Title: The Changing Role of Infant Statistical Learning Across the First Two Years
Over the last two decades, the discovery that human infants readily identify patterns among cooccurring elements in a stream of sights or sounds has profoundly reshaped our estimation of infants’ learning capacities. Nowhere has this discovery been more influential than in the study of early speech segmentation. In much of our work, my students and I focus not only on how infants track statistical regularities to find word-like units in continuous speech, but also how they represent, remember, and interpret these newly extracted word forms during speech segmentation and word learning tasks across the first two years of life. In this talk, I will present data that suggests that at 8 months, although infant memory for words newly segmented from fluent speech is initially fragile (Karaman & Hay, 2018), infants’ representations of these newly segmented word-like units share common characteristics with more familiar words to include
information that is most relevant for word meaning (Parvanezadeh Esfahani & Hay, under review). By 17 months, these statistical learning processes feed into subsequent word learning such that labels with stronger internal coherence are more readily mapped to meaning than those with weaker coherence (Pelucchi, Hay, Graf Estes, & Saffran, 2011), suggesting a facilitative of statistics at this age. However, as infants approach their second birthday, we have found that they no longer appear to require labels to have strong statistics to map them to meaning (Shoaib, Wang, Hay, & Lany, Cognitive Science, 2018). Instead, statistics appear to play an important role in language learning by inhibiting the mapping of labels with weak statistics, especially after a delay (Karaman, Lany, & Hay, under review). Our findings bolster the notion that statistical learning plays an integral role in early language development, but in a more nuanced way than earlier work has suggested.
Dr. Hay is an Associate Professor at the University of Tennessee, Knoxville.
NACS Seminars are free and open to the public.