Explicitness of recasts, learner responses, and l2 development of Korean relative clauses: an experimental study

Abstract

In the previous research on recasts, two interesting claims have been made. First, the absence of learners' immediate responses following recasts may limit the effectiveness of recasts, and therefore other interactional feedback moves that encourage uptake, such as prompts or negotiation, would be more beneficial for learners than recasts. In response to this claim, some researchers have argued that recasts might have delayed effects on L2 development, and thus the efficacy of recasts should not be discounted due to the lack of immediate responses (Mackey & Philp, 2003; McDonough & Mackey, 2006). The second claim is that, among the many operational definitions of recasts utilized in previous studies, the more explicit ones may be more effective than the more implicit ones (Ellis & Sheen, 2006; Nicholas, Lightbown, & Spada, 2001), since the former lead to a higher rate of learner uptake and repair (Loewen & Philp, 2006; Sheen, 2006) as well as greater noticing of the recasts (Egi, 2007; Kim & Han, 2007). These two issues are investigated in this study examining the effects of explicit (i.e., declarative) and implicit (i.e., interrogative) recasts on L2 development of Korean relative clauses. In total, 63 KSL/KFL learners at the beginning-high to intermediate level participated in this study. The impacts of explicit and implicit recasts are examined with respect to various learner responses in the discourse (i.e., immediate uptake and primed production) as well as subsequent L2 development, measured on two different occasions (i.e., immediate and delayed posttests) by employing three different tasks (i.e., an oral production task, a sentence combination task, and a grammaticality judgment task). The results showed that all three groups (i.e., declarative, interrogative, and control groups) significantly improved over time as a result of the interaction they engaged in during the experiments, regardless of the explicitness of the recasts and even the presence or absence of recasts. However, significantly higher repair rates followed declarative recasts, supporting the argument that declarative recasts are more explicit than interrogative ones. With regard to the relationships between the various learner responses and L2 development, only primed production was significantly associated with posttest scores.