Put a confident B2 speaker beside a cautious A2 learner and give them a shared worksheet. The stronger speaker can read the instructions, propose every sentence, repair each error, and finish the task while sounding admirably helpful. The pair produces good English. One member produces very little of it. Task roles can change that distribution when the quieter learner holds information the partner needs.
The information has to matter.
Proficiency does not decide the role
Mixed-level pairing is often treated as automatic support. A stronger learner supplies language; a lower-proficiency partner observes a better model and receives help. That benefit is possible. So is a division in which one person becomes the author and the other agrees.
Zabihi and Ghahramanzadeh studied 54 learners completing a joint composition in low-low, low-high, and high-high pairs. They examined language-related discussion, including whether partners challenged contributions and how problems were resolved. The study found that interlocutor proficiency shaped engagement with language, with pair types showing different patterns. Its focus on one composition task does not yield a rule that mixed-level pairs always fail or always succeed.
The useful teaching question is smaller: what must each person contribute before the pair can finish?
A worksheet shared by both learners may demand nothing from the quieter partner. Give that learner the customer complaints, map landmarks, interview notes, or missing figures, and silence now blocks completion. The stronger partner has a reason to ask, wait, check, and build from an answer.
Put the lower-proficiency learner in charge of content
Dao and McDonough tested task roles with 60 Vietnamese university learners in mixed-proficiency pairs. Pairs completed a story retell, created an ending, and wrote the story together. The researchers manipulated whether the lower-proficiency learner held or received the information. Their analysis asked whether the role changed discussion of language and pair dynamics, especially equality and mutual engagement. The study reports that assigning the lower-proficiency partner a consequential role can improve the chance that both learners participate and engage with each other's contributions.
This is a design lever, with boundaries. Holding information can increase pressure. A learner who cannot express the content may start pointing, switch entirely to Mandarin, or hand over the sheet. The task needs language support and a manageable information load.
Imagine a Taipei travel-planning task. Partner A, the quieter learner, receives five visitor constraints: arrival at 11:17, a NT$320 lunch budget, a knee injury, a request for one temple, and a train from Taipei Main Station at 18:42. Partner B receives a map and opening times. B cannot make the itinerary alone. A can read or paraphrase each condition, and B has to ask follow-up questions.
Those ugly numbers help. They make checking necessary and prevent a vague exchange from passing as completion.
Give the language before the floor
Information ownership creates a reason to speak. It does not supply the forms needed to speak. Before the task, rehearse four moves that the information holder can use:
- “One condition is that ...”
- “We still need to include ...”
- “That option conflicts with ...”
- “Could you change the plan because ...?”
The list has four items on purpose. A learner needs a way to introduce, add, reject, and request. Put the stems on Partner A's card only, so the support stays beside the role. After one round, remove two stems and repeat with changed details.
Choc Education's article on shared control in pair work describes equality and mutuality. Information ownership is one concrete route toward that condition. It changes the task before a teacher has to interrupt an unequal pair.
Watch the interaction, not the final page
A polished answer can hide a poor learning process. During pair work, sample 47 seconds of each pair and mark three things: who initiates a content move, who asks a question, and who changes the shared answer. A simple tally gives the teacher evidence that the role design worked.
Do not demand equal word counts. One learner may need longer turns because their role includes explaining the map. Look for access to consequential turns. Did the quieter learner introduce information that changed the plan? Did the partner respond to it? Did both people make at least one language choice?
Teacher feedback can target the interaction. “A supplied four constraints, and B checked three of them” is more useful than “good teamwork.” If B still took over, the next round can restrict the materials: B may write, while A alone can see the requirements. Then switch roles with a new case.
When the role should stay flexible
Some learners dislike being designated as weaker, and teachers often misjudge who will be quiet on a given topic. Assign roles from observed participation in the previous round, without labelling proficiency. Call the roles “brief holder” and “planner,” or “client” and “consultant.” Both carry status and a real contribution.
The design also suits same-level pairs. One talkative member can dominate despite equal test scores. Give the less vocal person the evidence, and the activity redistributes conversational rights without a public correction.
There are cases where information-holder status creates too much load: a first lesson, a highly anxious learner, a topic with 19 unfamiliar words, or instructions that already strain working memory. Reduce the brief to two facts, allow 45 seconds of planning, and keep the useful sentence stems visible. Role strength should rise with readiness.
Change one card
Pair work often gets repaired through teacher reminders: “Let your partner speak.” That instruction fades as soon as the task makes one learner optional. Material design lasts longer. The person who holds necessary content has a legitimate reason to enter, correct, and redirect the conversation.
For the next pair task, print two genuinely different cards. Put the map or options on one. Put four constraints and two sentence stems on the other. Give the constraint card to the learner who took fewer consequential turns last time. When the pair submits its answer, ask each person to circle the one detail that came from their own card.