Executive functions in two-way dual-language education: A mechanism for academic performance

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=ubrj20

Bilingual Research Journal

The Journal of the National Association for Bilingual Education

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/ubrj20

Executive functions in two-way dual-language

education: A mechanism for academic

performance

Alena G. Esposito

To cite this article: Alena G. Esposito (2021): Executive functions in two-way dual-language

education: A mechanism for academic performance, Bilingual Research Journal, DOI:

10.1080/15235882.2021.1874570

To link to this article: https://doi.org/10.1080/15235882.2021.1874570

Published online: 03 Feb 2021.

Submit your article to this journal

View related articles

View Crossmark data

RESEARCH ARTICLE

Executive functions in two-way dual-language education: A

mechanism for academic performance

Alena G. Esposito

Clark University, USA

Children across the United States are increasingly learning academic content through two-way dual-

language education (http://www.cal.org/twi/). This education model provides instruction through two

languages in classrooms comprised of approximately equal numbers of native and non-native English

speakers. For both language groups, this educational model is an eﬀective approach for

achieving second-language ﬂuency (García & Náñez, 2011; Lindholm-Leary & Genesee, 2014).

Importantly, both native and non-native English speakers in dual-language education programs

perform as well or better academically than their peers in mainstream English classrooms (e.g.,

Marian, Shook, & Schroeder, 2013; Steele et al., 2017). However, the mechanisms that explain this

academic advantage remain to be understood. We examined the possibility that enhanced executive

functions through second-language exposure underlie the academic beneﬁts of dual-language educa-

tion in a rural, low-income, sample of elementary school students.

Dual-language education and participating school system

Bilingual education is an umbrella term that encompasses two-way dual-language education, one-way

dual-language education, and immersion education models among others. Dual-language education

models teach academic content through two languages, such that children learn both the languages

and the content as they progress through school. The 3 primary goals for dual-language programs are

to support academic achievement, develop bilingualism and biliteracy, and foster socio-cultural

competence (Howard et al., 2018). In two-way dual-language education, classes aim for a 50/50

composition between speakers of the paired languages (language pairings vary; English/Spanish is

common in the U.S.) and at least 50% of content provided through the partner language (some models

provide up to 90%, initially, tapering down to 50%). Two-way dual-language diﬀers from one-way

dual-language in the 50/50 composition of speakers. Dual-language diﬀers from immersion models in

that there is typically less content provided through the partner language (often 90–100% and typically

comprised non-native speakers of that language; e.g., English speakers in a French immersion in

Montreal). Thus, dual-language refers to both one-way and two-way programs and bilingual educa-

tion is a broader term encompassing dual-language and immersion models.

The speciﬁc collaborating school system for the reported work developed a two-way dual-language

program as an optional strand within the schools to address the needs of their community (approxi-

mately one-third

are native Spanish speaking). In this particular model, instruction alternates days

between Spanish and English, thus providing a 50–50 model of content instruction through each

language. Children enter through two lotteries, one for native English speakers and one for native

Spanish speakers, ensuring the 50/50 composition of native speakers of each of the partner languages.

CONTACT Alena G. Esposito [email protected] Clark University, 950 Main Street, Worcester, MA 01610, USA.

Dr. Alena G. Esposito is a developmental psychologist focused on cognitive development. Her research concentration is malleable

factors inﬂuencing learning and subsequent academic achievement. Esposito is a former elementary educator who earned her Ph.D.

in Lifespan Developmental Psychology from North Carolina State University.

BILINGUAL RESEARCH JOURNAL

https://doi.org/10.1080/15235882.2021.1874570

Students not in the program are placed in a mainstream English education model within the same

school.

Dual-language education and academic performance

Multiple studies indicate that children in bilingual education models (including dual-language and

immersion models) have academic outcomes that match or even exceed those of their peers in

mainstream education models, especially in later elementary grades, (e.g., Cobb, Vega, & Kronauge,

2006; Lindholm-Leary & Genesee, 2014; Padilla, Fan, Xu, & Silva, 2013; Steele et al., 2017). For

example, Marian et al. (2013) investigated the academic achievement of students in grades 3, 4, or 5

(approximately ages 8–10 years), a portion of which were enrolled in a two-way dual-language

program. They found an advantage in academic performance in math across all three grade levels

and reading in 3

grade. Similarly, Watzinger-Tharp, Swenson, and Mayne (2018) examined growth

in over 2000 4

grade students in either mainstream English education or a dual-language education

model (comprised of both one-way and two-way models across three partner languages). In

a matched-sample of mainstream and dual-language students, the dual-language students showed

greater growth in math achievement across the 4

grade year.

Despite seemingly robust evidence for an academic advantage for bilingual education participants,

the eﬀect is still in question. In a meta-analyses of 10 studies reporting academic performance for

students in one-way or two-way dual-language programs compared to mainstream programs, Hill

(2018) determined the eﬀect to be null. The results indicated a small positive eﬀect that Hill proposed

could be easily nulliﬁed by the inclusion of a few studies with even small negative eﬀects. He also

questioned whether the reported results that show an advantage to those in dual-language education

are due to their participation or can be explained by changing demographics as a result of attrition.

Attrition, he reports, is likely to positively aﬀect the socio-economic status of the group because low-

income families are more transitory and more likely to relocate, leaving predominantly higher socio-

economic status students in the program. Socio-economic status is a known predictor of academic

performance (e.g., Hoﬀ, 2013; Nesbitt, Baker-Ward, & Willoughby, 2013). The question of attrition

also arises as students who are struggling academically may be more likely to leave the program for

mainstream education models. It is, therefore, important to examine academic achievement with

consideration for student intelligence and family socio-economic status.

Dual-language education and executive functions

The “bilingual advantage” refers to higher performance by bilingual compared to monolingual

individuals in executive functions (for a review, see Bialystok, Craik, & Luk, 2012). Executive functions

are the top-down processes that are required for eﬀortful cognition such as reasoning, problem-

solving, and planning and include the core components of inhibition, interference control, working

memory, and cognitive ﬂexibility (Diamond, 2013). Executive functions are positively correlated to

both socio-economic status (SES; e.g., Nesbitt et al., 2013) and academic performance (for review, see

Serpell & Esposito, 2016). Bilingualism may be a protective factor for children from low-SES back-

grounds, providing an advantage that oﬀsets the disadvantages associated with their economic

conditions. The bilingual advantage is thought to result from constant practice in managing two

languages, which enhances mental ﬂexibility and controlled attention. The advantage has been found

across age groups, languages, and geographic locations. However, the speciﬁc conditions under which

a bilingual advantage is and is not found have not been elucidated (e.g., Yang, Hartanto, & Yang, 2016;

Valian, 2015). While some research shows beneﬁts to executive functions after short periods of intense

training (e.g., Janus, Lee, Moreno, & Bialystok, 2016), others show that beneﬁts only emerge after

a threshold of proﬁciency in both languages is met (e.g., De Cat, Gusnanto, & Serratrice, 2018). Thus,

the second-language exposure gained through the speciﬁc context of two-way dual-language educa-

tion may not be suﬃcient for an executive functions advantage to develop.

