A Scoping Review of Bilingualism, Socioeconomic Status and Executive Function Development in Children and Adolescents: Implications for Educational Practice and Research

Abstract

Socioeconomic status (SES) strongly predicts children’s executive-function (EF) performance and academic achievement. Although findings remain inconsistent, bilingualism has been proposed as a potential moderator of SES-related differences. This scoping review systematically maps evidence on bilingualism, SES, and the three core EFs (inhibition, working memory, and cognitive flexibility) in children and adolescents. Following PRISMA-ScR guidelines, five databases were searched without publication-date restrictions on 9 February 2026, with a supplementary search on 17 September 2026. Eligible studies were peer-reviewed empirical articles examining bilingual participants aged 17 years or younger, including a monolingual comparison group, interpretable measures of SES and at least one core EF. Data were charted using a review-specific form by one author and checked by another. Thirty-nine studies published between 2007 and 2026 were included. Findings were highly heterogeneous. Inhibition was the most frequently examined EF and yielded comparatively more reports of bilingual advantages than the other domains. Working-memory findings were predominantly null, while cognitive flexibility was less extensively investigated and produced mostly null or mixed evidence. Rather than supporting a universal bilingual advantage, the findings suggest that EF outcomes depend on bilingual experience, SES, developmental stage, and task variability. Longitudinal research using multidimensional measures of bilingualism and SES is needed to clarify the possible role of bilingualism in reducing educational inequalities.

1. Introduction

Central to the cognitive development of a child are a set of cognitive abilities known as executive functions (EFs), the three core EFs being inhibition, working memory and cognitive flexibility (Miyake et al., 2000). These cognitive abilities are fundamental for everyday functioning (Diamond, 2013) and have awakened considerable attention in the fields of developmental psychology and educational research due to their central role in children’s and adolescents’ academic performance and achievement.

EFs develop gradually over an extended period spanning from early childhood to young adulthood, with the early years of life constituting an especially sensitive period. Among the many factors that have been seen to influence this development, socioeconomic status (SES) stands out as one of the strongest predictors of EF development. Children from low SES backgrounds frequently show greater academic disadvantage, contributing to what is widely known as the ‘achievement gap’ (Bradley & Corwyn, 2002; Noble et al., 2005).

Given the strong association between SES and EF development, increasing attention has been given to identifying factors that may reduce SES-related differences in cognitive development, among them bilingualism. Researchers suggest that the continuous management of two or more languages in bilingualism places high demands on regions of domain-general cognitive control processes, leading to enhanced executive functioning (Bialystok, 2017; Green, 1998). However, although promising, evidence for a bilingual advantage remains mixed: while numerous studies have reported advantages in inhibition, working memory and cognitive flexibility (e.g., Bialystok et al., 2004; Cox et al., 2016), others have failed to replicate these findings (e.g., Duñabeitia et al., 2014; Paap & Greenberg, 2013; Paap et al., 2015) resulting in an ongoing debate regarding the existence, extent and conditions under which bilingual advantages may emerge (see Antoniou, 2019).

A number of explanations for the inconsistency in these findings have been proposed. Studies show considerable methodological heterogeneity and differ substantially in their operationalisation of bilingualism, SES, participant characteristics, EF measures, and analytical approaches, making comparisons difficult. In particular, SES has often been treated exclusively as a control variable rather than as a factor that may meaningfully interact with the bilingual experience. This shortcoming was pointed out by Morton and Harper (2007), who argued that many reported bilingual advantages were a result of uncontrolled SES differences rather than caused by bilingualism itself. Carlson and Meltzoff (2008), in turn, demonstrated the importance of considering SES as a key developmental factor alongside bilingualism, marking a shift in how the relationship between these variables was viewed and, consequently, conceptualised. Since then, increasing attention has been devoted to understanding not only the independent effects of SES and bilingualism, but also how they interact to shape EF development, still yielding, however, mixed findings.

More recently researchers have begun to move beyond treating both SES and bilingualism as simple categorical variables. Instead, SES is recognised as a multidimensional construct including educational, occupational, financial, and environmental factors. At the same time, bilingualism is increasingly viewed as a multidimensional experience characterised by differences in language proficiency, exposure, language use, and interactional context. Furthermore, growing attention has been devoted to understanding how these two experiences interact throughout development as opposed to examining only their independent effects.

Previous meta-analyses have studied bilingual–monolingual differences in executive functioning and have considered SES to varying degrees (Donnelly et al., 2019; Gunnerud et al., 2020; Lehtonen et al., 2018; Lowe et al., 2021). However, their aims and scope differ from those of the present review. Lehtonen et al. (2018) focused on particular populations or EF outcomes, such as executive functioning in adults, while Donnelly et al. (2019) examined interference control across age groups. Others considered SES principally as a study-level moderator or methodological control. Gunnerud et al. (2020), for example, compared bilingual–monolingual effect sizes across broad SES categories and found a small bilingual advantage among middle-SES samples, but not among low- or upper-middle-SES samples. However, the small number of studies representing the latter groups limited the conclusions that could be drawn. On the other hand, Lowe et al. (2021) found that the apparent effects favouring bilingual children disappeared when only studies that had objectively measured and controlled for SES were considered. Even though these meta-analyses demonstrate that SES can influence conclusions about bilingual–monolingual EF differences, they do not specifically synthesise the evidence concerning the relationship between bilingualism and the three core EFs in children and adolescents across different socioeconomic contexts. Addressing this gap is the central aim of the present scoping review.

The present review therefore aims to map the available evidence concerning the relationship between bilingualism, SES, and the development of inhibition, working memory, and cognitive flexibility during childhood and adolescence, with particular attention to children from socioeconomically disadvantaged backgrounds. It explores whether and under what conditions bilingualism may be associated with stronger EF performance or may reduce SES-related differences. It also identifies the developmental and educational contexts represented in the literature, alongside the main methodological limitations and areas in need of further investigation.

2. Methodology

2.1. Methodological Framework

This scoping review was conducted following the methodological framework proposed by Arksey and O’Malley (2005) and subsequently refined by Levac et al. (2010), which suggests the following approach: (i) identifying the research question; (ii) identifying relevant studies; (iii) the selection of the studies; (iv) data extraction and organisation; (v) synthesis and reporting of the findings. To ensure methodological transparency, reproducibility, and comprehensive reporting, the review also followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) guidelines across all stages of the review process. No review protocol was registered for this scoping review.

This scoping review addresses the following research question:

What evidence does the literature provide about the potential relation between bilingualism and the development of the three core executive functions (inhibition, working memory, and cognitive flexibility) across different socioeconomic contexts in children and adolescents?

2.1.1. Search Strategy

A comprehensive systematic search was conducted on 9 February 2026 using the following five databases: PsycINFO, ERIC, Scopus, Web of Science, and MLA International Bibliography. The search strategy was designed to identify studies examining the relationships between bilingualism, SES, and executive functions in children and adolescents. Boolean operators were used to combine three key concepts: (i) the target population (children and adolescents); (ii) bilingualism and SES; and (iii) executive functions. To do so, the following search string was applied:

(child OR “school-aged” OR adolescen*) AND (socioeconomic OR SES) AND (bilingual* OR multilingual* OR “second language”) AND (“executive function*” OR inhibition OR “working memory” OR “cognitive flexibility”).

This search strategy was adapted to the indexing structure and functionality of each database. In Web of Science and Scopus, Topic and TITLE-ABS-KEY fields were used, respectively, to search for titles, abstracts, and keywords. In PsycINFO, searches were conducted in the title and abstract fields. Due to differences in indexing and search functionality, broader All Fields searches were conducted in ERIC and MLA International Bibliography to maximise the retrieval of potentially relevant studies.

2.1.2. Study Selection and Eligibility Criteria

All retrieved records were imported into the Mendeley reference management software to facilitate the identification and removal of any duplicate records after the database search. A study selection was then conducted using the following criteria: (i) included children or adolescents; (ii) investigated bilingualism or multilingualism, socioeconomic status (SES), and at least one of the three core executive functions (inhibition, working memory, or cognitive flexibility); (iii) reported original empirical findings; and (iv) was published in peer-reviewed journals.

Studies were excluded if they had the following characteristics: (i) were not published in peer-reviewed journals; (ii) non-empirical publications; (iii) involved clinical populations, including individuals with psychiatric disorders, neurodevelopmental disorders, or sensory disabilities; (iv) did not measure or include as a key variable either bilingualism, any of the three core EFs or the variable of SES; (v) explored research questions unrelated to the interaction between bilingualism, SES and EF development; (vi) included participants over the age of 17, unless findings for participants aged 17 years or younger could be extracted separately; (vii) did not include a monolingual control group; (viii) were written in a language other than English. The upper age limit of 17 was selected to preserve the review’s developmental focus on childhood and adolescence and to exclude young adults, particularly university student samples, which represent a distinct developmental and educational population. Although continuous within-bilingual designs provide valuable evidence about how variation in bilingual experience relates to executive functioning (Rothman et al., 2023; Zingaretti et al., 2026), this does not mean that bilingual–monolingual comparisons are uninformative in every research context. Indeed, Rothman et al. (2023) acknowledge that monolingual comparisons may be appropriate when their inclusion is explicitly justified by the research question. In the present review, requiring a monolingual comparison group provided a consistent framework for examining whether bilingual children performed differently from monolingual children across socioeconomic contexts and for classifying the published findings as advantage, mixed, or null. Monolingual participants were therefore treated as a theoretically relevant comparison group, not as a normative standard or true experimental control. This criterion was retained because it ensured comparability across the evidence synthesised, although its implications for the scope of the review are acknowledged in the limitations.

A total of 414 records were identified by the database searches. Due to export limitations, 412 records were successfully imported into Mendeley. Following the removal of 179 duplicate records and one corrigendum, 232 records remained for screening. An additional four records were identified through citation searching, resulting in a total of 418 records identified and 236 records entering the screening process, which is detailed in the PRISMA-ScR flowchart in Figure 1 (Page et al., 2021).

Figure 1. PRISMA 2020 flow diagram of the study selection process.

2.1.3. Screening Process and Data Extraction

The selection process consisted of title and abstract screening, full-text eligibility assessment, and data extraction. Full-text eligibility assessment was completed in two phases, with the second phase serving as a final eligibility verification during preliminary data charting. All stages were conducted using the predefined inclusion and exclusion criteria.

During the title and abstract screening stage, the titles and abstracts of the 236 retrieved records were assessed for eligibility. A total of 104 records were excluded, while 132 articles progressed to full-text assessment. One report could not be retrieved, resulting in 131 full-text articles being assessed for eligibility. Title and abstract screening was conducted independently by the two authors. Each author assessed the same records against the predefined criteria and recorded an include, exclude, or uncertain decision in a Microsoft Excel spreadsheet. Following screening, the authors compared their decisions. Discrepancies and uncertain cases were discussed with reference to the eligibility criteria until consensus was reached. A third reviewer was not required because all disagreements were resolved through discussion. Although the screening decisions were compared, no formal inter-reviewer agreement statistics were calculated.

