The Role of Prior Beliefs in The Rational Speech Act Model of Pragmatics: Exhaustivity as a Case Study
Résumé
This paper examines the interaction between prior beliefs and pragmatic inferences, focusing on exhaustivity effects. We present three experiments that tests how prior beliefs influence both interpretation and production of language, and compare the results with the predictions of the Rational Speech Act model, a Bayesian model of linguistic interpretation. We find that prior beliefs about conditional probabilities have no affect on language production, but do affect interpretation, producing anti-exhaustivity effects. We find that the RSA model achieves a relatively good fit both for the human production and interpretation data, but only for highly-implausible utterance costs.
Origine : Fichiers éditeurs autorisés sur une archive ouverte