This paper presents evidence that dialect switching can pose a variable cognitive load that modulates success in verbally mediated tasks. A Bayesian Markov Chain Monte Carlo model is used to explore and confirm the hypothesis that the morphosyntactic organization of African American English (ME) has significant, variable effects on second grade African American students' performance on mathematical reasoning tests conducted orally in Mainstream American English (MAE). These effects correlate with students' productions of AAE. Neither measures of spatial reasoning nor span measures of children's working memory correlated with this aspect of test performance, but certain types of representational mismatches did. These findings are consistent with other work suggesting that mathematical reasoning and language draw from a common working memory store, and that processing difficulties are linked to manipulating representations rather than limits on storage capacity.
- working memory
- Markov Chain Monte Carlo model
- dialect switching
- African American English