Baum and Titone's Keynote Article is a welcome addition to the ongoing and flourishing study of the neurocognitive effects of bilingualism. By focusing their review on executive control in bilinguals from the perspectives of both aging and neuroplasticity in general they have broadened our thinking in useful ways. Because my own research has focused on the early stages of aging and bilingualism in preschool and school-age children, I have chosen to comment on their review with reference to these kinds of learners. I will also focus on the language acquisition part of this discussion because at the heart of the issue is how and to what extent acquisition of more than one language affects cognitive development. My comments reinforce what I believe are important observations in their review and simultaneously seek to extend their call to expand our frames of reference in research on bilinguals even further.
Baum and Titone end their Keynote with three suggestions on how to conduct research to advance our understanding of bilingualism, aging, and neuroplasticity. Of particular significance, they point out that “bilinguals differ in ways that matter,” and they call on researchers to examine this variability by better characterizing learner and background characteristics of bilinguals that might matter by engaging in large-scale epidemiological studies that systematically identify the ways that bilinguals’ experiences matter and by using statistical techniques that go beyond traditional between-group analyses to examine in more detail the links between variability among learners and outcomes of interest. In a very early review of research on the cognitive consequences of bilingualism, Cummins (Reference Cummins1976) likewise pointed out that not all bilinguals are alike. He proposed the threshold hypothesis to explicate why some studies at the time reported advantages for bilinguals in comparison to monolinguals, while other studies reported no differences or even a monolingual advantage. He noted that a bilingual advantage was evident in studies that included bilinguals who were highly proficient in both languages whereas studies that reported a relative cognitive disadvantage among bilinguals usually included bilingual participants whose proficiency in one or both languages was relatively low. He went further to propose alternative explanations for these threshold effects. Variability among bilinguals was also highlighted by Lambert (Reference Lambert and Hornby1977) in his notion of additive and subtractive forms of bilingualism. He argued that positive cognitive and academic outcomes among young learners were likely to result from additive forms of bilingualism, which he characterized as learning environments that foster high levels of competence in both languages. In contrast, he argued, limited or negative cognitive and academic outcomes were likely to result from subtractive bilingualism experiences that are associated with learning environments in which acquisition of a second language displaces or overtakes acquisition of the native language, as in the case of many immigrants who are educated in a second, societally dominant language.
What is particularly significant about Lambert's notions of additive and subtractive forms of bilingualism is the decidedly sociocultural nature of the distinction. In effect, he was hypothesizing that sociocultural factors external to the learner might be expected to influence learner outcomes related to language and cognitive development. This is significant because it makes the case that it is not only intraindividual variation that might be important in understanding bilingualism but also variation in the broader sociocultural environment in which bilinguals grow up. Notwithstanding our lack of understanding of the precise ways in which this might occur, education researchers have long argued that understanding variations in learners’ backgrounds is indeed critical for understanding their linguistic and academic achievement. A specific case in point concerns bilingual versus English-only forms of education for minority language students, often referred to as English language learners (ELLs), in the United States, for example, students who speak only or mainly Spanish at home.
