Authors

Mojmír Dočekal¹;  ¹ Masaryk University, Czech Republic

Discussion

Recent research on Negative Polarity Items (NPIs) and neg-words (two critical sets of polarity-dependent expressions) followed the experimental turn, and many valuable observations were made. But still, little is known about their synchronic variation (for neg-words, some work was done in syntax: @burnett2018structural, e.g.).

In this talk, I add to this research new experimental evidence from the Czech. Both NPIs and neg-words appear in Czech, but since Czech (like all other Slavic languages) is a strict negative concord language, the distribution of strong NPIs and neg-words overlaps considerably. Nevertheless, some constructions distinguish between strong NPIs and neg-words even in strict negative concord languages, Neg-Raising being one of them.

The core of my talk will be reporting the results of two experiments targeting Czech strong NPIs and neg-words. Both expressions were tested for acceptability in Neg-Raising clauses, contexts with manipulated likelihood/noteworthiness scales, equatives, fragment answers, etc. The experimental results show that both kinds of polarity-dependent expressions are differentiated semantically and pragmatically (strong NPIs are preferred to neg-words if modifying expressions on low end-points of a scale; neg-words win over strong NPIs if they modify high end-point of a scale; in equatives, neg-words are much more acceptable than strong NPIs, etc.) But, most notably, there are (by the speaker) negative correlations between the acceptability of strong NPIs in one context and their rejection in another.

I will discuss various demographic factors from one of the experiments that were shown before to contribute to the neg-words variation but which seem to have a lesser impact in Czech than in previous work on negative concord in Romance languages.

Next, I will discuss the experiments' consequences on our current neg-words theories. I propose that the speaker variation is most easily handled by semantic theories of negative-concord (original formulation: @ovalle2004double, modern reincarnation: @kuhn2022dynamics) where there is a postulated not at-issue meaning part of neg-words semantics (which can be formalized as a post-supposition requirement for no discourse referents of neg-words). This theoretical stance allows us to analyze neg-words close to strong NPIs, which, in theories like @crnic2011getting, must be licensed by association with covert *even* triggering a likelihood presupposition. Therefore, the speaker variation and other problematic data patterns can be elegantly handled as switching presuppositions. The semantic theories of neg-words also give more promising answers (compared to the standard syntactic theory of negative concord; see @zeijlstra2004sentential a.o.)  to some old puzzles concerning the acceptability of neg-words in equatives, elliptical answers, etc.