Given the small mean effect size and small-study effects, this meta-analysis gives little support for a bilingual advantage on overall executive function

Gunnerud, H. L., ten Braak, D., Reikerås, E. K. L., Donolato, E., & Melby-Lervåg, M. (2020). Is bilingualism related to a cognitive advantage in children? A systematic review and meta-analysis. Psychological Bulletin, Sep 2020. https://doi.org/10.1037/bul0000301

Abstract: Bilingual people are often claimed to have an advantage over monolingual people in cognitive processing owing to their ability to learn and use two languages. This advantage is considered to be related to executive function (EF). However, no consensus exists as to whether this advantage is present in the population or under which conditions it prevails. The present meta-analysis examines the bilingual advantage in EF of children aged 18 years and under for different components of inhibition (hot; rewarding stimuli/cold; neutral stimuli), attention, switching, monitoring, working memory, and planning in 143 independent group comparisons comprising 583 EF effect sizes. The bilingual advantage in overall EF was significant, albeit marginal (g = 0.06), and there were indications of publication bias. A moderator analysis showed significant group differences on EF in favor of bilinguals for studies of children from middle-class socioeconomic backgrounds and studies from one specific lab. The EF components of cold inhibition, switching, and monitoring expressed significant bilingual advantages, but monitoring and cold inhibition were affected by publication bias. As for switching, this remained significant after controlling for publication bias. Thus, given the small mean effect size and small-study effects, this meta-analysis gives little support for a bilingual advantage on overall EF. Still, also after the moderator analysis, there was a large heterogeneity of true effects and a large amount of unexplained heterogeneity in the effect sizes. Thus, there might be bilingual advantages (or disadvantages) under conditions that this study is not able to identify through the analysis of 12 moderators