【作者简介】毕向阳，中央民族大学民族学与社会学学院

【文章来源】《社会学评论》2026年第2期

【内容提要】随着计算社会科学的发展，融合自然语言处理的计算扎根理论为质性研究开辟了新路径，但其应用规范与边界仍存争议。本文基于经典扎根理论，整合词向量字典扩展技术，构建了敏捷化的计算扎根理论操作框架，并以学生评教文本中教师魅力的性别刻板印象分析为案例，探索并优化该框架的应用。实证表明，此路径通过透明、可迭代的计算流程，在显著提升分析效率的同时，较好保持了质性研究的解释深度，实现了数据驱动与理论驱动的融合；又能在扎根文本的基础上充分发挥研究者的理论敏锐性，提供兼具敏捷性、透明度并能实证检验理论假设的研究方式。最后，本文从方法论层面反思了关键问题，旨在为大数据时代的社科文本分析提供一种规范、可参照的操作范式。

【关键词】学生评教 / 计算扎根理论 / 刻板印象 / 自然语言处理

【全文链接】http://src.ruc.edu.cn/CN/Y2026/V14/I2/32

Towards Agile Computational Grounded Theory: A Methodological Review, Reflection, and Research Practice

Abstract: With the development of computational social science, computational grounded theory integrated with natural language processing has opened a new path for qualitative research; however, its application norms and boundaries remain contentious. Building on classical grounded theory, this paper develops an agile operational framework for computational grounded theory by integrating word vector dictionary expansion techniques. Taking the analysis of gender stereotypes in teacher charisma within student evaluation texts as a case, it explores the application optimization of this framework. Empirical results show that through a transparent and iterative computational process, this approach significantly improves analysis efficiency while preserving the interpretive depth of qualitative research, thereby achieving the integration of data-driven and theory-driven methodologies. This paradigm is text-centered, leverages researchers’theoretical sensitivity, and offers a research method that is agile, transparent, and capable of empirically testing theoretical hypotheses. Finally, this paper reflects on key methodological issues, aiming to provide a standardized and referable operational paradigm for social science text analysis in the era of big data.