Neurocognition of Language/The Neurocognition of Bilingualism

Introduction
The previous chapter has dealt with language acquisition in monolingual infants and young children by introducing distinct developmental stages. On that account the interested reader might have wondered whether these developmental stages are similar in bilingual children and whether there are in general differences between bilingual and monolingual language processing and production. This chapter seeks to answer questions in this regard.

First, before introducing neurological evidence based on experiments employing electroencephalography (EEG) or functional magnetic resonance imaging (fMRI), the term bilingualism itself should be outlined and defined in detail, as there are many different approaches to do so, which have been more or less liberal (Obler & Gjerlow, 1999). Did you know, for instance, that someone who has learnt English at school could also be called ‘bilingual’? Subsequently, aphasia in bilinguals will be introduced, as studies on aphasic bilinguals have been the first attempt to localize multiple languages in the brain (Abutalebi, Cappa, & Perani, 2001). On that account different kinds of recovery patterns will be introduced.

Then, one section is concerned with EEG studies on phonetic processing as well as semantic and syntactic processing in bilinguals and results will be compared to monolingual processing of these events (c.f. Moreno, Rodriguez-Fornells, & Laine, 2008). This section will be followed by studies on the neural activation due to different language input. At the end, the question whether age of acquisition or proficiency account for differences in the neural organization in bilinguals should be answered by this chapter.

Definition: Bilingualism
It is a common belief that young children can learn another language much faster than older children or adults (Abutalebi et al., 2001; Halsband, 2006; Harley, 2008; Obler & Gjerlow, 1999). Moreover, it has been reported that infants can adjust more easily to different accents, as they still are able to integrate and distinguish the intonation of the other language, i.e. children are able to regulate their articulatory tract, in contrast to adults (Halsband, 2006; Obler & Gjerlow, 1999). In general, phonological and morphological aspects appear more difficult to be acquired later in life, albeit it is the opposite for the acquisition of a lexicon (Abutalebi et al., 2001). Thus, should only children up to a certain age be entitled to receive the label bilingual or should the importance of acquiring a lexicon in a foreign language be emphasized further?

As already mentioned in the introduction of this chapter, the term bilingualism has encountered many different kinds of definitions, which have been more or less restrictive (Obler & Gjerlow, 1999). On the one hand, some people call a person ‘bilingual’ if he or she is able to communicate in another language than their native one, regardless of the age of acquisition or proficiency (Abutalebi et al., 2001; Fabbro, 2001a; Obler & Gjerlow, 1999; Paradis, 2003). But on the other hand, people might rather tend to call another person ‘bilingual’ only if that person is able to communicate in another language like a monolingual native speaker of that language (Obler & Gjerlow, 1999). Yet, as some researchers have pointed out, bilingualism does imply the fact that a bilingual individual combines the knowledge of two monolingual persons, which means the knowledge in both languages does not have to be perfect (Fabro, 2001a). Instead, each bilingual consists of a "unique and complex neural system, which may differ in individual cases" (Abutalebi et al., 2001, p. 188). Consequently, the following definition taken from Harley (2008) appears quite reasonable, where he defines bilinguals as “someone brought up in a culture where they are exposed to two languages from birth” (p. 153) with no need to be fluent in both languages, but to be at least very proficient in the second language. With this kind of definition in mind a lot of people could be called bilinguals, as it has become very common in certain parts of the world to be exposed to more than just one language (e.g. Canada, Switzerland, Belgium, USA, India). As it can be seen finding a proper and especially universal definition of the ‘term’ bilingualism is not as easy as it sounds and therefore some scientists have agreed up on a more fine-grained definition of this concept by splitting it up into different categories.

