A learner gives a two-minute answer about a difficult day at work. The teacher then replaces it with a fresh topic, hoping that variety will keep the lesson lively. That choice throws away useful work. For an intermediate learner who can explain an idea slowly yet loses control under time pressure, repeating the same speaking task should come before changing the topic. The second attempt gives the learner a rare chance to improve language that has already been tested in real time.
This is deliberate repetition.
What changes on attempt two
Speaking draws on several processes at once. A learner has to decide what to say, retrieve words, arrange grammar, shape sounds, watch the listener, and repair trouble. A familiar task removes part of the first burden because the message has already been planned once. Attention can move towards a phrase that arrived too late, an article that disappeared, or a pause that split one idea in half.
Nel de Jong and Charles Perfetti tested this logic with 24 learners in English as a second language classes. Across three training sessions, participants recorded speeches lasting 4, 3, and 2 minutes. One group repeated a topic, while another spoke on different topics. Both groups became faster during training. Only the repetition group maintained its fluency gain on later tests. The researchers linked the result to proceduralisation: familiar words and sentence-building routines became easier to run under time pressure.
The sample was small, and the activity involved recorded monologues. A Taipei office worker preparing for a client call still has to manage turn-taking, clarification, and another person's accent. The study supports a classroom sequence, rather than a promise that three monologues will produce conversational fluency.
Repetition also makes progress visible at sentence level. On the first recording, a learner may abandon the clause we had expected the supplier to deliver and restart with the supplier was late. The second attempt gives that unfinished clause another chance, after meaning has stopped consuming all available attention. A teacher can compare the two versions, point to an actual change in phrasing that survived the full message, and avoid offering the vague verdict that the learner merely sounded more confident. This close comparison turns a repeated task into evidence for the next lesson, especially when the smoother sentence still contains a tense error that speed can hide.
Repetition needs a job
A second attempt can become empty if the instruction is simply "do it again". The learner needs a narrow target drawn from the first performance. One repeat might focus on keeping pauses between clauses instead of inside them. Another might require two accurate past-tense verbs. A third might ask the learner to replace very difficult with the phrase they searched for earlier, such as hard to coordinate.
I'd choose one target. Five corrections turn the repeat into a memory test about feedback, and the speaker stops attending to the message. This is where teacher judgement matters: the teacher listens for the bottleneck that most limits this learner's next attempt. That same principle sits behind Choc's view that a smaller English class works only when the lesson changes. More available speaking time helps when someone uses it to diagnose and redesign the next turn.
Mohammad Javad Ahmadian and Mansoor Tavakoli studied a more controlled version with 60 intermediate Iranian learners. They assigned learners to four conditions combining task repetition with either pressured or careful online planning. The condition that combined repetition and careful planning produced higher measures of accuracy, complexity, and fluency. The design cannot separate every classroom variable, and its narrative task differs from a job interview or GEPT speaking prompt. It does show why speed alone is a poor goal. Time to monitor language changed what repetition allowed learners to do.
Keep the content, change the pressure
Exact repetition works best over a short span. Imagine a B1 learner on the MRT preparing to describe a delayed software project in Monday's meeting. On the first attempt, they record a 97-second explanation without a script. They listen once and mark the longest pause. On the second attempt, they keep the same facts and aim for clearer clause boundaries. On the third, a partner interrupts after 43 seconds to ask, "What caused the delay?"
The message stays stable while the communicative pressure changes. That progression matters because an unchanged monologue can improve through simple familiarity with one performance. A later variation tests whether the easier phrasing can survive a new demand. Saraswati Dawadi's study of 40 EFL learners gives a useful warning. Repeating a narrative after three days produced more fluent and complex speech, alongside lower accuracy. Familiarity freed resources, yet learners did not spend those resources in the same way. A teacher still has to decide which dimension the next attempt should protect.
A compact classroom sequence
A practical cycle can fit inside 18 minutes. Give the learner two minutes to inspect the task, followed by a first performance with no script. Use three minutes for feedback on one language feature and one delivery feature. Repeat with the same listener. Then change one condition: the listener, the time limit, the purpose, or a single fact in the prompt.
The teacher should keep the first and final recordings when consent and classroom policy allow it. Compare pause placement, repair, and successful phrases. Raw words per minute can reward rushed speech, so it belongs beside comprehensibility and accuracy. A learner who moves from 83 to 96 words per minute while swallowing every final consonant has solved the wrong problem.
There are clear counterexamples. A beginner with too little language to complete the first task needs input and rehearsal before repetition. A learner who memorises a polished script may sound smooth on one topic and freeze on the first follow-up. Someone practising a safety briefing needs accurate content before any pressure to speak faster.
For tomorrow's lesson, resist the fresh prompt for six more minutes. Let the learner return to the sentence that broke down, say the whole message again, and then give the listener permission to interrupt at second 43.