Exhaustive and non-exhaustive variation with free choice and referential vagueness: Evidence from Greek, Catalan, and Spanish

This paper has two major goals. First, we want to critically assess the “universal free choice” (UFCA) analysis as it has been formulated in Menéndez-Benito (2010) for Spanish Free Choice Items (FCIs), while updating the dependent indefinite analysis of FCIs proposed originally in Giannakidou, 1997 and Giannakidou, 2001. We find the UFCA empirically inadequate for FCIs, failing to capture their correct distribution, and making wrong predictions about their interpretation. The dependent indefinite analysis that we defend here is found to be superior empirically and conceptually. Our second goal is to distinguish the Greek, Catalan and Spanish FCI from another type of anti-specific indefinite that we call referentially vague. The English equivalent is some-or-other. Unlike the FCI, the referentially vague indefinite requires non-exhaustive variation in the value-drawing domain. In Greek, we find a referentially vague indefinite that is also a Negative Polarity Item (NPI)—and we discuss briefly a similar item in Korean. Overall, our discussion suggests that we gain a better understanding of anti-specificity phenomena such as free choice and referential vagueness if we treat them as manifestations of referential deficiency or low referentiality (as suggested in Giannakidou's work, see also Partee, 2008), and it is unnecessary to appeal to propositional alternatives.

► The dependent indefinite analysis is empirically and conceptually superior to the universal free choice analysis.
► Free Choice Items are to be distinguished from referentially vague indefinites.
► Referentially vague indefinites require non-exhaustive variation in the value-drawing domain.
► Anti-specificity phenomena must be understood as manifestations of referential deficiency or low referentiality.
► The licensing of free choice items in the syntax-semantics must be distinguished from the free choice effect, which is pragmatic.