Speaker
Description
AI language models like ChatGPT can serve as automated tools for essay scoring, potentially transforming evaluation and feedback processes in both scholarly and practical contexts (Mizumoto & Eguchi, 2023). However, without specific training for generating feedback, ChatGPT fails to provide effective feedback on coherence and cohesion for English Language Learner (ELL) students (Yoon, Miszoglad, & Pierce, 2023). The objectives of this study are twofold: (1) to train ChatGPT to enhance its ability to provide feedback on the coherence of students' argumentative essays; and (2) to assess the quality of the feedback provided by this specially trained version of ChatGPT. For this purpose, ChatGPT was specifically tailored and trained to assess coherence, using Toulmin's model (2003) as the framework. The evaluation involved a set of 50 second language (SL) students' argumentative essays. The feedback provided was first analyzed by categorizing each sentence into argumentative components such as claims, data, or warrants based on their function. Subsequently, the logical connections between these components were assessed. The analysis revealed that the feedback was highly specific and offered concrete suggestions for improvement, highlighting the success of the training approach.
Biography
NTN
| Online Profile | NTN |
|---|
