Automated Feedback to Peer Tutors Increases Their Talk Moves and Learner Reasoning but Not the Balance of Talk

Dorottya Demszky; Yann Hicke; Joy Yun; Mariah Olson; Max Karpawich; Anees Iqbal; Matt Wu; Jimmy Huang; Akshay Ravikumar; Drew Bent

AI summary · arXiv: 2611.07072v1 · cs.CL

Automated Feedback to Peer Tutors Increases Their Talk Moves and Learner Reasoning but Not the Balance of Talk

Dorottya Demszky, Yann Hicke, Joy Yun, Mariah Olson, Max Karpawich, Anees Iqbal, Matt Wu, Jimmy Huang,

and 2 more authors

Akshay Ravikumar, Drew Bent

Original paper Submitted to arXiv on 04 Nov. 2026 License: CC BY 4.0 Also in: fr

How to read this summary? Illustrated guide

The essentials

Automated feedback increased tutors’ elicitation and uptake of learner ideas and learner reasoning, but did not reduce tutors’ share of talk or increase learner-facing effects.

Based on the abstract only: check the details in the original paper.

Question

The study asks whether automated discourse feedback improves participation and reasoning in small-group volunteer tutoring, and whether giving learners participation feedback strengthens those effects. It also examines why feedback may or may not change the balance of talk.

Method

The authors conducted a randomized controlled trial across two four-week SAT mathematics bootcamps on Schoolhouse.world, involving 1,030 volunteer tutors, 7,443 learners, and 6,508 recorded sessions. Tutors were assigned to no feedback, feedback on their own discourse, or tutor feedback combined with learner participation feedback under one of two motivational framings; interviews with 17 tutors and 13 learners supplemented the trial.

Findings

Result Tutor-facing automated feedback increased the elicitation of learner ideas, uptake of learner contributions, and learner reasoning by 8–15%.
See the passage in the paper
Tutors were randomly assigned to receive no feedback, automated feedback on their own discourse, or that feedback together with participation feedback delivered to their learners under one of two motivational framings. Feedback to tutors increased eliciting of learner ideas, uptake of learner contributions, and learner reasoning by 8-15%, and the pattern replicated across both bootcamps.
Result Tutor feedback did not change the proportion of session talk attributable to tutors, which was high in the control group.
See the passage in the paper
However, tutor talk time, which averaged 91.5% in the control group, did not change in any condition, and learner-facing feedback did not further change discourse.
Result Adding learner-facing participation feedback did not further change discourse in the sessions.
See the passage in the paper
However, tutor talk time, which averaged 91.5% in the control group, did not change in any condition, and learner-facing feedback did not further change discourse.
Authors’ interpretation Tutor interviews suggest tutors used the recommended talk moves but encountered limited learner responses.
See the passage in the paper
Interviews with 17 tutors indicate that they adopted the moves the feedback recommended but were limited by learners who did not respond.
Authors’ interpretation Learners’ preferences and the group-level nature of the feedback may help explain its limited effect.
See the passage in the paper
Interviews with 13 learners suggest why the learner-facing feedback had little effect: it reported a group-level rather than a learner-level statistic, some learners preferred to listen, and learner-to-learner interaction was rare in these sessions.

Limitations

The study is limited to volunteer tutoring sessions in two four-week SAT mathematics bootcamps, so its findings do not establish effects in other instructional settings.
The authors report that learner participation feedback used a group-level statistic, which may have limited its relevance to individual learners.
See the passage in the paper
it reported a group-level rather than a learner-level statistic

Why it matters

The authors argue that automated feedback can encourage tutors to invite learner thinking and improve reasoning, but changing how much learners talk likely requires interventions beyond feedback. Their findings distinguish improving the quality of tutor moves from shifting the overall balance of talk.

Written by AI from the paper. Passages are checked word for word against the text we analysed; the rest may contain errors.