First of all, there is a distinction between simultaneous, early sequential and late bilingualism (Harley, 2008; Paradis, 2003). Simultaneous bilingualism implies that the native language as well as the second language are learnt at the same time, as it is often the case if one parent comes from another country than the other (Harley, 2008). The terms of early sequential and late sequential bilingualism appear somewhat self-explanatory, as they refer to children learning the second language early in childhood and individuals acquiring the second language in adolescence (Harley, 2008). Often there is even just a differentiation between early and late bilingualism as for instance Paradis (2003) describes early bilinguals as individuals who have obtained both languages at the same time during infancy. In contrast to the so-called early bilingual the notion of a so-called late bilingual or fluent second-language speaker exists (Paradis, 2003). Individuals belonging to this group have acquired their second language later in time instead of simultaneously (Paradis, 2003). Eventually, productive and receptive bilingualism are differentiated (Harley, 2008)

As the term bilingualism has been described in detail in this section, the questions now arising are on the one hand what makes bilinguals in general different from monolinguals and on the other what distinguishes early bilinguals from late bilinguals considering the fact that they have acquired their second language at a different point in time? Can a distinction be made on account of neurophysiological evidence? These questions shall be answered in the following parts of this chapter.

Bilingual aphasia
Research on bilingual aphasia has shed light on the organization of language within the brain and enables description of dissociations between languages (Fabbro, 2001b; Obler & Gjerlow, 1999). These studies have been the first attempt to localize different languages in one brain (Abutalebi et al., 2001). In case you have already read other chapters of this book the term aphasia should be familiar to you (see Chapter on Aphasia). As bi- or multilingual individuals are able to comprehend and/ or produce two or more languages the question arises what happens in case of a stroke and subsequent aphasia?

In theory, if each language would be localized or represented in the same brain area, aphasic individuals should display the same deficit in any of their acquired languages (Obler & Gjerlow, 1999). On the same account recoveries should be denoted to the same extend in all of their languages, implying that a parallel recovery of all languages takes place (Fabbro, 2001a; Obler & Gjerlow, 1999; Paradis 2004). Yet, also deviating patterns of recovery have been reported and sometimes only one language seems to be affected (Abutalebi et al., 2001; Paradis, 2004). Some individuals, for instance, recover one language at a time, with one language recovering faster than the over, known as ‘differential recovery’ (Obler & Gjerlow, 1999; Paradis, 2004). The pattern of the so-called ‘antagonistic recovery’ represents a pattern where only one language is available at first, albeit this language disappears or regresses as soon as another language recovers (Fabbro, 2001a; Obler & Gjerlow, 1999; Paradis 2004). If this pattern repeats itself, it is called an ‘alternating antagonism’. Furthermore, sometimes languages are blended into another, i.e. ‘blending recovery’, which means that while speaking the individual mixes up words from both languages (Paradis, 2004). In this regard, there are rules attempting to explain the differential recovery of languages. Either the first-learnt language, which has been acquired first, i.e. native language, is the one, which will be recovered first, or the language that has been employed most at the time of the stroke/ aphasia (Fabbro, 2001a; Obler & Gjerlow, 1999). However, so far there has been no real explanation for different patters of recovery in bilingual aphasics (Fabbro, 2001a), yet it has been inferred that due to the existence of differential recovery patterns, different languages are organized in different areas of the brain (Abutalebi et al., 2001).

So far, it has been shown that aphasia in bilingual individuals entail very different repercussions, as a mixed picture of recovery patterns has been described. Consequently, so far evidence based on examinations with bilingual aphasics points into the direction of different languages being localized in different parts of the brain. However, before investigating this issue from another angle, i.e. by employing imaging techniques, the next section shall outline to what extent monolinguals and bilinguals differ from another in their language processing on basis of ERP components.

Event-related potentials of language processing in bilinguals
The previous chapter on language development has already mentioned the importance of studies of ERP components in language processing and has introduced some of the most relevant components. Hence, this section will shed light on components, which have already been introduced then, from another angle, i.e. from a “bilingual” point of view.

Phonetic discrimination: Mismatch negativity in bilingual infants
From the chapter on language development one should have kept in mind that the ability to distinguish non-native phonetic contrasts declines with age while the ability to distinguish native phonetic contrasts increases, yet this was not true for all subjects (Kuhl, 2004; Kuhl et al., 2008, Moreno, Rodriguez-Fornells, & Laine, 2008). Maybe while having read this subchapter you have asked yourself whether this may also be the case for bilingually raised individuals. On that account, Garcia-Sierra et al. (2011) report that only very few studies have been conducted with bilingual subjects on that issue. Therefore, they conducted a study with Spanish-English bilingual children with the aim of exploring the relationship of ERPs on phonetic discrimination of both languages, exposure to each language and the subsequent bilingual word production skills of each child (Garcia-Sierra et al., 2011).

