Using more than one language shapes brain structure, a phenomenon widely observed in bilingual groups (Abutalebi et al., 2015a; DeLuca et al., 2019, 2020; Grundy et al., 2017; Luk et al., 2011; Marin-Marin et al., 2022; Mechelli et al., 2004; Olulade et al., 2016; Pliatsikas et al., 2015). Research consistently shows that bilingual groups develop greater gray matter volume than monolinguals in regions such as the anterior cingulate cortex (ACC), inferior parietal lobule (IPL), inferior frontal gyrus (IFG), and basal ganglia (Abutalebi et al., 2015a,b; Burgaleta et al., 2016; DeLuca et al., 2024; Mechelli et al., 2004). These areas play central roles in language control and executive functions (Green and Abutalebi, 2013; Grundy et al., 2017).

Researchers have also reported similar bilingualism-related differences in white matter tracts. Aged and young bilingual groups show better white matter integrity than their monolingual counterparts in tracts such as the bilateral superior longitudinal fasciculus (SLF), the inferior fronto-occipital fasciculus (IFOF), and the corpus callosum (Luk et al., 2011; Pliatsikas et al., 2015). These structural adaptations—including increased gray matter volume in executive control regions and enhanced white matter integrity—appear to support cognitive advantages often observed in bilinguals, such as improved attentional control, task switching, and working memory (Alrwaita et al., 2024; Bialystok, 2011; DeLuca et al., 2020; Green and Abutalebi, 2013; Miyake et al., 2000).

Although bilingualism promotes structural adaptations, these changes vary across learners’ backgrounds and language experiences. For instance, while bilingualism generally relates to extended cortical volume, evidence suggests that this relationship changes with the stage of second language (L2) immersion (Grundy et al., 2017; Pliatsikas et al., 2020). Simultaneous interpreters, who are typically bilinguals at later stages of L2 immersion, demonstrate decreased volume in regions such as the left ACC, bilateral IFG, and left insula, changes that correlate with cumulative hours of interpreting (Elmer et al., 2014). Moreover, DeLuca et al. (2020) proposed that specific experience-based bilingual factors uniquely shape brain structure. They found that the intensity and diversity of language use influence cortical changes in regions supporting language processing and executive control, whereas the duration of L2 usage promotes neural efficiency, reflected in changes to subcortical structures and white matter pathways (DeLuca et al., 2024). These findings highlight the need to move beyond treating bilinguals as a homogeneous group and examine how specific aspects of language experience such as language proficiency, age of acquisition (AoA), and daily use frequency, drive neural adaptation (e.g., Bialystok et al., 2012; DeLuca et al., 2020; Luk and Bialystok, 2013).

Despite these advances, most neuroimaging studies focus on traditional bilinguals who naturally acquire two (or more) languages and use them regularly in interactional contexts. Far less is known about groups with non-immersive dual-language experiences, those who neither learn a second language naturally nor are immersed in a bilingual environment. This study addresses that gap by focusing on a large but understudied group: long-term foreign language learners with limited daily exposure to the target language. These learners typically study their foreign language—English—in formal educational settings for at least 12 years, but rarely use it in everyday life. As a result, they often achieve high proficiency but accumulate little practical experience. Such profiles are especially common in East Asia, including China and South Korea, where English education is extensive but real-world exposure to English outside the classroom remains minimal. Despite the growth of bilingualism research, scholars have largely overlooked how this distinct type of language experience relates to neural adaptation.

This foreign language learner population differs from traditional bilinguals and most studied second language learners in two significant ways. First, the context and goals of their learning differ. Their language learning occurs primarily in academic settings and also arises from achievement-oriented goals rather than organic exposure (Shin et al., 2018). Since they do not live in multilingual environments, they lack opportunities for spontaneous, real-time oral interaction in the target language. It remains unclear whether such differences in language experience lead to different brain restructuring. Second, East Asian learners typically study a foreign language, mostly English, over an extended period. Previous neuroimaging studies on L2 learning have mostly examined short-term training effects (e.g., Legault et al., 2019; Mårtensson et al., 2012). Whether long-term foreign language learning produces comparable or distinct neural effects remains an open question.