During the full-text screening stage, the 131 retrieved articles were evaluated in greater detail against the eligibility criteria. The same two authors assessed the full texts independently, recording an eligibility decision and, where applicable, the principal reason for exclusion. Their decisions were compared after screening, and discrepancies were resolved through discussion and re-examination of the full article until consensus was reached. At this stage, 72 studies were excluded, leaving 59 articles for further assessment. Reports that clearly failed to meet the criteria were excluded, whereas reports for which eligibility could not yet be determined confidently were retained for closer examination rather than excluded prematurely. The main reason for exclusion was that SES, bilingualism, or executive functions were not examined as key variables. All exclusion criteria used at this stage can be seen in the PRISMA-ScR flow diagram (Figure 1). The main reason for excluding each report was documented in the screening spreadsheet.

The 59 reports retained after the initial full-text assessment then underwent preliminary data charting and a second, more detailed eligibility verification. The authors examined the study aims, variables, measures, analyses, and discussion to determine whether bilingualism was a main variable, whether the outcomes represented one or more of the three core executive functions, whether SES provided interpretable evidence concerning executive functioning, and whether the study included a monolingual comparison group. The eligibility criteria did not change between the two phases; the second phase involved applying the same criteria using the more detailed information obtained during preliminary charting. This assessment resulted in the exclusion of 28 further reports: 19 were not aligned with the review objective and nine did not include a monolingual comparison group. These decisions were based on study characteristics and not on the direction or statistical significance of the findings. The final sample therefore comprised 31 studies. In total, 205 records or reports were removed from the review pathway, including the report that could not be retrieved. The complete selection process is presented in the PRISMA flow diagram (Figure 1).

The data-charting form was developed specifically for this review from the research question and eligibility criteria. It was not formally piloted on a predefined subset of studies; instead, it was refined iteratively during preliminary charting to ensure that all relevant information was captured consistently. After the final sample had been confirmed, data from the 31 included studies were fully extracted and charted by one author, and the recorded information was reviewed by the other author. The form included study characteristics (e.g., authors, publication year, country, study design and participant characteristics), alongside the main characteristics and metrics of the three key variables (SES, EFs and language measures), which are summarised in Table 1. The information regarding EF tasks used in the studies and the results reported was also charted and is presented in Table 3.

Table 1. Characteristics of the 39 studies included in the scoping review.