For their experimental design Garcia-Sierra and her colleagues used a double-oddball paradigm, which is known to elicit an ERP component called mismatch negativity (MMN) in case a deviating phonetic contrast occurs (Moreno et al., 2008; See chapter on language development).

First, the brain responses of bilinguals to phonetic input differed from the responses monolinguals displayed (Garcia-Sierra et al., 2011; Rivera-Gaxiola, Silva-Pereyra, & Kuhl, 2005). There was no neural discrimination of either English or Spanish contrasts at the age of 6 to 9 months, whereas this discrimination was reported at the age of 10 to 12 months in bilinguals (Garcia-Sierra et al., 2011). From there on, bilinguals showed continuous improvement (Garcia-Sierra et al., 2011). Monolingual infants in contrast would have already displayed a significant MMN for both contrasts, native English and non-native Spanish one, at the age of 6 to 9 months and would display an MMN for their native language only at the age of 10 to 12 months (Rivera-Gaxiola et al., 2005).

Furthermore, in order to investigate effects of language exposure, Garcia-Sierra and her colleagues (2011) reported language exposure scores prior to the investigation. This score emanated from in-home interviews and a bilingual questionnaire (Garcia-Sierra et al., 2011). Then, infants had been divided into different groups, according to their given language exposure score, with one representing the high and the other one representing the low exposure group to a language (Garcia-Sierra et al., 2011). Especially infants in the high exposure group of each language displayed strong MMN at the age of 10 to 12 months, which implies a better phone discrimination reflected by an increased negativity (Garcia-Sierra et al., 2011). In opposition to this finding, individuals belonging to the low exposure group did not display an MMN (Garcia-Sierra et al., 2011). Eventually, a follow-up word production test at the age of 15 months indicated a relation between word production and early neural discrimination and exposure to the second language (Garcia-Sierra et al., 2011).

Yet, this study on the phonetic discrimination in bilinguals encounters certain limitations, as bilinguals had not been confronted with the contrast of a third unfamiliar language, as the application of such a third contrast would be expected to give additional information on whether bilingual individuals are in general more likely to remain neural plasticity on the account of language processing or whether this applies for the acquired languages only (Garcia-Sierra et al., 2011).

The up-coming section shall now investigate the aspect of semantic and syntactic processing in bilinguals compared to monolinguals. The former has been related to the N400 component, which is a negativity appearing about 400ms after stimulus onset subsequent to semantic violations (Kutas & Hillyard, 1980). The latter, in contrast, has been related to two components an early left anterior negativity (ELAN) appearing about 150ms and a positivity at 600ms (P600) after stimulus onset subsequent to syntactic violations (Friederici, Hahne, & Mecklinger, 1996).

Semantic and syntactic processing in bilinguals
One of the first studies to investigate semantic processing in bilingual individuals was the study of Ardal and colleagues (Ardal, Donald, Meuter, Muldrew, & Luce, 1990). They investigated 48 bilinguals with their either second language having been acquired between the age of 3 to 10 (early group) or the age of 13 to 17 years (late group) (Ardal, et al., 1990). Aim of the study was to examine ERPs in first and second language of bilinguals and monolinguals subsequent to semantic violations in order to find an explanation for the slowing in lexical decision responses in a second language (Ardal, et al., 1990; Moreno et al., 2008). The authors report that latency of the N400 component is slightly delayed in bilinguals. In fact, monolinguals appear to have the shortest mean latency of the N400, followed by the first language of the bilinguals with their second language having the longest latency (left: 395ms vs 408ms vs 438ms - approximate deviation: 40ms; Ardal, et al., 1990). In addition to that, semantic violations in the second language entail a reduced amplitude of the frontal negativity (Ardal, et al., 1990). In fact if the second language is hardly ever used (low fluency/ less proficient group) the amplitude of this negativity is even smaller (Ardal, et al., 1990). However, contrary to the suggestion of the so-called critical period, no effect was found as a function of age of acquisition (Ardal, et al., 1990).

