More about PEECh
Given the situation below, why would the speaker choose one utterance over the other?
The project aims to understand, given exactly the same piece of evidence in a situation, what determines a speaker’s choice of how to qualify their statement: by expressing the degree of confidence they have (e.g. “must”) or by indicating the type of evidence available (e.g. “appear to”).
A speaker may say something about a proposition p (“John is at home”) without being fully committed to it. evidential and epistemic expression commonly serve to reduce the social responsibility for making such a statement: by ascribing to (the fallibility of) an external source or an estimation of likelihood. The project seeks to identify the pragmatic (discourse or interactional) factors driving the speaker’s decision in providing epistemic and evidential qualification.
To ensure empirical clarity, the project adopts controlled production and comprehension experiments, constraining the possible space of interpretation by systematically isolating key variables that are independently shown to inform crosslinguistic grammatical marking of evidentiality and epistemicity.
Apart from PEECh
I’m generally interested in seeking different precise ways to quantify interlocutors' presumptions about their beliefs, and to predict which utterances or actions they may prefer. So far, Rational Speech Act (RSA)-based modeling has been a good friend and a fruitful formalism for exploring this topic. Previously, my collaborators and I investigated a class of speaker-oriented items which signal prior-posterior differences during the interaction between the speaker and the listener. We were able to formulate their use under a Bayesian framework, in which the participants' knowledge states are represented probabilistically.