Does long-term dual-language immersion affect children’s executive functioning?

Highlights

• Dual-language immersion majority-language children were compared to monolinguals.

• Bilinguals initially better inhibited a response than monolingual children.

• After a year, monolingual children performed similarly to bilingual children.

• Effects of DLI on executive functioning are limited in scope and transitory.

Abstract

Dual-language immersion (DLI) experience has been linked to enhanced reading and math skills in minority- and majority-language elementary school children. However, it remains unclear whether DLI experience can also enhance executive functioning. The current study took a longitudinal approach to this question and examined the effect of DLI experience on the development of executive function skills in majority-language children over a 1-year period. In total, 33 monolingual children attending English-only classrooms (M_age = 9.17 years, SD = 1.03) and 33 English–Spanish bilingual children attending DLI classrooms (M_age = 9.27 years, SD = 0.94) matched on age, gender, nonverbal IQ, and socioeconomic status were tested twice, 1 year apart, on nonverbal measures of inhibition, shifting, switching, and monitoring. Results revealed a significant interaction between group and year only on the response inhibition task, with bilinguals showing superior inhibition in Year 1 but not in Year 2. The two groups performed equivalently on all other measures at both time points. Results suggest that classroom DLI has a minimal impact on executive functions, at least as tested in the current study.

Introduction

Dual-language immersion (DLI) programs bring together native English speakers and English-language learners and allow them to acquire language and literacy skills in two languages. Although DLI experience has been associated with enhanced academic and language outcomes in both the native language (L1) and second language (L2) (Lindholm-Leary and Genesee, 2014, Marian et al., 2013), less is known about whether DLI experience yields executive function (EF) advantages (Carlson and Meltzoff, 2008, Kalia et al., 2019). In this study, we examined the effect of DLI experience on the development of EF skills in majority-language children—native speakers of the dominant language in society—to ensure group comparability with monolinguals, especially in terms of socioeconomic status (SES) and native language skills. The bilingual children in our study were native English-speaking children attending Spanish–English DLI programs in the United States. We measured children’s progress over the course of 1 year to better capture the trajectory of EF development in intensive L2 immersion.

DLI programs allow children to develop proficiency in a second language while maintaining and developing their skills in their native language and learning grade-appropriate content (Calderon and Minaya-Rowe, 2003, Cloud et al., 2000). Two-way immersion (TWI) is a type of DLI that brings together children who are native speakers of the dominant language in society (i.e., majority-language speakers) and learners of that language (i.e., minority-language speakers). In TWI, classroom makeup consists of a ratio of approximately 50:50 majority- and minority-language speakers, facilitating interactions with native speakers of both languages. TWI leads to positive outcomes for both groups of children in terms of native language development, second language acquisition and proficiency attainment, and academic achievement (Calderon and Minaya-Rowe, 2003, Esposito and Bauer, 2018, Genesee, 2004, Lindholm-Leary and Genesse, 2010, Lindholm-Leary and Genesee, 2014). However, intensive L2 exposure may also be associated with some linguistic challenges in the L1.

Indeed, compared with monolinguals in non-immersion programs, majority-language students in 90:10 DLI programs (where children in the first year of the program receive 90% exposure in the minority language) develop their L1 reading, writing, speaking, and listening skills at a slower rate (Genesee, 2004). A similar process has been observed in young adults tested 3 months into foreign-language immersion; whereas comprehension and production fluency increase in the L2, access to the L1 is reduced (Linck, Kroll, & Sunderman, 2009). The lag in L1 development in bilingual children might be explained by the fact that a bilingual’s two languages compete for selection (Costa et al., 2003, Kroll et al., 2014, Marian and Spivey, 2003, Poarch and van Hell, 2012), creating additional cognitive demands (e.g., Bialystok, 2009, Kalia et al., 2019, Poarch and van Hell, 2012). A potential consequence of these increased demands in the linguistic domain that require children to practice inhibiting one language while using the other language is a cognitive functioning advantage. In general, bilingual children frequently need to switch between their two languages and therefore need a control mechanism to activate the relevant language in context while “tuning down” or inhibiting the other language (e.g., Bialystok, 2007, Bialystok and Craik, 2010, Green, 1998). Inhibiting, shifting, switching, and monitoring are some of the processes that constitute executive functioning. EFs can be defined as “general-purpose control mechanisms that modulate the operation of various cognitive subprocesses and … regulate the dynamics of human cognition” (Miyake et al., 2000, p. 50). Simultaneous acquisition of two languages in parallel from birth or within a few years from birth (Kohnert, 2010) may have an impact on children’s executive functioning (e.g., Bialystok, 1999, Bialystok, 2010, Bialystok and Martin, 2004, Carlson and Meltzoff, 2008, Martin-Rhee and Bialystok, 2008), although the literature on bilingual EF advantages has become increasingly contentious (e.g., Morton and Harper, 2007, Morton and Harper, 2009, Namazi and Thordardottir, 2010, Paap and Greenberg, 2013).

