The traditional approach towards (negative) polarity items is to answer the question in which contexts NPIs are licensed. The inspiring approaches of Kadmon & Landman (1990, 1993) (K&L) and Krifka (1990, 1992, 1995) go a major step further: they also seek to answer the question of why these contexts license NPIs. To explain the appropriate use of polarity items in questions, however, we need to answer an even more challenging question: why is a NPI used in a particular utterance in the first place? Kadmon & Landman and Krifka go some way to answer this question as well in terms of an entailment‐based notion of strength, but I seek to give the question a somewhat ‘deeper’ explanation. Strength will be though of as ‘relevance’ or ‘utility’, which only in special cases reduces to entailment. In questions, the information theoretical notion of ‘entropy’ will play a crucial role: NPIs are used in a question to increase the average informativity of its answers. To account for the rhetorical effect of the use of some NPIs in questions, I propose a domain widening analysis of so‐called ‘even NPIs’.
Stored only when both papers are in the corpus — 240 citations exist upstream.