Search Abstracts | Symposia | Slide Sessions | Poster Sessions
The language mosaic: MRI evidence from task-independent measures in monolinguals and bilinguals
Poster Session B, Friday, October 25, 10:00 - 11:30 am, Great Hall 3 and 4
Xuanyi (Jessica) Chen1, Esti Blanco-Elorrieta; 1New York University
In recent years, there has been active consideration that bilingualism may produce cognitive and neural benefits (Bialystok, 2017) as indexed by neural adaptations in several anatomical regions (e.g., supramarginal gyrus, prefrontal regions, white matter tracts; Pliatsikas, 2020). However, these results have been marked by inconsistency across experiments, suggesting that there may be qualitative differences between monolingual and bilingual individuals that are grounded in biological underpinnings, but offering little clarity to what those may be. The reasons for these inconsistencies may be two-fold. First, researchers have often relied on a simplified binary framework and made categorical distinctions between monolingual and bilinguals instead of establishing a bilingualism spectrum and assessing adaptations that result from the degree of bilingualism. Second, researchers have mostly focused on a handful of regions and reported the one region where groups vary in each project, preventing us from obtaining a complete, whole-brain account of how neural machinery adapts as a function of acquiring two languages. Here, we evaluate the impact of bilingualism on whole-brain anatomy by combining neuroscience and machine learning to create a comprehensive mosaic that reflects where and when neuroplasticity associated with multilingualism develops. We acquired structural, resting-state and diffusion-weighted images for 25 English monolinguals, and 25 Mandarin-English bilinguals. To overcome the simplified binary framework, we defined bilingualism qualitatively (categorical: mono vs bilingual) and quantitatively (continuous: by calculating a compound score for the degree of multilingualism). Specifically, we constructed a theoretical-driven multilingual scale that captured the ‘multilingual-ness’ of each subject accounting for their AoA, ability rating, and the total number of languages they know. From the MRI data, we obtained 1) gray matter, white matter, and subcortical ROI volume in 116 regions with voxel-based morphometry and 2) 168 ROIs with surface-based morphometry, 3) cortical thickness in 68 gray matter regions; 4) fractional anisotropy in 90 ROIs, and 5) resting-state connectivity between 164 ROIs. For each measure and ROI, we tested whether it showed a categorical or continuous effect of bilingualism using linear regression, corrected and uncorrected for total brain volume. To capture potential structural differences distributed across brain areas (i.e., to obtain a multi-dimensional ‘mosaic’ of anatomical profiles), we used supervised and unsupervised machine learning (i.e., K-means clustering, hierarchical clustering, and classification analyses with SVM) for each of the five metrics mentioned above. The results showed no consistent differences in any of the measures and ROIs after FDR correction. We found a significant difference in cortical volume of right precuneus only when the total brain volume was entered as a covariate. Additionally, our machine learning algorithms did not identify distinct anatomical patterns for monolinguals and bilinguals except for our SVM for cortical volume and thickness in surface-based analyses. However, follow-up analyses with simulated data showed that the SVM relied on a single varying dimension instead of reflecting an overall distinct anatomical profile. In all, the exhaustive analyses of structural differences show an overarching similarity between monolingual and bilingual brains, whether bilingualism is defined qualitatively and quantitatively, offering no evidence for multilingualism-induced neuroplasticity.
Topic Areas: Multilingualism,