Language exposure induced neuroplasticity in the bilingual brain: A follow-up fMRI study

Abstract

Although several studies have shown that language exposure crucially influence the cerebral representation of bilinguals, the effects of short-term change of language exposure in daily life upon language control areas in bilinguals are less known. To explore this issue, we employed follow-up fMRI to investigate whether differential exposure induces neuroplastic changes in the language control network in high-proficient Cantonese (L1)–Mandarin (L2) early bilinguals. The same 10 subjects underwent twice BOLD-fMRI scans while performing a silent narration task which corresponded to two different language exposure conditions, CON-1 (L1/L2 usage percentage, 50%:50%) and CON-2 (L1/L2 usage percentage, 90%:10%). We report a strong effect of language exposure in areas related to language control for the less exposed language. Interestingly, these significant effects were present after only a 30-day period of differential language exposure. In detail, we reached the following results: (1) the interaction effect of language and language exposure condition was found significantly in the left pars opercularis (BA 44) and marginally in the left MFG (BA 9); (2) in CON-2, increases of activation values in L2 were found significantly in bilateral BA 46 and BA 9, in the left BA44, and marginally in the left caudate; and (3) in CON-2, we found a significant negative correlation between language exposure to L2 and the BOLD activation value specifically in the left ACC. These findings strongly support the hypothesis that even short periods of differential exposure to a given language may induce significant neuroplastic changes in areas responsible for language control. The language which a bilingual is less exposed to and is also less used will be in need of increased mental control as shown by the increased activity of language control areas.

Introduction

Whether L1 and L2 are represented in different brain regions, has always been an interesting issue for neuroscientists. So far, many studies have investigated brain activities of bilinguals involved in the processing of L1 and L2, and as summarized by Indefrey (2006) and Abutalebi (2008), differences between L1 and L2 are usually reflected at the brain level by more extended activation patterns of L2 within or surrounding those regions responsible for L1 processing. As to why L2 engages more widespread brain activity, several studies have also suggested that the brain activity for language processing is modulated by specific demographic, linguistic and/or behavioral variables, such as language exposure in daily life (Abutalebi, Brambati, et al., 2007, Jeong et al., 2007, Perani et al., 2003), proficiency level (PL) (Briellmann et al., 2004, Chee, Caplan, et al., 1999, Chee, Tan, et al., 1999, Frenck-Mestre et al., 2005, Gandour et al., 2007, Golestani et al., 2006, Illes et al., 1999, Kotz, 2009, Perani et al., 1996, Perani et al., 1998), age-of-acquisition (AoA) of L2 (Bloch et al., 2009, Dehaene et al., 1997, Hernandez et al., 2007, Kim et al., 1997, Mayberry et al., 2011, Perani et al., 2003, Wartenburger et al., 2003), cross-linguistic similarities/dissimilarities (Jeong et al., 2007, Saur et al., 2009), and syntactic complexity (Suh et al., 2007, Yokoyama et al., 2006). Following Abutalebi and Green (2007), the extra-activity of L2 would be induced by an apparent lack of automaticity such as in cases where the proficiency for L2 is low or, likewise, when L2 is learnt later in life.

However, as underlined by Perani and Abutalebi (2005), there is an apparent lack of interest to one of these variables, i.e., language exposure in daily life. This fact is even more surprising when considering that language exposure in daily life was actually the first factor to be well described in the neurological literature pertaining to bilingualism and bilingual aphasia (Pitres, 1895). Language exposure was by then hypothesized to predict language outcome in bilingual aphasia, in the sense that the language the patient used more prior to the brain insult, was the one with the highest possibilities for recovery (see for discussion in Green & Abutalebi, 2008). To the best of our knowledge, only very few functional neuroimaging studies have analyzed the effects of language exposure upon the cerebral representation of languages in bilinguals (Abutalebi, Brambati, et al., 2007, Jeong et al., 2007, Perani et al., 2003). For example, Perani et al. (2003) carried out an fMRI study in high proficient Catalan-Spanish early bilinguals performing a lexical search and retrieval task, and found additional brain activity in the left middle frontal gyrus (MFG) (BAs 10, 46) and left inferior parietal lobule (IPL) (BA 40) only for the L2 to which subjects were less used. On the contrary, Perani et al. (2003) also observed less extensive engagement of prefrontal regions for the language to which subjects used more, and they attributed this to the strengthened neocortical connections by repeated activation of the cerebral representation of a more exposed language. Abutalebi, Brambati, et al., 2007, Abutalebi, Keim, et al., 2007 performed an event-related fMRI study to detect brain activity of language switching in early highly proficient Italian-French bilinguals, and observed the engagement of the left caudate nucleus and bilateral anterior cingulate gyrus (ACC) (BA 32) when switching into the less-used language during the auditory perception of language switches. The authors suggested that the activity found in the caudate nucleus and the ACC was specifically related to cognitive control mechanism and may reflect the switching cost when switching into a less exposed language. Jeong et al. (2007) examined the brain activation of Korean trilinguals in auditory sentence comprehension tasks in Korean, Japanese, and English, and found the brain activation in the opercular portion of left inferior frontal gyrus (IFG) (BA 44) and in the right cerebellum were significantly negatively correlated with the length of stay in English-speaking country. The authors suggested that this effect might be caused by the linguistic distance between languages at the syntactical level. Interestingly, most of the additional brain activities in the above-mentioned studies, i.e., left MFG (BA 10, 46), left IPL (BA 40), left caudate nucleus and bilateral ACC (BA 32), are all related to the language control network (Abutalebi and Green, 2007, Abutalebi and Green, 2008), suggesting that the influence of language exposure upon the brain affects mostly language control processes. However, the effects of a relatively brief period of differential language exposure upon the brain remain largely unknown since the aforementioned studies mainly investigated lifelong differences in language exposure.

