Bilingualism

Bilingualism

Albert Costa, ... Cesar Ávila, in Neurobiology of Language, 2016

35.4 The Neural Correlates of Language Control: A Frontal, Parietal, and Subcortical Network

A number of neuropsychological studies have presented patients showing selective linguistic alterations and/or pathological language switching (Abutalebi, Miozzo, & Cappa, 2000; Fabbro, Skrap, & Aglioti, 2000; Paradis, 2001). These cases suggest the existence of specific brain areas involved in language control. With the complementary aid of neuroimaging techniques, brain areas such as the left caudate, the ACC, the lateral prefrontal cortex, and the left inferior parietal cortex have been proposed to conform the language control network (Abutalebi & Green, 2007). These brain areas were not specifically dedicated to this aim but acquired a specific role in language control as a part of its general function.
The brain lesion most frequently related to pathological language switching in bilinguals is that affecting the left basal ganglia (Abutalebi et al., 2000; Adrover-Roig et al., 2011; Aglioti & Fabbro, 1993). The general role of this area is to integrate information from multiple brain regions to shape motor learning. In the case of bilinguals, this area is required to establish the adequate motor language program including language planning, selection, and switching. A second area involved in language control is the ACC, which participates in the monitoring of different response alternatives during conflict processing (e.g., Botvinick, Braver, Barch, Carter, & Cohen, 2001; Botvinick, Cohen, & Carter, 2004; Braver, Barch, Gray, Molfese, & Snyder, 2001). In the case of bilinguals, the ACC may participate in the control process of L1 or L2, and in error detection and selective attention during language monitoring. Consistent with this role, Fabbro et al. (2000) reported a case of pathological language switching after a lesion in the left ACC. A third area involved in language control is the left lateral prefrontal cortex including the dorsolateral and ventrolateral parts. These parts exert general executive control functions over behavior in response to stimuli. For instance, when bilinguals named pictures or read words aloud in their L2, this area was more activated in highly proficient bilinguals than in monolinguals (Jones et al., 2012). Thus, the left lateral prefrontal cortex participates actively in the language control network by exerting a role in response selection and inhibition and in working memory. The last part of the brain proposed to participate in the network is the left inferior parietal lobe that participates in the maintaining of language representations in working memory.
The language control network described seems to be somehow affected by the language proficiency level in the second language. For example, bilinguals with high proficiency in the two languages (Garbin et al., 2011) engage different brain areas when switching between L1 and L2 as compared with low-proficient bilinguals (Wang et al., 2007). Note also that the language control network seems to be functionally influenced by sociolinguistic aspects other than language proficiency. For example, Perani et al. (2003) examined the role of age acquisition and language exposure/usage in early and highly proficient Catalan-Spanish bilinguals by means of a fluency task performed in the two languages. Authors revealed the lexical retrieval in the language acquired earlier in life was associated with less extensive activation in the brain. Language exposure/usage modulated the brain areas involved in lexical retrieval. In fact, in one of two groups of participants, those who were less exposed to the L2 showed larger activations during L2 lexical retrieval. These results indicated that in addition to language proficiency, also age of language acquisition and language exposure/usage are crucial factors in determining the neural pattern during lexical processing in bilinguals. This conclusion is in accord with recent ERP evidence suggesting that different linguistic experiences (early versus late acquisition of the L2) in bilinguals may affect the way in which the brain recruits language control (Martin et al., 2013).
Taken together, the network described, which consists of the prefrontal cortex, the ACC, the posterior parietal cortex, and the basal ganglia, constitutes an efficient brain network for language selection and control. In the next section, we describe how this network is specifically engaged when bilinguals have to switch between languages, because language switching relies heavily on cognitive control (Monsell, 2003) and how L2 proficiency can influence this activity.