Studies testing the effects of bilingualism on EFs often measure executive functioning using nonverbal tasks that require applying attentional control to focus on relevant information to complete the task while inhibiting irrelevant information. Many tasks have been used to measure inhibitory control, which is both the ability to suppress a dominant response and the capacity to resist interference from distracting input (e.g., Tiego, Testa, Bellgrove, Pantelis, & Whittle, 2018). For example, simultaneous bilingual children aged 5 to 8 years were shown to perform better than monolinguals on the nonverbal Simon task, which requires suppression of interfering information (e.g., children must press the left button when seeing the word “left” even if the word is presented on the right side of the screen) (Martin-Rhee & Bialystok, 2008). In another nonverbal task, the flanker, children must focus on the direction of a target item at the center of a line of five items (e.g., fish, arrows) while inhibiting the direction in which the neighboring items are pointing. These items may be pointing in the same direction (congruent trials) or in opposite directions (incongruent trials). Bilingual children aged 8–11 years outperformed monolinguals on incongruent trials, suggesting better inhibitory control (Poarch & Bialystok, 2015). Simultaneous bilingual children have also been shown to outperform their monolingual peers on measures of task shifting. One such measure is the Dimensional Change Card Sort (DCCS) task (Zelazo, Frye, & Rapus, 1996), a widely used omnibus executive control task that requires switching, shifting, and monitoring rules to sort cards (Bialystok, 1999, Bialystok and Martin, 2004).

The DCCS involves three conditions: a pre-shift in which participants know and apply only one rule, a post-shift in which participants must apply the new rule, and a mixed condition in which either of the two rules must be applied. From these conditions, three cost measures can be derived: shifting costs, switching costs, and mixing costs. Shifting costs are derived from the difference between the post-shift and pre-shift conditions and index the capacity to overcome rule perseveration (Frye, Zelazo, & Palfai, 1995). Switching costs are derived from the difference between switch and non-switch trials in the mixed condition and index the ability to switch back and forth between two rules (Prior & MacWhinney, 2010). Mixing costs are derived from the difference between non-switch trials in the mixed condition and pre-shift trials and index the ability to monitor the application of a rule while knowing more than one rule (Prior & MacWhinney, 2010). Evidence supporting a bilingual advantage on these cost measures has been found in several studies (e.g., Bialystok, 1999, Bialystok and Martin, 2004). For instance, in Okanda, Moriguchi, and Itakura (2010), bilingual children were more accurate than monolingual children on the shifting measure of the DCCS. Prior and MacWhinney (2010) showed that bilinguals exhibited lower switching costs, indicating that they were faster than monolinguals to respond to switch trials. Barac and Bialystok (2012) found that bilinguals exhibited lower mixing costs than monolinguals, suggesting a bilingual advantage on monitoring skills. However, bilingual advantages on the DCCS cost measures have not always been replicated (e.g., Morton and Harper, 2007, Paap and Greenberg, 2013). For example, Paap and Greenberg (2013) did not find any significant group differences between bilinguals and monolinguals on the same task used in Prior and MacWhinney (2010).

