. But if I understand everything correctly, I think they predict that restitutive scope should just GENERALLY be impossible with change-of-state predicates, if the stative predicate head-moves to eventive v in causative and inchoative predicates. That head-movement should cause predicate-composition to apply to the State+v combo, and drag the denotation of v down into the trace position of State, preventing ‘again’ from EVER isolating the stative meaning component of such verbs and yielding the restitutive reading. I think this is a really difficult problem for them to solve with their account. If the requirement for ‘again’ to have only a matrix repetitive reading in long passives arises from head-movement to the matrix verb, then I think it should arise from head-movement in simplex clauses as well.

But really, this is a super-interesting and important paper. I really love it so much, also because it turns everything you think about movement and scope interpretations on its head: movement of the embedded verb to a higher position causes interpretive LOWERing of the MATRIX verb. It’s not scope-expanding, it’s scope-narrowing, kind of! SO COOL.