2. Definitions and models of FL aptitude

FL aptitude is a factor which, next to age of onset (AO), accounts for the largest proportion of variation in the outcomes of FL learning among all individual difference (ID) variables. It concerns a wide variety of cognitively-based learner differences and remains a reliable predictor of success after the closure of sensitive periods (Long Reference Long, Granena and Long2013: 13). A whole gamut of FL aptitudes enclosed in the domain of IDs has attracted the attention of SLA researchers (Abrahamson & Hyltenstam Reference Abrahamsson and Hyltenstam2008; Granena & Long Reference Granena and Long2013; Moyer Reference Moyer2013), with the effect that the construct has been updated and reconceptualized, largely thanks to the developments in the fields of SLA, psychology and neurolinguistics.

The first and the most influential model of FL aptitude was developed by Carroll (Reference Carroll1959, Reference Carroll and Diller1981), who described the construct in terms of four, relatively independent, subcomponents, namely: phonetic coding ability, grammatical sensitivity, inductive language learning ability, and associative memory. The four-component view of aptitude and the famous measurement tool, the Modern Language Aptitude Test (MLAT) (Carroll & Sapon Reference Carroll and Sapon1959), have provided a crucial foundation for all subsequent FL aptitude research, giving rise to the implementation of other instruments, such as those devised, among others, by Pimsleur (Reference Pimsleur1966), Grigorenko, Sternberg & Ehrman (Reference Grigorenko, Sternberg and Ehrman2000), Meara (Reference Meara2005), and Doughty et al. (Reference Doughty, Campbell, Mislevy, Bunting, Bowles, Koeth, Prior, Watanabe and Lee2010, see also Doughty Reference Doughty, Granena and Long2013).

At present, two models can be regarded as the most important: Skehan's (Reference Skehan1998, Reference Skehan and Robinson2002) and Robinson's (Reference Robinson and Robinson2002, Reference Robinson and DeKeyser2007), both of which view FL aptitude as a set of cognitive abilities and describe it from the perspectives of psycholinguistics and cognitive science. In contrast to previous models, which considered FL aptitude to be a self-contained area, Skehan (Reference Skehan1998) proposed a conceptualization, where aptitude consists of components that could be related to different stages of information processing in SLA. Thus, phonetic coding ability corresponds to input processing, language analytic ability, which includes Carroll's (Reference Carroll1959) grammatical sensitivity and inductive language learning, is related to central processing, and memory-as-retrieval is tied to output and fluency. Skehan's (Reference Skehan and Robinson2002) updated model includes attentional control and working memory, both extremely significant in contemporary research on FL aptitude, and identifies four stages of SLA processing: noticing, patterning, controlling and lexicalizing. Such a conceptualization is compatible with a cognitive view on SLA (Dörnyei & Skehan Reference Dörnyei, Skehan, Doughty and Long2003: 596; Linck et al. Reference Linck, Hughes, Campbell, Silbert, Tare, Jackson and Doughty2013a).

Robinson constructed a model in which FL aptitude is seen as consisting of a number of aptitude complexes that are differentially related to FL learning under different conditions, an assumption that is termed the Aptitude Complex Hypothesis (Robinson Reference Robinson and DeKeyser2007: 274). In line with this hypothesis, some FL learners might possess strengths in abilities facilitative under specific learning conditions but less efficient in others. Robinson (Reference Robinson and Robinson2002) identifies four complexes of aptitudes. Primary, or first-order, abilities are directly measured by psychological tests and involve such constructs as working memory capacity, pattern recognition, grammatical sensitivity and speed of processing, implemented by cognitive factors, including attention, working-, short- and long-term memory. Second-order abilities are combinations of primary abilities, which comprise such cognitive constructs as noticing the gap, memory for contingent speech, memory for contingent text, as well as metalinguistic rule rehearsal. While the model is in many respects an extension of traditional research on aptitude in assuming its hierarchical structure (see Carroll Reference Carroll1959, Reference Carroll1993), relating aptitude complexes to instructional options and settings (see Wesche Reference Wesche and Diller1981; Snow Reference Snow, Sternberg and Wagner1994; Skehan Reference Skehan1998), and assigning learners to memory-oriented and analytically-oriented categories (see Skehan Reference Skehan1989, Reference Skehan1998), its unique contribution lies in the complexity and meticulousness of analysis of aptitude-treatment-interactions (ATI) (see Vatz et al. Reference Vatz, Tare, Jackson, Doughty, Granena and Long2013).

In contemporary research on cognitive factors, FL aptitude is retheorised as a dynamic construct, which interacts with the environment, depends on evolutionary development (see Sternberg & Grigorenko Reference Sternberg and Grigorenko2000) and is potentially trainable (see Sternberg Reference Sternberg and Robinson2002; Dweck Reference Dweck2006; Mercer Reference Mercer2012). The most important implication of Sternberg's (Reference Sternberg and Robinson2002) theory is that abilities are not stable but can be developed. Consequently, individual differences with respect to FL aptitude should correspond to appropriate instructional approaches (see Robinson Reference Robinson and Robinson2002). This conviction finds its reflection in a method of testing proposed by Sternberg (Reference Sternberg and Robinson2002), dynamic testing, where learners are given feedback during the test in order to improve their scores. This idea provided a basis for a new FL aptitude test, the Cognitive Ability for Novelty in Acquisition of Language (CANAL-FT) (Grigorenko et al. Reference Grigorenko, Sternberg and Ehrman2000), which incorporates Sternberg's main assumptions with reference to human abilities. Such a dynamic perspective is in line with a position on the development of talent, which was proposed by Dweck (Reference Dweck2006) and later applied to the field of SLA by Mercer (Reference Mercer2012). Drawing upon the interactionist stance and complexity theory (Larsen-Freeman & Cameron Reference Larsen-Freeman and Cameron2008), Dweck questions the traditional view of ability as fixed and stable, and introduces the concept of mindset, which represents a mental attitude that determines how individuals interpret and respond to various situations.

3. Cognitive giftedness: A contemporary psychological perspective

There are many definitions of giftedness in psychology; however, most researchers describe it in rather broad terms, with reference to various cognitive and non-cognitive qualities. The IQ (intelligence quotient) score is considered an indispensable, but inadequate, measure of giftedness. This is because other factors, such as motivation, high self-concept and creativity, are considered to be vital qualities in the development of giftedness into talent (Renzulli Reference Renzulli, Sternberg and Davidson1986; Gagné Reference Gagné2000; Simonton Reference Simonton, John, Robins and Pervin2008; Erard Reference Erard2012). Two models of giftedness are often referred to in psychological literature, those proposed by Gagné (Reference Gagné2000) and Renzulli (Reference Renzulli, Sternberg and Davidson1986), even though terminological problems abound.

