Studies in Second Language Learning and Teaching Department of English Studies, Faculty of Pedagogy and Fine Arts, Adam Mickiewicz University, Kalisz SSLLT 5 (1). 2015. 13-40 doi: 10.14746/ssllt.2015.5.1.2 http://www.ssllt.amu.edu.pl Neurology of foreign language aptitude Adriana Biedroń Pomeranian University, Słupsk, Poland [email protected] Abstract This state-of-the art paper focuses on the poorly explored issue of foreign language aptitude, attempting to present the latest developments in this field and reconcep- tualizations of the construct from the perspective of neuroscience. In accordance with this goal, it first discusses general directions in neurolinguistic research on for- eign language aptitude, starting with the earliest attempts to define the neurological substrate for talent, sources of difficulties in the neurolinguistic research on foreign language aptitude and modern research methods. This is followed by the discussion of the research on the phonology of foreign language aptitude with emphasis on functional and structural studies as well as their consequences for the knowledge of the concept. The subsequent section presents the studies which focus on lexical and morphosyntactic aspects of foreign language aptitude. The paper ends with a dis- cussion of the limitations of contemporary research, the future directions of such research and selected methodological issues. Keywords: foreign language aptitude, neurology, neurolinguistics, individual differences 1. Introduction In the research on individual differences, foreign language aptitude (FL aptitude) has recently become one of the most often debated topics among scholars not 13 Adriana Biedroń only in the field of SLA and language education but also neurolinguistics. The research on the construct has always been considerably inspired by the sciences of cognitive psychology, genetics and neurology; however, only in the recent twenty years have the developments in neurology allowed genuine progress in the field (cf. Long, 2013, p. 33). As early as the 1980s, researchers trying to find the source of exceptional linguistic abilities concentrated on the neurological basis underlying talent for learning languages (Fein & Obler, 1988; Novoa, Fein, & Obler, 1988; Obler, 1989; Schneiderman & Desmarais, 1988a, 1988b). In their classic study of gifted foreign language learners, Schneiderman and Desmarais (1988a, 1988b) suggested that linguistic talent denotes greater neurocognitive flexibility as well as bilateral processing of the brain. Currently, the first part of this intuitive hypothesis referring to brain flexibility has been confirmed by ex- perimental research conducted by Susanne Reiterer and her coworkers (Reiterer, Hu, Sumathi, & Singh, 2013), who, as a result of functional neuroim- aging, provided evidence that phonetically talented subjects are more neu- rocognitively flexible than less gifted individuals. Recently, the knowledge of human cognitive abilities has greatly expanded owing to new discoveries in related science fields such as psychology of individual differences, cognitive science, neuroscience and genetics, with the effect that the construct has been updated and reconceptualized. FL aptitude is now de- fined as a conglomerate of various cognitive abilities (Carroll, 1993; Dörnyei, 2010), subject to the same biological, that is, genetic and neurological, princi- ples as all other abilities, such as mathematical or musical ones. The functioning of the neural system is a basis for individual differences in cognitive abilities. In this respect, there are three sources of ability differentiation: neural conduction velocity, neural efficiency, and gray and white matter volumes. As Jensen (1997, 2002), a major proponent of the hereditarian position, argues, all the variation in mental performance has a biological basis. He explains that there is a negative correlation between the intelligence quotient (IQ), which is a measure of gen- eral cognitive ability, and the reaction time of a person. According to this corre- lation, the higher the IQ level of a person is, the less time he or she needs to solve a problem or to learn something. His arguments rest on interdependen- cies between the results obtained using functional magnetic resonance imaging (fMRI), electroencephalogram (EEG), event related potential (ERP), emission to- mography (PET), and studies of nerve conduction velocity and IQ scores. Thanks to these neuroscientific methods of analysis, it has been well evidenced that intelligence is related to both brain functioning and structure. For example, an fMRI study demonstrated that the general cognitive factor appears to be based on the volume and location of gray matter tissue in the brain (Haier, Jung, Yeo, Head, & Alkire, 2004). Many studies have converged on the view that the frontal 14 Neurology of foreign language aptitude lobes are essential for fluid intelligence, a distinctive role being attributed to the lateral prefrontal cortex (Schoenemann, Sheehan, & Glotzer, 2005). Conse- quently, at the moment there is no academic discussion about individual differ- ences, especially cognitive factors, without neuroscientific research. Applied lin- guists and language educators cannot fail to include these breakthroughs from neuroscience into FL aptitude research. 