Megan Gross & Margarita Kaushanskaya (University of Wisconsin – Madison)
Language control; Cross-language errors; Task-shifting; Mixing costs; Switching costs; Executive function; Child bilingualism; English-Spanish; Picture naming
To communicate effectively with a variety of conversation partners, bilinguals must develop language control, the ability to adjust their language choice based on the situation. This ability to switch between languages and control interference from the language-not-in-use has been associated with task-shifting in adult bilinguals (e.g., Prior & Gollan, 2011). However, it remains unclear at what level of lexical selection domain-general task-shifting skills may be recruited for language control. The goal of the current study was to examine the relationship between the shifting component of executive function and language control, as measured at different levels, in bilingual children.
The study included 68 English-Spanish bilingual children, ages 5-7. Language control was assessed with a timed picture-naming paradigm. Children named pictures in English or Spanish, according to an auditory cue, in a single-language context (English-only; Spanish-only) and a mixed-language context in which the cued language varied unpredictably. Language control was measured at two levels. Cross-language errors, when children responded in the opposite language from the cue, indicated a failure of language control. For correct responses, increased naming latency indexed how taxing it was for children to successfully exert language control. At both levels, two kinds of costs were assessed: mixing costs (performance in a mixed-language vs. single-language context) and switching costs (performance when switching languages vs. staying in the same language during the mixed condition). The shifting component of executive function was measured with the Dimensional Change Card Sort (DCCS; Zelazo, 2006). The DCCS required children to sort colored squares and circles first by color, then by shape, and then to switch between sorting rules during a mixed phase. Accuracy during the mixed phase indexed shifting.
Mixed-effects analyses of cross-language errors revealed both mixing costs (b=2.30, t=2.81), and switching costs (b=1.08, t=2.56). Children were more likely to make cross-language errors in a mixed-language than a single-language context, and within the mixed context they were more likely to make cross-language errors when they were cued to switch languages. Children with better shifting (DCCS performance) were less likely to produce cross-language errors overall (b=-0.056, z=-5.48); this effect did not vary by context or switch status. Naming latency analyses revealed significant mixing costs (b=0.19, t=6.64) but no significant switching costs. The effect of DCCS performance on naming speed was modulated by an interaction with context (b=0.003, t=1.97), with a significant effect of DCCS in the single-language but not the mixed-language context. DCCS performance did not explain any additional variance in the switch-cost analysis of the mixed-language condition.
Together, these findings suggest that nonlinguistic shifting skills contribute to language control at the level of selecting the correct language. Once the correct language is selected, more nuanced costs in speed may be localized to the language system.