Gagné's (Reference Gagné2000) Differentiated Model of Giftedness and Talent elucidates how outstanding natural abilities develop into specific expert skills or talents. Giftedness is an untrained innate ability, evident in no less than one aptitude domain, to a degree that places a child at least among the upper 15% of peers. Talent signifies superior mastery of methodically upgraded abilities, skills or knowledge in at least one domain, which categorizes the individual's achievements within at least the top 15% of peers. There exist five domains of aptitude, that is intellectual, creative, socio-affective, sensorimotor and other factors. Natural abilities are inborn and can be identified in intellectual undertakings, one of which is L2 learning. These abilities can be observed in children, adolescents and adults through their capacity for acquiring new skills and the pace at which this happens. Abilities progressively transform into talents through systematic training, with the caveat that two key conditions have to be met: one is the presence of well-above-average natural ability and the other is systematic effort devoted to training and practising.

In Renzulli's (1986) Three Ring Model, three indispensable clusters of human traits form a gift, namely: above-average ability, high levels of task commitment and high levels of creativity, where above-average ability can be specified as either general or domain-specific cognitive ability. To make the implementation of all three elements successful, an ample variety of educational opportunities is required. Renzulli declares that it is not the ‘strength’ of particular clusters of traits, but rather the interaction occurring between them that shapes the development of talent.

Evidently, the terms ability, aptitude, giftedness and talent are not clearly differentiated in psychology and are often used interchangeably. Gagné (Reference Gagné2000) equates aptitude with giftedness, Renzulli (Reference Renzulli, Sternberg and Davidson1986) does not seem to differentiate between giftedness and talent, and there are those who opt for the term genius (Simonton Reference Simonton, John, Robins and Pervin2008: 683), which is highly controversial and subjective. In SLA research, FL aptitude is identified as a complex set of cognitive abilities, whereas giftedness/talent refers to an extremely high level of aptitude (see Skehan Reference Skehan1998; Robinson Reference Robinson and DeKeyser2007). An attempt at resolving this terminological confusion is made in Table 1.

Table 1

Classification of terminology (based on Renzulli Reference Renzulli, Sternberg and Davidson1986; Carroll Reference Carroll1993; Gagné Reference Gagné2000)

Irrespective of the specific terms used, in most theories, giftedness is not defined in a dichotomous way, but, rather, as a continuum. Such terms as ‘bright’, ‘able’, ‘more able’, ‘highly able’, ‘gifted’ and ‘talented’, referring to degrees of exceptionality, are present in all definitions of this concept. Procedures for identifying giftedness are varied because there is no conceptual clarity as to what exactly it comprises, with qualitative and quantitative criteria being proposed (see Kerr Reference Kerr2009; Erard Reference Erard2012). Teacher-, parent-, peer-, or self-nomination based upon academic results typically represent the former, while data concerning students’ school history and standardized test outcomes (e.g. intelligence or aptitude) are reflective of the latter. The process of classification of gifted individuals is approached from a number of perspectives, taking into account a variety of factors, such as specific and general cognitive ability together with cultural, gender and language differences (Kerr Reference Kerr2009: xxviii).

An important consideration in selecting talents is a decision as to how gifted individuals differ from what is regarded the norm. The norms proposed in the literature vary, as evident in recommendations that individuals placed within the top 15% (Renzulli Reference Renzulli, Sternberg and Davidson1986; Gagné Reference Gagné2000) or 5% (Selinker Reference Selinker1972) of the normal curve be considered talented. Despite these controversies and relative subjectivity of the discussed terms, researchers generally agree that the most important criterion for talent is remarkable achievement. Evidently, the definitions of talent or giftedness proposed in the field of psychology, although relevant to the domain of LG, are insufficient as a point of reference for SLA researchers who have to develop their own criteria for identifying and measuring linguistic talent. A modified definition of a gifted L2 learner that we propose might read as follows: A gifted FL learner is a person who, owing to his/her exceptional inborn gift for learning languages, especially capacious verbal working memory, as well as expertise in L2 learning, is able to learn any foreign language to a near-native level of competence, given proper motivation, time and conducive environment. Consequently, we will use the terms linguistic giftedness (LG) and gifted foreign language learners with respect to such individuals. However, other terms, such as foreign language high achievers, superlearners, hyperpolyglots and exceptional foreign language learners will be used as well when describing specific studies to give justice to the perspective that particular researchers adopt to refer to this group or the research foci they wish to emphasize.

4. Review of research on gifted FL learners

Recently, there has been a growing interest in gifted FL learners with respect to second or foreign languages. The most promising areas of research are studies examining ultimate attainment of late learners (e.g. Ioup et al. Reference Ioup, Boustagui, El Tigi and Moselle1994; Moyer Reference Moyer1999; Bongaerts, Mennen & van der Silk. 2000; Nikolov Reference Nikolov2000; van Boxtel, Bongaerts & Coppen Reference Van Boxtel, Bongaerts, Coppen, Foster-Cohen and Pekarek-Doehler2003; Birdsong Reference Birdsong, Davies and Elder2004; Bongaerts Reference Bongaerts2005; van Boxtel et al. Reference Van Boxtel, Bongaerts and Coppen2005; Birdsong Reference Birdsong, Bohn and Munro2007; Moyer Reference Moyer2007; Abrahamsson & Hyltenstam Reference Abrahamsson and Hyltenstam2008; DeKeyser, Alfi-Shabtay & Ravid Reference DeKeyser, Alfi-Shabtay and Ravid2010; Granena & Long Reference Granena and Long2013), those investigating the role of social-psychological factors in gifted FL learners (e.g. Moyer Reference Moyer2007; Rota & Reiterer Reference Rota, Reiterer, Dogil and Reiterer2009; Moyer Reference Moyer2010; Forsberg Lundell & Sandgren Reference Forsberg Lundell, Sandgren, Granena and Long2013; Moyer Reference Moyer2013), and those focusing on maximally advanced multilinguals, that is super-learners, polyglots, and savants (e.g. Smith et al. Reference Smith, Tsimpli, Morgan and Woll2011; Treffert Reference Treffert2011; Erard Reference Erard2012; Hyltenstam Reference Hyltenstam2014a, Reference Hyltenstam2014b; Hyltenstam Reference Hyltenstam and Hyltenstam2016a, Reference Hyltenstam and Hyltenstam2016b forthcoming). Particularly interesting is neurolinguistic research (e.g. Golestani, Price & Scott Reference Golestani, Price and Scott2011; Reiterer et al. Reference Reiterer, Pereda and Bhattacharya2011a, Reference Reiterer, Hu, Erb, Rota, Nardo and Grodd2011b; Díaz et al. Reference Díaz, Mitterer, Broersma and Sebastian-Galles2012; Sebastián-Gallés et al. Reference Sebastián-Gallés, Soriano-Mas, Baus, Díaz, Ressel, Pallier, Costa and Pujol2012; Hu et al. Reference Hu, Ackermann, Martin, Erb, Winkler and Reiterer2013), which represents a major advance on sole reliance on behavioral data. Researching exceptional talents is difficult because such talents are quite rare and it is difficult to assemble a group that would be large enough for statistical analysis. Besides, there exist idiosyncratic paths of development of gifted individuals that fall outside any classifications, the criteria adopted for the choice of such individuals are inconsistent, and the research methodology varies (see Abrahamsson & Hyltenstam Reference Abrahamsson and Hyltenstam2008; Long Reference Long, Granena and Long2013).

