|Bibliographic information:||Full paper in pdf|
|Experiment 1:||Experiment 2:||Experiment 3:|
|Result files in csv format||Result files in csv format||Result files in csv format|
|R script for the main analyses||R script for the main analyses||R script for the main analyses|
Abstract: Numerous linguistic phenomena have been described as giving rise to truth value gaps: presuppositions, homogeneity, vagueness, as well as various sorts of implicatures. The sim- ilarities and differences between these phenomena are most traditionally studied through their projection properties, investigating whether these various sources of gap give rise to the same behavior under negation, modals, in the restrictors and scopes of various quan- tifiers, etc. We propose to add a new method in the linguist toolbox to address the same comparative question, asking how probabilities are assigned to sentences under the risk of a truth value gap. This question is directly imported from the classical issue in philosophy of language concerned with deciding what the probability of a conditional sentence is, and we also import experimental methods which have been used to illuminate this issue. In three experiments, we show that homogeneity and vagueness consistently behave alike by this measure, while presuppositions behave differently, with surprising differences within the class of presupposition triggers, however.
Keywords: probabilities; presupposition; homogeneity; vagueness; scalar implicatures