2. Neurolinguistic research on foreign language aptitude Neurolinguistics has become the most informative and ground-breaking source of knowledge about SLA, complementing earlier dependence on behavioral rec- ords (cf. Long, 2013, p. 33). The number of studies on neurological substrates of FL aptitude is growing and the data obtained from them are becoming more consistent and replicable (cf. Chee, Soon, Lee, & Pallier, 2004; Díaz, Mitterer, Broersma, & Sebastián-Gallés, 2012; Golestani, Price, & Scott, 2011; Hu et al., 2013; Pereda, Reiterer, & Bhattacharya, 2011; Reiterer et al., 2011a; Reiterer, Pereda, & Bhattacharya, 2011b; Sebastián-Gallés et al., 2012). Nevertheless, there are many neglected or poorly investigated areas and, generally, the re- definition of the construct is far from complete. This situation originates from a variety of sources. The most important are the following: the heterogeneity and extension of the FL aptitude construct, the high level of individualization of the brain, and, last but not least, a small number of researchers interested in FL aptitude and specialized in neurolinguistics. One of the most important obstacles in examining FL aptitude is the het- erogeneity of the construct. To start with Carroll’s (1959) classic model of FL aptitude, which conceptualized the construct as comprising four distinct and relatively independent abilities:phonetic coding ability,grammatical sensitivity, inductive language learning ability androte memorization ability, all the succes- sive models (Grigorenko, Sternberg, & Ehrman, 2000; Robinson, 2002; Skehan, 2002; Sparks, Javorsky, Patton, & Ganschow, 1998) have attached extra apti- tudes reflecting current views and advances in the domain of SLA. Skehan’s (2002) aptitude model underscores the importance of incorporating develop- ments in SLA research to update FL aptitude theory, while Robinson’s (2002) aptitude complexes framework highlights the dynamic interactions between FL aptitude profiles, task features and their implications for L2 instruction. More- over, both models lay emphasis on the role of the memory factor in language acquisition. Grigorenko et al.’s (2000, see also Sternberg & Grigorenko, 2000) CANAL-F theory stresses the ability to cope with novelty and ambiguity when learning a foreign language, whereas Sparks et al.’s (1998)linguistic coding dif- ferences hypothesis (LCDH) emphasizes the dynamic nature of FL aptitude and 15 Adriana Biedroń postulates that native language (L1) skills are essential for predicting foreign language (L2) learning. Besides, such constructs as working memory (WM), pho- nological short-term memory and noticing ability have been incorporated in all the contemporary models of FL aptitude (Robinson, 2002; Skehan, 2002), which extends the FL aptitude research to the fields usually associated with psychol- ogy. Particularly, the proposal to include WM in the array of FL aptitudes seems to have gained increasing attention among SLA researchers in recent years (DeKeyser & Koeth, 2011; Doughty, 2013; Juffs & Harrington, 2011; Wen & Skehan, 2011; Williams, 2012). Overall, the whole construct of FL aptitude is highly complex and multifaceted, which is reflected by Dörnyei’s (2005, p. 33) statement that it has become an umbrella-term for a number of cognitive fac- tors creating a composite gauge regarded as the general capacity to master a foreign language. This has effects on the research on the neurology of FL apti- tude, where some mechanisms which serve language learning behavior are bet- ter investigated than others. For example, neural mechanisms for procedural and declarative memory, memory consolidation and attention (Schumann, 2004a, p. 1), phonological abilities (Reiterer et al., 2013), and the congenital na- ture of L1 and L2 aptitude (Díazet al., 2012) are often investigated. Others, for example analytic aptitude required for grammar processing, the ability to learn vocabulary, noticing ability, WM as FL aptitude, pragmatic ability and semantic fluency, remain neglected. It seems that these disproportions largely reflect weak areas in the theory of FL aptitude. Another major problem that complicates the foundation of a unified neu- rological picture of FL aptitude is a high level of the individualization of the brain. According to Schumann (2004b, p. 7), “all brains are different—as differ- ent as faces . . . and these differences have consequences for learning.” Some differences result from genetic inheritance; for example, greater brain plasticity (cf. Díazet al., 2012; Golestani, 2012; Sebastián-Gallés & Díaz,2012;Sebastián- Gallés et al., 2012). Some others are considered adaptive changes in the brain occurring in response to experience (Golestani, Molko, Dehaene, LeBihan, & Pallier, 2007; Green, Crinion, & Price, 2006). Accordingly, high FL aptitude might be a consequence of both inborn functional and structural/ anatomical charac- teristics as well as an individual brain response to an idiosyncratic experience of learning a language. De Bot (2006) expresses his opinion on this interrelation- ship in the following way: There are individuals who will have both exceptional language skills and deviant brain structures. . . . it is likely that learning might have an impact on brain structures, although it is unclear how plastic the brain is and to what extent specific teaching and learning methods might enhance plasticity or make optimal use of it. (p. 130) 16 Neurology of foreign language aptitude According to Schumann (2004b), there are five sources of variation among brains, which result in differences in FL aptitude, namelygenetic,devel- opmental,experiential,degeneracy andindividual appraisal system. His claim is in most part based on classic theories of heritability (cf. Jensen, 1997; Plomin, 1997), ascribing significant genetic contributions to cognitive abilities. Genetic variance in a child attributable to parental genes accounts for about 50% of cor- relation between siblings and is higher for monozygotic twins (about .86) and lower for fraternal twins (.60) and for regular siblings (.48), which means that genes are the most influential factor in the development of cognitive ability. The second source is the specific chemical environment during the embryonic stage of development. As a result, human brains are similarly constructed but differ sig- nificantly at the microstructural level. The third source of variation are the inter- actions with the environment, with the effect that they channel the brain’s anat- omy, that is, the increase of neurons and connections among them. Because each individual has idiosyncratic environmental experience, these influences contrib- ute to additional microstructural variation in the neural structure. The fourth pro- cess which contributes to variation between brains is called degeneracy. This term describes a situation when two or more different neural systems subserve the same goal, that is to say, when the same behavior can be achieved by differ- ent underlying processes. These alternate systems discriminate individual brains (Indefrey & Gullberg, 2006; Schumann, 2004b). The fifth source of variance are idiosyncratic preferences and aversions, that is, an individual appraisal system (Scherer, 1984). Individual experiences and affective reactions are stored in memory and used to evaluate future experiences, and consequently affect indi- vidual choices. Moreover, people seek environments fitting their genotype, which in turn influences their abilities (Jensen, 1997). Jensen (1997) also points to the fact that randomness or luck should be considered another source of variation. The development of FL aptitude might be a consequence of evolutionary selection processes (Schumann, 2004b), which means that individuals can be differently prepared to respond to environmental changes, and, consequently, to survive and to transmit their genes. Adaptation to the environment can gen- erate ahypertrophy, which is a structural (anatomical) abnormality in the brain (Van den Noort, Nordby, Bosch, & Hugdahl, 2005), which, in turn, can result in a specific ability. The brain of a talented individual with a particular hypertrophy responds to the learning process strengthening certain neural connections or creating new neural pathways. This, in sequence, facilitates learning, and, con- sequently, the talented individual might achieve high expertise in the field of study (cf. Golestani et al., 2011; Perani, 2005; Reiterer et al., 2013). Generally, neurological differences between foreign language learners, which might be assigned to different domains of FL aptitude, are divided into 17 Adriana Biedroń functional (i.e., those connected with brain activation) andstructural(i.e., those connected with brain anatomy). These differences are associated with distinct aspects and levels of language processing starting with simple perceptual/cog- nitive functions such as nonnative sound learning and articulation, and phonetic expertise, through more complex ones such as WM for verbal and lexical infor- mation, to the most compound processes including reading, syntax, bilingual functioning and executive control over linguistic fluency. Most of the research in the domain of language has focused on brain functioning using such methods as fMRI, EEG, and magnetoencephalography (MEG). However, over the last 17 years the number of studies examining brain structure or a change over time, that is, plasticity, has grown significantly thanks to the development of