Prediction in a changing world

Arielle Borovsky - Department of Speech, Language, and Hearing Sciences - Purdue University

Prediction in a changing world

Arielle Borovsky, Purdue University
Numerous language processing models emphasize the importance of listeners’ ability to predict upcoming information for efficient language comprehension and learning. Much of the evidence for these models is derived from studies of comprehension in well-known or familiar (i.e. predictable) contexts. However, speakers are pressed to prioritize novel information, suggesting that everyday conversation does not typically rehash redundant events. In developmental and learning contexts this problem may be compounded by the fact that listeners may still be learning about the language and the world. Therefore, these listeners may not have sufficient knowledge to generate predictions. In all of these circumstances, prediction might be counter-productive for comprehension.
I will discuss several studies that measure whether and how adult and child listeners deploy predictive mechanisms while learning about new events and words. Together, these studies paint a broader picture that fluent adult listeners become exquisitely sensitive and flexible to changing context and circumstances and can rapidly shift their predictions in the face of change. In children, these skills develop gradually over an extended period. I will argue that by incorporating developmental insights and learning paradigms into studies of linguistic prediction, we can develop more nuanced models of how and when prediction supports everyday communication and learning.

 

 

Arielle Borovsky: CV