Negative Polarity

Abstract

Negative polarity items (NPIs), for example the indefinites any , kanenas (in Greek), quoi que ce soit (in French), minimizers (e.g., lift a finger ), the adverb yet , and so forth have a limited distribution; they appear to stand in a licensing relationship to other expressions, such as negation, in the same sentence. This chapter offers an overview of some of the empirical challenges and theoretical issues raised by NPIs. It retraces a line of research that started out as an investigation of the right “licensing condition(s)” of NPIs: it sought to characterize licensing operators and used some semantic property (affectivity, downward‐entailingness and variants thereof, or non‐veridicality) to do so. It gradually shifted to exploring the very sources of polarity sensitivity: current theories no longer view NPIs as being in “need” of licensing, and instead hold that their contribution to meaning leads to semantic or pragmatic deviance in certain environments, from which they are thus barred. We first present the so‐called Fauconnier–Ladusaw approach to NPI‐licensing, and its licensing condition based on the notion of downward‐entailingness. This condition proves to be both too strong and too weak: it is too strong because NPIs can be licensed in the absence of a downward‐entailing operator, and it is too weak, in view of intervention effects caused by certain scope‐taking elements. We explore two ways of weakening the original condition: replacing downward‐entailingness with another semantic property, non‐veridicality; and Strawson downward‐entailingness. The upshot of this discussion is that NPIs are not licensed by operators , but rather need to fit into certain environments . Finally, we show how the restrictions on the distribution of NPIs (specifically indefinites) can be explained by the interaction between certain features of the meaning of NPIs and semantic properties of their environment.