A Computational Modeling of Statistical Learning for Island Effects by Korean L2 Learners of English

Abstract

Kim, Euhee, Whang, Man-Soo, and Park, Myung-Kwan. 2015. A Computational Modeling of Statistical Learning for Island Effects by Korean L2 Learners of English. Korean Journal of English Language and Linguistics 15-1, 043-073. This paper examines L2 learning of the well-known (syntactic island constraints on) wh-dependencies in English. Following the methodologies that Sprouse, Wagers, and Phillips (2012) and Pearl and Sprouse (2013) use to explicate the nature of the L1 acquisition of such dependencies, we first investigate L2 learners' acceptability ratings on them. We then examine the correlation between the perceived strength of island effects (i.e., island violations) and L2 learners' working memory capacity. We then move on to design and implement a computational learning algorithm capable of successfully learning the pattern of L2 learners' acceptability ratings on wh-dependencies, building on realistic input available to the learners. We finally discuss the kinds and natures of the learning biases/predispositions that make the proposed learning algorithm work in accounting for the target state of L2 learning on wh-dependencies.