Colloquium with Rebecca Lewis, UMass Amhert

Title: 

 I’ve elided more of this title than you would have done:  variable extraction out of do-ellipsis supports full phase ellipsis

Abstract:

"Traditional" Verb Phrase Ellipsis (TVPE) involves deletion of (minimally) a lexical verb and any complements (1), and has been argued to target either the clause-internal phase or its complement in English (Gengel 2007, Harwood 2013, Boskovic 2014). In British English there is, in addition to TVPE, an elliptical construction whereby the gap site is preceded by a non-finite form of do (do or done).
(1)    a. He has read the paper but she hasn’t <read the paper>.
b. He has been reading the paper but she hasn’t (been) <reading the paper>.        TVPE
(2)    He has read the paper but she hasn’t done.
(3)    He will read the paper but she won’t do.                              do-ellipsis
Interestingly, extraction out of do-ellipsis is constrained in ways that it is not out of TVPE. For example, it is possible to move subjects of unaccusative and raising verbs out of do-ellipsis (4a-b), but not the subject of passives (4c). Topicalization, quantifier raising and pseudo-gapping are all possible with do-ellipsis (5a-c), but wh-movement of an object or long-distance subject is not (5d). Not all wh-movement is disallowed, however: local subject wh-movement is licensed (5e). All of these extractions are allowed with TVPE.
(4)    a. He has arrived early but shei hasn’t done <arrived ti early>.
b. He seems to have enjoyed the talk but shei hasn’t done <seemed ti to enjoy the talk>.
c. *He has been criticized but shei hasn’t (been) done < (been) criticized ti>.

(5)    a. Pride and Prejudice, I will read. Emma, I won’t do.
b. A student will visit every poster and Opi  a professor will do <visit every posteri> too.
c. Some students have read Pride and Prejudice while others have done <read ti> Emmai. 
d. *I know that he has read Emma but I don’t know whati she has done <read ti>.
e. A: He wouldn’t write the review at such short notice. 
B: Whoi would do < ti write the review at such short notice> ?
In this talk, I show that the crucial difference between TVPE and do-ellipsis is that while TVPE may be phase or phase complement ellipsis, do-ellipsis is obligatorily phase ellipsis. In particular, I argue that the extraction facts regarding do-ellipsis can be explained if three things are true: (i) non-finite do can be inserted to host stranded non-finite affixes in British English, (ii) do-ellipsis is obligatorily deletion of the entire clause-internal phase, (iii) not all wh-movement targets Spec,CP. That is, I first claim that do-ellipsis is an extension of the prototypical finite ‘dummy do’ insertion found in all Englishes (Ramchand and Svenonious 2014, Ramchand 2018). The next claim, that do-ellipsis involves obligatory phase ellipsis, ensures that any movement that has a final landing site below the next phase head (C), which triggers spell-out/deletion of the lower phase, is allowed. Finally, the fact that local subject wh-movement is allowed out of do-ellipsis, but other wh-movements are not, provides additional evidence for Boskovic’s (2021) claim that local wh-subjects move to a position below SpecCP. The somewhat puzzling extraction facts regarding do-ellipsis receive a fairly straightforward explanation here as the interaction between uniform processes operating in different domains: movement operations that target different projections, deletion of (a particular piece of) syntactic structure, and post-syntactic insertion of material to host stranded affixes.