Abstract
This study investigated the effects of bilingualism on set‐shifting and
working memory in children with autism spectrum disorders (ASD). Bilinguals
with ASD were predicted to display a specific bilingual advantage in
set‐shifting, but not working memory, relative to monolinguals with ASD. Forty
6‐ to 9‐year‐old children participated (20 ASD, 20 typically‐developing).
Set‐shifting was measured using a computerized dimensional change card sort
(DCCS) task, and by parent report of executive functioning in daily life.
Results showed an advantage for bilingual relative to monolingual children with
ASD on the DCCS task, but not for set‐shifting in daily life. Working memory
was similar for bilinguals and monolinguals with ASD. These findings suggest
that bilingualism may mitigate some set‐shifting difficulties in children with
ASD.
The impact of bilingualism on cognition is an ongoing topic of debate.
Of particular interest is whether the use of two languages confers an advantage
to executive functions (EF). A number of researchers have reported enhanced
performance on EF tasks for typically‐developing bilingual children (e.g.,
Bialystok & Viswanathan, 2009; Carlson & Meltzoff, 2008),
whereas others have failed to find significant differences between monolingual
and bilingual groups (e.g., Morton & Harper, 2007; Namazi &
Thordardottir, 2010). EF abilities have also been investigated in monolingual
children with autism spectrum disorders (ASD), generally showing poorer
performance relative to typically‐developing peers. However, it is not known if
bilingualism may hold consequences for the EF of children with ASD. In the present
study, we test for the first time whether the bilingual advantage hypothesis
extends to children with ASD. Specifically, we compare four groups of children:
bilingual and monolingual children with ASD or with typical development. We
examine whether cognitive flexibility, or set‐shifting skills, may be enhanced
in bilinguals relative to monolinguals, and include working memory skills as a
control EF not expected to be affected by bilingualism.
Executive Functions in Autism
Spectrum Disorders
ASD are neurodevelopmental disorders characterized by impairments in
social communication and social interaction, accompanied by restricted and
repetitive behaviors and interests (American Psychiatric Association, 2013).
Executive functioning has been a major area of interest in autism research
(e.g., Robinson, Goddard, Dritschel, Wisley, & Howlin, 2009), with the
theory of executive dysfunction in autism being proposed as an explanatory
account for the condition (Ozonoff, Pennington, & Rogers, 1991; see Hill,
2004for a critical review of this account). EF comprise an array of cognitive
abilities necessary for goal‐oriented problem solving in everyday life,
including planning, inhibition, working memory, and set‐shifting (Eigsti, 2011;
Gioia, Isquith, Kenworthy, & Barton, 2002). These abilities have neural
bases in the frontal cortex and frontostriatal and frontoparietal pathways
(Schmitz et al., 2006). The EF of set‐shifting (also referred to as cognitive
flexibility) is particularly relevant to ASD. Set‐shifting is the EF domain
exhibiting the most pronounced impairment in daily life for children with ASD
relative to both children with typical development (Granader et al., 2014; in a
sample with over 400 participants in each group), and those with other developmental
disorders (Gioia et al., 2002). Although not included in the diagnostic
criteria for ASD, set‐shifting difficulties are closely related to the second
defining domain of ASD symptomology: repetitive or restricted behaviors and
interests (American Psychiatric Association, 2013). Symptoms falling in this
domain that are commonly (but not universally) experienced by people with ASD
include difficulty with transitions, narrowness of focus, perseveration on
interests and activities, desire for sameness in the environment, and
inflexible adherence to routines (Leekam, Prior, & Uljarevic, 2011).
Set‐shifting has been defined as the ability to accurately switch back
and forth between tasks given specific demands (Miyake et al., 2000). A
quantitative review of set‐shifting in individuals with ASD by Leung and
Zakzanis (2014) included 72 studies, most of which tested adults but also
including the few studies that have tested children and adolescents, yielding a
combined sample of 2,137 individuals with ASD and 2,185 controls. This review
covered 19 laboratory‐based experimental (also referred to as performance)
tasks for which multiple specific measures of set‐shifting were evaluated.
Across these studies, neurotypical participants outperformed participants with
ASD with mean effect sizes from 0.30 to 1.88, as measured by Cohen's d (with
benchmarks of 0.2 for small, 0.5 for medium, and 0.8 for large effects, Cohen,
1992). Interestingly, the one informant‐based measure of set‐shifting in daily
life that was included in the review (Behavior Rating Inventory of Executive
Functioning [BRIEF]; Gioia, Isquith, Guy, & Kenworthy, 2000a), bore
much larger effects, with a mean effect size of 2.14 for children based on
parent report and 5.04 from one study using self‐report with adults. This
review highlights three key points: (a) across a broad range of samples, tasks,
and measurement constructs (but not in each study included) individuals with
ASD exhibited an impairment in set‐shifting, (b) this effect was highly
variable in magnitude, which is not surprising given the scope of the review,
but as such did not meet the authors’ stringent criteria for use of
set‐shifting as a clinical marker for ASD (d ≥ 3.0 and < 5% of overlap
in data points between distributions), and (c) effect size is consistently and
substantially larger for set‐shifting difficulties in daily life relative to
that observed on experimental tasks. This divergence can be attributed to the
fact that experimental tasks isolate set‐shifting skills from real‐life
demands, imposing a standard structure and simplifying the problem space
relative to set‐shifting situations encountered in daily life.
A number of factors have been highlighted in the literature to explain
variable findings on EF tasks in individuals with ASD. One just mentioned is
that of task selection (e.g., report of EF in daily life vs. on performance
tasks, and across performance tasks with more or less explicit rules or
structure, Leung & Zakzanis, 2014; Van Eylen et al., 2011). A second is
with respect to testing modality (e.g., live experimenter administration vs.
computer administration). Specifically, multiple studies have reported that
individuals with ASD perform better on a given task when presented by computer
rather than live administration by an experimenter, whereas
typically‐developing comparison groups perform similarly across modalities
(reviewed by Kenworthy, Yerys, Anthony, & Wallace, 2008). This has been
reported for set‐shifting (Ozonoff, 1995; Pascualvaca, Fantie, Papageorgiou,
& Mirsky, 1998), as well as planning tasks (Kenworthy et al., 2008),
and has been attributed to (a) a decrease in the social demands of completing
the task and/or (b) the motivation typically‐developing children derive from
live interaction that their counterparts with ASD may not display. A third is
the massive heterogeneity among participants with ASD in terms of nonverbal IQ
(NVIQ) and language abilities, necessitating proper characterization of samples
and the use of appropriate matching procedures with comparison groups (Russo et
al., 2007).
A commonly employed experimental task of set‐shifting in children is
the dimensional change card sort (DCCS) task (Frye, Zelazo, & Palfai,
1995; Zelazo, 2006). In a first “pre‐switch” phase, children are asked to sort
a series of simple images depicted on cards (e.g., boats and rabbits) according
to one dimension (e.g., color). Afterwards, in a “post‐switch” phase, they are
asked to sort the same images according to a different dimension (e.g., shape).
Therefore, the task requires that children disengage from the previously used
rule and switch to a new dimension that is explicitly stated (Prior &
Macwhinney, 2010). Typically‐developing 5‐year‐olds are able to perform the
switch correctly, while younger children tend to continue sorting by the first
dimension (Zelazo, 2006). Children who perform correctly during the post‐switch
phase are then administered a more advanced phase in which mixed rather than
blocked trials are presented (i.e., the border version; Zelazo, 2006). In this
advanced phase, the task is to sort images by one of the previously used
dimensions depending on a visual cue (e.g., sort by color if the image has a
border, but by shape if the image has no border). As reported by Zelazo (2006),
different measures (e.g., reaction time [RT]) from the border version have been
used to assess set‐shifting skills in both school‐age children and adults
(e.g., Bialystok & Martin, 2004; Diamond & Kirkham, 2005).
The DCCS, in different testing modalities, has been employed to examine
set‐shifting abilities in monolingual children with ASD (e.g., Dichter et al.,
2010; Faja & Dawson, 2014; Yi et al., 2012). Faja and Dawson (2014)
studied the performance of 20 typically‐developing children and 23 children
with ASD matched on age (6–8 years), gender, and IQ using the standard card
presentation of the DCCS task with a live examiner. Results showed that
accuracy on the post‐switch phase for the children with ASD was comparable to
that of typically‐developing children. However, the typically‐developing
children outperformed their peers with ASD on the border version. Yi et al.
(2012) reported that eighteen 3‐ to 9‐year‐olds with ASD, matched on verbal
mental age to a group of 31 typically‐developing children (3–6 years old),
exhibited deficits on both the post‐switch phase and border versions of a
computerized DCCS task. Group differences on the post‐switch phase were likely
found because it was still developmentally sensitive to this younger sample.
Dichter et al. (2010) employed a computerized card sort task (with a modified
administration procedure, including written words to explicitly provide the
rule for each block) with 6‐ to 17‐year‐olds with and without ASD who were
matched on chronological age but who differed on NVIQ. These authors reported
that participants with ASD performed similarly to the typically‐developing
comparison group on key measures of switching, but their results are difficult
to interpret given lack of matching on NVIQ, a very broad age range, and
differences in task design.
While impaired performance on set‐shifting tasks has been widely
documented, other domains of EF, such as verbal working memory, have been shown
to be spared in children with ASD (Williams, Goldstein, & Minshew,
2006; however see Schuh & Eigsti, 2012 for an alternative view).
