【作 者】：
魏小东， 习鹏飞
【关 键 词 】：
大语言模型； 虚拟数字人； 第二语言教育； 生成式人工智能； 教育应用
【栏 目】：
课程与教学
【中文摘要】：
随着大语言模型技术在多个方面不断取得突破，大语言模型驱动的虚拟数字人辅助第二语言学习能够显著提升第二语言学习者的学习体验和学习效率。文章通过文献梳理总结了虚拟数字人在构建第二语言学习环境、生成第二语言学习资源、模拟第二语言交流对象和提供第二语言学习路径方面的作用，梳理了大语言模型和虚拟数字人存在的生成内容偏差、语用情感缺失、学习反馈不当和应用成本过高的问题，提出了人机协同内容生成构建可信学习资源、角色场景融合实现多元反馈方式、实时学习行为分析支持情感表达和整合通用设备接口促进教育应用普及的虚拟数字人第二语言教育应用策略，并通过案例开发与实施验证了应用策略的有效性，以期为我国第二语言教育在数字时代的可持续发展提供路径选择。
【英文摘要】：
With the continuous breakthrough of large language model technology in many aspects, virtual digital human driven by large language model to assist second language learning can significantly improve the learning experience and learning efficiency of second language learners. Through literature review, this paper summarizes the role of virtual digital human in constructing second language learning environment, generating second language learning resources, simulating second language communication objects and providing second language learning paths. It also sorts out the problems of large language model and virtual digital human, such as poor generated content, lack of pragmatic emotion, inappropriate learning feedback and high application cost. This paper puts forward the application strategy of virtual digital human-assisted second language education, which includes human-computer collaborative content generation to build trusted learning resources, role-scene integration to achieve multiple feedback modes, real-time learning behavior analysis to support emotional expression, and integration of universal device interfaces to promote the popularization of educational applications. The effectiveness of the application strategy has been verified through case development and implementation, with a view to providing choices for the sustainable development of China's second language education in the digital age.