The cerebral basis of bilinguals has interested researchers for more than a century. Indeed, following the discoveries of language areas in the human brain in the mid-nineteenth century, brain neuroscientists have been trying to give a specific identity also to bilingualism. The origins of this research field may be dated back to early accounts of selective loss and recovery in bilingual aphasia (Pitres, 1896) which gave rise to vivid discussions concerning potential different brain locations for multiple languages. With the recent advent of newly developed functional neuroimaging techniques, we are now able to trace an identikit of the neural correlates of bilingualism. This possibility stems largely from developments in neuroimaging technologies, among them most notably positron emission tomography (PET) and functional magnetic resonance imaging (fMRI). These techniques can capture "in vivo" images of the physiology of language processes. For instance, they may show us how specific regions of the brain "light up" when subjects are engaged in linguistic activities, such as listening to stories or producing words or even switching between different languages. Hence, they provide us with a powerful tool to investigate the neural basis of linguistic processes in the brain, and in particular the cerebral organization of bilingualism. For a basic understanding of its principles, the reader should know that functional neuroimaging techniques measure the regional cerebral blood flow (rCBF). It is widely believed that rCBF reflects synaptic activity following the equation that local increases in blood flow are necessary to replace the energy consumed by neurons. These changes in regional cerebral blood flow have been demonstrated to be closely related to changes in neural activity in both space and time. In functional neuroimaging studies, images of blood flow are collected in at least two different conditions (e.g., while generating words and while at rest). The perfusion data are then compared, in order to find areas where the experimental task is associated with increased cerebral blood flow in comparison with the control task. These areas of increased perfusion are typically referred to as 'activations'. As the reader will realize below we will often refer to these activations throughout the text.
The main focus of the present work is to provide an overview of the most relevant results that have so far been achieved in the field of the neurolinguistics of bilingualism. As the reader will realize there has been an impressive effort to unravel whether two languages share the same neural basis in the human brain. The paper is organize as follows: we will start with a brief introduction about essential neuroanatomy related to language processing and of the contribution of functional neuroimaging to the field of neurolinguistics. We will then consider those neuroimaging studies which have been specifically addressed to enlighten the cerebral organization of bilingualism. In particular, our main aim is not to focus on the question whether two languages are differentially represented in the brain but rather how the brain processes two languages, i.e, the neurodynamics of bilingualisms. This will be done in a schematic manner for each of the major components of language processing, i.e., grammatical processing, phonological processing and lexico-semantic processing. A particular emphasis will be put on the neural mechanism that allow us to "acquire" a second language (L2) with respect to an already existing L1 system. Due to a lack of research papers focusing on prosody and pragmatics in bilingual language processing, these latter aspects are not covered in our review.
1.1 A brief overview on the neuroanatomy of language
There are many ways to study the brain and language relationship. From a neuroscientific standpoint, the first approach was the study of aphasics (i.e., patients that suffer a brain injury or stroke causing language deficits). From the study of aphasia in monolinguals we know that the language areas include large areas of the associative cortex and of the sub-cortical white matter of the frontal, parietal and temporal lobes around the sylvian fissure of the left hemisphere (see figure 1). Noteworthy, among left-handers only one third of individuals have language sites located in the right hemisphere. Language areas are divided into perisylvian and marginal areas (respectively in red and blue in figure 1). The former includes Broca's area (Ba 44, i.e. Brodmann area 44), Wernicke's area (Ba 22) and the arch-shaped fasciculus arcuatus that connects Wernicke's to Broca's area. This set of areas was traditionally thought to be important for the phonological and phonetic-articulatory level of language processing. Wernicke's area is crucial for the identification and correct selection of phonemes, Broca's area for the combination of phonemes into verbal-motor speech sequences. Noteworthy, the perisylvian areas are also the first to maturate during development. Indeed, infants below the age of one year are generally only capable to generate phonemes such as 'ta-ta' or 'ba-ba'. The 'marginal' language areas are located around the perisylvian areas. In the frontal lobe they include Ba 45, which was thought to be important for the combination of words into appropriate morpho-syntactical sequences in order to construct coherent sentences. In posterior brain areas of the left hemisphere the 'marginal' language areas comprise the posterior parts of the middle and inferior temporal lobe (Ba 37), the supramarginal gyrus (Ba 40) and the angular gyrus (Ba 39) located at the temporal-parietal-occipital transition area. The angular gyrus is a multi-modal region of the brain and connects Wernicke's area with other associative cortices, playing a central role in the functioning of the semantic code, making it possible to understand and evoke meaningful words. A role of sub-cortical structures (i.e, grey matter nuclei that are located deep within the white matter of the brain) of the dominant hemisphere, such as the caudate and the thalamus (see Figure 2), in language function has not yet been clearly defined. It has been hypothesized that they play a role in semantic decision processes (Cappa & Abutalebi, 1999). As illustrated in figure 1, the above mentioned areas underlying language were delineated by the anatomo-clinical method , carried out first on the autopsy table and successively with structural neuroimaging (CT and MRI, respectively Computed Tomography and Magnetic Resonance Imaging).
However, nowadays we know that the neural architecture of language is much more extended and complex than the system described above. Indeed, our current knowledge derives largely from the new insights gained by the application of functional neuroimaging techniques in healthy normal subjects. Although some of the first lesion-related findings have never been totally invalidated (e.g., the involvement of Broca's and Wernicke's area, respectively in speech production and comprehension), there are several reasons that make the interpretation of lesion studies complex and, in some cases, quite impossible (Dmonet and colleagues (2005). In the first place, language is a most complex cognitive function and as such it is embedded in complex and highly interconnected networks. Anatomo-clinical studies may demonstrate whether a certain brain region is necessary for a given language component, but not the broader network of which that region may form a part. These networks may be specialized for distinct language functions such as syntactical or lexical processing. It is unlikely that an accidental lesion in a certain brain region knocking out a given linguistic function such as syntactical processing may permit us to localize that function at the lesion site. Consider, for instance, that syntactical processing may be carried out by the interplay of a neural network containing at least three brain areas that could ideally be Broca's area, the left basal ganglia and to a certain degree Wernicke's area. A single lesion in each of these structures or even in the connections among these structures could give rise to the same linguistic deficit and, hence, a sufficient correspondence between a specific lesion site and a symptom may not be established. Given the complexity and the limitations of the classical anatomo-clinical approach to the study of the neural basis of language, functional neuroimaging techniques may offer a better way of evidence since they may provide us with a view of the distributed anatomy of language. Indeed, it is used in normal (i.e., healthy) subjects to identify where the different domains of language processing are localized with considerable precision. Indeed, functional neuroimaging opened a number of new discoveries leading to substantial revisions of traditional concepts. One of those discoveries is a new perspective upon Broca's area: recent imaging evidence reports that the traditional Broca's area in the left inferior frontal gyrus can be functionally sub-divided into three regions: one that is posterior and superior and is involved in the sound structure (phonology) of language; a second, anterior and ventral that is concerned with the meaning of words (semantics); and a third, lying in-between the first two regions, that is involved in meaning conveyed by sentence structure (syntax)(Bookheimer, 2002). Some authors consider even Broca's area as the main neural correlate of morpho-syntactical aspects of language processing (Moro et al., 2001; Tettamanti et al., 2002). Likewise, also for Wernicke's area and the superior temporal gyrus new discoveries are challenging the classical view of language areas. A fundamental review of functional activation studies by Rise (2003) proposed that there are two connections between posterior and anterior language areas (instead of the long-held assumption that only the arcuate fasciculus connects these areas). One involves a 'rostral route' from the anterior temporal cortex via the uncinate fasciculus to the prefrontal cortex (figure 1) while the other involves a 'caudal route' via the well-known arcuate fasciculus to the prefrontal cortex. Although these two routes are unlikely to be independent, the former may process word meaning and the latter, carried out by the arcuate fasciculus, word sound structure (phonetics and phonology). Many other important discoveries have been achieved such as language-related brain activations outside the classical language areas (see reviews in Price, 1998 and Indefrey & Levelt, 2000; Dmonet et al., 2005). However, the discussion of these discoveries would surely go beyond the main scope of the present manuscript. Nonetheless, it is worth citing that functional neuroimaging is also a suitable tool to investigate neural plasticity of language areas, an issue that was totally impossible since 20 years ago. Neural plasticity may be observed, for instances, following learning of an artificial language (Opitz & Friederici, 2004) or following the acquisition of literacy (Abutalebi et al., 2007). In both cases, there is evidence that at the initial stages the role of the left hippocampus (a brain structure that is located in the mesial part of the temporal lobes and is important for making new associations and memories) is predominant underlining its role for language learning. In the later stages, activity shifts more to left hemispheric language areas while once the consolidation of learning is achieved these left hemispheric language areas decrease the mean peak of activity.
Another field in which the contribution of functional neuroimaging has been very fruitful and which is the main focus of our manuscript is the study of bilingualism. Before the advent of functional neuroimaging, many researchers postulated that bilinguals may have languages represented in different brain areas or even in different hemispheres (Albert & Obler, 1978). This picture was essentially based because it is not rare to observe a bilingual aphasic who recovers only with one language, while the other is lost. The latter observation gave rise to the hypothesis that the brain area for one language was damaged and that for the other was not. However, fMRI studies have so far contradicted this assumption (see for review, Perani & Abutalebi, 2005; Indefrey, 2006; Abutalebi 2008). In the next sections we will analytically review the studies that have been carried out in order to address to some crucial questions: (1) Having acquired L1 during childhood, does a second language relies on the same well-organized neurolinguistic L1 system or rather ( 2) will a L2 rely on different neural devices?; and if so (3) are putative neural differences really differences in neuroanatomical representation (i.e., a specific neural representation for L1 and one specific for L2) or rather reflecting differences of how the brain processes two languages (i.e., different usage of auxiliary cognitive resources such as more cognitive control, attentional control and memory resources for processing a L2).