Studies on gifted L2 learners can be roughly divided into three groups. The first comprises early research on gifted individuals (e.g., Novoa et al. Reference Novoa, Fein, Obler, Obler and Fein1988; Schneiderman & Desmarais Reference Schneiderman and Desmarais1988a, Ioup et al. Reference Ioup, Boustagui, El Tigi and Moselle1994). The second includes research on accomplished L2 post-pubescent learners (e.g., Moyer Reference Moyer1999; Bongaerts et al. Reference Bongaerts, Mennen and van der Silk2000; DeKeyser Reference DeKeyser2000; Nikolov Reference Nikolov2000; Birdsong Reference Birdsong, Davies and Elder2004; Bongaerts Reference Bongaerts2005; van Boxtel et al. Reference Van Boxtel, Bongaerts and Coppen2005; Birdsong Reference Birdsong, Bohn and Munro2007; Moyer Reference Moyer2007; Abrahamsson & Hyltenstam Reference Abrahamsson and Hyltenstam2008; Sáfár & Kormos Reference Sáfár and Kormos2008; Abrahamsson & Hyltenstam Reference Abrahamsson and Hyltenstam2009; DeKeyser et al. Reference DeKeyser, Alfi-Shabtay and Ravid2010; Biedroń Reference Biedroń2011a, Reference Biedroń2012a, Reference Biedroń and Pawlak2012b; Granena Reference Granena, Granena and Long2013; Granena & Long Reference Granena and Long2013). The third brings together empirical investigations of L2 super-learners, or hyperpolyglots (Erard Reference Erard2012), two ongoing studies of savants, those of Christopher (Smith & Tsimpli Reference Smith and Tsimpli1991; Morgan et al. Reference Morgan, Smith, Tsimpli and Woll2007; Smith et al. Reference Smith, Tsimpli, Morgan and Woll2011) and Daniel Tammet (Treffert Reference Treffert2011), and a study on polyglots conducted by Hyltenstam (Reference Hyltenstam2014a, Reference Hyltenstam2014b). The following sections present studies representing these three lines of inquiry.

5. Native language abilities and foreign language aptitude

Since 1973, when Carroll formulated a hypothesis that FL aptitude might be a residue of L1 learning ability, evidence for this suggestion has accumulated. A relationship between aptitudes for L1 and L2 was demonstrated in the Bristol Language Project (Wells Reference Wells1981), which revealed that variation in L1 development in 125 children participants correlated with differences in FL aptitude, examined twelve years later (see Skehan Reference Skehan and Meara1986, Reference Skehan1998). The assumption that native language abilities and FL aptitude are related also lies at the core of Sparks & Ganschow's (Reference Sparks and Ganschow1991, Reference Sparks and Ganschow2001) Linguistic Coding Deficit Hypothesis, based on research findings showing that at-risk FL learners have linguistic coding difficulties, which has a detrimental effect on both L1 and L2 acquisition. As Sparks & Ganschow speculate, ‘Inefficiency of the language processing codes may produce interference resulting in individual differences in FL acquisition. [. . .] We suggest, then, that native language factors are likely to be implicated as the main variable in FL learning’ (Reference Sparks and Ganschow1991: 10). Empirical evidence for this position comes from a number of studies (e.g. Sparks et al. Reference Sparks, Javorsky, Patton and Ganschow1998; Sparks et al. Reference Sparks, Patton, Ganschow, Humbach and Javorsky2006; Sparks Reference Sparks2012), which suggest that native-language skills operate as a basis for L2 learning for phonology, orthography, grammar and semantics, and, consequently, any problems occurring in these areas will have a negative impact on both L1 and L2 learning.

Unfortunately, very little is known about the native language abilities of gifted language learners. We know that the case described by Ioup et al. (Reference Ioup, Boustagui, El Tigi and Moselle1994) – Julia– started to speak in full sentences at the age of 18 months, which may suggest high verbal abilities. In another study, conducted by Biedroń (Reference Biedroń2012a), who aimed to create cognitive-affective profiles of gifted language learners, 14 out of the total of 44 participants declared that they started to speak around their first year of life, and 30 reported starting to read around the fourth or fifth year of life. Such information confirms that FL aptitude and L1 abilities have a common source, with the caveat that the data were obtained through self-reports. A more valuable source of information on the relationship between L1 and L2 abilities is provided by neurolinguistic studies conducted by Díaz and her collaborators (Díaz et al. Reference Díaz, Baus, Escera, Costa and Sebastián-Gallés2008; Díaz et al. Reference Díaz, Mitterer, Broersma and Sebastian-Galles2012), the results of which suggest that native phonetic abilities can predict successful learning of a new phonetic system.

6. Intelligence in gifted FL learners

In research on individual differences, FL aptitude and intelligence have been regarded as separate (Skehan Reference Skehan1998). However, the contribution to FL learning proficiency of a general factor, which seems to underlie all other cognitive abilities, has been recognized by some researchers. Carroll's (Reference Carroll1993) Cognitive Abilities Model classifies language abilities as primary cognitive abilities subsumed under a general cognitive ability factor, that is intelligence. Additionally, as Dörnyei (Reference Dörnyei2005) underscores, both the MLAT (Carroll & Sapon Reference Carroll and Sapon1959) and the Pimsleur Language Aptitude Battery (PLAB) (Pimsleur Reference Pimsleur1966) include a test of L1 vocabulary, which is also a fundamental element in measuring intelligence. Skehan (Reference Skehan1998) claims that there is a relationship between specific language learning ability and general cognitive ability (i.e. intelligence), which is especially strong for analytic abilities and far less evident for memory and phonetic coding ability. The study by Sasaki (Reference Sasaki1996) cast some light on the controversial relationship between FL aptitude and intelligence. She investigated to what extent and in what respects FL aptitude depends on the general factor of intelligence and in what respects it is independent of it, that is language-specific. Sasaki concluded that intelligence and analytic ability are interrelated, while phonetic coding ability and memory factors are separate components of FL aptitude, independent of intelligence (see Skehan Reference Skehan1998).

Very little is known about the IQ of linguistically gifted individuals. All such individuals described in the literature manifested average IQ and they performed higher on verbal than performance scales (see Skehan Reference Skehan1998). The self-reported IQ score for the 172 people who reported having learnt over six languages in Erard (Reference Erard2012) oscillated between 120 and 140. In Biedroń's (Reference Biedroń2012a) study, the general IQ, as measured by Wechsler scale, of the gifted FL learners was 125, with the score on the verbal scale being higher than those on the non-verbal and memory scales (130 as opposed to 118 and 128, respectively). Also Hyltenstam (Reference Hyltenstam2014b) lists high general cognitive ability among the characteristics of polyglots.