very ad- vanced technologies such as anatomical magnetic resonance imaging (aMRI) and diffusion tensor imaging (DTI; Golestani, 2012, p. 2). Neuroimaging tech- niques are described in Table 1. Table 1 Neuroimaging techniques Technique Definition PET (positron emission tomogra- Used for localization of different neural functions by means of injection phy) of radioactive tracers. More active brain areas have higher levels of blood flow and, consequently, of the tracer. By creating pictures of the tracer distribution, a neuroscientist can obtain a pattern of brain func- tioning. PET has high spatial resolution (Goswami, 2004: 5-6). fMRI (functional magnetic reso- Gives similar results to PET, but relies on measuring the magnetic reso- nance imaging) nance signal generated by the protons of water molecules in neurons. fMRI has high spatial resolution (Goswami, 2004: 5-6). ERP (event related potential) ERP is, unlike PET and fMRI, based not on localization of neural activity, but on the timing of neural events. ERP has high temporal resolution. Electrodes placed on the skin of the scalp record activity of the brain. This experimental technique is based on EEG (encephalography) (Go- swami, 2004: 5-6). MEG (magnetoencephalography) A diagnostic technique which measures the level of magnetic signals as a result of electrical activity in the brain. MEG has high temporal resolution. (http://psychologydictionary.org/magnetoencephalography-imegl/) aMRI (anatomical magnetic reso- A high resolution technique which can be used to describe the shape, nance imaging) size and integrity of grey and white matter structures in the brain. (http://fmri.ucsd.edu/Research/whatisfmri.html) DTI (diffusion tensor imaging) It can be used to map white matter fibre tracks (http://fmri.ucsd.edu/Research/whatisfmri.html) Traditionally, it is believed that changes in brain functioning are rapid whereas those in brain structure take longer. However, these new methods of brain investigation have revealed that also structural changes can occur rapidly, basically within hours (Golestani, 2012). Another important discovery is that, gen- erally, the same regions which functionally subserve cognitive processes involved 18 Neurology of foreign language aptitude in language processing also structurally correlate with these processes. As a re- sult, a number of anatomical differences have been found in more versus less proficient foreign language learners. For example, Mechelli et al. (2004) discov- ered that the acquisition of multiple languages results in an expansion of grey matter in the left parietal cortex. Green et al. (2006) studied anatomical changes implicated in processing a language among simultaneous interpreters as com- pared to monolingual, bilingual and multilingual speakers. What they found was higher grey matter density in interpreters in three regions: bilateral putamen, the inferior and superior colliculi, and the bilateral dorso-medial thalami, a phe- nomenon ascribed to long-term effects of the acquisition of a very advanced linguistic skill, which, in turn, makes the acquisition of succeeding languages easier. Stein and colleagues’ (Stein et al., 2012) study provided evidence for brain structural plasticity as a result of second language learning. They con- ducted a longitudinal study by means of aMRI on native speakers of English learning German prior to and after five months of learning. As a result, they discovered structural changes over time in the left inferior frontal gyrus and in the left anterior temporal lobe, which positively correlated with individual dif- ferences in the increase in second language proficiency during training. Gener- ally, the differences in the left inferior parietal cortex and in the left inferior frontal cortex associated with bilingualism are related to the age of acquisition and predict second language proficiency (Golestani, 2012, p. 20). Interesting as they are, these studies explain differences in proficiency between learners, but proficiency does not equal aptitude. Accordingly, Reiterer, Pereda and Bhattacharya (2009, p. 98) point to the fact that “language proficiency” is an ambiguous term involving various factors including aptitude for languages. Therefore, most of the studies presented in this review must be interpreted as indirect evidence of differences in FL aptitude. All the above mentioned research provides evidence for brain plasticity as a result of experience. However, many studies offer an alternative interpretation of this phenomenon, tracing the roots of anatomical specificity to genetic factors (cf. Golestani, 2012). The example of a polyglot Emil Krebs (1867-1930), who flu- ently spoke more than 60 languages, is presented as classic evidence for a pecu- liar inborn brain architecture that facilitates FL aptitude. Apparently, the cell structure in his Broca’s area was significantly different from a normal brain cell structure (Amunts, Schleicher, & Zilles, 2004). In contrast, no plausible explanation for talent has been discovered in the brain of a linguistic savant, Christopher (Smith, Tsimpli, Morgan, & Woll,2011). The discussion of the origination of hy- pertrophies will be addressed at greater length in the following section. For the sake of clarity, the following review will present both functional and anatomical studies in the fields of phonology, grammar and lexis with respect to FL aptitude. 