Effects of Pedagogical Agents’ Language Styles and Their AI-Generated Approach on the Video Learning for Pupils with Different Cognitive Styles

doi:10.12139/j.1672-0628.2026.03.013

Abstract

Abstract:

Technological advances have improved the language capabilities of pedagogical agents, enabling the creation of AI-generated pedagogical agents. However, it is unclear which language style of pedagogical agent is appropriate for students with different cognitive styles, and further exploration is needed into the effectiveness of their AI-generated pedagogical agents. Therefore, Experiment 1 examined the effects of conversational and formal style pedagogical agents on video learning among pupils with different cognitive styles. The results indicated no significant differences in social presence, cognitive load, or test scores between the two conditions for field-independent students. However, field-dependent students exhibited higher social presence and test scores in the conversational style condition than in the formal style condition; the social presence in field-dependent students fully mediated the relationship between language style and test scores. Experiment 2 further examined whether the conversational style pedagogical agents created using AI-generated voice, AI-generated voice and human-like appearance, or AI-generated voice and cartoon appearance were equally effective for field-dependent students, compared to real-human conversational style pedagogical agents. The results indicated that only the conversational style pedagogical agents created using AI-generated voice and cartoon appearance further improved the test scores of field-dependent students. These findings suggest that conversational style pedagogical agents are more effective for field-dependent students, and their effectiveness can be further enhanced when they are created using AI-generated voice and cartoon appearance.

Key words: pedagogical agent, language style, cognitive styles, AI-generated pedagogical agents, pupils

摘要：

技术的进步提高了教学代理的语言能力，也使创建AI生成教学代理成为可能。但不同认知风格学生适合何种语言风格教学代理尚不清晰，其AI生成教学代理是否有效也有待探究。为此，实验1考察了会话和正式风格教学代理对不同认知风格小学生视频学习的影响。结果发现，场独立学生在两种条件下的社会临场感、认知负荷和测验成绩均无显著差异；场依存学生在会话风格条件下的社会临场感和测验成绩更高，且场依存学生的社会临场感在教学代理语言风格和测验成绩间起完全中介作用。实验2进一步探究了相较于真人会话风格教学代理，采用AI生成语音、AI生成语音及仿真形象和AI生成语音及卡通形象方式创建的会话风格教学代理对场依存学生是否同样有效。结果发现，AI生成语音及卡通形象的会话风格教学代理进一步提高了场依存学生的测验成绩。上述结果表明，会话风格教学代理对场依存学生更有效，AI生成语音及卡通形象的会话风格教学代理能进一步提高其学习效果。

关键词: 教学代理, 语言风格, 认知风格, AI生成教学代理, 小学生

CLC Number:

B849

Xuejiao DU, Zifan MENG, Yao LI, Chuwen WANG, Yinghong DONG. Effects of Pedagogical Agents’ Language Styles and Their AI-Generated Approach on the Video Learning for Pupils with Different Cognitive Styles[J]. Studies of Psychology and Behavior, 2026, 24(3): 384-392.

杜雪娇, 孟子凡, 李瑶, 王楚雯, 董颖红. 教学代理语言风格及AI生成方式对不同认知风格小学生视频学习的影响[J]. 心理与行为研究, 2026, 24(3): 384-392.