Finally, Ardal and colleagues conclude that the N400 latency can be regarded as some kind of outer limit when it comes to evaluating the "amount of time it takes the central nervous system to determine the incongruity of a given word" (p. 188). Thus, already before the onset of a N400 the stimulus must have been processed and evaluated as some kind of language by the brain (Ardal, et al., 1990). Consequently, bilinguals could be regarded as displaying a reduced processing speed in both of their languages (Ardal, et al., 1990; Moreno et al., 2008).

Weber-Fox and Neville (1996) provide another study on this subject. They investigated semantic and syntactic processing in 61 Chinese-English bilinguals (Weber-Fox & Neville, 1996). The groups differed from another according to the age of arrival in the US, i.e. point in time when the individual had encountered their second language (Weber-Fox & Neville, 1996). All age groups demonstrated a N400 component, yet with small differences between one another (Weber-Fox & Neville, 1996). The group of bilinguals who came to the US at the age of 1 to 3 years displayed an N400 component similarly to English monolingual subjects (i.e. negativity subsequent to semantic violations in posterior regions which was somewhat larger over the right hemisphere; Weber-Fox & Neville, 1996). Bilinguals in the group of 4 to 6 years of age and of 7 to 10 years of age upon the arrival in the US showed a comparable pattern, albeit no differences with regard to hemisphere could be revealed (Weber-Fox & Neville, 1996). It was assumed that this could also be due to handedness of the subjects (Weber-Fox & Neville, 1996). However, the last group of bilinguals, who entered the US after the age of 16 years, displayed a broader latency window and also a larger mean amplitude (Weber-Fox & Neville, 1996). Subsequently, the results of each group were compared to the performance of monolingual subjects of an earlier study (Weber-Fox & Neville, 1996). The results indicated no significant effect for the younger groups, but peak latencies of the groups of individuals having arrived in the US at the age of 11 to 13 years and older than 16 years were delayed in comparison to monolinguals (Weber-Fox & Neville, 1996). Again, as it has been the case in the study of Ardal and colleagues (1990), Weber-Fox and Neville (1996) indicate that this effect could depict slowing in processing, which could indicate reduced fluency compared to monolinguals. When it comes to syntactic violations a different pattern emerges (Weber-Fox & Neville, 1996). Syntactic processing appears to be more susceptible to age of acquisition of language, as no ELAN is displayed by the youngest group in this experiment (Weber-Fox & Neville, 1996). Furthermore, P600 was already delayed in the youngest group and reduced in subjects of the groups 11-13 years and older than 16 years at arrival in the US. The authors assume that these subjects do not attempt to fix anomalies (Weber-Fox & Neville, 1996).

Altogether, Weber-Fox and Neville (1996) demonstrated that semantic processing differs in individuals in case the second language has been acquired after puberty while syntactic processing appears to be affected much earlier by age of second language acquisition. Consequently, it has been assumed that different neural mechanisms are responsible for semantic and syntactic processing (Hahne & Friederici, 2001). Another study comprising the comparison of (late) bilinguals and monolinguals with respect to semantic and syntactic processing, was conducted by Hahne and Friederici in 2001. They investigated sentence comprehension in learners of a second language by looking at ERP components (Hahne & Friederici, 2001). Their subjects were native Japanese and had started to learn German after puberty (Hahne & Friederici, 2001). In the experimental setting the subjects had to listen to correct, semantically or syntactically incorrect sentences as well as to a combination of both (i.e. semantically and syntactically incorrect sentences; Hahne & Friederici, 2001). Furthermore, subjects had to make judgments on the grammaticality after each sentence (Hahne & Friederici, 2001). The ERPs have eventually been compared to the ERPs on native German speakers (Hahne & Friederici, 2001). Subsequent to correct sentences second language learners displayed a greater positivity than native German speakers (Hahne & Friederici, 2001). The authors suggest this to be a "reflection of greater difficulties in syntactic integration" (Hahne & Friederici, 2001,p.123). Semantic violations, in contrast, entailed a comparable ERP pattern in both groups (N400 component), whereas second language learners did not display significant ERP components as expected after syntactic violations (Figure 1; Hahne & Friederici, 2001). Moreover, as the evaluation of grammaticality was better in the syntactically and semantically combined task than in the syntactic task alone, it was inferred that second language learners mainly rely on the evaluation of semantic information (Hahne & Friederici, 2001).