35.4.1 The Neural Correlates of Language Switching

Neuroimaging techniques like positron emission tomography (PET) and fMRI have been used during the past 20 years to measure brain activity. They provide accurate information about time course of brain activity, and especially about brain localization of this activity. Initial studies on bilingualism were more focused on the representation problem, that is, in knowing whether both languages were represented in the same or different parts of the brain (Kim, Relkin, Lee, & Hirsch, 1997; Perani et al., 2003). Once the great overlapping in the brain activity for L1 and L2 representations was determined, the remaining question was to investigate how language control is exerted in the brain (i.e., how these overlapping brain areas are recruited for language selection, language inhibition, and language switching).
As in the ERP studies, the different fMRI experiments conducted to investigate language switching have been designed to investigate “sustained” and/or “transient” control mechanisms in language control. As previously introduced, both “sustained” and “transient” control processes are important for language control and both processes may be best characterized in a qualitatively different way and subserve different aspects of language control (Christoffels et al., 2007).
“Sustained” activity was studied by Hernandez, Martinez, and Kohnert (2000) and Hernandez, Dapretto, Mazziotta, and Bookheimer (2001) by comparing brain activation in early and highly proficient Spanish–English bilinguals during naming blocks in one language with naming in mixing blocks in either L1 or L2 (fixed order). Mixed compared with blocked naming conditions increased the activation of the left inferior frontal cortex and the bilateral dorsolateral prefrontal cortex for the two languages. These results suggest that switching between languages requires the extra participation of executive control areas. Similar results (bilateral activation of dorsolateral prefrontal cortex) were obtained by Wang, Kuhl, Chen, and Dong (2009) in late, low-proficient, Chinese–English bilinguals using a single digit naming task in mixed and blocked conditions. Also, they found an additional activation in the supplementary motor area (SMA) (see Guo et al., 2011 for similar results in low-proficient bilinguals). Interestingly, Wang et al. (2009) revealed some dissociation between the two languages for sustained control. The mixed condition elicited the activation of the left middle frontal gyrus and right precuneus relative to blocked naming in L1 (Chinese). The mixed condition as compared with the blocked naming in L2 (English) instead revealed the activation of a large network of brain areas: the bilateral middle frontal gyri, the cerebellum, the left inferior frontal gyrus, and the SMA. This recruitment of frontal areas for both languages may reflect proactive processes necessary to regulate the activation level of the two languages when the interference is high (i.e., mixed condition) (Braver et al., 2003). In accord with this observation, Ma et al. (2014) have explored sustained control in a group of Chinese–English unbalanced bilinguals, relatively highly proficient in their L2. In detail, authors revealed that switching between languages (mixed condition) as compared with naming in L1 (blocked L1 condition) elicited the activation of two large clusters. The first one included the left inferior frontal gyrus, the SMA bilaterally, the left insula, and the basal ganglia (including caudate and putamen portions). The second big cluster peaked into the left inferior parietal gyrus and extended to the left supramarginal gyrus and the angular gyrus and precuneus. Switching between languages (mixed condition) as compared with naming in L2 (blocked L2 condition) elicited activations mainly in the left inferior frontal gyrus, the bilateral precentral gyrus and SMA, the bilateral inferior parietal gyrus, the bilateral fusiform, the left lingual gyrus, the left inferior temporal gyrus, as well as the hippocampus bilaterally. These results revealed a neural dissociation that suggests that the sustained mechanisms for L1 and L2 involve different levels of control demands. This may be taken to reflect that when competition increases because of the need to alternate the two languages, the level of activation of the L1 needs to be reduced to favor L2 production. According to Ma et al. (2014), this might be achieved through the coupling between the frontal and basal ganglia brain regions.
These evidences suggest that language proficiency may affect the way in which sustained control processes are applied to regulate the availability of words of the two languages during speech production.
In a different study, Abutalebi et al. (2008) studied highly proficient bilinguals (university students of the Translation Department) using a similar procedure with a random presentation of cues to name in the L1 or in the L2. Besides the activation of left Broca’s area, it was reported that naming in L1 in the bilingual context (where subjects had to select L1 or L2 nouns following a cue) compared with monolingual contexts (where subjects had to select L1 nouns or L1 verbs following a cue) induced an increased activation in the left prefrontal cortex and specifically engaged the left caudate and the ACC. Strikingly, this pattern of activity was absent for the same L1 nouns when the same subjects were placed in a monolingual context, therefore highlighting the crucial role of these neural structures in language control and particularly in language switching.
As in ERPs, different studies have used the first instantiation of the language switching paradigm designed to study “transient” effects. For example, Wang et al. (2007) applied this paradigm to a group of late Chinese–English bilinguals. The overall switching condition when compared with non-switching condition activated the language control network, including the left and right dorsolateral prefrontal cortex, the right ACC, and right caudate. When considering directional changes, the authors observed an asymmetry in behavioral switch costs, such as a larger cost present when switching to the L1 than when switching to the L2. The notion that there is competition between languages predicts that increased executive processes are recruited to allow L2 production compared with L1, especially in the case of low-proficient bilinguals. In line with this prediction, Wang et al. (2007) observed that switches from L1 to L2 (“forward switching”) activated the left ACC/SMA, whereas the switches from L2 to L1 (“backward switching”) did not activate any brain area within the language control network (Figure 35.2).