Author(s) & Year	Participants		Country/Region	Study Design	Educational Setting	Number of SES Groups	SES Indicators	Statistical Role of SES	Formal Bilingualism × SES Interaction Tested	Linguistic Profile	Language Metrics
	Sample (n=)	Age Range (Mean)
Abdurokhmonova et al. (2026)	80 (Behavioural sample) 44 (fMRI subsample)	4–7 yrs (M = 5.82 yrs)	USA (Boston metropolitan area)	Cross-sectional	Lower-SES charter schools & general-public/community recruitment in metropolitan Boston	Continuous income-to-needs ratio (ITN); classified into 5 descriptive U.S. Department of Housing and Urban Development (HUD) income categories	Federal ITN and average parent education years	Independent Predictor & Moderator	Yes	2 groups: Monolingual English (n = 51) vs. Bilingual (n = 29; Spanish n = 21, Haitian Creole, Cape Verdean Creole, Mandarin, Greek, Russian, German, Catalan)	LSBQ parent questionnaire (5-point scale) and daylong LENA audio exposure recordings
Bablekou et al. (2025)	145	8.4–12.10 yrs (M = 10; 4)	Greece (Attica and Thessaloniki)	Cross-sectional	Greek state schools and Albanian community schools on weekends	3 groups: low, middle, and high SES (92.5% of bilinguals were low SES)	Composite score: parental education (0–6 scale), occupation type, and occupational status (0–6 scale)	Control (Covariate)	No	2 groups: monolingual Greek vs. Albanian-Greek bilingual (natural/heritage acquisition at home)	Questionnaires for background; Renfrew Word Finding Vocabulary Test for proficiency; Greek vocabulary
Bialystok and Shorbagi (2021)	234	~6.5 yrs	Canada (Toronto)	Cross-sectional	Public Schools	Two groups: High vs. Middle SES via median split	Parental Education (5-point scale)	Categorical Grouping Factor & Independent Predictor & Moderator	Yes	2 groups: “more monolingual” vs. “more bilingual” (41 heritage languages; natural acquisition)	Language and Social Background Questionnaire (LSBQ) factor score
Blom et al. (2014)	120	5–6 yrs	The Netherlands	Longitudinal	Public Schools, Preschools, & Daycares	Two groups: Bilinguals (Low SES) and Monolinguals (High SES)	Mean parental educational level	Control (Covariate)	No	2 groups: Dutch monolingual vs. Turkish-Dutch Bilingual (natural sequential acquisition; home/community)	Dutch and Turkish receptive vocabulary (Toets Tweetaligheid/Taaltoets Alle Kinderen (TAK))
Bosman and Janssen (2017)	86 (38 Turkish–Dutch, 48 native-Dutch)	~7 yrs (Turkish–Dutch M = 7; 4, native-Dutch M = 7; 2)	The Netherlands	Cross-sectional	Primary schools (1st grade classrooms in inner-city low-income neighbourhoods)	Uniformly Low-income	Contextual low-SES classification based on neighbourhood recruitment; no individual SES measure	Contextual/Sample selection criteria	No	2 groups: Native-Dutch monolinguals vs. Turkish–Dutch bilinguals (Turkish L1; 80% exposed to some Dutch before age 3)	Productive Vocabulary and Sentence Imitation tests in Dutch and Turkish; Reynell Language Comprehension Test
Buac and Kaushanskaya (2014)	85 (Monolingual n = 46; Bilingual n = 39)	Monolinguals M = 8; 5/Bilinguals M = 8; 25 yrs	USA	Cross-sectional	English-speaking schools; 33% attending Spanish-English dual-immersion schools	Two groups: Study 1: Monolinguals (Middle-Upper SES) Study 2: Bilinguals (Low SES)	Primary caregiver level of education on an 8-point scale	Descriptive Variable	No	2 population studies: Monolingual English vs. Spanish–English Bilingual (immigrant families; natural heritage acquisition)	Grammaticality Judgment (GJ) task; Clinical Evaluation of Language Fundamentals–Fourth Edition (CELF-4) standardised language assessments
Buac et al. (2016)	82 (36 monolingual, 46 bilingual)	5–7 yrs	USA	Cross-sectional	Local Schools (Madison, WI): English-speaking schools; 48% Spanish–English dual immersion	Two groups: monolinguals (high SES), bilinguals (low SES)	Primary caregiver years of education.	Descriptive Variable	No	2 groups: Monolingual English vs. Spanish–English Bilingual (Spanish heritage home language; some instructional dual-immersion)	Standardised expressive/receptive vocabulary tests in English and Spanish (Peabody Picture Vocabulary Test–3rd Edition (PPVT-III), Woodcock-Johnson III Tests of Achievement (WJ-III), Batería III Woodcock–Muñoz Pruebas de aprovechamiento (TVIP, Batería III))
Calvo and Bialystok (2014)	175	6–7 yrs	Canada	Cross-sectional	Public Schools	Two groups: Working Class (WC) (Low) vs. Middle Class (MC)	Maternal/Paternal Education and Boyd’s Income Index (percentile ranking based on average occupational earnings)	Categorical grouping factor & Moderator	Yes	4 sub-groups (WC Monolingual, WC Bilingual, MC Monolingual, MC Bilingual). Monolingual English vs. Bilingual (26 heritage languages; natural acquisition)	Language and Social Background Questionnaire (LSBQ) parent questionnaire for language environment; Peabody Picture Vocabulary Test–3rd Edition (PPVT-III) for receptive vocabulary
Carlson and Meltzoff (2008)	50	4.83–6.9 (M = 6) yrs 1	USA	Cross-sectional	Public Kindergarten (Standard and L2 Immersion)	3 groups: Bilingual (Lower SES), Emerging bilinguals through participation in immersion programme (Higher SES); Monolinguals (Higher SES)	Maternal/Paternal education; Annual family income	Control (Covariate)	No	3 groups: Native Bilingual (Spanish–English; natural home acquisition) vs. Monolingual vs. L2 Immersion (Spanish/Japanese instructional)	Language Background Questionnaire; Expressive One-Word Picture Vocabulary Test (Spanish–English Bilingual Edition)
Choi et al. (2018)	825 (FACES 2009 4-year-old cohort)	Head Start entry: M = 4.49 yrs 1; followed for approximately 1.5 years to kindergarten	USA (Nationwide)	Longitudinal (3 waves: Head Start entry, end of Head Start year, Kindergarten)	Head Start preschools transitioning to kindergarten	Uniformly Low-income	Family Risk Index composite: maternal education (<high school), poverty status, and single-parent household	Control (Covariate)	No	3 groups: Spanish–English bilinguals (n = 210), dual language learners with limited English skills (DLLs-LES) (n = 140), and English Monolinguals (n = 475)	Parent-reported home language; Preschool Language Assessment Scales (PreLAS) 2000 Simon Says and Art Show English screeners; Expressive One-Word Picture Vocabulary Test/Expressive One-Word Picture Vocabulary Test—Spanish Bilingual Edition (EOWPVT/EOWPVT-SBE) expressive vocabulary
Clayden et al. (2023)	16,134 (Millennium Cohort Study analytic sample)	3 & 11 yrs	UK (England, Scotland and Northern Ireland)	Longitudinal (MCS)	Nationwide School & Home Contexts (MCS)	4 groups: Distributed by Maternal Education (None, GCSE, A Level, Higher) and equivalised household income	Maternal Education; OECD equivalised income	Independent Predictor & Moderator	Yes	2 groups: With MLE (Multiple Language Exposure) vs. Without MLE (no non-English exposure at home)	British Ability Scales Second Edition (BAS II) Naming Vocabulary (Age 3) and Verbal Similarities (Age 11); Bracken School Readiness Assessment (BSRA-R) (Age 3); Language spoken at home reports (MLE)
Czapka et al. (2020)	168	9 yrs	Germany	Cross-sectional	Inclusive Primary Schools	2 groups: Monolinguals (Higher SES) vs. Multilinguals (Lower SES)	Educational levels of mother and father according to ISCED	Pairwise Matching	No	2 groups: Monolingual German vs. Multilingual (50 bilingual, 7 trilingual); natural home acquisition	Parent questionnaire (AoA, proficiency, usage frequency); German lexicon size (expressive/receptive)
De Cat et al. (2018)	174	5–7 yrs	UK (North of England)	Cross-sectional	Primary Schools	Continuous measure (48-point scale), descriptively split into High, Mid, and Low)	Highest current parental occupation and education level according to NS-SEC occupation ranking scale (48-point)	Independent Predictor & Moderator	Yes	2 groups: Monolingual English vs. Bilingual (28 home languages; natural home acquisition)	Bilingual Proficiency Index (BPI); Sentence repetition
Dicataldo and Roch (2020)	111	~5 yrs	Italy	Cross-sectional	Preschool	Continuum (Low to High SES)	Composite (PCA): Income and Education	Independent Predictor & Moderator	Yes	2 groups: Monolingual Italian vs. Multilingual (28 families with non-Italian L1). Natural home acquisition	Length of exposure to the language context (LELC) and Daily exposure the language context (DELC)
Engel de Abreu et al. (2012)	80	7–9 yrs	Luxembourg (Immigrant) and Portugal (Source)	Cross-sectional	Primary School	One group: Low SES	ISEI score (parent occupation/education); income and household size/possessions	Matching Variable	No	2 groups: Portuguese Monolingual vs. Portuguese–Luxembourgish Bilingual; natural/home acquisition	Daily use reports; conceptual vocabulary (Expressive One-Word Picture Vocabulary Test—EOWPVT) in both languages)
Engel de Abreu et al. (2013)	60	M = 7; 1 yrs	Luxembourg (Immigrant) and Brazil (Monolingual context)	Cross-sectional	Multilingual State Classrooms (Luxembourg)	One group: Matched pairwise for Low- to Medium- SES	Caregiver education (ISCED converted into years of schooling)	Matching Variable	No	3 groups: Portuguese language-minority (Multilingual) vs. Luxembourgish language-majority (Multilingual) vs. Monolingual Portuguese; natural/home acquisition	Questionnaire; Home/School words production/comprehension
Esposito (2021)	288 full sample; 136 Coarsened Exact Matching (CEM) sample (68 matched pairs)	Grades K-1: M = 6; 11; Grades 4–5: M = 10; 9	USA (Southeastern Rural)	Cross-sectional/Quasi-experimental (CEM matched-pairs)	Rural public elementary schools: Spanish–English Two-Way Dual-Language programme (TWDL) vs. Mainstream English education	Uniformly Low-income	Caregiver education level (7 levels) and contextual rural poverty	Matching Variable	No	2 educational groups: Spanish–English Two-Way Dual-Language (TWDL) and mainstream English education	Educational-programme assignment (Spanish–English Two-Way Dual-Language (TWDL) vs. mainstream English); Spanish Category Fluency Task
Esposito and Baker-Ward (2013)	113	6–10 yrs	USA	Cross-sectional	Rural Title 1 Schools (primary and elementary)	One group: Uniformly Low (almost 90% free lunch eligibility)	Poverty proxy (90% free/reduced lunch eligible)	Contextual	No	2 status groups: Instructional: Dual-language programme (50:50 Spanish–English instructional) vs. Traditional (English monolingual)	Native Spanish/English via years in dual-language programme and parental reports.
Federico et al. (2023)	135	7–11 yrs (M = 9; 5)	Italy	Cross-sectional	Primary School in Suburbs	Two groups: Low and Medium-Low SES	Parental education years and occupation scores	Control (Covariate)	No	2 groups: Monolingual Italian vs. Bilingual (natural acquisition from birth/before age 4)	Parent questionnaire (proficiency, daily use, AoA); Peabody Picture Vocabulary Test, Fourth Edition (PPVT-IV) for receptive vocabulary
Filippi et al. (2020)	209	7–12 yrs (M = 9.5)	UK (London and Cambridge)	Cross-sectional	Primary School	Continuous composite score (Middle to High) Later clustered into 3 groups for analysis	Composite: Parental education, employment, income	Matching & Descriptive Indicator in cluster analyses	No	2 groups: English Monolingual vs. Multilingual (English + one or more of 16+ languages); natural home acquisition	Online parent questionnaire (history/exposure); British Picture Vocabulary Scale: Third edition (BPVS-III) for English receptive vocabulary
Filippi et al. (2022)	859	11–15 yrs (Test 1: M = 12, Test 2: M = 14.2)	UK London	Longitudinal (Study of Cognition, Adolescents and Mobile Phones (SCAMP))	Secondary School (State and Independent) (SCAMP)	Two groups: High vs. Low SES (via cluster analysis).	Cluster variables: School type, Carstairs postcode deprivation index, parental university attendance, and ONS occupation status	Categorical Grouping Factor & Moderator	Yes	2 groups: English Monolingual vs. Multilingual natural family acquisition	Questionnaires identifying first language acquired and current languages spoken with family
Frechette et al. (2023)	1021	1.9–5.1 yrs	USA	Cross-sectional	Head Start (Low income)	One group: Uniformly Low SES	Family SES Risk: sum of risks (household income, single-parent status, parent education lower than Associate of Arts degree); neighbourhood vulnerability z-score	Independent Predictor & Moderator	Yes	3 groups: English Monolingual vs. Spanish-English Dual Language Learners (DLL) (split into “Proficient” and “Emergent” Bilinguals); natural home acquisition	Parent report at enrolment; Emergent vs. Proficient bilingual status; standardised English performance on the Peabody Picture Vocabulary Test, Fourth Edition (PPVT-4)
Gathercole et al. (2010)	193 (Tapping Task)/258 (Stroop Task)	M = 6; 4–8; 11 & M = 12; 4–16; 0 yrs	UK North Wales and Chester	Cross-sectional	Primary/Secondary Schools	Continuum (Low to High SES) analysed via correlations	Parental education (1–5 scale) and occupational status (1–3 white-collar to blue-collar scale)	Correlate	No	4 groups: English Monolingual vs. Welsh–English Bilinguals (OWH: Only Welsh home; WEH: Both; OEH: Only English home); natural home acquisition	Questionnaires for history/usage; Receptive vocabulary (British Picture Vocabulary Scale (BPVS)/Welsh test); Battery of 13 receptive grammar structures
Grote et al. (2021)	60	4 yrs	USA	Cross-sectional	Head Start/Preschools	One group: Uniformly Low-SES	Annual family income categories (<$20,000 to $39,999)	Contextual	No	3 groups: Monolingual English vs. Monolingual Spanish vs. Spanish–English Bilingual. Natural home acquisition	Peabody Picture Vocabulary Test (PPVT) English/Spanish
Hartanto et al. (2019)	18,200	5–7 yrs	USA	Longitudinal (Early Childhood Longitudinal Study–Kindergarten Class of 2010–2011 (ECLS-K))	State/Public Schools (Kindergarten-First Grade)	Continuous variable with 3 representative levels: Low, Middle and High SES)	Composite: Education, occupation, income; and components separately	Independent Predictor & Moderator	Yes	2 groups: Monolingual English vs. Bilingual (various home languages). Natural home acquisition	Dummy coding based on household language; English proficiency (Preschool Language Assessment Scales (preLAS))
Huang et al. (2023a)	74	3–5.5 yrs (M = 4; 2) yrs 1	USA	Cross-sectional	Preschool (assessed online)	One group: Uniformly Low-SES	Household income-to-needs ratio (INR < 2)	Eligibility criterion & Covariate	No	2 groups: English Monolingual vs. Spanish–English Bilingual; natural home acquisition	Quick Interactive Language Screener (QUILS) (English) and Quick Interactive Language Screener: English–Spanish (QUILS:ES) (English-Spanish) standardised assessments
Huang et al. (2023b)	68 (35 English monolinguals, 33 Spanish–English bilinguals)	3–5.5 years (Monolinguals: M = 4.08 years; Bilinguals: M = 4.33 years 1)	USA (Nationwide online recruitment)	Cross-sectional	Preschool; specific educational setting not reported	Uniformly Low-income; ITN analysed continuously (44.1% poverty, 55.9% near-poverty)	Household income-to-needs ratio (INR) calculated from family size and household income	Eligibility criterion & Covariate	No	2 groups: Low-income English Monolinguals vs. Low-income Spanish–English Bilinguals	Language Experience and Proficiency Questionnaire (LEAP-Q) parent report; Quick Interactive Language Screener (QUILS) (English) & QUILS:ES (Spanish); Degree of Bilingualism score
Krizman et al. (2016)	62	M = 14; 6 yrs	USA	Cross-sectional	Public High School	Two groups: Low vs. High SES	Maternal level of education	Categorical Grouping Factor & Moderator	Yes	2 groups: English Monolingual vs. Spanish–English Bilingual. Natural home/heritage acquisition	Parent report; neural consistency metrics (brainstem responses to speech); The Language Experience and Proficiency Questionnaire (LEAP-Q); Wechsler IQ
Ladas et al. (2015)	110	Experiment 1: 6–12 yrs; Experiment 2: 6–8 yrs	Greece	Cross-sectional	Public Schools	One group: Uniformly Low-SES	Parental Education and Occupation Status (Hellenic Statistical Authority)	Matching	No	2 groups: Greek Monolingual vs. Greek–Albanian Bilingual (heritage acquisition)	Self/parent reports; Vocabulary subtest of the Wechsler Intelligence Scale for Children—Version III (WISC-III); objective Language-Switching switch cost asymmetry
Morton and Harper (2007)	34	M = 6; 9 yrs	Canada	Cross-sectional	French schools (Bilinguals) vs. birth announcements (Monolinguals)	One group: Predominantly Middle Class (to High)	Composite: Parental education and total income	Matching	No	2 groups: Monolingual English vs. English–French Bilingual; natural home acquisition	Daily Language Use Questionnaire; Peabody Picture Vocabulary Test-Revised (PPVT-R) (English) and Échelle de Vocabulaire en Images Peabody (EVIP) (French) for vocabulary
Palomino and Brudvig (2022)	43	4.7 yrs	USA	Cross-sectional	Head Start Preschool	One group: Economically disadvantaged/at-risk	Eligibility for Head Start programme (economically at-risk)	Contextual	No	2 groups: English Monolingual vs. Bilingual. Natural home acquisition.	Parent reports of status; Peabody Picture Vocabulary Test, Fourth Edition (PPVT-IV) for receptive English vocabulary
Panesi et al. (2026)	91 (44 monolinguals, 47 bilinguals; SES composite available for 83 children)	4.42–6.17 years 1 (M = 5.35 years)	Italy (Northwestern region)	Cross-sectional	Two public preschools in Genoa	Continuous composite (monolinguals had significantly higher SES than bilinguals)	Composite z-score: maternal and paternal education level (0–6) and occupation level (1–8)	Control (Covariate)	No	2 groups: Monolingual Italian vs. Bilingual (Italian at school, home language: Spanish 40%, Albanian 29.8%, Arabic 10.6%, French, English, Chinese, Sinhala, Polish, Tagalog, Urdu)	Parents’ questionnaire on home language/nationality; Peabody Picture Vocabulary Test (PPVT) for Italian receptive vocabulary/comprehension
Perry et al. (2026)	1673 pupils (SCAMP cohort)	11–16 yrs (SATs at 11, cognitive testing at 12 & 14, GCSEs at 16)	UK (Greater London, England)	Longitudinal cohort study	State secondary schools	Continuous factor score (Monolinguals higher SES than multilingual groups)	Number of university-educated parents; parental occupational prestige; Free School Meal status; area-level deprivation (Carstairs Index)	Covariate	No	3 groups: Monolinguals (English), Simultaneous Multilinguals (English and another language), and Multilingual Learners (English not L1)	SCAMP self-report language-status question; classification checked for consistency with National Pupil Database records
Santillán and Khurana (2018)	1146	4–6 yrs	USA	Longitudinal Family and Child Experiences Survey (FACES)	Head Start Preschool	One group: Uniformly Low SES	SES Risk Index (Poverty, single parent, maternal education) based on parent reports and the Family and Child Experiences Survey (FACES) indicators	Control (Covariate)	No	3 groups: English monolinguals vs. early Spanish–English bilinguals vs. emerging bilinguals (children who actively acquired bilingualism during the course of the study)	Parent reports; Preschool Language Assessment Scales (preLAS) (Simon Says and Art Show) for English proficiency
Tonér and Nilsson (2021)	47	4–6 yrs	Sweden	Cross-sectional	Preschool (Suburban)	One group: Measured as a continuum (10-grade scale); sample was predominantly High SES, but SES of bilingual children was considerably lower than of monolingual group	Composite: Parental annual income and education level	Independent Predictor	Yes	2 groups: Monolingual Swedish vs. Bi/multilingual (various home languages; natural home acquisition)	Narrative language (Bus Story Test); parent-reported vocabulary and morphology (Swedish Communicative Development Inventory III—SCDI-III)
Tonér et al. (2021)	431	4–6 yrs	Sweden	Cross-sectional	Preschool	One group: Continuous categories; predominantly higher-SES	Parental income and education separately	Independent Predictor	Yes	2 groups: Monolingual Swedish vs. Bi-/multilingual (43 languages); natural home acquisition	Parental questionnaire on home language use and strongest language; Swedish vocabulary (Peabody Picture Vocabulary Test—PPVT) and grammar assessments; Swedish Communicative Development Inventory (Swedish Communicative Development Inventory III—SCDI-III).
Trebits et al. (2022)	39	9–10 yrs	Germany	Longitudinal	Primary Schools: Regular vs English Immersion	One group: Diverse Range: Population analysed continuously via HISEI	HISEI (Highest occupation level)	Independent Predictor	Yes	2 groups: German Monolingual vs. Sequential Bilingual (Instructional Immersion).	School programme status; L2 English vocabulary and grammar assessments
Tsimpli et al. (2020)	687	M = 9; 11 yrs	India	Cross-sectional	Government Primary Schools (Slum/Non-slum)	One group: Continuous score: Uniformly Low-SES	Asset-based score: 8-item index: gadgets, tap water, homework, chores, and mother’s literacy/education	Descriptive variable	No	2 groups: Monolingual vs. Bilingual (natural acquisition; e.g., Hindi, Telugu)	Questionnaire on conversational partners; sociolinguistic diversity composite score
White and Greenfield (2017)	303 (148 Spanish–English bilinguals, 83 English monolinguals, 72 Spanish-dominant emerging bilinguals)	3–5 yrs (M = 4.36 yrs 1)	USA (Miami-Dade County, Florida)	Cross-sectional	Head Start preschool centres (7 centres)	Uniformly Low-income	Head Start poverty enrolment verification	Contextual	No	3 groups: Spanish–English bilinguals, English monolinguals, and Spanish-dominant emerging bilinguals	Woodcock Muñoz Language Survey–Revised, Normative Update (WMLS-R NU) Picture Vocabulary subtest in English and Spanish

