Is a linguistic model needed to build abstract event representations?

A central question in cognitive development is whether language simply expresses pre-existing event concepts or plays a critical role in their construction and use. Recent findings from studies with infants, preschoolers and adults have raised the possibility that generic two-place relations (e.g., cats push rabbits) can only be represented when people have access to the transitive sentences that express them. This suggests that these concepts could be constructed as we acquire a pre-existing, external language that expresses them. To explore this hypothesis, we tested whether adult homesigners-individuals without exposure to a pre-existing language-could construct such concepts in a nonverbal imitation task. Participants viewed three instances of a given generic event (with either one or two participants), then they were given new exemplars of the same kinds (e.g., new rabbit and cat) and prompted to act. Their performance was compared to English-speaking five-year-olds. Both groups performed well in the critical two-participant condition, consistently mapping figurines of the right kind to each role. There were no group or event-type differences. Thus, homesigners have the representational resources needed to support role binding. These findings demonstrate that abstract representations of generic two-place relations can emerge without exposure to a language that models these constructions or a set of shared linguistic conventions.