Conversation practice often assumes that both speakers will understand enough to keep going. A learner in a Taipei video meeting may catch seven words out of ten, miss the date that controls the whole decision, and reply “OK” because stopping a senior colleague feels risky. English lessons should teach a small set of repair phrases and require learners to use them under pressure. Repair is a speaking skill with observable moves, not a personality trait.
“Please repeat” is too blunt
“Could you say that again?” has a place. It also asks the other person to reproduce an entire turn, perhaps at the same speed. Skilled repair narrows the missing piece. If the missed material was one number in a 38-word explanation, repeating all 38 words creates another search problem. Naming the uncertain number tells the partner where to slow down, what to contrast, and when the listener is ready to continue. The difference becomes obvious in practice during a noisy call, where every extra sentence competes with the same bad connection.
Compare four moves:
- “Did you say the thirteenth or the thirtieth?” offers two candidates.
- “What does rollover mean here?” identifies one unknown item.
- “So the Taipei team sends the draft, and Manila approves it?” checks an interpretation.
- “I got the budget part. Could you repeat what happens after approval?” marks the boundary of understanding.
Each sentence gives the partner information about the breakdown. That information makes the response easier to adjust.
Strategy instruction can change behaviour
Yasuo Nakatani studied explicit communication-strategy teaching in a 12-week communicative EFL course with 62 female learners. One group received awareness-raising work on oral strategies while a comparison group followed implicit instruction. The 2005 study reported stronger oral-test performance for the strategy group and a positive relationship between awareness and language development. The same sample size appears in Nakatani's 2010 classroom study, which examined Japanese college students through test transcripts, an oral communication strategy inventory, and retrospective reports. Strategies for maintaining discourse and negotiating meaning were associated with better communicative performance. The learners produced relatively few examples of modified output, suggesting that getting clarification does not automatically produce much work on grammatical form.
These were small higher-education cohorts in Japan, and one study involved only women. Self-report data can also differ from what speakers actually do. The evidence supports teaching and practising repair moves in communicative tasks. It does not establish one universal list of phrases or promise broad proficiency gains from memorising them.
Teach a sequence, not a poster
A classroom wall can display “Sorry?”, “Pardon?”, and “Could you repeat that?” for a year without changing a single conversation. Learners need a sequence that links a listening diagnosis to a response. Begin with the size of the gap, offering alternatives such as “Was that fifteen or fifty?” when one sound is uncertain, naming a single unknown expression, or paraphrasing the full message when every word was audible but the intention remained unclear. If the whole turn vanished, request repetition and specify a helpful change: “Could you say the last sentence again, a little more slowly?”
Then wait.
Then listen to the repair. A learner who asks a good question and mentally leaves the conversation has completed only half the move. The partner's new wording must lead to a confirmation, a revised understanding, or another focused request. Short phrases help because the listener can retrieve them while attention is already overloaded. Still, scripts can sound abrupt across status relationships. “What?” may work between close friends and land badly in a meeting with a new client. “When you said X, did you mean Y?” costs several more words and signals care.
Put missing information in the task
Repair practice fails when every learner holds the same worksheet. If both partners can see Flight BR716 leaves at 09:37, nobody needs to check the number. Give Partner A a departure board and Partner B a disrupted itinerary. Add one smudged gate, one similar number, and a connection constraint. The task now creates reasons to clarify.
The teacher can score visible actions without pretending to measure a vague quality called confidence. In a six-minute pair task, record whether each learner:
- identified a specific gap at least twice,
- checked one interpretation by paraphrasing,
- responded to a partner's repair request with changed wording rather than louder repetition,
- confirmed the repaired meaning before moving on.
Four criteria are plenty. The first round can use phrase cards. The second round removes them and changes BR716 to CX421. A third round one week later can move from travel to a product-delivery call.
This design pairs with Choc's article on information-gap speaking tasks. The gap creates the need. Repair phrases keep learners working when the exchange goes wrong.
The partner needs training too
Courses sometimes train the person who misunderstands and ignore the speaker who caused the problem. Repair is joint work. Repeating the same sentence louder may preserve the same unknown word, the same compressed pronunciation, and the same ambiguity. Give the responding partner three options: repeat a shorter span, replace the difficult expression, or add an example. If “We need to roll it over” caused the gap, a useful response might be “We move the unused budget into next month.” For a number, write it in the chat after saying it. For a name, spell it once.
There is a boundary here. Learners do not need to interrupt every time one article or adjective escapes them. Excessive checking can break the thread and burden the partner. Ask for repair when the missing material changes the decision, blocks the next turn, or seems likely to cause a false agreement.
Practise the uncomfortable moment
Status changes the task. It is easier to ask a classmate to repeat than to stop a visiting director. Role cards should name the relationship and the consequence: a teammate discussing lunch, a customer confirming NT$18,600, or a supervisor moving a deadline from October 13 to October 30. The language and hesitation will differ. Teacher feedback can focus on one missed opportunity. At 02:47, the learner said “OK” after an ambiguous date. Rewind eight seconds, let both speakers choose a repair move, and resume from there. The correction remains attached to a communicative cost.
During a role-play, a learner who hears the deadline unclearly must decide in real time whether to risk interrupting, which phrase fits the relationship, and whether the partner's second version has actually resolved the uncertainty. For the next pair task, hide one number and one unfamiliar expression from each partner. Give them 90 seconds to review four repair stems, then start the call. Count the moments when a vague “OK” becomes a precise check.