Types of Predicates and the Representation of Existential Readings

Abstract

The main theoretical proposal to be made in this paper is that existential force may have two sources.It may be contributed by either indefinite arguments or by certain predicates.I will argue against the current view that subsumes BNPs under the analysis of indefinites and assume instead that BNPs are property-denoting expressions (see also McNally (1995» , which as such cannot contribute existential force.The obvious problems regarding semantic composition raised by this hypothesis can be solved by defining a class of predicates that are able to apply to property-denoting arguments.I will argue that the crucial property of such predicates is that they allow (some of) their argument-variables to be bound by existential closure and correlatively translate as lambda-abstracts over the properties that restrict those argument-variables.This proposal allows us to derive the (correct part) of the Milsark-Carlson constraints on the existential reading of BNPs from type-matching constraints. The Representation of Indefinites and Bare PluralsMy analysis of BNPs is close to that of Farkas (1985) and Dobrovie-Sorin&Laca (1996), who argue that the existential and generic readings of BNPs cannot be given a unified account.I will assume that the two interpretations rely on property-and kind-denotation, respectively.All theorists agree that insofar as BNPs lack determiners, they are to be analyzed as predicates, i.e. as property-denoti ng « e,t>-type) expressions. 1 It is however currently assumed that this analysis cannot be correct for the BNPs that occupy A-positions, because it would give rise to type-mismatch between semantic type and distribution (A-positions can only be occupied by entity-denoting DPs or by DPs that denote generalized quantifiers, e and « e,t>,t>-type expressions, respectively).It is for this reason (which is however not always explicitely acknowledged) that most theorists (see Kratzer (1988, 1995), Diesing (1992), etc.) extend the Lewis-Kamp-Heim analysis of indefinites to BNPs: on a par with indefinites, BNPs are assumed to contribute variables.In what fo llows I will go against this trend: BNPs cannot contribute LF variables, but are to be analyzed as property-denoting (see also 'McNally (1995)).The type-mismatch problem can be solved by assuming that certain predicates may be represented in such a way that they may apply to property-denoting expressions (Section 2).This analysis of BNPs leads to a revision of the analysis of indef mites itself.Since (i) BNPs may occur in A-positions and nevertheless preserve property denotation, and (ii) indefinites at large may be property-denoting (see their occurrence in predicate position), we are led to conclude that indefinites at large may preserve property-denotation when they occur in A-positions.2But in contradistinction to existential BNPs, other indefinites may also contribute LF variables.Turning now to generic BNPs, I will fo llow Dobrovie-Sorin&Laca (1996) in assuming Carlson's (1977) hypothesis according to which they fu nction as proper names of kinds.Since I do not think that a Carlsonian analysis can be maintained for

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 56 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.