19 Adriana Biedroń 3. Neurology of phonological aptitude The phonological aspect is the best investigated of all the components of FL apti- tude (Christiner & Reiterer, 2013; Díaz et al., 2012; Golestani et al., 2011). As far as anatomy is concerned, differences in the phonological cognitive functioning include the auditory cortex, the parietal cortices and the inferior frontal gyrus, all of which are related to such levels of phonetics as auditory processing, the per- ception of nonnative sounds, the use of tonal information, and the ability to imi- tate nonnative sounds. Differences in auditory processing have been found in left Heschl’s gyrus (HG) anatomy, which means that higher gray matter density is as- sociated with better performance (Sutherland et al., 2012; Warrier et al., 2009). A significant factor related to language aptitude is phonetic perception, which is required for phonetic production, accent imitation, verbal WM, as well as semantic perception and production. Many studies have confirmed substantial in- dividual differences among people in the perception, recognition and learning of foreign sounds (Golestani et al., 2007; Golestani, Paus, & Zatorre, 2002; Golestani et al., 2011; Sebastián-Gallés et al., 2012). As a result of the examination of brain structure in expert phoneticians, Golestani and her team discovered that phono- logically talented learners have more grey matter and white matter in parietal re- gions, in particular in the left hemisphere. Their results suggest that this morpho- logical difference is inborn and might have existed before the onset of phonetic training thus affecting career choices of the subjects. As they explain, complemen- tary influences of inborn predispositions and experience-dependent brain pliability interact in determining not only how experience shapes the human brain, but also why some individuals become engaged in certain fields of expertise (Golestani et al., 2011, p. 4213). Left parietal cortex is pertinent to phonetic tasks and is the lo- cation of phonological verbal WM; therefore, the anatomy fundamental for WM in the left auditory cortex also predicts phonological aptitude. The researchers explain the asymmetry in the amount of white matter in more talented learners in terms of greater myelineation, that is, an increase in myelin volume (white matter), which indicates a better isolation of the transport of electric signals, which, in turn, leads to faster and more efficient neural processing vital in learning the phonetics of a language. The researchers conclude that morphological differences in parietal white matter can predict the pace and efficiency of learning new sounds. There is a number of other hypertrophies that differentiate more from less able L2 learners, mostly related to the anatomy of the HG. For example, higher white matter density has been found in the left HG, as well as in a split or a dupli- cate of the HG, in more able learners. In fact, there can be two or three HG per hemisphere. Additionally, the right insula and HG are more superiorly located in slower learners (Golestani et al., 2011). What is more, a larger volume of grey 20 Neurology of foreign language aptitude matter in the HG has been found in musicians, which positively correlates with musical aptitude (Schneider et al., 2002; cf. Christiner & Reiterer, 2013). Generally, a global displacement of components of the language area in the left hemisphere can predict the learning of speech sounds. There is also evidence that variation in perisylvian anatomy is related to oral language ability. Abnormal- ities have been found in children with dyslexia and other language disorders. Ab- normal asymmetry of the planum temporale has been detected in people with poor verbal ability (Golestani et al., 2007). Moreover, an increase in grey matter has been observed in the mid-body of the corpus collosum which connects the two hemispheres in highly proficient L2 speakers (Coggins, Kennedy, & Arm- strong, 2004; Van den Noort, Bosch, & Hugdahl, 2006). Interestingly, the differ- ences lie not only in the auditory cortex, but also in the more general language network and even in the right hemisphere. For example, greater white matter density has been observed in certain visual brain regions, which means that those are also engaged in phonological processing (Golestani et al., 2007). Sebastián-Gallés et al. (2012) examined neuroanatomical markers of indi- vidual differences in vowel perception. They