As an introductory example we will just cite the case of grammar processing. Consider for instance, that some authors consider Broca's area along with the left basal ganglia as the main neural correlate of morpho-syntactical aspects of language processing in monolinguals (Moro, Tettamanti, Perani, Donati, Cappa, & Fazio, 2001; Tettamanti, Alkadhi, Moro, Perani, Kollias, & Weniger , 2002; Caplan, Alpert, Waters, & Olivieri, 2000). It is worth underlining that the same brain structures are also involved in implicit memory. Consider now these implications for L2 processing. Will Broca's area, along with the basal ganglia, be responsible only for L1 grammar processing or will it underlie also the processing of L2 grammar? Following Ullman (2001; 2004), L1 is acquired implicitly, mediated perhaps by an innate language learning mechanism only triggered during a critical period whereas an L2 is generally acquired explicitly via formal instruction and, hence, grammatical knowledge for L2 may not be processed through neural structures related to implicit processing such as Broca's area and the basal ganglia, as it is the case for L1 grammar. Hemodynamic studies, such as fMRI investigations, offers a unique opportunity to assess directly the neural representation of L2 subcomponents such as grammar, and hence to test suppositions such as the one raised by Ullman (2001; 2004). In the next sections we will specifically deal with these issues.
2. The Neural Representation of L2
2.1. Acquisition and Processing of L2 Grammar
Based on the assumption of a critical period for language acquisition (Lenneberg, 1967; Johnson & Newport, 1989; Nowak, Komarova, & Niyogi, 2001; Birdsong, 2006; Hernandez & Pi, 2007), many psycholinguists argue that grammar knowledge of a L2 learned after such a critical period may be represented differently, at a cognitive level, from that of L1. The rationale for this assumption is that an L1 is generally acquired implicitly whereas an L2 is acquired explicitly in the sense that its grammar may be taught. The Declarative/Procedural model (Ullman, 2001) which is a re-elaboration of Paradis' implicit/explicit model (Paradis, 1994), provides a rationale for the supposition of differential representation. Its essence is that in normal monolinguals, words are represented in a declarative (i.e., explicit) memory system whereas grammatical rules are represented in a cognitive system that mediates the use of procedures (i.e., implicit memory that is processed without conscious awareness). Moreover, when an L2 is acquired after the critical period it cannot rely on the implicit resources that are used for L1 grammatical processes. Grammatical processing in L2 would be carried out by explicit resources (Ullman, 2001; 2004). Since implicit and explicit knowledge are mediated by distinct neural systems (i.e., Broca's area and the basal ganglia for the first type and left temporal areas for the second), Ullman argues that late L2 acquisition bilinguals would use posterior brain areas whenever they compute L2 grammar (Ullman, 2001). Ullman's supposition on neural differences between L1 and L2 are strongly in contrast with Green's convergence theory (Green, 2003). Green argues instead that the acquisition of L2 arises in the context of an already specified, or partially specified language system (i.e., L1). If Broca's area has learned to compute grammatical processing for L1 during the initial stages of language acquisition, it will perform the same kind of computation also for an L2. However, Green (2003) states that in the initial stages of L2 acquisition, there may a need of additional brain areas for processing the newly acquired L2 but once L2 receives a comparable proficiency, its neural representation will converge to that of the language learned as L1 (Broca's area and nasal ganglia in the case of grammar processing).
However, to what extent is there functional neuroimaging evidence of these different claims? A strong source of evidence in favor of the convergence theory comes from studies of artificial grammar learning. In a landmark study, Opitz & Friederici (2004) investigated with fMRI the acquisition of language-like rules of an artificial language. Increased proficiency for the artificial language was associated with increased recruitment of Broca's area. In a further study, Friederici, Bahlman, Helm, Schubotz & Anwander (2006) confirmed those findings. These results may support the notion that the acquisition of an L2 (albeit an artificial one) is achieved through an existing network mediating syntax in L1. Also in real settings (i.e., studies investigating bilinguals) the results of at least a dozen functional neuroimaging studies have so far contradicted the predictions of the Declarative/Procedural Model (Ullman, 2001; 2004). In fact, the available evidence clearly points out that either low L2 proficiency than high proficiency bilinguals engage for grammatical processing the same neural structures responsible for L1 processing. For example, studies investigating single word processing in L2 such as verb conjugation (Sakai, Miura, Narafu & Muraishi, 2004) and past tense word processing (Tatsuno & Sakai, 2005), showed increased activity around the areas mediating L1 syntax (i.e., Broca's area). Specifically, the study by Sakai et al. (2004) showed that the acquisition of grammatical competences in late bilingual twins is achieved through the same neural structures for processing L1 grammar. Twins were used as subjects in order to investigate whether shared genetic factors influence their language abilities and neural substrates for Japanese (L1) and English (L2). For 2 months, the students participated in intensive training in English verbs (either regular than irregular verbs) as part of their standard classroom education. The authors reported that despite notable differences between L1 and L2 in the students' linguistic knowledge and in their performance in conjugating verbs", Broca's area was responsible for coniugating verbs in L2. (cfr. Sakai et al., 2004, page 1233). These findings suggest a cortical mechanism underlying L2 grammar acquisition identical to that of L1. Similar conclusions were made by Indefrey, Hellwig, Davidson & Gullberg (2005) who investigated Chinese adults immigrated to the Netherlands where they attained L2 (Dutch) classroom courses. Follow-up fMRI during a grammatical judgment task was employed after 3, 6 and 9 months of L2 teaching. Interestingly, the authors reported that as soon as 6 months after the onset of L2 learning, L2 recruited brain areas related to L1 syntax processing such as Broca's area and surrounding areas. Along similar lines, two further fMRI studies in adults reported comparable evidence for shared brain structures underlying native language and the acquisition of L2 grammar (Tettamanti, Alkadhi, Moro, Perani, Kollias, & Weniger, 2002; Musso et al. , 2003). Strikingly, Musso and colleagues (2003) highlighted Broca's area as a crucial structure in the acquisition of rules from a foreign language, but not for rules that are inconsistent with natural languages. A further relevant study is the fMRI investigation by Golestani, Alario, Meriaux, Le Bihan, Dehaene and Pallier (2006) that required moderately fluent late bilinguals in French and English either to read, covertly, words in L1 or in L2 or to produce sentences from these words, again covertly, in either L1 or in L2. There was no systematic difference in the left prefrontal region activated in L1 as compared to L2 and no shifts in the extent of activation with increased syntactical proficiency (measured outside the scanner). But interestingly, the distances between peak activation (i.e., strength of activity) converged with an increase in L2 proficiency. The authors suggested that such convergence for L2 might reflect the use of neural regions more tuned to syntax. A further relevant finding of the Golestani et al.'s study (2006) was that increased proficiency in L2 was correlated to increased involvement of the basal ganglia. Golestani et al. (2006) claim that the involvement of the basal ganglia may be for rule-based processing. However, again, such a finding is, not consistent with the direct application of the Declarative/Procedural model to the bilingual case because the model proposes that the basal ganglia are not involved in syntactic encoding in L2.
To sum up, the available evidence shows that independently of language proficiency (low or high degree of L2 proficiency), late L2 acquisition bilinguals engage for grammatical processing the same neural structures responsible for L1 processing (e.g., Rueschemeyer, Zysset & Friederici, 2006; Suh, Yoon, Lee, Chung, Cho, & Park, 2007). Within these brain structures there may be differences concerning the extension and/or the peak activation of brain activity in the sense that a late learned L2 may recruit additional neural resources around the areas mediating L1 syntax (Wartenburger, Heekeren, Abutalebi, Cappa, Villringer, & Perani, 2003; Sakai et al., 2004; Rueschemeyer, Fiebach, Kempe, & Friederici, 2005; Tatsuno & Sakai, 2005; Golestani et al., 2006; Jeong, Sugiura, Sassa, Haji, Usui, Taira et al., 2007). Noteworthy, early bilinguals (Wartenburger et al., 2003) activate exactly the same brain areas for L1 and L2. Similar results were reported by Luke, Liu, Wai, Wan, & Tan (2002) with subjects that learned L2 at the relative end-state of the critical period and by Hernandez, Hofmann, & Kotz (2007).
What may we conclude from these studies investigating grammatical processing in bilinguals and what are the repercussions on the neural basis of L2 grammatical competence acquisition? First of all, these studies show us that the same set of brain areas responsible for L1 grammar acquisition and processing is also involved for acquiring and processing an L2, independently of age of L2 acquisition and proficiency. Second, there may be differences regarding the extent of brain activity in the sense that a L2 that is processed with a lower degree of proficiency would entail additional areas located around Broca's area (but not in posterior brain areas as suggested by Ullman, 2001). Third, the differences of the extent of brain activity seem to vanish once the proficiency is comparable to that of L2, hence the neural representation of L2 converges to that of L1 as suggested by Green (2003). However, it remains to be determined why exactly there is more activity for an L2 (especially when spoken with a low or not native like proficiency) in Broca's area and surrounding areas. Following Indefrey (2006), bilinguals might compensate for lower efficiency by driving this region more strongly. It is noteworthy, that a recent functional connectivity analysis (i.e., a statistical approach to measure the strength of connections between brain areas) reported that the strength of connections of brain areas related to syntax production is stronger during L2 sentence production than during L1 (Dodel, Golestani, Pallier, ElKouby, Le Bihan, & Poline, 2005). An alternative interpretation is based upon the principles of "the efficiency of neural organization", i.e., the amount of neurons necessary to perform a given task. In the latter case, performance can be negatively correlated with either with the extend than with the peak of activation. Studies reporting negative longitudinal changes (i.e., a decrease of brain activity) such as when a more extended L2 network converges to that of the L1 following a learning period (Sakai et al., 2004; Tatsuno & Sakai, 2005) may support the notion of the efficiency of neural organization.
