Individual Differences in Rule-Based Grammaticality Judgment Behavior: A Bayesian Modeling Approach

Abstract

　　The aim of the present study is to examine the source of individual differences in the rule-based grammaticality judgment behavior of foreign language learners. It has been recognized that learners with certain specific traits benefit relatively more from rule-based knowledge representations in their grammaticality judgment performance, while another type of learner tends to rely on their unconscious and non-symbolic type of knowledge, which is often called intuition. Generally, the former type of learner exhibits a relatively higher discriminability (d') in a grammaticality judgment task than the latter type. However, there remain individual differences unexplained in such rule-based grammaticality judgment behavior. The present study focused on grammatical carefulness (GC) as a key factor behind the variance inspected, as GC is considered as one of the latent variables associated with the speed-accuracy tradeoff and meta-linguistic behavior. Using the type II signal detection paradigm and Bayesian modeling, the present study, which targeted Japanese university students (N = 40) who engaged in a grammaticality judgment task (K = 48), found that learners with higher GC scores are more prone to benefit from rule-based knowledge representations than those with lower GC scores, who benefit from their intuition. It indicates that individuals' psychological, behavioral, and meta-cognitive traits moderate grammaticality judgment behavior in a foreign language. The practical implications concerning testing and assessing grammatical performance in a foreign language were also discussed.