Research on the relationship between FL aptitude and intelligence is clearly inadequate, given the importance of this factor for human cognitive abilities in general and FL aptitude in particular (see Carroll Reference Carroll1993; Grigorenko et al. Reference Grigorenko, Sternberg and Ehrman2000). There are two main theories which dominate the discussion. The first reduces the effect of intelligence exclusively to analytic abilities (see Sasaki Reference Sasaki1996; Skehan Reference Skehan1998), while the other, based on early research on gifted L2 learners (see Novoa et al. Reference Novoa, Fein, Obler, Obler and Fein1988), entirely ignores its role in the development of LG. It has to be emphasized that most studies examining the relationship between LG and other cognitive abilities have focused on performance and verbal IQ, neglecting the third IQ domain, memory, which is surprising given its relevance to LG. Contemporary operationalizations of intelligence, such as Wechsler's (Reference Wechsler1997), offer a variety of scales and in interpreting the results much importance is attached to disproportions between particular test scores on particular scales, which enables researchers to conduct in-depth, clinical diagnosis of intellect. Apart from the traditional two-factor solution, three-factor, four-factor and six-factor solutions have been developed. The most up-to-date six-factor solution includes verbal comprehension, perceptual organization, working memory, processing speed, auditory memory and visual memory (Tulsky, Ivnik, Price & Wilkins Reference Tulsky, Ivnik, Price, Wilkins, Tulsky, Saklofske, Chelune, Heaton, Ivnik and Bornstein2003). Applying modern methods of testing may change the long-lasting stereotypical understanding of the role of IQ in the development of LG.

7. Memory in gifted FL learners

Skehan (Reference Skehan1998: 233) argued that very talented learners are not qualitatively different from simply high-aptitude learners. The most significant characteristic of exceptionally successful learners is unusual verbal memory. Indeed, outstanding memory for verbal material was the most striking characteristic of all the described cases of talented individuals (see Novoa et al. Reference Novoa, Fein, Obler, Obler and Fein1988; Schneiderman & Desmarais Reference Schneiderman and Desmarais1988a; Ioup et al. Reference Ioup, Boustagui, El Tigi and Moselle1994; Erard Reference Erard2012). Savants, like Christopher, have selective memory, and what makes them really peculiar is that they are able to recall things after a time interval even with greater accuracy than soon after memorizing them. Working memory (WM) (Baddeley & Hitch Reference Baddeley, Hitch and Bower1974; Baddeley, Gathercole & Papagno Reference Baddeley, Gathercole and Papagno1998; Baddeley Reference Baddeley2003) has recently moved to the top of the research agenda as a significant factor determining learning outcomes and a component of FL aptitude (e.g. Papagno & Vallar Reference Papagno and Vallar1995; Miyake & Friedman Reference Miyake, Friedman, Healy and Bourne1998; Sawyer & Ranta Reference Sawyer, Ranta and Robinson2001; Robinson Reference Robinson, Doughty and Long2003; Doughty et al. Reference Doughty, Campbell, Mislevy, Bunting, Bowles, Koeth, Prior, Watanabe and Lee2010; DeKeyser & Koeth Reference DeKeyser, Koeth and Hinkel2011; Wen & Skehan Reference Wen and Skehan2011; Biedroń & Szczepaniak Reference Biedroń and Szczepaniak2012a; Skehan Reference Skehan, Gass and Mackey2012; Linck et al. Reference Linck, Hughes, Campbell, Silbert, Tare, Jackson and Doughty2013a; Linck et al. Reference Linck, Osthus, Koeth and Bunting2013b).

Biedroń & Szczepaniak (Reference Biedroń and Szczepaniak2012a), for instance, tested short-term and WM abilities in gifted FL learners. Twenty-seven such individuals were compared to 36 mainstream philology students with the help of: (1) three memory subtests of the Wechsler Intelligence Scale, that is Digit-Span, Digit-Symbol Coding and Arithmetic, which comprise the Memory and Resistance to Distraction Index, (2) two short-term memory tests in the MLAT that is Number learning and Paired associates, and (3) a WM test, the Polish Reading Span (PRSPAN), designed by the researchers (Biedroń & Szczepaniak Reference Biedroń, Szczepaniak, Badio and Kosecki2012b). The analysis revealed that short-term memory and WM abilities for gifted FL learners were higher than for mainstream philology students. Moreover, the former obtained higher scores on memory tests based on linguistic material than those based on numbers. The researchers concluded that two components of WM, namely the phonological loop and the central executive, are significant factors in determining FL learning outcomes.

The idea of WM as FL aptitude has been reflected in newly developed measurement tools. The High-Level Language Aptitude Battery (Hi-LAB) (Doughty et al. Reference Doughty, Campbell, Mislevy, Bunting, Bowles, Koeth, Prior, Watanabe and Lee2010; Doughty Reference Doughty, Granena and Long2013; Linck et al. Reference Linck, Hughes, Campbell, Silbert, Tare, Jackson and Doughty2013a) is an innovative, computer-delivered test, designed to predict high-level attainment in post-critical period FL learning. Doughty et al. (Reference Doughty, Campbell, Mislevy, Bunting, Bowles, Koeth, Prior, Watanabe and Lee2010) maintain that the Hi-LAB is the first instrument to incorporate advances in memory functioning, particularly, WM. The test comprises constructs hypothesized to underlie high-level FL aptitude, such as memory, including short-term, long-term and WM, perceptual acuity, speed, primability, induction, pragmatic sensitivity and fluency. Linck et al. (Reference Linck, Hughes, Campbell, Silbert, Tare, Jackson and Doughty2013a) conducted a study aimed at examining whether the cognitive and perceptual abilities measured by the Hi-LAB can predict very high levels of attainment in L2. The subjects comprised two groups of adults: one high-attainment and the other mixed-attainment, with high-level attainment being operationalized as highly proficient L2 performance measured by the Defense Language Proficiency Tests and/or as high-level job performance in single or multiple languages. High-level aptitude was defined as ‘a composite of domain-general cognitive abilities and specific perceptual abilities that, together, can support or constrain one's ability to attain high-level proficiency as an adult learner’ (Linck et al. Reference Linck, Hughes, Campbell, Silbert, Tare, Jackson and Doughty2013a: 535). It was concluded that the test is an efficient tool for distinguishing very successful L2 learners from other individuals, with high-level attainment being positively correlated to WM, associative learning and implicit learning.

