Abstract

Dermot Lynott (University of Lancaster): What can you do with 1 trillion words combined with embodied experience?

Previous research has shown that people use both embodied perceptual simulations and linguistic distributional statistics during conceptual processing. Linguistic information is seen as especially useful for shallow tasks and rapid responding, while embodied perceptual knowledge is more useful for tasks that require deeper processing over longer
timescales. In this talk, I will present some examples that first show a generalised role for embodied semantics in classic language processing and memory tasks. Then, using co-occurrence data from a large-scale corpus, I provide evidence for distributional linguistic information in deeper as well as shallower conceptual tasks, and also demonstrate a novel role for distributional linguistic information in predicting implicit attitudes. These findings support theories
that argue for complementary roles of the linguistic and perceptual simulation systems in conceptual processing.