Empirically, it is well known to bilingual speakers that a period of full immersion in an almost single language context may result in less automaticity for word-finding in the non-exposed language. The possible reason is that in an immersion environment where a single language provides the dominant context, the mental control of the non-exposed language will be hampered. Recently, Abutalebi and Green (2007) proposed an explanation in terms of mental control over limited resources (such as less proficiency to a given language), which contributes to the inhibition or activation of one language or the other. Besides, Paradis' Activation Threshold Hypothesis (Paradis, 1988) proposed that this inhibition is conveyed at different thresholds of activation such that the language to which the bilingual speaker is less exposed becomes less available due to its raised activation threshold. On the other hand, the more a bilingual speaker is exposed to a given language, the lower its activation threshold will be; hence, the easier that language is available for production. As to the cognitive basis of language exposure induced neuroplastic effects, we suggest that it may act on the automaticity with which a language is processed. The less automatic a language becomes (i.e., due to less usage of that particular language), the less available it is for spontaneous production (as compared to the more exposed language). More importantly, language use in bilingual speakers yields dual language activation and so requires them to select a response in the face of competing demands from the current non-target language. If automaticity in lexical access decreases with reduced exposure, there will be increased competitive demand to select that language, i.e., more inhibition for the more-exposed language and more neural effort for the initiation of the less-exposed language. We may, thus, hypothesize that the language a bilingual uses less for a short time may be in greater need of mental control.

In the present study, our a-priori hypothesis was that the effects of differential language exposure would be manifest upon the network that governs language control, i.e., a network built by the following areas: bilateral PFC, bilateral caudate nuclei, and the ACC (Green & Abutalebi, 2013). Indeed, our main aim was to detect the fast effects of language exposure on the language control mechanism that governs L1 and L2 processing in Cantonese-Mandarin bilinguals. For the sake of clarity, we refer in this study to language exposure as the percentage of L1 and L2 usage measured as the time spent in listening, speaking, reading, writing, and thinking of the subjects in their everyday lives. This was estimated by the recordings of the subjects' daily L1 and L2 usages in minutes in a home designed table (see Table S1 in Supplementary Materials). Considering putative confounding effects, we tried to isolate language exposure from other variables of language context. When recruiting subjects, we tried to avoid the effect of AoA of L2 by selecting subjects who were exposed to two languages from birth (simultaneous bilinguals) and tried to avoid the effects of syntactic complexity by using a natural language paradigm, a silent narration task, which involves only daily oral expression and therefore minimize additional control demands of processing structurally complex sentences (Suh et al., 2007, Yokoyama et al., 2006). In order to adequately control language exposures, we recruited a group of high-proficient bilinguals living in Guangzhou, a Cantonese (L1)-Mandarin (L2) bilingual region. Each subject attended two BOLD-fMRI scans when performing language tasks in an interval of approximately 3 months. The 1st fMRI scan was carried out during the semester when the subjects had been exposed in CON-1, in which they had comparable usage of L1 and L2. Approximately, two months after the 1st scan, subjects started their summer vacation during which they had significantly more usage of L1 than of L2 for at least 30 days (i.e., CON-2). The 2nd fMRI scan was immediately performed after the summer holidays in order to provide us a condition with differential language exposure.