WM plays a crucial role in a number of complex cognitive abilities, such as language learning, reasoning, comprehension and cognitive control, and the outcomes of WM measures are indicators of general cognitive ability. Two subsystems of WM are especially significant in SLA: the phonological loop and the central executive. The phonological loop, viewed as an equivalent of a language acquisition device, plays a crucial role in learning phonological forms of new words. The central executive directs attentional processes that create conditions for goal-directed behaviour by maintaining relevant information in an active state or retrieving it under conditions of interference, distraction or conflict. Friedman et al.'s (Reference Friedman, Miyake, Young, DeFries, Corley and Hewitt2008) study on twins indicates that executive functions are among the most heritable psychological traits. WM capacity underlies noticing ability, which facilitates SLA in general. Although WM is generally stable, Juffs & Harrington (Reference Juffs and Harrington2011: 157) suggest that it can be modified through experience, thus constituting a combination of a trait and state variable. Summing up, the extraordinary memory abilities of gifted L2 learners reported in research confirm Skehan's (Reference Skehan1998) assumption that the most significant characteristic of exceptionally successful learners is unusual verbal memory. However, it is not possible to decide on the basis of the available empirical data whether superior memory abilities in gifted L2 learners are inborn or, rather, evolve as a result of multiple experiences of FL learning. Probably the two possibilities are not mutually exclusive.

8. Personality factors in gifted foreign language learners

Two different trends can be distinguished in research on FL aptitude and ultimate attainment in late bilinguals. One adopts a narrow scope, confined to three variables, that is AO, FL aptitude and LOR (see Abrahamson & Hyltenstam Reference Abrahamsson and Hyltenstam2008; DeKeyser et al. Reference DeKeyser, Alfi-Shabtay and Ravid2010; Bylund, Hyltenstam & Abrahamsson Reference Bylund, Hyltenstam, Abrahamsson, Granena and Long2013; Granena & Long Reference Granena and Long2013; Long Reference Long, Granena and Long2013). The other embraces a wider perspective, which can be called socio-psychological, as it extends the scope of research to the role of instruction, motivation, attitudes and other variables believed to explain high levels of L2 attainment (see Nikolov Reference Nikolov2000; Moyer Reference Moyer2007; Rota & Reiterer Reference Rota, Reiterer, Dogil and Reiterer2009; Moyer Reference Moyer2010, Reference Moyer2013, Reference Moyer2014; Forsberg Lundell & Sandgren Reference Forsberg Lundell, Sandgren, Granena and Long2013; Muñoz Reference Muñoz, Pawlak and Aronin2014).

The first stance is clearly represented by Granena & Long (Reference Granena and Long2013), who convincingly argue that non-native-like attainment in L2 learning is typical of most non-native speakers despite favorable conditions, such as very long residence in the TL country, being married to a native speaker, using L2 on a daily basis, as well as absence of any attitudinal or emotional barriers, accompanied by strong motivation. As they point out, the influence of psychological and social factors or the LOR on acquisition of phonology and morphosyntax is minimal, the AO being the sole factor that reliably predicts ultimate attainment. They comment: ‘Age effects are robust, even in situations where the quantity and quality of input available to learners are not at issue, and regardless of learners’ social-psychological profile’ (2013: 312). They concede, however, that for late learners, those who start learning an L2 after the closure of sensitive periods, the decline is more gradual and other factors, such as FL aptitude, motivation, L2 use and exposure, or pronunciation training may play a role.

An opposite view is held by Moyer (Reference Moyer2007, Reference Moyer2010, Reference Moyer2013, Reference Moyer2014), who believes that the role of factors traditionally accepted as critical in the acquisition of near-native pronunciation is overstated. She argues that social-psychological variables, such as identity, motivation, empathy, concern for native-likeness, LOR and contacts with native speakers are as important as aptitude and AO. Although maturational constraints are a function of neurological changes, the rule of linearity of age-related learning outcomes ceases after maturation, at about12, and the deterioration in L2 learning afterwards hinges on the aforementioned factors. Moyer also emphasizes the role of formal instruction and contacts with native speakers, which allow the provision of feedback. Positive attitudes are also among the most significant variables in acquiring L2 pronunciation. Nonetheless, evidence for such claims is less convincing than that offered by her opponents (see Abrahamsson & Hyltenstam Reference Abrahamsson and Hyltenstam2008; Long Reference Long, Granena and Long2013).

There are several studies which have examined the role of personality factors in FL aptitude. For example, an ongoing study on phonetically talented L2 learners, conducted by Reiterer (Reference Reiterer, Dogil and Reiterer2009) and co-workers, has provided remarkable insights into the correlation between phonetic abilities and personality. No correlation was found between pronunciation talent and extraversion, openness to experience or neuroticism, whilst a moderate positive correlation was revealed for conscientiousness and agreeableness. The researchers ascribe this finding to the separateness of phonetic aptitude, which does not require social competence, from other aptitudes affecting oral L2 use (Hu & Reiterer Reference Hu, Reiterer, Dogil and Reiterer2009). Moyer (Reference Moyer2007) examined 50 immigrant learners of English in the USA and found that acquiring authentic pronunciation was a function of reasons for learning English, perceived ability to improve accent, self-confidence in English, assimilation with the TL culture, and an intention to reside in the USA for a long time. Moreover, attitudes towards the TL were more important than those towards culture. Two factors, experience and positive orientation, are particularly important for achieving greater authenticity in accent. Similarly, Rota & Reiterer (Reference Rota, Reiterer, Dogil and Reiterer2009) found positive correlations between talent for pronunciation and such ID variables as empathy, mental flexibility and phonological WM. Hu et al. (2013) discovered that empathy constitutes a better predictor of pronunciation aptitude in advanced learners than phonological short-term memory. Kaufman et al. (Reference Kaufman, DeYoung, Gray, Jimenez, Brown and Mackintosh2010) suggested that openness and intuition are strongly correlated with implicit learning. Biedroń (Reference Biedroń2011a) compared two groups of individuals, 44 gifted L2 learners (highly proficient multilinguals) and 37 non-gifted L2 learners (English philology students) with respect to personality factors, adopting the Five Factors model (Costa & McCrae Reference Costa and McCrae1992). She found that openness to experience was significantly higher in the gifted than in non-gifted learners, but no differences were detected for the remaining four factors, namely neuroticism, agreeableness, extraversion and conscientiousness. Forsberg Lundell & Sandgren (Reference Forsberg Lundell, Sandgren, Granena and Long2013) examined the relationship between production of collocations, FL aptitude and personality in late L2 French learners. The personality dimensions of cultural empathy and open-mindedness were significantly correlated with collocations and Llama D (phonetic memory) (Meara Reference Meara2005). As the researchers explain, both of these factors involve a cognitive capacity to adopt other people's perspective and to be tolerant.

