Bilingualism

by Soo Jung Lee

=**1 General Overview **= toc

Just like the word, bilingualism is defined as acquiring more than one native language simultaneously. Bilingual language acquisition is different from second language acquisition. One method to distinguish these two is the presence of a critical period or sensitive period. This is an early period of development that a child can absorb certain behaviors rather than actively learning them. The critical period for language acquisition is still in debate, but most linguists agree that children start to acquire language at age 1 and complete language development before age 3. In other literature, bilingualism may refer to simply speaking more than one language in a non-specific way.

 Bilingualism is one way to look at brain plasticity and it is highly stable across languages. There has been a great interest in the neural correlates in the isocortex of managing two languages. There are some known facts about language and brain structures: lateralization to the left hemisphere, some subcortical structures specific to language switching, and different cortical recruitment that matters with Age of Acquisition (AoA). However, recent studies suggest distributed neural networks including the right hemisphere rather than lateralization to only the left or one-to-one relationship.

 There are some known advantages of bilingualism. White matter integrity has been found to be better in bilingual people. Corresponding to this study, bilingual Alzheimer’s Disease (AD) patients showed better cognitive reserve.

=**2 Neural Basis of Language switching** =

__2-1 Language switch__
The existence of language switch was first proposed by Penfield and Roberts (1959). In their study, they observed no anatomical separation while the children switched from one language to another language. They defined the ‘switch’ as a conditioned reflex.



2-2a General Structures for language
Subcortical structures involved in aspects of language are the basal ganglia, thalamus, subcortical white matter pathways and the cerebellum. Other suggested areas are the caudate nucleus, putamen, globus pallidus, anterior and posterior internal capsule and subinsular. Lesions in these areas (striatocapsular strokes) result in language disorders [7]. These lesion-based studies provided crucial information for a valid neurocognitive model [3]. Contradicting results were reported from a bilingual aphasia study: some individuals had severe disorders with their first language while their second language was left intact. It is likely to be the degree of language automatization that is related to recovery; less automatization had more cerebral cortex representation [8].

2-2b Switch-specific areas
Language switching involves one’s brain inhibiting the non-targeted language and successfully activating the targeted language. There are two kinds of language switching. Switching from a First language (L1) to a Second Language (L2) (a //forward// switch) and the reverse of this (a //backward// switch). In both cases, presented and produced language will be different. According to recent meta-analysis of 128 studies, eight brain regions were found to be the major structures underlying neural correlates for language switching; the left inferior frontal gyrus (l-IFG), the left middle temporal gyrus, the left middle frontal gyrus, the right precentral gyrus, the right superior temporal gyrus (r-STG), the midline pre-supplementary motor area (SMA) and the bilateral caudate nuclei .Other suggested areas based on neuroimaging studies are: the left caudate nucleus, the prefrontal cortex (PFC), the anterior cingulate cortex (ACC), the left temporal-parietal areas including superior temporal gyrus, and the supramarginal gyrus.

2-2c Age factor
Different subcortical recruitment patterns were also found according to the AoA. A correlation between AoA of a second language and subcortical organization was also reported. Some early papers might have mixed participants of early and late bilingualism. Recent papers started considering AoA as a conditioning factor.

__2-3 Late bilingual studies__
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;">It is still hard to differentiate between when language acquisition for the first language stops and when second language acquisition begins. However, the most common method for distinguishing bilingualism is the proficiency. Methods used to assess language proficiency are self-assessment accompanied with an English Vocabulary Test, a New National Adult Reading Test, and a Graded Naming Test. Some studies included age as a condition, mostly no later than early teenage. Thus, studies possess varying degree of language proficiency.

<span style="color: #000000; font-family: Arial,Helvetica,sans-serif;">2-3a High proficiency
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;"> Although not necessarily early proficient, some highly proficient bilingual people showed activation in a similar set of brain regions irrespective of forward or backward switching [10]. This suggests that underlying neural circuits between two languages are highly overlapping in highly proficient bilingual people. This data does not however show how the brain controls language switching [10]. According to Crinion et al. (2006), semantic priming—the relevance of paired words in different or the same language—was only observed in the left caudate head (fMRI and PET were used). They concluded that the head of left caudate controlled and monitored the language in use. Subject languages were Japanese, non-Indo-European (IE), and German and English that were both IE. Thus, it may shed light on the universal role of the head of the left caudate across the languages.



