EFL proficiency significantly shapes chatbot interaction: high-proficiency learners over-engage in dialogue while low-proficiency learners over-rely on feedback requests¶

Claim · All claims
Evidence · 1 study (2 entries) · associational r1–r2 · q2
Currency · newest study 2026 · checked against its sources 2026-10-20 (automated, 2 of 2 entries) · not reviewed by a person

Subclaims¶

q2 i? Behavioural state frequencies differed significantly by proficiency (χ2 = 25.4, p < .003), with high-proficiency learners over-represented in user_chat and low-proficiency learners over-represented in check_button. → Steve Woollaston 2026 q2 i? Low-proficiency learners were significantly under-represented in conversational engagement (user_chat residual between −2 and −4). → Steve Woollaston 2026 (2)

Evidence¶

Steve Woollaston 2026¶

Steve Woollaston, Brendan Flanagan, Yuko Toyokawa and Hiroaki Ogata. (2026). Penny: Transition Network Analysis of Learner-Chatbot Interactions in Scaffolded EFL Writing. Preprint, accepted at the 1st Workshop on Transition Network Analysis (TNA) at LAK '26. https://arxiv.org/abs/2607.14575 doi:10.48550/arxiv.2607.14575

q2 · i? · associational · r1

Chi-squared test of independence on behavioural state frequencies by proficiency group (High n = 37,722 events; Low n = 25,606), visualised in a mosaic plot of standardised residuals; the test "yielded a 𝜒2 value of 25.4 (𝑝 < .003)".

"High-proficiency learners displayed a distinct over-representation in conversational states. Specifically, user_chat showed a positive association (residual between 2 and 4), indicating these learners engaged in significantly more dialogue than expected."

Steve Woollaston 2026 (2)¶

Steve Woollaston, Brendan Flanagan, Yuko Toyokawa and Hiroaki Ogata. (2026). Penny: Transition Network Analysis of Learner-Chatbot Interactions in Scaffolded EFL Writing. Preprint, accepted at the 1st Workshop on Transition Network Analysis (TNA) at LAK '26. https://arxiv.org/abs/2607.14575

q2 · i? · associational · r2

Same chi-squared mosaic-plot analysis, low-proficiency group pattern: significant over-representation of check_button and under-representation of user_chat (residual between −2 and −4), described as "a higher dependency on general feedback".

"This group showed a statistically significant over-representation for check_button (residual between 2 and 4), indicating a higher dependency on general feedback."

Discussion¶

Related Claims¶

• Revision after chatbot feedback predominantly results in successful uptake, indicating effective error correction — related
• Learner-chatbot EFL writing interaction follows two dominant behavioural loops: a Revision Loop and a Chat Loop — related
• Strategy-specific effects: time affects self-repetitions and marginally clarification requests; proficiency pairing affects confirmation checks and negative feedback; language affects other-repetitions and negative feedback — related
• In CPS dialogues, low-achievers requested more support while high-achievers provided more support and reminders, though group differences were not statistically significant — related
• Language proficiency, beliefs, feedback literacy and learning objectives moderate students' engagement with and uptake of WCF — related
• Low-proficiency EFL undergraduates report feeling more competent in English tasks through ChatGPT-assisted writing feedback, conversational practice, and translation support — related