Figures/Tables 8

References 0

	李艳芬, 赵宁宁, 周铁民. 不同内隐学习任务在不同认知风格个体上的成绩差异. 心理与行为研究, 2017, 15 (5): 606- 612. DOI
	吕馨, 刘景瑶, 魏柳青, 张学民. 目标数量与运动框架旋转角度对不同场认知风格个体多目标追踪表现的影响. 心理学报, 2019, 51 (1): 24- 35.
	宋合义. (1988). 认知方式图形测验的编制与修订说明. 见谢斯骏, 张厚粲 (编), 认知方式一个人格维度的实验研究 (pp. 261–276). 北京: 北京师范大学出版社.
	宋淑娟, 刘华山. 数学-性别刻板印象对女生的威胁效应: 场认知风格的调节作用. 心理与行为研究, 2015, 13 (3): 302- 305, 319.
	王福兴, 李文静, 谢和平, 刘华山. 多媒体学习中教学代理有利于学习吗?——一项元分析研究. 心理科学进展, 2017, 25 (1): 12- 28. DOI
	Almeida, L. M. C. G., Münzer, S., & Kühl, T. More personal, but not better: The personalization effect in learning neutral and aversive health information. Journal of Computer Assisted Learning, 2024, 40 (5): 2248- 2260. DOI
	Bai, J., Cheng, X. L., Zhang, H., Qin, Y. H., Xu, T., & Zhou, Y. Can AI-generated pedagogical agents (AIPA) replace human teacher in picture book videos? The effects of appearance and voice of AIPA on children’s learning. Education and Information Technologies, 2025, 30 (9): 12267- 12287. DOI
	Castro-Alonso, J. C., Wong, R. M., Adesope, O. O., & Paas, F. Effectiveness of multimedia pedagogical agents predicted by diverse theories: A meta-analysis. Educational Psychology Review, 2021, 33 (3): 989- 1015. DOI
	Chen, C. M., Li, M. C., & Chen, Y. T. The effects of web-based inquiry learning mode with the support of collaborative digital reading annotation system on information literacy instruction. Computers & Education, 2022, 179, 104428. DOI
	Du, X. J., Chen, C., & Lin, H. X. The impact of working memory capacity on collaborative learning in elementary school students. Frontiers in Psychology, 2022, 13, 1027523. DOI
	Evans, C., Richardson, J. T. E., & Waring, M. Field independence: Reviewing the evidence. British Journal of Educational Psychology, 2013, 83 (2): 210- 224. DOI
	Fynes-Clinton, S., Marstaller, L., & Burianová, H. Differentiation of functional networks during long-term memory retrieval in children and adolescents. NeuroImage, 2019, 191, 93- 103. DOI
	Ginns, P., Martin, A. J., & Marsh, H. W. Designing instructional text in a conversational style: A meta-analysis. Educational Psychology Review, 2013, 25 (4): 445- 472. DOI
	Kühl, T., & Münzer, S. Learning about a serious disease: When a personalized message is harmful unless you are happy. Journal of Computer Assisted Learning, 2021, 37 (5): 1312- 1323. DOI
	Li, H. Y., & Graesser, A. C. The impact of conversational agents’ language on summary writing. Journal of Research on Technology in Education, 2021, 53 (1): 44- 66. DOI
	Li, J., Kizilcec, R., Bailenson, J., & Ju, W. Social robots and virtual agents as lecturers for video instruction. Computers in Human Behavior, 2016, 55, 1222- 1230. DOI
	Li, W. J., Wang, F. X., & Mayer, R. E. Increasing the realism of on-screen embodied instructors creates more looking but less learning. British Journal of Educational Psychology, 2024, 94 (3): 759- 776. DOI
	Lin, L. J., Ginns, P., Wang, T. H., & Zhang, P. L. Using a pedagogical agent to deliver conversational style instruction: What benefits can you obtain. Computers & Education, 2020, 143, 103658. DOI
	Lin, L. J., Lin, X., Zhang, X. F., & Ginns, P. The personalized learning by interest effect on interest, cognitive load, retention, and transfer: A meta-analysis. Educational Psychology Review, 2024, 36 (3): 88. DOI
	Moreno, R., & Mayer, R. E. Engaging students in active learning: The case for personalized multimedia messages. Journal of Educational Psychology, 2000, 92 (4): 724- 733. DOI
	Netland, T., von Dzengelevski, O., Tesch, K., & Kwasnitschka, D. Comparing human-made and AI-generated teaching videos: An experimental study on learning effects. Computers & Education, 2025, 224, 105164. DOI
	Nightingale, S. J., & Farid, H. AI-synthesized faces are indistinguishable from real faces and more trustworthy. Proceedings of the National Academy of Sciences of the United States of America, 2022, 119 (8): e2120481119. DOI
	Paas, F. G. Training strategies for attaining transfer of problem-solving skill in statistics: A cognitive-load approach. Journal of Educational Psychology, 1992, 84 (4): 429- 434. DOI
	Pataranutaporn, P., Danry, V., Leong, J., Punpongsanon, P., Novy, D., Maes, P. & Sra, M. AI-generated characters for supporting personalized learning and well-being. Nature Machine Intelligence, 2021, 3 (12): 1013- 1022. DOI
	Pi, Z. L., Deng, L. X., Wang, X., Guo, P. R., Xu, T., & Zhou, Y. The influences of a virtual instructor’s voice and appearance on learning from video lectures. Journal of Computer Assisted Learning, 2022, 38 (6): 1703- 1713. DOI
	Piaget, J. (1952). The origins of intelligence in children (M. Cook, Trans.). New York: W. W. Norton & Company.
	Polat, H., Taş, N., Kaban, A., Kayaduman, H., & Battal, A. Human or humanoid animated pedagogical avatars in video lectures: The impact of the knowledge type on learning outcomes. International Journal of Human-Computer Interaction, 2025, 41 (14): 8912- 8927. DOI
	Reichelt, M., Kämmerer, F., Niegemann, H. M., & Zander, S. Talk to me personally: Personalization of language style in computer-based learning. Computers in Human Behavior, 2014, 35, 199- 210. DOI
	Sikström, P., Valentini, C., Sivunen, A., & Kärkkäinen, T. How pedagogical agents communicate with students: A two-phase systematic review. Computers & Education, 2022, 188, 104564. DOI
	Stiller, K. D., & Jedlicka, R. A kind of expertise reversal effect: Personalisation effect can depend on domain-specific prior knowledge. Australasian Journal of Educational Technology, 2010, 26 (1): 133- 149.
	Tinajero, C., Castelo, A. M., Guisande, M. A., & Páramo, M. F. Self-regulated learning in female students with different cognitive styles: An exploratory study. Perceptual and Motor Skills, 2010, 111 (1): 31- 44. DOI
	Xu, T., Liu, Y., Jin, Y. R., Qu, Y. Y., Bai, J., Zhang, W. L., & Zhou, Y. From recorded to AI-generated instructional videos: A comparison of learning performance and experience. British Journal of Educational Technology, 2025, 56 (4): 1463- 1487. DOI
	Zhao, F. Z. & Mayer, R. E. Improving multimedia learning with emotional design: Depicting key elements with positive features. Journal of Computer Assisted Learning, 2025, 41 (3): e70028. DOI