Is anything known about the personalities of hyperpolyglots? Are, as some researchers (see Dörnyei Reference Dörnyei2009) suggest, super-learners more open to new experience? Erard's research suggests that they are indeed. They claim to be good observers and mimics, not only of accent but also of body language and behaviour, which means that they are good at adaptation. As one of the hyperpolyglots put it, ‘The good language learner has the ability to accept the role of a child [. . .]. of being naïve, foolish, stumbling, and inarticulate but also curious, open-minded, and full of energy’ (Erard Reference Erard2012: 238). As far as learning styles are concerned, hyperpolyglots seem to be intuitive sharpeners (see Ehrman Reference Ehrman and Griffiths2008: 68) and systemizers (Baron-Cohen et al. Reference Baron-Cohen, Richler, Bisarya, Gurunathan, Wheelwright, Frith and Hill2003), which means that they tend to organize and systemize the learned material. They like patterns, discovering rules, making predictions and looking for exceptions. Their main characteristics are self-confidence, motivation (Hyltenstam Reference Hyltenstam2014b), perseverance and diligence. Supposedly, they have the ability to engage in flow (Csikszentmihalyi Reference Csikszentmihalyi1990) and actively seek experience of this kind. Flow is a state experienced by people performing at the peak or beyond their limits and it involves experiencing spontaneous joy when individuals become utterly absorbed in a task. It is a state of self-forgetfulness, a prerequisite for mastery in any profession or art, and also in language learning.

SLA researchers recognize the need to investigate personality factors in studies on multilinguals (see Dewaele & Furnham Reference Dewaele and Furnham2000; Deweale Reference Dewaele2002, Reference Dewaele, Ritchie and Bhatia2009) and gifted FL learners (see Hu & Reiterer Reference Hu, Reiterer, Dogil and Reiterer2009; Biedroń Reference Biedroń2012a; Moyer Reference Moyer2013). However, there is very little research on the relationship between FL aptitude and personality traits, possibly due to disappointing correlations between success in L2 and personality dimensions (see Dörnyei Reference Dörnyei2005, Reference Dörnyei2009). Even less is known about personalities of linguistically gifted individuals. Dörnyei (Reference Dörnyei2009, Reference Dörnyei and Macaro2010) addresses the problem of the relationship between cognitive, motivational and emotional processes, and their cumulative impact on human cognitive functioning. He argues that a modular view of individual differences which involves multiple, discrete ID factors has limited value, and suggests that it is possibly more beneficial to focus on higher-order trait complexes that act as integrated wholes (Dörnyei Reference Dörnyei and Macaro2010: 248). In keeping with this position, Dörnyei recommends dynamic systems theory (see Larsen-Freeman & Cameron Reference Larsen-Freeman and Cameron2008) as a paradigm which can best account for individual differences. In his view, tracing individual paths of growth of gifted learners is likely to offer more useful insights into their personalities than traditional quantitative methods. This position, however, is not problem-free, not least because, as Gregg (Reference Gregg2010) claims, Larsen-Freeman & Cameron's (Reference Larsen-Freeman and Cameron2008) perspective on complexity theory lacks any logical or scientific basis in excluding mathematics, refusing to restrict the number of variables and proposing to focus on all possible details. He argues that if complexity theory is to be applied to the domain of language, it must be tested empirically. If such criticisms are accepted, it is difficult to see how dynamics systems theory can offer a basis for investigating the contribution of personality factors to LG.

9. Neurology of linguistic giftedness

In recent years neurolinguistics has become the most revealing and revolutionary source of information about SLA, supplementing earlier reliance on behavioral data (see Long Reference Long, Granena and Long2013: 33). The good news is that the number of studies on neurological substrates of FL aptitude is on the rise and the data obtained allow more in-depth analysis of this domain (e.g. Chee et al. Reference Chee, Soon, Lee and Pallier2004; Golestani et al. Reference Golestani, Price and Scott2011; Reiterer et al. Reference Reiterer, Pereda and Bhattacharya2011a; Reiterer et al. Reference Reiterer, Hu, Erb, Rota, Nardo and Grodd2011b; Díaz et al. Reference Díaz, Mitterer, Broersma and Sebastian-Galles2012; Sebastián-Gallés et al. Reference Sebastián-Gallés, Soriano-Mas, Baus, Díaz, Ressel, Pallier, Costa and Pujol2012; Hu et al. Reference Hu, Ackermann, Martin, Erb, Winkler and Reiterer2013). According to Schumann (Reference Schumann and Schumann2004), FL aptitude might be a result of evolutionary selection processes. Some individuals are better neurologically prepared to respond to rapid changes in the environment, and, thus, to survive and to spread their genes. Anatomical differences in brains, for example hypertrophies,Footnote ⁶ underpinning specific abilities, are triggered by the environment. If a talented individual with a particular hypertrophy begins learning in a conducive environment, the brain acts in response to such conditions intensifying particular neural connections. This, in turn, aids learning, making it effortless and faster. Accordingly, such an individual might achieve high expertise in a specific field of study. All of this shows that high FL aptitude might result from both inborn functional and structural characteristics, and an individual brain response to a specific experience of learning a language.

Opinion is divided as to whether neural correlates of L1 and L2 are the same or different. For example, Dehaene et al. (Reference Dehaene, Dupoux, Mehler, Cohen, Paulesu, Perani, van de Moortele, Lehéricy and Le Bihan1997) discovered that, in contrast to L1, which always activates the same areas in the left hemisphere, L2 activates a highly variable network of both right and left hemispheres, with this tendency being observed in late L2 learners. Variability in the cortical representation of an L2 can be attributed to AO as well as level of proficiency. Generally, more intensive right hemisphere activation was observed in late, low-proficiency L2 learners, whereas in early, more proficient learners, areas of L1 and L2 activation overlap (Kim et al. Reference Kim, Relkin, Lee and Hirsch1997). Similar results were obtained by Golestani & Zatorre (Reference Golestani and Zatorre2004), Reiterer et al. (Reference Reiterer, Pereda and Bhattacharya2011a, Reference Reiterer, Hu, Erb, Rota, Nardo and Grodd2011b), Sebastián-Gallés et al. (Reference Sebastián-Gallés, Soriano-Mas, Baus, Díaz, Ressel, Pallier, Costa and Pujol2012), and Wong, Perrachione & Parrish (Reference Wong, Perrachione and Parrish2007), who report more extended or bilateral activation in poorer L2 learners. Indefrey & Gullberg (Reference Indefrey, Gullberg, Gullberg and Indefrey2006) postulate that L2 proficiency level, not AO, is the strongest predictor of the degree of similarity between late learners and native speakers, which means that as L2 proficiency increases, the processing profile in L2 becomes similar to that in L1. Recently, a moderate view, referred to as partial overlap (Reiterer Reference Reiterer, Dogil and Reiterer2009: 160), has been gaining popularity. In line with this position, there is an essential core overlap for L1 and L2 processing, with additional brain areas being activated for L2, possibly as a function of fluency or proficiency. The explanation is that increased ‘control effort’ causes higher activation in lower-proficiency L2 speakers (Reiterer Reference Reiterer, Dogil and Reiterer2009; Reiterer et al. Reference Reiterer, Hu, Erb, Rota, Nardo and Grodd2011b) and subsequent automatization lowers it. It is possible that speakers compensate for lower L2 efficiency by driving this region more strongly, or, alternatively, a greater number of neurons might be employed to perform a particular task. Consequently, the efficiency of the neural system might constitute a neurological basis for FL aptitude. Since reduced effort results from higher TL proficiency, it is likely that FL aptitude is an outcome of an interplay between inborn capacities, early experience and training.

