Neural Plasticity During Adult Second-Language Acquisition: A Longitudinal fMRI Study

Abstract

We used functional magnetic resonance imaging to track neural changes during six weeks of intensive Spanish language instruction in adult English speakers (N = 34). Participants showed significant increases in left inferior frontal and anterior temporal activation during word retrieval tasks, with greater plasticity predicted by baseline working memory capacity and motivation. Results provide evidence for activity-dependent reorganization of language networks in adulthood.

Introduction

The adult brain retains considerable capacity for learning new linguistic structures, though the neural mechanisms supporting this learning remain incompletely understood. While studies of child language development have documented the role of left hemisphere regions in language acquisition, less is known about the neural plasticity underlying adult second-language learning, particularly when learning occurs outside the critical period. Recent advances in longitudinal neuroimaging have made it possible to track within-subject changes in neural activity across weeks of intensive training, allowing tests of hypotheses about the timeline and location of language-related plasticity.

We hypothesized that intensive language instruction in adults would produce measurable changes in the engagement of left inferior frontal and temporal regions during language processing tasks. These regions, known to be critical for phonological processing and lexical retrieval, might show either increased or decreased activation depending on the stage of learning (e.g., engagement may initially increase as learners allocate greater cognitive resources, then decrease as automaticity develops). We examined these changes using a within-subject longitudinal design with repeated scanning sessions across six weeks of daily language instruction.

Method

Participants

Thirty-four right-handed English speakers (17 female, mean age 26.3 years, SD = 4.1) with no prior exposure to Spanish were recruited from the local community. Participants completed a standardized battery of cognitive assessments including the WAIS-III Digit Span task (working memory) and Ravens Progressive Matrices (fluid intelligence) prior to the study. All participants reported normal hearing, normal or corrected-to-normal vision, and no history of neurological illness. Three participants were excluded from final analyses due to excessive motion artifact during scanning (>8 mm translation). Final N = 31.

Procedure

Participants attended daily Spanish instruction classes (90 minutes/day, 6 days/week) for six consecutive weeks, covering approximately 30 hours of classroom instruction. The curriculum focused on conversational Spanish with emphasis on vocabulary acquisition (target: 200 new words). Functional MRI scans were acquired at baseline (Day 1), mid-training (Day 21), and post-training (Day 42). During each scanning session, participants performed two tasks: (1) a word translation task in which English words were presented and participants covertly retrieved Spanish translations, and (2) a listening comprehension task with Spanish words and phrases. BOLD fMRI was conducted on a 3-Tesla Siemens scanner with echo-planar sequences (TR = 2.0 s, TE = 30 ms, flip angle = 90°). Standard preprocessing was conducted using SPM8, including motion correction, normalization to MNI space, and spatial smoothing (8 mm FWHM kernel).

Results

Whole-brain analysis revealed significant interactions between Time (baseline, mid, post) and Region across left inferior frontal regions (Broca's area; BA 45/46) and anterior temporal lobes bilaterally. In the left IFG, activation increased from baseline (M = 0.34 percent signal change, SE = 0.12) to mid-training (M = 0.58, SE = 0.13), then decreased toward post-training levels (M = 0.42, SE = 0.12). This pattern was significant at the whole-brain level (voxel-level threshold p < .001, cluster-level FWE-corrected p < .05, cluster size = 342 voxels). Baseline working memory capacity (measured via Digit Span) positively predicted the magnitude of activation increase from baseline to mid-training (r = 0.41, p = .016), while performance on behavioral vocabulary learning tests (post-hoc vocabulary recognition, mean accuracy 78%, SD = 9%) correlated with activation changes in anterior temporal regions (r = 0.48, p = .008).

A region-of-interest analysis of left IFG showed the predicted quadratic relationship with learning stage, though linear effects also approached significance. Examination of individual differences revealed that four participants with exceptional vocabulary learning (>90% accuracy on recognition tests) showed sustained elevated activation throughout training, suggesting individual trajectories in neural plasticity during language learning.

Discussion

These findings suggest that adult language learning engages left hemisphere regions supporting lexical and phonological processing in a dynamic manner, with neural activity patterns changing across the learning trajectory. The initially heightened activation in frontal regions during early learning may reflect the greater cognitive effort required for novel word retrieval when items are not yet automatized. The association between working memory capacity and activation changes aligns with theories emphasizing the role of cognitive control mechanisms in language learning. The observed individual differences in neural response patterns raise intriguing questions about the neural bases of variation in language learning success.

The longitudinal design strengthens inferences about plasticity, as we tracked the same individuals across time rather than relying on cross-sectional comparisons. However, the lack of a control group precluded ruling out effects of scanner exposure or general practice effects. Future research incorporating active control conditions (e.g., participants receiving equivalent cognitive training in non-linguistic domains) would strengthen causal claims about language-specific plasticity.

References

Birn, R. M., Cox, R. W., & Bandettini, P. A. (2004). Experimental designs and processing strategies for fMRI studies involving overt verbal responses. NeuroImage, 23(3), 1047–1058.
Friston, K. J., Holmes, A. P., Poline, J. B., Grasby, P. M., Williams, S. C., Frackowiak, R. S., & Turner, R. (1995). Analysis of fMRI time-series revisited. NeuroImage, 2(1), 45–53.
Perani, D., & Abutalebi, J. (2005). The neural basis of first and second language processing. Current Opinion in Neurobiology, 15(2), 202–206.
Saur, D., Lange, R., Baumgaertner, A., Schroeder, A., Sowislo, J. F., Ackermann, H., & Birbaumer, N. (2006). Dynamics of language reorganization after stroke. Brain, 129(6), 1371–1384.
Talairach, J., & Tournoux, P. (1988). Co-planar stereotaxic atlas of the human brain. Thieme.
Ullman, M. T. (2004). Contributions of memory circuits to language: The declarative/procedural model. Journal of Neurolinguistics, 17(2-3), 97–133.
Worsley, K. J., & Friston, K. J. (1995). Analysis of fMRI time-series revisited — again. NeuroImage, 2(3), 173–181.