<span style="color: #000000; font-family: Arial,Helvetica,sans-serif;">2-3b Varying proficiency
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;">In vascular brain imaging studies, proficiency in L2 affected blood-oxygen-level-dependent (BOLD) signals : The less proficient, the larger the BOLD signal that was shown in various structures [15]. This also coincided with AoA studies: in late bilinguals AoA was a confounding factor for proficiency [15]. Also in late bilinguals with various proficiency, asymmetrical brain activity for forward or backward in phonological judgement was observed [13]. In the study conducted by Van Heuven et al. (2010), backward (from L2 to L1) had no significant increase in brain activity; whereas, forward (from L1 to L2) elicited greater activation in the right PFC, left IFG, left STG, ACC, and caudate nucleus in phonological judgement (switching direction sensitivity) [15]. In late and low proficient bilinguals pre-SMA/ACC was also found to be involved in suppression of L1 to make L2 output [11] and forward language switching (L1 to L2).

<span style="font-family: Arial,Helvetica,sans-serif; font-size: 110%;">__ 2-4 Early Bilingual studies __
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;">Most studies provided profiles of language proficiency yet had insufficient information regarding AoA. Ambiguity and lack of comprehension of the critical or sensitive period might blur the confounding effect of age. Lack of participants could have also been restricting.

<span style="color: #000000; font-family: Arial,Helvetica,sans-serif;">2-4a Early and high proficiency
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;">Highly proficient bilinguals have no difficulty understanding and producing complex sentences in L2. In early proficient bilinguals, individuals practice and continue using two languages interchangeably from birth or infancy [17]. Garbin et al. (2011) chose individuals with very early AoA to eliminate age as a confounding factor; all of their candidates acquired a second language before the age of 4 and had sufficient experience with both languages [17]. This experience accompanied with a special sociolinguistic environment- where an individual went through language switching throughout life- developed high proficiency in both languages [17]. This filtering could have been an attempt to seek for greater automatization. <span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;"> They found greater activation in the head of the left caudate for forward switching and pre-SMA/ACC for backward switching [17]. This is a contradiction to what was previously found by Wang et al. (2007) and Van Heuven et al. (2008) (see above varying proficiency). Overall, Garbin et al. (2011) showed different neural networks in early and high proficient bilinguals.

=<span style="font-family: Arial,Helvetica,sans-serif; font-size: 120%;">** 3 Cognitive Reserve ** =

<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;">In the course of diagnosing AD and other neurodegenerative disorders, clinicians are often faced with individuals with considerable brain atrophy yet exhibiting less cognitive functional disorders. This mismatch is attributed to the concepts of brain reserve and cognitive reserve.

<span style="color: #000000; font-family: Arial,Helvetica,sans-serif;">3-1a Fractional Anisotropy (FA)
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;">Using diffusion tensor imaging (DTI), FA values can give insight to white matter integrity since its values are known to be associated with variations in axon features (number, size, degree of myelination) and the FA values correlate with information transmission properties and cognitive and information processing speed.

<span style="color: #000000; font-family: Arial,Helvetica,sans-serif;">3-1b White matter tracts in early childhood
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;"> White matter tracts in bilinguals and monolinguals showed different FA values [19]. In the study conducted by Mohades et al. (2012), fourty early or under teenage children were grouped into three categories: simultaneous bilinguals (exposed to both language since birth), sequential bilinguals (acquired L2 after the age of 3), and monolinguals. Out of four major white matter pathways, two of them showed no significant differences while the other two showed differences. Among the two pathways with distinct FA values, left inferior occipitofrontal fasciculus (lIFOF) in simultaneous bilingual children had higher mean FA values compared to monolingual and sequential bilingual children. Whereas,the anterior part of the corpus callosum projecting to the orbital lobe (AC-OL) in simultaneous bilingual children had lower mean FA values than monolinguals. This provided the first evidence of white matter microstructures exhibiting adaptation related to bilingualism.



