Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Two Streams, One Brain: Cortical Tracking of Statistical Learning in Monolingual and Bilingual Populations
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Konstantina Zacharaki1,2, Bianca Franzoia1, Ruth de Diego-Balaguer1,2,3; 1Department of Cognition, Development and Educational Psychology, Institute of Neurosciences, University of Barcelona, 2Cognition and Brain Plasticity Unit, Institut de Recerca Biomèdica de Bellvitge, 3Institució Catalana de Recerca i Estudis Avançats, Barcelona, Spain
Humans are adept at extracting statistical regularities from speech, a mechanism known as statistical learning (SL) that is critical for language acquisition (Saffran et al., 1996). While word segmentation research on SL typically tests monolingual participants extracting co-occurring syllables from a single artificial speech stream, real-world environments are increasingly multilingual. To address this gap and better approximate naturalistic language exposure, our ongoing study investigates word segmentation in monolingual and bilingual 8-month-old infants and adults exposed to two sequential artificial speech streams. Previous research on bilingual SL has primarily relied on offline behavioral methods with infants (e.g., Antovich & Graf Estes, 2018) and adults (e.g., Wang & Saffran, 2014). To capture the continuous dynamics of learning as it unfolds, we use EEG to record online cortical tracking of word and syllable frequencies during exposure to two artificial languages (Batterink & Paller, 2017; Choi et al., 2020). Learning will be measured as cortical tracking at the word rate versus at the syllable rate. Adults’ and infants’ language profiles will be assessed using language questionnaires. We define bilingual adults as individuals who were exposed to Catalan and Spanish during the first years of life, and bilingual infants as those with more than 20% exposure to a second language. The two artificial languages have been created using Catalan and Spanish corpora to resemble two distinct but phonologically similar languages. Critically, our experimental design incorporates a partial phonetic overlap between the two streams. Therefore, if participants simply calculate transitional probabilities (TPs) across the combined streams without segregating the two distinct input sources, they will fail to demonstrate SL, evidenced by a lack of emergent cortical tracking at the word rate. At test, we will use age-appropriate behavioral measures. Infants’ reactions will be measured using a gaze-contingent central fixation paradigm and adults will complete familiarity rating and syllable detection tasks. To explore potential individual differences in the adult sample, we will also collect measures of inhibition (flanker task) and working memory (digit span). Our goal is to collect data from 24 monolingual and 24 bilingual participants in each age group. As data collection is currently ongoing, we present this work in the Sandbox Series to discuss our motivation, methodological design, and the interpretation of our planned analyses. Based on our design, we hypothesize that monolinguals will show cortical tracking at the word level only to the first stream, showing limited or no learning of the second language at both the behavioral and neural levels. We expect bilinguals, who are accustomed to managing multiple phonetic systems, to successfully extract words from both streams. Furthermore, we anticipate no significant differences in the overall cortical tracking trajectory patterns between adults and infants, pointing toward an SL mechanism that operates similarly across developmental stages. Interpreting these potential outcomes will provide valuable insights into how early and ongoing language experience shapes the neurobiology of continuous speech segmentation in multi-stream environments.
Topic Areas: Multilingualism, Language Development/Acquisition