Converging evidence suggests that even limited exposure to a foreign language can influence cognitive and neural development. Infants exposed to a non-native language can discriminate phonetic features absent in their native tongue (Kuhl et al., 2003). Young monolingual children with minimal Spanish exposure successfully acquired novel words from Spanish speakers (Akhtar et al., 2012). Children with limited non-native language input have also shown greater openness to dual labeling (Rojo and Echols, 2018), particularly in socially interactive contexts (Lee and Song, 2024). Among adults, short-term L2 training enhances cognitive performance (Wong et al., 2019), strengthens functional connectivity (Bubbico et al., 2019), and even induces structural brain changes (Mårtensson et al., 2012). Together, these findings suggest that, even without immersive environments, long-term foreign language learning may yield measurable neuroplastic changes.

Building on this work, the present study investigates how gray matter volume and white matter integrity are associated with age of acquisition, foreign language proficiency, and daily use experiences among long-term English learners in South Korea, a predominantly monolingual environment. The participants in this study began learning English since childhood, primarily through formal instruction with minimal opportunities for everyday communicative use, giving them a distinct profile from those in previous studies. We hypothesize that, despite their restricted daily exposure, they will exhibit structural adaptations linked to language-related factors in gray matter volume and white matter integrity in regions involved in language and cognitive control, following patterns observed in traditional bilinguals.

Table 2 presents descriptive statistics and intercorrelations for self-reported proficiency, AoA, and daily use of English. First, participants reported relatively high English proficiency (M = 25.73, SD = 5.71 out of 40). To supplement these self-reports, we collected certified English test scores, including the Test of English as a Foreign Language (TOEFL), Test of English for International Communication (TOEIC), commonly used in Asia and Europe, and Test of English Proficiency (TEPS), developed by Seoul National University in Korea. We converted all scores to TOEFL equivalents using the official conversion table: https://www.teps.or.kr/InfoBoard/ConversionTable#. The mean converted score reached 102.34 (SD = 9.81), substantially higher than the 2024 global average TOEFL score of 86 (ETS, 2025). Certified scores correlated strongly with self-reported proficiency (r = 0.70, p < 0.001), confirming the validity of the self-report measure.

Second, participants reported low daily English use (M = 17.41, SD = 9.52 out of 80). We derived this score from 20 LSBQ items that measured English use across diverse situations and activities, as well as language switching within conversations. The mean score represents less than one-quarter of the maximum possible, reflecting participants’ reliance on Korean in most personal and media contexts and their infrequent engagement in language switching. None of the participants had lived abroad before adulthood, and all reported fewer than 2 years of overseas experience afterward. Together, the low daily use scores from the LSBQ and the minimal residence abroad operationalize “limited exposure,” capturing participants’ scarce opportunities for naturalistic, immersive English use despite their relatively high proficiency.

Third, English proficiency correlated significantly with AoA (r = −0.64, p = 0.001), suggesting a link between earlier acquisition of English and higher proficiency. In contrast, neither AoA nor proficiency correlated significantly with daily English use (r = 0.03, n.s.; r = 0.23, n.s., respectively).

After establishing the behavioral relationships among AoA, proficiency, and daily use, we examined whether these language-related experiences correlated with brain structure. Using VBM analysis, we tested whether gray matter volume in bilingualism-related regions varied with these language experiences (Table 3). Volume in the left angular gyrus increased with earlier AoA (uncorrected p < 0.001), while volume in the right angular gyrus increased with greater daily English use (uncorrected p = 0.001; Figure 1A). Daily English use also predicted reduced volume in the right ACC (uncorrected p < 0.001; Figure 1B). Among the subcortical structures, later AoA predicted reduced volume in the left thalamus (uncorrected p < 0.001; Figure 1C). Overall, earlier AoA correlated with greater gray matter volume in several regions, whereas daily English use predicted distinct patterns of structural adaptations. We did not observe other significant relationships. Supplementary Table 1 provides a complete list of significant and non-significant VBM results across the ROIs.

