L2 Learners’ Perceptions and Preferences of Automated Corrective Feedback

Abstract

<p>Form-focused automated Corrective Feedback (CF) is widely used in general-purpose and specialized writing software and research indicates a positive effect for automated CF on second language (L2) learning (AbuSeileek, 2013; AbuSeileek & Abualsha’r, 2014). As any educational practice, preferences and perceptions of learners are expected to influence how automated CF is used by L2 learners (Amrhein & Nassaji, 2010; Brown, 2009; Schulz, 2001). However, preferences and perceptions of automated CF are poorly understood due to paucity of relevant research. This study contributes to this line of investigation by exploring three pertinent topics. First, it explores L2 learners’ preferences and perceptions of automated CF and four different CF strategies: identification, direct correction, metalinguistic CF, and graduated CF. Second, it examines learners’ preferences between different CF timings and frequency choices. Third, it explores learners’ past experience with AWE tools and investigates if past Automated Writing Evaluation (AWE) experience affects learners’ preferences and perceptions of automated CF.</p> <p>To accomplish these objectives, the present study surveyed and interviewed 30 learners at an intermediate to advanced English as Second Language (ESL) proficiency level. It calculated descriptive statistics of the surveys and employed exploratory factor analysis to identify the underlying relationships between different variables measured by the survey. For interview analysis, it employed a grounded theory approach to identify major concerns and perceptions of automated CF not accounted for in the survey. Results revealed a strong preference for direct correction followed by metalinguistic CF, identification, and graduated CF respectively. Factor analysis identified a close association between clarity and usefulness perceptions and preferences for CF strategies, between comprehensive CF and direct correction and between the frequency of AWE use and identification. The interviews revealed two major concerns with potential influence on preferences for CF strategies: time and learning. Based on preference data, the time required to use CF successfully for error correction is a more important factor for most learners than learning from CF. In other words, CF strategy preferences appear to be mainly shaped by the time factor. This dissertation concludes with specific implications of these findings for developers of AWE tools and L2 educators. Specifically, developers should be mindful of the wide range of concerns that shape L2 learners’ preferences and perceptions of CF in order to design and deliver CF that is timely, desirable, and positively perceived by L2 learners. Furthermore, L2 educators should exert the effort to mitigate L2 learners’ concerns that undermine the value of CF qualities and strategies that were empirically proven to be effective for L2 development.</p>