This exploration of the syntax-semantics interface is concerned with negative indefinites like English ‘nobody’, ‘nothing’, etc. and their counterparts in other languages. A cross-linguistically unified analysis is proposed and applied to several languages. While negative indefinites are standardly assumed to be semantically negative quantifiers, this work argues for a different analysis. It is motivated by three phenomena, which negative indefinites give rise to in different languages and which are unexpected under the negative quantifier analysis. The first, negative concord, has been widely ... More

This exploration of the syntax-semantics interface is concerned with negative indefinites like English ‘nobody’, ‘nothing’, etc. and their counterparts in other languages. A cross-linguistically unified analysis is proposed and applied to several languages. While negative indefinites are standardly assumed to be semantically negative quantifiers, this work argues for a different analysis. It is motivated by three phenomena, which negative indefinites give rise to in different languages and which are unexpected under the negative quantifier analysis. The first, negative concord, has been widely discussed in both semantic and syntactic literature. The fact that in many languages negative indefinites can co-occur with other seemingly negative elements without contributing a negation to the semantics motivates the assumption that these expressions are not inherently negative. Following recent work on negative concord, an analysis is elaborated that is based on the assumption that negative indefinites are semantically non-negative and must be licensed by a — possibly covert — negation. This analysis explains the behaviour of negative indefinites in a number of languages. In a next step, this analysis is extended to languages that do not exhibit negative concord. Motivation for this comes from the fact that even in non-negative concord languages, the negative quantifier analysis cannot account for the semantics of negative indefinites. Crucial evidence comes from the existence of split readings, in which another operator takes scope in between the negative and the indefinite meaning component. Moreover, in many languages the distribution of negative indefinites is subject to syntactic restrictions. It is shown how this follows from the proposed analysis and independently motivated syntactic properties.