Bilingualism and the Development of Executive Function: The Role of Attention

Bilingual Effects on Cognitive Development

There is substantial evidence that the language environment that children experience influences the quality of the cognitive systems they develop (6), so it should not be surprising that bilingualism is an important factor in developmental outcomes. The earliest systematic evidence for a beneficial effect of bilingualism came from studies of children’s metalinguistic awareness (7). Bilingual children generally outperformed monolinguals on tasks assessing their understanding of abstract language structure, but the implications of these findings became apparent when it was found that these metalinguistic advantages were largely confined to tasks that included conflict and required control to manage that conflict (8). Hence, the reported bilingual advantage in metalinguistic ability was less about language processing and more about cognitive ability.

The shift from an interest in examining the effect of bilingualism on language-related outcomes to its effect on cognition led to a body of research documenting tasks in which bilingual children outperformed their monolingual peers (review in 9, meta-analysis in 10). Most of the tasks in which bilingual advantages are found are considered to be indicators of executive function (but see discussion below). The usual explanation is that both languages are always active in bilinguals, so the domain-general executive function system is incorporated into language processing to direct attention to the target language and in so doing becomes reorganized, fortified, or both (11). Thus, bilingualism “trains” executive function through its constant recruitment for language selection. An enhancement of executive function is not trivial: executive function is a major predictor of academic success (12) and academic success predicts long term health and well-being (13).

One well-accepted view of executive function is the tripartite model proposed by Miyake and colleagues consisting of inhibition, updating (working memory), and shifting (14, 15). Following this model, if executive function is involved in language processing for bilinguals, then it would be important to identify the precise component that is involved and possibly boosted through this experience. Several researchers have proposed candidates for this effect, with the most common suggestion being inhibition on the assumption that the non-target language is suppressed to avoid interference. However, clear evidence endorsing any one of these components, including inhibition, has not emerged (16). Tasks (e.g., flanker task) and conditions (e.g., incongruent trials) that clearly require inhibition are indeed performed better by bilinguals than monolinguals, but this is often in the context of similar bilingual advantages on other tasks or conditions for which no inhibition is required (e.g., congruent trials) (17) or bilingual advantages in some types of inhibition but not others (e.g., response inhibition versus interference suppression) (18). Thus, there is little evidence for the specificity of bilingual effects on inhibition. What has been reported instead is a wide range of tasks on which bilingual children typically (but not always) outperform monolinguals.

Identifying the Source of Bilingual Advantage

In an attempt to identify the components of executive function affected by bilingualism, several studies have used a version of the flanker task, with the most frequent choice being the children’s Attentional Network Task (19). In this task, children see a line of five fish in which the central fish is designated as the target and the four flanking fish point in either the same (congruent trials) or opposite (incongruent trials) direction. Children are required to “feed” the target fish by indicating the direction it is facing with a keypress. Studies generally report faster or more accurate performance by bilingual children (20). However, as in the adult literature, the bilingual advantages in this task are typically found on both congruent and incongruent trials (21) even though no explicit inhibition is required on congruent trials since no misleading information is presented.

Instead of inhibition, therefore, some researchers have proposed that the source of the bilingual advantage is in monitoring (22, 16), a concept similar to shifting in Miyake’s model but broader than any individual component. In some sense, inhibition is included in monitoring; as one shifts across options, the irrelevant cue or response must be suppressed. One task that incorporates both inhibition and monitoring is the dimension change card sort task (DCCS; 23). A set of cards depicting bivalent stimuli (e.g., colored shapes) needs to be sorted first by one dimension (color) then re-sorted by the other (shape). Young children find this difficult and fail to reclassify the stimuli in the second sorting round. Successful performance requires that children ignore the previous dimension (inhibition) and shift attention to the newly relevant dimension (monitoring). In several studies, bilingual children have been more successful than their monolingual age-mates on this task (24, 25), extending bilingual advantages to multiple components of executive function.

Monitoring also includes the notion of working memory in that successful monitoring requires holding a rule in mind over a set of procedures. Studies assessing possible working memory differences in monolingual and bilingual children have produced mixed results, with some showing no difference between groups (26) and others showing better performance by bilinguals (27). In the adult literature, language group differences in working memory are also inconsistent, although bilingual advantages are more likely to be found when the working memory task is based on nonverbal materials than verbal stimuli (28). Thus, under some conditions working memory is also improved by bilingualism.