2.2. Acquisition and Processing of L2 Phonology
Languages differ in many properties, including their phoneme inventories. English, for example, contains a contrast between /r/ and /l/ which is lacking in Japanese, but English lacks the retroflex /D/ versus dental /d/ distinction that is used in Hindi and other South Asian languages. Correctly perceiving and producing the sounds of a second language is a very difficult task, as evidenced both by widespread anecdotal evidence and by a number of formal studies (see for a review Strange, 1995). Problems of this kind are observed even in those who have been exposed to a second language for considerable periods of time and who have therefore had plenty of opportunities to learn its sounds (Flege, Yeni-Komshian, & Liu, 1999; Pallier, Bosch, & Sebastian-Galles, 1997; Sebastian-Galles & Soto-Faraco, 1999). Adults find it difficult to discriminate acoustically similar phonetic contrasts that are not used in their native language whereas young infants discriminate phonetic contrasts even if they are not used in the language they are learning. Therefore, similar to grammar processing (see above), critical or sensitive periods were advocated also for phonological processing (Johnson & Newport, 1989, and see for review Singleton, 2005). However, the ease with which foreign sounds are perceived and produced may vary; the degree of difficulty may depend on the phonetic similarity between L1 and L2 sounds (Flege, 1995).
From a neuroanatomical perspective, in monolinguals phonetic perception and production appears to involve specialized networks in the left hemisphere chief among them the Wernicke's area in the left temporal lobe, the supramarginal and angular gyrus in the parietal lobe, and also Broca's area in the frontal lobe as shown by functional neuroimaging either in adults (Zatorre, Evans, Meyer, & Gjedde, 1992; and see for review Indefrey & Levelt, 2004 and Dmonet et al., 2005) or in young infants (e.g., Dehaene-Lambertz & Baillet, 1998; Dehaene- Lamberts, Dehaene, & Hertz-Pannier, 2002). Recent neuroimaging studies also provide evidence for a shared neural network either for production than perception of phonemes (e.g., Heim & Friederici, 2003). The temporal dynamics in this network during perception shows a primacy for Wernicke's over Broca's area while the reversed pattern occurs for producing phonemes. Heim & Friederici (2003) interpreted this finding with respect to the functionality of the different regions within the shared network, with Wernicke's area being the sound form store and Broca's area a processor necessary to extract relevant phonological information from that store.
As mentioned above, correctly perceiving and producing phonemes of a second language may be a very difficult task, and there may be time-constraints such as critical periods for achieving native-like performance (Flege, Yeni-Komshian, & Liu, 1999). Thus, similarly to grammar processing we may again question whether L2 phonological representations are acquired, stored and processed in a different manner at the brain level.
A series of studies specifically investigating phonological processing are now available in the literature. For instances, Callan, Jones, Callan, & Yamada (2004) investigated with fMRI the neural processes underlying the perception of phonemes. The same phonemes (i.e., English syllables starting with a /r/, /l/, or a vowel) were used for native English speakers and English-L2 speakers (i.e., low proficient Japanese-English bilinguals). Greater activity for second- over native-language speakers during perceptual identification of /r/ and /l/ relative to vowels was found in Broca's area, Wernicke's area, and parietal areas (including the supramarginal gyrus) while more extended activity for native-language speakers was found in the anterior parts of the temporal lobe. In the former case, Callan et al., (2004) suggested that the more extended involvement of neural structures for non-native phoneme identification may be related to the use of internal models of speech articulation and articulatory-orosensory representations that are used to overcome the difficulty in L2 phoneme identification.
A further finding of the Callan et al. study (2004) was the greater activity for /r/ and /l/ relative to vowel perceptual identification between native Japanese and native English speakers in right frontal regions. These findings are consistent with the findings of previous fMRI studies by Callan et al. (2003), Pillai, Araque, Allison, Sethuraman, Loring, Thiruvaiyaru et al. (2003), and Wang, Sereno, Jongman, & Hirsch (2003) who, apart from a greater involvement of left frontal areas, reported also greater recruitment of right frontal regions for L2 phonetic processing. Specifically, in the study of Callan et al. (2003) native Japanese speakers underwent one month of perceptual identification training in order to learn the /r-l/ phonetic contrast in English as compared to a relatively easy /b-g/ contrast. Brain activity was present to a much greater extent in Broca's area, and other frontal lobe areas for the difficult /r-l/ contrast than for the easy /b-g/ contrast even before training. Enhancement in brain activity after relative to before training occurred bilaterally in several brain regions including Wernickes's area, the supramarginal and angular gyrus, in Broca's area and other frontal lobe areas. Noteworthy, the authors reported also the engagement of the basal ganglia, the anterior cingulate cortex and the dorsolateral prefrontal cortex (i.e., two areas in the frontal lobe that are related to cognitive control processes) which might be related to selection and control processes (see below the section related for the neural basis of control processes). Interestingly, the change in brain activity with learning the /r-l/ contrast did not resemble that of the easily discriminated /b-g/ contrast, suggesting that different and/or additional neural processes may be used for processing difficult phonetic contrasts even as performance improves.
Learning non-native phonetic contrasts was also investigated by Golestani & Zatorre (2004) who studied with fMRI ten native English monolinguals while performing an identification task before and after training with a Hindi dental- retroflex non-native phonetic contrast. Successful learning of the non-native phonetic contrast resulted in the recruitment of the same areas involved during the processing of native contrasts, including the left frontal and temporal lobe areas. Additionally, Golestani & Zatorre (2004), showed that the degree of success in learning is accompanied by more efficient neural processing in the Broca's area and the caudate nucleus which, as suggested by the authors, may be related to the allocation of potential control resources for processing the newly learned foreign language speech sounds. In line with this assumption, it was shown that low proficient bilinguals engage Broca's area to a greater extent for L2 during a phonological task (i.e., non-word processing) (Marian, Shildkrot, Blumenfeld, Kaushanskaya, Faroqi-Shah, & Hirsch, 2007).
Moreover, Golestani & Zatorre (2004) found also a positive correlation between a behavioral learning measure and activity in the parietal lobes (i.e., left and right angular gyri): there was less deactivation of these regions for better learners than for worse learners. The strong positive correlation between learning and activation in the left angular gyrus supports the idea that activity in this region is modulated by learning, such that poorer learners deactivate this region more than faster learners do. This observation does well fit with a series of voxel-based morphometry (VBM)  studies conducted by Golestani and co-workers (Golestani, Paus, & Zatorre, 2002; Golestani, Molko, Dehaene, Le Bihan, Pallier, 2007; Golestani, & Pallier, 2007) who have shown structural brain differences within the parietal lobes for fast phonetic learners as compared to slow phonetic learners. In general, fast phonetic learners (i.e., subjects who successfully learn to distinguish or produce phonetic contrasts not present in their native language) have increased white matter density in both parietal lobes but more evident in the left parietal lobe. While the VBM studies of Golestani and coworkers investigated the white matter density of brain volumes, a different VBM study showed gray matter density differences in the left parietal lobe between different groups of bilinguals (Mechelli, Crinion, Noppeney, O'Doherty, Ashburner, Frackowiack, & Price, 2004). Early bilinguals had increased grey matter density within this area. Noteworthy, also late bilinguals may have comparable grey matter density in this brain area, but only when L2 proficiency is high. These striking results of the above mentioned VBM studies fit with data from the classical aphasiological literature, suggesting that the left inferior parietal lobule is the site of the so-called "language talent" area in bilinguals (Poetzl, 1930).
In summary, although the functional neuroimaging literature on phonological processing in bilinguals is rather limited (as compared to the multitude of studies investigating grammatical and lexico-semantic processing), some tentative conclusions may be drawn. Similarly to grammatical processing, the available evidence shows that an L2 is essentially acquired and processed through the same neural structures mediating L1 phonology. As to the extension of brain activity, the studies reviewed above indicate overall that an L2 is in need of greater recruitment of neural resources. One reason for this observation might be that all studies reported above used late and low proficient bilinguals (i.e., Callan et al., 2003; Pillai et al., 2003; Wang et al., 2003; Callan et al., 2004), or monolinguals subjects who for the purpose of the experiment had to learn a phonetic contrast in a foreign language (Golestani & Zatorre, 2004). In the latter case, it is possible that the greater brain activity may reflect the accommodation of a new set of sounds by the existing native speech system rather than acquisition of a nonnative phonetic contrast in a second-language context. In the former case (i.e. L2 speakers), it is plausible that processing sounds in the less proficient language is subserved by less well-tuned neural representations and/or may require greater cognitive effort and it therefore requires greater neuronal activity than processing sounds in L1. Both neural mechanism proposed by Indefrey (2006; and see above) to interpret the stronger activity of L2 may thus hold on also for phonological processing.
2.3. The Lexical-Semantic Domain
Many contradictory findings in early research about the organization of the bilingual language system stemmed from the confusion between lexical and semantic word representations (Kroll, 1993). Studies focused on word meanings mostly produced evidence for a single language system shared by both languages, whereas studies that primarily addressed lexical processes appeared to provide evidence for two distinct, language-specific systems.
