【作 者】：
赵 姝， 康春晨， 叶思婷， 胡万山， 王娟利
【关 键 词 】：
写作助手智能体； 写作焦虑； 写作行为； 情感分析； 基础教育
【栏 目】：
课程与教学
【中文摘要】：
为了探究写作焦虑水平对小学生使用写作助手智能体过程中的行为、认知与情感表现的影响，研究设计了一款以启发式提问为核心功能的写作助手智能体，采用认知网络分析与情感分析方法，对不同写作焦虑水平学生在三次写作活动中的成绩变化、行为模式、认知特征和情感体验进行了系统考察。结果显示：低写作焦虑组学生写作成绩显著优于高写作焦虑组，低写作焦虑组成绩稳定上升，高写作焦虑组成绩不稳定；行为上，高写作焦虑组表现为“提问—推理—要求智能体推理和总结”的行为模式，低写作焦虑组表现为“探索—加工—自主总结大纲”的行为模式；认知上，低写作焦虑组保持了稳定的高阶认知水平，高写作焦虑组认知水平有所提升，但依赖于智能体引导；情感上，两组均表现出积极的情感体验。研究提出，在设计写作助手智能体时，应采用启发式提问，提供差异化的认知支架和情感支架。研究为智能体设计提供了实证依据，为分析不同心理特征学生的行为及认知表现提供了分析思路。
【英文摘要】：
To explore the impact of writing anxiety levels on elementary school students' behavioral, cognitive and emotion performances in the process of using writing assistant agents, the study designed a writing assistant agent with heuristic questioning as its core function. It adopted cognitive network analysis and emotional analysis to systematically examine performance changes, behavioral patterns, cognitive characteristics, and emotional experience of students with different levels of writing anxiety during three writing activities. The results indicate that students in the low writing anxiety group have significantly better writing performance than those in the high anxiety group, with the former's performance steadily improving and the latter's being unstable. In terms of behavior, students in the high writing anxiety group exhibit a pattern characterized by "questioning-reasoning-requesting the agent for reasoning and summarization", while students in the low writing anxiety group follow a trajectory of "exploration-processing-independent outline construction". Cognitively, students in the low writing anxiety group maintain a stable higher-order cognition, while students in the high writing anxiety group show some improvement in cognitive levels, but rely on the guidance of the agent. Emotionally, both groups express generally positive experiences. The findings suggest that the design of writing assistant agents should incorporate heuristic questioning and offer differentiated emotional and cognitive scaffolds. This study provides empirical evidence for agent design and offers analytical ideas for analyzing students' behavioral and cognitive performance with different psychological traits.