Enhanced efficiency in the bilingual brain through the inter-hemispheric cortico-cerebellar pathway in early second language acquisition

The study investigates how bilingual experience, and particularly the age of acquisition (AoA) of a second language (L2), shapes the organization of functional brain networks. Prior work shows bilingualism can positively impact working memory, cognitive control, speech-in-noise perception, attention, cognitive aging, and clinical outcomes, and that bilingualism affects brain structure and function. Moving beyond localizationist models, contemporary frameworks emphasize distributed network organization and the roles of integration and segregation. The authors hypothesized that early L2 learning would enhance whole-brain global efficiency and potentially modularity, and examined how cortico-cerebellar systems are impacted by L2 acquisition at different ages. Using whole-brain connectomes derived from resting-state fMRI, they tested whether learning two languages (early or simultaneous vs. late) compared to one language changes global integration and segregation and the interactions between cerebellar and cortical systems.

Functional neuroimaging shows language tasks engage right posterolateral cerebellum and left temporal/frontal regions, implicating cortico-cerebellar circuits. Bilingual processing extends beyond classic left-lateralized language areas to fronto-parietal systems and cerebellum, with roles in cognitive control and grammar processing. Network neuroscience demonstrates the brain’s modular (segregated) and integrative properties, with global efficiency linked to cognition and language performance. Prior bilingual studies reported higher connectivity in inferior frontal gyrus and fronto-parietal-temporal regions, AoA-related inter-hemispheric frontal connectivity, and proficiency-related changes in attention networks. However, whole-brain network effects beyond language/control networks are underexplored, and previous work often omitted cerebellum. The present study addresses these gaps by including cortical, subcortical, and cerebellar regions, focusing on how AoA influences integration and cortico-cerebellar interactions.

Sample: Resting-state fMRI from 151 right-handed, healthy participants (French/English context) scanned at the Montreal Neurological Institute on the same scanner and parameters. Groups based on self-reported L2 AoA: simultaneous bilinguals (AoA from birth), early bilinguals (AoA ≤ 5 years), late bilinguals (AoA > 5 years), and monolinguals (limited L2 exposure). Exclusions: language/hearing/vision impairments, TBI, medical/neurological disorders, MRI incompatibility. Ethics: Approved by MNI REB; informed consent obtained. MRI acquisition: 3T Siemens TrioTim, 32-channel coil. Resting-state EPI: 132 volumes, TR/TE=2260/30 ms, flip angle=90°, matrix=64×64, FoV=224 mm, 38 slices, thickness=3.5 mm, duration ~5:04 min; eyes open fixation. T1 MPRAGE: TR/TE=2300/2.98 ms, flip angle=9°, matrix=256×256, FoV=256 mm, slice thickness=1 mm. Preprocessing: CONN v20b (SPM12 routines). Steps: realignment/unwarping, outlier detection (ART; intensity >3 SD, motion >0.5 mm), segmentation/normalization to MNI, 6 mm FWHM smoothing. Nuisance regression: aCompCor components (WM/CSF), motion parameters, outlier scans. Parcellation and networks: 374 ROIs (333 cortical from Gordon et al., plus 15 subcortical and 26 cerebellar from Harvard-Oxford). Organized into 14 modules (12 cortical networks + subcortical + cerebellum). Edges: Pearson correlations of ROI-averaged time series, Fisher r-to-z transformed; negative edges discarded; weighted analyses. Network measures: Global efficiency (E) computed as average nodal efficiency (based on inverse shortest paths via Dijkstra’s algorithm) across nodes; modularity (Q) via Newman’s algorithm assessing proportion of within-module vs. all connections. Implemented with R igraph. Statistics: One-way ANOVA for group effects on whole-brain E and Q; Welch’s two-sample t-tests for post-hoc comparisons with Bonferroni–Holm correction. AoA effects tested via Pearson correlation within early and late bilinguals. Network-Based Statistics (NBS; R NBR v0.1.5) to identify clusters of connections contributing to whole-brain effects, with edgewise a priori threshold p<0.01 (two-sided) and 1000 permutations, FWE-corrected; consistency checked with p<0.05 threshold. Recursive NBS at ROI level within significant network clusters. Additional analyses: Linear regressions testing relationships of years of L2 experience (YoE = age − AoA) and self-reported L2 proficiency (Likert 1–7 across modalities) with E, Q, and NBS connectivity. Music training included as binary covariate (26 participants reported training) in whole-brain and network analyses.