¹ The authors provide the age in months. For purposes of comparability, this measure was transformed into years.

Following data charting, the extracted information was compared across studies to identify patterns, similarities and differences, and was grouped according to shared characteristics such as EF domain, SES, age, bilingual experience, and methodological design. Studies with similar characteristics were then grouped under broader themes to facilitate cross-study comparisons and aid in identifying themes of discussion, which will be presented, alongside the results, in the following sections.

A recommended supplementary search was conducted on 17 September 2026 to identify relevant studies published or indexed since the original search. The following expanded search string was used: (child* OR “school-aged” OR adolescen*) AND (socioeconomic OR “socio-economic” OR SES OR poverty OR income OR disadvantage* OR deprivation OR “free lunch” OR “Head Start” OR “maternal education” OR “parental education” OR occupation OR “social class”) AND (bilingual* OR multilingual* OR “second language” OR “dual language” OR “heritage language” OR immersion) AND (“executive function*” OR inhibition OR “inhibitory control” OR “working memory” OR “cognitive flexibility” OR shifting OR switching). The syntax was adapted where necessary to meet the requirements of each database, as previously mentioned.

The supplementary search identified 964 records. After removing 412 records matching those identified in the original search, 552 records remained. A further 178 duplicates were removed, leaving 374 records for title and abstract screening. Of these, 326 were excluded, 48 reports were sought for retrieval, and two could not be retrieved. The remaining 46 reports underwent full-text eligibility assessment, of which 38 were excluded, resulting in eight additional included studies. The complete selection process is presented in Supplementary Figure S1. When combined with the 31 studies identified through the original search, these eight additional studies brought the final review sample to 39 studies.

3. Results

3.1. Characteristics of the Included Studies

The characteristics of the 39 studies included in this scoping review are summarised in Table 1. The studies are described according to their year of publication, geographical location, participant characteristics, educational settings, conceptualisation and measurement of bilingualism, operationalisation of socioeconomic status (SES), and executive functions (EFs) assessed.

3.1.1. Publication Period

As can be seen in Figure 2, the articles reviewed were published between 2007 and 2026, with most appearing after 2018. This distribution reflects a steady interest in understanding the relationship between bilingualism, SES, and EF development during childhood and adolescence.

Figure 2. Publications per year.

3.1.2. Geographical Distribution

The reviewed studies included data from 12 countries, with most research conducted in North America and Western Europe. The largest number of studies were carried out in the United States (n = 16), followed by the United Kingdom (n = 6). Canada and Italy contributed three studies each, while Greece, Germany, The Netherlands, Sweden, and Luxembourg were represented in two studies each. India was represented in one study. The two studies conducted in Luxembourg additionally included monolingual comparison samples from Portugal and Brazil, respectively. Overall, 25 of the 39 studies were conducted primarily in the United States, the United Kingdom, or Canada, reflecting the continued predominance of English-speaking contexts in research on this topic and the comparatively limited representation of other linguistic and sociocultural settings. However, this pattern should be interpreted in light of the review’s restriction to English-language publications and the coverage of the databases searched, which may have increased the representation of research conducted in English-speaking contexts.

3.1.3. Participants

Great variation can be observed in sample sizes, ranging from 34 participants to more than 18,000 participants in large national cohort datasets, as well as the age groups studied. Table 2 reflects the distribution of studies according to the age of the participants:

Table 2. Distribution of studies according to age of participants.

Early Childhood (2–6)	Middle-Childhood (7–11)	Adolescence (12–17)	Multi-Stage Samples
18	8	3	10

The studies were unevenly distributed across developmental stages. Eighteen studies focused on early childhood, eight on middle childhood, three on adolescence, and ten included multi-stage samples. Thus, the evidence remains considerably stronger for early and middle childhood than for adolescence. Conclusions concerning adolescence should therefore be regarded as tentative, and further research focusing specifically on bilingual adolescents from different socioeconomic backgrounds is needed.

3.1.4. Sites of Data Collection

Data were collected in a wide variety of settings, including preschools, primary and secondary schools, bilingual and dual-language immersion programmes, Head Start centres, lower-SES charter schools, inner-city and rural public schools, community-based settings, and remote or home-based assessment contexts. Several studies also used large-scale longitudinal cohorts and educational databases.

3.1.5. Language Status

Following the eligibility criteria used in this review, all studies included a monolingual comparison group. The predominant methodological approach was a bilingual–monolingual comparison (e.g., Bablekou et al., 2025; Hartanto et al., 2019; Morton & Harper, 2007), although a subset of studies also incorporated multilingual profiles, including simultaneous multilinguals, multilingual learners, and children with different levels of dominance or proficiency (e.g., Bialystok & Shorbagi, 2021; Czapka et al., 2020; Engel de Abreu et al., 2013; Perry et al., 2026; White & Greenfield, 2017).

The bilingual/multilingual vs. monolingual condition was established through different means. Mostly, children were classified according to the level of exposure and use of their home language via parent-report questionnaires (e.g., Federico et al., 2023; Gathercole et al., 2010). Many studies also used tests of language proficiency to measure bilingualism, such as standardised vocabulary tests (e.g., Bosman & Janssen, 2017; Buac & Kaushanskaya, 2014; Buac et al., 2016, Choi et al., 2018), while others took into account age of acquisition, duration of bilingual experience, or participation in immersion or dual-language educational programmes (e.g., Dicataldo & Roch, 2020; Trebits et al., 2022). On the other hand, a smaller number of studies treated bilingualism as a multidimensional construct by combining measures of language proficiency, exposure, and use across both languages (e.g., De Cat et al., 2018; Krizman et al., 2016). Finally, most studies focused on Spanish–English bilinguals (e.g., Frechette et al., 2023; Esposito, 2021; Huang et al., 2023a).

Spanish–English was the most frequently represented bilingual pairing (e.g., Frechette et al., 2023; Huang et al., 2023a; White & Greenfield, 2017). Nevertheless, the evidence base also included Turkish–Dutch, Italian bilingual, and linguistically diverse multilingual samples involving a broad range of heritage and home languages (e.g., Abdurokhmonova et al., 2026; Bosman & Janssen, 2017; Panesi et al., 2026; Perry et al., 2026; Tonér et al., 2021; Tsimpli et al., 2020).

3.1.6. Operationalisation of Socioeconomic Status

SES was also measured and incorporated into the main analysis in a wide variety of ways. The most widely used proxies were parental education, household income, and parental occupation, either individually or combined into composite SES indices (e.g., Bablekou et al., 2025; De Cat et al., 2018; Filippi et al., 2020; Panesi et al., 2026; Tonér & Nilsson, 2021). Some studies calculated an income-to-needs ratio based on household income and family size (e.g., Abdurokhmonova et al., 2026; Huang et al., 2023b), while others used broader multidimensional measures that additionally incorporated poverty status, family structure, eligibility for free or reduced-price lunch, Head Start participation, cumulative socioeconomic risk, or neighbourhood and area-level deprivation (e.g., Choi et al., 2018; Esposito & Baker-Ward, 2013; Frechette et al., 2023; Palomino & Brudvig, 2022; Perry et al., 2026; Santillán & Khurana, 2018).

Several studies examined uniformly low-income or socioeconomically disadvantaged samples, with SES established through Head Start eligibility, neighbourhood or school context, poverty thresholds, or recruitment criteria (e.g., Bosman & Janssen, 2017; Choi et al., 2018; Esposito, 2021; Huang et al., 2023b; White & Greenfield, 2017). These restricted-range designs provide evidence concerning bilingualism and EFs within disadvantaged populations, but they do not allow for a direct study of whether the connection between bilingualism and EFs varies across SES levels.

The analytical role of SES also differed considerably across studies. SES was treated as an independent predictor (e.g., Buac & Kaushanskaya, 2014; Dicataldo & Roch, 2020; Perry et al., 2026), moderator (e.g., Filippi et al., 2022; Krizman et al., 2016), covariate (e.g., Blom et al., 2014; Carlson & Meltzoff, 2008; Panesi et al., 2026), matching variable (e.g., Engel de Abreu et al., 2012; Esposito, 2021; Ladas et al., 2015), grouping factor (e.g., Bialystok & Shorbagi, 2021; Krizman et al., 2016), eligibility or sample-selection criterion (e.g., Bosman & Janssen, 2017; Huang et al., 2023b), or contextual or descriptive characteristic of the sample (e.g., Filippi et al., 2020; White & Greenfield, 2017).

Across the 39 included studies, 13 formally tested whether SES moderated the association between bilingualism and executive functioning by including a bilingualism × SES interaction. Eight studies used SES only as a control or covariate. Restricted-range studies were defined as those whose samples were confined to a single or narrow socioeconomic range, including uniformly low-SES, economically disadvantaged, and low-to-medium-low-SES samples. According to this definition, 16 studies used restricted-range samples. These categories were not mutually exclusive, as some restricted-range studies also controlled statistically for the remaining variation in SES. In addition, 10 studies entered SES as an independent predictor or main effect. This methodological heterogeneity highlights the absence of a standardised approach to conceptualising and measuring SES and limits direct comparison of findings across studies.

