Effects of Corpus-Based Instruction on Students' Academic Writing Development and Their Perceptions toward Learning with Corpora: A Focus on Self-Compiled Corpora

Abstract

This study aimed to examine the effect of corpus-based instruction (CBI) on EFL learners’ academic writing development and their perceptions toward CBI, on third-year Software Engineering students (N = 50) of Woldia University, Ethiopia in 2024. Mixed-methods approach with an interrupted time series quasi-experimental research design was employed. For the 36-hour classroom intervention, a self-compiled corpus of 200 IELTS Band 9 essays was exploited with AntConc software. The data gathering tools used in the study were Test scores, a questionnaire, and semi-structured interviews. Using an explanatory sequential data analysis technique, a paired samples t-test, percentage, and mean were used using SPSS version 26 for quantitative data and thematic analysis for qualitative ones. The analysis revealed significant statistical differences in all pairs of posttest scores at a value of p=.000. Besides, low, medium, and high scorers’ posttest writing performance showed a mean difference of 37.35, 29.41, and 14.82, respectively, from their initial test achievements. These findings indicate that CBI effectively enhanced academic writing, especially for low and medium scorers. Participants also expressed positive views on CBI’s role in grammar and vocabulary development, despite its moderate impact on task achievements and text coherence. However, further studies shall be done on different academic writing genres.