On Being Certain that Presuppositions don’t Project Universally

Abstract

In this paper, we present new experimental evidence that presupposition projection out of quantified environments can be modeled in a framework that uses a trivalent logic like Strong Kleene. Assuming a quantificational analysis, our experiment focuses on projection out of attitude reports, which are understudied from the empirical point of view. Theories that assume generalized universal projection, which are viewed as standard, cannot capture our data, which are instead compatible with the idea that presupposition failures can take part in the meaning computation. In the data analysis, we show how modeling standard deviations using a distributional model allows for new insights by enabling the detection of the squeamishness characteristic of presupposition failure in participants’ response behavior.