<span style="color: #000000; font-family: Arial,Helvetica,sans-serif;">3-1c White matter integrity in adulthood
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;">In the study conducted by Luk et al. (2011), higher FA mean values in the corpus callosum (CC) extending to the superior and inferior longitudinal fasciculi were found in lifelong bilingual older adults. Bilinguals’ higher functional and structural connectivity was the first evidence that well-kept white matter integrity was related to lifelong naturally occurring experience. Their work provided supporting evidence of lifelong bilinguals having higher levels of cognitive control than comparable monolinguals. This may provide a neural basis for ‘brain reserve’.

<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 110%;"> media type="custom" key="13703566" align="right"

<span style="font-family: Arial,Helvetica,sans-serif; font-size: 110%;">__ 3-2 Alzheimer’s Disease __
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;"> In a recent study conducted by Schweizer et al. (2011), when monolingual and bilingual AD patients were matched for cognitive performance and number of years of education, and compared, it was found that bilingual patients exhibited substantially greater amounts of brain atrophy than monolingual patients in areas traditionally used to distinguish AD patients [18]. Researchers selected monolingual and bilingual AD patients who had similar years of education and cognitive functions measured by Behavioral Neurology Assessment (BNA). BNA measures cognitive functions such as memory, language attention, visuospatial function, naming and executive function. There were no significant difference of BNA and other behavioral test scores between monolinguals and bilinguals; their cognitive functional disorders were at the same level. However, patients' non-contrast CT head scan revealed that bilingual patients' had greater atrophy in Medial Temporal Lobes (MTL). CT scans showed that bilinguals had significantly higher atrophy in left, right and largest temporal horn. This showed bilingualism as a possible protective measure, since they exhibited higher level of cognitive function than what would have been predicted from their level of atrophy [18]. Bilingual patients had the mismatch of brain atrophy and cognitive functional disorders. To put it simply, bilingual patients were capable of delaying the cognitive symptoms even though their atrophy were greater.

=<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 120%;">**4 Criticism** =

<span style="font-family: Arial,Helvetica,sans-serif;">__ 4-1 Brain imaging methods __
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;"> Vascular brain imaging measures, such as fMRI, are only indirect measures of neuronal activity relying on the BOLD signals. This response is slower with a time-lag of 6 seconds after stimulus onset. While temporal resolution is low, that of spatial resolution, like event-related potentials (ERP), is high [3]. Thus, it is crucial to combine data from two imaging techniques.

<span style="font-family: Arial,Helvetica,sans-serif;">__ 4-2 Research paradigms and designs. __
<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px;"> Bilingual studies in neuroimaging and behavioral studies use different research paradigms. While behavioral studies conducted make complex comparisons using a variety of sentences, neuroimaging studies were restricted to limited complexity due to movement related artifacts such as eye movements. For neuroimaging studies, common methods were short paired words comparisons [16]. Only in a few recent studies were these accompanied by behavioral data collections in a separate session [16].

Introduction Second Language Acquisition Language acquisition has been intensely studied since it is a critical developmental stage. It is a very demanding cognitive task as it involves learning of memorization of tens of thousands of words in a language’s vocabulary, recognition based on visual and auditory cues, acquisition of literacy (for example, reading comprehension) and manipulation of prosody. From the neuroscience perspective, language acquisition therefore yields a great deal of questions of how different cognitive processes are accomplished by underlying neurological pathways. Studying of such underlying mechanisms has been difficult due to the fact that spoken and written language is an exclusively human behavior and the lack of appropriate animal models limits researchers to non-invasive methods. To date, our knowledge regarding language acquisition involves specific brain regions that have been observed to show increased activity (based on brain imaging) during certain linguistic tasks. Moreover, second language acquisition is defined as learning of a non-native language. Many studies, using non-invasive brain imaging, tried to compare the neurological activities involved in manipulation of the two languages. It has been suggested that, on the neurobiological basis, the acquisition and manipulation of the second language is at least in part separate from the native language.1 Lastly, several studies showed levels of activity in certain brain region seem to predict the proficiency in cognitive verbal tasks.


