[PS-1.23]The neural basis of a frequency-based learning mechanism: an fNIRS study with 7-month-old infants

de la Cruz-Pavía, I. 1, 2 , Werker, J. F. 2 & Gervain, J. 1

1 CNRS - Université Paris Descartes
2 University of British Columbia

The high frequency of occurrence of functors (the,of) in natural languages and their order with respect to content words (turtle,walk) has been proposed to potentially enable infants to segment the input into phrases and discover basic word order, as their relative order as found in the language input correlates with the basic word order of a language. Indeed, by 8 months of age, infants segment a structurally ambiguous artificial language according to the relative order of frequent (functors) and infrequent elements (content words) in their native languages.
We examine the neural basis of this frequency-based learning mechanism in an artificial language learning study with English- & French-learning 7-month-olds using near-infrared spectroscopy (NIRS). Studies using NIRS have shown that frontal and temporal regions respond differentially to the presence of different types of regularities. We familiarize infants with an artificial language consisting of a strict alternation of frequent and infrequent elements, and then present them with blocks of either frequent-initial or frequent-final items. Preliminary analysis of six participants reveals canonically shaped hemodynamic responses for both conditions, but no differences in activation between conditions so far. As in previous literature, this response is strongest in the temporal channels and present in both hemispheres.