Download Full PDF

Nanxi Bian

University of Macau, China

 

Qingyu Gao

Shenzhen MSU-BIT University, China

 

Abstract

Traditional “one-size-fits-all” HSK preparation models may constrain proficiency development by insufficiently addressing diverse learner needs. This study investigates the efficacy of a data-driven intervention for HSK Level 4 implemented through an intelligent practice system. Grounded in SLA theories, the system uses big data analytics to analyse learner errors, weaknesses, and behaviours. It then dynamically generates personalised “i+1” exercises for HSK Level 4’s core item types, enabling precise learning interventions. An 8-week quasi-experiment with 49 Russian-speaking students employed a natural usage-based grouping to examine the association between system engagement and learning outcomes. Quantitative analysis revealed a strong positive correlation between system usage intensity and score gains (r = .811, p < .001). This study provides empirical evidence suggesting that big data intervention may help address the limits of homogenised instruction. By enabling realtime tracking, diagnosis, and personalised feedback, it outlines a data-informed pathway that may inform HSK preparation practices and ongoing digital transformation efforts in language education.

 

Keywords

Data-driven language learning, HSK preparation, intelligent tutoring system, second language acquisition, personalised learning

 

数据驱动的语言能力发展路径构建：——基于 HSK4 级 AI 自动出题工具的干预研究

 

边楠茜

澳门大学，中国澳门特别行政区

 

高清宇

深圳北理莫斯科大学，中国

 

摘要

传统的 HSK 备考模式普遍采用同质化的训练方式，难以充分回应学习者个体差异，进而可能制约学习者语言能力的发展。本研究考察一项面向 HSK4 级的数据驱动学习干预的有效性。该干预依托自动出题系统实施，在二语习得理论的指导下，借助大数据分析诊断学习者的错误类型与薄弱环节，并针对 HSK4 级核心题型动态生成个性化“i+1”练习，以实现精准化学习干预。研究以 49 名俄语母语来华留学生为对象，开展了为期 8 周的准实验研究，并依据自然使用强度进行分组，考察系统参与度与学习成效之间的关联。量化结果表明，系统使用强度与成绩提升呈显著正相关（ r = .811， p < .001）。研究结果表明，数据驱动干预可为缓解同质化教学的问题提供智能化解决方案。通过实现对学习过程的实时追踪、诊断与个性化反馈，本文提出了一条数据驱动的语言能力发展路径，可为 HSK 备考实践及语言教育数字化转型提供参考。

 

关键词

数据驱动语言学习， HSK4 级，智能教学系统，二语习得，个性化学习