2 A. G. ESPOSITO

The research investigating whether an advantage develops for children in bilingual education

models has found mixed results. Studies examining immersion education have found support for

emerging beneﬁts to executive functions after 3 or more years of participation (e.g., Bialystok & Barac,

2012; Nicolay & Poncelet, 2013, 2015). In contrast, several studies have failed to ﬁnd diﬀerences

between two-way dual-language and mainstream monolingual education models (e.g., Kaushanskaya,

Gross, & Buac, 2014; Poarch & van Hell, 2012) with at least one study ﬁnding a disadvantage (Purić,

Vuksanović, & Chondrogianni, 2017). These studies, however, examined children with less than

2 years of experience in dual-language programs and, where reported, children were from middle-

class backgrounds. Hartanto, Toh, and Yang (2018) found that the bilingual advantage was only

evident for children from low-SES backgrounds. Thus, if the conditions of two-way dual-language

education are in-line with developing a bilingual advantage, it is more likely to emerge for children

who have more than 2 years of experience in the program and are from low-SES backgrounds who can

most beneﬁt from an intervention. In support of this hypothesis, a study examining controlled

attention for children in an area of marked poverty enrolled in either mainstream monolingual

education or two-way dual-language education for more than 3 years found evidence for emerging

beneﬁts (Esposito & Baker-Ward, 2013).

As documented in an extensive literature, executive functions correlate with indices of academic

performance (for review, see Serpell & Esposito, 2016). For example, Best, Miller, and Naglieri (2011)

measured academic achievement and executive functions in a nationally representative sample of

children aged 5–17 years and found a consistent relation between executive functions performance

and academic achievement in math and reading. Thus, if two-way dual-language education conveys

a beneﬁt to executive functions, that beneﬁt could support greater academic achievement.

The present study

We examined whether enhanced executive functions gained through participation in a two-way dual-

language program are a mechanism through which an academic advantage emerges. In sum, there is

evidence for an academic advantage in bilingual education models, but the advantage is still in

question and the mechanism is unknown. We propose that two-way dual-language education fosters

executive functions similar to the advantage found in bilingual individuals and that well-developed

executive functions are a mechanism for an academic advantage. In the present cross-sectional study,

we recruited a sample of primary and intermediate elementary students in either two-way dual-

language education (Spanish/English) or mainstream English education in an area of rural poverty

within the same schools. The participating school system provided academic data. Parents provided

information about the home environment and family demographics. We met with students individu-

ally to measure executive functions as well as variables that permitted us to create a matched-sample of

students in two-way dual-language education and the mainstream English model. The full-sample

provided a larger sample size and included all participants, including those on the ends of the score

distribution. The matched-sample, however, allowed us to address three criticisms of previous

literature, namely that few studies equated groups on individual intelligence, family socio-economic

status, and sample size.

The present study had four research questions: 1) is there an academic advantage for children

enrolled in two-way dual-language compared to mainstream English education in an area of rural

poverty; 2) is the second-language experience provided through a 50/50 two-way dual-language

education model suﬃcient to beneﬁt executive functions; and, if so, 3) is there evidence that executive

functions are a mechanism through which the academic advantage for two-way dual-language

participants emerges? Relatedly, 4) does the pattern of results in a full-sample analyses replicate in

a matched-sample controlling for participant intelligence and family demographics? We predicted

that we would replicate the dual-language academic advantage in late elementary students, especially

in math where the ﬁndings of an advantage appear to be the most robust. We also predicted a dual-

language advantage in executive functions for late elementary students. We predicted that executive

BILINGUAL RESEARCH JOURNAL 3

functions would mediate the relation between educational model and academic performance, provid-

ing evidence for a mechanism of the academic advantages found in models of bilingual education. We

predicted the reduced sample size and elimination of the extreme ends of the score distributions of the

matched-sample would reduce the power and eﬀect sizes of the results, but that the pattern of results

would be similar to those found in the full-sample.

Method

Participants and school system

The participants were 288 children (primary = grades K-1, n= 175, Mean age = 6 years, 11 months;

intermediate = grades 4–5, n= 113, Mean age =10 years, 9 months), enrolled in a rural public school

system in the southeastern United States (see Table 1). The matched-sample was a subset exact

matched on grade and we used Coarsened Exact Matching (CEM) on verbal and non-verbal intelli-

gence as well as parent/guardian education level (n=136). The school system includes 266 mi

and all

students within the county attended the same 2 schools as part of a continuous progression through

the county grade school program (K-2 primary school, early elementary; and 3–5 intermediate school;

Table 1. Full Sample and Matched Sample by School.

Primary Intermediate

Dual-Language Mainstream Dual-Language Mainstream

Full CEM Full CEM Full CEM Full CEM

n (female) 50 (27) 42 (22) 125 (62) 42 (16) 34 (22) 26 (13) 79 (36) 26 (17)

Mean WASI

Vocabulary

(SD)

19.56

(9.99)

19.13

(9.43)

18.82

(8.94)

19.23 (9.17) 38.00 (8.16) 37.62 (7.74) 35.38 (6.58) 36.50 (6.74)

Mean WASI

Blocks (SD)

8.67 (6.08) 7.10 (4.74) 6.59 (4.58) 7.28 (4.92) 21.48 (12.01) 21.29 (12.20) 16.95 (10.14) 18.65 (10.19)

Caregiver

Education

Level (SD)

4.33 (1.78) 4.38 (1.86) 4.08 (1.49) 4.05 (1.55) 4.43 (2.02) 4.61 (1.94) 4.26 (1.62) 4.29 (1.27)

Hours Spent with

an Adult on

Homework

3.32 (1.78) 3.31 (1.83) 3.38 (2.31) 3.40 (2.41) 2.13 (1.78) 2.37 (1.86) 2.95 (2.25) 2.64 (1.94)

Hours Spent with

an Adult

Reading

2.94 (1.81) 3.08 (1.90) 3.20 (2.17) 3.46 (2.49) 1.52 (1.17) 1.47 (1.17) 2.45 (2.05) 2.44 (2.06)

Participation in

extra-

curriculars (%

of participants

indicating at

least one)

42.00 35.70 55.20 57.10 50.00 50.00 58.20 50.00

Individual

Education Plan

(IEP; % of

participants

with one)

12.00 7.10 14.40 21.50 14.70 7.70 21.50 11.50

Number of Adults

in the Home

1.78 (.71) 1.73 (0.60) 2.00 (.73) 1.98 (0.91) 2.08 (0.50) 2.11 (0.46) 2.20 (1.22) 2.21 (1.32)

Number of

Children in the

Home

1.44 (1.34) 1.19 (0.94) 1.65 (1.28) 1.76 (1.26) 1.21 (0.88) 1.11 (0.81) 1.73 (1.43) 1.72 (1.51)

Spanish Fluency

task (SD)

7.88 (5.35) 7.78 (5.46) 2.54 (4.89) 2.11 (4.56) 12.86 (8.25) 14.48 (8.55) 3.98 (7.27) 1.95 (6.03)

Caregiver education level was coded as completed elementary school =1; completed middle school = 2; completed high school =3;

some school beyond high school =4; completed an associate degree or other training program =5; completed a bachelor’s

degree =6; beyond college =7.

4 A. G. ESPOSITO

late elementary). Reﬂecting the diversity of the community, the sample was comprised of an approxi-

mately equal number of Black (n = 91), non-Hispanic White (n = 78), and Hispanic White (n = 92)

participants, with an additional 27 participants identifying as multiracial. We recruited participants

with letters distributed by their teachers and only those children whose parents/guardians provided

written consent participated (52% of population; total providing consent n = 454). The reported

sample (n= 288) reﬂects all students whose family consented and completed the parent/guardian

questionnaire.

As described, the school system oﬀers two educational tracks: mainstream English (monolingual)

or two-way dual-language (TWDL; Spanish/English; 50% split instructional time). All children whose

data were included in this study enrolled in their education model in kindergarten and maintained

a stable placement. The TWDL program is housed within the school and TWDL and Mainstream

English classrooms are alongside each other with the same resources. Entrance into the TWDL

education model is by lottery at kindergarten registration. Not all families enter the lottery and the

percentage of parents who do is not available. All children placed in the TWDL program are placed by

lottery, but not all children in Mainstream English entered the lottery for TWDL placement. In light of

the lack of complete randomization, we implemented Coarsened Exact Matching (CEM; Iacus, King,

& Porro, 2012) to create a matched sample and examine diﬀerences in performance due to educational

program assignment. This results in a quasi-experimental design (Shadish, Cook, & Campbell, 2002).

