Chatbots in ELT
Methodology
Chatbots are conversational AI systems that simulate human dialogue through natural language input. In ELT, they serve as interactive CALL tools, providing speaking practice, writing feedback, vocabulary drilling, and grammar correction on demand, with the key advantage of being available 24/7 without the social pressure of a human interlocutor.
Two Generations
| Type | Logic | Interaction | ELT Example |
|---|---|---|---|
| Rule-based | Scripted if-then paths | Structured, menu-driven | Early Duolingo chatbots |
| LLM-powered | Neural language models | Open-ended, adaptive | ChatGPT, Andy English Bot |
Rule-based bots offer predictability and controlled input; LLM-powered bots offer natural conversation but may produce inaccurate language or hallucinate information.
Pedagogical Applications
| Skill | Application |
|---|---|
| Speaking | Dialogue practice, role-play, pronunciation drills, simulated real-world scenarios |
| Writing | Iterative feedback on grammar, coherence, and vocabulary; revision support |
| Vocabulary | Contextual drilling, adaptive quizzes, nuance and connotation explanations |
| Grammar | Error correction, metalinguistic explanations, targeted practice |
| Listening | Exposure to synthesised speech; limited by unnatural prosody |
Chatbots work well as pre-task warm-ups (activating background knowledge, reducing anxiety before peer speaking) and post-task extension (writing reflections, extended argument development).
Research Evidence
Meta-analytic reviews confirm medium positive effects on L2 learning:
- Lyu (2025): significant medium effect on L2 performance (g = 0.608)
- Kızıl (2025): systematic review of 33 studies (2020–2024): consistent benefits for L2 acquisition
- ReCALL meta-synthesis (2024): 57 studies on voice-based AI chatbots: improvements in speaking fluency, Willingness to Communicate, and self-perceived communicative competence
- Anxiety: strong, consistent finding: chatbots reduce speaking anxiety by providing judgment-free practice environments
- Writing: improved coherence, grammatical variety, and structure through iterative feedback loops
Popular Tools
| Tool | Strength |
|---|---|
| ChatGPT | Versatile; open-ended conversation, grammar Q&A, writing feedback |
| ELSA Speak | Pronunciation-focused; analyses intonation, rhythm, individual sounds |
| Duolingo roleplay | Structured adaptive dialogue; linguistically engineered |
| Replika | Social companion chatbot; reduces anxiety, promotes conversational fluency |
| Andy English Bot | Grammar and vocabulary through interactive chat |
Limitations
- Pragmatics ceiling: chatbots correct grammar effectively but struggle with discourse coherence, register, implicature, and cultural appropriateness
- Unnatural prosody: synthesised voices flatten elision, assimilation, and natural rhythm, limiting listening input value
- ASR failure: speech recognition breaks down when learners produce errors, exactly when feedback is most needed
- Hallucination: LLM chatbots may generate plausible but inaccurate language explanations
- Equity: requires devices, internet access, and often subscriptions
- Overreliance: learners may depend on automated feedback rather than building autonomous error-detection
Design Principles
- Start with the skill objective: choose the chatbot tool to serve the goal, not the reverse
- Scaffold the interaction: give learners sentence starters, role cards, or prompt frames; fully open tasks often produce minimal language
- Combine feedback types: immediate grammatical correction + delayed pragmatic reflection
- Pair with human follow-up: teachers interpret chatbot data and address discourse/pragmatics chatbots cannot handle
- Build critical AI literacy: train learners to spot inaccurate chatbot output and evaluate feedback quality
Classroom Integration Patterns
- Think-Pair-Chatbot: chatbot replaces the initial pair before whole-class discussion, reduces anxiety and gives learners something to build on
- Role-play rehearsal: practice a scenario with chatbot first, then perform with a peer
- Writing conference simulation: draft → chatbot feedback → revise → teacher conference
Key References
- Lyu, B. (2025). Effectiveness of chatbots in improving language learning: A meta-analysis. International Journal of Applied Linguistics, 35(1).
- Kohnke, L. (2023). A pedagogical chatbot: A supplemental language learning tool. RELC Journal, 54(2), 537–550.
- Kızıl, A. Ş. (2025). A systematic review of recent research on the usefulness of chatbots for language education. Journal of Computer Assisted Learning.
- Fryer, L. K. et al. (2020). Chatbot learning partners: Connecting learning experiences, interest and competence. Computers in Human Behavior, 93, 279–289.
- Godwin-Jones, R. (2022). Partnering with AI: Intelligent writing assistance and instructed language learning. Language Learning & Technology, 26(2), 5–24.