Experimental part with Lyn Tieu (Western Sydney University) and Emmanuel Chemla (LSCP, CNRS)

We argue that most of the typology of inferences in natural language (including implicatures, presuppositions, supplements, expressives and homogeneity inferences) can be replicated with pro- and post-speech gestures, i.e. gestures that fully replace or follow spoken words. In some cases, such as implicatures, this result is expected: if gestures introduce informativity scales, standard theories predict that they should trigger implicatures, and they do. In other cases, such as presuppositions, the result is unexpected for several theories, and highlights the need for a ‘triggering algorithm’ that can predict which part of a new informational content is treated as presuppositional. More generally, iconic gestures make it possible to create new words ‘on the fly’ while insuring that their informational content is immediately understood. This makes it possible to test instances of ‘zero-shot learning’ in semantics, and assess the productivity of rules by which new information content is divided among the various components of the inferential typology.

References

Schlenker, Philippe: to appear, Gestural Semantics: Replicating the typology of linguistic inferences with pro- and post-speech gestures.  Natural Language & Linguistic Theory.   https://ling.auf.net/lingbuzz/003485
Tieu, Lyn; Schlenker, Philippe; Chemla, Emmanuel: 2018, Linguistic inferences without words: Replicating the inferential typology with gestures. Manuscript.