These studies investigating bilingual advantages in inhibition, monitoring, and working memory tend to use simple tasks based on specific aspects of processing, in part because the goal is to identify one component of executive function as uniquely responsible for developmental differences in bilingual children. Using this approach, no single component has emerged as decisive. However, another group of studies has taken a broader approach and used tasks that incorporate more integrated reasoning ability. These tasks are difficult to categorize in terms of individual executive function components although they certainly require executive functioning for their solution. The general result is that tasks that are effortful and include perceptually conflicting information are performed better by bilinguals.

One set of studies in this category is investigations of theory of mind. There is considerable debate about the correct interpretation of theory of mind and the role of language proficiency, social awareness, and other factors in its development, but many accounts highlight the central role of executive functions in performing these complex tasks (29 but see 30 for a different view). The tasks are perceptually misleading as well; the appearance-reality task deliberately distorts the identity of an object by making it look like something else and the false belief task alters the function of a known visual target in a brief narrative. But theoretical debate aside, there is an established set of tasks that most researchers agree assess this complex ability. Research comparing performance of monolingual and bilingual children has generally reported better performance by bilinguals (31, 32). Interestingly, an adult adaptation of a false belief task that used eye-tracking methodology showed that bilingual adults looked less at the incorrect option than monolinguals, paralleling error performance in children (33). To the extent that executive function is involved in theory of mind performance, its definition must be based on a more holistic conception than is conveyed by the components inhibition, shifting, and working memory as none of these individual components is obviously primary.

Other conceptually complex visual tasks such as creating novel drawings (34), resolving the dual representation in ambiguous figures (35, 36), and calculating visual perspective (37) have all been shown to be performed better by bilingual children than by their monolingual peers. Although these tasks involve some form of monitoring and inhibiting, they are not traditionally considered to be tests of executive function. Both the lack of consensus for the responsibility of a single executive function component and evidence for bilingual advantages in tasks that are more integrative leave unsolved the precise link between bilingual experience and the reported cognitive advantages.

Connection to Language Use

The assumption in the adult literature is that the bilingual advantage in nonverbal executive functioning can be traced to the use of that system to resolve conflict from jointly-activated languages, making it more efficient across a range of tasks. Support for the claim of joint language activation comes from behavioral (38), eye-tracking (39), ERP (40), and fMRI studies (41). There are two implications of the view that this management of language conflict is the primary mechanism for the bilingual effects on cognition.

The first implication follows from the interpretation that these effects are essentially caused by experience-dependent training and so requires evidence that the advantage increases with more bilingual experience. In a recent study, monolingual participants who undertook a yearlong university-level course in either Introductory Spanish or Introductory Psychology were tested before and after the course on executive function tasks with ERP recordings (42). One task was a nonverbal go-nogo task for which two previous studies have shown ERP waveform differences between monolinguals and bilinguals (but no behavioral differences), with bilingual electrophysiology consistent with better performance (43, 44). In the first session, all participants were equivalent on this measure, but in the second session, the ERP results for participants in the Spanish group significantly shifted towards those reported for bilinguals in the previous studies. Thus, even a small amount of experience learning a second language produced changes in these fundamental processes even in the absence of behavioral differences.

A relation between outcomes and degree of bilingual experience was also found in a study with children. Bialystok and Barac (45) reported two studies of children who were learning a second language through immersion education. None of the children was fully bilingual but they had spent different lengths of time in the program and achieved different levels of proficiency in the second language. Using regression analysis to relate background and learning variables to outcomes, both studies showed that language proficiency predicted performance on metalinguistic tasks, but the length of time spent in the immersion program predicted performance on nonverbal executive function tasks. Thus, for both adults and children, the bilingual advantage in nonverbal executive function emerges with more bilingual experience.

The second implication is that if bilingual advantages depend on managing linguistic conflict, then performance differences between monolinguals and bilinguals should not be found until the individual has built up adequate linguistic representations to create competition between them and sufficient experience in managing them to affect the developing executive function system. The youngest children in the early studies reporting bilingual advantages were 3½- or 4-years old (e.g., studies on theory of mind or DCCS). By this age, children are reasonably verbal, and bilingual children can communicate effectively in both languages. But would bilingual advantages be found in younger children? The first study to investigate this question examined children from 29- to 60-months old performing a number of simple tasks that involved different aspects of control (46), such as a tapping task (if the experimenter taps once, the child taps twice) and reverse categorization (put big animals in the bucket marked baby and little animals in the bucket marked mommy). Even at the youngest age, bilingual children outperformed monolingual children on most tasks.