Measures

The task battery consisted of measures included for the purpose of creating matched-pairs, measures

to examine group diﬀerences, and those that were the target of the investigation. For matching

purposes, we included verbal and non-verbal intelligence as well as parent/guardian education level.

Group diﬀerence measures included parent/guardian report of academic involvement and household

density as well as a measure of Spanish ﬂuency. Target variables included academic measures obtained

from the participating school system and measures of executive functions, both computerized and

teacher report.

Wechsler abbreviated scale of intelligence (WASI)

The WASI was designed to be a quick and reliable measure of general intelligence appropriate for ages

6–90, has reliabilities ranging from .92-.98, test–retest stability of .88 for children aged 6–11, and has

been validated with other tests in the Weschler library (Weschler, 1999). We administered the

Vocabulary subtest to measure word knowledge and verbal concept formation. We also administered

the Block Design to measure nonverbal concept formation. Importantly, the WASI was not adminis-

tered as a diagnostic tool and should not be interpreted as a clinical intelligence measure. This measure

was administered for comparative purposes and was used as a matching variable. Both tests were

conducted entirely in English. The total score was recorded for each subtest, resulting in two

independent variables, which were used to create the matched sample.

Parent and teacher data

Parents completed a written questionnaire including family demographics and students’ activities (See

Table 2). With parental authorization, the school guidance counselors provided academic data for

participating children (see Table 2). The measures collected diﬀered by grade. In primary (K and 1

grade), language arts (reading and writing) performance was measured with the mCLASS: Text

Reading Comprehension (TRC). The TRC tracks reading development across the elementary years.

Children read leveled text passages and complete comprehension questions. Expected progress is

a level D, or level 6, by the end of kindergarten and a level J, or 18, by the end of ﬁrst grade. The math

measure was year-end-grade achieved by each student as an accumulation of classroom assessments

across the academic year. For intermediate (4

and 5

grades), we collected year-end grades for

language arts, math, science, and social studies for a total of four means. Grades are on a 100 point

BILINGUAL RESEARCH JOURNAL 5

scale. We also collected scores on the state-standardized end-of-grade tests (EOG). The tests are

administered beginning in the third grade in all public schools in North Carolina. Scores range from 1

to 4, with a three the minimum to advance to the next grade without remediation. Both 4

and 5

grade students take the language arts and math tests, for a total of 2 means.

Spanish category ﬂuency

Although time restrictions from the collaborating school system prevented a full language test battery,

the ﬂuency task allowed us to get an indication of whether children in the TWDL model were

developing Spanish language skills at a higher rate than their peers in the mainstream model, an

important premise of bilingual education and the bilingual advantage. Tasks such as these, although

administered in English, are often part of intelligence scales (such as the Weschler Intelligence Scale for

Children, Revised; Weschler, 1974). In this task children were ﬁrst asked to name as many animals

within a minute as they could, using Spanish, “such as perro [dog] or gato [cat]” and then to name as

many things to eat or drink as they could within one minute, using Spanish, “such as leche [milk] or

pan [bread].” Experimenters recorded correct, unique, responses and the total number was summed

for one measure of Spanish Category Fluency.

Computerized executive functions measures

Computerized tasks used the Psychology Experiment Building Language (PEBL: Mueller, 2011)

delivered on Compaq Presario CQ60 laptop computers attached to HP Compaq L2105 21.5 inch

color touch screen monitors.

Switching. We administered a version of the Trail Making Task developed speciﬁcally for children

(Delis, Kaplan, & Krames, 2001). Children completed three sequencing trails. The ﬁrst trail required

numeration of numbers from 1 to 16 (1-2-3). In the second, the sequence was alphabetical from

A-K (A-B-C). In the last, the sequence alternated between a number and a letter (1-A-2-B-3-C).

Table 2. Mean scores (and SDs) on tests of academic achievement by grade level and educational model.

Primary Intermediate

Dual-Language Mainstream Dual-Language Mainstream

Academic Means Full CEM Full CEM Full CEM Full CEM

Language Arts (SD)

Kindergarten TRC 6.89 (3.43) 7.35 (3.25) 5.88 (3.43) 5.80 (3.47)

First Grade TRC 16.46

(5.64)

16.33

(4.56)

19.02

(6.37)

20.22

(5.82)

Year-average 93.24

(3.59)

93.72

(2.87)

88.05

(6.49)

90.16

(5.28)

End-of-grade 2.15

(0.91)

2.08

(0.81)

1.81

(0.83)

2.08

(0.86)

Math (SD)

Kindergarten Year-

average

84.64

(9.35)

84.18

(9.22)

82.71

(13.99)

82.13

(14.10)

First Grade

Year = average

80.08

(11.39)

81.39

(11.67)

83.77

(12.83)

82.17

(15.70

Year-average 94.15

(3.29)

94.36

(2.72)

85.91

(7.50)

87.88

(6.90)

End-of-grade 2.41

(0.89)

2.42

(0.90)

1.80

(0.96)

1.96

(0.98)

Science (SD)

Year-average 94.73

(1.86)

94.72

(2.01)

90.74

(5.05)

92.40

(3.42)

Social Studies (SD)

Year-average 96.15

(2.99)

96.56

(2.65)

90.65

(6.32)

91.32

(5.13)

TRC is Text Reading and Comprehension. Year-average grades refer to those determined by classroom teachers based on school-

wide assessments. End-of-grade refers to standardized state tests.

6 A. G. ESPOSITO

Response time in seconds is calculated for each trail, with the last trail being a measure of switching

performance. The PEBL version of this task has been tested for validity (Piper et al., 2012) and higher

test–retest reliability than paper versions (r= .61-.74 vs. r= .45, respectively; Piper et al., 2015).

Bilingual children have previously shown an advantage over monolingual children on this task

(Bialystok, 2010; Esposito & Baker-Ward, 2013).

Inhibitory control. We included both a stimulus-stimulus conﬂict measure as well as a stimulus-

response conﬂict measure of inhibitory control. The Bivalent Shape task (stimulus-stimulus conﬂict:

Mueller & Esposito, 2014) was developed to measure the ability to ignore salient features of a stimulus

and act on the relevant information. The task has been used to measure executive functions in this age

range previously (e.g., Esposito & Bauer, 2018). Two active buttons are at the bottom of the screen,

a red circle, and a blue square. Stimuli (circles and squares in red, blue, or a black outline for six

possible test items) appeared in the center of the screen and participants are directed to match the

shape. Congruent stimuli matched in both color and shape, incongruent stimuli matched in shape but

not in color. Neutral items did not have color, and thus no facilitating or distracting element. The task

consisted of practice blocks for each type of stimuli followed by a mixed block in which the three types

of stimuli were presented in a set randomized order with 10 of each type for a total of 30 trials.

Children had up to 3 seconds to respond. Mean reaction times on correct trials for each trial type

produce three dependent variables (congruent, incongruent, and neutral). Bilingual children have

shown an advantage over their monolingual peers on this task (Esposito, Baker-Ward, & Mueller,

2013).

The Simon task represents a stimulus-response conﬂict. Participants responded to the appearance

of a green or orange rectangle on the left or right side of the monitor by pressing the right or left shift

buttons in response to the color, ignoring the location of the object on the screen (Lu & Proctor, 1995;

Simon & Wolf, 1963). The task has good reliability (Cronbach’s alpha = .88, with adults in short form;

Cevada, Conde, Marques, & Deslandes, 2019) and has been validated as a marker of attention deﬁcit

along with other executive function tasks in children (e.g., Mullane, Corkum, Klein, & McLaughlin,

2009). Individuals respond more quickly if the object is on the same side of the screen as the shift

button that correlates to the color and they respond more slowly when the position of the object is on

the opposite side of the appropriate shift button. The Simon task has a history of elucidating a bilingual

advantage, especially in children (e.g., Martin-Rhee & Bialystok, 2008). Mean reaction times for

correct trials were calculated for congruent and incongruent trial types for a total of two dependent

variables.