Similarly, mixed results have been found on the go/no-go task when comparing bilingual and monolingual children’s response inhibition abilities. The go/no-go paradigm requires participants to respond to stimuli with the exception of specific items, the “no-go” stimuli. Both the flanker and go/no-go measure inhibition, albeit different aspects of it; the flanker measures the ability to resist interference, whereas the go/no-go measures the ability to inhibit prepotent responses (Bunge et al., 2002, Kaushanskaya et al., 2017). In a study using both behavioral and electrophysiological measures to assess potential differences between bilingual and monolingual 5-year-old children on simple response inhibition (gift delay) and complex response inhibition (go/no-go) tasks, bilinguals outperformed monolinguals on the go/no-go task, suggesting higher executive control capacity for bilinguals (Barac, Moreno & Bialystok, 2016). However, this bilingual advantage was not found when comparing monolingual and bilingual groups of children who were on average 9 years old (Bonifacci, Giombini, Bellocchi, & Contento, 2011).

Large-scale studies have found mixed evidence of the putative bilingual advantage on executive functioning. For example, in a study involving 252 monolinguals and 252 bilinguals matched on various knowledge and cognitive measures, children completed verbal and nonverbal Stroop tasks (Duñabeitia et al., 2014). Both tasks involved inhibiting irrelevant information when responding to stimuli. Findings showed that both groups performed similarly on each task, suggesting no differences in inhibitory control capacity across monolinguals and bilinguals. It has been suggested that inconsistencies in the degree to which bilingual and monolingual groups are matched on several knowledge and cognitive measures can explain some of the discrepant results in the bilingual EF literature.

In a review of studies on bilingualism and nonlinguistic executive functioning published from 2010 to 2014, Hilchey, Saint-Aubin, and Klein (2015) concluded that the variability in sociolinguistic factors across monolinguals and bilinguals across the lifespan limits replication of studies and confidence in associating bilingualism with specific cognitive advantages. Similarly, Paap, Johnson, and Sawi (2015) discussed factors that interact with measures of EFs in bilinguals such as SES and cultural differences. For example, Engel de Abreu, Cruz-Santos, Tourinho, Martin, and Bialystok (2012) found that bilinguals outperformed monolinguals on monitoring and inhibiting measures. However, the bilinguals lived in Luxembourg, whereas the monolinguals lived in Portugal, introducing confounds in terms of SES and cultural differences that might account for some of the bilingual advantage observed. Indeed, in two large-scale studies by Antón et al., 2014, Duñabeitia et al., 2014 involving Basque–Spanish bilinguals of the same age and older than bilinguals in Engel de Abreu et al. (2012), matched on SES and all native residents of Spain, no advantages were found for the bilinguals on executive control measures. Regarding cultural differences, a study by Carlson and Choi (2009) showed that when comparing Korean–English bilinguals living in the United States and American English monolinguals on six different measures of executive processing, bilinguals exhibited cognitive advantages. However, when these bilinguals were compared with Korean monolinguals, the advantages disappeared, suggesting that cultural differences over and above mono- or bilingualism might be contributing the executive processing differences observed.

One way to limit these confounds is to compare children who live in a similar environment but who differ in their language experiences. Such an environment is created by DLI classrooms, which provide majority-language children with extensive L2 experience. In the current study, English-speaking children attending Spanish–English DLI classrooms and English-speaking children attending English-only classrooms all spoke English as their first language, came from the same town, attended the same schools, and shared a cultural background. Their families were highly comparable in their SES, and the children were highly similar in their English-language skills. A comparison of these two groups on EF skills therefore affords a rare opportunity to test the effect of bilingual language experience on EFs outside of the sociocultural and linguistic correlates of bilingualism inherent in testing simultaneous bilingual children (vis-à-vis monolingual children).

