In lieu of an abstract, here is a brief excerpt of the content:

  • Indefinites in Verb Phrase Ellipsis
  • Bernhard Schwarz

The purpose of this squib is to show that verb phrase ellipsis data elucidate the semantics of indefinite noun phrases in ways not described previously. Specifically, ellipsis data are shown to provide counterexamples to the thesis (formulated in Fodor and Sag 1982) that all indefinites that appear to be nonlocallyscoped existential quantifiers actuallydenote individuals. Ellipsis data also reveal that these counterexamples are of two kinds. In some cases, the indefinite may be analyzed as nonquantificational, through reference to a function (as in the theoryproposed in Kratzer 1998). But in other cases, the indefinite resists such treatment and indeed appears to introduce nonlocal existential quantification. The squib therebyvindicates conclusions drawn in previous literature from related observations regarding the interaction of indefinites with quantifiers (Farkas 1981, Chierchia 2001, Schwarz 2001). It moreover examines potential evidence from ellipsis for the existence of individual-denoting indefinites, but argues that in this case the evidence is inconclusive.

1 Ambiguity and Scope

Fodor and Sag (1982) argue that indefinite noun phrases are ambiguous between quantificational and referential interpretations. A quantificational indefinite is construed as an existential (generalized) quantifier, whereas a referential indefinite is taken to denote an individual-whatever individual the speaker intends to refer to in the utterance situation. This proposal assimilates referential indefinites to deictic demonstrative noun phrases, which are also assigned their referents through the nonlinguistic context.

Fodor and Sag moreover propose that, just like the scope of other quantifiers, the scope of quantificational indefinites is local in that it is usually clause bound and can never escape from syntactic islands. They argue that any indefinite that may appear to be an existential quantifier with nonlocal scope is in fact to be analyzed as referential.

While it is uncontroversial that indefinites can be existential, there are questions about the need for referential interpretations. Fodor and Sag offer the ellipsis example in (1) to illustrate the referential use. They report (p. 376) that the first sentence can "quite easily be understood as ascribing to Sandya belief about a particular person that the speaker beat at chess."

(1) Sandy thinks that every student in our class plays chess better than a guy I beat this morning. Chris does, too.

It is moreover possible-in fact, natural-to understand the elliptical [End Page 344] second sentence as ascribing to Chris a belief about the very same person. I will adopt the terminology of Oku (2000), who refers to this way of understanding an elided verb phrase whose antecedent hosts an indefinite noun phrase as its definite reading.

The definite reading of the elided verb phrase in (1) finds a straightforward account in the assumption that a guy I beat this morning has a referential reading. Whatever person the speaker has in mind as the referent, the highest verb phrase of the first sentence in (1) will denote the property of thinking that every student plays chess better than that person. The elided verb phrase in the second sentence, being anaphoric to the highest verb phrase in the first, will then denote the same property, hence will ascribe to Chris a belief about that person as well.

Definite readings thus provide a potential argument for the existence of referential indefinites. As it stands, however, the argument is of course incomplete. For it remains to be seen whether definite readings could not just as well be derived in a different way.1

We will briefly return to the issue of definite readings in section 3. For the time being, let us accept the existence of referential indefinites as given and focus on the second claim made by Fodor and Sag: that quantificational indefinites are subject to island constraints on scope and that examples that suggest otherwise can be credited to referential readings.

2 Are There Existential Indefinites with Nonlocal Scope?

An example that could be analyzed in terms of nonlocal scope of an existential indefinite is given in (2).

(2) Tom ate the cookies some woman had brought.

If the indefinite some woman is construed as a narrow scope existential, the sentence entails that Tom ate all the cookies brought by women. While (2) may well be understood in this way...

pdf

Share