The Revised Hierarchical Model (RHM) of Kroll and colleagues (1994) captures the connections between lexical and conceptual representations across languages as learners become more proficient in the L2 and tries to explain how lexical and semantic representations interact when words are translated from L1 to L2 (forward translation) and vice versa (backward translation). It assumes two language-specific lexical stores and a common semantic system to account for how word-to-concept mappings are developed and accessed during language processing. Unlike earlier models (Weinreich, 1953), the lexical and semantic units of the RHM are fully interconnected even though the weight of these connections varies and can be asymmetrical. During early stages of second language acquisition, words in the L2 are hypothesized to be associated to their translation equivalents. The activation of the translation equivalent in L1 facilitates access to meaning for the new L2 words, as words in L1 access directly their respective meanings. The model also assumes that L1 word forms are more strongly connected to the meaning they represent with respect to links between L2 word forms and their semantic representations. On the other hand, the lexical connections between the two word forms are thought to be stronger from L2 to L1 than the other way around, as the acquisition of L2 words is initially mediated by associating them with L1 translations. At the lexical level, there might be some feedback that enables direct translation from the L1 to the L2, but the model assumes that the strong conceptual connections from L1 to meaning will increase the likelihood that translation from the L1 to the L2 is conceptually mediated. Thus, in accordance with this model forward translation is more likely to engage semantic mediation than backward translation, at least during the first stages of L2 language acquisition. However, as L2 proficiency increases, L2 lexico-semantic links become stronger as words appear in many meaningful contexts. These predictions give rise to the so-called 'developmental hypothesis' which assumes that once the degree of proficiency in L2 is high enough a transition from a lexical to a semantic mediation occurs as the connections between L2 words and their semantic representation are assumed to become strong enough to influence backward translation. For instance, while translating from English to Italian during early stages of language learning, the Italian word 'gatto' is hypothesized to be associated to the translation equivalent 'cat' in English. The English word 'cat' will have direct access to the meaning; thus, the word-to-concept link is stronger in the L1 than in the L2. As proficiency increases in the L2, the model hypothesizes that the connection between 'gatto' and the concept will strengthen and the dependency on the L1 translation equivalent will diminish.
The evidence for the RHM comes primarily from experiments on translation performance. According to the RHM, translation in the forward direction, from L1 to L2, should be conceptually mediated, but translation in the opposite direction, from L2 to L1, should be lexically mediated. To test this prediction, Kroll and Stewart (1994) measured the performance of highly proficient Dutch-English bilinguals while translating in both directions. In one condition, the words to be translated were grouped in blocks by semantic category, and in the other, they were randomly mixed. They found that translation from L1 to L2 was slower in the context of the semantically categorized lists than in the mixed conditions, but the semantic manipulation did not affect translation from L2 to L1. Additionally, even for these highly proficient bilinguals, there was a translation asymmetry, with longer latencies in the L1 to L2 direction than in the L2 to L1 direction. As for the developmental hypothesis of the RHM, Talamas, Kroll and Dufour (1999) found greater interference of semantically related false translations in a translation recognition task when participants were highly proficient in L2, whereas less proficient bilinguals suffered more from fillers that shared a related word form. This suggests that the latter group of participants relied more on lexical information for the translation task and that the overlap in the lexical representations between translation equivalents is more important in the early stages of second language acquisition that in later stages. Only with increasing L2 proficiency are L2 learners able to access the meanings of L2 words directly.
The RHM clearly delves into the nature of the lexico-semantic connections between L1 and L2 and makes very explicit predictions about the strength and the weight of interlanguage connections. According to psycholinguistics, in summary, one may conclude that during the early stages of L2 acquisition there may be a dependency on L1 to mediate access to meaning for L2 lexical items (Kroll & Stewart, 1994). This would occur because an L2 is generally acquired with reference to existing L1 concepts (i.e., an L2 is mediated through L1 translation while L1 is concept-mediated). Increasing L2 proficiency may result in a less L1-dependency. Higher levels of proficiency in L2 produce lexical-semantic mental representations that more closely resemble those constructed in L1 and according to Green's 'convergence hypothesis' (2003), many of the qualitative differences between native and L2 speakers may disappear as proficiency increases.
It is therefore interesting to see whether also functional neuroimaging evidence supports the psycholinguistic notions of a shared system for a L2 and L1 when the proficiency of the former increases. Irrespective of the experimental paradigm employed (such as picture naming, verbal fluency, word completion, and word repetition), functional neuroimaging studies consistently reported common activations in similar left frontal and temporo-parietal brain areas, when the degree of L2 proficiency was comparable to L1 (Klein, Zatorre, Milner, Meyer, & Evans, 1994; Klein, Milner, Zatorre, Meyer, & Evans, 1995; Klein, Milner, Zatorre, Zhao, & Nikelski 1999; Chee, Tan, & Thiel, 1999, Hernandez, Martinez, & Kohnert 2000; Hernandez, Dapretto, Mazziotta, & Bookheimer, 2001; Pu, Liu, Spinks, Mahankali, Xiong, Feng, et al., 2001; Ding, Perry, Peng, Ma, Li, Xu, et al., 2003; Perani, Abutalebi, Paulesu, Brambati, Scifo, Cappa, & Fazio, 2003; Klein, Watkins, Zatorre, & Milner, 2006). Noteworthy, the same set of areas are commonly engaged also when monolinguals perform the same task. The activations found for L2 were similar, if not identical, with those underlying L1 lexical retrieval in the same individuals underlining the fact that a bilingual can utilize the same neural structures to perform identical tasks for both languages. Moreover, this happened irrespective of differences in orthography, phonology and syntax among languages (Chee et al., 1999). On the other hand, bilinguals with low proficiency in L2 engaged additional brain activity, mostly in prefrontal areas (Yetkin, Yetkin, Haughton, & Cox, 1996; Chee, Hon, Ling Lee, & Soon, 2001; De Bleser, Dupont, Postler, Bormans, Speelman, Mortelmans, et al., 2003; Pillai et al., 2003; Vingerhoets, Van Borsel, Tesink, van den Noort, Deblaere, & Seurinck, 2003; Briellmann, Saling, Connell,Waites, Abbott, & Jackson, 2004; Marian et al., 2007). It is also worth mentioning that these finding were also confirmed by employing paradigms such as lexical decision and semantic judgment tasks in bilinguals (for example, lexical decision: Illes, Francis, Desmond, Gabrieli, Glover, Poldrack, et al., 1999; Pillai, Araque, Allison, Sethuraman, Loring, Thiruvaiyaru, et al., 2003; semantic judgement: Chee et al., 2001; semantic judgments: Wartenburger et al., 2003, Rueschemeyer et al., 2005; Rueschemeyer et al., 2006). Indeed, low proficient bilinguals activated more extensively the prefrontal cortex.
Although not strictly pertinent to lexico-semantic processing, proficiency related neuroanatomical differences were also reported in tasks such as story comprehension and sentence production. For the comprehension paradigm, Perani et al. (1998) reported that low proficient bilinguals, when compared to high proficient bilinguals, activated less neural substrate in the left temporal lobe suggesting a less elaborated linguistic comprehension of the verbal material in L2. On the other hand, in a sentence production task, Kim, Relkin, Lee, & Hirsch, (1997) observed a differential engagement of Broca's area for late L2 learners as compared to early bilinguals. However, since the authors did not provide a background on the subjects' level of L2 proficiency, it is not easy to properly interpret these data. As previously reported, the age of L2 acquisition seems to have no major role in the lexico-semantic domain (Perani & Abutalebi, 2005; Indefrey, 2006; Abutalebi & Green, 2007; Abutalebi, 2008).
Apart from the degree of language proficiency, also the amount of relative exposure towards a given language may have an impact upon the cerebral organization of languages (Abutalebi, Tettamanti, Perani, 2009). Two of our own fMRI investigations have reported that exposure rather then proficiency may determine specific activity patterns in the bilingual brain (Perani, Abutalebi, Paulesu, Brambati, Scifo et al., 2003; Abutalebi et al., 2007). In the former study, two groups of early high proficient bilinguals living in Barcelona (either Spanish-born and Catalan-born individuals) were scanned with fRMI while performing a word fluency task. Strikingly, Spaniards living in Barcelona (Catalonia) and hence mostly exposed to Catalan, as assessed by an extensive questionnaire, activated less the left prefrontal cortex for word generation in L2 than Catalans, who were less exposed to Spanish (their L2). As to the role of exposure, it is worth underlining that L2 can even replace L1 when bilinguals are no more exposed to L1 (Pallier, Dehaene, Poline, LeBihan, Argenti, Dupoux, & Mehler, 2003); behavioural and fMRI findings carried out in Korean adoptees suggested that indeed L2 might replace L1.
In line with our conclusions on grammatical and phonological processing, we may draw also parallels to the lexical-semantic domain. Again, L2 is essentially processed through the same neural networks underlying L1 processing. L2 related differences are found for low proficiency and/or less exposed bilinguals in terms of greater engagement of the left prefrontal cortex or the selective engagement of prefrontal areas located outside the classical language areas (more anterior to language areas). As to our initial question whether L1 and L2 rely upon a shared system, we may conclude that this holds on also for the lexico-semantic domain since there is no brain evidence that two different system are responsible for two different languages. A remarkable study to this regard is the study by De Bleser et al. (2003) who reported that brain activity for producing L1 words and L2 cognates exactly overlap, while L2 non-cognates were in need of additional brain activity around the same brain areas mediating L1 word production (i.e. left prefrontal cortex).
In general, how may we interpret the overall finding not only in lexical tasks but also, as shown above, during grammatical and phonological tasks, of the greater engagement of the left prefrontal cortex when processing a second language which is not mastered in a native-like fashion? Elsewhere, we have argued that the activity within the prefrontal cortex may reflect executive control over access to short- or long-term memory representations such as grammatical, phonological or lexical representations to assist L2 processing (Abutalebi, 2008). The main idea is that a low proficient L2 will be processed trough neural pathways related to "controlled processing" (i.e. with the active engagement of brain areas related to cognitive control). These brain areas (of which the prefrontal cortex is a main component) are responsible for a conscious control of our actions and in the case of bilingualism they would allow us, for instances, to block potential interferences from the dominant language while speaking the weaker language. On the other hand, a "strong" L2 system (i.e., a high proficient L2) is processed in a more native-like fashion and, hence, in a more automatic manner without the engagement of brain areas related to cognitive control. In the next sections, we will show that cognitive control is a chief component of bilingualism per se, especially during the process of L2 acquisition. For a proper understanding, we have first to characterize the so-called prefrontal effect (Abutalebi & Green, 2007; Abutalebi, 2008).
