AI-Driven Scaffolding and Affective Support in ESL Argumentative Writing: A Multimodal Analytics Approach

Abstract

The paper suggests an ESL system of writing an argumentative essay, combining AI-based scaffolding and emotional support. This mechanism incorporates multimodal pretense extraction and gives a blend of text, speech, and motion expression of faces to estimate the emotional feelings and writing patterns of learners. The Transformer architecture with the emotion recognition model is used to complete dynamic intervention framework. It was experimentally determined that such a system also managed the quality of writing with emotion management and reduced cognitive load. Moreover, statistical modeling was also used to find a correlation between multimodal characteristics and writing performance. The findings indicate that this system has a positive influence on the performance of ESL students in terms of writing and, at the same time, offers adaptive and personalized feedback to the learners.