Working memory is defined as the ability to temporarily maintain and manipulate
information to accomplish a specific task (Baddeley, 1992). Experimental
paradigms have been used to investigate working memory using visual as well as
auditory stimuli (e.g., Ozonoff & Strayer, 2001). The available
evidence suggests that individuals with ASD show difficulties with complex
working memory tasks that encompass high processing load (Minshew &
Goldstein, 2001), especially in the visual domain (e.g., Williams, Goldstein,
Carpenter, & Minshew, 2005). In contrast, performance on simple verbal
working memory tasks, such as word recall (e.g., Russell, Jarrold, &
Henry, 1996) and digit recall (Bennetto, Pennington, & Rogers, 1996;
Faja & Dawson, 2014; Williams et al., 2005, 2006), has been shown to be
unimpaired.
Overall, children with ASD exhibit an uneven profile in their EF
abilities, with consistent impairments on set‐shifting skills alongside
relatively preserved verbal working memory for simple tasks (as demonstrated by
Faja & Dawson, 2014 in 6‐ to 8‐year‐olds). So far, studies examining EF
in ASD have only included monolingual children. We used Faja and Dawson's
(2014) methods as a model to examine whether bilingualism may confer a specific
advantage in characteristically impaired set‐shifting skills, but not verbal
working memory, in children with ASD.
The Bilingual Advantage Hypothesis
Some researchers have reported that typically‐developing bilinguals
exhibit advantages in particular EF relative to monolinguals (e.g., Bialystok
& Martin, 2004; Bialystok & Viswanathan, 2009). The bilingual
advantage hypothesis posits that living as a bilingual and using two linguistic
systems involves increased cognitive demands. In particular, set‐shifting
skills are involved when (active and proficient) bilinguals exert control over
their two linguistic systems to continuously switch between two competing
languages to successfully meet the demands of their communication context
(e.g., Bialystok & Viswanathan, 2009). The ongoing practice of
switching between languages is thought to yield global advantages in
set‐shifting that can be observed in nonlinguistic domains (Bialystok, 2009;
Green, 1998; Prior & Macwhinney, 2010). In contrast, it has been
suggested that a bilingual advantage is not observed in other EF domains such
as working memory (e.g., Bialystok, 2009; Engel de Abreu, 2011; however see
Delcenserie & Genesee, 2016 findings with adults). A potential cause
for the bilingual advantage effect being more easily observed on set‐shifting
tasks is that these paradigms involve greater cognitive demands (e.g.,
switching between sets, updating, and inhibiting a previously established rule)
than simple working memory tasks (although the latter also requires cognitive
control to some degree, Engel de Abreu, 2011).
Of greatest relevance to the current study, a bilingual advantage in
set‐shifting has been reported for typically‐developing children in studies
using the DCCS task previously discussed or adaptations thereof (Barac
& Bialystok, 2012; Bialystok, 1999; Carlson & Meltzoff, 2008).
Barac and Bialystok (2012) administered a computerized color–shape switch task
to three groups of bilingual 5‐ to 7‐year‐olds (Chinese–English,
French–English, and Spanish–English bilinguals) and to a group of English
monolingual children. Findings revealed that whereas monolingual and bilingual
groups had similar RT for blocks in which switching was not required, for mixed
blocks (e.g., border version of the DCCS task) the three bilingual groups
outperformed the monolingual group by exhibiting shorter RTs.
It should be noted that the bilingual advantage hypothesis has been
challenged by other researchers who have found comparable performance on EF
tasks for bilinguals and monolinguals (e.g., Morton & Harper, 2007;
Namazi & Thordardottir, 2010; Paap & Greenberg, 2013). These
authors ascribed the enhanced performance reported elsewhere for bilingual
participants to higher socioeconomic status (Morton & Harper, 2007),
better memory skills (Namazi & Thordardottir, 2010), or shortcomings of
the statistical analyses conducted such as the presence of Type I errors (Paap
& Greenberg, 2013), among others. Consequently, whether bilingualism
confers and advantage to EF abilities remains a debated topic which warrants
further research.
Although our focus in the present study is on EF, it is important to
address language ability in relation to ASD, as well as in relation to
bilingualism. First, when all children with ASD are grouped, they are
significantly delayed in language development (e.g., Luyster, Kadlec, Carter,
& Tager‐Flusberg, 2008; Weismer, Lord, & Esler, 2010).
Importantly, this group‐level delay coexists with massive individual
variability in language skills among children with ASD. One study found that by
school age approximately half of children with ASD were language impaired as
assessed by standardized tests, a quarter had borderline skills within 2 SD of
the normal range, and the last quarter scored in the normal range or above
(Kjelgaard & Tager‐Flusberg, 2001). Therefore, we carefully
characterize the language abilities of our sample and include children with ASD
who have a range of language abilities (see Participants). Research questions
focusing on the language skills of the current sample are addressed in a
separate manuscript (Gonzalez‐Barrero & Nadig, 2017a). Second, in
children who are bilingually exposed, language skills are also heterogeneous.
Bilinguals’ language skills vary along a continuum related to the amount of
language exposure received in each of their languages, with optimal mastery of
two languages occurring with balanced exposure, that is, between 40% and 60%
exposure to each language (Elin Thordardottir, 2011). Given that the starting
point for the bilingual advantage hypothesis is living as a proficient
bilingual, we ensured this in our sample by employing a very stringent
definition of bilingualism as described in Method under “Bilingual Status”.
A Bilingual Advantage in ASD?
As reviewed earlier, set‐shifting is a domain of executive functioning
that has been consistently reported to be impaired in individuals with ASD
(e.g., Gioia et al., 2002; Granader et al., 2014; Hill, 2004; Leung &
Zakzanis, 2014). Does bilingualism hold benefits for this specific EF in
children with ASD, as has been reported for typically‐developing children
(e.g., Bialystok & Viswanathan, 2009; Carlson & Meltzoff, 2008;
Poulin‐Dubois, Blaye, Coutya, & Bialystok, 2011)? That is, can being
bilingual mitigate the set‐shifting impairment observed in children with ASD?
This was our novel research question, for which we posed the following
hypotheses: (a) all children with ASD will exhibit impaired performance
relative to typically‐developing children on measures of set‐shifting; (b)
however, bilingual children with ASD will perform better on set‐shifting tasks
relative to monolingual children with ASD, who do not need to manage two
linguistic systems; (c) bilinguals and monolinguals with ASD will not differ on
tasks that assess verbal working memory, included as a control EF not expected
to be affected by bilingualism. We evaluated these hypotheses using rigorous
group matching, a stringent, multisource definition of bilingualism, and
comprehensive evaluation of a potential bilingual advantage. Specifically, we
measured set‐shifting with an experimental task alongside parent report of
children's EF in daily life to provide a more ecologically valid view of EF
skills, as called for in the literature (Kenworthy et al., 2008).
Method
Participants
Forty children with a chronological age range of 6–9 years participated
in the present study (chronological age M = 8.0 years, SD = 9 months). Data
were collected between February 2014 and April 2016. There were 20
typically‐developing children (henceforth TYP, 10 monolinguals, 10 bilinguals)
and 20 children with ASD (10 monolinguals, 10 bilinguals). Exclusion criteria
for the TYP group were parental report of history of language, learning, or
developmental difficulties; physical, visual, or hearing limitations; or any
family members who had been diagnosed with ASD. All children in the TYP group
attended regular schools.
Participants with ASD, who did not have intellectual disability, were
recruited from autism organizations, schools, therapy programs, and a database
from previous studies. They all had a formal clinical diagnosis obtained from
licensed clinicians or multidisciplinary groups from health care institutions,
and parents were asked to provide a copy of the diagnostic report to confirm
the participant's status. In addition, the Social Communication Questionnaire
(SCQ; Rutter, Bailey, & Lord, 2003) was used to confirm ASD symptoms.
This questionnaire consists of 40 questions which provide information about the
child's social functioning and early communication abilities. A score of 15 or
higher on the SCQ is consistent with a diagnosis of ASD. A one‐way analysis of
variance (ANOVA) revealed a statistically significant difference on SCQ scores
between groups, F(3, 35) = 45.50, p < .001. Post hoc tests showed that
as expected, participants with ASD exhibited higher scores than TYP children (p
< .001). However, there was no statistically significant difference
between the monolingual and bilingual TYP groups (p = .11) or between the
monolingual and bilingual ASD groups (p = .98).
The study was conducted in Montreal, Canada, a multicultural city where
French is the official language though many people also speak English. A
quarter of participants had at least one immigrant parent (TYP: 0 monolinguals,
3 bilinguals; ASD: 3 monolinguals, 4 bilinguals). Children were speakers of
English, French, or Spanish or a combination of any of these languages.
Bilingual speakers of other languages were not included in order to control for
language typology and because of the availability of the same standardized measures
in these three languages. Some participants in the bilingual groups (two in the
TYP bilingual and three in the ASD bilingual group) had minimal exposure (i.e.,
< 10%) to a third language over their lifetime, but were not proficient
in this language. The participants’ language of instruction was either French
or English. However, given that French is the official language in the province
of Quebec, it is common that children attending English schools receive some
amount of French language instruction.