Bilingual education for ELLs consists of up to 90% of instruction in the primary grades in the home language followed by a combination of instruction in both the home language and English. Opponents of bilingual education have used evidence from research on Canadian immersion programs to argue that education in English only is the preferred and most effective approach for ELLs and that providing education through the home language detracts from their acquisition of English. Research conducted in Canada has consistently shown that majority group English-speaking students in early French immersion programs, who are initially educated only or primarily in French, attain age-appropriate native language and academic skills while also acquiring advanced levels of functional proficiency in French (Genesee, Reference Genesee, Bhatia and Ritchie2004). Quite simply, the argument made by opponents of bilingual education in the United States is that if education through a second language is so effective for English-speaking students in Canada, then education in English only will be equally effective for ELLs. Yet, it has been found repeatedly that ELLs in English-only programs in the United States consistently score significantly below native English-speaking students and below state and national norms on standardized tests of English and academic achievement (US Department of Education, 2008). In contrast, ELLs in bilingual programs have been shown to achieve superior English language and academic outcomes in comparison to comparable students in English-only programs, and at the same time, they acquire significantly more competence in their native language (for a review, see Genesee & Lindholm-Leary, Reference Genesee, Lindholm-Leary, Harris, Graham and Urdan2012). Space limitations preclude delving into the precise explanations that have been proposed for the differential effectiveness of education through a second language for majority versus minority language students (for detailed explanations, see Genesee & Lindholm-Leary, Reference Genesee, Lindholm-Leary, Harris, Graham and Urdan2011). Suffice it to say that these cross-national comparisons demonstrate that even broadly defined sociocultural factors can help explain the differential effects of different forms of bilingual education on students with different ethnolinguistic backgrounds (see also Paradis, Genesee, & Crago, Reference Paradis, Genesee and Crago2011, p. 8). In brief, background differences among learners in bilingual environments can make a difference in both language development and academic success and, by implication, cognitive development.
The vast majority of research on bilingualism and cognition has made comparisons between bilinguals and monolinguals, although this is beginning to change (for examples see Mercier, Pivneva, & Titone, Reference Mercier, Pivneva and Titone2013; and Iluz-Cohen & Armon-Lotem, Reference Iluz-Cohen and Armon-Lotem2013). Although such comparisons may be a useful starting point for studying bilinguals, this approach has limitations. It is often assumed in such studies that monolingual performance is the norm or default against which to understand other forms of language learning and its effects. From this perspective, differences between bilinguals and monolinguals are often interpreted as advantages in favor of one group or the other. However, such interpretations fail to recognize the particularities of each group. The limitations of this approach are particularly evident when differences favor monolinguals and, accordingly, are interpreted as deficits on the part of bilinguals. For example, evidence that bilinguals often score lower than monolinguals on vocabulary tests is often interpreted as a deficit on the part of bilinguals. However, this interpretation fails to consider the distributed nature of bilingual acquisition and use or the evidence that the cost for switching between languages risks viewing the bilingual as “two monolinguals in one” rather than as possessing a different and possibly unique neurocognitive architecture that allows for complex and interactive forms of language representation, processing, and use. Differences in favor of bilinguals that are interpreted as advantages are arguably equally problematic. In short, simple comparisons between bilinguals and monolinguals risk making the assumption that there is only one normal way to learn and use language, the monolingual way, and that variations on that norm result in deviations. From a neurobiological point of view, this seems an untenable assumption.
That simple comparisons between monolinguals and bilinguals are limited also comes from a growing body of research that suggests that what is often taken to be nativelike language ability or performance may actually be unachievable except in cases of monolingual acquisition. For example, research by Abrahamsson and Hyltenstam (Reference Abrahamsson and Hyltenstam2009) found that less than 40% of the children in their study who had immigrated to Sweden during the preschool years demonstrated nativelike (i.e., monolingual) performance on a battery of demanding language tasks even 20 years after immigration. Our own research has found significant lags in the language development of internationally adopted children in comparison to nonadopted monolingual children, even though the adoptees began learning the adopted language exclusively between 12 and 24 months of age and were tested up to 10 years post adoption. The adopted children demonstrated no lags in general health or in cognitive, socioemotional, attentional, or nonverbal development, indicating that their lags in language were not due to more general developmental delays (Delcenserie, Genesee, & Gauthier, Reference Delcenserie, Genesee and Gauthier2013). That the learners in these studies failed to demonstrate the nativelike abilities demonstrated by the monolingual participants suggests that the gold standard of monolingual nativelike competence is misguided because it fails to recognize the complex circumstances in which language learning, especially when more than one language is involved, can take place (for similar points of view, see also Cook, Reference Cook1992; and Grosjean, Reference Grosjean1992). We need alternative conceptualizations of nativelike competence or at least different frames of reference for understanding language learning in different contexts.