AI-generated keywords: automated feedback volunteer tutoring classroom discourse learner participation SAT mathematics

Read further

Comprehensive Summary
Layman's Summary
Blog article
Original abstract

The paper examines whether automated feedback that has improved classroom discourse in other settings can also change interaction in small-group volunteer tutoring—and whether giving learners participation feedback makes that change more effective. The study focuses on SAT mathematics bootcamps on Schoolhouse.world, where volunteer tutors lead online sessions. It distinguishes between encouraging tutors to invite and build on learners’ thinking and changing the overall balance of who speaks. The researchers report a randomized controlled trial spanning two four-week bootcamps. The study involved 1,030 volunteer tutors, 7,443 learners, and 6,508 recorded sessions. Tutors were assigned to one of three broad conditions: no feedback, automated feedback about their own discourse, or tutor feedback combined with participation feedback for learners. The learner-facing feedback used one of two motivational framings. The abstract does not specify the feedback measures, statistical analyses, or the particular framings. Tutor-facing automated feedback increased several discourse behaviors: tutors more often elicited learners’ ideas and built on their contributions, while learners showed more reasoning. These improvements were reported as increases of 8–15%, and the pattern was replicated across both bootcamps. The findings suggest that feedback can influence tutors’ questions and responses and may encourage more learner reasoning. However, these changes did not shift the overall distribution of talk. Tutors in the control group spoke for an average of 91.5% of session time, and tutor talk time did not change in any condition. Adding participation feedback for learners did not produce further discourse changes compared with tutor feedback alone. Thus, the intervention appears to have affected particular instructional moves without increasing learners’ overall share of the conversation. Interviews offer possible explanations for this gap. Interviews with 17 tutors suggest that some adopted the recommended moves but encountered learners who did not respond. Interviews with 13 learners indicate that group-level, rather than individual, participation statistics may have made the learner-facing feedback less relevant. Some learners also preferred listening, and learner-to-learner interaction was uncommon. These accounts help explain why prompting tutors to invite learners’ thinking could increase reasoning without substantially increasing how much learners spoke, and why learner-facing feedback did not add a measurable effect. The authors conclude that automated feedback can help tutors invite and build on learner thinking in volunteer tutoring, but that feedback alone may not be enough to redistribute talk. Increasing learner talk may require additional changes, such as addressing learners’ willingness or opportunities to respond and supporting interaction among learners. The available source consists only of the paper’s title and abstract; the full text was not available. As a result, the supplied material does not establish details needed to assess the study fully, including how outcomes were measured, the size and uncertainty of individual effects, assignment and analysis procedures, or the scope of the interview evidence. The results come from two SAT mathematics bootcamps on one platform, so the abstract alone cannot show how well they generalize to other subjects, tutoring settings, or learners.

Can Automated Feedback Change Who Speaks in Volunteer Tutoring?

In a small-group tutoring session, asking learners to explain their ideas is not quite the same as giving them more time to speak. A tutor might ask a thoughtful question and build on a learner’s response, yet still do most of the talking. A study titled “Automated Feedback to Peer Tutors Increases Their Talk Moves and Learner Reasoning but Not the Balance of Talk” examines this distinction in online SAT mathematics bootcamps.

The researchers studied volunteer-led sessions on Schoolhouse.world. They asked whether automated feedback could help tutors use discussion moves that draw out learners’ thinking—and whether feedback given to learners about participation could make those changes more effective. The results point to a meaningful but limited effect: tutors changed some of their instructional behavior, and learners showed more reasoning, but learners did not gain a larger overall share of the conversation.

Inside the Study

The randomized controlled trial spanned two four-week bootcamps. In total, it involved 1,030 volunteer tutors, 7,443 learners, and 6,508 recorded sessions. Tutors were assigned to one of three broad conditions:

No feedback.
Automated feedback about the tutor’s own discourse.
Tutor feedback combined with participation feedback for learners.

The learner-facing feedback used one of two motivational framings. The available abstract does not describe the specific feedback measures or explain those framings in detail, so it is not possible to tell exactly what participants saw.

What Changed—and What Did Not

Tutors used more discussion moves

Compared with the control condition, tutors who received automated feedback more often elicited learners’ ideas and built on their contributions. Learners also showed more reasoning. The reported increases in these behaviors were 8–15%, and the pattern appeared in both bootcamps.

That distinction matters. The findings suggest that feedback can influence the way tutors ask questions and respond to learners, rather than simply reminding them to cover material. Encouraging tutors to invite ideas and follow up on them may help make reasoning more visible in a session.