Details of participant characteristics and demographic information are
presented in Table 1. To ensure matching on key variables, a series of one‐way
ANOVA were conducted confirming no significant difference between groups on the
following four characteristics: NVIQ (NVIQ > 80), chronological age (6–9
years), maternal education as a proxy for socioeconomic status (mothers from
all groups had attained at least a high school degree), and gender (the
majority of children were boys, given the higher prevalence of ASD in males,
Baio, 2012). NVIQ was assessed using the Leiter–R (Roid & Miller,
1997), which is a completely nonverbal test appropriate for the assessment of
children from different language backgrounds as well as for children with ASD.
Note
TYP = typically‐developing; ASD = autism spectrum disorders; NVIQ =
nonverbal IQ composite score from the Leiter–R; Receptive vocabulary = standard
score from the Peabody Picture Vocabulary–4 or its French version; SCQ = Social
Communication Questionnaire total score.
aSignificant difference between monolingual TYP and bilingual TYP.
bSignificant difference between monolingual TYP and monolingual ASD.
cSignificant difference between monolingual TYP and bilingual ASD. dSignificant
difference between bilingual TYP and monolingual ASD. eSignificant difference
between bilingual TYP and bilingual ASD.
Two standardized tests were used to assess participants’ language
skills. Receptive vocabulary was measured with the Peabody Picture Vocabulary
Test, 4th ed. in English (Dunn & Dunn, 2007), French (Échelle de
vocabulaire en images Peabody; Dunn, Theriault‐Whelan, & Dunn, 1993),
and Spanish (Test de Vocabulario en Imagenes Peabody; Dunn, Padilla, Lugo,
& Dunn, 1986). Vocabulary measures were included given the influence of
vocabulary on EF (e.g., Bialystok, Craik, & Luk, 2008; Buac, Gross,
& Kaushanskaya, 2016). The recalling sentences subtest from the
Clinical Evaluation of Language Fundamentals, 4th ed. (CELF–IV; Semel, Wiig,
& Secord, 2003), along with its French (Évaluation clinique des notions
langagières fondamentales; Wiig, Secord, Semel, Boulianne, & Labelle,
2009), and Spanish versions (CELF–IV Spanish Edition; Semel, Wiig, &
Secord, 2006) were used to determine if participants had language impairment.
Sentence repetition tasks such as this one are considered a useful clinical
marker for language impairment and have been used in studies examining
structural language disorder in ASD (Riches, Loucas, Baird, Charman, &
Simonoff, 2010). Four of 10 participants in the monolingual ASD group and 3 of
10 participants in the bilingual ASD group met criteria for language impairment
as indicated by scores 1 SD below the mean on the recalling sentences subtest.
This cutoff provides adequate sensitivity and specificity for the
identification of language impairment (Conti‐Ramsden, Botting, &
Faragher, 2001; Elin Thordardottir et al., 2011). Participants with ASD and
concomitant language impairment were included to reflect the heterogeneity of
language abilities in this population (Kjelgaard & Tager‐Flusberg,
2001). All children in the TYP groups had scores within the average range on
the recalling sentences subtest.
Bilingual Status
A comprehensive language background questionnaire (Gonzalez‐Barrero
& Nadig, 2017a) including participants’ history of language exposure
over their lifetime, current amount of language exposure, and proficiency in
all languages of exposure, was administered to parents to aid in the
determination of participants’ bilingual status. However, percent of lifetime
exposure to a second language on its own did not accurately reflect fluent
bilingualism, especially for the ASD participants. Given this, determination of
bilingual status was based on a combination of four indices: (a) > 20%
of lifetime exposure to each language (i.e., dominant and nondominant language)
according to parent report, (b) the ability to complete at least 5 of the 8
language‐based tasks of the testing protocol in both languages, (c) a score of
3 or 4 on a 4‐point scale of language proficiency in each of the bilingual
child's languages, as rated by parents, and (d) mean ratings of 2 or above on a
4‐point scale of language proficiency according to the assessment of three
external raters. These raters were blind to bilingual status and based their
ratings on observation of video records of the testing sessions. The 20%
minimum exposure cutoff to determine bilingual status was based on evidence
from Pearson, Fernandez, Lewedeg, and Oller (1997), who found that it may be
difficult to elicit utterances in a language for children with language
exposure lower than 20%. Concerning age of first exposure to each language,
there were seven simultaneous bilingual children in the group with ASD (i.e., children
who have been exposed to both languages before 3 years of age; Paradis,
Genesee, & Crago, 2011) and three early sequential bilingual children
(mean age of first exposure to L2 = 4.3 years), whereas in the TYP bilingual
group there were eight simultaneous bilinguals and two early sequential
bilinguals (mean age of first exposure to L2 = 3 years).
Monolingual children were defined as those who (a) had not been exposed
to a language other than English (or French for the French‐L1 participants) for
more than 20% of their lifetime, (b) if exposed to an L2, had scores of 1 or 2
on a 4‐point scale of language proficiency completed by parents, or (c) could
not complete the testing protocol in both languages, even if they did not meet
criteria 1 and 2. The vast majority of monolingual children who had some
language exposure to an L2 over their lifetime (nine in the TYP group and eight
in the group with ASD) did not have an amount of exposure > 20%. One
child in the monolingual ASD group had a history of exposure to a second
language of 22%, however, his parent proficiency rating in the L2 was low (1 of
4) and this child could not maintain a simple conversation with a research
assistant in the second language. Therefore, he was kept in the monolingual group.
Procedure
Ethics approval was obtained from a university institutional review
board. Parental consent and children's assent was also obtained. This project
was part of a larger study examining the effects of bilingualism on cognition
and language in children with ASD. Participants were tested individually by
trained research assistants in a quiet room at a university laboratory or at
the participant's home. Testing consisted of one 2‐hr session for monolingual
participants, and two sessions for bilingual participants (because language
tests were administered in separate sessions for each language). The protocol
of language tests, described in detail in Gonzalez‐Barrero and Nadig (2017a),
included subtests from the CELF–IV, a nonword repetition task, a picture
description task, and a free‐play language sample. Sessions for the bilingual
participants took place approximately 2 weeks apart, and one language (i.e.,
English, French, or Spanish) was used exclusively during each session. Breaks
and snacks were provided as needed and participants received a small gift upon
completion of each session.
EF measures (BRIEF, DCCS, and verbal working memory) were administered
during the first testing session. Experimental EF tasks (DCCS task and verbal
working memory) were administered in the participant's dominant language.
During the first testing session, parents completed the BRIEF in their
preferred language (English, French, or Spanish), as well as the SCQ and the
child language background questionnaire.
EF Assessments
Parent ratings of EF in daily life
The BRIEF (parent form; Gioia et al., 2000a) was administered to
parents. The BRIEF can be used to assess EF in daily life for TYP children as
well as children with neurodevelopmental disorders, aged 5–18 years (Gioia et
al., 2000a). Its reliability (e.g., internal consistency = 0.80–0.98) and
validity are considered adequate (Gioia, Isquith, Guy, & Kenworthy,
2000b). This questionnaire has been used in previous research examining EF in
children with ASD (e.g., Leung, Vogan, Powell, Anagnostou, & Taylor,
2016) and is considered an ecologically valid measure of executive functioning
(Kenworthy et al., 2008). The test consists of 86 questions that assess eight
clinical dimensions: inhibit, shift, emotional control, initiate, working
memory, plan/organize, organization of materials, and monitor. These subscales
can be grouped into two indexes (i.e., Behavioral Regulation Index and
Metacognition Index), which, when combined, comprise an overall EF score (i.e.,
Global Executive Composite). T‐Scores are calculated and provide information
about whether the child exhibits clinically significant executive dysfunction
in daily life (i.e., t‐score > 65). Special attention was paid to the
shift and working memory subscales given our research questions.
Performance measure of
set‐shifting
Set‐shifting was assessed using a computerized version of the DCCS task
(Zelazo, 2006) developed with E‐Prime 2.0 software (Psychology Software Tools,
2012). The task was presented on a Toshiba Satellite A350 laptop with a 16‐in.
screen. The computerized testing modality was used to decrease the social
demands of face‐to‐face interactions for children with ASD, as suggested by
prior research on set‐shifting in this population (e.g., Ozonoff, 1995;
Pascualvaca et al., 1998). Furthermore, the use of a computerized task allowed
us to gather precise RT.
Our setup of the computerized task followed Bialystok and Martin (2004)
and Barac and Bialystok (2012). A sticker with a red boat was placed on the
right side of the laptop keyboard (“p” key) and a sticker with a blue rabbit
was placed on the left side (“w” key). A black cover on the keyboard was used
to hide all other keys from view. Stimuli presented on the screen consisted of
red rabbits and blue boats, thus the stimuli to be sorted were never identical
to the indicators on the keyboard. Participants sat in front of the laptop, the
experimenter sat next to it.
Our computerized DCCS task included seven phases, adhering to the
administration guidelines of Zelazo (2006). In the first phase (i.e.,
demonstration phase), the experimenter introduced the task and explained the
rules to the participant. In the second phase (i.e., practice phase, three
trials), the child was asked to perform the task: The stimuli appeared on the
center of the screen and participants were instructed to press the appropriate
key to categorize the image (e.g., “We are going to play the color game, in the
color game all the blue ones go here—pointing to the key with the blue
rabbit—and all the red ones go here—pointing to the key with the red boat. Here
is a blue one, where does it go?”). The image remained on the screen until the
participant pressed a key to respond. An intertrial interval of 2,000 ms
preceded the next trial. During the practice phase, visual feedback on the
screen (e.g., a cartoon of a checkmark signaling “good work”) and verbal
feedback from the experimenter were provided after each trial. To continue to
the pre‐switch phase, the child was required to succeed on at least two of
three practice trials. In the third phase (i.e., pre‐switch phase, six trials),
the experimenter reminded the child about the rule (e.g., sort by color) to
reduce memory demands, and the participant was again asked to sort images
following this rule. No feedback was provided during the experimental trials.