3. Effects of successful language acquisition and the prefrontal response
As reported above, differential activity found for a low proficient L2 is located (i) in the same L1-related language areas which are, however, more extensively activated (either in the extent than the peak of brain activity) or/and (ii) in brain areas located more anterior to the classical language areas such as in BA 9, 46, 47 (i.e, brain areas located in the prefrontal cortex) that are related to cognitive control (Miller & Cohen, 2001). We have also reported that once a native-like proficiency is achieved these prefrontal activations disappear strongly supporting the neural convergence hypothesis (Green, 2003; Perani and Abutalebi, 2005). Consider also, that the neural representation of an L2 converges with that of an L1 does not deny that in certain cases, the reverse will apply. For instance, when individuals learn to read in L2 first, the substrate for reading L1 will converge with that of L2 (see Abutalebi, Keim, Brambati, Tettamanti, Cappa, De Bleser, and Perani, 2007).
Now, establishing evidence of neural convergence requires that we consider the effects of proficiency on L2 processing. A suitable example is lexical retrieval. An L2 learner will necessarily struggle to produce the correct name for a picture or to name a word and such difficulty may have a number of reasons. The neural connections between the concept, lemma and word form may be still weak and, in general, lexical retrieval may take more time for a low proficient L2 (Snodgrass, 1993; Kroll & Stewart, 1994). Such differences in relative strength of one language system over the other may offer one source to expect the difference in prefrontal activation and potentially the change with increasing L2 proficiency. A second potential reason of difficulty may be interference from a dominant concept name. The L2 learner must block unwanted "prepotent" L1 lexical items during L2 word production. As aforementioned, the "prefrontal effect" may reflect between language competition involving the controlled, rather than the automatic processing of L2. Certainly, once a speaker achieves higher levels of proficiency in L2, overt intrusions (Poulisse & Bongaerts, 1994) become less frequent. A decrease in interference is to be expected to the extent the system underlying the use of L2 is differentiated from that of L1 (Hernandez, Li & McWhinney, 2005). A further strong reason may be that, the actual process of generating a lexical item will be more practiced and so demand less cognitive effort. We may expect then that with increasing proficiency, the L2 learner may be less in need of controlled processing in normal language. In between language competition can be resolved more automatically.
At the brain level, to be less dependent on control mechanisms is translated by the fact that prefrontal activity decreases. A strong message that we wish to provide here to the reader is that it would be false to infer that L2 is differentially represented (at the neural level) from L1 on the basis of neuroimaging data. Consider the following very simple question: Why should the L2 learner have her L2 more extensively represented at the brain level (in terms of more brain areas, i.e., the prefrontal effect)? Consider that a low proficient L2 speaker such as our learner may know only, for example, not more than 1500 words as compared to the 15000 or so words for her native language. Following any principle of neural efficiency it would be a paradox that these 1500 words are represented in larger brain areas. As a consequence, the prefrontal effect cannot be a question of language specific neuroanatomical representations (L2 lexical items differentially represented from L1 lexical items), but necessarily an issue of differences of processing demands! The neural effort to process a weak system such as the L2 of an learner is higher than to process a strong system, such as the L2 of high proficiency L2 speaker.
Consider also that competition may occur between a weak L2 and a prepotent L1. The prefrontal response is linked to the need of controlling the languages in bilinguals, i.e, to resolve competition with inhibitory control (Abutalebi & Green, 2007; 2008). Abutalebi et al., 2000; and Abutalebi & Green (2007) argued for the central role of the prefrontal cortex for language processing in bilinguals because of its strategic position and interconnectivity with a multitude of cortical and subcortical areas (see for review Miller & Cohen, 2001). In line with this assumption, Rodriguez-Fornells, Balaguer, Muente (2006) have proposed that two interrelated control/inhibitory mechanisms might regulate competition in bilinguals: (i) a top-down control inhibitory mechanism implemented by the prefrontal cortex when language schemas are activated and (ii) a prefrontal selection/inhibition mechanism could interact with a more local and bottom-up inhibitory mechanism that regulates the level of activation of the non-target language during competition.
As outlined above, it is important to distinguish between controlled and automatic retrieval. Once sufficient L2 proficiency is achieved, retrieval, correct selection and maintenance of lexical items will become more tuned and more automatic because subjects are familiar with the task and, therefore, the prefrontal response would not be necessary anymore because L2 is processed in a more automatic manner.
In the following section, we will illustrate that controlled processing (we will refer to it as "language control" in bilinguals) is not achieved solely through the intervention of the prefrontal cortex but, like many other cognitive functions, through a dedicated network of brain areas. Interestingly, this network of brain areas is responsible also for language functions specific to bilingualism such as language switching and translation. Since, language control constitutes a particular and specific aspect of bilingualism, we will provide a psycholinguistic background of the phenomenon which is then followed by the neural evidence of how language control is achieved.
4. Controlling two languages
4.1. Psycholinguistic Background
Language control in the field of bilingualism may be best exemplified in the domain of lexical processing because of the multitude of available psycholinguistic studies. Obviously, this does not deny the importance of L2 control during grammatical or phonological tasks (e.g., Segalowitz & Hulstijn, 2005). Indeed, some psycholinguistic studies specifically emphasize controlled processing either in the domain of grammar processing (Doepke (1998) for grammatical competition during language acquisition in early bilinguals and Bordag (2004) for gender competition during grammatical encoding) than in the domain of phonology (i.e., Jared & Kroll, 2003; Jared & Szucs, 2002; Roelofs & Verhoef, 2006).
In general, how do bilingual individuals control the use of their two languages? Bilinguals must possess an effective means to correctly select the intended language. Consider also that bilinguals can also switch between languages on demand or translate between them. How bilinguals perform such tasks can help reveal the nature of the cognitive architecture that supports language use. We will consider the simple case of picture naming that is the experimental paradigm commonly used to investigate language control: the bilingual speaker must identify the picture and , hence, access a conceptual representation of it), understand its meaning (access a semantic and syntactic representation for it, its lemma) and map this meaning onto a suitable word (access a suitable word form) and specify its phonology. As mentioned above, for bilinguals conceptual representations are linked to two different words, i.e., different lexico-semantic representations (Gollan & Kroll, 2001; Kroll & Stewart, 1994; Francis, 1999). A number of psycholinguistic models suggest that in planning to name an object in one language rather than another, individuals specify the language goal as a cue that is part of the conceptual representation of the intended utterance (de Bot & Schreuder, 2003; La Heij, 2005; Green, 1986; Green, 1998; Poulisse & Bongaerts; 1994; Hermans, 2000) The purpose of the cue is to direct activation to lexical representations in the target language butvthere is good evidence that alternative lexico-semantic representations in both languages are active at least briefly (e.g., Colom, 2001; Costa, Miozzo & Caramazza, 1999; Costa & Caramaza, 1999). The current psycholinguistic debate surrounds the issue how, effectively, the bilingual system handles such unintended activation.
In a recent review of the literature, Abutalebi & Green (2008) have outlined three broad possibilities. The first possibility is that the intention to speak one language rather than another, expressed in terms of a language cue in a preverbal specification of the utterance, is sufficient enough to differentiate words in the intended language (La Heij, 2005). A second possibility (Costa et al., 1999; Costa, 2005) supposes that although words in both languages may be active during the planning of an utterance, the intention to speak in one language rather than another effectively restricts selection to words in the target language. Competition on this view only occurs within the target language (i.e., within language competition). An alternative third possibility is that active lexical representations in both languages compete for selection and that such competition is in part managed by mechanisms external to the lexicon (e.g., Green, 1998). The language cue on this account can bias the activation of representations in the target language but does not necessarily prevent candidates in the non-target language competing for selection (i.e., between language competition). Hence, in summary, competition may arise between the goal of speaking in the first language (L1) and the goal of speaking in the second language (L2) (LA Heij, 2005; de Bot & Schreuder, 1993; Green, 1998; Roelofs, 2003); between lemmas in different languages (Green, 1998) or between word forms of the chosen language (Finkbeiner, Gollan & Caramazza, 2006).
Currently no model accounts transparently for the full range of behavioral data. Empirically, there may be no single site at which language selection occurs (see Kroll et al., 2006), To complicate the issue, competition may arise also at more than one site (Green, 1986; 1998; Kroll, Bobb & Wodniecka, 2006) and, if so, different cognitive and neural mechanisms will be involved in managing and in resolving competition. As underlined by Paradis (1994; 2004), bilinguals may initially use deliberate control and monitoring in production tasks. Changes in proficiency may increase the linkage between concepts and lemma items (Kroll & Stewart, 1994) within a language and mediate the feeling of "thinking in a language" and so increase the efficiency of language cue in selecting the appropriate response. At the neural level, as we will see below, different brain regions are engaged in the context of language selection and control and functional imaging data suggest that the precise nature of any conflict may alter with proficiency (Abutalebi & Green, 2007) consistent with a change from controlled to automatic L2 processing. As we will see neuroimaging studies may indicate that control occurs at different levels and may, hence, be useful for resolving the psycholinguistic debate.
4.2. How the brain deals language control in bilinguals
Using various language paradigms, functional neuroimaging studies carried out in bilingual subjects have started characterizing the neural basis of language control processes. We will consider paradigms such as language switching, language translation and language selection because these paradigms have in common an important cognitive load: a current task must be inhibited (i.e, speaking in language A) in favor of the new task (speaking in language B) in the case of switching and translating and withholding a potential prepotent response (i.e., from a non-target dominant language) when selecting items of a weaker language in the case of language selection. Thus, these tasks are thought to heavily rely upon cognitive control mechanisms.