 * 1) Native Language Acquisition

Numerous studies of functional imaging have mapped out major cortical regions activated during linguistic processes. Since language is a multifaceted skill, different aspects of linguistic knowledge involve different neural correlates needed for different computational demands and therefore elicit different patterns of activation of the cortex.
 * 1.1 **** a. Cortical Organization **
 * Lexicon ** (inventory of vocabulary): Studies investigating neurological processes involving lexical tasks elicits high levels of activation in the Broca’s Area (left Inferior Frontal Gyrus), Wernicke’s area (left Superior Temporal) and fusiform gyri, 1 all of which constitute the classical model of language faculty within the brain.
 * Syntax ** (grammatical rules of a language): Summarizing from functional imaging and lesion studies, Friederici’s model of syntax neural processes involves three stages: the first stage involves superior temporal gyrus, the second stage of lexical and syntactic integration involves Broca’s area (middle temporal gyrus, inferior frontal gyrus) and the third stage of syntax reanalysis involves the activation of posterior superior temporal gyrus and the basal ganglia. 2

The majority of brain activity involved in language is elicited in the left hemisphere. Many functional imaging and Event Related Potential (EPR) studies show asymmetrical activation of the two hemispheres during linguistic tasks (e.g. speech production.) 3 Generally, neurological processes are much more frequently represented in the left hemisphere instead of the right hemisphere. Cerebral dominance, therefore, refers to the specialization of one hemisphere managing certain functions or computational tasks. 4 Cerebral lateralization of language function begins at the age of 5 5 and the on-set of lateralization is unclear until recently, Badzakova suggested that a single gene may be responsible for lateralization of many functions in the human brain. 4
 * 1.1 **** b. Functional Lateralization **

2 . Second Language Acquisition As mentioned earlier, Age of Acquisition (AOA) of second language (L2) is a significant influencing factor for acquisition efficiency, proficiency and the neural representation of L2. Different languages can exceptionally vary in terms of grammar and vocabulary (e.g. Spanish vs. Korean,) therefore learning an additional language possibly leads to additional recruitment of neural networks and/or modifications of existing ones. Through functional imaging studies, second language acquisition and utilization share most of the same neural correlates representing the native language. At the same time, however, scientists noted the second language of a late learner elicit higher activation in a few key regions and demonstrate a much less degree of functional lateralization.


 * 2.1 a. Cortical Organization **

__ Convergent neural networks: __ Most brain imaging studies agree that people acquired a second language at later stages of life (late bilinguals) share the basic linguistic cortical representations with people who only know one language. 6 Some studies such as conducted by Illes et al. also showed that the majority of neurological activations by L1 and L2 occur in the same regions in late bilinguals. 7 These findings suggest that aspects of two languages are likely to be processed in similar ways, perhaps by same populations of neurons in very similar neural circuits. In a study by Park et al., brain activation was compared during verbal tasks of monolingual speakers of English and bilingual speakers of Macedonian (L1) and English(L2). 1 It was found that most of the activated sub-cortex for L2 in bilinguals closely resembles the sub-cortex activated by L1 tasks, which is also very similar to the monolingual control.

__ Divergent neural networks __ : While majority of the sub-cortex activation pattern is the same for the native and the second language, many scientist point out that there are key differences between the bilingual and monolingual brain. In the same study done by Park et al., the investigators found that during verbal tasks involving L2, the inferior and middle frontal gyri, the superior parietal lobule and the angular gyri on the right hemisphere is significantly more activated in the right hemisphere compared to the monolingual brains. 1 Bilingual brain lesion studies show that damage to the frontal cortex will result in inappropriate switching (of a few words) between languages 8, suggesting that this region may be used for organizational demand such as separation of languages. Other additional neural networks are recruited for the additional demands involved in knowing and using one more language.