In the fourth phase (i.e., post‐switch phase, six trials), the child was asked
to change the sorting rule and to now classify the images by another dimension
(e.g., “okay, now we are not going to play the color game anymore; now we are
going to play the shape game …” Zelazo, 2006, p. 297). The order of
presentation of sorting dimension was counterbalanced, with half of children
receiving color first and the other half receiving shape first. Following
Zelazo (2006), participants passed the post‐switch phase if they were able to
correctly sort on five of the six trials.
A more advanced phase was administered to participants who passed the
post‐switch phase. This “border version” (Zelazo, 2006) was demonstrated in a
fifth phase where children were asked to sort by color if the image displayed
on the screen had a border, or by shape if the image had no border,
consequently referred to as “mixed condition” in the literature. A practice
border version with feedback was presented (i.e., sixth phase, two trials). If
the child obtained at least one correct response, he or she was asked to
complete the border version (seventh phase, 12 trials). Children were
considered to pass the border version if they correctly sorted on at least 9 of
the 12 trials.
Accuracy (i.e., passing or failing) on the pre‐switch, post‐switch, and
border version phases of the DCCS task along with RT in milliseconds were
analyzed. Following Diamond and Kirkham (2005), for RT analyses only correct
trials were considered and trials that were < 200 ms or 2.5 SD above the
mean for each group were not included as these are considered to be either too
quick or too slow to reflect processing of the stimuli. For TYP participants,
after removing incorrect trials (17%) and trials that did not meet the response
time criteria just noted (3%), 80% of possible trials remained for inclusion in
RT analyses. For participants with ASD, after removal of incorrect trials
(15%), and trials excluded given response time criteria (3%), 82% were kept for
RT analyses. The mean RT of the last two trials of the pre‐switch phase was
subtracted from the mean RT of the first two trials of the post‐switch phase to
obtain a RT switch cost difference score. This approach to RT analysis follows
that used by Dichter et al. (2010) with a group of monolingual children with
ASD. In addition, mean RT for the border version (i.e., mixed condition) was
examined to investigate if the bilingual advantage reported by Barac and
Bialystok (2012) for TYP bilingual children could be replicated in the present
study.
Performance measure of verbal
working memory
To assess short‐term and working memory, the number repetition subtest
of the CELF–IV (Semel, Wiig, & Secord, 2003), along with its French
(Évaluation clinique des notions langagières fondamentales; Wiig, Secord,
Semel, Boulianne, & Labelle, 2009) or Spanish versions (CELF–IV Spanish
Edition; Semel, Wiig, & Secord, 2006) were used. Administration and
scoring followed the guidelines provided in the CELF–IV manual. In this task
there are eight number lengths (ranging from two to nine digits, with two items
at each length) that the child was asked to repeat immediately after the
examiner (i.e., number repetition forward). In addition, there are seven number
lengths (ranging from 2 to 8 digits, with two items at each length) that the
child was asked to repeat in reverse order (i.e., number repetition backwards).
Number repetition forward is considered a measure of short‐term memory, whereas
number repetition backwards is a more complex measure that taps into both
short‐term and working memory (e.g., Eigsti, 2011; Engel de Abreu, 2011).
Although we were interested in the working memory measure of number repetition
backwards, both subtests were given to maintain standard administration
involving a gradual increase in complexity. Only the scaled score for number
repetition backwards was considered for analysis.
Our research question was whether bilingualism can mitigate
set‐shifting difficulties in children with ASD, via a bilingual advantage. To
recap, we hypothesized that both monolingual and bilingual children with ASD
will exhibit impaired performance relative to typically‐developing children on
measures of set‐shifting. Given prior findings (Leung & Zakzanis,
2014), we expected set‐shifting impairments to be larger on a rating measure of
EF in daily life (BRIEF) than in an experimental task (DCCS task). The critical
hypothesis was that, despite this background of a general set‐shifting
impairment, bilinguals with ASD will perform better on set‐shifting than
monolinguals with ASD. Finally, we hypothesized that bilinguals and monolinguals
with ASD will not differ with respect to the control EF domain of verbal
working memory. We did not have specific hypotheses as to how ratings of EF in
daily life and performance on experimental measures of EF would compare, as
they have not previously been compared directly when investigating a bilingual
advantage.
Results
Data Analysis
To examine the impact of bilingualism on EF, two‐way (2 Diagnostic
Group [TYP, ASD] × 2 Language Status [monolingual, bilingual]) ANOVA were
conducted on continuous variables (i.e., shifting and working memory scores
from the BRIEF, DCCS RTs, and working memory scaled score from the CELF). For
categorical variables (i.e., DCCS task accuracy/passing), Fisher's exact tests
were performed. In addition, correlation analyses were conducted to explore the
relation between scores based on parent report and direct testing of EF
measures. The assumptions of statistical tests were checked and parametric
tests were used where appropriate. In cases where the assumption of normality
was not met, Kendall's tau was used for correlations. Following a conservative
approach, Bonferroni corrections were used for measures where multiple tests
were performed (e.g., DCCS task and BRIEF) to avoid Type I errors. Therefore,
the applicable alpha level for significance is presented for each measure.
Means and standard deviations on all EF measures are provided in Table
2. To assess the validity of the responses provided by parents on measures of
EF in daily life, we used the inconsistency scale from the BRIEF. This scale
compares similar items to which parents are expected to provide consistent
answers. A score derived from this scale indicates whether the questionnaire
can be considered acceptable, questionable, or inconsistent. Results showed
that for the majority of children, scores were consistent. However, two
participants in the monolingual ASD group showed questionable scores and were
therefore excluded from analyses involving the BRIEF. Two statistical analyses
were conducted using the BRIEF scores (i.e., shift subscale and working memory
subscale); therefore, the alpha level for significance was set at p <
.025.
TYP = typically‐developing children; ASD = autism spectrum disorders;
BRIEF = Behavior Rating Inventory of Executive Functioning; DCCS = dimensional
change card sort task.
aMain effect of diagnostic group, bMain effect of language,
cSignificant difference between monolingual TYP and bilingual TYP, dSignificant
difference between bilingual TYP and bilingual ASD.
Set‐Shifting (BRIEF)
A 2 (diagnostic group) × 2 (language status) ANOVA on shift subscale
scores, obtained via parent report on the BRIEF, revealed only a significant
main effect of diagnostic group, F(1, 34) = 43.12, p = < .001, partial
η2 = .56, where TYP participants obtained lower scores than participants with ASD.
As previously described, higher scores on the BRIEF (i.e., t‐scores >
65) reflect a greater degree of difficulty in EF (Gioia et al., 2000a).
Conversely, there was no significant main effect of language status, F(1, 34) =
.08, p = .78, partial η2 = .00, nor was the interaction between diagnostic
group and language status significant, F(1, 34) = .88, p = .36, partial η2 =
.03.
Set‐Shifting (DCCS Task
Accuracy/Passing)
Results showed that children in all groups passed the pre‐switch phase,
and the majority of participants in each group passed the post‐switch phase
(passing criteria: 5 correct trials out of 6). Consequently, following Diamond
and Kirkham (2005), statistical analyses were not conducted on accuracy data in
these phases (available in Table 2) given the minimal variation observed. The
percentage of participants passing the pre‐switch, post‐switch, and border
versions of the DCCS task using the criteria of Zelazo (2006) are presented in
Figure 1. For the border version (passing criteria: 9 correct trials out of
12), a Fisher's exact test showed a significant difference across the four
groups (p = .024). To further examine this difference, pairwise Fisher's exact
tests were performed. Results revealed that, consistent with our second
hypothesis, the number of bilinguals with ASD who passed the border version was
significantly higher than the number of monolinguals with ASD who passed (p =
.026). In fact, the bilingual ASD group also outperformed the TYP bilingual
group on the border version of the DCCS task (p = .009). On the other hand, the
TYP groups did not differ significantly by language status (p = .347). The
monolingual TYP and monolingual ASD groups also did not differ significantly (p
= .619).
Figure 1 Percentage of participants passing each phase of the
dimensional change card sort task by group.
Open in figure viewerPowerPoint
Percentage of participants passing each phase of the dimensional change
card sort task by group.
Set‐Shifting (DCCS RTs)
The significance level for analyses concerning RTs was set at p
< .025, given two statistical analyses conducted, one on switch cost
between pre‐ and post‐switch phases and the other on mean RT of the border
phase. A two‐way ANOVA was performed on RT switch cost (i.e., mean RT of first
two trials from post‐switch phase minus mean RT of last two trials from
pre‐switch phase). The main effect of diagnostic group was not significant,
F(1, 33) = 2.97, p = .09, partial η2 = .08, although a trend was observed in
which the TYP groups were faster than the groups with ASD. The main effect of
language status, F(1, 33) = .46, p = .50, partial η2 = .01, was not
statistically significant, contrary to previous studies with TYP children
(e.g., Barac & Bialystok, 2012). Finally, the interaction between
diagnostic group and language status was not significant, F(1, 33) = .46, p =
.50, partial η2 = .01.
For the border version, mean RTs were analyzed using a two‐way ANOVA.
