Predicting scalar diversity in a Bayesian model of pragmatic inference with lexical uncertainty

Team: Chao Sun (ZAS), Richard Breheny (University College London), Nicole Gotzner (SiGames, ZAS) and Anton Benz (SiGames, ZAS)

Recent experimental studies have found that different scalar expressions give rise to scalar inferences (SIs) at different rates (Doran et al., 2009, 2012; van Tiel et al., 2016). This phenomenon has become known as scalar diversity.

In this project, we aim to predict the variability of inferences rates across scalar expressions in a Bayesian model with lexical uncertainty. We adopt the framework, rational speech acts model with lexical uncertainty (RSA-LU), set out by Bergen et al. (2016). The objective of the first part of our project is to obtain data on the prior probability of the literal interpretation of each scalar term. We will then use our data to build a Bayesian model that makes qualitative predictions about the relative strength of the scalar inference in van Tiel et al.’s inference task. The objective of the second part of the project is to relate factors underlying negative strengthening to RSA-LU approaches to scalar diversity.