Behavioral executive functions measure

Teachers completed the Behavioral Rating Inventory of Executive Functions (BRIEF; Gioia, Isquith,

Guy, & Kenworthy, 2015). The BRIEF is an 86 item behavioral checklist normed for children aged

5–18. Respondents mark each behavior as “never,” “sometimes,” or “always” and the responses are

summed into a Global Executive Composite that documents executive dysfunction. Reliability as

measured by Cronbach’s alpha ranges from .8-.98. The test is validated across age, socio-economic,

race, and ethnic groups within the United States. A higher score indicates greater dysfunction.

Procedures

Children were tested individually in a quiet classroom within their school during a single session

lasting approximately 30 minutes (the time allotted by the participating school system). Experimenters

were six female psychology students with intermediate to advanced Spanish ability. All participating

children provided verbal assent. The university institutional review board and participating school

system school board reviewed and approved all procedures. The computer tasks were administered in

random order. All computer tasks are designed to be non-verbal and instructions were administered

in the child’s preferred language (children were asked their preference; English or Spanish). Computer

BILINGUAL RESEARCH JOURNAL 7

tasks were followed by the Spanish ﬂuency task. We concluded with the WASI, which was adminis-

tered in English as per the manual. All assessments, reports, and teacher ratings were completed in the

last month of the school year.

Matching

In order to control for individual and group diﬀerences that could inﬂuence the outcome variables of

interest as well as control for diﬀerences in sample size between groups, we created a matched-sample

using Coarsened Exact Matching. CEM analysis utilized an R plugin (R version 3.3.0; CEM Extension

Bundle). Each TWDL participant was paired to a child enrolled in mainstream English education.

Grade was exactly matched while verbal intelligence, non-verbal intelligence, and parent/guardian

education level were coarsened. Of the 288 participants, 68 pairs were created for a total of 136

participants included in the CEM subset. Means are reported in Table 1, with matching variables

outlined for easy identiﬁcation and separated by Education Model (TWDL vs. Mainstream English)

and School (primary vs. intermediate).

Results

The results are reported in three parts; examination of group diﬀerences, academic performance, and

executive functions. Evidence for executive functions as a mediator for the relation between education

model and academic performance in then examined. All reported analyses are two-tailed and were

conducted using SPSS 24 software.

Group dierences

We ﬁrst examined whether the educational groups diﬀered in observable ways that could impact

academic performance (means reported by group in Table 1). Group diﬀerences in verbal and non-

verbal intelligence, parent/guardian level of education, parent/guardian academic involvement (home-

work help and reading), participation in extra-curricular activities, child has an individual educational

plan (IEP; implemented for special needs), and the density of the home environment were tested with

a 2 (Educational Model) x 2 (School) multivariate analyses of variance (MANOVA). There was a main

eﬀect of School, F(9, 185) = 24.42, p < .001, ŋ

= .54, such that the intermediate students, compared to

the primary students, scored higher in verbal (M = 19.01 vs. 36.63; F(1, 197) = 160.44, p < .001, ŋ

.45) and non-verbal intelligence (M = 7.75 vs. 19.59; F(1, 197) = 85.24, p < .001, ŋ

= .31). Families with

children in the intermediate school, compared to the primary school, reported spending fewer hours

supervising homework (M = 2.44 vs. 3.36; F(1, 197) = 5.56, p = .02, ŋ

= .03) and reading with their

child (M = 1.98 vs. 3.12; F(1, 197) = 8.81, p = .003, ŋ

= .04). There were no other main eﬀects or

interactions.

The CEM procedure was intended to minimize diﬀerences between education groups on socio-

economic and general aptitude. However, we recognize that the measures included are a proxy and

incomplete. We, thus, examined whether the educational groups diﬀered in observable ways that could

impact academic performance but were not included in the matching variables in a 2 (Education

Model) x 2 (School) MANOVA with parent/guardian academic involvement (homework help and

reading), participation in extra-curricular activities, child has an IEP, and the density of the home

environment as dependent variables. The results were similar to those of the full-sample analyses.

There was a main eﬀect of School, F(6, 96) = 2.72, p = .02, ŋ

= .15, such that families in the

intermediate school reported spending less time supervising homework (M = 2.51 vs. 3.36; F(1,

105) = 5.71, p = .02, ŋ

= .05) and reading with their children (M = 2.01 vs. 3.31; F(1, 105) = 12.44,

p = .001, ŋ

= .11). There was also a main eﬀect of Education Model, F(6, 96) = 2.26, p = .04, ŋ

= .12,

such that families in the dual-language program reported fewer children in the home compared to

8 A. G. ESPOSITO

those in mainstream education (M = 1.16 vs. 1.74; F(1, 105) = 4.77, p = .03, ŋ

= .05). There were no

other main eﬀects and no interactions.

We also examined the Spanish Fluency of participants. We examined this in a 2 (Education Model)

x 2 (School) Analysis of Variance (ANOVA).

In the full sample, there was a main eﬀect of Education

Model such that students in the TWDL program had higher scores than their peers in the mainstream

model, F(1, 238) = 65.74, p < .001, ŋ

= .22. There was a main eﬀect of School such that the

intermediate students had higher performance compared to the primary students, F(1, 238) = 13.44,

p < .001, ŋ

= .05. There was also a signiﬁcant interaction, F(1, 238) = 4.08, p = .04, ŋ

= .02. Follow-up

univariate tests revealed that the dual-language intermediate students had signiﬁcantly higher perfor-

mance than the primary students, F(1, 77) = 10.49, p = .002, whereas there was no diﬀerence between

educational programs for those in the mainstream education program, F(1, 159) = 2.25, p = .14.

The same analyses with the CEM sample replicated these results. There were signiﬁcant main

eﬀects of Education Model, F(1, 119) = 61.86, p < .001, ŋ

= .35, and School, F(1, 119) = 7.98, p = .006,

= .07, and a signiﬁcant interaction, F(1, 119) = 8.77, p = .004, ŋ

= .07. Follow-up univariate tests

also replicated the previous analyses with dual-language intermediate students producing signiﬁcantly

more words than primary students, F(1, 62) = 14.06, p < .001, ŋ

= .19, while there was no diﬀerence

for those in mainstream education, F(1, 57) = 0.01, p = .91.

In summary, there were few diﬀerences found between the Education models with the exception of

Spanish ﬂuency performance. There were also few diﬀerences between the full sample and the CEM

sample.

Academic performance

Primary school

Academic performance is reported by grade and education model in Table 2.

Primary school

academic achievement was represented with a reading measure (TRC) and math measure (year-end

classroom grade). For the primary school, we had access to kindergarten data for all participants

(including current ﬁrst-grade students). Thus, we were able to analyze kindergarten data for all

primary students (n = 156) as a MANOVA with Education Model as a predictor. The MANOVA

revealed no signiﬁcant diﬀerences in reading or mathematical performance between Education

Models, F(2, 153) = 1.43, p = .24. In addition to kindergarten data, ﬁrst-grade students also had

data from their ﬁrst-grade year. This enabled a short-term longitudinal examination of the growth

between kindergarten and ﬁrst grade for children in the TWDL program compared to those in the

mainstream education model. We did a repeated measures MANOVA with Education Model as

a predictor. There was a main eﬀect such that math scores decreased between kindergarten and ﬁrst

grade (M = 1.98 vs. 3.12; F(1, 72) = 55942.04, p < .001, ŋ

= .99). It is important to note that the math

measure is based on the content from that year, so a lower grade does not indicate less knowledge

overall, but poorer performance on the more advanced content. There were no other signiﬁcant main

eﬀects or interactions. The interaction between reading and Education Model approached signiﬁcance

(p = .08) such that children in the TWDL program showed less growth in reading compared to their

peers in mainstream education.