The main effect of diagnostic group was not significant, F(1, 28) = 2.82, p =
.10, partial η2 = .09, and the main effect of language status revealed that,
contrary to previous findings, the monolingual TYP and ASD groups responded
faster than the bilingual TYP and ASD participants in the border version, F(1,
28) = 6.32, p = .018, partial η2 = .18. The interaction between diagnostic
group and language status was not statistically significant, F(1, 28) = .02, p
= .89, partial η2 = .00 (see Figure 2).
Figure 2 Mean reaction time for border version of the dimensional
change card sort task by group.
Open in figure viewerPowerPoint
Mean reaction time for border
version of the dimensional change card sort task by group.
Verbal Working Memory (BRIEF)
A two‐way ANOVA conducted to examine working memory in everyday life,
as assessed by parent report on the corresponding BRIEF subscale, revealed a
significant main effect only for diagnostic group, F(1, 34) = 12.63, p = .00,
partial η2 = .27, with the ASD group having higher scores, reflecting poorer
performance, relative to the TYP groups. The main effect of language status,
F(1, 34) = .005, p = .95, partial η2 = .00, and the interaction, F(1, 34) =
1.39, p = .25, partial η2 = .04, were not statistically significant.
Verbal Working Memory (CELF–IV)
Participants’ performance on the experimental verbal working memory
measure (i.e., scaled score from the number repetition backwards subtest of the
CELF–IV) was examined. In line with our third hypothesis, results revealed that
there was no significant main effect for diagnostic group, F(1, 36) = 0.40, p =
.53, partial η2 = .01, or for language status, F(1, 36) = 0.96, p = .34,
partial η2 = .03, nor a significant interaction between diagnostic group and
language status for this variable, F(1, 36) = 0.003, p = .95, partial η2 = .00.
Correlations Between BRIEF
Scores and Experimental Tasks
Finally, Kendall's tau‐b correlations were run to examine the relation
between parent report of EF behaviors in daily life as measured by the BRIEF
and experimental tasks of EF. For the set‐shifting variables (i.e., BRIEF
shifting subscale and DCCS task switch cost), there was not a statistically
significant correlation for the TYP group, τb = .122, p = .505, or for the
group with ASD, τb = −.226, p = .252. Similarly, results for the working memory
measures (i.e., BRIEF working memory subscale and number repetition subtest
from the CELF–IV) revealed no significant correlations for the TYP group, τb =
.108, p = .530, or for the group with ASD, τb = −.226, p = .206.
Discussion
In this study, we tested for the first time whether being proficiently
bilingual can mitigate the set‐shifting impairment observed in children with
ASD. Our first hypothesis was that both monolingual and bilingual children with
ASD would exhibit impaired performance relative to typically‐developing
children on measures of set‐shifting. This was the case for a parent report
measure of set‐shifting (BRIEF), which confirmed clinical levels of difficulty.
However, we did not observe a decrement at the level of diagnostic group on an
experimental task of set‐shifting (the DCCS task), given the high rate of
passing by bilingual children with ASD. Our second hypothesis proposed that if
the bilingual advantage extends to ASD, bilingual children with ASD should
perform better on set‐shifting tasks relative to monolingual children with ASD,
who do not need to coordinate two linguistic systems. Our results from the DCCS
task support this hypothesis: A significantly higher percentage of bilingual
children than monolingual children with ASD passed the border version of the
DCCS task. However, no bilingual advantage was observed in parent report of
set‐shifting behaviors in daily life on the BRIEF. Our third hypothesis was that
bilingual children with ASD would not differ from their monolingual peers with
ASD on tasks that assess verbal working memory, included as a control EF not
expected to be affected by bilingualism. This hypothesis was supported by our
results on both a performance task (number repetition backwards) and parent
report of working memory in daily life. We now consider each of these findings
in turn.
Set‐Shifting in Daily Life
Regardless of language status, and as anticipated, children with ASD
exhibited set‐shifting difficulties in everyday life relative to their same‐age
TYP peers, as shown by results from the BRIEF. Both groups with ASD scored
within the dysfunction range on this subscale, whereas the TYP participants’
scores were in the average (typical) range. This confirmation of set‐shifting
difficulties in daily life on a rating measure is consistent with a substantial
body of work indicating that set‐shifting is a “peak” area of EF impairment in
ASD (Gioia et al., 2002; Granader et al., 2014), and that rating measures are
the most sensitive to detect them (Leung & Zakzanis, 2014; Toplak,
West, & Stanovich, 2013). By establishing set‐shifting impairment, this
first finding sets the backdrop for testing whether bilingualism may mitigate
these difficulties in ASD. At the same time, the similar performance of
bilingual and monolingual children with ASD on the BRIEF was contrary to our
second hypothesis of a bilingual advantage in set‐shifting skills. We did
however observe a bilingual advantage for children with ASD on the DCCS task
discussed next. We return to the (lack of) relation between rating and
performance or experimental task measures later in the Discussion.
Set‐Shifting in an Experimental
Task
The key finding of this study was a bilingual advantage on the advanced
border version of the DCCS task, where bilingual children with ASD passed at a
significantly higher rate than monolingual children with ASD, upholding our
second hypothesis. The fact that we detected this difference on ability to complete
the border, rather than simpler post‐switch phase is likely due to the age of
our participants. Notably, this is also the phase of the DCCS task where
significant effects were reported by Faja and Dawson (2014), who tested
monolingual children with ASD of similar ages. Although novel and revealing,
this finding should be taken as preliminary evidence given our small sample
size. Replication with a larger but similarly well‐characterized sample to our
own (i.e., stringent confirmation of bilingual status, matching on NVIQ and
maternal education as well as age, description of language abilities, etc.) is
needed to confirm a bilingual advantage in children with ASD.
Two other aspects of our DCCS task results, however, ran counter to our
predictions. Following our first hypothesis, we expected typically‐developing
children to outperform both groups of children with ASD. Yet, in our sample,
the DCCS task did not capture the impairment in set‐shifting reported by
previous studies using the same task with monolingual children with ASD (Faja
& Dawson, 2014; Yi et al., 2012). This is likely because the DCCS task
is not a highly sensitive measure of set‐shifting impairment, with a 95%
confidence interval of [−1.82, 2.85] pooled over the two studies just mentioned
(Leung & Zakzanis, 2014). That is, the DCCS task does not reliably show
better performance for typically‐developing children relative to children with
ASD. Other studies have also reported spared set‐shifting abilities in children
and adults with ASD on some experimental tasks (e.g., Edgin &
Pennington, 2005; Hill & Bird, 2006). Second, while we detected a
bilingual advantage in children with ASD, this was not the case for
typically‐developing children.
Why might the bilingual advantage be observable only in children with
ASD and not in typically‐developing children in our study? One possibility is
related to testing modality (e.g., the computer presentation employed here, vs.
live administration by an experimenter). As discussed in the Introduction,
children with ASD have been found to exhibit better performance on EF tasks
that employ computerized versus live administration by an examiner (Kenworthy
et al., 2008; Ozonoff, 1995; Pascualvaca et al., 1998), which has been
attributed to the reduced social demands required of them. Typically‐developing
children, on the other hand, do not perform better on computerized tasks and
may actually benefit from social motivation when interacting with an
experimenter (Kenworthy et al., 2008). In fact, during the administration of
the computerized DCCS task in our study, it was observed that some TYP children
(five monolinguals, two bilinguals) looked at the experimenter as if awaiting
confirmation on their performance during the border version (although no
feedback was provided on experimental trials). Although this was occasionally
observed, this behavior was less prominent among the ASD groups (one
monolingual, two bilinguals). In addition, children in both ASD groups seemed
more motivated and engaged with the computerized task than did their TYP peers;
some of them verbally expressed high levels of enjoyment and even wanted to
continue “playing the computer game.”
A second possible reason for the enhanced performance of the bilingual
ASD group relative to the bilingual typically‐developing group is the degree of
explicitness of the task. It has been reported that individuals with ASD
perform better on tasks that provide clear and explicit rules (e.g., DCCS task)
relative to tasks where rules are implicit and need to be inferred (e.g.,
Wisconsin Card Sorting Task), (Van Eylen et al., 2011). Since sorting criteria
for the DCCS task were explicitly stated, this may explain the relatively high
performance of the ASD groups in the pre‐switch and post‐switch phases. Yet, it
was on the border version, where the presence or absence of the visual cue of a
border indicated the rule to use, that bilinguals with ASD outperformed monolinguals
with ASD as well as typically‐developing bilinguals. It is possible that this
visual cue was perceived as “explicit” and easier for bilingual (but not
monolingual) children with ASD to use given their bilingual advantage, but was
not perceived as an explicit cue by typically‐developing children.
A third contributor to the lack of a bilingual advantage for
typically‐developing children in the current study may be related to the
participants included in the monolingual typically‐developing group. All of
these children spoke French, whereas the other three groups of participants
included both French and English (dominant) speakers. This difference was
unavoidable because the study was conducted in Montreal, where it is possible
to recruit monolingual French‐speaking children (as defined in the Method). It
was not, however, possible to find typically‐developing school‐age children who
were English monolinguals without a significant amount of exposure to French,
as children in regular English language schools receive French instruction and
hear French in multiple social contexts. English and French school boards are
separate in Quebec and may provide different background experiences.
Consequently, the comparison between monolingual and bilingual typically‐developing
children was not as well matched in this respect as it was for children with
ASD.