An unfortunate side effect of the monolingual bias is a conceptualization of languages as closed systems, with the languages of bilinguals being compartmentalized modules, and of language learning as a relatively invariant progression to some idealized nativelike level of competence, with minor perturbations due to variations in input and major perturbations due to relatively rare cases of endogenous impairment, such as specific language impairment. Research on young bilinguals, including research on code-switching, communication in bilingual contexts, and language acquisition in preschool bilinguals, challenges these assumptions (Genesee, Reference Genesee2009). This growing body of research is beginning to provide a picture of the language faculty that is more flexible, complex, and supple than that based on data from only monolingual acquisition. Surely, the ultimate goal of research on bilinguals, aside from understanding bilinguals, is to contribute to the development of a unified theory of language acquisition, representation, and use that encompasses all learners. Accomplishing this ultimate goal will require going beyond current conceptualizations of language and language learning in order to facilitate the integration of empirical evidence from monolinguals and bi/multilinguals. This in turn calls for the kind of multivariate approach that Baum and Titone recommend, an approach that views learning in monolingual and bi/multilingual contexts in terms of differences in variation rather than differences in kind. In short, I could not agree more with Baum and Titone that future research needs to embrace research methods and statistical techniques that examine “the inherent individual variability among bilinguals and monolinguals (my addition) in all its glory.” Baum and Titone's recommendations for more attention to learner variability is an important and critical step in that direction. I hope that these recommendation will be taken more seriously than similar recommendations in the past. There are encouraging signs that this is beginning to happen. For example, recent work by Bialystok has begun to explore metalinguistic awareness and executive control changes in subgroups of bilinguals who differ in exposure to and proficiency in the second language. (Bialystok & Barac, Reference Bialystok and Barac2012). In addition, a very welcome advance in the field is the research by Titone and her students that has been examining how individual differences in executive control influence language processing among bilinguals (Mercier et al., Reference Mercier, Pivneva and Titone2013).
You have
Access
Baum and Titone's Keynote Article is a welcome addition to the ongoing and flourishing study of the neurocognitive effects of bilingualism. By focusing their review on executive control in bilinguals from the perspectives of both aging and neuroplasticity in general they have broadened our thinking in useful ways. Because my own research has focused on the early stages of aging and bilingualism in preschool and school-age children, I have chosen to comment on their review with reference to these kinds of learners. I will also focus on the language acquisition part of this discussion because at the heart of the issue is how and to what extent acquisition of more than one language affects cognitive development. My comments reinforce what I believe are important observations in their review and simultaneously seek to extend their call to expand our frames of reference in research on bilinguals even further.
Baum and Titone end their Keynote with three suggestions on how to conduct research to advance our understanding of bilingualism, aging, and neuroplasticity. Of particular significance, they point out that “bilinguals differ in ways that matter,” and they call on researchers to examine this variability by better characterizing learner and background characteristics of bilinguals that might matter by engaging in large-scale epidemiological studies that systematically identify the ways that bilinguals’ experiences matter and by using statistical techniques that go beyond traditional between-group analyses to examine in more detail the links between variability among learners and outcomes of interest. In a very early review of research on the cognitive consequences of bilingualism, Cummins (Reference Cummins1976) likewise pointed out that not all bilinguals are alike. He proposed the threshold hypothesis to explicate why some studies at the time reported advantages for bilinguals in comparison to monolinguals, while other studies reported no differences or even a monolingual advantage. He noted that a bilingual advantage was evident in studies that included bilinguals who were highly proficient in both languages whereas studies that reported a relative cognitive disadvantage among bilinguals usually included bilingual participants whose proficiency in one or both languages was relatively low. He went further to propose alternative explanations for these threshold effects. Variability among bilinguals was also highlighted by Lambert (Reference Lambert and Hornby1977) in his notion of additive and subtractive forms of bilingualism. He argued that positive cognitive and academic outcomes among young learners were likely to result from additive forms of bilingualism, which he characterized as learning environments that foster high levels of competence in both languages. In contrast, he argued, limited or negative cognitive and academic outcomes were likely to result from subtractive bilingualism experiences that are associated with learning environments in which acquisition of a second language displaces or overtakes acquisition of the native language, as in the case of many immigrants who are educated in a second, societally dominant language.