The first study carried out was a PET (i.e., positron emission tomography)  study on bilinguals performing translation and switching tasks based on visually presented words (Price, Green, & von Studnitz, 1999). The authors reported that switching between languages increased activation in Broca's area and the left supramarginal gyrus (i.e. a brain area located in the parietal lobule). Conversely, word translation increased activation in the anterior cingulate cortex and basal ganglia structures. The involvement of the basal ganglia along with activity in the left prefrontal cortex was also reported by the fMRI study of Lehtonen, Laine, Niemi, Thomson, Vorobyev, and Hughdal (2005) during sentence translation in a group of Finnish-Norwegian bilinguals. Language switching in picture naming (compared to non-switching trials) increased fMRI responses in the left prefrontal cortex (Hernandez et al., 2000; Hernandez et al., 2001; Wang et al., 2007). Noteworthy, when switching into the less proficient-language the prefrontal activity is paralleled by activity in the anterior cingulate cortex (Wang et al., 2007).
Two further studies showed that, when controlling interference from the non-target language during naming (Rodriguez-Fornells, van der Lugt, Rotte, Britti, Heinze, & Muente, 2005) and during reading (Rodriguez-Fornells, Rotte, Heinze, Noesselt, & Muente, 2002) in the target language, there is specifically brain activity in the left prefrontal cortex. Similar findings were found by employing an adaptation paradigm (Chee, 2006). In adaptation paradigms, similar stimuli such as items belonging to the same language are contrasted to stimuli belonging to two different languages. For instances, Chee et al. (2003) studied word repetition within and across languages and, noteworthy, only the 'across language' condition led to more extended left prefrontal activity (see also Klein et al., 2006, for similar findings). In a further adaptation paradigm, Crinion, Turner, Grogan, Hanakawa, Noppeney, Devlin, et al., (2006) reported that left caudate activity was sensitive to the "across-language" but not to a "within language" condition.
In line with these findings, Abutalebi, Annoni, Zimine, Pegna, Seghier, Lee-Jahnke, et al. (2008) have shown that the specific activity of the left caudate is confined in bilinguals to the situational context. Only naming in L1 in a bilingual context (i.e., where L2 stimuli may have occurred to create a fully bilingual setting) increased activation in the left caudate and anterior cingulate cortex. Strikingly, this pattern of activity was absent when the same bilingual subjects were placed in a purely monolingual L1 naming context (i.e., by using the same L1 stimuli that occurred also in the bilingual context). As to the validity of these findings and their practical and theoretical application to bilingualism, we underline that the above mentioned studies investigated only language production, and moreover, with the exception of the study of Lehtonen et al. (2005), production was investigated only at the single word level. It is therefore remarkable that in a recent study focusing on the auditory perception of language switches during comprehension of narratives (Abutalebi, Brambati, Annoni, Moro, Cappa, & Perani, 2007) a neural network consisting of the ACC and the left caudate was reported when subjects perceived a switch into the weaker language (i.e. the less exposed language). In general, a language comprehension is thought to be a more passive and automatic task than language production (Abutalebi, Cappa, & Perani, 2001). Ultimately, the finding of a cognitive control network during comprehension strongly highlights the fact that the bilingual brain is equipped with a dedicated cognitive mechanism responsible for the correct selection of the intended language. We also emphasize that there is now ample clinical evidence of the language control mechanism in bilinguals. Clinical studies have consistently reported that lesions to a left prefrontal-basal ganglia circuit, not only cause involuntarily switching between languages, but also may also cause interferences from the non-target language during naming tasks (e.g., Abutalebi, Miozzo, & Cappa, 2000; Marien, Abutalebi, Engelborghs, & De Deyn, 2005; Abutalebi et al., 2009).
In summary, the emerging picture shows that language control in bilinguals is achieved through a set of brain areas, i.e. the caudate nucleus, the prefrontal cortex, the anterior cingulate cortex and eventually the supramarginal gyrus (see figure 3). These brain regions are classically related to executive control in humans (Braver & Barch, 2006; Dosenbach, Visscher, Palmer, Miezin, Wenger, Kang, et al., 2006; Abutalebi & Green, 2007). However, as to our fundamental question if language control is a chief component of bilingual language processing and, if yes, at which level it occurs, the data here review do not favor those psycholinguistic models that to do not postulate inhibition and control mechanism during language selection (Colom, et al, 2003; Costa et al., 1999; Costa & Caramazza, 1999; Roelofs, 2003, Finkbeiner et al., 2006). If a dedicated cognitive control network in the brain is necessary to achieve correctly language selection, than it is desirable that also cognitive models take into consideration these neurobiological evidence. Furthermore, the available neural evidence on language control in bilinguals shows also that multiple neural levels of control (prefrontal - anterior cingulate cortex - basal-ganglia and parietal) are involved and so cognitive accounts that focus on a single level of control (e.g., competition between lemmas or competition between goals) may be insufficient to explain properly lexical retrieval in bilinguals (see for discussion Abutalebi & Green, 2007). As to the effects of proficiency upon the cognitive control mechanism, there are good indications that cognitive control networks are specifically engaged when it comes to the task of processing a low proficient L2. For instances, the studies that disentangled the single switching trials in order to observe whether it is more difficult to switch into L1 or into L2 have so far reported that prefrontal along with anterior cingulate cortex and caudate activity is even more necessary when switching into a less proficient L2 (Wang et al., 2007; Abutalebi, Khateb et al., 2008).
In this manuscript, we have reviewed and analyzed to what extend functional neuroimaging may show us how the brain acquires and processes an L2. First of all, the available neurobiological evidence clearly underlines that those brain areas responsible for L1 processing are also involved in the acquisition of an eventual L2. Moreover, the above reviewed works support a dynamic view concerning language acquisition. "Dynamic" because, as aforementioned, there may be proficiency related changes in the brain, i.e., once an L2 learner has gained sufficienct proficiency, she will engage exactly the same brain areas as she would do for her native language. We underline that this fact was also observed for grammar and phonology acquisition in late L2 learners contrary to what one may expect because of the notion of critical periods. As we have seen, at the initial stages of L2 acquisition there are neural differences particularly prominent in the left prefrontal cortex. Indeed, two kinds of neural differences between L1 and L2 were observed: on the one hand, increased L2-related brain activity in and around the areas mediating L1 such as Broca's area, and on the other, the specific engagement of additional brain areas such as areas related to cognitive control (i.e., left prefrontal cortex, ACC, basal ganglia). In the first case, as proposed by Indefrey (2006), L2 speakers could compensate for lower efficiency by driving these region more strongly and the greater activity observed for L2 may reflect the number of neurons necessary to perform a given task. In the second case, the specific engagement of control structures may underline the nature of L2 processing, that is, more controlled processing as compared to L1 (Abutalebi & Green, 2007). Again, once a native-like proficiency is achieved, our L2 learner may rely less on control structures for L2 processing; we may then suppose that L2 is processed in a more "native-like" fashion. It should be emphasized that these two interpretations are not contradictory in their essence since it is plausible that they may co-exist.
Finally, we have shown that tasks specific to bilingualism such as switching, translation and language selection are carried out through a network dedicated to executive control (to which we refer as the 'language control' network in bilinguals). Again, the role of language proficiency seems to be prominent since low proficient bilinguals activate stronger the language control network. As we have suggested in a recent paper (Abutalebi et al., 2007), the observation of the engagement of this control network may be a neural signature of a "non-native like L2 processing".
To conclude, what type of main message can the field of neurobiology provide to the field of language teaching? According to the evidence here reviewed, it is never too late to learn an L2, even in late learners the brain adapts with a strong correspondence between proficiency and extent of brain activity.
Abutalebi J, Della Rosa PA, Tettamanti M, Green DW, & Cappa SF (2009). Bilingual aphasia and language control: A follow-up fMRI and intrinsic connectivity study. Brain & Language, 109, 141-156.
Abutalebi, J., Annoni, J.M., Seghier, M., Zimine, I., Lee-Jahnke, H., Lazeyras, F., Cappa, S.F., & Khateb A. (2008). Language control and lexical competition in bilinguals: an event-related fMRI study. Cerebral Cortex, 18, 1496-505
Abutalebi, J. (2008). Neural processing of second language representation and control. Acta Psychologica, 128, 466-478.
Abutalebi, J., & Green, D. (2008). Control mechanisms in bilingual language production: Neural evidence from language switching studies. Language and Cognitive Processes, 23, 557 - 582.
Abutalebi, J., Brambati, S. M., Annoni, J. M., Moro, A., Cappa, S. F., & Perani, D. (2007). The neural cost of the auditory perception of language switches: An event-related fMRI study in bilinguals. Journal of Neuroscience, 27, 13762-13769.
Abutalebi, J., & Green, D. (2007). Bilingual language production: The neurocognition of language representation and control. Journal of Neurolinguistics, 20, 242-275.
Abutalebi, J., Keim, R., Brambati S.M., Tettamanti M., Cappa S.F., De Bleser, R., & Perani, D. (2007). Late acquisition of literacy in a native language. Human Brain Mapping, 28, 19-33.
Abutalebi, J., Cappa, S.F., and Perani, D. (2001). The bilingual brain as revealed by functional neuroimaging. Bilingualism: Language and cognition, 4, 179-190.
Abutalebi J, Cappa SF, & Perani D. (2001). The bilingual brain as revealed by functional neuroimaging. Bilingualism: Language and cognition, 4, 179-190.
Albert, M.L., & Obler, L.K. (1978). The bilingual brain: neuropsychological and neurolinguistic aspects of bilingualism. London: Academic Press.
Birdsong, D. (2006). Age and Second Language Acquisition and Processing: A Selective Overview. Language Learning, 56, 9-49.
Bookheimer, S. (2002). Functional MRI of language: new approaches to understanding the cortical organization of semantic processing. Annu Rev Neurosci, 25, 151-188.
