One learner explains a project delay to a partner, hears a request for the supplier's name, then gives the explanation again to somebody else. The second account keeps useful wording from the first and has to work for a listener with different knowledge. Audience rotation gives repeated speaking a communicative reason: the message stays stable enough for refinement, and the listener changes enough to demand a fresh response. This design suits learners who can prepare a short explanation yet lose fluency when a conversation moves beyond the script. It also gives a small class a way to increase speaking turns without producing three unrelated topics.
Reuse with a live purpose
Exact repetition can help a speaker retrieve words and organise clauses faster. It can also become recital. Once the same partner already knows the answer, a second performance may have little information to carry. Tony Lynch and Joan Maclean studied a different arrangement in an English for Specific Purposes course. Fourteen learners prepared academic posters, then answered questions from changing visitors in a poster carousel. The researchers analysed two learners at markedly different proficiency levels in detail. Both reused content across visits and changed vocabulary, grammar or delivery as new questions arrived. One learner's correct subject-verb patterns moved from 3 of 9 relevant cases on the first visit to 8 of 10 on the sixth, though the sequence also contained a dip on visit four.
The audience still has work.
Those figures come from a case analysis, not a controlled trial. They cannot tell us how much improvement audience rotation would cause in another class. They show the process at close range: repeated content met different questions, and the speaker's language changed during the encounters.
Questions become revision material
Emi and Masaki Kobayashi later examined three rounds of poster presentations with different audiences. Their 2018 ethnographic multiple-case study followed Japanese university students through classroom observations, recordings, interviews and documents. Some presenters used audience feedback to refine later performances. Others carried a question from one audience into the next exchange and invited the new listeners to contribute. These variations matter. A request for the survey duration identifies missing content. A listener who mishears fifteen as fifty identifies a pronunciation risk. A blank look after a long noun phrase identifies a packaging problem. The speaker now has a concrete change to make before the next visitor arrives.
Risk can freeze later rounds.
Some students in the Kobayashi study avoided changing later performances because they wanted to reduce risk. Repetition alone did not guarantee experimentation. Teachers need to make one manageable revision visible between rounds.
Design a four-listener circuit
Use one message with a real information gap. A 34-year-old engineer in Hsinchu might practise a 52-second update about a delayed component, the revised delivery date and the decision needed from a manager. Each listener receives one role card: purchasing colleague, client, new team member or manager. The speaker keeps the core update and adapts to four sets of background knowledge without needless delay. Before round one, give 90 seconds to prepare six content words and one useful phrase. Choc Education's article on practising useful phrases before a speaking task offers a compatible routine. Preparation should supply access to language without writing a script. A card containing push the delivery back by two days can support the first attempt. Six complete sentences will invite reading.
Run the circuit this way:
- Speaker A gives the update and answers one genuine follow-up from Listener 1.
- Listener 1 writes one content gap or one phrase that was hard to follow.
- Speaker A chooses one change, gets 23 seconds to mark the prompt card, and moves to Listener 2.
- After four rounds, Speaker A records one final version for a listener who has only the role title.
Other learners rotate at the same time, so nobody waits through four private performances while one speaker gets all the practice. In an online class, breakout-room assignments can change every four minutes, the teacher can visit one room each cycle, and recurring listener questions can be recorded for the final review.
Keep the message partly unstable
Fresh listeners should alter the exchange. If every listener reads the same printed prompt, the task still resembles a drill. Give roles different priorities. The client wants a firm date. The new team member needs the acronym explained. The manager wants two options and a recommendation. Core facts remain fixed: order 417B, a faulty sensor, delivery moved from 6 to 8 November. Stable facts let the speaker retrieve some language again. Different priorities force choices about detail, politeness and sequence. That combination is the reason to rotate listeners. Listener roles should also change what counts as a successful answer, because a client may need certainty about the date, a new colleague may need the process explained, and a manager may accept uncertainty if the speaker names two options and the evidence behind them. One generic fluency score would miss those adaptations. A simple role-specific checklist can capture them without slowing the circuit.
Do not correct five features between rounds. One revision target preserves momentum. If the first listener missed the date, the speaker might stress “the eighth of November” in round two. If the explanation began with four pieces of background, the speaker might lead with the delay. A later round can address verb tense.
One improving account is enough.
I would rather hear one account become clearer across four listeners than collect four first attempts on unrelated prompts.
The repeated message gives both teacher and learner something comparable.
Measure adaptation, not polish
Fluency in the fourth performance may reflect memory for the content. That is part of the design, yet it is poor evidence of broad speaking development. Separate speed from listener adaptation in the record: one column can hold clause-internal pauses, another can hold clarification requests, and a third can note whether the role received its required fact. A faster account that omits the client's date has failed that round, even if its speech rate looks much better on a tidy progress chart today. Score features tied to the changing listener:
- The speaker answers the follow-up without abandoning the message.
- A recurring misunderstanding disappears in a later round.
- The speaker alters detail or wording for the role.
- The final listener recovers the date, cause and requested decision.
Keep the first and fourth recordings. Count clause-internal pauses, successful answers and listener requests for clarification. Avoid turning the last round into a polished score. The research base includes small, qualitative studies, and transfer to a new topic remains uncertain. Next lesson should test that boundary. Give a new problem with the same structure: cause, consequence, date and decision. If the learner can organise it for two fresh listeners, the classroom has evidence beyond a rehearsed story. Use the same checklist so changed content remains the only major difference.
For tomorrow, four role cards and one timer are enough.
Put “Detail needed by this listener” at the bottom of every card. After the first exchange, that note determines the speaker's next 23 seconds.