Probabilistic pragmatics, rational analysis and data-driven inference

Michael Franke – Philosophische Fakultät – Seminar für Sprachwissenschaft – Universität Tübingen

Probabilistic modeling of experimental data of pragmatic language use and interpretation has received increased attention in the recent years. I present key ideas behind a class of like-minded approaches, subsumed here as ’probabilistic pragmatics’, and contrast it with, in particular, more traditional approaches to pragmatic phenomena. I present a number of case studies in order to demonstrate how probabilistic modeling of a pragmatic core component can seamlessly be integrated into a custom-made Bayesian data-generating model, tailored specifically to the experimental data and the theoretical questions of interest. Time permitting, I speculate about steps towards deriving processing-related predictions from state-of-the-art probabilistic pragmatics models.

If you would like to meet with the speaker, please contact Vera Demberg.

Michael Franke’s personal website.