Deletion Versus Pro-Forms: A False Dichotomy

        A common dichotomy in the study of ellipsis is the dichotomy between null pro-forms, which are assumed to be generated essentially as base-generated empty pronouns,  lacking internal structure,  and  null phrases that become null as a result of deletion (see Johnson (2001) for a discussion of this distinction). In this paper,  I will show that the  dichotomy is often not pure, but a result of deletion of a small enough part of a phrase that creates the appearance that the anaphoric item is a null pro-form.  I shall illustrate this with two anaphoric constructions in English,  British English do and so-called NP-ellipsis.

   On the face of it, both of these constructions look like excellent candidates for pro-form status. While British speakers take (1) and (2) to be synonymous, with (1) involving VP-ellipsis, as in American English, and (2) being an example of British English do.  However,  there are differences; while VP-ellipsis allows wh-traces (as in (3)), British English do does not (4).  Furthermore,  while VP-ellipsis allows an object quantifier within the elided VP to  take inverse scope over the subject (5),  British English do only allows direct scope (6).  These differences might lead one to posit British English do as a pro-form, and take VP-ellipsis to be deletion of a phrase with internal syntactic structure.  However,  British English do has characteristics that require it too to have internal structure. For example, sloppy readings are considered a symptom of a need to compose a complex meaning of an element,  so  that the sloppy reading in (7) is thought to be represented by a logical form as in (8).  Just as (7) has the sloppy  bound-variable reading, (9), with British English do,  has this reading as well.  Furthermore,  British English do must allow the interpreted  VP to contain a DP that had  undergone A-movement out of it. For example,  (10) requires interpretaton of a VP that  has  been extracted from by subject-to-subject raising,  and (11) requires interpretation of an unaccusative VP (the  VPs contain a resultative construction, and Levin & Rappaport-Hovav (1995) have shown that resultatives can only be predicated of direct objects;  this resultative is therefore a diagnostic for unaccusativity).

         We have a seeming paradox, therefore, with respect to British English do, in that it can’t be looked into for wh-dependencies and quantifier scope, but can be for A-movement and sloppy identity.  The paradox can be resolved if,  following a suggestion of  Chris Collins (personal communication), we analyze do as v in a Larsonian shell  (Larson (1988)), and say that all VPs  are introduced as complements to v (Legate (2003), Collins (1997)),  and postulate British English do as involving deletion of the VP complement to this v.  Hence,  (2) will involve a structure with the characteristics in (12).  If we assume  Chomsky’s (2000), (2001) theory of phases, in which syntactic operations take place phase-by-phase,  and the  phases are vP and CP (and perhaps DP-more on this below), we can adopt Legate’s analysis of wh-movement to [Spec, CP] as first involving movement of the wh-phrase  to  the edge of the vP phase.   Deletion of the VP,  if it occurs in the syntax and not at PF,  will cause the VP and everything that it dominates to lose all formal features,  including those that drive movement.  Presumably, QR which involves inverse scope would also require movement to the edge of vP.

     This account would allow us to account for the hybrid nature of British English do, in that it is analyzable with respect to some processes but not others.  We can actually see the same situation in so-called NP-ellipsis, as in (13). If this is ellipsis, it is very different from VP-ellipsis. For one thing,  NP-ellipsis can’t contain a wh-trace (compare (14) with its non-NP-elided counterpart  (15).  For another,  a genitive DP can allow inverse scope, as in (16),  but when the NP complement of the genitive is elided,  inverse scope becomes impossible (17).  Therefore, we have the same situation with respect to NP-ellipsis as with British English do,  with the same temptation to analyze the anaphoric construction as involving a pro-form without internal syntactic structure.  Again,  however,  we have evidence that there must be internal structure.  Elbourne (2005) has found a sentence , given as (18),  that requires that the interpretation of the null NP must first involve interpretation of an elided VP,  and we know that VP-ellipsis requires internal syntactic  structure.  Therefore, NP-ellipsis must as well,  and involves the same mechanisms of deletion of a smaller constituent than a phase edge that we saw for British English do.


 

(1)  John will read the book, and Fred will ___too.

(2)     John will read the book, and Fred will do___, too.

(3)       Although I don’t know which book  John will read, I do know which book Fred will___.

(4)     *Although I don’t know which book John will read,  I do know which book Fred will do.

(5)       Some man will read every book, and some woman will___,too.  (Some<every,  every<some)

(6)       Some man will read every book, and some woman will do, too. (Some<every, *every<some).

(7)      John likes himself, and Fred does __too.

(8)       lx (x likes x) John and ly (y likes y) Fred.

(9)      John might like himself, and Fred might do, too.

(10)  John might seem to enjoy this, and Fred might do,too.

(11)   The river might freeze solid, and the lake might do, too.

(12)     [vP [DP Fred ][v’ [v do][VP read the book]]]

(13)    Although I didn’t  see John’s pictures of  Sally,  I did see Fred’s_____.

(14)  Although I don’t know who he saw many pictures of,  I do know who he saw some pictures of__.

(15)  * Although I don’t know who he saw many pictures of, I do  know who he saw some___.

(16)  Although I don’t enjoy some linguist’s discussion of every problem,  I do enjoy some philosopher’s discussion of every problem . (Some<every, Every<some).

(17)  Althouth I don’t enjoy some linguist’s discussion of every problem,  I do enjoy some philosopher’s___. (Some<every, *Every<some).

(18)  When John wanted to cook, he met some people who didn’t want him to; and when he wanted to clean he met some too.

 

  References:

1.        Chomsky, Noam (2000). “Minimalist Inquiries; The Framework” , in D. Michaels, R. Martin, & J. Uriagureka, eds., Step by Step: Essays in Honor of Howard Lasnik,  MIT Press, Cambridge, Mass., pp. 89-155

2.        Chomsky, Noam (2001). “Derivation by Phase”, in M. Kenstowicz, ed., Ken Hale: A Life in Language, MIT  Press, Cambridge, Mass., pp. 1-52

3.        Collins, Chris (1997). Local Economy, MIT Press, Cambridge, Mass.

4.         Elbourne, Paul (2005). “The Semantics of Ellipsis”, unpublished ms., University of Konstanz

5.         Johnson, Kyle (2001). “Why VP-Ellipsis Can Do What It Can, But Not Why”, in M. Baltin & C. Collins, eds., The Handbook of Contemporary Syntactic Theory, Blackwell & Sons, Oxford, U.K., pp. 439-479

6.        Larson, Richard (1988) “On the Double Object Construction”, Linguistic Inquiry,   Vol. 19, No. 2, pp. 334-391

7.        Legate, Julie (2003), “Some Interface Properties of the Phase”, Linguistic Inquiry, Vol. 34, No. 3, pp. 506-516

8.