Studies investigating the effects of bilingualism on cognitive function have often yielded conflicting results, which may stem in part from the use of arbitrary criteria to categorize participants into groups based on language experience. The present study addresses this limitation by using a machine learning algorithm, known as cluster analysis, to identify naturally occurring subgroups of participants with similar language profiles. In a sample of 169 participants with varying degrees of first- and second-language proficiencies and ages of acquisition, the cluster analysis yielded four bilingual subgroups: late-unbalanced, early-unbalanced, late-balanced, and early-balanced. All participants completed the NIH Toolbox Cognition Battery. Results revealed that early-balanced and early-unbalanced bilinguals scored higher than late-unbalanced bilinguals on the cognitive flexibility and inhibitory control subtests of the NIH Toolbox Cognition Battery, whereas late-unbalanced bilinguals scored higher than early-balanced bilinguals on the verbal working memory subtest of the NIH Toolbox Cognition Battery. Bilingual language experience did not impact performance on measures of processing speed, episodic memory, and English vocabulary. These findings demonstrate the utility of data-driven approaches to capture the variability in language experience that exists in the real world. We conclude that different bilingual experiences can shape a wide range of cognitive abilities, from working memory to inhibitory control.
Read full abstract