Queen Mary, University of London
There are four phenomena that are particularly troublesome for theories
of ellipsis: the existence of sloppy readings when the relevant pronouns
cannot possibly be bound; an ellipsis being resolved in such a way that
an ellipsis site in the antecedent is not understood in the way it was
there; an ellipsis site drawing material from two or more separate an-
tecedents; and ellipsis with no linguistic antecedent. These cases are
accounted for by means of a new theory that involves copying syntacti-
cally incomplete antecedent material and an analysis of silent VPs and
NPs that makes them into higher order deﬁnite descriptions that can be
Keywords: VP-ellipsis, NP-deletion, deﬁnite descriptions
There is a common view of ellipsis according to which an elided phrase
between elided phrase and antecedent is one of identity of Logical Form (LF)
or meaning (Sag 1976, Williams 1977).
Hankamer and Sag (1976) argued that ellipsis had to have a linguistic an-
tecedent on that basis of examples like these:
Previous versions of this work were presented in talks at NYU and CUNY in Fall 2004 and
for their comments, especially to Sigrid Beck, Dianne Bradley, Robert Fiengo, Katja Jasin-
skaja, Elke Kasimir, Stephen Neale, Uli Sauerland, Peter Staudacher, Anna Szabolcsi and Eytan
Zweig. Naturally all errors are my own. This research was funded by the Deutsche Forschungs-
gemeinschaft as part of Sonderforschungsbereich 632 (Information Structure).
I use the term elided phrase simply as a descriptive term, without wishing to advocate the
goes, later, for my use of the term NP-deletion.
Ishihara, S., M. Schmitz, and A. Schwarz (eds.)
c 2005 Paul Elbourne
(Hankamer attempts to stuff a 9-inch ball through a 6-inch hoop.)
Sag: It’s not clear that you’ll be able to.
(Sag produces a cleaver and makes as if to hack off his left hand.)
Hankamer: Don’t be alarmed. He never actually does.
We are supposed to imagine these examples being acted out, as it were, so that
there is no linguistic antecedent for the ellipsis in each case, only an accompany-
ing action. Hankamer and Sag claim that (1) cannot felicitously be understood
as “It’s not clear that you’ll be able to push that ball through that hoop,” even
though it is obvious what action is being referred to. An analogous claim is
made for (2). Hence the requirement for linguistic antecedents.
Ellipsis is thought to be based on the meaning or LF of the antecedent phrase
because of cases like the following, which is taken from Heim and Kratzer
1998. In (3), the elided VP can only be understood in the same way as the
scopally ambiguous antecedent VP. That is, if we understand the antecedent
with a drawing scoping above every teacher, we have to understand the elided
VP this way too; and analogously for the other scopal construal.
Laura showed a drawing to every teacher, but Lena didn’t.
tecedent and requires us to understand the same meaning at the ellipsis site, it
is obvious that we can achieve this result. We can also achieve this result by
supposing that ellipsis resolution makes reference to the syntactic level of LF,
where the quantiﬁers will have moved to positions that reﬂect their scopal order-
ing in the semantics (May 1977, 1985). For example, we might have something
like (4) as an LF representation of (3).
LF into the ellipsis site, or of optionally not pronouncing a piece of syntactic
This view is commonly linked to an account of strict and sloppy readings
that sees them as deriving from the pronouns in the antecedent being referen-
tial and bound respectively. The strict and sloppy readings of a representative
example are given in (5) (Ross 1967).
John loves his mother and Bill does too.
‘. . . Bill loves Bill’s mother.’ (Sloppy)
‘. . . Bill loves John’s mother.’ (Strict)
If the pronouns in the VPs are ambiguous between referential and bound, we
have a neat account of this ambiguity (Keenan 1971). We can suppose that the
sloppy reading results from the pronoun being bound, as in (6), and the strict
reading results from the pronoun being referential, as in (7), where it is to be
understood that the sentence is interpreted with respect to a variable assignment
that maps 1 to John.
mother] and Bill [λ
for the correct interpretation to be obtained at the ellipsis site.
The view that the strict-sloppy ambiguity is to be dealt with in terms of ref-
erential versus bound pronouns is independent in principle of the theses that
ellipsis requires a linguistic antecedent and that the relationship between an-
tecedent and elided phrase is one of identity of meaning or LF; but in practise
these three theories are often combined. I will call the composite view that as-
sumes all three the common view.