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

The study investigates how learning a second language (L2) at different ages influences the organization of intrinsic functional brain networks. Grounded in evidence that neuroplasticity and skill acquisition depend on age of learning, the authors hypothesize that early L2 acquisition enhances whole-brain network integration (global efficiency) and possibly modular segregation (modularity). Given mounting evidence implicating cortico-cerebellar circuits in language and cognitive control, the study specifically examines whether cerebello-cortical connectivity differs across monolinguals and bilinguals with varying ages of L2 acquisition. The purpose is to clarify whether timing of bilingual exposure shapes global and local properties of resting-state functional networks and to identify the network pathways driving any observed effects, with potential implications for understanding language development and brain efficiency.

Prior work shows bilingualism can influence working memory, cognitive control, attention, speech-in-noise perception, and aging cognition, though effects depend on factors such as age and task demands. Neuroimaging studies report that bilingualism shapes both functional and structural brain organization. Task fMRI consistently implicates left frontal/temporal regions and the right posterolateral cerebellum in language performance, underscoring the importance of cortico-cerebellar systems. Beyond classical language areas, fronto-parietal systems and the cerebellum contribute to bilingual cognitive control and rule processing. Contemporary models emphasize distributed, interacting language networks rather than localizationist accounts. Graph-theoretic analyses reveal modular (segregated) and integrative properties of brain networks; global efficiency has been linked to cognition and language performance. Previous bilingual studies often targeted specific regions or networks (language/control) and white-matter tracts; relatively few adopted a whole-brain network perspective including the cerebellum. Some studies reported higher IFG connectivity and interhemispheric frontal connectivity in early bilinguals, and associations between age of acquisition (AoA) and callosal structure. However, the relationship between whole-brain segregation/integration and early vs late L2 learning, particularly involving cerebello-cortical interactions, remained unclear.

Design and participants: Retrospective compilation of resting-state fMRI data from 151 right-handed, healthy adults scanned at the Montreal Neurological Institute (single 3T Siemens TrioTim, consistent parameters). Groups by self-reported age of acquisition (AoA) for English/French: monolinguals (N=19; limited L2 exposure), late bilinguals (AoA >5 years; N=59), early bilinguals (AoA ≤5 years; N=34), simultaneous bilinguals (AoA from birth; N=39). All reported no other languages beyond English/French. Exclusions: language/hearing/visual impairment, traumatic brain injury, medical/neurological disorders, MRI incompatibility. Ethics approvals and informed consent obtained. MRI acquisition: Resting-state fMRI (eyes open, fixation): T2*-EPI, 132 volumes, TR/TE=2260/30 ms, flip angle 90°, matrix 64×64, FoV 224 mm, 38 slices, 3.5 mm thickness, duration 5:04 min. Structural: T1-weighted MPRAGE TR/TE=2300/2.98 ms, flip angle 9°, matrix 256×256, FoV 256 mm, 1 mm isotropic. Preprocessing: CONN v20b with SPM12 routines. Steps: realignment/unwarping, outlier detection (ART conservative; outliers defined as >3 SD intensity deviation or >0.5 mm composite motion from previous volume), segmentation/normalization to MNI space, 6 mm FWHM smoothing. Nuisance regression included aCompCor components from WM/CSF, motion parameters, and identified outlier scans; confound control followed established approaches for motion/physiological noise. Connectome construction: Time series averaged within 374 ROIs (333 cortical from Gordon parcellation plus 15 subcortical and 26 cerebellar from CONN Harvard–Oxford). ROIs grouped into 14 modules (12 cortical networks plus subcortical and cerebellum). Edges computed as Pearson correlations between ROI time series and Fisher z-transformed; negative edges discarded; weighted analyses used. Network metrics and statistics: Whole-brain network features: global efficiency (E; integration) computed via standard shortest-path algorithms; modularity (Q; segregation) via Newman’s algorithm. Group effects tested via ANOVA and post hoc tests with multiple-comparisons control (Bonferroni-Holm). Age of acquisition (AoA), L2 proficiency, years of bilingual experience (YoE), and music training examined as covariates/predictors. Network-based statistics (NBS) assessed group differences in connectivity between functional networks and at the edge level; family-wise error (FWE) control applied. Motion was quantified via mean framewise displacement; groups did not differ significantly in motion. Sample characteristics summary (Table 1): N per group as above; mean ages approximately 24–26 years with no significant group difference; sex distributions reported; AoA differed by design; years of education and in-scanner motion not significantly different across groups.