Bordag, D. (2004). Interaction of L1 and L2 systems at the level of grammatical encoding: Evidence from picture naming. EUROSLA Yearbook, 4, 203-230.
Briellmann, R.S., Saling, M.M., Connell, A.B., Waites, A.B., Abbott, D.F., & Jackson, G.D. (2004). A high-field functional MRI study of quadrilingual subjects. Brain & Language, 89, 531-542.
Callan, D.E., Tajima, K., Callan, A.M., Kubo, R., Masaki, S., Akahane- Yamada, R. (2003). Learning-induced neural plasticity associated with improved identification performance after training of a difficult second language phonetic contrast. NeuroImage, 19, 113- 124.
Callan, D.E., Jones, J.A., Callan, A.K., & Yamada, R.A. (2004). Phonetic perceptual identification by native- and second-language speakers differentially activates brain regions involved with acoustic phonetic processing and those involved with articulatory-auditory/orosensory internal models. NeuroImage, 22, 1182- 1194.
Caplan, D., Alpert, N., Waters, G., & Olivieri, A. (2000). Activation of Broca's area by syntactic processing under conditions of concurrent articulation. Hum Brain Map, 9, 65-71.
Cappa, S.F., Abutalebi, J. (1999). Subcortical Aphasia. In: Fabbro F, editor. The Concise Encyclopedia of Language Pathology. Oxford (UK): Pergamon Press: p. 319-327.
Chee, M.W.L. (2006). Dissociating language and word meaning in the bilingual brain. Trends in Cognitive Sciences, 10, 527-529.
Chee, M.W.L., Tan, E.W.L., & Thiel, T. (1999). Mandarin and English single word processing studied with functional Magnetic Resonance Imaging. Journal of Neuroscience, 19, 3050-3056.
Chee, M.W.L., Hon, N., Ling Lee, H., & Soon, C.S. (2001). Relative language proficiency modulates BOLD signal change when Bilinguals perform semantic judgments. NeuroImage, 13, 1155-1163.
Chee, M.W.L., Soon, C.S., & Lee, H.L. (2003). Common and Segregated Neuronal Networks for Different Languages Revealed Using Functional Magnetic Resonance Adaptation. Journal of Cognitive Neuroscience, 15, 85-97.
Chee, M.W.L., Soon, C.S., Lee, H.L., & Pallier, C. (2004). Left insula activation: A marker for language attainment in bilinguals. PNAS,101, 15265-15270.
Costa, A., & Caramazza, A. (1999). Is lexical selection in bilinguals language-specific? Further evidence from Spanish-English bilinguals and English-Spanish bilinguals. Bilingualism: Language and Cognition, 2, 231-244.
Costa, A., Miozzo, M., & Caramazza, A. (1999). Lexical selection in bilinguals: Do words in the bilingual's two lexicons compete for selection? Journal of Memory and Language, 41, 365-397.
Crinion, J., Turner, R., Grogan, A., Hanakawa, T., Noppeney, U., Devlin, J.T., Aso, T., Urayama, S., Fukuyama, H., Stockton, K., Usui, K., Green D.W., Price C.J. (2006). Language control in the bilingual brain. Science, 312, 1537-1540.
De Bot, K., & Schreuder, R. (1993). Word production and the bilingual lexicon. In R. Schreuder & B. Weltens (Eds.), The bilingual lexicon (pp. 191-214). Amsterdam/Philadelphia: John Benjamins.
De Bleser, R., Dupont, P., Postler, J., Bormans, G., Speelman, D., Mortelmans, L., Debrock, M. (2003). The organisation of the bilingual lexicon: a PET study. J of Neurolinguistics, 16, 439-456.
Dmonet, J.F., Thierry, G., and Cardebat, D. (2005). Renewal of the neurophysiology of language: functional neuroimaging. Physiological Review, 85, 49-95.
Ding, G. S., Perry, C., Peng, D. L., Ma, L., Li, D. J., Xu, S. Y., et al. (2003). Neural mechanisms underlying semantic and orthographic processing in Chinese-English bilinguals. NeuroReport, 14, 1557-1562.
Dodel, S., Golestani, N., Pallier, C., ElKouby, V., Le Bihan, D., Poline, JB. (2005). Condition-dependent functional connectivity: syntax networks in bilinguals. Phil. Trans. R. Soc. B, 360, 921-935.
Doepke, S. (1998). Competing language structures: the acquisition of verb placement by bilingual German-English children. Journal of Child Language, 25, 555-584.
Finkbeiner, M., Gollan, T.H., & Caramazza, A. (2006). Lexical access in bilingual speakers: What's the (hard) problem? Bilingualism: Language and Cognition, 9, 153-166.
Francis, W.S. (1999) Cognitive integration of language and memory in bilinguals: semantic representation. Psychological Bulletin, 125, 193-222.
Friederici, A.D., Bahlmann, J., Helm, S., Schubotz, R.I., Anwander, A. (2006). The brain differentiates human and non-human grammars: functional localization and structural connectivity. Proceedings of the National Academy of Sciences of the United States of America, 103, 2458-2463.
Golestani, N., Paus, T., & Zatorre, R.J. (2002). Anatomical Correlates of Learning Novel Speech Sounds. Neuron, 35, 997-1010.
Golestani N., & Zatorre, R.J. (2004). Learning new sounds of speech: reallocation of neural substrates. NeuroImage, 21, 494- 506.
Golestani, N., Alario, F.X., Meriaux, S., Le Bihan, D., Dehaene, S., Pallier, C. (2006). Syntax production in bilinguals. Neuropsychologia, 44, 1029-1040.
Golestani, N., & Pallier, C. (2007). Anatomical Correlates of Foreign Speech Sound Production. Cerebral Cortex, 17, 929-934.
Golestani, N., Molko, N., Dehaene, S., LeBihan, D., & Pallier, C. (2007). Brain structure predicts the learning of foreign speech sounds. Cerebral Cortex,17, 575-582.
Gollan, T. H., and Kroll, J.F. (2001) Lexical access in bilinguals. In B. Rapp (Ed.) A handbook of cognitive neuropsychology: what deficits reveal about the human mind (pp. 321-345). New York: Psychology Press.
Green, D.W. (2003). The neural basis of the lexicon and the grammar in L2 acquisition. In R. van Hout, A. Hulk, F. Kuiken and R. Towell (Eds.), The interface between syntax and the lexicon in second language acquisition. John Benjamins: Amsterdam.
Green, D.W. (1986). Control, activation and resource. Brain and Language, 27, 210-223.
Green, D.W. (1998) Mental control of the bilingual lexico-semantic system. Bilingualism: Language and Cognition, 1, 67-81.
Heim, S.T., & Friederici, A.D. (2003). Phonological processing in language production: time course of brain activity. NeuroReport, 14, 2031-2033.
Hermans, D. (2000). Word production in foreign language. Unpublished doctoral dissertation. University of Nijmegen, Nijmegen. The Netherlands.
Hernandez, A.E., Dapretto, M., Mazziotta, J., & Bookheimer, S. (2001). Language switching and language representation in Spanish-English Bilinguals: an fMRI study. Neuroimage, 14, 510-520.
Hernandez, A.E., Martinez, A., & Kohnert, K. (2000). In Search of the Language Switch: An fMRI Study of Picture Naming in Spanish-English Bilinguals. Brain & Language, 73, 421-431.
Hernandez, A.E., Hofmann, J., & Kotz, S.A. (2007). Age of acquisition modulates neural activity for both regular and irregular syntactic functions. NeuroImage, 36, 912-923.
Hernandez, A.E., & Li, P. (2007). Age of acquisition: Its neural and computational mechanisms. Psychol Bull., 133, 638-50.
Illes, J., Francis, W.S., Desmond, J.E., Gabrieli, J.D.E., Glover, G.H., Poldrack, R., Lee, C.J., & Wagner, A.D. (1999). Convergent cortical representation of semantic processing in bilinguals. Brain and Language, 70, 347-363.
Indefrey, P. (2006). A Meta-analysis of Hemodynamic Studies on First and Second Language Processing: Which Suggested Differences Can We Trust and What Do They Mean? Language Learning, 56, 279-304.
Indefrey, P., & Levelt, W.J.M. (2004). The spatial and temporal signatures of word production components. Cognition, 92, 101-144.
Indefrey, P., Hellwig, F., Davidson, D., & Gullberg, M. (2005). Native-like hemodynamic responses during sentence comprehension after six months of learning a new language. Poster presented at the 11th Annual Meeting of the Organization for Human Brain Mapping, Toronto, Canada.
Jared, D., & Szucs, C. (2002). Phonological activation in bilinguals: Evidence from interlingual homograph naming. Bilingualism: Language and Cognition, 5, 225-239.
Jared, D., & Kroll, J.F. (2001 ). Do Bilinguals Activate Phonological Representations in One or Both of Their Languages When Naming Words? Journal of Memory and Language 44, 2-31.
Johnson, J.S., & Newport, E.L. (1989). Critical period effects in second language learning: the influence of maturational state on the acquisition of English as a second language. Cognit. Psychol, 21, 60-99.
Kim, K.H.S., Relkin, N.R., Lee, K.M., & Hirsch, J. (1997). Distinct cortical areas associated with native and second languages. Nature, 388, 171-174.
Klein, D., Zatorre, R., Milner, B., Meyer, E., & Evans, A. (1994). Left putaminal activation when speaking a second language: evidence from PET. Neuroreport, 5, 2295-2297.
Klein, D., Milner, B., Zatorre, R., Meyer, E., & Evans, A. (1995). The neural substrates underlying word generation: A bilingual functional-imaging study. Proc Natl Acad Scien USA, 92, 2899-2903.
Klein, D., Milner, B., Zatorre, R.J., Zhao, V., & Nikelski, J. (1999). Cerebral organization in bilinguals: A PET study of Chinese-English verb generation. NeuroReport, 10, 2841-2846.