in low-proficient bilinguals using a single digit naming task. They replicated behavioral asymmetries of switch costs (larger switch cost to L1 than to L2) and that switching from L1 to L2 (“forward switching”) activated the language control network (left SMA, left dorsolateral prefrontal cortex, and left inferior parietal lobe), but the results for overall switches and switches from L2 to L1 (“backward switching”) did not show activations in areas of the language control network. In a further study using the first instantiation of language switching paradigm, Garbin et al. (2011) studied a group of early and highly proficient bilinguals and found a different pattern of results. Switches from L1 to L2 (“forward switching”) activated the left caudate, whereas switches from L2 to L1 (“backward switching”) activated the SMA/ACC (Figure 35.3).

At the neural level, different brain areas appear to be engaged during language switching and, interestingly, functional data indicate that the nature of this may alter with L2 proficiency, consistent with a change from controlled to more automatic L2 processing (Abutalebi & Green, 2007).
Abutalebi et al. (2012, 2013) used a different strategy based on the comparison of the process of language switching in bilinguals to a within-language switching task in monolinguals. The first study focused on the ACC and described this brain area as tuned by bilingualism to resolve cognitive and language conflicts (Abutalebi et al., 2012). The second study compared multilinguals with high proficiency in the L2 and poor proficiency in the L3 with monolinguals⁵ while performing the first instantiation of the language switching paradigm twice (once with L1 and L2, and the other with L1 and L3). In this study, authors were particularly interested in evaluating how language proficiency modulates the brain network of bilingual language control during a language switching task. The crucial finding of this study is that the pre-SMA/ACC participated in language switching regardless of individual proficiency, whereas the left caudate only was active when switching from L1 to the less proficient language (L3). Contrasting with these results, no neural significant differences were found in this study when comparing direction of switches within the multilingual group.
These studies suggest that language proficiency has an important role in determining how brain areas are recruited for language control, both for the “sustained” and “transient” components.
Besides these adaptations of the switching behavioral paradigms, other studies have used perceptive tasks. Crinion et al. (2006) used a classical semantic priming procedure to compare neural activation when the prime and the target belong to the same language or to different languages. Results showed an activation of the left caudate during language switching compared with nonlanguage switching trials across three different groups of highly proficient bilinguals. A different perceptive task was used by Abutalebi et al. (2007) in a sample of early, highly proficient bilinguals with more experience in L2. They used narratives that included unpredictable changes of languages that might be regular or irregular depending on the respect or violation of the constituents of sentence structure. The comparison of switch and nonswitch trials activated more language areas, including the left inferior frontal gyrus and middle temporal gyrus. Switches into L1 (the less exposed language) activated the left caudate and the ACC, whereas switches into L2 did not activate the language control network.
In general terms, the reviewed studies provide an acceptable uniformity of the results highly consistent with brain areas proposed to form the language control network. This uniformity is greater if we attend to the special role played by variables, such as the type of task and proficiency of bilinguals. The studies designed to investigate “sustained” processes in language control showed consistent activations in the dorsolateral prefrontal cortex and the SMA/ACC, whereas those more related to the “transient” process circumscribed the neural basis of language switching to the SMA/ACC. As in other conflict processes, the language conflict seems to require the participation of ACC that may exert the functions of monitoring and resolution. The dorsolateral prefrontal cortex would remain a global cognitive control mechanism of language conflicts. The role of the left caudate seems to be restricted to language in early and highly proficient bilinguals in tasks that involve language production and comprehension (see Ma et al., 2014 for involvement of caudate in late bilinguals more proficient in L2). Even though its specific role is not clear because in some studies it participated in “backward switching” and in others it appeared to be more related to “forward switching,” the diverse studies were consistent in showing a role in detection of language switching when both languages were acquired early. Finally, the left inferior parietal lobe is the brain area of the language network less often detected, probably due to the low working memory demands of the used paradigms. The activation of this area has only been found in tasks using the same stimuli (i.e., numbers) and responses. Maybe its role would be more prominent in translation tasks (Price, Green, & Von Studnitz, 1999).