NB: In this talk I will present a research proposal that I developed together with Riccardo Fusaroli for a Marie Curie fellowship grant application. I will start by presenting my own previous computational modelling work and a review of the experimental literature that has led me to this research question. The rest of the talk will consist of an outline of the research proposal and experimental design, on which we would really appreciate your feedback.

Abstract:
How did humans learn to communicate successfully? Natural languages come with a significant amount of ambiguity and underdeterminacy: any given utterance can have an infinite set of potential intended meanings (Carston, 2002; Wasow et al., 2005). If I say “I’m tired”, this can mean anything from “let’s go home” to “I want to end this relationship”. Two mechanisms have been argued to help us overcome this challenge: perspective-taking, which allows us to take into account our interlocutor’s mental states (Clark & Murphy, 1982), and conversational repair, which helps us signal trouble when communication breaks down (Schegloff et al., 1977). However, neither of these mechanisms can be assumed to have been in place when language first emerged in our hominin ancestors. Computational modelling has shown that (i) perspective-taking can improve through the establishment of a conventional signalling system (Woensdregt et al., forthcoming), and (ii) conversational repair and systematic structure in language may depend on each other to work efficiently (Woensdregt & Dingemanse, 2020). The aim of the current project is to experimentally test these predictions by innovatively combining insights from computer simulations, iterated learning experimental techniques, and cognitive modelling.

Zoom: https://aarhusuniversity.zoom.us/j/5636106271

About the speaker

Marieke Woensdregt, Postdoc

Centre for Language Studies

Radboud University Nijmegen