We then tested whether white matter integrity was associated with language experiences using TBSS analysis (Table 4). We found that AoA positively predicted MD in the splenium of the CC (FWE-corrected p = 0.009), whereas daily use negatively predicted MD in the body of the CC (FWE-corrected p = 0.030; Figure 2). We observed no significant effects for FA or in other TOIs. Supplementary Table 2 provides the complete list of significant and non-significant TBSS results.

The present study investigated whether long-term foreign language learning under conditions of limited daily exposure relates to structural brain adaptations. We focused on a unique and understudied group: long-term foreign language learners in South Korea, a linguistically homogeneous country where opportunities for foreign language use are scarce. These learners acquired English in formal educational settings over an extended period but had little exposure in everyday communication. Unlike traditional bilinguals frequently studied in previous research, who use their second language regularly in immersive environments, this group allows us to examine neuroplasticity that emerges under constrained experiential input. Prior studies show that even minimal exposure can shape language learning patterns (Bubbico et al., 2019; Kuhl et al., 2003; Lee and Song, 2024; Wong et al., 2019), leading us to hypothesize that sustained but limited experience can still induce meaningful neurocognitive changes. Accordingly, we examined how gray matter volume and white matter integrity in bilingualism-related regions varied with age of acquisition (AoA), English proficiency, and daily use.

Our VBM analyses revealed significant differences in the bilateral IPL, the right ACC, and the left thalamus. Specifically, earlier AoA (i.e., lower numerical AoA values) predicted greater gray matter volume in the left angular gyrus, a part of IPL known to play a central role in language learning and semantic processing (Barbeau et al., 2017; Van Ettinger-Veenstra et al., 2016). In our study, AoA correlated negatively with gray matter volume in this region, meaning that participants who learned English earlier had greater gray matter volumes. Daily use then positively predicted gray matter volume in the right angular gyrus. These findings align with previous research on bilingualism. For instance, Mechelli et al. (2004) found that early and late bilinguals exhibited increased gray matter density in the bilateral IPL and that L2 AoA correlated negatively with left IPL density. Similarly, Abutalebi et al. (2015b) reported that the left IPL correlated with L2 proficiency, whereas the right IPL linked to L2 exposure, results highly consistent with ours. Together, our findings replicate and extend prior work in bilinguals, showing that even long-term foreign language learning with limited exposure can reshape bilateral IPL structures involved in language processing.

We also found that higher daily English use predicted reduced gray matter volume in the right ACC, consistent with results from advanced bilinguals (e.g., Elmer et al., 2014). Grundy et al. (2017) proposed that highly proficient bilinguals exhibit reduced yet more efficient activation and decreased gray matter volume in the anterior regions, suggesting a neural optimization effect. The Dynamic Restructuring Model (Pliatsikas, 2020; Pliatsikas et al., 2020) similarly suggests that reductions in gray matter volume among proficient bilinguals may reflect increased efficiency in language and cognitive control mechanisms. Supporting these models, structural and functional studies show that bilinguals recruit the ACC more efficiently during cognitive tasks, with decreased activation linked to reduced volume, indicating improved language conflict management (Abutalebi et al., 2012). Therefore, our findings suggest that long-term foreign language learning, even with minimal daily exposure, strengthens cognitive control systems. An alternative interpretation is that reduced volume may reflect weakened control in this area; however, converging behavioral and neuroimaging evidence makes this less likely. Notably, Liu et al. (2021) found that learners with similar profiles to our participants exhibited decreased left ACC volume associated with better performance in language-switching tasks, reinforcing the interpretation that reduced ACC volume reflects functional efficiency. Although our results involved the right ACC, they likely reflect a similar optimization process.

In the subcortical regions, earlier AoA predicted greater gray matter volume in the left thalamus. This finding aligns with previous research (e.g., Burgaleta et al., 2016; Korenar et al., 2023), which identified the thalamus as a key relay structure supporting language and domain-general learning (Ayyildiz et al., 2025; Bulut and Hagoort, 2024; Grillner and Robertson, 2016). Researchers also report thalamic adaptations as a later stage of neuroplasticity, supporting more effective information relay and cognitive control across distributed networks (Green and Abutalebi, 2013; Grundy et al., 2017; Pliatsikas et al., 2020). Along with the observed reduction in right ACC volume, these findings together further suggest that long-term learning with limited or absent immersive L2 environments can trigger advanced stages of structural adaptation resembling those in experienced bilinguals.

