Wiechmann, D. 1 , Kerz, E. 2 & Terhorst, D. 2
1 University of Amsterdam
2 RWTH Aachen University
The related fields of statistical learning (SL) and implicit learning (IL) research have demonstrated that humans can acquire regularities embedded in input upon brief incidental exposure. While traditionally IL is considered to be not as susceptible to individual differences (IDs) as explicit learning (Reber et al. 1991), recent studies using semi-artificial language systems have begun to explore potential relationships of IL with ID measures, notably working memory (WM) (e.g., Grey et al. 2015; Tagarelli et al. 2016). WM accounted for little of the variance in learning outcomes. However, these studies employed measures of WM that were developed in accordance with the view of WM as capacity. We report on two crowdsourcing experiments (N = 200 participants) based on a semi-artificial language system that investigated whether the IL of L2 morphosyntax interacts with WM as controlled attention (Engle 2001). Learning outcomes were assessed using an acceptability judgment task and a two-alternative forced choice task (2AFCT). Attentional control was measured by a Stroop color-word test. We found the IL effect only in the 2AFCT, suggesting that learning was modulated by task demands. Crucially, we found that learning outcomes were predicted by Stroop task performance, suggesting that IL is mediated by attentional processes.