Teaching evaluation is an indispensable and important part of the whole classroom teaching process. With the development of network information technology, the multimodal theoretical teaching mode has entered the classroom teaching. The traditional single-modal teaching evaluation mode is too boring and low in interactivity, which can no longer meet the requirements of modern teaching mode. In order to solve this problem, this paper introduces the multimodal theory under the artificial intelligence scenario into language teaching, and builds a student-centered multimodal theoretical language teaching evaluation system based on the multimodal discourse analysis theory. Through 18 weeks of unimodal, bimodal and multimodal language teaching in the three classes respectively, the students in the three classes were tested after the teaching. The results show that the students’ attention in class has been improved by multimodal teaching, and the students’ performance has improved by 8.8%. The satisfaction of teachers and students has been improved to a certain extent. Language teaching based on multimodal theory can fully mobilize and give full play to students’ subjective initiative and enthusiasm for learning, improve teaching quality, which makes teaching evaluation more abundant, comprehensive, objective and effective.