Finally, a fourth reason why a bilingual advantage may not have been
observed for typically‐developing 6‐ to 9‐year‐olds in our study is that the
task may no longer have been developmentally sensitive for them. Bialystok
(1999) used the DCCS task (live card administration, original procedure) and
found a bilingual advantage for typically‐developing 4‐ and 5‐year‐olds. Barac
and Bialystok (2012) used a computerized color–shape switching task with
typically‐developing 5‐ to 7‐year‐olds, also documenting a bilingual advantage
in switch cost. As our primary research questions focused on children with ASD,
our TYP sample was older, in line with prior studies on children with ASD, and
we employed a very simple task of set‐shifting as a starting point to observe
potential differences related to bilingualism.
DCCS Response Time Measures
In research on typically‐developing children and adults, the bilingual
advantage in set‐shifting is most commonly found in RT measures (e.g., Prior
& Macwhinney, 2010; but see Morton & Harper, 2007). The general
finding is that bilinguals exhibit smaller switching costs (e.g., delays when
changing from one rule to another, or on switch blocks compared to same rule
blocks) relative to monolinguals during set‐shifting paradigms (Prior &
Macwhinney,2010). It is important to note that our examination of RT measures
was exploratory given that we employed the original procedure of the DCCS task
although it was computer administered, and RT measures are not involved in the
conventional scoring of this task. We presented a maximum of 24 trials compared
to 200 trials over five blocks (with either the same rule or a switch in rule)
that are used in many card sort tasks focusing on RT measures. Keeping that in
mind, we did not find a bilingual advantage for either the ASD or TYP groups
with respect to switch cost between pre‐ and post‐switch phases of the task,
which has been reported previously for typically‐developing children in other
types of card sort tasks (e.g., Bialystok & Viswanathan, 2009; Prior
& Macwhinney, 2010). There was a trend for children with ASD to present
increased switch cost times relative to children with typical development,
which is in line with previous studies reporting overall slower RTs in
monolingual children with ASD relative to typically‐developing controls on
computerized card sort tasks (e.g., Dichter et al., 2010).
Similarly, for mean RT in the border version (a phase where the rule to
be used was mixed), no bilingual advantage was found, contrary to prior reports
for typically‐developing children using color–shape card switch tasks (Barac
& Bialystok, 2012). In fact, the reverse was found: In our study, both
groups of monolinguals tended to respond more quickly on border version trials
than did bilinguals with typical development or ASD. Once again, this finding
should be interpreted with caution since the task procedure was not comparable
to other studies analyzing RTs. However, at least for the bilingual ASD group,
who were significantly better at passing this phase, this finding appears to
represent a speed/accuracy trade‐off.
Working Memory in Daily Life and
on an Experimental Task
Results from the BRIEF working memory subscale revealed that children
with ASD exhibited difficulties in this domain in their everyday life, as
reflected in somewhat elevated scores. However, while set‐shifting scores of
children with ASD were in the executive dysfunction range, this was not the
case for working memory scores, replicating the finding that set‐shifting is a
peak area of difficulty in ASD (Gioia et al., 2002; Granader et al., 2014).
Finally, our third hypothesis that working memory should not be affected by
being proficiently bilingual was upheld, both on the BRIEF and on our
performance measure of verbal working memory. Scores on the number repetition
backward subtest of the CELF–IV showed comparable performance between all four
groups of participants. This is also in line with previous work showing verbal
working memory to be a spared aspect of EF in children with ASD (Faja &
Dawson, 2014).
Relation Between Experimental
Tasks and Skills in Daily Life
The dissociation between impairments in EF skills in everyday life and
intact performance on the experimental EF tasks has been discussed by autism
researchers (e.g., Kenworthy et al., 2008). These researchers have advocated
for more comprehensive and ecologically valid approaches to the study of cognitive
flexibility in ASD. Our study was novel in employing both experimental or
performance measures and a rating measure of EF skills in daily life to
investigate a bilingual advantage. We found weak, nonsignificant correlations
between switch cost on the DCCS task and the shift subscale of the BRIEF (TYP,
τb = .122, ASD, τb = −.226). It is important to point out that the dissociation
between performance‐based and rating measures of EF is not specific to ASD and
is widely documented (Isquith, Roth, & Gioia, 2013), serving as the
subject of a practitioner review by Toplak et al. (2013). These authors
reviewed 20 studies that compared both performance‐based and rating measures of
EF, including child and adult, nonclinical and clinical samples, particularly
attention deficit hyperactivity disorder and traumatic brain injury. Across
these studies, they report a median correlation of only .19 between
performance‐based and rating measures of EF, with only 24% of the correlations
reported in these studies reaching significance (Toplak et al., 2013). The lack
of a strong relation between these measures across a range of populations
indicates that they in fact measure different constructs. Performance‐based
measures provide optimal conditions and structure the problem space with a
well‐defined task and often an explicit rule to follow. In contrast, the use of
analogous EF abilities in daily life requires applying them as the need arises
to meet ones’ goals. Toplak et al. (2013, p. 140) deem these complementary measures
given that each provide important information about EF abilities, with
performance‐based measures serving as an index of efficiency of processing, and
rating measures reflecting success in rational goal pursuit in natural
environments.
How can we bridge the divide between the significant impairment in EF
abilities reported in daily life, and the good or even enhanced (in the case of
bilingual children with ASD) performance on experimental tasks? It may be
possible to use tasks where children with ASD present spared performance to
improve EF abilities in daily life. For instance, if children with ASD are able
to switch between sets using a computer‐based task, this ability could be built
on through intervention. One could gradually shift from computer‐based
cognitive flexibility paradigms to the integration of EF skills in daily life,
for example, by adding one additional dimension of complexity at a time.
Contributions and Limitations of
the Current Study
The present study is the first to examine the effects of bilingualism
on the set‐shifting and working memory skills of proficient bilingual children
with ASD. To ascertain children's bilingual status, this study employed a
rigorous four‐step procedure including parent report of amount of language exposure,
parent ratings of language proficiency, direct testing in both languages of the
bilingual participants, and ratings of language proficiency by external judges.
In addition, all four groups of participants were matched on NVIQ, age, and
level of maternal education, and to the extent possible, dominant language.
Participants were from the same city and for the most part shared similar
cultural and educational backgrounds. Importantly, both experimental and rating
measures of set‐shifting and working memory in daily life were gathered.
Taken together, these methodological strengths allow us to interpret
results as due to bilingual exposure rather than confounding factors. However,
this methodological rigor, as well as the difficulties inherent in recruiting
special populations, limited the sample size in each group. It is possible that
with only 10 children per subgroup we did not have enough power to detect
significant differences between monolingual and bilingual typically‐developing
children, yet as discussed earlier, we did have enough power to detect
differences among children with ASD. Potential differences in the strength of a
bilingual advantage across typically‐developing children and those with
developmental disorders should be explored in future research.
Generalizability of Bilingual
Advantage in EF for Children With ASD
We present the first evidence that a bilingual advantage (Bialystok
& Martin, 2004; Bialystok & Viswanathan, 2009) can be observed
in children with ASD. It is important to consider to whom and where this effect
may generalize. Our participants with ASD were school age and did not have
intellectual delay, with 7/20 meeting research criteria for language
impairment. Crucially, to have bilingual status in our study participants had
to be proficiently bilingual as described earlier, and they lived in a society
where the two languages in question were official languages used in multiple
contexts.
The effect of bilingualism on EF in ASD was selective: there was an
advantage on an experimental task of set‐shifting but not for verbal working
memory. The selective influence of bilingualism on specific cognitive skills
had been discussed and this same pattern has been found in typically‐developing
individuals (e.g., Bialystok, 2009).
We detected a bilingual advantage in set‐shifting in children with ASD
using a relatively brief and simple computerized task that reduced social
demands and provided explicit rules. This task was also untimed, that is,
participants could respond when ready. There was a trend for children with ASD
to demonstrate longer switch costs. The bilingual children with ASD, who were
the most accurate in the advanced border phase of the DCCS task, also tended to
be slower on all response time measures. This longer response time raises the
possibility that an ASD bilingual advantage may not be observed in tasks or
situations involving time pressure.
We also presented evidence that the bilingual advantage in set‐shifting
for children with ASD does not extend to ratings of set‐shifting in daily life.
As discussed earlier, this reflects a more general phenomena where performance
on experimental EF tasks is minimally related to the application of EF in daily
life (Toplak et al., 2013). It is possible that the cognitive processes engaged
and enhanced by the proficient use of two languages are not related to the
broader EF skills tapped by rating measures. Alternatively, any potential
bilingual advantage effects in ratings of EF in daily life may only be observed
at a much later age, with accumulated bilingual experience. This is another
important question for future research.
In a separate study with a largely overlapping sample (Gonzalez‐Barrero
& Nadig, 2017b), we report a similar bilingual advantage on a verbal
fluency task (e.g., “name all of the animals you can think of”), which assesses
both lexical skills and multiple EF skills. Thus, it appears this bilingual
advantage in ASD is not limited to the nonverbal domain or to the particular
task we employed in this study. On the verbal fluency task, bilinguals
outperformed monolinguals with ASD in the total number of words produced, but
not on switching between semantic clusters of words. Consequently, we suggested
that the bilingual advantage observed on the verbal fluency task reflects
generativity and initiation, rather than set‐shifting. Further work is clearly
warranted to explore the extent and reliability of bilingual advantages for
children with ASD.
