Proceedings of the Annual Meeting of the Cognitive Science Society

Artificial language studies using reaction time-based measures have suggested grammar learning even in participants with-out awareness of underlying grammatical rules (Leung & Williams, 2011; Batterink, Reber, & Paller, 2014). However,traditional linear analyses of reaction times might not capture qualitative differences between participants with/withoutconscious rule awareness (Rouder, Lu, Speckman, Sun & Jiang, 2005; Rousselet & Wilcox, in press). In a partial repli-cation of one study (Batterink et al., 2014), participants were exposed to pseudoword articles that were predictive of anaccompanying English noun’s living/non-living status. Linear analyses showed that both rule-aware and rule-unaware par-ticipants exhibited slowdowns to rule-violating trials, indicating grammar learning. Hierarchical Weibull distribution anal-yses suggested that rule-unaware and rule-aware participants differed in the underlying cognitive mechanisms involved:rule-violating trials affected the processing architecture for both groups but only affected processing speed for rule-awareparticipants. These results illustrate the potential of yet-underused distribution-modeling approaches for second languagepsycholinguistics.