Degree

Doctor of Philosophy (PhD)

Department

The Lutrill & Pearl Payne School of Education

Document Type

Dissertation

Abstract

This study examined the short-term effects of access to a custom AI-supported translanguaging writing assistant on English as a Second Language (ESL) students’ academic writing performance and metacognitive awareness. It also explored how students perceived using the assistant for course-related writing and for planning, monitoring, and evaluating their learning. The study addressed a gap in research at the intersection of translanguaging pedagogy, generative artificial intelligence, and metacognitive awareness in ESL writing instruction.

The study used a quasi-experimental design with a supplemental qualitative interview component. Participants were adult international students enrolled in Level 5 and Level 6 writing courses in an intensive English program in the western United States. The final quantitative analytic sample included 43 students, with 20 students in the intervention group and 23 students in the comparison group. Students in the intervention group received access to a custom AI-supported translanguaging writing assistant built on the ChatGPT platform. Students in the comparison group completed the same writing curriculum without access to the custom assistant.

Quantitative data included pre-intervention and post-intervention writing scores based on the program’s analytic writing rubric and pre-test and post-test scores on the Metacognitive Awareness Inventory. The main analyses used analysis of covariance to compare post-intervention outcomes while controlling for baseline scores and course level. Supplemental analyses included rubric-category and MAI-component analyses, gain-score checks, and mixed-effects sensitivity models. Qualitative data came from semi-structured interviews with five intervention-group students and supplemental in-tool feedback.

The quantitative findings did not show statistically significant short-term differences between the intervention and comparison groups in writing performance or metacognitive awareness. The qualitative findings showed that students generally perceived the assistant as useful for writing-process support, including planning, organization, vocabulary, grammar, sentence clarity, translation, examples, and revision. Students also described selective use of English and their first languages, continued ownership of their writing, and critical judgment of AI responses. Overall, the findings suggest that the custom assistant functioned more as a mediational writing-process tool than as a short-term achievement intervention.

Date

7-10-2026

Committee Chair

Jennifer Qian

LSU Acknowledgement

1

LSU Accessibility Acknowledgment

1

Download

Available for download on Friday, July 08, 2033

Share

COinS