One of the key findings of the study by Park et al. was that the degree of left hemisphere lateralization is significantly reduced in late bilinguals. 1 During verbal task involving the native language, subjects showed more activity in a number of right hemisphere cortical regions compared to the monolinguistic controls (such as right superior occipital gyrus, superior parietal lobule, insular lobe and middle temporal gyrus.) When utilizing L2, the bilingual subjects showed greater bilateral activation in superior temporal gyrus and other regions such as right inferior frontal occipital gyrus and the right cerebellum. 1 Such change in lateralization and recruitment of right cortical structures can be the response to new storage or processing demands of a new language. However, some of the activated right hemisphere structures are not previously known to be involved in linguistic functions and they could be recruited by higher computational demands of dealing with a new language (depending on the L2 proficiency.)
 * 2.1. b. Degree of lateralization: **

The level of proficiency of the second language is also an important moderating factor. Different bilingual individuals with varied levels of proficiency in L2 will different neurological representations.
 * 1) Proficiency

One study by Abutalebi et al. found that the patterns of activation were different between individuals of high and low language proficiencies 9. It was concluded, therefore that proficiency (rather than age) was the most important factor determining the linguistic cortical organization and neurological processing. In terms of syntactic processing, studies have shown that as the language proficiency of a late bilingual improves, activation level increases in superior temporal gyrus at the expense of activity of inferior frontal gyrus. This trade off of activity is not observed in monolingustic individuals.
 * 3.1 **** Cortical organization of L2 also depends on proficiency **

A study by Tan et al. has demonstrated that during verbal tasks, the activity levels in left caudate nucleus and fusiform gyrus effectively predict individual’s proficiency level at L2 and their language learning ability (confirmed by a follow up experiment.) 10 Late bilinguals that had higher levels of activation in those regions also had performed better in L2 verbal tasks. It is proposed that activation of the observed caudate-fusiform network is responsible for suppression of native language during the second language learning, and thus makes L2 acquisition more effective. This hypothesis is supported by the fact that involuntary and inappropriate language switch was observed in multilingual patients with caudate damage.
 * 3.2 **** A neurological marker for proficiency **


 * 1) 1. Park, H.R., Badzakova-Trajkov, G. & Waldie, K.E. Language lateralisation in late proficient bilinguals: A lexical decision fMRI study. //Neuropsychologia// **50**, 688-695 (2012).
 * 2) 2. Friederici, A.D. & Kotz, S.A. The brain basis of syntactic processes: functional imaging and lesion studies. //NeuroImage// **20 Suppl 1**, S8-17 (2003).
 * 3) 3. Toga, A.W. & Thompson, P.M. Mapping brain asymmetry. //Nature reviews. Neuroscience// **4**, 37-48 (2003).
 * 4) 4. Badzakova-Trajkov, G., Haberling, I.S., Roberts, R.P. & Corballis, M.C. Cerebral asymmetries: complementary and independent processes. //PloS one// **5**, e9682 (2010).
 * 5) 5. Szaflarski, J.P., Holland, S.K., Schmithorst, V.J. & Byars, A.W. fMRI study of language lateralization in children and adults. //Human brain mapping// **27**, 202-212 (2006).
 * 6) 6. Brauer, J., Anwander, A. & Friederici, A.D. Neuroanatomical prerequisites for language functions in the maturing brain. //Cereb Cortex// **21**, 459-466 (2011).
 * 7) 7. Illes, J.//, et al.// Convergent cortical representation of semantic processing in bilinguals. //Brain and language// **70**, 347-363 (1999).
 * 8) 8. Fabbro, F., Skrap, M. & Aglioti, S. Pathological switching between languages after frontal lesions in a bilingual patient. //Journal of neurology, neurosurgery, and psychiatry// **68**, 650-652 (2000).
 * 9) 9. Abutalebi, J. Neural aspects of second language representation and language control. //Acta psychologica// **128**, 466-478 (2008).
 * 10) 10. Tan, L.H.//, et al.// Activity levels in the left hemisphere caudate-fusiform circuit predict how well a second language will be learned. //Proceedings of the National Academy of Sciences of the United States of America// **108**, 2540-2544 (2011).

=<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 90%;">Reference: =

<span style="color: #000000; font-family: Arial,Helvetica,sans-serif; font-size: 12px;">

<span style="color: #000000; display: block; font-family: Arial,Helvetica,sans-serif; font-size: 14.66px; height: 1px; left: -40px; overflow-x: hidden; overflow-y: hidden; position: absolute; top: 1660.5px; width: 1px;"> Table 1. Demographic and behavioral characteristics of monolingual and bilingual patients.