Klein, D., Zatorre, J.R., Chen, J.K., Milner, B., Crane, J., Belin, P., & Bouffard, M. (2006). Bilingual brain organization: A functional magnetic resonance adaptation study. NeuroImage, 31, 366 - 375.
Klein, D., Watkins, K.E., Zatorre, R.J., & Milner, B. (2006). Word and Non-word Repetition in Bilingual Subjects: A PET Study. Human Brain Mapping, 27, 153-161.
Kroll. J.F., Bobb, S., & Wodniecka, Z. (2006). Language selectivity is the exception, not the rule: Arguments against a fixed locus of language selection in bilingual speech. Bilingualism: Language and Cognition, 9, 119-135.
Kroll, J.F., & Stewart, E. (1994). Category Interference in Translation and Picture Naming: Evidence for Asymmetric Connections Between Bilingual Memory Representations. Journal of Language and Memory, 33, 149-174.
La Heij, W. (2005). Selection processes in monolingual and bilingual lexical access. In J.F. Kroll and A.M.B. de Groot (Eds.), Handbook of bilingualism: Psycholinguistic approaches (pp. 289-307). New York: Oxford.
Lehtonen, M., Laine, M., Niemi, J., Thomson T., Vorobyev, V.A., and Hughdal, K. (2005). Brain correlates of sentence translation in Finnish-Norwegian bilinguals. NeuroReport, 16, 607-610.
Lenneberg, E.H. (1967). Biological foundations of language. New York: Wiley.
Marian, V., Shildkrot, Y., Blumenfeld, H.K., Kaushanskaya, M., Faroqi-Shah, Y., & Hirsch. (2007). Cortical activation during word processing in late bilinguals: Similarities and differences as revealed by functional magnetic resonance imaging. J of Clinical and Experimental Neuropsychology, 29, 247 - 265.
Marin P, Abutalebi J, Engelborghs S, & De Deyn PP. (2005). Acquired subcortical bilingual aphasia in an early bilingual child: pathophysiology of pathological language switching and language mixing. Neurocase, 11, 385-398.
Mechelli, A., Crinion, J.T., Noppeney, U., O'Doherty, J., Ashburner, J., Frackowiack, R.S., & Price, C.J. (2004). Structural plasticity in the bilingual brain. Nature Brief Communications, 431: 757.
Miller, E.K., & Cohen, J. (2001) An integrative theory of prefrontal cortex function. Annual Review of Neuroscience 24, 167-202.
Moro, A., Tettamanti, M., Perani, D., Donati, C., Cappa, S. F., & Fazio, F. (2001). Syntax and the brain: Disentangling grammar by selective anomalies. Neuroimage, 13, 110-118.
Musso, M., Moro, A., Glauche, V., Rijntjes, M., Reichenbach, J., Buechel, C., & Weiller, C. (2003). 'Broca's area and the language instinct'. Nature Neuroscience, 6, 774-81.
Nowak, M.A., Komarova, N.L., & Niyogi, P. (2001). Evolution of universal grammar. Science, 291, 114-118.
Opitz, B., and Friederici, A.D. (2004). Interactions of the hippocampal system and the prefrontal cortex in learning language-like rules. NeuroImage, 19, 1730-1737.
Poetzl, O. (1930). Aphasie und Mehrsprachigkeit. Zeischrift fuer die gesamte Neurologie und Psychiatrie, 124, 145-162.
Paradis, M. (1994). Neurolingustic aspects of implicit and explict memory: implications for bilingualism. In N. Ellis (Ed.), Implicit and explicit learning of second languages (pp. 393-419). London: Academic Press.
Paradis, M. (2004). A neurolinguistic theory of bilingualism. John Benjamins Publishing Company: Amsterdam/Philadelhpia.
Perani, D., & Abutalebi, J. (2005). Neural basis of first and second language processing. Current Opinion of Neurobiology, 15, 202-206.
Perani, D., Paulesu, E., Sebastian-Galles, N., Dupoux, E., Dehaene, S., Bettinardi, V., Cappa, S.F., Fazio, F., & Mehler, J. (1998). The bilingual brain: Proficiency and age of acquisition of the second language. Brain, 121, 1841-1852.
Perani, D., Abutalebi, J., Paulesu, E., Brambati, S., Scifo, P., Cappa, S.F., & Fazio F. (2003). The role of age of acquisition and language usage in early, high-proficient bilinguals: a fMRI study during verbal fluency. Human Brain Mapping, 19, 170-182.
Pillai, J., Araque, J., Allison, J., Sethuraman, S., Loring, D., Thiruvaiyaru, D., Ison, C., Balan, A, & Lavin, T. (2003). Functional MRI study of semantic and phonological language processing in bilingual subjects: preliminary findings. NeuroImage, 19, 565-576.
Pitres, A. (1895). Etude sur l'aphasie chez les polyglottes. Revue de mdecine, 15, 873-899.
Poulisse, N., and Bongaerts, T. (1994). First language use in second language production. Applied Linguistics, 15, 36-57.
Price, C.J. (2000). The anatomy of language: contributions from functional neuroimaging. J Anat, 197, 335-59.
Price, C.J., Green, D., and von Studnitz, R. A. (1999). Functional imaging study of translation and language switching. Brain, 122, 2221-2236.
Pu, Y., Liu, H.L., Spinks, J.A., Mahankali, S., Xiong, J., Feng, C.M., Tan, L.H., Fox, P.T., & Gao, J.H. (2001). Cerebral hemodynamic response in Chinese (first) and English (second) language processing revealed by event-related functional MRI. Magnetic Resonance Imaging, 19, 643-647.
Rodriguez-Fornells, A., Rotte, M., Heinze, H.J., Noesselt, T., and Muente, T.F. (2002). Brain potential and functional MRI evidence for how to handle two languages with one brain. Nature, 415, 1026-1029.
Rodriguez-Fornells, A., van der Lugt, A.,Rotte, M., Britti, B. Heinze, H.J., and Muente, T.F. (2005). Second language interferes woth word production in fluent bilinguals: Brain potential and Functional imaging evidence. Journal of Cognitive Neuroscience, 17, 422-433.
Roelofs, A. (2003). Goal-referenced selection of verbal action: modelling attentional control in the Stroop task. Psychological Review, 110, 88-125.
Roelofs, A., & Verhoef, K. (2006). Modeling the control of phonological encoding in bilingual speakers. Bilingualism: Language and Cognition (2006), 9: 167-176.
Rueschemeyer, S.A., Fiebach, C.J., Kempe, V. & Friederici, AD. (2005). Processing lexical semantic and syntactic information in first and second language: fMRI evidence from German and Russian. Human Brain Mapping, 25, 266-286.
Rueschemeyer, S.A., Zysset, S., & Friederici, A.D. (2006). Native and non-native reading of sentences: An fMRI experiment. NeuroImage, 31 354 - 365.
Sakai, K.L., Miura, K., Narafu, N., & Muraishi, Y. (2004). Correlated Functional Changes of the Prefrontal Cortex in Twins Induced by Classroom Education of Second Language. Cerebral Cortex, 14, 1233-1239.
Singleton, D. (2005). The Critical Period Hypothesis: A coat of many colours. International Review of Applied Linguistics in Language Teaching, 43, 269 - 285.
Snodgrass, J.G. (1993). Translating versus Picture Naming. In R. Schreuder and B. Weltens (Eds.), The bilingual Lexicon (pp. 83-114). Amsterdam: John Benjamins.
Suh, S., Yoon, H.W., Lee, S., Chung, J.Y., Cho, Z.H., & Park, H.W. (2007). Effects of syntactic complexity in L1 and L2: An fMRI study of Korean-English bilinguals. Brain Research, 1136, 178-189.
Tatsuno, Y., & Sakai, K.L. (2005). Language related activations in the left prefrontal regions are differentially modulated by age, proficiency and task demands. Journal of Neuroscience, 16, 1637-1644.
Ullman, M.T. (2001). A neurocognitive perspective on language: the declarative/procedural model. Nat. Rev. Neurosci, 2, 717-726.
Ullman, M.T. (2004). Contributions of memory circuits to language: the declarative/procedural model. Cognition, 92, 231-270.
Wang, Y., Sereno, J.A., Jongman, A., Hirsch, J., 2003. fMRI evidence for cortical modification during learning of mandarin lexical tone. J. Cogn. Neurosci. 15, 1019- 1027.
Wang, Y., Xue, G., Chen, C., Xue, F., & Dong, Q. (2007). Neural bases of asymmetric language switching in second-language learners: An ER-fMRI study. NeuroImage, 35, 862-870.
Wartenburger, I., Heekeren, H.R., Abutalebi, J., Cappa, S.F., Villringer, A., & Perani, D. (2003). Early Setting of Grammatical Processing in the Bilingual Brain. Neuron, 37, 159-170.
Yetkin, O., Yetkin, F.Z., Haughton, V.M., & Cox, R.W. (1996). Use of functional MR to map language in multilingual volunteers. American Journal of Neuroradiology, 17, 473-477.
Vingerhoets, G., Van Borsel, J., Tesink, C., van den Noort, M., Deblaere, K., Seurinck, R., et al. (2003). Multilingualism: an fMRI study. NeuroImage, 20, 2181-2196.
 The anatomo-clinical method is the correlation method that was widely used in clinical neurology and neuropsychology for which a particular deficit (such as a linguistic deficit) was linked to the location of a specific brain lesion observed on the autopsy table and/or with CT and MRI techniques.
 Voxel-based morphometry is a recent neuroimaging technique by which one can measure the density respectively of white matter and grey matter in the human brain.
 PET is a technique similar to fMRI, but it relies on the intravenous injection of a radiotracer to detect brain activity. It is, hence, invasive and more a more expensive tool to study cognition in humans.