Evidence has been provided that native phonetic abilities underlie L2 phonetic abilities. Díaz et al. (Reference Díaz, Baus, Escera, Costa and Sebastián-Gallés2008), using the ERP method, found individual differences between more and less talented learners with respect to phonetic discrimination ability, with those differences being observed in both native and additional languages. The researchers hypothesized that FL phonetic abilities are based on native phonetic abilities; in addition, these abilities result from language-specific rather than general acoustic abilities (see Díaz et al. Reference Díaz, Mitterer, Broersma and Sebastian-Galles2012; Sebastián-Gallés & Díaz Reference Sebastián-Gallés, Díaz, Roberts and Meyer2012; Sebastián-Gallés et al. Reference Sebastián-Gallés, Soriano-Mas, Baus, Díaz, Ressel, Pallier, Costa and Pujol2012). A number of anatomical differences have been found in more proficient L2 learners in comparison to less proficient ones. Golestani and her colleagues (Golestani, Paus & Zatorre Reference Golestani, Paus and Zatorre2002; Golestani et al. Reference Golestani, Molko, Dehaene, LeBihan and Pallier2007; Golestani et al. Reference Golestani, Price and Scott2011; see also Lopez-Barroso et al. Reference Lopez-Barroso, de Diego-Balaguer, Cunillera, Camara, Münte and Rodriguez-Fornells2011) discovered that more phonologically talented learners possess more grey matter and white matter in parietal regions, especially in the left hemisphere. This finding is important, as left parietal cortex has a role to play in phonetic tasks and is the location of phonological verbal WM. The anatomy underlying WM in left auditory cortex predicts phonological aptitude. Greater asymmetry in the amount of white matter in successful learners can be ascribed to greater myelineation,Footnote ⁷ which results in faster and more efficient neural processing essential in successful learning of phonological aspects of a language. This asymmetry is probably not limited to the ability to learn speech sounds, but to a more general ability to integrate rapidly changing acoustic information. The researchers suggest that this anatomical feature is independent of training; therefore, morphological differences in parietal white matter can predict the effortlessness of learning new sounds. They suggest that the human brain is shaped by the complementary influences of inborn predispositions and experience-dependent brain pliability.

Some neurolinguistic studies have addressed the problem of morphosyntactic attainment in late learners (Lopez-Barroso et al. Reference Lopez-Barroso, de Diego-Balaguer, Cunillera, Camara, Münte and Rodriguez-Fornells2011; Wong et al. Reference Wong, Morgan-Short, Ettlinger and Zheng2012; Wood Bowden et al. Reference Wood Bowden, Steinhauer, Sanz and Ullman2013). Wood Bowden et al. (Reference Wood Bowden, Steinhauer, Sanz and Ullman2013) claim that native-like brain processing (and maybe also ultimate mastery) of syntax can be achieved by university learners. Using the ERP method, the researchers examined 32 late learners of Spanish, who were divided into two groups: low-intermediate, with little experience, and advanced, with more experience. The researchers found that L2 syntactic processing initially differs from L1 processing, but can become native-like as a result of adequate proficiency and exposure. Wong et al. (Reference Wong, Morgan-Short, Ettlinger and Zheng2012) tapped the neurogenetic source of variability in language learning, especially syntax, in particular the role of the dopaminergic system, which is associated with WM, attention and learning in general. The researchers hypothesize that dopamine-related genes are capable of influencing variation in grammar learning.

Summing up, much is known about the neurology of phonological aptitude, whereas scant information is available about morphosyntax, lexis and pragmatics processing in the brains of poor versus successful FL learners, and these areas need to be addressed in future research. Moreover, most researchers suggest that hypertrophies are inborn rather than acquired, but subject to development along with growing expertise in L2 learning (see Perani Reference Perani2005). This suggests the possibility of an increase in FL aptitude thanks to training and practice. Generally, it seems that research on the neurology of FL aptitude, although promising, is still in its infancy.

10. Strategies of gifted foreign language learners

Extremely scant information is available when it comes to the strategies employed by gifted individuals. As we learn from Erard's (Reference Erard2012) study and Hyltenstam's review (forthcoming Reference Hyltenstam and Hyltenstam2016a, Reference Hyltenstam and Hyltenstam2016b), hyperpolyglots are often self-didactics who devise a number of their own learning strategies, even though they appreciate classroom learning. At the same time, teaching methods do not seem to be of great significance to them as very talented individuals seem to succeed no matter what instructional options are employed. Moreover, some learners have negative experiences with formal instruction and their diverse learning activities are undertaken outside institutional education as this gives them the freedom to experiment. On the whole, hyperpolyglots seem to be masters of learning strategies. One of the techniques used by Alexander, a hyperpolyglot examined by Erard, was shadowing, or pronouncing FL sounds simultaneously with hearing them from a recording, which was possible thanks to his excellent memory. Another strategy is to collect and make use of all sorts of dictionaries, books and materials helpful in learning. Hyperpolyglots read a lot. They rely on all the possible types of techniques to enhance their learning, such as playing games, singing, visualizing, talking to themselves, seeking error correction, finding rules and patterns, using recordings to learn sounds and words, following structured reviews, or inventing mnemonic devices. The most prominent among the numerous strategies used by hyperpolyglots are memory strategies. These usually involve repeating patterns many times so that they become internalized, automatic and accessible with little cognitive effort. They demonstrate expertise in learning languages and they are ingenious when it comes to discovering patterns or ‘systems’, especially in learning difficult aspects of a language, such as the tones of Mandarin. This expertise also applies to metacognitive strategies, such as planning and monitoring. They spend many hours a day practising the languages they learn, which might involve performing memory exercises, translating, writing and reading practice, doing grammar exercises and all possible sorts of revisions (Erard Reference Erard2012). Gifted individuals also seem to express a preference for explicit learning strategies (Hyltenstam Reference Hyltenstam2014b).

