Linguistic statistical learning abilities in bilingual and at-risk population

Project Information

LingStats

Grant agreement ID: 101108884

DOI

10.3030/101108884

Project closed

EC signature date 2 May 2023

Start date 15 October 2023

End date 14 October 2025

Funded under

• Marie Skłodowska-Curie Actions (MSCA)

Total cost

No data

EU contribution

€ 165 312,96

Coordinated by

UNIVERSIDAD POMPEU FABRA
Spain

Periodic Reporting for period 1 - LingStats (Linguistic statistical learning abilities in bilingual and at-risk population)

Reporting period: 2023-10-15 to 2025-10-14

Summary of the context and overall objectives of the project

The speech signal is a continuous stream, where, contrary to written language, words are not separated. Yet, despite such a complex signal, by 6 months, infants can already recognize some words and by the end of their first year they can recognize a sizable amount (30 words on average). One mechanism that has been proposed they can rely upon is the so-called Statistical Learning (SL). In short, infants track regularities present in the speech signal and take advantage of such regularities to segment it (divide fluent speech stream into units that correspond to the words of the language). For instance, young infants can track frequency of occurrence of transitions between syllables (transitional probabilities): the probability of two syllables following each other is higher within a word than between words. There is a plethora of studies showing that infants are indeed sensitive to different types of distributional information at an early age.
The present project aims to investigate the linguistic SL abilities during the first two years of life and how they relate to later language development in different infant populations (monolingual infants, bilingual infants, and pre-term infants). Therefore, this project investigates the specificities underlying the acquisition of multiple phonological systems in the case of infants growing up in a specific bilingual environment (in which the phonological systems overlap, Catalan/Spanish). In addition, this project examines SL abilities in preterm infants (if birth occurs before 37 gestational weeks). Evidence has shown worse cognitive and behavioral outcomes in preterm children and adolescents and a higher risk for developmental delay compared to their full-term peers. These infants might not use statistical information as effectively as their peers and are therefore a crucial population to investigate whether SL abilities might be predictive of later language outcomes.

The research goals (RO) of this project are the following:
Objective 1. To compare the role of linguistic statistical learning abilities in early language development
and their potential as a predictor for later language development in different infant populations (monolingual, bilingual, and preterm)
Objective 2. To identify early neural and cognitive risk markers for developmental disorders and explore how/whether these correlate to statistical learning abilities and other linguistic abilities.
Objective 3. To develop appropriate experimental designs and data analysis tools that will be used through pupillometry measures and that can be used cross-linguistically and with young infants

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

- A new paradigm to test statistical learning with pupillometry was designed, together with an analysis pipeline.
- A total of 90 adults (Spanish, Catalan and Spanish-Catalan bilinguals) were first tested in a statistical learning task to test the paradigm and methodology, as well as stimuli. Pupil dilation and keyboard responses indicated that the three groups perform above chance in the non-word condition, showing that they can distinguish between words and non-words. However, differences arise between the groups in the part-word condition. Further analyses exploring whether differences Catalan-dominant and Spanish-dominant participants can be observed are on-going.
- A total of 50 infants (Catalan-learning, Spanish-learning and Spanish-Catalan-learning infants) were tested so far at the ages of 8 months; and we followed up 20 of them so far. We also obtained their receptive and expressive vocabulary via the BVQ (García-Castro et al., 2023) at 15 months. Because of the longitudinal nature of the study, the data collection is still on-going, but a preliminary analysis showed differences in pupil dilation responses between the bilingual and the Catalan-learning infants.
- Additionally, a small group of 10 infants born prematurely have been tested at the age of 8 months. The data collection is still on-going.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

The project advances the state of the art in the study of bilingual language acquisition and statistical learning by combining behavioral and physiological measures (pupil dilation) in both adult and infant populations, within a longitudinal and cross-linguistic framework. The longitudinal design linking early learning mechanisms (at 8 and 15 months) with later vocabulary development (at 15 months) represents a novel contribution to understanding how early monolingual and bilingual experience impacts language trajectories.
Once the data collection and analyses are finished, the results of this project have the potential to inform early diagnostic and educational strategies for bilingual and premature infants, supporting evidence-based approaches to early intervention and language development assessment.
The methodology used in the project, pupillometry, has been shown to be sensitive in evaluating language development and could therefore be a highly valuable tool for further experimental research and developmental assessments.

Word statistical segmentation