3.1.7. Measurement of Executive Functions

A wide range of tests was used to test the EFs, with some studies applying multiple measures for one EF (e.g., Blom et al., 2014 or Engel de Abreu et al., 2012; Panesi et al., 2026), while others used single tasks (e.g., Bialystok & Shorbagi, 2021; Choi et al., 2018; De Cat et al., 2018). The methods used to measure these functions included traditional researcher-administered behavioural tasks, computerised and online tasks, standardised test batteries, and batteries from which composite or latent EF scores were derived (e.g., Abdurokhmonova et al., 2026; Huang et al., 2023b; White & Greenfield, 2017). Studies also show a wide variety of tasks to measure the EFs, with more than 20 distinct task paradigms used to measure inhibition alone, therefore limiting direct comparability across findings. Backward Digit Span or Backward Digit Recall was a notable exception to this variability, being the most frequently used measure of WM.

3.2. Executive Function Outcomes

The EF outcomes reported across the 39 studies are summarised in Table 3. To make comparison across studies easier, findings were classified as advantage, mixed, or null. Advantage was used when the study reported significantly better EF performance or development among bilingual participants. Mixed findings were identified when bilingual effects were conditioned by specific contextual factors, including SES, age, bilingual proficiency, developmental stage, task characteristics, or specific subcomponents of the EF. Finally, null referred to studies reporting no bilingual advantage, either because no significant bilingual–monolingual differences were found or because monolingual participants performed significantly better.

Table 3. Descriptive summary of studies reviewed: executive function outcomes and tasks used.

Author(s) & Year 1	EF Measure and Task Used	Key Findings and Interactions Regarding Bilingual Advantage by EF Domain
Abdurokhmonova et al. (2026)	Composite EF (inhibition, WM, Flexibility): Head–Toes–Knees–Shoulders (HTKS) task; Hearts and Flowers (HF) task	Composite EF: Mixed; Parent-reported bilingual knowledge was not significantly associated with EF-composite performance after controlling for SES. However, more balanced home bilingual exposure was significantly associated with better EF-composite performance. Higher SES independently predicted better EF, but neither bilingual knowledge nor bilingual exposure moderated the relationship between SES and EF. Thus, there was no evidence of a bilingual protective effect at lower SES.
* Bablekou et al. (2025)	Inhibition: Attention Network Test (ANT), WM: Backward Digit Task/2-back updating task, Flexibility: How Many—What Number task	Inhibition: Null; Accuracy and RT were predicted by age and non-verbal IQ; no significant effect of bilingualism. WM: Null; Verbal working memory (backward span) was predicted by SES, not language status Flexibility: Null; Age was the dominant predictor; no moderating effect of bilingualism was found.
Bialystok and Shorbagi (2021)	Inhibition: Simon Picture Task	Inhibition: Mixed; Interaction found; higher-SES bilinguals outperformed all other groups, but middle-SES results were null.
Blom et al. (2014)	WM: Odd-One-Out (Visuospatial), Backward Digit (Verbal)	WM: Advantage; At age 6 years, bilinguals outperformed monolinguals Backward Digit Recall after controlling for SES and vocabulary.
Bosman and Janssen (2017)	WM: Automated Working Memory Assessment (AWMA) Listening Recall; AWMA Backward Digit Recall	WM: Null; No bilingual advantage was found within the uniformly low-SES sample. Native-Dutch monolingual children significantly outperformed Turkish–Dutch bilingual children on Dutch Listening Recall and Backward Digit Recall. Within the bilingual group, performance did not differ significantly between the Turkish and Dutch versions of either working-memory task.
* Buac and Kaushanskaya (2014)	Inhibition: Flanker Task (non-linguistic), Grammaticality Judgment Task (linguistic)	Inhibition: Null; No bilingual advantage was found, even though domain-general cognitive control was found to be instantiated differently by monolinguals and bilinguals.
* Buac et al. (2016)	WM: Backward Digit Span, Listening Span task	WM: Null: No difference between groups in WM tasks, despite significant difference in SES and English proficiency.
Calvo and Bialystok (2014)	Inhibition: Flanker, WM: Frog Matrices Task (FMT)	Inhibition: Advantage; Bilinguals outperformed monolinguals across both low and middle SES groups (independent and additive effects). WM: Advantage; Bilinguals were more accurate than monolinguals regardless of SES (independent and additive effects).
Carlson and Meltzoff (2008)	Task Battery Composite: Inhibition: ANT, Delay of Gratification WM: Visually Cued Recall, Flexibility: Advanced DCCS	Inhibition: Mixed significant relative advantage of native bilingual children in conflict tasks after controlling for age and SES, but none for emerging bilinguals. No advantage found in response tasks. WM: Mixed; significant relative advantage of native bilingual children after controlling for age and SES, but none for emerging bilinguals. Flexibility: Mixed; significant relative advantage of native bilingual children after controlling for age and SES, but none for emerging bilinguals.
Choi et al. (2018)	Inhibition: Pencil Tapping Task	Inhibition: Mixed; At Head Start entry, Spanish–English bilinguals and English monolinguals did not differ significantly in inhibitory control, although bilinguals outperformed DLLs with limited English skills (DLLs-LES). Across the following 1.5 years, both bilinguals and DLLs-LES showed significantly faster inhibitory-control growth than English monolinguals. By kindergarten, bilingual children had the highest inhibitory-control scores.
Clayden et al. (2023)	WM: CANTAB Spatial Working Memory (SWM)	WM: Null; Spatial WM was associated with maternal education. No direct effect of Multiple Language Exposure on WM at age 11 years was found once early abilities and SES were accounted for. However, data about bilingual experience and level of English are limited and may have influenced the results.
Czapka et al. (2020)	Inhibition: Go/Nogo, Bivalent Shape Task (BST), WM: N-back	Inhibition: Mixed; Interference Inhibition/Executive Processing: Advantage; Globally reduced RTs for multilinguals emerged after individual pairwise matching for background variables. Response Inhibition: Null; No differences found between monolingual and bilingual children on the Go/Nogo task WM: Null; Initial WM gaps were eliminated once children were matched for background variables.
De Cat et al. (2018)	Inhibition: Simon Task, WM: Backward Digit recall (WISC-III), Flexibility: Advanced Dimensional Change Card Sort DCCS	Inhibition: Advantage; global bilingual advantage in reaction time (RT); higher SES conferred a “pronounced advantage” in the bilingual group; higher amount of bilingual exposure predicted better performance overall. WM: Null; neither bilingual proficiency nor SES were significant predictors. Flexibility: Null; Found a significant interaction where SES predicted DCCS performance in bilinguals only, but no bilingual advantage
Dicataldo and Roch (2020)	Inhibition: Day and Night Test (Stroop condition), WM: Backward Digit Span, Flexibility: Dimensional Change Card Sort (DCCS)	Inhibition: Null; Found no specific influence of bilingual exposure on the accuracy or speed of preschoolers. WM: Advantage; Specific contribution of bilingual exposure to backward span over and above the effect of SES. Flexibility: Mixed; Contribution of bilingual exposure found for shifting proficiency but not accuracy.
Engel de Abreu et al. (2012)	Inhibition: Flanker Fish Task, Sky Search, WM: Odd-One-Out	Inhibition: Advantage; Bilingual children showed significantly better executive control/selective attention and interference suppression than monolingual children and were faster on congruent and incongruent flanker trials; performed better on the Sky Search attention score. WM: Null; Bilingual and monolingual children performed equivalently on the Odd-One-Out task
Engel de Abreu et al. (2013)	WM: AWMA Counting Recall and Backward Digit Recall	WM: Null; Digit-based tasks were resilient to SES disadvantage but showed no group difference.
Esposito (2021)	Inhibition: Bivalent Shape Task, Simon Task Flexibility: Trail Making Task	Inhibition: Null; The computerised inhibition tasks did not provide a robust bilingual advantage. Significant differences occurred primarily on neutral or congruent trials, rather than the conflict trials that specifically index inhibitory control. Flexibility: Null; Children in two-way dual-language and mainstream English education did not differ significantly on the Trail Making switching measure.
Esposito and Baker-Ward (2013)	Inhibition: Sun/Moon Task (response inhibition), Flexibility: Trail Making Task (TMT)	Inhibition: Null; Bilinguals and monolinguals performed comparably on the Sun/Moon task (response inhibition). Flexibility: Advantage; Dual-language students showed higher switching proficiency (Trail C) than monolingual peers.
Federico et al. (2023)	Inhibition: Attention Network Task WM: Backward Digit Span/Letter Numbering Sequencing	Inhibition: Mixed; Bilingual advantage found in the executive conflict network specifically when using social stimuli, but not for non-social stimuli. WM: Null; No significant differences found for letter-number sequencing.
Filippi et al. (2020)	Inhibition: Auditory–motor task	Inhibition: Null; Results for verbal and non-verbal interference are comparable in multilingual and monolingual children
Filippi et al. (2022)	WM: Backward Digit Span Flexibility: Trail Making Test	WM: Advantage; General advantage of bilinguals in Backward Digit Span (verbal). Flexibility: Mixed; Bilingual advantage for Low but not high SES
Frechette et al. (2023)	Composite EF: Minnesota Executive Function Scale	Composite EF: Mixed; Proficient bilinguals had similar EF levels to monolinguals, while emergent bilinguals performed lower on EF skills
Gathercole et al. (2010)	Inhibition: Stroop Task Flexibility: Tapping sequence	Inhibition: Null; No bilingual advantage was found on the Stroop Task, for which SES was a significant predictor. Flexibility: Advantage; Welsh-dominant bilinguals outperformed monolinguals on the switching tapping task.
Grote et al. (2021)	Inhibition: Day/Night, Embedded Figures Task, WM: Concentration game and Colorforms task	Inhibition: Advantage; Bilingual preschoolers exhibited better inhibition. WM: Advantage; Bilingual preschoolers performed better in both visual–spatial memory tasks.
Hartanto et al. (2019)	WM: Numbers Reversed Subtest Flexibility: Dimensional Change Card Sort	WM: Mixed; No effect of bilingualism on verbal working memory; bilingual status only buffered low-SES effects at age 7 years (wave 4). Flexibility: Advantage; Bilinguals performed better in shifting task.
Huang et al. (2023a)	Inhibition: Animal Go–NoGo, Flexibility: Dimensional Change Card Sorting	Inhibition: Advantage; Bilingual preschoolers performed significantly better on response inhibition accuracy. Flexibility: Advantage; Bilingual preschoolers outperformed monolinguals in attention shifting (DCCS).
Huang et al. (2023b)	Inhibition: Animal Go–No Go Task Flexibility: Dimensional Change Card Sorting (DCCS) Task (online post-switch score).	Inhibition: Advantage; Low-income Spanish–English bilingual children performed significantly better than low-income English monolingual children on the Animal Go–NoGo task. The bilingual advantage remained significant after controlling for age, income-to-needs ratio, and English vocabulary. Flexibility: Advantage; Low-income Spanish–English bilingual children performed significantly better than low-income English monolingual children on the Dimensional Change Card Sorting task. The bilingual advantage remained significant after controlling for age, income-to-needs ratio, and English vocabulary.
Krizman et al. (2016)	Inhibition: Integrated Visual and Auditory Continuous Performance Test	Executive control (Inhibition): Advantage; Heightened executive control in bilinguals independent of their specific low/high SES standing.
Ladas et al. (2015)	Inhibition: Attentional Network Task (ANT), Scalar Implicature (SI) task	Inhibition: Mixed; No advantage for bilinguals as a group, but correlation between level of bilingualism and ability to resolve conflict
Morton and Harper (2007)	Inhibition: Simon Task	Inhibition: Null; The bilingual lead disappeared when groups were matched for identical middle-class backgrounds; argued reported advantages were SES artifacts.
Palomino and Brudvig (2022)	Composite EF: Minnesota Executive Function Scale (MEFS), Regulation focus	Composite EF: Advantage; Bilingual status was a significant predictor of EF skills.
Panesi et al. (2026)	Inhibition: Day/Night Stroop Task; Simon Says Task Working Memory: Backward Word Span (BWS); Magic House Task Flexibility: Dimensional Change Card Sort (DCCS)	Inhibition: Null; Monolingual children initially outperformed bilingual children on Simon Says, while no significant group difference emerged on the Day/Night Stroop task. After controlling for SES, neither inhibition task differed significantly by linguistic status. Therefore, there was no bilingual advantage. WM: Null; Monolingual children initially outperformed bilingual children on the Magic House updating task, whereas no significant differences emerged on Backward Word Span. After controlling for SES, linguistic status did not significantly predict performance on any working-memory task. Therefore, there was no bilingual advantage. Flexibility: Null; Monolingual and bilingual children did not differ significantly on the DCCS, either before or after controlling for SES.
Perry et al. (2026)	WM: Backward Digit Span; Spatial Working Memory Task	WM: Mixed; Simultaneous multilinguals demonstrated a cognitive advantage at ages 12 and 14 years, with verbal working memory showing the closest association with multilingual status. However, language status did not significantly impact rate of development of verbal working memory from ages 12 to 14 years, and no clear multilingual advantage was reported for spatial working memory.
Santillán and Khurana (2018)	Inhibition: Pencil-Tapping Task	Inhibition: Advantage (Growth); Early bilinguals showed higher initial levels and steeper growth in inhibitory control; developing bilinguals started with lower levels but showed same rate of growth as early bilinguals, and higher rate of growth than monolinguals.
* Tonér and Nilsson (2021)	“Unity” EF: Inhibition: Head–Shoulders–Knees–Toes (HTKS), Flanker Fish Task), WM: Backward Digit Span, Flexibility: Dimensional Change Card Sort (DCCS)	Inhibition: Null; No difference between bilinguals and monolinguals was observed, but bilinguals were from significantly lower SES. WM: Null; No difference between bilinguals and monolinguals was observed, but bilinguals were from significantly lower SES. Flexibility: Null; No difference between bilinguals and monolinguals was observed, but bilinguals were from significantly lower SES. EF composite: Null; No difference between bilinguals and monolinguals was observed, but bilinguals were from significantly lower SES.
Tonér et al. (2021)	Inhibition: Flanker Fish Task, Head–Shoulders–Knees-and-Toes task (HTKS), WM: Backward Digit Span, Flexibility: Dimensional Change Card Sort (DCCS)	Inhibition: Null; Flanker scores correlated with WM but showed no unique bilingual lead. WM: Null; BDS scores correlated with language skills but showed no unique bilingual lead. Flexibility: Null; High-SES sample showed DCCS scores correlated with WM but no unique lead.
Trebits et al. (2022)	WM: Letter-Number Sequencing, Backward Digit Span	WM: Advantage; Immersion students outperformed monolingual peers on all WM tests at the end of the study, but not at the beginning.
Tsimpli et al. (2020)	WM: 2-back task	WM: Advantage; Bilinguals outperformed monolinguals on 2-back task.
White and Greenfield (2017)	Composite Latent EF Factor: Inhibition: Spatial Conflict Arrows (SCA), Go–No-Go (GNG), Silly Sounds Stroop (SSS) WM: Working Memory Span Houses (WMSH), Pick the Picture (PTP) Flexibility: Something’s the Same (STS)	Composite EF: Mixed; Spanish–English bilingual children demonstrated significantly higher latent EF than English monolingual children after controlling for age, gender, and processing speed. Spanish-dominant emerging bilinguals obtained intermediate EF scores but did not differ significantly from either established bilinguals or English monolinguals.

