Back

NEUROLINGUA: A Neuroimaging Database Tailored to Unravel the Complexity of Multilingual Comprehension

Quinones, I.; Carrion-Castillo, A.; Diez-Zabala, I.; de Frutos-Sagastuy, L.; Costello, B.; Carcedo, D.; Manso-Ortega, L.; Slivka, M.; Sanchez, A.; Schüller, A.; Caballero Gaudes, C.; Paz Alonso, P. M.; Carreiras, M.

2025-06-06 neuroscience
10.1101/2025.06.06.658050 bioRxiv
Show abstract

The neural mechanisms underlying language processing involve a well-defined brain network, including mainly left perisylvian areas. Yet, the extent of individual variability remains largely unexplored, particularly in bilingual and multilingual contexts. Differences in linguistic profiles (e.g., age of acquisition, exposure, proficiency) provide an opportunity to assess how network topology is shaped by sociolinguistic factors. To address this, we developed NEUROLINGUA, a comprehensive database of functional and structural MRI data, enriched with sociodemographic, sociolinguistic, and behavioral information. It includes 725 healthy individuals aged 18-82 immersed in a Basque-Spanish multilingual environment, ranging from near-monolinguals to highly proficient multilinguals. Participants completed a functional MRI language localizer with both auditory and visual comprehension tasks, enabling cross-modal comparisons, as well as sentences involving arithmetic problem-solving. Exploratory analyses confirmed associations between structural MRI, sociodemographic, and cognitive measures. We demonstrate that NEUROLINGUAs functional MRI data localize the language comprehension network and thus capture linguistic profile effects. This integrative dataset offers an unparalleled resource to investigate factors influencing language network adaptability and variability in diverse sociolinguistic contexts.

Matching journals

The top 2 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.