Intermediate school

Intermediate school academic achievement was represented with both standardized test scores (EOG)

in mathematics and language arts and classroom grades in mathematics, language arts, science, and

social studies. Both 4

and 5

grade students completed these measures and the measures are normed

within grade, meaning we would not expect to see growth across grade levels. We analyzed the

standardized test scores in a MANOVA with Education Model as a predictor. Education Model was

signiﬁcant, F(2, 105) = 4.53, p= .01, ŋ

= .08. Univariate analyses revealed that students in the TWDL

program had signiﬁcantly higher performance on the standardized math score, F(2, 108) = 9.13, p=

.003, ŋ

= .08, and the standardized language arts score neared signiﬁcance in the same direction, F(2,

BILINGUAL RESEARCH JOURNAL 9

108) = 3.58, p= .06, ŋ

= .03. A MANOVA predicting classroom grades in math, language arts, science,

and social studies with educational program as the predictor revealed similar results. Educational

program was signiﬁcant, F(4, 102) = 9.26, p < .001, ŋ

= .27, and univariate analyses revealed higher

performance for TWDL students in all subjects, Fs(1, 107) ≥362.25, ps < .001, ŋ

s ≥ .15.

Primary school CEM

We used the same plan of analyses to examine academic performance within the CEM sample.

A MANOVA examining kindergarten reading and math performance replicated the full sample in

that there was not a signiﬁcant diﬀerence by Education Model, F(2, 75) = 1.70, p = .19. The repeated

measures MANOVA examining kindergarten and ﬁrst-grade reading and math performance pre-

dicted by Education Model also replicated the ﬁndings in the full sample. There was a main eﬀect of

math such that math scores decreased between kindergarten and ﬁrst grade, F(1, 28) = 2891.51, p <

.001, ŋ

= .99, but no other signiﬁcant eﬀects.

Intermediate school CEM

In the MANOVA examining intermediate student standardized math and language arts performance,

Education Model was not a signiﬁcant predictor of performance, F(2, 47) = 1.93, p= .12. The

MANOVA predicting classroom grades in math, language arts, science, and social studies with

Educational Program as the predictor replicated results with the full sample. Educational program

was signiﬁcant, F(4, 45) = 6.11, p = .001, ŋ

= .35, and univariate analyses revealed higher performance

for TWDL students in all subject areas, Fs(1, 50) ≥8.57, ps ≤ .005, ŋ

s ≥ .15.

In summation, there were no signiﬁcant diﬀerences between educational programs in performance

for primary students in either the full or CEM samples. Intermediate students in the TWDL program

signiﬁcantly outperformed their mainstream-educated peers on the standardized math exam and on

classroom grades in all subjects in the full sample. In the CEM sample, students in TWDL education

had signiﬁcantly higher scores in classroom grades.

Executive functions

Executive functions were measures both with computerized tasks as well as a teacher reported

inventory. Scores on executive functions tasks are shown in Table 3. Preliminary analyses indicated

that the executive functions tasks (TMT, BST, Simon, BRIEF) were measuring unique attributes of

executive functions as intended, rs < .46.

Therefore, we ran separate analyses for the outcome

variables of each measure: TMT (Trail A, Trail B, Trail C), BST (neutral, congruent, incongruent),

and Simon (congruent and incongruent), and BRIEF. In all four,

we utilized 2 (Education Model) x 2

(School) MANOVAs or an ANOVA (for BRIEF) to test for main eﬀects of Education Model and

School as well as an interaction. To ensure that age did not have an undue inﬂuence on the models, we

calculated Z-scores (M = 0, SD = 1) for each of the computerized executive function measures within

each grade. The conversion to Z-scores allowed us to examine relative ranking within each grade. The

BRIEF is measures based on grade-level expectations and did not require Z-score conversion.

Full sample

We ﬁrst analyzed the data from the full sample. There were no signiﬁcant main eﬀects or interactions

for the TMT, Fs(3, 237) ≤2.45, ps ≥ .06, or the BST, Fs(3, 233) ≤2.36, ps ≥ .07. There were no signiﬁcant

main eﬀects on the Simon task, Fs(2, 236) ≤1.40, ps ≥ .25, but there was a signiﬁcant interaction, F(2,

236) = 3.48, p = .03, ŋ

= .03. Univariate analyses revealed that intermediate TWDL students were

signiﬁcantly faster in the congruent trials compared to their mainstream educated peers, F(1, 241) =

6.90, p = .009, ŋ

= .03. Thus, though the means indicate the intermediate students were faster in the

computerized executive functions tasks, this diﬀerence only reached signiﬁcance on one part of one

task.

10 A. G. ESPOSITO

In a 2 (Education Model) x 2 (School) ANOVA predicting the Global Executive Composite score,

there was a main eﬀect of Education Model such that children in the TWDL program exhibited fewer

disordered behaviors related to executive functions, F(1, 259) = 15.44, p < .001, ŋ

= .06. There was not

a main eﬀect of School, F(1, 259) = 1.10, p = .30, and no interaction, F(1, 259) = 0.48, p = .49.

CEM sample

We next analyzed the data from the CEM sample. Replicating the full sample, there were no signiﬁcant

main eﬀects or interactions for the TMT, Fs(3, 116) ≤0.96, ps ≥ .04. In contrast, there was a signiﬁcant

main eﬀect of Education Model in the BST, F(3, 113) = 2.79, p = .04, ŋ

= .07, such that children in the

TWDL program were faster at responding to neutral trials, F(1, 115) = 6.37, p = .01, ŋ

= .05, and

trending in that direction for incongruent trials, F(1, 115) = 3.75, p = .05, ŋ

= .03, compared to their

mainstream educated peers. There were no other main eﬀects or interactions for the BST, Fs(3, 113)

≤1.55, ps ≥ .21. The Simon task analyses again replicated the results of the full sample in that there were

no signiﬁcant main eﬀects, Fs(2, 117) ≤0.86, ps ≥ .43, but there was a signiﬁcant interaction, F(2, 117) =

4.60, p = .01, ŋ

= .07. Univariate analyses revealed that intermediate TWDL students were signiﬁcantly

faster in the congruent trials compared to their mainstream peers, F(1, 122) = 7.57, p = .007, ŋ

= .06.

In a 2 (Education Model) x 2 (School) ANOVA predicting the Global Executive Composite score

on the BRIEF, the results replicated those of the full sample. There was a main eﬀect of Education

Model such that children in the TWDL program exhibited fewer disordered behaviors, F(1, 120) =

13.88, p < .001, ŋ

= .11. There was not a main eﬀect of School, F(1, 120) = 1.30, p = .26, and no

interaction, F(1, 120) = 0.16, p = .69.

In summary, the computerized executive function tasks did not reveal a robust pattern of results. The

BRIEF, in contrast, indicated a main eﬀect such that the students in the TWDL program were exhibiting

behaviors consistent with more developed executive functions compared to their peers in the main-

stream education model. This pattern of results was found in both the full sample and the CEM sample.

Table 3. Mean reaction times in milliseconds for executive functions tasks by school, trial type, and education model.

