Unformatted text preview:

1G. ChierchiaU. of MilanRevised version- August 2000A Puzzle about Indefinites*1. Introduction.Long distance indefinites (i.e. indefinites whose scopal properties don't seem to obey the canonical constraints associated with quantified NPs) have been analyzed in recentproposals in terms of choice functions. There is, however, disagreement as to whether such functions are subject to quantificational closure (more specifically, existential closure) and if so, at which sites. On the one hand, Winter (1997) and Reinhart (1997) argue that choice functions are subject to existential closure freely, i.e. at any admissible scope site (including"intermediate" ones). On the other hand, Kratzer (1998) proposes that choice functions are left free and their interpretation is to be obtained through contextual clues; a closely related proposal can e found in Matthewson (1999), who argues on the basis of data from a Salish language that such functions are closed existentially but only at the topmost possible level. In this paper, I address this disagreement and point to a puzzle. First, I'll argue that a certain, familiar range of contexts (roughly, the downward entailing (DE) ones) require for their interpretation intermediate quantificational closure of choice functions. So, Reinhart and Winter's approaches appear to be right for such contexts. Second, I'll point to a number of other cases where the behavior of indefinites appears to be difficult to make sense of, if one has free quantificational closure over choice functions. So an approach like Kratzer's seems to be right for such cases. This situation is puzzling. One theory seems to be right fora set of cases (approximately, the DE contexts) the other seems to be right for other cases. The empirical generalization that emerges is quite interesting and, to my knowledge, it has gone unnoticed so far. The outcome of our discussion will be the proposal of a certain constraint on existential closure, constraint that seems to be formally simple and to have a broad empirical coverage. At present, however, I have no real explanation as to how such a constraint is to be derived.The present paper is structured as follows. In the remainder of this section, some relevant background will be provided. In section 2, we'll look at long distance indefinites in DE context and argue that without intermediate existential closure one cannot get their truthconditions right. In section 3, we'll first provide some new evidence in favor the idea that indefinites have hidden parameters and that existential closure is not freely available. Finally, we will discuss some possible ways in which the generalizations that emerge might be accounted for.2As discussed in much recent literature,1 the scope of indefinites is not subject to island constraints. There are a number of examples, by now fairly familiar, that show this:(1) a. Every linguist studied every conceivable solution that some problem might haveb. Everyone is convinced that if a friend of mine comes to the party, it will be a disaster.Sentence (1a) is a version of Abusch's "professor" example: every professor rewarded every student who read some book on his reading list. Sentence (1b) is a variant of examples discussed extensively by Reinhart and Winter. The indefinites some problem and a friend ofmine in (1) are inside islands. Yet they can be construed as having scope outside the islands that contain them. In particular, they can be construed as having scope either at the root (what traditionally has been called the "specific" or "referential" interpretation) or at some intermediate level within the scope of the matrix subject. The intermediate readings are those we are mostly interested in. They can be rendered in standard logical representations as follows.(2) a x[ linguist(x) y[ problem(y)  z [solution to (z,y)  studied (x,y) ]]]For every linguist x, there is a particular problem y (possibly a different one for eachlinguist) such that x studied every possible solution to y.2b. x [person(x)  y [friend of mine(y) convinced ( x, ^[ come to the party (y) it will be a disaster]) ]]3For every (relevant) person x, there is a certain friend of mine y (possibly, a different one for each person) such that x is convinced that if y comes to the party, it will be adisaster.1* The bulk of this paper was presented at the conference "Developments in Semantics" held in December 1997 at Institute for Advanced Studies at the Hebrew University in Jerusalem. I am indebted to that audienceand to the members of the Semantics Group hosted by the Institute in the Fall of 1997 for many helpful comments. I would like to thank, moreover, A. Bonomi, P. Casalegno, C. Cecchetto, O. Percus, and S. Zucchi for their comments. Insightful observations from K. Von Fintel, P. Schenkler and A. von Stechow have led to substantial revisions of the first draft of this paper. Errors that remain are mine. Much of the recent work on the scopal properties of indefinites developed as a reaction to Fodor and Sag (1982). See, e.g., King (1998), Ludlow and Neale (1991), Ruys (1992). A particularly influential paper is Abusch (1993).2 Formula (2a) can be made trivially true by choosing an unsolvable problem as value for y. Here and thoughout I assume that it is a general presuppositional requirement associated with quantifiers like every that their restriction be non empty (i;e. non trivial). In the case at hand, this means that olnly solvable problems are under consideration. The issue of existential requirements of strong quantifiers is an aspect of the projection problem for presuppositions.3 I use the symbol ‘’ to represent the Stalnaker/Lewis conditional.3It is also known that these readings (whose informal paraphrase is given under the respective formulas) can be favored by a variety of factors. One is the insertion of modifierslike particular or certain after some or a. Another is the presence of a pronoun in the nominal complement of the determiner (as in, e.g., "every linguist considered every conceivable solution that some problem that intrigued him or her had"). Yet another is a particular intonational contour. For example, the intermediate scope reading for some in (1a) is favored by a raising pitch on it (similar to the one that gives rise to "scope inversion"phenomena -- see, e.g., Krifka 1998 and references therein). But even in the absence of these factors, if the relevant


View Full Document

MIT 24 954 - A Puzzle about Indefinites

Download A Puzzle about Indefinites
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view A Puzzle about Indefinites and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view A Puzzle about Indefinites 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?