¹ In studies marked with an asterisk, comparable performance of bilinguals and monolinguals can be interpreted as actually indicating an advantage of the former, since bilingual/multilingual subjects in this study came from lower SES backgrounds (see p. 27).

As will be seen below, this classification was not without its challenges, since some studies compared bilingual and monolingual participants from different SES backgrounds, and these differences were not always accounted for consistently when interpreting the results. This could have led to the classification of the impact of bilingualism as “null” when, in fact, equal performance on EF tasks of lower SES bilinguals and higher SES monolinguals may be understood as potentially consistent with the buffering hypothesis, but not sufficient to demonstrate one. These studies have been identified through an asterisk (*) in Table 3, but it was decided to maintain the authors’ interpretations all the same.

As regards WM, only executive-loaded measures of this skill were taken into account for the results. This is due to the fact that these have generally been considered to provide a more representative assessment of WM since they require both storage and manipulation of information, as opposed to simple span tasks that are generally regarded as a measure of short-term storage. (Conway et al., 2005; Friedman & Miyake, 2017).

Only performance-based cognitive tasks were included in the domain-specific EF results; questionnaire-based ratings of EF-related behaviour were not counted. To avoid double-counting across EF domains, each task outcome was assigned to a single domain according to the primary construct assessed by the relevant score. Response-inhibition and interference-control measures were classified as inhibition, tasks requiring the storage and manipulation of information as working memory, and switching or rule-shifting measures as cognitive flexibility. A study could contribute findings to more than one domain when it used separate tasks, but the same task result was not counted under more than one domain. Composite or latent EF scores were classified separately when domain-specific results were not reported.

4. Discussion

Overall, the results of the studies reviewed are very varied. Table 4 summarises their findings and specifies the number of studies reporting an advantage (A), mixed results (M) and null results (N) for each EF, using the task-to-domain classification rule described in Section 3.2. It thus forms an overview of the current evidence, and the basis for the discussion presented in the following section. Consistent with the mapping purpose of this scoping review, no formal critical appraisal was conducted. The descriptive counts therefore provide an overview of the patterns reported across the included evidence base, with each study contributing equally to the counts in Table 4 regardless of its design or sample size, which ranged from 34 to over 18,000 participants. Consequently, these counts do not account for differences in statistical power, precision, or methodological strength and should be interpreted as descriptive patterns rather than weighted estimates of the evidence.

Table 4. Distribution of advantage (A), mixed (M) and null (N) results for each EF.

Inhibition (n = 25)			Working Memory (n = 22)			Cognitive Flexibility (n = 14)			EF Composite (n = 4)
A	M	N	A	M	N	A	M	N	A	M	N
8	6	11	7	3	12	5	3	6	1	3	0

4.1. Executive Function Patterns Across Socioeconomic Contexts

4.1.1. Inhibition

Inhibition was the most frequently assessed executive function, but its findings were heterogeneous: eight studies were classified as showing a bilingual advantage, six as mixed, and eleven as null. Differences between inhibitory processes may help explain some of the variation. Several studies reported bilingual advantages on interference-control tasks, including the Flanker and Simon tasks, or on selective-attention measures such as Sky Search (e.g., Calvo & Bialystok, 2014; De Cat et al., 2018; Engel de Abreu et al., 2012). However, findings from these types of tasks were not consistently positive: other studies reported null or mixed results on the Flanker, Simon, ANT, and Bivalent Shape tasks (e.g., Bablekou et al., 2025; Buac & Kaushanskaya, 2014; Esposito, 2021; Morton & Harper, 2007). Response inhibition measures also yielded both null findings and advantages: Go/No-Go and delay tasks were often null (e.g., Carlson & Meltzoff, 2008; Czapka et al., 2020), whereas advantages were reported on Animal Go/No-Go and pencil tapping measures in some samples (e.g., Choi et al., 2018; Huang et al., 2023a; Santillán & Khurana, 2018). The findings are therefore compatible with the proposal that bilingual language management engages attention and conflict resolution (see Bialystok, 2017). However, the mixed results across both conflict and response inhibition tasks prevent a firm conclusion that bilingualism is more strongly associated with one inhibitory process than another.

Several factors may contribute to these inconsistent findings. First of all, the tasks used often require simultaneous activation and engagement of inhibition, working memory and cognitive flexibility (see Bialystok & Shorbagi, 2021; Czapka et al., 2020). The second reason for the inconclusive results can be found in the composition of the samples, either being too small or not adequately controlled for confounding variables, such as intelligence, migration history, or educational background (see De Cat et al., 2018; Filippi et al., 2020). A third reason may relate to methodological decisions, including how studies account for socioeconomic status (SES), intelligence, and other background variables. For example, Morton and Harper (2007) found no bilingual advantage when comparing children with similar socioeconomic and ethnic backgrounds. Other studies suggested that the association between bilingual experience and inhibitory-control performance varied with SES (e.g., Bialystok & Shorbagi, 2021).

4.1.2. Working Memory

As previously mentioned, this review considered only executive-loaded tasks to measure WM, as their high demands on both the storage and manipulation of information directly index the executive component of WM (Friedman & Miyake, 2017; Conway et al., 2005). Findings were predominantly null (12 of 22 studies), although seven studies reported an advantage and three yielded mixed results. Both verbal and visuospatial executive-loaded tasks yielded a mix of advantages and null findings, indicating that neither modality is more sensitive to bilingual effects than the other.

In some longitudinal studies, WM advantages were observed at later assessments but not earlier ones (e.g., Blom et al., 2014; Trebits et al., 2022). Hartanto et al. (2019) likewise found that bilingualism buffered the association between low SES and verbal WM only at a later wave. On the contrary, Perry et al. (2026) reported cross-sectional differences associated with multilingual status but no significant difference in longitudinal verbal WM change. These findings suggest that age and bilingual experience may be relevant, although they do not establish that advantages consistently emerge over time.

4.1.3. Cognitive Flexibility

Cognitive flexibility (CF) represents the least extensively investigated executive-function domain in the present review, appearing in 14 studies. Five studies were classified as showing a bilingual advantage, three as mixed, and six as null. The lack of research concerning this executive function can be partially explained due to the attention given to inhibition and WM, both compelling topics of interest due to the extensively reported connection between inhibition and language suppression in bilinguals and the relationship between WM and academic success and school readiness. Added to this are the issues of task impurity, the ‘ceiling effect’ often shown by the DCCS in children older than 5 years, and the linguistic mediation of complex rules common to CF tasks (see De Cat et al., 2018), which can confound results with a child’s linguistic proficiency.

