You are viewing the SNL 2018 Archive Website. For the latest information, see the Current Website.


Poster B48, Thursday, August 16, 3:05 – 4:50 pm, Room 2000AB

Tracking the time course of statistical learning in pre-lingual infants: online evidence from neural entrainment

Laura Batterink1,2, Dawoon Choi3, Alexis Black3, Ken Paller1, Janet Werker3;1Northwestern University, 2University of Western Ontario, 3University of British Columbia

Both linguistic processing and implicit learning can be tracked via endogenous brain rhythms that synchronize to exogenous stimuli (Ding et al., 2016; Batterink and Paller, 2017). Statistical learning (SL) refers to the ability to detect structure in the environment and is one of the mechanisms by which infants may learn to segment words from a continuous stream of speech sounds (Saffran, Aslin, & Newport, 1996). Although statistical learning in infants has been demonstrated behaviourally through offline looking-time measures, these measures are influenced by memory retrieval processes and other peripheral factors, and are unable to track learning directly. To begin to bridge this gap, we used electroencephalography (EEG) to track SL in 6-month-old infants. Twenty-four infants (12 female) were exposed to a continuous stream of four repeating tri-syllabic nonsense words. SL was assessed by comparing neural entrainment at the frequency of the hidden embedded words relative to that of individual syllables. Importantly, this EEG neural entrainment measure is obtained during rather than after exposure to structured input, reveals the time course of learning, and tracks learning directly, without requiring offline tests that are additionally influenced by subsequent memory retrieval processes. Consistent with prior work in adults (Batterink & Paller, 2017), infants’ neural entrainment increased at the word level and decreased at the syllabic level as a function of increasing exposure, indicating that entrainment is a viable means of tracking SL in infants. We further characterized and directly compared infants’ and adults’ individual learning curves, as reflected by the progression of neural entrainment over the exposure period. Interestingly, neural entrainment to words reached a plateau in infants more quickly than in adults. Whereas infants and adults showed similar learning progressions during early phases of the exposure period (first ~90 s), only adults continued to exhibit further increases in word-level entrainment with additional exposure. These results suggest a briefer sampling period of environmental statistics in infant learners, which may arise from constraints on attention. This briefer environmental sampling in infants may represent an adaptive mechanism that prevents the overspecification of linguistic representations. More generally, monitoring EEG entrainment in infants provides a promising new avenue to assess the online course of SL, and may provide new insights into individual or population-level (e.g., developmental) differences in learning.

Topic Area: Language Development

Back