What is particularly significant about Lambert's notions of additive and subtractive forms of bilingualism is the decidedly sociocultural nature of the distinction. In effect, he was hypothesizing that sociocultural factors external to the learner might be expected to influence learner outcomes related to language and cognitive development. This is significant because it makes the case that it is not only intraindividual variation that might be important in understanding bilingualism but also variation in the broader sociocultural environment in which bilinguals grow up. Notwithstanding our lack of understanding of the precise ways in which this might occur, education researchers have long argued that understanding variations in learners’ backgrounds is indeed critical for understanding their linguistic and academic achievement. A specific case in point concerns bilingual versus English-only forms of education for minority language students, often referred to as English language learners (ELLs), in the United States, for example, students who speak only or mainly Spanish at home.
Bilingual education for ELLs consists of up to 90% of instruction in the primary grades in the home language followed by a combination of instruction in both the home language and English. Opponents of bilingual education have used evidence from research on Canadian immersion programs to argue that education in English only is the preferred and most effective approach for ELLs and that providing education through the home language detracts from their acquisition of English. Research conducted in Canada has consistently shown that majority group English-speaking students in early French immersion programs, who are initially educated only or primarily in French, attain age-appropriate native language and academic skills while also acquiring advanced levels of functional proficiency in French (Genesee, Reference Genesee, Bhatia and Ritchie2004). Quite simply, the argument made by opponents of bilingual education in the United States is that if education through a second language is so effective for English-speaking students in Canada, then education in English only will be equally effective for ELLs. Yet, it has been found repeatedly that ELLs in English-only programs in the United States consistently score significantly below native English-speaking students and below state and national norms on standardized tests of English and academic achievement (US Department of Education, 2008). In contrast, ELLs in bilingual programs have been shown to achieve superior English language and academic outcomes in comparison to comparable students in English-only programs, and at the same time, they acquire significantly more competence in their native language (for a review, see Genesee & Lindholm-Leary, Reference Genesee, Lindholm-Leary, Harris, Graham and Urdan2012). Space limitations preclude delving into the precise explanations that have been proposed for the differential effectiveness of education through a second language for majority versus minority language students (for detailed explanations, see Genesee & Lindholm-Leary, Reference Genesee, Lindholm-Leary, Harris, Graham and Urdan2011). Suffice it to say that these cross-national comparisons demonstrate that even broadly defined sociocultural factors can help explain the differential effects of different forms of bilingual education on students with different ethnolinguistic backgrounds (see also Paradis, Genesee, & Crago, Reference Paradis, Genesee and Crago2011, p. 8). In brief, background differences among learners in bilingual environments can make a difference in both language development and academic success and, by implication, cognitive development.
The vast majority of research on bilingualism and cognition has made comparisons between bilinguals and monolinguals, although this is beginning to change (for examples see Mercier, Pivneva, & Titone, Reference Mercier, Pivneva and Titone2013; and Iluz-Cohen & Armon-Lotem, Reference Iluz-Cohen and Armon-Lotem2013). Although such comparisons may be a useful starting point for studying bilinguals, this approach has limitations. It is often assumed in such studies that monolingual performance is the norm or default against which to understand other forms of language learning and its effects. From this perspective, differences between bilinguals and monolinguals are often interpreted as advantages in favor of one group or the other. However, such interpretations fail to recognize the particularities of each group. The limitations of this approach are particularly evident when differences favor monolinguals and, accordingly, are interpreted as deficits on the part of bilinguals. For example, evidence that bilinguals often score lower than monolinguals on vocabulary tests is often interpreted as a deficit on the part of bilinguals. However, this interpretation fails to consider the distributed nature of bilingual acquisition and use or the evidence that the cost for switching between languages risks viewing the bilingual as “two monolinguals in one” rather than as possessing a different and possibly unique neurocognitive architecture that allows for complex and interactive forms of language representation, processing, and use. Differences in favor of bilinguals that are interpreted as advantages are arguably equally problematic. In short, simple comparisons between bilinguals and monolinguals risk making the assumption that there is only one normal way to learn and use language, the monolingual way, and that variations on that norm result in deviations. From a neurobiological point of view, this seems an untenable assumption.