The present findings extend a bilingual advantage in set‐shifting
skills, as measured by an experimental task, to children with ASD. We
demonstrate that, contrary to common belief, bilingualism is not harmful for
children with ASD (e.g., with respect to their language abilities, Kay‐Raining
Bird, Genesee, & Verhoeven, 2016), and in fact, may provide some
advantages, such as mitigating prominent set‐shifting difficulties. There is
growing evidence that bilingualism acts as a protective factor that delays the
onset of dementia in normal aging (for a review, see Bialystok, Abutalebi, Bak,
Burke, & Kroll, 2016). Likewise, bilingualism, under the right
conditions, may act as a protective factor for certain EF difficulties in
populations with neurodevelopmental disorders. If replicated, this finding
could provide critical evidence to inform educational decisions taken by the
increasing number of families with children with ASD for whom the use of two or
more languages is a valued practice (Kay‐Raining Bird et al., 2016).
References
American Psychiatric Association. (2013). Diagnostic and statistical
manual of mental disorders (5th ed.). Washington, DC: Author.
Baddeley, A. (1992). Working memory. Science, 255, 556–559.
https://doi.org/10.1126/science.1736359
Baio, J. (2012). Prevalence of autism spectrum disorders: Autism and
developmental disabilities monitoring network, 14 Sites, United States, 2008.
Morbidity and Mortality Weekly Report, 61(3), 1–19.
Barac, R., & Bialystok, E. (2012). Bilingual effects on
cognitive and linguistic development: Role of language, cultural background,
and education. Child Development, 83, 413–422.
https://doi.org/10.1111/j.1467-8624.2011.01707.x
Bennetto, L., Pennington, B. F., & Rogers, S. J. (1996). Intact
and impaired memory functions in autism. Child Development, 67, 1816–1835.
https://doi.org/10.1111/j.1467-8624.1996.tb01830.x
Bialystok, E. (1999). Cognitive complexity and attentional control in
the bilingual mind. Child Development, 70, 636–644. https://doi.org/10.1111/1467-8624.00046
Bialystok, E. (2009). Bilingualism: The good, the bad, and the
indifferent. Bilingualism: Language and Cognition, 12, 3–11.
https://doi.org/10.1017/s1366728908003477
Bialystok, E., & Martin, M. M. (2004). Attention and inhibition
in bilingual children: Evidence from the dimensional change card sort task.
Developmental Science, 7, 325–339.
https://doi.org/10.1111/j.1467-7687.2004.00351.x
Bialystok, E., Craik, F., & Luk, G. (2008). Cognitive control
and lexical access in younger and older bilinguals. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 34, 859.
https://doi.org/10.1037/0278-7393.34.4.859bhbb
Bialystok, E., Abutalebi, J., Bak, T. H., Burke, D. M., &
Kroll, J. F. (2016). Aging in two languages: Implications for public health.
Ageing Research reviews, 27, 56–60. https://doi.org/10.1016/j.arr.2016.03.003
Bialystok, E., & Viswanathan, M. (2009). Components of
executive control with advantages for bilingual children in two cultures.
Cognition, 112, 494–500. https://doi.org/10.1016/j.cognition.2009.06.014
Buac, M., Gross, M., & Kaushanskaya, M. (2016). Predictors of processing‐based
task performance in bilingual and monolingual children. Journal of
Communication Disorders, 62, 12–29.
https://doi.org/10.1016/j.jcomdis.2016.04.001
Carlson, S. M., & Meltzoff, A. N. (2008). Bilingual experience
and executive functioning in young children. Developmental Science, 11,
282–298. https://doi.org/10.1111/j.1467-7687.2008.00675.x
Cohen, J. (1992). A power primer. Psychological Bulletin, 112, 155–159.
https://doi.org/10.1037/0033-2909.112.1.155
Conti‐Ramsden, G., Botting, N., & Faragher, B. (2001).
Psycholinguistic markers for specific language impairment (SLI). Journal of
Child Psychology and Psychiatry, 42, 741–748.
https://doi.org/10.1111/1469-7610.00770
Delcenserie, A., & Genesee, F. (2016). The effects of age of
acquisition on verbal memory in bilinguals. International Journal of
Bilingualism. https://doi.org/10.1177/1367006916639158
Diamond, A., & Kirkham, N. (2005). Not quite as grown‐up as we
like to think parallels between cognition in childhood and adulthood.
Psychological Science, 16, 291–297.
https://doi.org/10.1111/j.0956-7976.2005.01530.x
Dichter, G. S., Radonovich, K. J., Turner‐Brown, L. M., Lam, K. S. L.,
Holtzclaw, T. N., & Bodfish, J. W. (2010). Performance of children with
autism spectrum disorders on the dimension‐change card sort task. Journal of
Autism and Developmental Disorders, 40, 448–456. https://doi.org/10.1007/s10803-009-0886-1
Dunn, L. M., Padilla, E. R., Lugo, D. E., & Dunn, L. M. (1986).
Test de Vocabulario en Imagenes Peabody. Circle Pines, MN: American Guidance
Services.
Dunn, L. M., & Dunn, L. M. (2007). Peabody Picture Vocabulary
Test, 4th ed.. Bloomington, MN: Pearson Assessment.
Dunn, L., Thériault‐Whalen, C., & Dunn, L. (1993). Échelle de
vocabulaire en images Peabody: Adaptation française du Peabody Picture
Vocabulary Test. Toronto, Ontario, Canada: PsyCan.
Edgin, J. O., & Pennington, B. F. (2005). Spatial cognition in
autism spectrum disorders: Superior, impaired, or just intact? Journal of
Autism and Developmental Disorders, 35, 729–745. https://doi.org/10.1007/s10803-005-0020-y
Eigsti, I. M. (2011). Executive functions in ASD. In D. A. Fein (Ed.),
The neuropsychology of autism (pp. 185–203). New York, NY: Oxford University
Press.
Elin Thordardottir, E. (2011). The relationship between bilingual
exposure and vocabulary development. International Journal of Bilingualism, 15,
426–445. https://doi.org/10.1177/136700691140320
Elin Thordardottir, E., Kehayia, E., Mazer, B., Lessard, N., Majnemer,
A., Sutton, A., & Chilingaryan, G. (2011). Sensitivity and specificity
of French language and processing measures for the identification of primary
language impairment at age 5. Journal of Speech, Language, and Hearing
Research, 54, 580–597. https://doi.org/10.1044/1092-4388(2010/09-0196)
Engel de Abreu, P. M. (2011). Working memory in multilingual children:
Is there a bilingual effect?Memory, 19, 529–537.
https://doi.org/10.1080/09658211.2011.590504
Faja, S., & Dawson, G. (2014). Performance on the dimensional
change card sort and backward digit span by young children with autism without
intellectual disability. Child Neuropsychology, 20, 692–699.
https://doi.org/10.1080/09297049.2013.856395
Frye, D., Zelazo, P. D., & Palfai, T. (1995). Theory of mind
and rule‐based reasoning. Cognitive Development, 10, 483–527.
https://doi.org/10.1016/0885-2014(95)90024-1
Gioia, G. A., Isquith, P. K., Guy, S. C., & Kenworthy, L.
(2000a). Behavior rating inventory of executive function (BRIEF). Lutz, FL:
PAR.
Gioia, G. A., Isquith, P. K., Guy, S. C., & Kenworthy, L.
(2000b). Test review behavior rating inventory of executive function. Child
Neuropsychology, 6, 235–238.
Gioia, G. A., Isquith, P. K., Kenworthy, L., & Barton, R. M.
(2002). Profiles of everyday executive function in acquired and developmental
disorders. Child Neuropsychology, 8, 121–137.
https://doi.org/10.1076/chin.8.2.121.8727
Gonzalez‐Barrero, A. M., & Nadig, A. (2017a). Bilingual
children with autism spectrum disorders: The impact of language exposure on
lexical and morphological skills at school age. Manuscript submitted for
publication.
Gonzalez‐Barrero, A. M., & Nadig, A. (2017b). Verbal fluency in
bilingual children with Autism Spectrum Disorders. Linguistic Approaches to
Bilingualism, 7, 460–475.
Granader, Y., Wallace, G. L., Hardy, K. K., Yerys, B. E., Lawson, R.
A., Rosenthal, M., … Schultz, R. T. (2014). Characterizing the factor structure
of parent reported executive function in autism spectrum disorders: The impact
of cognitive inflexibility. Journal of Autism and Developmental Disorders, 44,
3056–3062.
Green, D. W. (1998). Mental control of the bilingual lexico‐semantic
system. Bilingualism: Language and Cognition, 1, 67–81.
https://doi.org/10.1017/s1366728998000133
Hill, E. (2004). Evaluating the theory of executive dysfunction in
autism. Developmental Review, 24, 189–233.
https://doi.org/10.1016/j.dr.2004.01.001
Hill, E. L., & Bird, C. M. (2006). Executive processes in
Asperger syndrome: Patterns of performance in a multiple case series.
Neuropsychologia, 44, 2822–2835.
https://doi.org/10.1016/j.neuropsychologia.2006.06.007
Isquith, P. K., Roth, R. M., & Gioia, G. (2013). Contribution
of rating scales to the assessment of executive functions. Applied
Neuropsychology: Child, 2, 125–132.
https://doi.org/10.1080/21622965.2013.748389
Kay‐Raining Bird, E., Genesee, F., & Verhoeven, L. (2016).
Bilingualism in children with developmental disorders: A narrative review.
Journal of Communication Disorders, 63, 1–14.
https://doi.org/10.1016/j.jcomdis.2016.07.003
Kenworthy, L., Yerys, B. E., Anthony, L. G., & Wallace, G. L.
