Seamus Donnelly, Patricia Brooks, Bruce Homer (City University of New York, Graduate Center)
sdonnelly@gc.cuny.edu
Bilingualism; executive control; inhibitory control; meta-analysis; multi-level models.
An open question in cognitive science is whether and under what circumstances bilinguals outperform monolinguals on interference control tasks. Results have been mixed on both interference costs (Hilchey & Klein, 2011) and global RTs (Paap & Greenberg, 2013). One explanation for this variability is variation across studies in tasks, age and labs. This poster reports on a meta-analysis synthesizing these studies and testing the effects of four potential moderators (described below). The analysis contains 73 comparisons from published studies, reflecting 5538 unique participants.
Moderators tested include RT cost, task, participant age, and lab. RT cost refers to whether the effect size reflects global RT or interference cost. These costs are thought to reflect separable components of executive function and are implicated in different models of bilingual language processing (Hilchey & Klein, 2011). Task refers to the type of interference control task included, Simon, Flanker, Stroop or other. While these tasks are all considered interference control tasks, they are often uncorrelated (Paap & Sawi, 2014). Age was coded categorically and includes three groups: children, younger adults and older adults. Finally labs that contributed more than four effect sizes were included as separate dummy coded variables.
The effects of these moderators were tested in a series of three-level meta-analyses. Three-level meta-analysis explicitly models the dependence between effect sizes from the same sample and was necessary since most samples contributed effect sizes for global RT and interference cost. A first model, which included no moderators, yielded an effect size of d = .39 (CI: .19 – .59). A second model, which included RT cost as a moderator, did not significantly improve fit (p = .52). However, as it is plausible that the effects of other moderators might be different for global RTs and interference costs, RT cost was included in all subsequent models, despite being non-significant on its own. Subsequent models revealed no main effect of task, and no interaction between task and RT cost. There was no main effect of age, but a significant interaction between age and RT cost; for older adults differences in interference costs were larger than global RTs while for children differences in global RTs were larger than differences in interference costs. There was a significant main effect of lab, but it did not interact with cost.
The significant interaction between age and RT cost is difficult to interpret. The significant main effect for lab may be driven by differences in subject pools, differences in defining and handling outliers in RT distributions or other methodological factors.
References
[1] Hilchey M, Klein R. Are there bilingual advantages on nonlinguistic interference tasks? Implications for the plasticity of executive control processes. Psychonomic Bulletin & Review [serial online], 18(4):625-658, 2011.
[2] Paap K, Greenberg Z. There is no coherent evidence for a bilingual advantage in executive processing. Cognitive Psychology [serial online], 66(2):232-258, 2013.
[3] Paap, K. R., & Sawi, O. Bilingual advantages in executive functioning: problems in convergent validity, discriminant validity, and the identification of the theoretical constructs. Frontiers in psychology, 5, 2014