教学代理语言风格	学生认知风格	测验成绩		社会临场感		心理努力		任务难度
教学代理语言风格	学生认知风格	M	SD	M	SD	M	SD	M	SD
正式风格	场独立	1.06	0.91	16.69	2.51	4.09	2.26	4.53	1.80
会话风格	场独立	0.88	0.71	17.66	3.68	4.31	2.09	4.78	2.03
正式风格	场依存	0.34	0.60	15.25	3.85	4.41	2.00	4.47	2.00
会话风格	场依存	0.91	0.86	19.31	4.24	5.06	1.78	5.28	1.51

教学代理语言风格	学生认知风格	测验成绩		社会临场感		心理努力		任务难度
教学代理语言风格	学生认知风格	M	SD	M	SD	M	SD	M	SD
正式风格	场独立	1.06	0.91	16.69	2.51	4.09	2.26	4.53	1.80
会话风格	场独立	0.88	0.71	17.66	3.68	4.31	2.09	4.78	2.03
正式风格	场依存	0.34	0.60	15.25	3.85	4.41	2.00	4.47	2.00
会话风格	场依存	0.91	0.86	19.31	4.24	5.06	1.78	5.28	1.51

结果变量	预测变量	β	SE	t	p	R	R²	F	p
测验成绩	教学代理语言风格	0.71	0.24	3.04	0.004	0.36	0.13	9.25	0.004
社会临场感	教学代理语言风格	0.90	0.22	4.01	<0.001	0.45	0.21	16.10	<0.001
测验成绩	教学代理语言风格	0.09	0.20	0.45	0.650	0.72	0.51	32.06	<0.001
	社会临场感	0.69	0.10	6.92	<0.001	0.72	0.51	32.06	<0.001

结果变量	预测变量	β	SE	t	p	R	R²	F	p
测验成绩	教学代理语言风格	0.71	0.24	3.04	0.004	0.36	0.13	9.25	0.004
社会临场感	教学代理语言风格	0.90	0.22	4.01	<0.001	0.45	0.21	16.10	<0.001
测验成绩	教学代理语言风格	0.09	0.20	0.45	0.650	0.72	0.51	32.06	<0.001
	社会临场感	0.69	0.10	6.92	<0.001	0.72	0.51	32.06	<0.001

	效应值	Boot标准误	Boot CI下限	Boot CI上限
直接效应	0.09	0.18	−0.27	0.44
中介效应	0.62	0.19	0.29	1.03
总效应	0.71	0.23	0.27	1.18