Weird VPs

Split off from Some Scope Creep.

"shoe kuq" as a constituent. It's clear why things in TP and above make scopes but shoe does not.

This VP structure, combined with generalized Predicate Modification, allows for adjuncts to appear between subject and object.

Laqme suggests this unusual syntactical structure for VPs. She thinks we should analyze serial verbs as constituents, rather than using movement. In that case, it's natural that shoe would not introduce scope; rather, shoe kuq can somehow be considered a V, reflecting that serials and atomic verbs really are syntactically the same. This would change the behavior of ru, but perhaps for the better (you can do serial-ru-serial).

The tree structure suggests that Shoe kuq jí súq and Shoe kuq jí are constituents — as is indeed sort-of evidenced by Underfilling, if we allow that?

Supposedly there is some "internal structure" to a serial V. We imagine that shoe can "apply to" kuq to get the meaning of shoe kuq, if you do a truly silly amount of type-shifting. Or maybe there is a covert operator in any serial that knows how to "do" the serialization — shoe kuq is really shoe x_shoe kuq where x_shoe is some unpronounced operator that implements merge-into for shoe's frame.

Scope is just left-to-right, serial verbs are just verbs, and there is no movement in VPs. Anaphora are no longer resolved via c-command. This is a big step away from Chomsky-style generative grammar, towards something more like Montague grammar / effectful parsing / continuations and type-checking-based semantics.

See Quantificational Binding Does Not Require C-Command (Barker, 2012) and A modular theory of pronouns (and binding) (Charlow, 2017) for more ideas, and the ECNLS slides. Also Continuations and Natural Language has a nice chapter about "the irrelevance of c-command".