Date of Award
Doctor of Philosophy
Applied Linguistics and Technology
This dissertation outlines the development of a system of corpus-based feedback and evaluates its effects on students’ error correction and language learning. The design of the corpus-based feedback used in the study was informed by two SLA theories: the Interactionist Approach and Skill-learning Theory. The dissertation theorizes all aspects of the feedback design and discusses the various features of corpus-based feedback from SLA theoretical perspectives. To evaluate the quality of corpus-based feedback, the study compared it to that of traditional coded feedback. Based on the logical framework for evaluating CALL corrective feedback and Chapelle’s evaluation criteria (2001), five research questions were developed concerning three qualities of CALL corrective feedback: language learning potential, learner fit, and impact.
To address the five research questions concerning language learning potential, learner fit, and impact, the study employed a mixed-methods approach to collect quantitative and qualitative data from 90 participants to evaluate and compare the effects of corpus-based and coded feedback. The quantitative data consisted of scores measuring the immediate effects of feedback in error correction exercises, scores measuring learning gains demonstrated by the pre- and post-tests, and students’ ratings of the effects of feedback in Likert-scale questionnaires. The qualitative data consisted of think-aloud exercises and semi-structured interviews, which played an important role in describing participants’ learning experiences with the intervention.
Compared to students using traditional coded feedback, students treated with corpus-based feedback had greater success correcting errors in two error correction practice exercises and transferring knowledge to correct errors in a new text during the post-test. The benefits of the corpus-based feedback were also confirmed by students in the think-aloud activities and interviews. In addition, the data yielded strong evidence that advanced-low ESL learners found corpus-based feedback to be more appropriate than coded feedback for enabling correction of syntactic errors; students perceived that corpus-based feedback exerted a more positive impact on their affective, cognitive, and intrinsic development towards error correction than the coded feedback. The research methodology and findings of this dissertation make extensive contributions to corrective feedback research and the utility of theoretical approaches in tool development and learning evaluation.
Chen, Mo, "Developing and evaluating corpus-based feedback" (2019). Graduate Theses and Dissertations. 17162.