Key points are not available for this paper at this time.
Scaffolding is crucial for supporting effective corpus use among low-proficiency second-language (L2) English learners, who often face cognitive overload during corpus-based vocabulary learning. This study designed a custom GPT and incorporated it into a corpus-based language pedagogy (CBLP) framework, using six scaffolding methods to address three key challenges faced by low-proficiency L2 English learners. A mixed methods case study compared two within-group conditions: CBLP with and without custom GPT scaffolding. Multiple data were collected, including vocabulary tests, GPT chat logs, and semi-structured interviews. The results show that GPT-scaffolded CBLP effectively improves vocabulary knowledge in low-proficiency learners, with the greatest gains in use, followed by form and meaning. Analysis of GPT chats revealed that the custom GPT was capable of delivering differentiated and synergistic scaffolds. Thematic analysis of interview transcripts further showed that GPT support fulfilled key scaffolding functions and embodied the characteristic of contingency. By integrating qualitative and quantitative findings, we found that the custom GPT provided adaptive scaffolding during learner-initiated interactions, especially in developing learners’ vocabulary use. Several implementation challenges were also identified. Based on these findings, the study offers pedagogical implications and suggests directions for future research to implement custom GPT-scaffolded CBLP in larger-scale classroom environments.
Jing et al. (Fri,) studied this question.