What is a polarity context?

Polarity items are defined by their distributions – by their patterns of occurrence or non-occurrence in particular sorts of sentences and utterance contexts. But it is not obvious what it is that makes the loose conglomeration of contexts that affect polarity licensing a natural class. In English the contexts which license NPIs and inhibit PPIs include clauses and constituents not just in the scope of negation, but also in the scope of a question, the premise of a conditional, the standard of a comparison, or in the complements of certain intensional predicates (e.g. doubt, deny, sad, sorry, surprised, and amazed). Presumably all these contexts share some common property which explains their common behavior as “affective,” NPI-licensing contexts.

Broadly speaking, linguists since Klima have followed two basic strategies for explaining affectivity. The traditional assumption was that affective contexts are defined by their relation to negation (Klima 1964; Baker 1970; Linebarger 1980; Progovac 1994). The idea is implicit in the term negative polarity item itself, and in the loose but common use of negative contexts for NPI licensors generally (Buyssens 1959; van der Wouden 1997). Although many licensors are neither semantically nor grammatically negative, there is nonetheless a common intuition that “negation is the most central of them, while the others have various kinds of semantic or pragmatic connection with negation” (Huddleston & Pullum 2002: 834).