Our diffusion-weighted imaging data revealed significant structural changes in white matter integrity. Specifically, earlier AoA and greater daily use predicted increased integrity in different parts of the CC, a major interhemispheric communication tract. Similar results have been widely reported in diverse bilingual populations (Luk et al., 2011; Pliatsikas et al., 2015), and even after short-term L2 training (Schlegel et al., 2012). Although the CC is not language-specific, it likely supports conflict management during language-switching, which may explain the observed structural differences. These results support our hypothesis that long-term L2 learning—even under minimal daily exposure—facilitates white matter reorganization that enhances efficiency in language and cognitive processing. However, diffusion metrics such as FA and MD provide only indirect indices of integrity, as multiple microstructural properties contribute to these metrics. Future studies should employ more biologically specific diffusion models to enhance interpretability.

Overall, our findings demonstrate that structural adaptations commonly observed in bilinguals also emerge in long-term foreign language learners with minimal exposure. Earlier AoA predicted increased gray matter volume in the left angular gyrus and left thalamus, along with enhanced white matter integrity in the splenium of CC. Greater daily use predicted increased gray matter volume in the right angular gyrus and enhanced integrity in the body of the CC, but reduced gray matter volume in the right ACC. Together, these patterns of combining increases, decreases, and white matter reorganization mirror the dynamic adaptations reported in bilingual populations.

We did not find any significant effects of the composite English proficiency score on gray matter regions or white matter tracts. One plausible explanation is that proficiency levels in our sample were relatively homogeneous, with most participants already achieving high academic proficiency, reducing variance. Another possibility is that the composite score averaged across speaking, comprehension, reading, and writing obscured skill-specific neural associations. To address this, we conducted an additional analysis entering each self-reported skill separately. Interestingly, we found that self-reported speaking proficiency correlated positively with gray matter volume in the left pars triangularis, a part of IFG (cluster size = 108 voxels, peak uncorrected p = 0.000, MNI coordinates x = −52, y = 34, z = 12). This region is a key part of Broca’s and is well-known to be involved in speech production and bilingualism (Green and Abutalebi, 2013; Indefrey and Levelt, 2004). Writing proficiency correlated positively with FA value in the body of the CC (cluster size = 22 voxels, peak uncorrected p = 0.001, FWE-corrected p = 0.034, MNI coordinates x = −9, y = −1, z = 28), suggesting enhanced white matter integrity. These findings imply that production-related foreign language skills may have a specific impact on neural adaptations. Nonetheless, the mechanisms underlying skill-specific effects remain unclear, as few studies have examined this directly. Addressing this gap will require future research that systematically explore how individual skills contribute to structural plasticity.

By extending bilingualism research to long-term foreign language learners, our study highlights an important but often overlooked population. In many predominantly monolingual social societies, such as South Korea, learners acquire a foreign language primarily through formal education with little naturalistic exposure. Our findings show that meaningful neuroplastic changes are not exclusive to traditional immersive bilinguals. Instead, sustained learning under limited exposure can also reshape brain structures, thus broadening definitions of bilingualism and underscoring the need to include diverse learner profiles in neurocognitive research.

Moreover, our results suggest that long-term learning without immersion can still drive structural changes. Previous studies have reported gray matter volume increases after short-term language training (e.g., Mårtensson et al., 2012; Stein et al., 2012). Building on this evidence, we observed white matter enhancements and gray matter reductions in anterior brain regions, changes typically associated with later stages of neuroplasticity. We do not argue that immersive L2 exposure is unnecessary; rather, our study emphasizes that limited daily exposure combined with long-term learning experience in educational settings can also have a substantial impact on structural adaptations which is potentially comparable to those associated with more immersive experiences.

Our findings also carry implications for language education. Educational settings can provide structured opportunities for language learning and use to drive significant structural changes that support more efficient language control. Given that language control shares neural resources with domain-general executive functions, such educational experiences may also confer indirect cognitive advantages like those observed in bilingualism (Bialystok et al., 2012; Green and Abutalebi, 2013).