As with inhibition, the heterogeneity of results suggests CF to also be highly task-dependent, making it difficult to draw firm conclusions regarding the existence of a consistent bilingual advantage. Across the included studies, bilingual advantages were most frequently reported in tasks requiring children to switch efficiently between mental sets or response rules, such as the Dimensional Change Card Sort (DCCS), the Trail Making Test (TMT), and other complex task-switching paradigms with brief switching costs that “tax” the executive system enough to reveal the bilingual’s refined control mechanisms, similar to what is required when managing two active language systems (e.g., Esposito & Baker-Ward, 2013; Hartanto et al., 2019; Huang et al., 2023a, 2023b). However, these same or closely related measures produced null findings in other studies (e.g., De Cat et al., 2018; Esposito, 2021; Panesi et al., 2026; Tonér et al., 2021). This variation indicates that task demands alone do not explain whether a bilingual advantage emerges. Differences in age, bilingual experience, SES, language proficiency, sample composition, and analytical controls may also contribute to the inconsistent results.

Developmental stage and accumulated bilingual experience may be factors to consider when interpreting findings regarding cognitive flexibility. Esposito and Baker-Ward (2013) found no programme difference among kindergarten children, while children in Grades 2 and 4 with longer exposure to dual-language education performed better on the Trail Making switching task. However, age and duration of programme participation were confounded in this cross-sectional study, and the available longitudinal evidence in this review does not show a consistent developmental progression (e.g., Hartanto et al., 2019; Tonér & Nilsson, 2021). This possibility should therefore be examined in longitudinal research that can distinguish the effects of age from those of the duration and intensity of bilingual experience.

Overall, the findings do not support a consistent bilingual advantage in cognitive flexibility. The small number of studies, variation in task selection and inconsistent results across similar measures limit the conclusions that can be drawn. Further longitudinal research is needed to examine cognitive-flexibility development across age groups while distinguishing developmental change from the duration and intensity of bilingual experience.

4.2. Developmental Stage

Beyond differences across EFs, an additional pattern appeared regarding participants’ developmental stage. Following Best and Miller (2010), early childhood, middle childhood and adolescence were established as 2–6 years, 7–11 years and 12–17 years, respectively. These age-ranges were selected because they broadly reflect recognised stages of cognitive development during which EFs undergo important developmental changes (see Best & Miller, 2010). Of the 39 included studies, 18 focused on early childhood, 8 on middle childhood, 3 on adolescence (12–17 years), and 10 spanned multiple developmental stages. Within the published evidence, advantages in inhibition were more commonly observed during early childhood and in cross-sectional studies (e.g., Carlson & Meltzoff, 2008; De Cat et al., 2018; Huang et al., 2023b). Choi et al. (2018) also found that bilingual children and dual language learners with limited English skills showed faster inhibitory-control growth than English monolinguals from Head Start entry to kindergarten, although the groups did not all differ significantly at initial testing. Findings regarding WM, and to a lesser extent CF, raise the possibility that bilingual effects may become more apparent with increasing age and accumulated bilingual experience (e.g., Blom et al., 2014; Hartanto et al., 2019; Esposito & Baker-Ward, 2013; Trebits et al., 2022). However, this pattern was not consistent across studies, and the available evidence does not establish that bilingual advantages reliably emerge or increase over time. Any developmental interpretation should therefore remain tentative, particularly because the evidence was unevenly distributed across age groups and only three studies focused specifically on adolescence.

A number of explanations can be provided for this developmental pattern. Firstly, the early development of inhibition may be due to the plasticity of the brain. This allows environmental experiences, including bilingualism, to influence cognitive development. Secondly, as has been mentioned in the literature and highlighted in this review, the EFs of WM and CF may require prolonged bilingual experience before measurable advantages emerge, supporting threshold or cumulative-experience accounts (see Hartanto et al., 2019; Trebits et al., 2022). Finally, the choice of task may interact with the developmental stage of the child, since some of the tests may be too easy or too complex for children, leading to ceiling or floor effects (see Bablekou et al., 2025; Tonér et al., 2021). All in all, these findings highlight the importance of adopting longitudinal designs to track EF development over time as opposed to solely relying on cross-sectional comparisons. The methodological implications of these observations, particularly the need for longitudinal designs and greater representation of adolescent samples, are discussed in Section 5.

4.3. The Role of Bilingual Experience

The developmental patterns identified across the three executive functions suggest that bilingualism cannot be understood as a simple categorical characteristic. Rather, the reviewed evidence shows that differences in proficiency, language use, accumulated experience, and educational context help explain why bilingual advantages emerge under some conditions but not others. This makes it necessary to conceptualise bilingualism as a multidimensional and dynamic experience, shaped by degree of language proficiency, balance between languages, intensity of language use, age of acquisition, language-switching habits, interactional context, and educational setting when measuring inhibition (e.g., Santillán & Khurana, 2018), WM (e.g., Blom et al., 2014; Dicataldo & Roch, 2020; Hartanto et al., 2019; Trebits et al., 2022), cognitive flexibility (e.g., Filippi et al., 2022), and composite EF scores (e.g., Abdurokhmonova et al., 2026; White & Greenfield, 2017). This shift regarding how bilingualism is conceptualised represents one of the greatest developments within the literature, offering one of the most compelling explanations for the great fluctuation in findings observed across studies exploring the relationship between SES, EFs and bilingualism.

Across the reviewed studies, and across all three executive functions, language proficiency emerged as a stronger predictor of EF performance than bilingual status alone, with a number of studies suggesting that EF advantages are more likely to be observed among balanced or highly proficient bilinguals than among children with limited bilingual experience (e.g., Blom et al., 2014; Bosman & Janssen, 2017; Carlson & Meltzoff, 2008; Frechette et al., 2023; Ladas et al., 2015). Such findings are consistent with the ‘Threshold Hypothesis’ (Cummins, 1976, 1979), which proposes that a minimum level of bilingual proficiency and sustained dual-language engagement is required before any cognitive benefits appear. However, not all authors agree with the notion of a threshold arguing that the relation between EF and bilingualism is a gradient experience, where small, continuous changes affect EF outcomes (see Bialystok & Shorbagi, 2021; Trebits et al., 2022).

The longitudinal studies included in this review raise the possibility of cumulative effects of bilingual experience, particularly for WM, as children accumulate greater proficiency (e.g., Blom et al., 2014). For example, Hartanto et al. (2019), who tracked 18,200 children from ages 5 to 7 years, found that bilingualism significantly weakened the impact of low SES on verbal working memory by Wave 4 (age 7 years). Choi et al. (2018) similarly found faster growth in inhibitory control among bilingual children and Dual Language Learners with limited English skills than among English monolinguals from Head Start entry to kindergarten. Developmental factors also appeared to play an important role for CF, suggesting that bilingual advantages may emerge progressively and not be present during the earliest stages of bilingualism, maturing throughout childhood and adolescence (see Bablekou et al., 2025; Filippi et al., 2022). De Cat et al. (2018) further argues that this cumulative bilingual experience is of great importance, with indices like the Bilingualism Profile Index (BPI) being able to predict inhibitory performance more accurately than simple group labels like monolingual or bilingual.

This ‘cumulative bilingual advantage’ could be due to the fact that practicing language switching and management causes a daily workout for the brain, strengthening its neural base (see Abdurokhmonova et al., 2026). However, alternative explanations have also been proposed, such as the ‘Controlled-Dose (or Routinisation) Hypothesis’ (Paap, 2018), which suggests that EF control demands may be greatest during the early stages of second-language acquisition and gradually diminish as control of the two acquired languages becomes more automatic (e.g., Bablekou et al., 2025).

Beyond proficiency, the reviewed studies consistently highlight the significance of how languages are used. Dicataldo and Roch (2020) highlighted in their study the importance of moving away from binary labels to use daily exposure as a continuous predictor, which explains the significant variance in WM. Abdurokhmonova et al. (2026) further states that objectively recorded balance in children’s everyday language exposure predicted EF performance when parent-reported bilingual knowledge did not. Consequently, factors such as active bilingual language use, frequent language switching, and sustained engagement with two languages were constantly connected to stronger EF performance, more so than passive exposure alone. The impact of language use lends support to the ‘Adaptive Control Hypothesis’ (Green & Abutalebi, 2013), which suggests that continuously managing two active languages strengthens EF processes, especially when children frequently change between languages in different situations and across different communicative contexts (e.g., Filippi et al., 2020, 2022). As a consequence, both the intensity and quality of the bilingual experience seem to be stronger predictors of EF development than bilingual status itself.

As regards the type of bilingualism developed by children, studies point towards an advantage of early bilinguals. However, the longitudinal study carried out by Trebits et al. (2022) suggested that immersion and dual-language education can also bolster EF development when exposure is intensive, sustained, and cognitively demanding. These authors observed cumulative effects of education in bilingual immersion programmes, where cognitive gains regarding WM emerged after 3 to 4 years of intensive dual-language schooling and concluded that these benefits appeared regardless of family SES. However, this conclusion should be interpreted cautiously because the sample was small and SES data were available for only 27 of the 39 participants. Similarly, Esposito and Baker-Ward (2013) carried out a cross-sectional study with a socioeconomically homogeneous sample of low-SES children that found that advantages in CF emerged only after several years of intensive exposure in dual-language (50:50) classrooms, suggesting that dual-language programmes could serve as a valuable intervention strategy for students from disadvantaged backgrounds.

4.4. The Role of Socioeconomic Status

SES emerged as one of the strongest influences on EF development across the 39 reviewed studies. Instead of simply functioning as a background characteristic, SES was very commonly associated with children’s access to cognitive stimulation, educational opportunities and environmental resources, with parental education and related SES indicators frequently being identified as stronger predictors of executive-function performance than bilingual status itself (e.g., Ladas et al., 2015; Morton & Harper, 2007; Filippi et al., 2020; Buac et al., 2016; Clayden et al., 2023; Tonér et al., 2021; Tonér & Nilsson, 2021). This reinforces the importance of considering SES when studying the relationship between bilingualism and EFs.

This review thus highlights an important conceptual shift within the literature on bilingualism and EFs. Whereas early studies often treated SES as a nuisance background variable in need of statistical control, more recent research increasingly considers it a key developmental factor capable of shaping or moderating the relationship between bilingualism and EF development. This has led to studies now examining not only the independent effects of SES, but also the way in which socioeconomic circumstances as a whole influence the consequences of bilingualism on cognition. As a result, the way in which SES is included in the design can strongly influence conclusions regarding bilingual advantages, suggesting that the lack of consensus within the literature can partly be explained by methodological differences in the conceptualisation and measurement of SES.

The reviewed studies argue that the three-way relationship between SES, EFs, and bilingualism is highly complex. For instance, Calvo and Bialystok (2014) reported independent and additive contributions of bilingualism and SES to inhibition (also see Krizman et al., 2016) and WM (also see Dicataldo & Roch, 2020). However, several studies also reported bilingual advantages within socioeconomically disadvantaged populations, including advantages in inhibition, WM, CF, or composite EF performance (Choi et al., 2018; Engel de Abreu et al., 2012; Huang et al., 2023b; Santillán & Khurana, 2018; White & Greenfield, 2017). However, because these studies sampled a restricted socioeconomic range, they demonstrate that bilingual advantages can occur within low-SES populations rather than establishing that bilingualism buffers SES-related disadvantages.