Primary Intermediate

Task Trial Type

Dual-Language Mainstream Dual-Language Mainstream

Full CEM Full CEM Full CEM Full CEM

BST

(SD)

Neutral 1.37 (0.22) 1.36 (0.23) 1.37 (0.21) 1.38 (0.24) 0.98 (0.12) 0.96 (0.12) 1.08 (0.14) 1.08 (0.14)

Congruent 1.40 (0.23) 1.41 (0.23) 1.35 (0.23) 1.36 (0.24) 0.98 (0.14) 0.97 (0.15) 1.03 (0.12) 1.03 (0.11)

Incongruent 1.43 (0.23) 1.42 (0.23) 1.43 (0.26) 1.48 (0.27) 1.04 (0.17) 1.03 (0.17) 1.10 (0.16) 1.11 (0.16)

TMT

(SD)

Trail A 27.17

(11.23)

27.96

(11.79)

23.75

(12.00)

26.93

(15.19)

11.41

(4.72)

10.80

(3.50)

13.58 (7.26) 12.46 (4.20)

Trail B 27.81

(16.04)

28.79

(16.98)

27.14

(15.23)

29.70

(15.01)

13.94

(6.51)

12.68

(3.66)

15.71

(11.66)

13.30 (3.78)

Trail C 52.08

(24.65)

53.35

(25.64)

53.76

(31.85)

58.63

(42.70)

24.98

(11.01)

25.27

(9.90)

31.42

(15.45)

27.93

(16.74)

Simon (SD)

Congruent 1.52 (0.44) 1.57 (0.45) 1.41 (0.30) 1.44 (0.34) 0.98 (0.11) 0.95 (0.11) 1.05 (0.14) 1.06 (0.13)

Incongruent 1.60 (0.42) 1.62 (0.44) 1.56 (0.34) 1.63 (0.40) 1.05 (0.13) 1.04 (0.12) 1.13 (.15) 1.11 (0.17)

BRIEF GEC

89.72

(21.75)

88.85

(19.05)

102.78

(31.12)

104.50

(31.63)

82.67

(17.87)

81.64

(18.92)

101.35

(31.87)

101.00

(27.09)

BST refers to Bivalent Shape Task. TMT refers to Trail Making Task. GEC refers to Global Executive Composite.

BILINGUAL RESEARCH JOURNAL 11

Mediation analyses

We next explored whether executive functions is a potential mechanism through which intermediate

TWDL students are showing an academic advantage over their mainstream educated peers. Speciﬁcally,

we examined whether the teacher reported BRIEF mediated the relation between being in the TWDL

program and academic performance on the standardized measure of math achievement. This analysis

was only conducted with the standardized math measure in the full sample for three reasons. First,

classroom teachers issue both classroom grades and the behavioral rating. Any relation between the

variables could be due to coming from the same source. Second, only math was assessed because the

standardized measure of language arts did not signiﬁcantly diﬀer between Education Models. Third, the

signiﬁcant diﬀerence in standardized math performance was only found in the full sample.

In this analysis, we followed the four steps outlined by Baron and Kenny (1986) to assess mediation

(see Table 4 for regression results). First, we examined whether Education Model signiﬁcantly

predicted executive functions. Second, we conducted a regression with Education Model predicting

standardized math performance. Third, we conducted a regression with executive functions predicting

math performance. Fourth, we examined whether the eﬀect of Education Model on math performance

was reduced when executive functions was included in the regression. Having determined mediation

with these steps, we next examined whether the mediation was statistically signiﬁcant with the Sobel

test (1982). The Sobel test indicated statistically signiﬁcant mediation, z= 2.47, p= .01.

Discussion

The present study investigated the academic achievement, executive functions, and the relation

between these variables in a cross-sectional design comparing students in primary or intermediate

elementary education in either a two-way dual-language program or a mainstream English education

program. The pattern of results supported an academic advantage for intermediate TWDL students.

The advantage in executive functions was less robust, emerging for TWDL students in behavioral

ratings but not in computerized measures. Using the behavioral rating measure of executive functions

and a standardized measure of math performance, we did ﬁnd evidence for executive functions as

a mechanism supporting the academic advantage.

We predicted an academic advantage for intermediate students in the TWDL program, who had

experienced the program for 5–6 years. As predicted, no diﬀerences were found in academic perfor-

mance for primary students. Intermediate TWDL students showed an advantage in both standardized

measures of achievement and classroom grades in the full-sample. The match-sample showed a similar

pattern, but the standardized measures failed to reach signiﬁcance. Classroom grades had a greater

range of performance compared to the narrow range of the standardized measures. This limited

variability could contribute to the diﬀerences in the results between the full sample that included

a larger sample and did not reduce participants on the extreme ends of the distribution (in comparison

to the matched sample).

Table 4. Regression models analyzing the potential of executive functions as a mediator for education model and math achievement.

Model n t p β F df p adj. R

Step 1; DV = Executive Functions 101 9.81 1, 99 .002 .08

Educational Program −3.13 .002 −.30

Step 2; DV = Math performance 101 8.33 1, 99 .005 .08

Educational Program 2.89 .005 .28

Step 3; DV = Math performance 101 14.83 1, 99 < .001 .12

Executive Functions −3.85 < .001 −.36

Step 4; DV = Math performance 101 9.48 2, 98 < .001 .15

Educational Program 1.93 .06 .19

Executive Functions −3.14 .002 −.31

12 A. G. ESPOSITO

While there were few diﬀerences between those in TWDL and mainstream education in the

computerized measures of executive functions, the behavioral rating measure revealed a signiﬁcant

diﬀerence between education models such that children in the TWDL program exhibited fewer

indicators of executive dysfunction in the classroom. The diﬀerence was present at both the primary

and the intermediate level. This could indicate a preexisting group diﬀerence such that children in the

TWDL program begin school with greater executive functions skills. The results could also indicate

that teachers in the TWDL program have diﬀerent expectations of students in regards to classroom

behavior and, thus, rate them more favorably. There could also be other group diﬀerences aside from

education program that our variables failed to capture. Alternatively, it could indicate that an

advantage in executive functions related to classroom behavior emerges early after only one or two

years of participation in a TWDL program. There are at least two indicators of the latter interpretation.

First, teachers in the TWDL program and the mainstream program are within the same community

with the same training opportunities and regularly rotate through teaching in the TWDL program.

Thus, diﬀerences in expected behaviors to the degree found in this study are unlikely. Second, the

diﬀerences in the scores predicted diﬀerences in academic achievement on a state standardized test.

This last point indicates that the diﬀerences in performance were not a systematic bias held by teachers

and were instead meaningful reﬂections of behavior predicting academic outcomes.

We analyzed whether executive functions is a potential mechanism for the academic advantage

often found for children in dual-language education models. The analysis was limited to intermediate

students who had a standardized measure of academic performance not issued by a teacher. This

eliminated the concern that both measures of interest were coming from the same source or any

concern regarding diﬀerent expectations across programs. The results did indicate that the academic

advantage found on the standardized math assessment for children at the intermediate level of the

TWDL program was mediated by executive functions behaviors exhibited in the classroom. Although

the analyses were limited to a subset of the participants and variables, the results support further

investigation of cognitive advantages emerging through TWDL education that could positively

inﬂuence academic achievement. Bilingual education is generally focused on bilingualism, biliteracy,

and cultural competence, but the present results are the ﬁrst evidence we are aware of that children in

a bilingual education program have advanced executive skills that positively impact their academic

achievement. Further research is needed to determine whether executive functions are developing

diﬀerently over time based on educational program placement, but these results are encouraging.

The executive functions advantage for TWDL students was present across all grade levels, yet the

academic advantage was only present in the intermediate students. If we assume that the executive functions

advantage is not the results of preexisting group diﬀerences, the results indicate that, in comparison, it takes

longer for the academic advantage to emerge. This is possibly because executive functions behaviors

supporting the academic advantage do not have an immediate eﬀect. It could also indicate that

a language threshold needs to be met in both languages before the academic advantage emerges. Future

research that includes full language assessments could help to understand the relation between language

development, academic achievement, and executive functions in dual-language education models.