• Whole-brain global efficiency: Significant group effect (F(3,147)=3.39; PFWE=0.04). Post-hoc: Early bilinguals vs monolinguals: t(40.3)=3.26; PFWE=0.014. Simultaneous bilinguals vs monolinguals: t(42.7)=2.77; PFWE=0.042. Late bilinguals vs monolinguals: t(36.2)=1.88; PFWE=0.272. No significant differences among bilingual groups (ps>0.074). No group effect for modularity (F(3,147)=0.53; p=0.66). • Age of acquisition (AoA): Significant negative correlation with global efficiency (ρ=-0.22; p-uncorrected=0.037), indicating earlier AoA associates with higher network efficiency; no correlation with modularity (ρ=-0.14; p=0.18). • Network-Based Statistics (NBS): Simultaneous vs monolinguals showed higher connectivity between cerebellum and association networks (default mode, dorsal and ventral attention, fronto-parietal, sensorimotor-hand) and increased ventral attention–salience connectivity (PFWE=0.001). Early vs monolinguals showed higher cerebellar connectivity with default mode, dorsal attention, fronto-parietal networks, and increased auditory–default mode and auditory–ventral attention connectivity (PFWE=0.001). • Inter-hemispheric cortico-cerebellar connectivity: In simultaneous bilinguals, 118 stronger connections than monolinguals; inter-hemispheric connections overrepresented vs intra-hemispheric (ratio=64:37=1.72; excluding 17 vermis-involving connections; χ²₁=7.22; p=0.0072). Right cerebellar lobules VI and VIII showed more inter-hemispheric connections to left dorsal attention areas. • No significant relationships of L2 proficiency or years of L2 experience (YoE) with global efficiency, modularity, or NBS connectivity. Group differences remained when controlling for music training.

Findings demonstrate that bilingualism, especially when L2 is acquired early or from birth, is associated with increased whole-brain functional integration (global efficiency) compared to monolingualism, while modular segregation (modularity) remains unchanged. This suggests that the organization of bilingual brain networks is characterized more by enhanced inter-network interactions than changes in within-module segregation. NBS analyses indicate that cortico-cerebellar circuits predominantly drive this increased efficiency, highlighting the cerebellum’s contribution beyond classical perisylvian language areas. The timing of L2 acquisition plays a critical role: earlier AoA is linked to higher global efficiency and stronger inter-hemispheric cortico-cerebellar connectivity, aligning with evidence of inter-hemispheric structural and functional differences in bilinguals and the cerebellum’s broad cortical projections. The study, conducted in a relatively homogeneous French/English bilingual environment, underscores the importance of developmental timing in shaping intrinsic functional patterns that support language learning and control and suggests that network-level integration underlies efficient bilingual language processing.

Using whole-brain network analyses of resting-state fMRI in a large sample spanning monolinguals and bilinguals with varied L2 AoA, the study identifies greater global efficiency in bilinguals—driven by early childhood L2 exposure—and heightened cortico-cerebellar and inter-hemispheric connectivity. Results emphasize that the developmental timeline of second language acquisition influences intrinsic functional organization, supporting more optimized mechanisms for achieving L2 competence when learning occurs during periods of heightened neuroplasticity. Proposed mechanistic interpretation suggests early bilingual environments may trigger biochemical and cellular cascades fostering neuroplasticity, yielding macro-structural and functional network changes that enhance efficiency. While neuroplasticity supports L2 learning across the lifespan, its nature likely differs when acquisition occurs later. Future research should explore how learning multiple languages and specific language pairings further shape these neural pathways.

Generalizability may be limited: the sample comprises French/English bilinguals in a specific Montreal linguistic environment, and findings may differ in contexts influenced by factors such as immigrant status and socioeconomic status. AoA and proficiency were self-reported, which may introduce reporting bias. The study focuses on resting-state functional organization and does not directly assess task-based language performance. Cerebellar contributions are inferred from functional connectivity; causality cannot be established.