In conclusion, this section has delineated the differential effects of bilingualism of language processing. While semantic processing appears to be generally intact, but only delayed in individuals as a result of age of acquisition or proficiency (Ardal et al., 1990; Weber-Fox and Neville, 1996), syntactic processing seems to be subject to another kind of neural mechanism (Hahne & Friederici, 2001). Moreno and colleagues (2008) assume that "early negativities, (…) may reflect rather highly automatic processes in native speakers, [which] might not be achieved by late language learners maybe due to developmental constraints in neural plasticity" (p. 489).

Functional Imaging in Bilinguals
The section on bilingual aphasia has demonstrated that the existence of differential recovery patterns has been indicated of different languages being differentially organized within the human brain (Abutalebi et al., 2001). A next step into investigating the neural organization of language has been the adaption of imaging techniques (Abutalebi et al., 2001). Imaging techniques such as fMRI and positron emission tomography (PET), have enabled scientists to look at this topic from another angle. Research on this account shall be reviewed in this final section of the chapter. Also, as it has been the main focus of the section on ERP components in bilingual individuals, we will turn to the question whether age of acquisition or proficiency is more important when it comes to neural differences in language processing.

First, some studies, which have reported differential activation patterns in first and second language processing, shall be introduced. The study of Perani and colleagues (Perani et al.,1996) has described slightly different neural activations for language processing in the first and the second language, while listening to stories in three different languages (first, second and an unknown one) in the PET. Next, in order to discover a neural substrate for the acquisition of a second language DeHaene and colleagues (DeHaene et al., 1997) have conducted a study on the neural representation of first and second language comprehension, since earlier studies so far have failed to do so (DeHaene et al., 1997). While listening to stories in their first language (French) all subjects displayed similar areas of activations, i.e. within the left temporal lobe and especially the left superior temporal sulcus (DeHaene et al., 1997). In contrast to that, listening to stories in the second language (English) resulted in great variable patterns of activations, which differed from subject to subject (DeHaene et al., 1997). Areas included in these activations have been left as well as right temporal and frontal areas, albeit the activation in the left temporal area was smaller for the second in comparison to the first language (DeHaene et al., 1997). Occasionally, activations have been reported for the right hemisphere exclusively in case of second language processing (DeHaene et al., 1997). Based on their evidence the authors concluded that the acquisition of the first language is related to a network within the left hemisphere, whereas no such distinct biological underpinning appears to exist for the acquisition of the second language (DeHaene et al., 1997). Kim, Relkin, Lee and Hirsch (1997) also examined the neural organization of different languages in the brain. As Dehaene and colleagues (1997) Kim et al. (1997) found different activations for first and second language, but only if the second language was acquired in adulthood could these differences be found. Yet, if the second language had been acquired early during development, second language as well as the native language appear to be neurally represented in similar frontal areas (Kim et al., 1997).

The study of Kim and colleagues (1997) already indicates that in some kind the ‘degree’ of bilingualism, i.e. early or late bilingualism, has an influence on whether languages are neurally organized in a similar or different way. However, one might wonder, whether it is only age what differentiates both groups or whether there is another aspect apart from age influencing the neural organization of language processing.