The adaptive control hypothesis (ACH) is useful in formulating specific predictions regarding how the language environment of majority-language speakers attending DLI might shape specific EFs (Green & Abutalebi, 2013). The ACH rests on the assumption that language comprehension and production require control processes. It follows that in the context of bilingual interactions, how the two languages are used poses variable demands on cognitive control, compelling language control processes to adapt to these demands. Green and Abutalebi (2013) identified eight types of control processes: goal maintenance, conflict monitoring, interference suppression, salient cue detection, selective response inhibition, task disengagement, task engagement, and opportunistic planning. They delineated three contexts in which these control processes will adapt differently: the single-, dual-, and code-switched language contexts. In a “dual-language context,” a bilingual’s two languages are used with different interlocutors in the same environment (e.g., classroom, family). As such, language switching is more likely to take place between conversations rather than within an utterance. In a “single-language context,” a bilingual uses one language at home and one language outside the home, and therefore switching between languages will not be frequent. In a “dense code-switching context,” bilingual speakers frequently mix both languages within an utterance. The authors emphasized that goal maintenance, conflict monitoring, and interference suppression all are necessary for a bilingual to choose what language to speak. As a result, both the single- and dual-language contexts make demands on these processes over and above those imposed by the dense code-switching context. However, the dual-language context makes additional demands in that the bilingual speaker must detect salient cues to control which language to use with a given interlocutor. If the inappropriate language in context is selected due to a robust prepotent response, the speaker will need to apply selective response inhibition to first trigger “task disengagement” (stop using the language) and subsequently generate “task engagement” (start using the other language). In the code-switching context, because dual-language use is expected at the utterance level, no such additional demands are made on these control processes. However, the speaker must apply “opportunistic planning” to adapt words from one language to fit in the frame of another language, a control process that is not readily needed in either of the other two contexts.

When considered from the perspective of the ACH, a DLI classroom for majority-language speakers in part resembles a dual-language context because majority-language children have the opportunity to speak the majority language with native speakers of the majority language and to speak the minority language with native speakers of the minority language. It is unclear, however, whether children actually maintain such a separation of their languages outside the classroom. Furthermore, for majority-language children, the exposure to the minority language happens only at school, with most children experiencing little input in the minority language in their homes. Thus, the DLI classroom also has characteristics of a single-language context for majority-language speakers. Given this conceptualization of DLI for majority-language speakers, who are likely to engage in minimal code switching, it can be hypothesized that the EFs most likely to be affected by their language experience would be monitoring, resisting interference and inhibiting prepotent responses.

However, similar to studies with simultaneous bilinguals, studies on DLI bilinguals, including majority- and minority-language children, present mixed evidence regarding cognitive advantages for bilingual children as a result of immersion (Carlson and Meltzoff, 2008, Kalia et al., 2019). In a study comparing monolingual native English speakers with both majority- and minority-language speakers who had been in English–Spanish DLI for at least 9 months, children aged 5 to 9 years were administered two measures of executive functioning: the DCCS and a Lexical Stroop Sort task (Kalia et al., 2019). The DLI bilinguals outperformed the monolinguals on both EF tasks. In addition, there were no differences in performance on the EF tasks between minority- and majority-language children in DLI (Kalia et al., 2019). In Poarch and van Hell (2012), four groups of children (5–8 years of age) were tested. Monolinguals were compared with simultaneous bilinguals, majority-language speaker DLI bilinguals (immersed on average for 1.3 years), and trilinguals on the Simon task and the Attentional Networks Task (ANT), measuring conflict detection and resolution ability. The simultaneous bilinguals and trilinguals, but not the majority-language DLI bilinguals, performed better on both EF tasks than the monolinguals, suggesting that a threshold of extended practice in inhibitory control must be passed to begin observing enhanced executive functioning in bi- and trilingual children.

Similarly, Carlson and Meltzoff (2008) compared monolinguals, simultaneous bilinguals, and majority-language speaker DLI bilingual children (4–6 years of age), who had been immersed for 6 months, on a battery of EF tasks. After statistically controlling for verbal ability, SES, and age, simultaneous bilinguals demonstrated significantly better performance on the EF tasks over the monolinguals and DLI bilinguals, who did not statistically differ from each other. In Kaushanskaya, Gross, and Buac (2014), monolingual children and majority-language speaker DLI bilinguals (immersed for ~ 2 years on average) (5–7 years of age) were compared on measures of word learning, verbal short-term and working memory, and a measure of nonverbal task shifting. No differences were found between groups on the task-shifting and verbal short-term memory measures. However, DLI bilinguals outperformed monolinguals on the verbal working memory and word learning measures. These findings suggest that although exposure to the L2 in DLI might not translate to executive control advantages in children, it does support the development of verbal memory and word learning ability.