That simple comparisons between monolinguals and bilinguals are limited also comes from a growing body of research that suggests that what is often taken to be nativelike language ability or performance may actually be unachievable except in cases of monolingual acquisition. For example, research by Abrahamsson and Hyltenstam (Reference Abrahamsson and Hyltenstam2009) found that less than 40% of the children in their study who had immigrated to Sweden during the preschool years demonstrated nativelike (i.e., monolingual) performance on a battery of demanding language tasks even 20 years after immigration. Our own research has found significant lags in the language development of internationally adopted children in comparison to nonadopted monolingual children, even though the adoptees began learning the adopted language exclusively between 12 and 24 months of age and were tested up to 10 years post adoption. The adopted children demonstrated no lags in general health or in cognitive, socioemotional, attentional, or nonverbal development, indicating that their lags in language were not due to more general developmental delays (Delcenserie, Genesee, & Gauthier, Reference Delcenserie, Genesee and Gauthier2013). That the learners in these studies failed to demonstrate the nativelike abilities demonstrated by the monolingual participants suggests that the gold standard of monolingual nativelike competence is misguided because it fails to recognize the complex circumstances in which language learning, especially when more than one language is involved, can take place (for similar points of view, see also Cook, Reference Cook1992; and Grosjean, Reference Grosjean1992). We need alternative conceptualizations of nativelike competence or at least different frames of reference for understanding language learning in different contexts.
An unfortunate side effect of the monolingual bias is a conceptualization of languages as closed systems, with the languages of bilinguals being compartmentalized modules, and of language learning as a relatively invariant progression to some idealized nativelike level of competence, with minor perturbations due to variations in input and major perturbations due to relatively rare cases of endogenous impairment, such as specific language impairment. Research on young bilinguals, including research on code-switching, communication in bilingual contexts, and language acquisition in preschool bilinguals, challenges these assumptions (Genesee, Reference Genesee2009). This growing body of research is beginning to provide a picture of the language faculty that is more flexible, complex, and supple than that based on data from only monolingual acquisition. Surely, the ultimate goal of research on bilinguals, aside from understanding bilinguals, is to contribute to the development of a unified theory of language acquisition, representation, and use that encompasses all learners. Accomplishing this ultimate goal will require going beyond current conceptualizations of language and language learning in order to facilitate the integration of empirical evidence from monolinguals and bi/multilinguals. This in turn calls for the kind of multivariate approach that Baum and Titone recommend, an approach that views learning in monolingual and bi/multilingual contexts in terms of differences in variation rather than differences in kind. In short, I could not agree more with Baum and Titone that future research needs to embrace research methods and statistical techniques that examine “the inherent individual variability among bilinguals and monolinguals (my addition) in all its glory.” Baum and Titone's recommendations for more attention to learner variability is an important and critical step in that direction. I hope that these recommendation will be taken more seriously than similar recommendations in the past. There are encouraging signs that this is beginning to happen. For example, recent work by Bialystok has begun to explore metalinguistic awareness and executive control changes in subgroups of bilinguals who differ in exposure to and proficiency in the second language. (Bialystok & Barac, Reference Bialystok and Barac2012). In addition, a very welcome advance in the field is the research by Titone and her students that has been examining how individual differences in executive control influence language processing among bilinguals (Mercier et al., Reference Mercier, Pivneva and Titone2013).
ACKNOWLEDGMENTS
I thank Johanne Paradis, University of Alberta, for comments on an earlier version of this article.