- Whole-brain integration: Significant group effect for global efficiency E (F(3,147)=3.39; PFWE=0.04). Post hoc: early bilinguals vs monolinguals t≈3.26, PFWE=0.014; simultaneous bilinguals vs monolinguals t≈2.77, PFWE=0.042; late bilinguals vs monolinguals t≈1.88, PFWE=0.272. No significant differences among bilingual groups (ps>0.074). - Segregation: No group effect for modularity Q (F(3,147)=0.53; p=0.66). - Age of acquisition: Within bilinguals, earlier AoA associated with higher global efficiency (correlation = −0.22; p-uncorrected=0.037); no association with modularity (correlation = −0.14; p=0.18). - Network-based statistics (functional networks): Relative to monolinguals, simultaneous bilinguals showed higher connectivity between cerebellum and association networks (default mode, dorsal/ventral attention, fronto-parietal, sensorimotor-hand) and higher ventral attention–salience connectivity (PFWE=0.001). Early bilinguals showed higher cerebellar–association network connectivity (default mode, dorsal attention, fronto-parietal) and higher auditory–default mode and auditory–ventral attention connectivity (PFWE=0.001). - Edge-level NBS (simultaneous vs monolinguals): 118 edges stronger in simultaneous bilinguals. Inter-hemispheric edges were over-represented versus intra-hemispheric (ratio 64:37=1.72; excluding 17 vermis connections; χ²(1)=7.22, p=0.0072). Right cerebellar lobules VI and VIII had numerous inter-hemispheric connections to left dorsal attention areas. - Null effects of experience measures: No significant relationships for L2 proficiency or years of bilingual experience with E, Q, or NBS connectivity. Group differences persisted when covarying music training.

Findings indicate that bilingualism—particularly when L2 is acquired very early—enhances whole-brain functional integration without altering modular segregation. This suggests that the bilingual brain’s organization is characterized by more efficient cross-network interactions rather than changes in the degree of within-network clustering. Network-based statistics localized these effects to cortico-cerebellar and association networks, highlighting the cerebellum’s crucial role beyond classical perisylvian language regions in supporting efficient language function and learning. The prominence of inter-hemispheric cortico-cerebellar connections, especially involving right cerebellar lobules VI/VIII and left dorsal attention areas, aligns with prior evidence for enhanced inter-hemispheric communication and callosal differences in bilinguals, and with task-based co-activation of right cerebellum and left frontal language regions. Importantly, AoA, rather than L2 proficiency or years of exposure, related to higher global efficiency, supporting models where early developmental timing confers enduring advantages in network integration due to heightened neuroplasticity. While the sample’s linguistic context (Montreal French/English) enhances homogeneity and control of confounds, it may limit generalizability; nonetheless, the implicated cerebello-cortical systems are likely central to language organization and AoA effects given their roles in speech-motor coordination, working memory, attention, sequencing, error monitoring, and syntax.

Using whole-brain network analyses of resting-state fMRI in a large, well-characterized cohort, the study shows that bilinguals—especially those exposed to two languages from birth or early childhood—exhibit higher global efficiency driven by stronger cortico-cerebellar and association network connectivity, with a marked inter-hemispheric pattern. These results underscore the influence of developmental timing of L2 acquisition on intrinsic brain organization and suggest a more optimized network mechanism for language skillfulness when learning occurs during periods of heightened neuroplasticity. Future research should test generalization across diverse bilingual contexts, examine how learning multiple languages and different language pairings modulate these pathways, and integrate longitudinal designs to disentangle developmental trajectories and causality.

Generalizability may be limited: the sample comprised French/English speakers from the Montreal linguistic environment, and findings may differ in other bilingual populations or contexts where factors such as immigrant status or socioeconomic status modulate bilingual experience. The study focuses on intrinsic connectivity from resting-state fMRI, and while groups were homogeneous and motion-controlled, directionality and causal mechanisms cannot be inferred from resting-state data. Open questions remain regarding effects of learning multiple languages and specific language pairings.