Overall, these studies involved children who had been in immersion for 6 months to 3 years, and less is known about how children perform in later years of the DLI. It is possible that the absence of EF advantages in DLI bilinguals who are majority-language speakers observed in prior studies is related to the relatively short period of time they had spent in the immersion classrooms. Therefore, in the current study, we examined how older children with more DLI experience (4 years) might perform on EF tasks. We focused on the majority-language group to avoid the issues previously discussed involved in matching bilingual and monolingual groups. In our sample, native speakers of the dominant language in DLI, the majority speakers, were similar to their monolingual peers in all aspects but classroom environment.

We examined the effect of intensive L2 exposure on the development of inhibiting, shifting, switching, and monitoring functions in children who had been in DLI for an average of 4 years at the beginning of the study. We focused on majority-language speakers in DLI in order to avoid the difficulty of matching children on SES or language proficiency. We tested children twice, over a period of 1 year, to examine the possibility that children’s maturation may contribute to differential progress in executive functioning depending on group—monolingual or bilingual.

To assess the potential impact of DLI experience on executive functioning, we tested children in the age range of 8–10 years. Within this age range, we expected children’s language functioning to largely stabilize and expected children to gain a significant degree of Spanish experience and skill. At the same time, their EFs would still be maturing (Anderson et al., 2010, Stuss, 1992), allowing for the possibility of movement in the EF skills as a result of language experience. Children in the two groups (DLI and mainstream classrooms) were native speakers of English, characterized by highly similar SES, ethnicity, language, and cultural profiles. At the time of the study, the bilingual children had been immersed in the DLI program for approximately 4 years. We predicted that if intensive L2 exposure in DLI affects performance on EF measures, then DLI bilinguals would perform better than monolinguals on these measures. Moreover, we examined the developmental trajectory of EFs in this age range and predicted that both groups would perform better on EF skills over time.

The benefit of the longitudinal design is its potential to capture group differences that may be moderated by children’s maturation levels and cognitive skills. EF tasks are notoriously sensitive to participants’ level of cognitive functioning, and the same task may yield optimal levels of performance in one age group and lead to floor effects in another age group. We considered the possibility that one potential reason for murky findings within the bilingual EF literature might be the different degrees of sensitivity of the same EF task at different ages and levels of cognitive maturity. Therefore, we tested children twice over a 1-year period, expecting that should our particular EF tasks be more or less sensitive to bilingual experience at different ages, we would be better able to capture this effect if we tested children longitudinally. The theoretical ramification of the longitudinal design is that it enables testing of the persistence of the bilingual effect on EFs. If DLI experience affords a lasting effect on EFs, we should observe bilingual advantages on our EF tasks at both testing time points. Thus, we conducted a 1-year longitudinal study to capture the potential long-term effects of DLI immersion on children’s EF skills. The children in the DLI programs became bilingual as a result of 4 years in DLI and experienced 5 years in DLI by the end of the study.

We focused on the EF measures that have played a central role in the previous bilingual EF literature—inhibitory control (as indexed by flanker and go/no-go tasks) (Gunnerud, Ten Braak, Reikerås, Donolato, & Melby-Lervåg, 2020) and task shifting (as captured by the different cost indices of the DCCS task). These particular EFs are expected to be salient in a DLI context when considered within the theoretical framework of the ACH (Green & Abutalebi, 2013). We did not make distinct predictions for each of the EF measures. Rather, we were interested in whether the effects of DLI experience would be specific to any one of these EFs or whether they would generalize across tasks. In general, we hypothesized that children in DLI must consistently inhibit one of their languages (because the two languages are present sequentially in classrooms and because code switching is generally absent in DLI classrooms, at least in teachers’ communication) (e.g., Spooner and Arias Olsen, 2017, Wei and Martin, 2009). Therefore, we hypothesized that should DLI experience have an effect on EFs, this effect would be strongest for the inhibitory control measures and for the monitoring aspect of EF and less so for the shifting and switching measures.