Large language models empower multimodal interaction: a study on enhancing oral fluency in Chinese as a second language learners in digital classrooms
大语言模型赋能多模态交互：数字课堂中汉语二语学习者口语流利度提升路径研究

摘要

数字化课堂环境下，汉语二语口语流利度提升面临着教学互动受限和技术应用不足两个核心问题。在教学层次上，互动方式单一，缺少真实情境，缺乏反馈；在技术层面上，语音识别的准确性、多模态的融合程度不高，以及大尺度的语言模型应用不完全。针对以上问题，本文提出以大语种模型为核心驱动力，从多模态沉浸训练场景构建、个性化实时反馈系统优化、分层高级训练框架构建、学习数据监控与优化机制四个方面优化路径，旨在系统提升汉语二语教学水平，为学习者口语流利程度的提升提供一体化解决方案。

Abstract

In the digital classroom environment, the improvement of fluency in spoken Chinese as a second language faces two core issues: limited teaching interaction and insufficient technological application. At the teaching level, the interactive mode is single, lacking real situations and feedback; On a technical level, the accuracy of speech recognition, the degree of multimodal fusion, and the incomplete application of large-scale language models are not high. In response to the above issues, this article proposes to use a large language model as the core driving force to optimize the path from four aspects: construction of multimodal immersive training scenarios, optimization of personalized real-time feedback systems, construction of hierarchical advanced training frameworks, and monitoring and optimization mechanisms for learning data. The aim is to systematically improve the level of Chinese second language teaching and provide an integrated solution for enhancing learners' oral fluency.