Pair Work That Stalls into English

The recordings show three different problems. First, proficiency: three-to-five-word chunks and few clarification requests are consistent with novice-high to intermediate-low interaction; students may not yet have the language to repair breakdowns. Second, task design: the handout lists questions in order, so students can complete the 'interview' without listening; there is no information gap and no reason to use the partner's answer. Third, norms and rubric: the rubric rewards accuracy and completion, both possible with memorized questions, and gives no credit for listening, turn-taking, or repair. The 38% English is therefore not simply motivation; it is a rational response to a task that can be completed without Japanese. The five students who asked for clarification may have had more repertoire or more tolerance for uncertainty. The 'awkward' comment points to a missing routine: students have not practiced the language of not understanding. The instructional inference is to teach interactional moves explicitly and change the task so the partner's answer is necessary. The English figure is a useful signal, but it cannot by itself tell you whether students lack language, lack a reason to listen, or lack norms.

Redesign as a two-day sequence. Day 1: pre-task. Model a 45-second exchange with a colleague or a recorded video; students identify three repair moves (「もう一度お願いします」, 「ゆっくりお願いします」, 「〜ですか」) and one follow-up move (「へえ、どうしてですか」). Practice in a circle with a single question, rotating partners every 90 seconds. Day 2: information-gap interview. Partner A has a weekend schedule with two missing details; Partner B has a different schedule. They must find the missing information and then report one surprising detail. Add a must-use requirement: at least two follow-up questions and one clarification. Provide a desk strip with question words, past-tense forms, and repair phrases, but not full scripts. The teacher circulates and notes evidence. Post-task: students listen to 30 seconds of their own recording and mark one place where they repaired or should have repaired. Justification: the gap creates a reason to listen; the model and practice make repair language accessible; the follow-up requirement pushes beyond chunk recall. The task remains accountable to the same vocabulary and structures, so it fits the unit rather than becoming an unrelated speaking activity.

Revise the rubric into four criteria: task completion and information exchange, comprehensibility, interaction (initiating follow-ups, responding to the partner, repairing), and language control (vocabulary and structures from the unit). Weight interaction at least 25%. Descriptors should distinguish 'asks the next scripted question regardless of the answer' from 'asks a follow-up based on the answer.' Collect evidence three ways: a 60-second recording at the start and end of the unit, teacher observation notes during the information gap, and a short self-assessment. For the midyear report, report growth on the same criteria, not a single holistic score. About the 38% English: do not ban first language outright; teach an 'English-free 90 seconds' routine and track it, but also analyze when English appears. If it clusters around repair, that is a language gap; if it clusters around off-task talk, that is a norms issue. Share the data with students and let them set a goal, such as two repair phrases per conversation. This keeps the focus on interaction rather than on policing language.