Michael Franke

The pragmatics of communicating causal information

Knowledge of causal processes is vital for all aspects of our lives. Yet, while strong evidence for causal relations requires systematic intervention, much of our causal knowledge is acquired not from individual experience or direct experimentation, but indirectly from cultural transmission via language. From this point of view, it is very curious that relatively little work in (formal/experimental) pragmatics has addressed the problem of communicating causal information. This talk therefore introduces a probabilistic model of communicating causal information, in which speakers choose utterances to inform listeners about causal facts, and in which listeners reason about alternative causal models based on the usual conversational assumptions of informative and relevant information exchange. The model shows how listeners can be justified to infer causal information reliably from expressions, like indicative conditionals or statements like “A is associated with B”, which arguably do not encode this information in their semantic meaning. Building critically but constructively on prior work by Gershman & Ullman 2023), we present novel experimental evidence that causal enrichments in the interpretation of “A is associated with B” are indeed a pragmatic inference.

Michael is affiliated with University of Tübingen, Germany, see his website.