More dramatic, however, is evidence that differences in performance as a function of language environment can be found in the first year of life. Kovacs and Mehler (47) recorded anticipatory eye movements to a reward that appeared on one side of a display following an auditory cue in 7-month old children who were being raised in monolingual or bilingual homes. After a learning set, the position of the reward changed, so infants needed to override their learned response and look to the opposite side. Only bilingual infants could achieve this; monolingual infants continued to execute the habitual response even though no reward was present. In another study, Singh and colleagues (48) studied habituation and concept formation in 6-month old infants and reported that infants being raised in bilingual homes performed better than monolingual infants on measures of stimulus encoding and recognition. In terms of executive function, the bilingual infants in these studies showed more flexibility and perhaps more inhibitory control over a simple behavior. There is no doubt that managing the conflict from jointly-activated languages is a crucial part of the explanation, but evidence from preverbal infants shows that such conflict management alone is inadequate to explain the emergence of nonverbal executive function differences between monolinguals and bilinguals in early childhood.

If not Language, then What?

It is clear that competition between languages is crucial for the emergence of bilingual advantages in executive function, but two factors challenge that view as the exclusive mechanism. First, monolingual adults routinely experience conflict from competing representations, even linguistic ones (e.g., “cup” vs. “mug”) but these conflicts are not considered to enhance executive functioning. Second, infants have only rudimentary representations of language yet differences between monolingual and bilingual infants are apparent by 7 months. What could be the trigger for the processing differences that lead to enhanced executive function in bilinguals, including infants?

One possibility comes from studies of infants processing a stimulus that is very salient in their environments: talking faces. In two studies, Werker and colleagues (49, 50) presented infants with a silent video of a face that was reading sentences in one language and after infants habituated, switched to a different language and continued reading. The question was whether infants could detect the language change from visual cues alone as evidenced by regained interest in the video. In both studies, the bilingual infants noticed the language switch but monolingual infants did not. This was the case both when the two languages were the same as those in the infants’ environment (50) and when they were completely different from those heard by the bilingual infants (49). Whatever infants were using to make this discrimination was more general than the facial features associated with known languages.

The possibility raised by these studies is that bilingual experience changes the way that attention is directed to the environment. For the infant, the presence of two languages that introduces two sets of sounds, cadences, structures, speakers, and facial configurations draws attention to the contrasts between the systems. Contrasts create novelty, thus attracting more attention and possibly more intense processing than similarity. Thus, bilingual babies may simply attend more carefully to subtle environmental differences. If so, these strategies both improve attentional processing and lead to the creation of more complex representational structure that includes two languages. Once two representational structures are established, the executive function is recruited to maintain attention to the target language. This account is different from the view that the non-target language is inhibited: infants are not resolving conflict between lexical features but rather identifying organized systems that are subtly different and require attentional processing to discriminate.

Older children and adults do not need to infer the presence of two language systems through bottom-up attentional processing because they know that the languages are distinct. However, just as infants direct attention to contrasts between the environmental languages, children and adults are drawn to the contrasting features of the jointly-activated languages. Therefore, it is not only that the two languages are jointly activated but also that bilinguals attend to both languages that creates the need for a general selection mechanism such as the executive function to be recruited into language processing to avoid interference. Put this way, the bilingual advantage is not in inhibition; rather it is the failure of bilinguals to inhibit attention to the non-target language that leads to the involvement of executive function and the eventual consequences for its development and function. Importantly, none of the evidence endorses a particular component of executive function as being responsible for this selection; instead, a more “unified” conception of effortful processing (cf., 14) appears to operate. The cognitive and neural dimensions of this unified executive function are the subject of current investigations.

When Peal and Lambert (3) demonstrated cognitive advantages for bilingual children, their research changed the assumptions about how this perfectly normal experience affected children’s development. Although there is still much we do not know, it is a sign of the success of this research that these assumptions have changed. Current studies that report no difference between groups present themselves as a challenge to claims that there are bilingual advantages rather than to claims that there are bilingual disadvantages, although both interpretations are statistically equivalent outcomes of a null result. We have come a long way from the time when the prevailing belief was that children could be harmed by the languages people spoke to them.