Although this issue was not specifically investigated by researchers working in the field of LLS, useful insight into the use of these strategies by talented individuals can be derived from studies examining the relationship between the application of LLS and TL proficiency (see, e.g. Takeuchi, Griffiths & Coyle Reference Takeuchi, Griffiths, Coyle, Cohen and Macaro2007; Pawlak Reference Pawlak, Arabski and Wojtaszek2011, for reviews). In their pioneering study, Bialystok & Fröhlich (Reference Bialystok and Fröhlich1978) found that, in the case of high school learners of French, combined use of the strategies of inferencing, practising and monitoring explained attainment in listening, reading and grammar. Subsequent research provided more evidence of a positive relationship between LLS use, either overall, with respect to specific categories, or particular strategies, and attainment, both in general and in specific areas (e.g. Green & Oxford Reference Green and Oxford1995; Dreyer & Oxford Reference Dreyer, Oxford and Oxford1996; Wharton Reference Wharton2000; Pawlak Reference Pawlak2009; Ying-Chun Reference Ying-Chun2009), with estimates that such use may account for between 30% and 70% in achievement (see Oxford Reference Oxford2011). On the other hand, some studies have failed to identify this relationship or indicated the impact of mediating variables, such as learner-related factors, a specific aspect of proficiency, the task, mode of instruction, or educational level (e.g. Politzer & McGroarty Reference Politzer and McGroarty1985; Oxford & Ehrman Reference Oxford and Ehrman1995; Griffiths Reference Griffiths2003; Nisbet, Tindall & Arroyo Reference Nisbet, Tindall and Arroyo2005; Magogwe & Oliver Reference Magogwe and Oliver2007). Another problem is that the relationship between strategy use and proficiency may in some cases be curvilinear rather than linear, because expert learners may automate strategies to such an extent that they are no longer aware of their employment (see Oxford Reference Oxford2011), or fall back on fewer strategies, which, however, trigger deeper processing (e.g. Alexander Reference Alexander, Day and Sternberg2004). The latter point is reflective of the possibility that it is not the sheer quantity of LLS use but, rather, its quality that is responsible for superior achievement in FL learning. More specifically, it is vital that strategies should be matched to task demands, be compatible with learners’ personalities and learning styles, and be adeptly combined into clusters or chains rather than haphazardly applied in isolation (see Ehrman, Leaver & Oxford Reference Ehrman, Leaver and Oxford2003). There is also the question of causality, since correlation and regression analyses cannot tell us whether strategy use impacts attainment or perhaps the other way around, which points to the need for studies based on structural equation modelling. Equally important is research that would seek to tap the use of strategies in the performance of specific tasks and involve reliance on multiple instruments, yielding both quantitative and qualitative findings, as this could shed light on what makes LLS useful in specific contexts for specific learners. Perhaps this approach would be particularly valuable in examining gifted language learners, as it would enable us not only to determine whether they use more strategies but also how exactly they use them, and with respect to different foreign languages.

11. Research on gifted FL learners – limitations, directions, methodological issues

Gifted FL learners comprise a distinctive group of talented individuals that has thus far been poorly scrutinized and described. First of all, there is no transparent definition of what a linguistic talent is, such that would take account of the specificity of SLA. The traditional definition (see Schneiderman & Desmarais Reference Schneiderman and Desmarais1988a; Skehan Reference Skehan1998), describing linguistically talented individuals as those who learn an L2 quickly post-puberty and to a native-like level, is imprecise, vague and untenable. In light of the results of contemporary empirical studies and the discussion concerning the nature of FL aptitude they inspired (see Abrahamsson & Hyltenstam Reference Abrahamsson and Hyltenstam2008, Long Reference Long, Granena and Long2013; Hyltenstam Reference Hyltenstam2014a, Reference Hyltenstam2014b), it is evident that there is a pressing need to reconceptualize the definition of a linguistically talented person. Even though, due to the remarkable paucity of relevant research, redefining exceptional LG poses a considerable challenge, it seems that such an attempt is necessary at least to provoke further discussion of this fascinating phenomenon. After our review of the literature, we are confident that a modified definition of a gifted L2 learner which we proposed in Section 3 – that is: a gifted FL learner is a person who, owing to his/her exceptional inborn gift for learning languages, especially capacious verbal working memory, as well as expertise in L2 learning, is able to learn any foreign language to a near-native level of competence, given proper motivation, time and conducive environment – will contribute to future research on LG. Obviously, this definition should be viewed as tentative, with the effect that further research is necessary to verify and extend it.

Future research on LG should primarily focus on issues that remain highly controversial or have been neglected, such as the development of linguistic talent, the contribution of WM and other types of memory, the role of L1, personality and LLS, as well as the processing of morphosyntax, lexis and pragmatics from a neurolinguistic perspective. There are also other intriguing issues that warrant undertaking more studies of linguistically gifted individuals. One of them is the relationship between LG and intelligence, the latter of which represents a general cognitive ability factor. Traditionally, it is believed that analytic abilities are related to intelligence, while memory and phonetic abilities are independent of it, which indicates that they are language-specific. While the relationship between general, as well as performance (non-verbal) and verbal IQ and FL aptitude has been, on the whole, well described (see Sasaki Reference Sasaki1996; Skehan Reference Skehan1998; Dörnyei Reference Dörnyei2005), there is a marked lack of research on the relationship between different FL aptitude components and other cognitive abilities, both higher-order and primary. This problem is especially perplexing in the studies of gifted L2 learners.

Aptitude for implicit learning (Granena Reference Granena, Granena and Long2013) is also worth further exploration. Another limitation of research on aptitude is the absence of focus on the pragmatic aspect of language, facets of communicative competence other than its systemic component, or patterns of use of LLS. A broader, albeit no less intriguing, issue is the nature of LG, operationalized in terms of FL aptitude, which, as measured by traditional instruments, is considered to be a relatively stable factor. This brings with it the assumption, evident in the concept of mindsets (Mercer Reference Mercer2012), that all abilities can be developed, modified and improved. Therefore, the distinctive characteristics of talented individuals are their passion, dedication, and, first and foremost, the manner in which they identify and tackle their weak points, with the effect that giftedness is neither fixed nor inborn, and cognitive ability is a quality that is modifiable and dynamic (see Sternberg Reference Sternberg and Robinson2002). Consequently, Mercer (Reference Mercer2012) argues that it is necessary to deconstruct the myth of a natural-born genius and replace it with an explanation of success as a joint product of the world an individual lives in and affordances resulting from the environment (see Aronin & Singleton Reference Aronin and Singleton2012). This, of course, calls for other types of research on LG than those predominantly used, and, perhaps, also studies that would seek to integrate the two opposing views, as this would allow us to examine more precisely the idiosyncratic paths of development of linguistically gifted learners.

This last point brings us to methodological issues, as it indicates the need to reconcile various theoretical perspectives in the study of individuals endowed with exceptional talent for L2 learning. Such reconciliation or integration could reasonably take place within a single study, where quantitative data could be collected by means of well-established instruments, and qualitative data could be obtained through recordings, interviews or stimulated-recall sessions. It could also occur through conducting separate studies, longitudinal in nature, the results of which could be synthesized with a view to generating a multi-faceted picture of LG. As regards more detailed recommendations, it is clear that more uniform criteria for selection of linguistic talents should be adopted. There needs to be more consistency in measurements of ultimate attainment or, more generally, FL mastery, and even in studies which adhere to the quantitative paradigm, data should routinely be gathered about L1 learning experiences, personality types, learning styles or LLS. Particularly promising is the use of data collection tools representing a neurolinguistic approach, although such research needs to be extended to cover not only phonology, but also morphosyntax, lexis and pragmatics. Studies conducted in such ways can help us better understand the nature of LG, the requirements for its occurrence, and the steps that can be taken to stimulate its development.