(2008). Understanding executive control in autism spectrum disorders in the lab
and in the real world. Neuro, 18, 320–338.
https://doi.org/10.1007/s11065-008-9077-7
Kjelgaard, M. M., & Tager‐Flusberg, H. (2001). An investigation
of language impairment in autism: Implications for genetic subgroups. Language
and Cognitive Processes, 16, 287–308. https://doi.org/10.1080/01690960042000058
Leekam, S. R., Prior, M. R., & Uljarevic, M. (2011). Restricted
and repetitive behaviors in autism spectrum disorders: A review of research in
the last decade. Psychological Bulletin, 137, 562–593. doi: 10.1037/a0023341
Leung, R. C., Vogan, V. M., Powell, T. L., Anagnostou, E., &
Taylor, M. J. (2016). The role of executive functions in social impairment in
autism spectrum disorder. Child Neuropsychology, 22, 336–344.
https://doi.org/10.1080/09297049.2015.1005066
Leung, R. C., & Zakzanis, K. K. (2014). Brief report: Cognitive
flexibility in autism spectrum disorders: A quantitative review. Journal of
Autism and Developmental Disorders, 44, 2628–2645.
https://doi.org/10.1007/s10803-014-2136-4
Luyster, R. J., Kadlec, M. B., Carter, A., & Tager‐Flusberg, H.
(2008). Language assessment and development in toddlers with autism spectrum
disorders. Journal of Autism and Developmental Disorders, 38, 1426–1438.
https://doi.org/10.1007/s10803-007-0510-1
Minshew, N. J., & Goldstein, G. (2001). The pattern of intact
and impaired memory functions in autism. Journal of Child Psychology and
Psychiatry, 42, 1095–1101. https://doi.org/10.1111/1469-7610.00808
Miyake, A., Friedman, N. P., Emerson, M. J., Witzki, A. H., Howerter,
A., & Wager, T. D. (2000). The unity and diversity of executive
functions and their contributions to complex “frontal lobe” tasks: A latent
variable analysis. Cognitive Psychology, 41, 49–100. https://doi.org/10.1006/cogp.1999.0734
Morton, J. B., & Harper, S. N. (2007). What did Simon say?
Revisiting the bilingual advantage. Developmental Science, 10, 719–726.
https://doi.org/10.1111/j.1467-7687.2007.00623.x
Namazi, M., & Thordardottir, E. (2010). A working memory, not
bilingual advantage, in controlled attention. International Journal of
Bilingualism Education and Bilingualism, 13, 597–616.
https://doi.org/10.1080/13670050.2010.488288
Ozonoff, S. (1995). Reliability and validity of the Wisconsin Card
Sorting Test in studies of autism. Neuropsychology, 9, 491–500.
https://doi.org/10.1037/0894-4105.9.4.491
Ozonoff, S., Pennington, B. F., & Rogers, S. J. (1991).
Executive function deficits in high‐functioning autistic individuals:
Relationship to theory of mind. Journal of Child Psychology and Psychiatry, 32,
1081–1105. https://doi.org/10.1111/j.1469-7610.1991.tb00351.x
Ozonoff, S., & Strayer, D. L. (2001). Further evidence of intact
working memory in autism. Journal of Autism and Developmental Disorders, 31,
257–263. https://doi.org/10.1023/A:1010794902139
Paap, K. R., & Greenberg, Z. I. (2013). There is no coherent
evidence for a bilingual advantage in executive processing. Cognitive
Psychology, 66, 232–258. https://doi.org/10.1016/j.cogpsych.2012.12.002
Paradis, J., Genesee, F., & Crago, M. B. (2011). Dual language
development & disorders: A handbook on bilingualism and second language
learning (2nd ed.). Baltimore, MD: Paul H. Brookes.
Pascualvaca, D. M., Fantie, B. D., Papageorgiou, M., & Mirsky,
A. F. (1998). Attentional capacities in children with autism: Is there a
general deficit in shifting focus? Journal of Autism and Developmental
Disorders, 28, 467–478. https://doi.org/10.1023/A:1026091809650
Pearson, B. Z., Fernandez, S. C., Lewedeg, V., & Oller, K.
(1997). The relation of input factors to lexical learning by bilingual infants.
Applied Psycholinguistics, 18, 41–58. https://doi.org/10.1017/S0142716400009863
Poulin‐Dubois, D., Blaye, A., Coutya, J., & Bialystok, E.
(2011). The effects of bilingualism on toddlers’ executive functioning. Journal
of Experimental Child Psychology, 108, 567–579.
https://doi.org/10.1016/j.jecp.2010.10.009
Prior, A., & Macwhinney, B. (2010). A bilingual advantage in
task switching. Bilingualism: Language and Cognition, 13, 253–262.
https://doi.org/10.1017/s1366728909990526
Psychology Software Tools. (2012). [E‐Prime 2.0]. Pittsburgh, PA.
Retrieved from http://www.pstnet.com
Riches, N. G., Loucas, T., Baird, G., Charman, T., & Simonoff,
E. (2010). Sentence repetition in adolescents with specific language
impairments and autism: An investigation of complex syntax. International
Journal of Language & Communication Disorders, 45, 47–60.
https://doi.org/10.3109/13682820802647676
Robinson, S., Goddard, L., Dritschel, B., Wisley, M., & Howlin,
P. (2009). Executive functions in children with autism spectrum disorders.
Brain and Cognition, 71, 362–368. https://doi.org/10.1016/j.bandc.2009.06.007
Roid, G. H., & Miller, L. J. (1997). Leiter international
performance scale‐revised. Wood Dale, IL: Stoelting.
Russell, J., Jarrold, C., & Henry, L. (1996). Working memory in
children with autism and with moderate learning difficulties. Journal of Child
Psychology and Psychiatry, 6, 673–686.
https://doi.org/10.1111/j.1469-7610.1996.tb01459.x
Russo, N., Flanagan, T., Iarocci, G., Berringer, D., Zelazo, P. D.,
& Burack, J. A. (2007). Deconstructing executive deficits among persons
with autism: Implications for cognitive neuroscience. Brain and Cognition, 65,
77–86. https://doi.org/10.1016/j.bandc.2006.04.007
Rutter, M., Bailey, A., & Lord, C. (2003). SCQ. The Social
Communication Questionnaire. Torrance, CA: Western Psychological Services.
Schmitz, N., Rubia, K., Daly, E., Smith, A., Williams, S., &
Murphy, D. G. (2006). Neural correlates of executive function in autistic
spectrum disorders. Biological Psychiatry, 59, 7–16.
https://doi.org/10.1016/j.biopsych.2005.06.007
Schuh, J., & Eigsti, I. M. (2012). Working memory, language
skills, and autism symptomatology. Behavioral Sciences, 2, 207–218.
https://doi.org/10.3390/bs2040207
Semel, E., Wiig, E. H., & Secord, W. A. (2003). Clinical
Evaluation of Language Fundamentals, 4th ed.. San Antonio, TX: Psychological
Corporation.
Toplak, M. E., West, R. F., & Stanovich, K. E. (2013).
Practitioner review: Do performance‐based measures and ratings of executive
function assess the same construct? Journal of Child Psychology and Psychiatry,
54, 131–143. doi: 10.1111/jcpp.12001
Van Eylen, L., Boets, B., Steyaert, J., Evers, K., Wagemans, J.,
& Noens, I. (2011). Cognitive flexibility in autism spectrum disorder:
Explaining the inconsistencies? Research in Autism Spectrum Disorders, 5,
1390–1401. https://doi.org/10.1016/j.rasd.2011.01.025
Weismer, S. E., Lord, C., & Esler, A. (2010). Early language
patterns of toddlers on the autism spectrum compared to toddlers with
developmental delay. Journal of Autism and Developmental Disorders, 40,
1259–1273. https://doi.org/10.1007/s10803-010-0983-1
Wiig, E. H., Semel, E., & Secord, W. A. (2006). Clinical
Evaluation of Language Fundamentals – Fourth Edition, Spanish. San Antonio, TX:
The Psychological Corporation.
Wiig, E., Secord, W., Semel, E., Boulianne, L., & Labelle, M.
(2009). Évaluation clinique des notions langagières fondamentales: Version pour
francophones du Canada (Clinical evaluation of language fundamentals: French
Canadian version). Toronto, Ontario, Canada: Pearson Canada Assessment.
Williams, D. L., Goldstein, G., Carpenter, P. A., & Minshew, N.
J. (2005). Verbal and spatial working memory in autism. Journal of Autism and
Developmental Disorders, 35, 747–756. https://doi.org/10.1007/s10803-005-0021-x
Williams, D. L., Goldstein, G., & Minshew, N. L. (2006). The
profile of memory function in children with autism. Neuropsychology, 20, 21–29.
https://doi.org/10.1037/0894-4105.20.1.21
Yi, L., Liu, Y., Li, Y., Fan, Y., Huang, D., & Gao, D. (2012).
Visual scanning patterns during the dimensional change card sorting task in
children with autism spectrum disorder. Autism Research and Treatment, 2012,
1–11. https://doi.org/10.1155/2012/123053
Zelazo, P. D. (2006). The dimensional change card sort (DCCS): A method
of assessing executive function in children. Nature Protocols, 1, 297–301.
https://doi.org/10.1038/nprot.2006.46
No hay comentarios:
Publicar un comentario
Aportes