compared brain morphology in two groups of highly proficient early bilinguals, equally proficient in an L2, but dif- fering in their ability to perceive both native and nonnative vowels. Voxel-based morphometry analysis revealed that there is a larger white matter volume in the right insulo/fronto-opercular region in poorer perceptual discriminators of native and nonnative vowels. The higher white matter volumes in poor perceiv- ers indicate a stronger activation of these areas which are used as a compensa- tory mechanism that enhances auditory discrimination abilities. This conclusion accords with similar results obtained by Reiterer et al. (2011a), Reiterer et al. (2011b) and Wong, Perrachione, and Parrish (2007), where a more extended or bilateral activation in poorer language learners was observed. Another group of studies refers to the use of tonal information linguistically. Wong and colleagues (Wong, Chandrasekaran, Garibaldi, & Wong, 2011; Wong et al., 2007; Wong et al., 2008) confirmed larger volume of the left HG in more successful learners using fMRI, aMRI and DTI. Moreover, Wong et al. (2011) found that white matter connectivity in the left temporoparietal region correlated positively with the use of tonal information. Summing up, there is a partial dissociation between the structural correlates of phonetic perception and production (Golestani, 2012, p. 15). Functional studies on phonological processing generally corroborate three hypotheses related to FL aptitude, that is (a) a stronger and bilateral acti- vation of brain areas of less gifted individuals in comparison to those of more gifted ones, (b) the dual genetic/environmental source of aptitude differences, and (c) the common neural basis for L1 and L2 aptitudes (Díaz, Baus, Escera, 21 Adriana Biedroń Costa, & Sebastián-Gallés, 2008; Golestani & Zatorre, 2004; Reiterer et al., 2011a; Reiterer et al., 2011b; Wong et al., 2007). One of the earliest questions asked by neurolinguists was whether neural correlates for an L1 and L2 are the same or different. Most studies have con- verged on the view that unlike an L1, which always activates the same areas in the left hemisphere, an L2 activates a very changeable network of both hemi- spheres (Dehaene et al., 1997). This observation is typically not ascribed to dif- ferences in aptitude but to the age of onset and level of proficiency. In many studies late-onset, low proficiency L2 learners have demonstrated greater right hemisphere activation, whereas areas of L1 and L2 activation tend to overlap in early-onset, more proficient learners (Kim, Relkin, Lee, & Hirsch, 1997). More recently, these results have been replicated by Golestani and Zatorre (2004), Indefrey and Gullberg (2006), Reiterer et al. (2011a); Reiterer et al. (2011b), Se- bastián-Gallés et al. (2012) and Wong et al. (2007), all of whom reported a more extended or bilateral activation in the brains of less successful language learn- ers. Specifically, more active cortical regions in less proficient learners during L2 processing concentrate in the left posterior inferior frontal gyrus (IFG) (Indefrey, 2006; Stowe, 2006; Van den Noort et al., 2005). In Indefrey’s (2006, p. 300) in- terpretation, the IFG is optimized for an L1 and less efficient for an L2. Effort increases activation, which means that learners might compensate for lower efficiency in an L2 by driving this region more strongly or activating a bigger number of neurons to perform a task, whereas automatized activities require less effort, and, consequently, less activation. All of this indicates that the effi- ciency of the neural organization, next to brain anatomy, might establish a neu- rological basis for FL aptitude. Indefrey and Gullberg (2006) postulate that with the increase in L2 proficiency, the processing profile in an L2 becomes similar to an L1. What causes higher activation in lower-proficiency L2 speakers is the in- creased “control effort” (Reiterer, 2009; Reiterer et al., 2011a). Generally, most contemporary researchers choose a moderate view termed partial overlap (Reiterer, 2009, p. 160). According to this opinion, there is a basic core overlap for L1 and L2 processing; however, in all probability, the level of proficiency or fluency triggers brain activation in additional areas for an L2. Golestani and Zatorre (2004) investigated changes in brain activity during phonetic processing by means of fMRI. Their subjects were ten monolingual Eng- lish-speaking individuals, who were scanned during performing an identification task of a sound unknown to them: a Hindi dental retroflex. The fMRI was con- ducted before and after five sessions of training. As a result, they confirmed that the successful learning of a nonnative phonetic contrast causes the employ- ment of the same areas that are active in the processing of native contrasts. More- over, frontal speech regions are less active in successful learners as compared to 22