Relationship between language switching experience and executive functions in bilinguals: an Internet-based study^*

* The research was conducted at the Abo Akademi University, Finland.

ABSTRACT

We examined the relationship between self-reported everyday language switching experience and the performance of early bilinguals in tasks measuring different executive functions. Our participants were Finnish–Swedish early bilinguals, aged 16–41 years (N = 66, Experiment 1) and 18–69 years (N = 111, Experiment 2). An earlier study using a sample from a similar population discovered a negative relationship between self-reported language switching and a mixing cost in error rates in a number–letter task. This finding was not replicated. Instead, we found that a higher rate of reported contextual language switching predicted larger switching cost reaction times in the number–letter task, and that a higher rate of reported unintended language switches predicted larger error rates in a spatial n-back task. We conclude that these results likely reflect individual differences in executive skills, and do not provide evidence for the hypothesis that language switching trains executive functions.

KEYWORDS:

• Bilingualism
• executive functions
• language switching
• Bilingual Switching Questionnaire

Acknowledgements

We would like to thank the BrainTrain research group at the Abo Akademi Psychology Department for discussions on the manuscript, and Daniel Wärnå and Tuomas Pellonperä for programming the Internet testing platform.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

* The research was conducted at the Abo Akademi University, Finland.

1 If we examine the individual predictors, we see in the model with the BSWQ variables as predictors that Contextual switches correlated negatively with the Simon effect reaction time, and Language switches correlated negatively with the N-back effect error rate. In the current type of analysis, however, many pairwise comparisons are conducted, which greatly increases the risk of Type I error. The overall model F is less affected by multiple comparisons, as it gives an overall estimate of the model fit. Here we apply the conservative strategy of examining individual predictors only if the overall model F is significant.

2 As noted by an anonymous reviewer, these interpretations rest on what the BSWQ factors are supposed to measure. The face validity of the individual questions assumedly measuring unintended switches or contextual switches could be questioned. For instance, item 3 in the BSWQ (“I tend to switch languages during a conversation”), which is assumed to measure contextual switches, may not distinguish between contextually driven switches and intentional switches. On the other hand, the individual questions presumably measuring unintended switches have higher face validity; for example, item 8: “It is difficult for me to control the language switches I introduce during a conversation”.

Additional information

Funding

This work was supported by the Emil Aaltosen Säätiö; University of Helsinki; Academy of Finland under grants [260276] and [288880]; the Abo Akademi University, and the Abo Akademi University Endowment (grant to the BrainTrain project); Kulttuurin ja Yhteiskunnan Tutkimuksen Toimikunta.