EUVIMEDEUVIMED entdecken
Systematische ÜbersichtOpen Access · CC BY-NC-ND 4.0

Rooted in and beyond interaction: A systematic review of interactive affordances of chatbots for language learning amidst the rise of large language models

Yunfei Du, Barry Lee Reynolds

Acta Psychologica · 2025-09 · 259 · 105307

DOI: 10.1016/j.actpsy.2025.105307 ↗

Kurz gefasst

Systematische Übersicht in der Zeitschrift Acta Psychologica, 2025-09. Verfasst von Yunfei Du und 1 weiteren Autor:innen. Beteiligte Einrichtung: School of Languages and Communication Studies, Chongqing University of Technology, Chongqing, China; Faculty of Education, University of Macau, Taipa…. Hauptthemen laut MEDLINE: Generative Artificial Intelligence, Language, Large Language Models, Learning. Die Arbeit wurde bisher 17-mal zitiert (Crossref) und nennt 99 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: Chongqing Municipal Education Commission.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

As chatbots evolve from scripted responders to reasoning companions, their role in language education is rapidly shifting. Yet, while their presence is expanding, a deeper understanding of how they interact-not just respond-with learners remains elusive. This systematic review repositions interaction as the core of chatbot-assisted language learning, drawing on 100 peer-reviewed studies published between 2009 and 2024. Through a three-phase grounded coding approach to extract and synthesize these literature, we identify four distinct interaction styles, such as interlocutor, narrator, entertainer, and facilitator, each shaping engagement across pedagogical, affective, and cognitive domains. We find that large language models (LLMs) have broadened the educational affordances of chatbots, particularly in writing, speaking, and emotional support. However, their potential to foster higher-order thinking remains largely untapped. Limitations such as hallucination, bias, and ethical risks are seldom addressed in practice, underscoring the need for more critical integration. We argue that effective use depends not only on technological sophistication, but on prompt design, emotional connection, and teachers as ethical gatekeepers.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Generative Artificial IntelligenceLanguageLarge Language ModelsLearning

Stichwörter der Autor:innen

ChatGPTChatbotInteractionLanguage learningLarge language models (LLMs)

Autor:innen und Institutionen

  • School of Languages and Communication Studies, Chongqing University of Technology, Chongqing, China; Faculty of Education, University of Macau, Taipa, Macau SAR, China; Centre for Data Science, University of Macau, Taip…
  • Faculty of Education, University of Macau, Taipa, Macau SAR, China; Centre for Data Science, University of Macau, Taipa, Macau SAR, China. Electronic address: barryreynolds@um.edu.mo.

Förderung

  • Chongqing Municipal Education Commission

Zitieren

APA

Du, Y., & Reynolds, B. L. (2025). Rooted in and beyond interaction: A systematic review of interactive affordances of chatbots for language learning amidst the rise of large language models. Acta Psychologica, 259, 105307. https://doi.org/10.1016/j.actpsy.2025.105307

Vancouver

Du Y, Reynolds BL. Rooted in and beyond interaction: A systematic review of interactive affordances of chatbots for language learning amidst the rise of large language models. Acta Psychologica. 2025;259:105307. doi:10.1016/j.actpsy.2025.105307

BibTeX
@article{Du2025,
  author = {Du, Yunfei and Reynolds, Barry Lee},
  title = {Rooted in and beyond interaction: A systematic review of interactive affordances of chatbots for language learning amidst the rise of large language models},
  journal = {Acta Psychologica},
  year = {2025},
  volume = {259},
  pages = {105307},
  doi = {10.1016/j.actpsy.2025.105307}
}
Kostenlos starten →