An interesting ﬁnding was the diﬀerence between the results of the computerized executive

functions measures and the behavioral ratings. Studies of executive functions, including those used

to identify the bilingual advantage and the contexts in which it emerges, rely heavily on computerized

measures (for a review, see Bialystok et al., 2012). The computerized measures, in theory, are

measuring abilities utilized in the classroom. However, performance on computerized tasks has not

shown good transfer to how children are actually behaving in classrooms. Children who are trained on

a computerized tasks in laboratories show improvements on that task and sometimes similar compu-

terized tasks (near transfer), but rarely show beneﬁts in areas such as academic performance (con-

sidered far transfer; Kassai, Futo, Demetrovics, & Takacs, 2019; Serpell & Esposito, 2016). The results

of the present study indicate that advantages in executive functions gained through bilingual educa-

tion may be more readily detected with an applied measure relevant to classroom behavior.

BILINGUAL RESEARCH JOURNAL 13

A recurring criticism of work identifying the bilingual advantage in executive functions is that it

may be based on diﬀerences in SES (see Valian, 2015, for review). A similar concern has arisen

regarding the academic advantage for children in dual-language education. The explanation is that

lower SES students, who are more likely to be transient, most commonly attrit out of the program,

leaving higher SES students in the later elementary years where the academic advantage is most often

found (e.g., Hill, 2018). As a way of addressing these concerns, we analyzed the data twice. First, we

included all consenting participants who completed the tasks of interests. In this full sample, we

examined group diﬀerences in participants’ intelligence, parent education level, and other indicators,

but we did not control for these measures. We then analyzed a subset of the data that was matched on

grade level (exact) as well as verbal and non-verbal intelligence and parent education level (coarsened

exact matching). The intention was to identify whether group diﬀerences found in the full sample

would still be found when controlling for individual and family-level variables known to aﬀect

academic outcomes and executive functions. Although there were some small changes in the results

regarding which analyses reached signiﬁcance, the overall pattern between the two samples was

similar. The results do not support a stance that advantages in academic achievement or executive

functions are the result of group diﬀerences in intelligence or SES, nor do they support attrition as an

explanation for later emerging academic advantages.

The present study is not without limitations. The study is a cross-sectional and quasi-experimental

investigation that captures a snapshot of the student performance. A longitudinal investigation that

included random assignment to educational program and measurement prior to school start would be

better able to capture the developmental changes attributed to the educational model. However, the

school system allows parents to opt-in, eliminating random assignment as a possibility. We were also

limited in time with each child by the participating school system and, thus, were unable to assess

language proﬁciency. As future directions, we recommend longitudinal work that will be able to assess

not only a snapshot of current performance, but the change over time as cross-sectional research

cannot identify preexisting group diﬀerences. The current work is also focused speciﬁcally on an

English-only model (teachers do not have proﬁciency in other languages) compared to a dual-

language model. Additionally, we recommend assessments of language proﬁciency in both the partner

languages as well as translanguaging practices (García & Lin, 2016) to identify the role of language

development in both academic performance and executive functions for both education models.

In conclusion, the present study found evidence for executive functions as a mechanism contributing

to the academic advantage shown by TWDL intermediate students compared to their mainstream

English education peers. The present study indicates that in addition to gaining second-language

ﬂuency, literacy, and greater cultural competence, two-way dual-language education models contribute

to cognitive development, speciﬁcally executive functions, in ways that support academic achievement.

Notes

1. Analyses, including in CEM models, were run with and without a variable reﬂecting home language as reported

by parents. The variable did not change pattern of results. The more parsimonious model is, therefore, reported.

2. Analyses were run with and without a variable reﬂecting English proﬁciency at kindergarten entry. Results did

not diﬀer between models. Therefore, the more parsimonious model is reported.

3. Latent variable analyses was attempted, but the model did not converge.

4. Analyses were run with and without a variable reﬂecting English proﬁciency at kindergarten entry. Results did

not diﬀer between models. Therefore, the more parsimonious model is reported.

Funding

This work was supported by a postdoctoral/predoctoral fellowship provided by the National Institute of Child Health

and Human Development (T32-HD07376) through the Center for Developmental Science, University of North Carolina

at Chapel Hill, and by a subaward of IES R305A150492 to Alena G. Esposito.

14 A. G. ESPOSITO

References

Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable distinction in social psychological research:

Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51(6), 1173–1182.

doi:10.1037/0022-3514.51.6.1173

Best, J. R., Miller, P. H., & Naglieri, J. A. (2011). Relations between executive function and academic achievement from

ages 5 to 17 in a large, representative national sample. Learning and Individual Dierences, 21(4), 327–336.

doi:10.1016/j.lindif.2011.01.007

Bialystok, E. (2010). Global–local and trail-making tasks by monolingual and bilingual children: Beyond inhibition.

Developmental Psychology, 46(1), 93. doi:10.1037/a0015466

Bialystok, E., & Barac, R. (2012). Emerging bilingualism: Dissociating advantages for metalinguistic awareness and

executive control. Cognition, 122(1), 67–73. doi:10.1016/j.cognition.2011.08.003

Bialystok, E., Craik, F. I., & Luk, G. (2012). Bilingualism: Consequences for mind and brain. Trends in Cognitive Sciences,

16(4), 240–250. doi:10.1016/j.tics.2012.03.001

Cevada, T., Conde, E., Marques, D., & Deslandes, A. C. (2019). Test-retest reliability of the simon task: A short version

proposal. Somatosensory & Motor Research, 36(4), 275–282. doi:10.1080/08990220.2019.1689114

Cobb, B., Vega, D., & Kronauge, C. (2006). Eﬀects of an elementary dual language immersion school program on junior

high school achievement. Middle Grades Research Journal, 1(1).

De Cat, C., Gusnanto, A., & Serratrice, L. (2018). Identifying a threshold for the executive function advantage in bilingual

children. Studies in Second Language Acquisition, 40(1), 119–151. doi:10.1017/S0272263116000486

Delis, D. C., Kaplan, E., & Krames, J. H. (2001). Trail making task. Delis Kaplan executive function system examiner

manual (pp. 39–54). San Antonio, Texas: The Psychological Corporation Harcourt Assessment Company.

Diamond, A. (2013). Executive functions. Annual Review of Psychology, 64(1), 135–168. doi:10.1146/annurev-psych

-113011-143750

Esposito, A. G., & Baker-Ward, L. (2013). Dual-language education for low-income children: Preliminary evidence of

beneﬁts for executive function. Bilingual Research Journal, 36(3), 295–310. doi:10.1080/15235882.2013.837848

Esposito, A. G., Baker-Ward, L., & Mueller, S. T. (2013). Interference suppression vs. response inhibition: An explana-

tion for the absence of a bilingual advantage in preschoolers’ Stroop task performance. Cognitive Development, 28(4),

354–363. doi:10.1016/j.cogdev.2013.09.002

Esposito, A. G., & Bauer, P. J. (2018). Building a knowledge base: Predicting self-derivation through integration in

6-to10-year-olds. Journal of Experimental Child Psychology, 176, 55–72. doi:10.1016/j.jecp.2018.07.011

García, E. E., & Náñez Sr, J. E. (2011). Bilingualism and cognition. American Psychological Association.

García, O., & Lin, A. M. (2016). Translanguaging in bilingual education. In O. García, A. Lin, & S. May (Eds.), Bilingual

and multilingual education (pp. 117–130), Encyclopedia of Language and Education 5. New York: Springer.

