Learning to understand: Statistical learning and infant language development

Infants rapidly develop from being naïve listeners, who experience spoken language as a sea of sounds, to understanding their native language(s). How does this remarkable learning process unfold? One potentially useful source of information lies in the statistical patterns that characterize natural languages, which signal structures ranging from phonemes to words to grammatical structures. Over the past two decades, researchers have demonstrated that infants are sensitive to myriad statistical regularities in language input. Beyond merely tracking these patterns, how might infants use statistical regularities to support language development? In my presentation, I will explore the hypothesis that infants exploit statistical regularities in the service of efficiently processing information in their linguistic environments. To this end, I will discuss studies from my lab examining the role of statistical learning in infants’ uncertainty reduction behaviors (including predictive processing, error-based learning, and active sampling). By learning to efficiently encode language input, infants become increasingly able to process their native language(s). I’ll conclude with a discussion of atypical developmental trajectories considered through the lens of statistical/predictive language learning.

Speaker Bio: 

Dr. Saffran has been on the faculty of the University of Wisconsin-Madison since she received her Ph.D. in 1997 (University of Rochester). Her research program focuses on how infants learn, particularly in the domain of language. She is a highly engaged teacher and mentor to students at all levels. Dr. Saffran is the recipient of numerous awards for her scholarship and teaching, including, most recently, the Jeffrey Elman Prize for Scientific Achievement and Community Building from the Cognitive Science Society (inaugural recipient). She was elected to the American Academy of Arts and Sciences in 2015.

Zoom Link: 

 https://mit.zoom.us/j/95066563449

Please note: You will need an MIT certification to join