The overall balance of talk stayed the same

Despite those changes, learners did not take up more of the session’s total speaking time. In the control group, tutors spoke for an average of 91.5% of session time. Tutor talk time did not change in any condition.

Adding participation feedback for learners also did not produce further discourse changes compared with tutor feedback alone. In other words, the intervention appears to have improved particular instructional moves without redistributing the conversation overall. Asking for learner thinking and giving learners a greater share of the floor are related goals, but they are not interchangeable.

Why Might More Invitations Not Mean More Learner Talk?

Interviews offer possible explanations. Seventeen tutors described cases in which they adopted the recommended moves but learners did not respond. A tutor can invite a contribution, but the invitation alone cannot ensure that a learner feels ready or willing to speak.

Interviews with 13 learners suggest that group-level participation statistics may have felt less relevant than feedback addressed to an individual. Some learners also preferred listening. Meanwhile, learner-to-learner interaction was uncommon. Together, these accounts help explain how tutor prompts might encourage more reasoning while leaving the overall distribution of talk largely untouched.

These interview findings are useful clues, not a complete explanation. They suggest that improving discussion may require attention to the conditions around a prompt: whether learners have a clear opportunity to respond, whether they want to speak, and whether the session supports them in engaging with one another as well as with the tutor.

What the Findings Suggest

The study’s central takeaway is that automated feedback can help volunteer tutors invite and build on learner thinking. But feedback by itself may not be enough to give learners more speaking time. If the goal is a more balanced conversation, future approaches may need to pair tutor guidance with ways to support learner willingness and opportunity to participate, as well as interaction among learners.

The evidence also has important limits. The available source is the paper’s title and abstract, not its full text. It does not establish how outcomes were measured, the size and uncertainty of individual effects, the assignment and analysis procedures, or the full scope of the interview evidence. The findings come from two SAT mathematics bootcamps on one platform, so the abstract alone cannot show whether the results generalize to other subjects, tutoring settings, or learner groups.

Even with those caveats, the study highlights a practical distinction for anyone designing tutoring support: better questions and responses may strengthen learner reasoning, but changing who speaks requires more than prompting tutors to ask. It may also mean making participation feel relevant, possible, and worthwhile for learners.

Created on 09 Nov. 2026

Was this synthesis useful?

Score: 0

Ask questions about this paper to our AI assistant

How to chat with this paper? Illustrated guide

Compare with the PDF

You can also chat with multiple papers at once here.

AI assistant instructions?

Tools for this paper

Explain it differently 1 free rewrite left today

Save to a collection Group papers in collections to compare them, chat with them and export them. Continue with Google Create a free account Log in

Export BibTeX · RIS (Zotero, Mendeley, EndNote) · Obsidian (Markdown) · Notion (Markdown)

Advanced tools Premium

For researchers who follow a field closely: citation alerts, journal-club slides, version comparison and Zotero export. Discover Premium

Zotero

Citation alerts Citation alerts guide

Who works on it? The people and institutions who publish on its topic

Weekly letter Presented in "Artificial intelligence this week" of 9 November 2026. Artificial intelligence this week: When better scores miss the point

Journal club Premium+ Slides to present it Propose for the journal club

Versions What changed between versions

Talk about it in public Premium+ Press kit: release, questions, hype check Press kit guide

Disclaimer: The AI-based summarization tool and virtual assistant provided on this website may not always provide accurate and complete summaries or responses. We encourage you to carefully review and evaluate the generated content to ensure its quality and relevance to your needs.

Automated Feedback to Peer Tutors Increases Their Talk Moves and Learner Reasoning but Not the Balance of Talk

The essentials

Question

Method

Findings

Limitations

Why it matters

Read further

Definitions

Can Automated Feedback Change Who Speaks in Volunteer Tutoring?

Inside the Study

What Changed—and What Did Not

Tutors used more discussion moves

The overall balance of talk stayed the same

Why Might More Invitations Not Mean More Learner Talk?

What the Findings Suggest

Ask questions about this paper to our AI assistant

Tools for this paper

Advanced tools Premium

Papers around this one

Closest papers

Prior work it builds on

Go further

Cite this paper (BibTeX)