The Semantics of Ellipsis
95
(a)
its denotation describes an action or thing made salient by an
obvious sensory clue; or
(b)
it is a copy of a phrase of the same syntactic category drawn
from the linguistic environment.
This theory should be seen as part of general linguistic competence, but it will
be used by speakers and hearers in different ways. The hearer, whose job it is
to try to work out what the speaker was saying, might not arrive at the precise
unpronounced LF phrase that is present in the mind of the speaker, especially
in cases where there is no linguistic antecedent; but communication will have
proceeded well enough if the speaker comes up with something with the same
or a relevantly similar meaning. See Neale 2005 for salutary discussion of the
asymmetric roles of speaker and hearer.
Before leaving the current theory, there are two loose ends that should be
tied up. Firstly, the theory of ellipsis outlined here is naturally one way of
spelling out some details left obscure in my previous work (Elbourne 2001,
forthcoming) in which I claimed that E-type anaphora was NP-deletion; but this
latter thesis is independent of any particular analysis of NP-deletion.
Secondly, we should revisit example (60), repeated here as (100), to make
sure that we do not end up predicting that it is good.
(100)
The garbage can is full. *I hope that you will, for a change.
In the absence of a strong sensory clue of the kind exemplified earlier, we have
to work with the antecedent VP is full. The example presumably fails because
there is no sufficiently salient relation R, a possible first argument of T
HE
, that
could combine with the meaning of this VP and give the desired interpretation.
Note that if a strong visual clue is offered, the example becomes better. If, for
example, I hand you the brimming garbage can and utter the last sentence of
(100), the sentence dramatically improves.
96
Paul Elbourne
3 Previous Literature
In this section I briefly compare the theory advocated in this article with some
other theories that try to cover some or all of the tricky cases dealt with here. I
will not attempt a detailed review of the literature on ellipsis, which would be a
mammoth undertaking.
12
3.1
Rooth 1992
Rooth’s influential paper proposes using entailment-like relations involving fo-
cus to characterize the relationship between elided phrase and antecedent. He
claims that VP-ellipsis is permitted only if two conditions are met: first, the lex-
ical content of the elided VP at LF must be the same as that of an antecedent
VP, modulo indices on pronouns and traces; and second, the elided VP must be
embedded in a constituent β such that there is a constituent α containing the
antecedent VP such that the ordinary semantic value of α is a member of the
set of focus alternatives generated by β. A variant of Rooth’s condition is to be
found in work by Merchant (2001).
Some of the details of Rooth’s theory are less than satisfactory, however. He
gives an analysis of binderless sloppy readings like (101) that crucially involves
the DP John scoping out of its containing DP. This already seem dubious, since
DP is normally an island. And we surely cannot extend this idea to (102) and
12
Two theories worthy of note that I do not deal with in the main text are the higher order
unification theory of Dalrymple, Shieber and Pereira (1991) and the discourse grammar theory
of Pr¨ust, Scha and van den Berg (1994). The former deals with binderless sloppy readings and
might be extended to deal with (62a), the sentence about Mary swimming the English Channel
and climbing Kilimanjaro; but it does not attempt to deal with ellipsis-containing antecedents
and the other split antecedent cases, and I see no way of extending it so that it would. Pr¨ust,
Scha and van den Berg (1994) also account for cases like (62a) but they do not attempt to deal
with the other split antecedent cases, the binderless sloppy readings or the ellipsis-containing
antecedents, and again I see no way of extending their theory to achieve better coverage.
The Semantics of Ellipsis
97
(103), where the relevant DP would have to scope out of a relative clause or an
if -clause.
(101)
John’s coach thinks he has a chance, and Bill’s coach does too.
(102)
The policeman who arrested John read him his rights, but the policeman
who arrested Bill didn’t.
(103)
If John has trouble at school, I’ll help him, but if Bill does, I won’t.
Rooth does not offer any analysis of ellipsis-containing antecedents or split an-
tecedents.
3.2
Fiengo and May 1994
Fiengo and May (1994) attempt to account for binderless sloppy readings (and
other data) by means of a complex system that exploits isomorphism of patterns
of indices in trees. I will not attempt to summarize this theory here.
13
Fiengo and May also analyze split antecedent cases such as (104).
(104)
I did everything that Mary did. Mary swam the English Channel, and
Mary climbed Kilimanjaro, and I did too.
About this example, they say the following (1994:195):
In this sentence, what is elided are occurrences of the VPs swim
the English Channel and climb Kilimanjaro (and an occurrence of
and). This is all we need to know to “recover” the ellipsis—that
13
We should note, however, that Rooth already in his 1992, page 18, had published the fol-
lowing counterexample to Fiengo and May’s theory:
(i) Yesterday John’s boss told him to shape up, and today Bill’s boss did.
(ii) Yesterday the guy John works for told him to shape up, and today Bill’s boss did.
Since John and Bill do not occupy isomorphic positions in their respective sentences in (ii), it
is unclear that Fiengo and May’s account correctly predicts the sloppy reading here.
Dostları ilə paylaş: |