Gioia, G. A., Isquith, P. K., Guy, S. C., & Kenworthy, L. (2015). BRIEF: Behavior rating inventory of executive function.

Lutz, FL: Psychological Assessment Resources.

Hartanto, A., Toh, W. X., & Yang, H. (2018). Bilingualism narrows socioeconomic disparities in executive functions and

self-regulatory behaviors during early childhood: Evidence from the early childhood longitudinal study. Child

Development, 90(4), 1215–1235. doi:10.1111/cdev.13032

Hill, S. R. (2018). A meta-analysis of the eﬀect of ten language immersion programs on academic achievement. Spanish

and Portuguese Review, 4, 23–50.

Hoﬀ, E. (2013). Interpreting the early language trajectories of children from low-SES and language minority homes:

Implications for closing achievement gaps. Developmental Psychology, 49(1), 4. doi:10.1037/a0027238

Howard, E. R., Lindholm-Leary, K. J., Rogers, D., Olague, N., Medina, J., Kennedy, B., . . . Christian, D. (2018). Guiding

principles for dual language education (3rd ed.). Washington, DC: Center for Applied Linguistics.

Iacus, S. M., King, G., & Porro, G. (2012). Causal inference without balance checking: Coarsened exact matching.

Political Analysis, 20(1), 1–24. doi:10.1093/pan/mpr013

Janus, M., Lee, Y., Moreno, S., & Bialystok, E. (2016). Eﬀects of short-term music and second-language training on

executive control. Journal of Experimental Child Psychology, 144, 84–97.

Kassai, R., Futo, J., Demetrovics, Z., & Takacs, Z. K. (2019). A meta-analysis of the experimental evidence on the near-

and far-transfer eﬀects among children’s executive function skills. Psychological Bulletin, 145(2), 165. doi:10.1037/

bul0000180

Kaushanskaya, M., Gross, M., & Buac, M. (2014). Eﬀects of classroom bilingualism on task-shifting, verbal memory, and

word learning in children. Developmental Science, 17(4), 564–583. doi:10.1111/desc.12142

Lindholm-Leary, K., & Genesee, F. (2014). Student outcomes in one-way, two-way, and indigenous language

immersion education. Journal of Immersion and Content-Based Language Education, 2(2), 165–180. doi:10.1075/

jicb.2.2.01lin

Lu, C. H., & Proctor, R. W. (1995). The inﬂuence of irrelevant location information on performance: A review of the

Simon and spatial Stroop eﬀects. Psychonomic Bulletin & Review, 2(2), 174–207. doi:10.3758/BF03210959

Marian, V., Shook, A., & Schroeder, S. R. (2013). Bilingual two-way immersion programs beneﬁt academic achievement.

Bilingual Research Journal, 36(2), 167–186. doi:10.1080/15235882.2013.818075

BILINGUAL RESEARCH JOURNAL 15

Martin-Rhee, M., & Bialystok, E. (2008). The development of two types of inhibitory control in monolingual and

bilingual children. Bilingualism: Language and Cognition, 11(1), 81–93. doi:10.1017/S1366728907003227

Mueller, S. T. (2011). The psychology experiment building language (Version 0.11) [Computer software]. Retrieved from

http://pebl.sourceforge.net

Mueller, S. T., & Esposito, A. G. (2014). Computerized testing software for assessing interference suppression in children

and adults: The Bivalent Shape Task (BST). Journal of Open Research Software, 2(1), e3.

Mullane, J. C., Corkum, P. V., Klein, R. M., & McLaughlin, E. (2009). Interference control in children with and without

ADHD: A systematic review of Flanker and Simon task performance. Child Neuropsychology, 15(4), 321–342.

doi:10.1080/09297040802348028

Nesbitt, K. T., Baker-Ward, L., & Willoughby, M. T. (2013). Executive function mediates socio-economic and racial

diﬀerences in early academic achievement. Early Childhood Research Quarterly, 28(4), 774–783. doi:10.1016/j.

ecresq.2013.07.005

Nicolay, A. C., & Poncelet, M. (2013). Cognitive advantage in children enrolled in a second-language immersion

elementary school program for three years. Bilingualism: Language and Cognition, 16(3), 597–607. doi:10.1017/

S1366728912000375

Nicolay, A. C., & Poncelet, M. (2015). Cognitive beneﬁts in children enrolled in an early bilingual immersion school:

A follow up study. Bilingualism: Language and Cognition, 18(4), 789–795. doi:10.1017/S1366728914000868

Padilla, A. M., Fan, L., Xu, X., & Silva, D. (2013). A Mandarin/English two-way immersion program: Language

proﬁciency and academic achievement. Foreign Language Annals, 46(4), 661–679. doi:10.1111/ﬂan.12060

Piper, B. J., Li, V., Eiwaz, M. A., Kobel, Y. V., Benice, T. S., Chu, A. M., . . . Mueller, S. T. (2012). Executive function on the

psychology experiment building language tests. Behavior Research Methods, 44(1), 110–123. doi:10.3758/s13428-011-

0096-6

Piper, B. J., Mueller, S. T., Geerken, A. R., Dixon, K. L., Kroliczak, G., Olsen, R. H., & Miller, J. K. (2015). Reliability and

validity of neurobehavioral function on the Psychology Experimental Building Language test battery in young adults.

PeerJ, 3, e1460. doi:10.7717/peerj.1460

Poarch, G. J., & van Hell, J. G. (2012). Executive functions and inhibitory control in multilingual children: Evidence from

second-language learners, bilinguals, and trilinguals. Journal of Experimental Child Psychology, 113(4), 535–551.

doi:10.1016/j.jecp.2012.06.013

Purić, D., Vuksanović, J., & Chondrogianni, V. (2017). Cognitive advantages of immersion education after 1 year: Eﬀects

of amount of exposure. Journal of Experimental Child Psychology, 159, 296–309. doi:10.1016/j.jecp.2017.02.011

Serpell, Z. N., & Esposito, A. G. (2016). Development of executive functions: Implications for educational policy and

practice. Policy Insights from the Behavioral and Brain Sciences, 3(2), 203–210. doi:10.1177/2372732216654718

Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental and quasi-experimental designs for generalized

causal inference/William R. Shedish, Thomas D. Cook, Donald T. Campbell. Boston: Houghton Miﬄin.

Simon, J. R., & Wolf, J. D. (1963). Choice reaction times as a function of angular stimulus-response correspondence and

age. Ergonomics, 6(1), 99–105. doi:10.1080/00140136308930679

Sobel, M. E. (1982). Asymptotic intervals for indirect eﬀects in structural equations models. In S. Leinhart (Ed.),

Sociological Methodology (pp.290–312). San Francisco: Jossey-Bass.

Steele, J. L., Slater, R. O., Zamarro, G., Miller, T., Li, J., Burkhauser, S., & Bacon, M. (2017). Eﬀects of dual-language

immersion programs on student achievement: Evidence from lottery data. American Educational Research Journal, 54

(1_suppl), 282S–306S. doi:10.3102/0002831216634463

Valian, V. (2015). Bilingualism and cognition. Bilingualism: Language and Cognition, 18(1), 3–24. doi:10.1017/

S1366728914000522

Watzinger-Tharp, J., Swenson, K., & Mayne, Z. (2018). Academic achievement of students in dual language immersion.

International Journal of Bilingual Education and Bilingualism, 21(8), 913–928.

Weschler, D. (1974). Weschler intelligence scale for children-revised. New York, NY: Psychological Corporation.

Weschler, D. (1999). Weschler Abbreviated Scale of Intelligence (WASI). San Antonio: Psychological Corporation.

Yang, H., Hartanto, A., & Yang, S. (2016). The importance of bilingual experience in assessing bilingual advantages in

executive functions. Cortex, 75, 237.

16 A. G. ESPOSITO