Evidence directly supporting the buffering hypothesis was reported by Hartanto et al. (2019) and Filippi et al. (2022), who formally examined interactions between bilingualism and SES. Hartanto et al. (2019) found that bilingualism generally attenuated SES-related differences in CF, while its buffering effect on verbal WM emerged only at the later assessment, at approximately age seven. These findings led Hartanto et al. (2019) to suggest that lower-SES children may be particularly responsive to the benefits of bilingual experience because they otherwise had comparatively fewer opportunities for cognitive development. Filippi et al. (2022) similarly found that a verbal WM advantage emerged among low-SES multilingual adolescents later in development, at ages 13–15 years, potentially alongside increasing second-language proficiency. These findings further suggest that the relationship between bilingualism and SES may change over time.

Conversely, several studies found that bilingual advantages disappeared after rigorously controlling for or matching SES (e.g., Bablekou et al., 2025; Esposito, 2021; Ladas et al., 2015; Filippi et al., 2020; Morton & Harper, 2007; Panesi et al., 2026). Abdurokhmonova et al. (2026) found that balanced naturalistic bilingual exposure predicted better behavioural EF performance but did not buffer the effect of low SES, while Perry et al. (2026) found no effect of multilingual status on longitudinal change in verbal WM. Others proposed that favourable socioeconomic environments may act as enabling conditions for bilingual cognitive benefits to emerge (e.g., Bialystok & Shorbagi, 2021). Taken together, these findings indicate that the interaction between bilingualism and SES is dynamic and context-dependent rather than uniformly compensatory.

Lastly, methodological differences in the measurement of SES further contribute to the heterogeneity of the findings. A wide range of factors and indicators were used throughout the reviewed studies to measure SES, including parental education, occupation, household income, neighbourhood characteristics, and composite indices. Increasingly, authors argue that SES should not be conceptualised as a simple categorical variable (high- vs. low-SES), but as a continuous and multidimensional construct including educational, occupational, financial, and environmental resources (e.g., Hartanto et al., 2019).

5. Priorities for Future Research and Limitations of the Current Evidence

The present review highlights a number of important limitations that, according to the authors, continue to hinder a homogeneous view of the SES-EF-bilingualism relationship. Consequently, to provide a more comprehensive understanding of the conditions under which bilingualism and SES interact to influence EF development, it is essential to address these limitations by means of more rigorous methodological approaches and study designs.

As was mentioned above, highly diverse definitions and indicators were used across the reviewed studies, which limited the possibility of comparison across investigations. Authors reviewed in this study suggest looking at the role of environmental factors—including bilingualism—in the development of language and cognitive skills in a more continuous, demographically representative manner (Buac & Kaushanskaya, 2014). It is also suggested that research moves away from between-group comparisons and rather uses matching procedures where subjects are matched by demographic variables to allow for comparison of results in the tests (e.g., Czapka et al., 2020).

The assessment of EFs is another important limitation. Despite the considerable attention given to inhibitory control and WM, comparatively little importance has been given to CF. Regarding test structure, many existing studies continue to rely on single-task measures to measure the three core EFs despite the well-recognised problem of task impurity. Future investigations should consequently use broader executive-function batteries, latent variables and more ecologically valid assessments that better capture children’s cognitive functioning in everyday contexts (e.g., Bablekou et al., 2025; Czapka et al., 2020; Tonér & Nilsson, 2021).

Another major priority for future research is the use of longitudinal research designs. While cross-sectional studies provide valuable insights into group differences, there is still very limited information regarding developmental trajectories or causal mechanisms, which this review has highlighted as highly consequential (e.g., Clayden et al., 2023; Santillán & Khurana, 2018). Longitudinal studies are fundamental to examine how bilingual experience accumulates over time (the cumulative experience), the correlation between proficiency and EF advantage, and how bilingualism and SES conditions interact throughout development and across age ranges (e.g., Bablekou et al., 2025; Krizman et al., 2016). Such designs are also essential to clarify the still ongoing debate of whether bilingualism influences EF growth or whether children with stronger executive functions are more likely to become proficient bilinguals.

Future research should also address the uneven representation of developmental stages within the literature. Most studies focus on early and middle childhood, with relatively few studies exploring adolescence, strongly limiting our understanding of how bilingualism and SES influence EF development beyond primary school. Providing greater representation across all developmental stages, alongside longitudinal designs, would enable researchers to determine more precisely whether bilingual advantages are temporary and limited to certain age ranges, remain stable or increase throughout childhood and adolescence. This would also serve to identify which periods would benefit the most from educational interventions and bilingual programmes, particularly among children from socioeconomically disadvantaged backgrounds.

Several authors in the review have also called for the imperative need to broaden the nature of study samples (e.g., Bablekou et al., 2025; Clayden et al., 2023; Frechette et al., 2023; Huang et al., 2023a). Most of the existing evidence comes from Western contexts, therefore representing specific types of bilingual populations, mostly Spanish–English speakers. Greater attention should be given to different cultural contexts, such as children living in non-Western settings with different language combinations, refugee populations, and other underrepresented groups.

Limitations of the Present Review

Limitations of the present review should also be acknowledged. Firstly, even though the research strategy used tried its best to identify all relevant studies in multiple databases and through a citation search, it is possible that some studies were not identified because they were either not indexed in the selected databases or not findable through the selected search strategy. Thus, evidence that might have been relevant may not have been included. The same is true for the decision to exclude books and book chapters from the search. The aim was to ascertain that the studies were peer-reviewed, but depending on publishing policies, this may also be true for books and book chapters. While acknowledging these limitations, the review nevertheless provides a first overview of the current evidence and highlights key patterns, methodological challenges, and fundamental considerations for future research.

As regards the fact that studies selected had to include a monolingual comparison group, this may have led to the exclusion of 22 studies, some of which could have proven relevant, and that were probably more careful in the description of their samples. As has been mentioned throughout the present study, one of the problems in research in the field of bilingualism that has increasingly been criticised is the fact of using language status as a categorical variable (monolingual vs. bilingual) rather than as points on a gradient, reflecting the reality of language exposure and use in more realistic ways. In future studies this exclusion criterion should be modified to better reflect the relative, continuous nature of language status.

A further limitation concerns publication bias. Since the review was restricted to published, peer-reviewed studies, its findings may overrepresent evidence supporting a bilingual advantage and underrepresent null or negative findings. De Bruin et al. (2015) found that 68% of conference studies reporting evidence in favour of a bilingual advantage were subsequently published, compared with 29% of studies reporting no bilingual effect or a bilingual disadvantage. Accordingly, the advantage, mixed, and null classifications reported in this review should be understood as a descriptive mapping of the published evidence rather than as an estimate of the prevalence or strength of bilingual advantages across all research conducted in the field.

Finally, the uneven distribution of studies across developmental stages represents an important limitation of the review. Whereas 18 studies focused on early childhood, only three specifically examined adolescence. Although ten studies included multi-stage samples, the inclusion of several age groups does not necessarily mean that developmental differences were formally analysed. The limited evidence on adolescents restricts the extent to which conclusions can be drawn about the relationship between bilingualism, SES, and executive functioning during this developmental period. Future research should therefore prioritise adolescent samples and employ longitudinal designs capable of examining whether the relationship between bilingualism, SES, and the core executive functions changes across development.

6. Practical Educational Implications

This review suggests some implications for educational policy. Even if evidence does not support a universal bilingual advantage, it has been suggested that bilingualism can represent a valuable cognitive resource, particularly for children from socioeconomically disadvantaged backgrounds. For this reason, both educators and policymakers should be aware of the potential of growing up with more than one language to support EF development within disadvantaged SES contexts rather than, as sometimes occurs, viewing bilingualism as a risk factor.

One of the strongest implications relates to bilingual education. Two studies in this review have set forth that structured bilingual programmes such as dual-language and immersion programmes can provide cognitively enriching environments that not only develop language learning but also enhance EF development (Esposito & Baker-Ward, 2013; Trebits et al., 2022). Esposito and Baker-Ward (2013) reported CF advantages within a low-SES sample, whereas Trebits et al. (2022) observed WM gains that appeared across socioeconomic backgrounds. However, the evidence base remains thin, consisting of one cross-sectional study with 113 participants and one longitudinal study with 39 participants. These findings should therefore be regarded as preliminary. Although they do not establish that participation in bilingual education causes improvements in EF or reduces SES-related inequalities, they raise the possibility that sustained bilingual educational experience may support EF development under certain conditions and identify bilingual educational settings as a promising context for larger, controlled longitudinal studies. Although not part of the evidence base included in this review, Dosi and Papadopoulou (2020) provide relevant contextual evidence suggesting that educational setting may form an important dimension of bilingual experience, with differences in the balance, quality, and active use of the two languages across school contexts being associated with performance.

Additionally, even though their studies were not carried out specifically in language immersion programmes, the results provided by authors like Blom et al. (2014) and Hartanto et al. (2019) further suggest the potential of bilingualism to level the playing field for low-SES students regarding EF development. However, despite these promising findings, it is important to highlight that bilingual education should not be seen as a universal solution to the educational disadvantages stemming from low-SES, but rather as a tool that, in combination with broader educational, social, and environmental support, can particularly benefit children growing up in disadvantaged circumstances. Also, authors like Filippi et al. (2022) and Trebits et al. (2022) actively advocate for the maintenance of heritage languages, encouraging educators to both support and include them in their classroom.

Finally, these findings could also have implications for assessment. Several studies encourage considering bilingual profiles when assessing, instead of using a single language to measure performance and interpreting lower scores as a result of cognitive or linguistic impairment, since bilingual children often distribute their knowledge across two languages (see Tonér & Nilsson, 2021). It is also proposed that, when possible, cultural and linguistically appropriate measures should also be taken into consideration to minimise background bias.

Overall, the findings propose that educational policies and practices should move beyond asking whether bilingualism is beneficial and instead focus on creating environments that support the low-cost, high-impact intervention that some authors consider bilingualism to be (see Engel de Abreu et al., 2012). Ideally, such an approach could not only promote multilingualism but also contribute to reducing SES inequalities in education, thus helping to narrow the achievement gap.

7. Conclusions

This scoping review was guided by the following research question: What evidence does the literature provide about the potential relation between bilingualism and the development of the three core executive functions (inhibition, working memory, and cognitive flexibility) across different socioeconomic contexts in children and adolescents? The findings indicate a more complex relationship than a simple bilingual advantage framework would suggest: rather than supporting a single, consistent effect of bilingualism, they reveal substantial variability in the development of inhibition, working memory, and cognitive flexibility. Of the three core EFs, inhibition was the most frequently examined and produced the largest number of studies reporting a bilingual advantage; however, findings remained heterogeneous form all three EFs.

This variability can be related to differences in the research methodology used, the developmental stage of the participants and the definition and measurement of bilingualism and SES. Indeed, one of the main findings of this review is the need to look at the two variables of bilingualism and SES as multidimensional and placed on a continuum rather than as categorical variables. On the other hand, the duration of participants’ exposure to two languages has been found to have an impact on the results of the studies, which calls for longitudinal research to confirm this pattern.

As regards the effect of bilingualism and whether it can potentially act as a protective factor against the negative effects associated with low SES, evidence is not conclusive. Further research is needed in this area to establish if, when, for whom and under what conditions bilingualism may influence executive-function development. This knowledge can then inform educational policies and interventions aimed at reducing SES-related inequalities and promote more equitable educational opportunities.