One of the first studies to have employed imaging techniques on the account of bilingual language processing was the one of Klein and colleagues (Klein, Milner, Zatorre, Meyer, & Evans, 1995). They used PET in order to examine word generation in the native (English) and the second language (French) of their subjects (Klein et al., 1995). Moreover, they found a substantial overlap in the frontal activations between both languages, i.e. these activations were identical (Klein et al., 1995). Therefore, this study does not offer any evidence for a different neural representation of first and second language in an individual (Klein et al., 1995). In another study of Perani and colleagues (Perani et al., 1998) no differences in activations could be found either between processing in the native and in the second language. In this study their participants were individuals belonging to two different groups (Perani et al., 1998). One group consisted of Italian-English bilinguals who had started to learn their second language after the age of ten, but gained high proficiency in that second language (Perani et al., 1998). The other group consisted of Spanish-Catalan bilinguals who had obtained their second language before the age of four years (Perani et al., 1998). As no significant differences in could be found between processing the first and second language in both groups, i.e. early and late bilinguals, the assumption of age of acquisition being a crucial factor becomes questionable and the role of attained proficiency in the second language is emphasized (Perani et al., 1998). Comparable results could be obtained by an fMRI study on Chinese-English bilinguals conducted by Chee, Tan and Thiel (1999). Their subjects had to perform a word generation task in each language while being scanned (Chee et al., 1999). Prefrontal, temporal and parietal regions as well as the motor area were activated in both language conditions (Chee et al., 1999). As neither peak activations nor certain activations turned out differ significantly from another, the authors concluded that both Chinese and English do not entail activations in different neural areas (Chee et al., 1999). Also, neither did early and late bilinguals differ from another in their neural pattern leading to the assumption that age of acquisition of language is not relevant for the neural organization of language (Chee et al., 1999). Finally, the authors assume that neural plasticity for language does not change with increasing age, but instead fluency might be a more influential factor in this regard (Chee et al., 1999). This assumption has been supported by the study of Yetkin, Yetkin, Haughton and Cox (1996). They acquired participants who were fluent in two languages (L1 and L2) and learnt a third language where they were not as fluent as in the first two. Participants had to perform a task in all three languages while being scanned (Yetkin et al., 1996). The left temporal lobe was activated in all conditions. Moreover, the number of activated pixels was highest for the language the participants were least fluid in (Figure 2), showing that processing is indeed depending on the level of proficiency or fluency (Paradis, 2003; Yetkin et al., 1996). However, there was no significant difference between first and second language in the activation patterns (Yetkin et al., 1996). Consequently, it could be inferred that "when a language is not used regularly, a larger network may be necessary for its processing" (Abutalebi et al., 2001, p. 183).



In summary, after having reviewed studies on the account of neural organization in bilingual individuals in comparison to monolingual individuals, it has become obvious that a clear statement on whether bilinguals or monolinguals differ from another in their activation patterns cannot be made. Instead, the different definitions of the term bilingualism have to be considered. Actually, at this point the usability of the distinction between early and late bilinguals must be recognized. Yet, it appears that age cannot be the ultimate solution in explaining the emergence of distinct neural organization in bilinguals. It was assumed that areas, which have been related to control functioning in humans, are the ones, which are engaged additionally in the processing of a second language, and function as some kind of compensating mechanism (Abutalebi et al., 2001). Yet, if the individual has improved his/ her skills in the second language so that he/ she has become as proficient as a native-speaker, the bilingual might employ these regions less (Abutalebi et al., 2001). Therefore, the attempt to trace differences in neural activation back to language proficiency appears more reasonable than age of acquisition accounting for these differences.

Summary
The chapter on ‘neurocognitive perspective on bilingualism’ shall introduce the reader to the complex concept of bilingualism and thereby outline different kinds of definitions, which have been developed so far. The importance of distinct meanings of the term ‘bilingualism’ will become evident as soon as studies on the account of ERP components and imaging studies in bilinguals in comparison to monolinguals will be reviewed. The former have indicated that neural plasticity sustains for a longer period in bilingual infants in comparison to monolingual ones (Garcia-Sierra et al., 2011) and that bilinguals tend to rely more on semantic cues instead of syntactic cues when it comes to language processing (Hahne & Friederici, 2001).

Reports on different patterns of recovery in bilingual aphasics and some research employing imaging techniques are reviewed. They suggest that multiple languages in general are represented in different regions of the brain, resulting in different patterns of activation for the latter (Abutalebi et al., 2001; Dehaene et al., 1997; Kim et al., 1997; Perani et al.,1996 ). Yet, as other studies on that subject report no differences between different age groups but relate these differences to proficiency in the second language, another approach attempting to resolve the issue of language organization in bilinguals, is introduced (Chee et al., 1999; Perani et al., 1998; Yetkin et al., 1996)