From webinar to conversation: redesigning a synchronous online course

Three layers, and they need separating because they call for different responses.

Platform constraints (real but not causal). Mobile data limits video; breakout rooms fail for three learners. These degrade quality and explain why some participants cannot use the richest features. They do not explain 70% teacher talk or silent rooms in the fifteen other cases.

Teaching behaviour. Talking for 70% of a session is a choice, and usually a stress response: silence online feels longer and more ambiguous than in a room, so the teacher fills it. That is understandable and it is also the single biggest driver of the 'webinar' perception. The camera mandate is the same reflex - trying to recreate physical presence instead of designing for the medium.

Task design. Silent breakout rooms usually mean the task was not achievable without the teacher, had no defined product, or had no audience afterwards. 'Discuss these questions' produces silence on screen and often in rooms too. The chat being used only for technical complaints is diagnostic: there was no task that gave the chat a purpose.

Why the distinction matters. If the problem is framed as the platform, the solution is a licence the provider has already said it cannot buy, and the course stays broken. If it is framed as teaching behaviour and task design, the fixes are free: pre-task asynchrony, defined products, and accountability back to the whole group. Note also that cameras raise equity and privacy issues and do not create interaction; requiring them adds friction without addressing the cause.

Design principle: move input and thinking off the live hour so the live hour is only for interaction. The most valuable and least replaceable thing about synchronous time is other people being present at the same moment; use it for nothing else.

Asynchronous pre-task, 48 hours before. A six-minute audio or video prompt plus one question, answered as a written comment or a voice note in a shared thread. This generates the language the live session will work on, gives quieter learners thinking time, and means everyone arrives with something to say.

Live 90 minutes

• 0–10 Check-in, chat-first. Greetings and one quick question answered in text. Low bandwidth, everyone can participate, room warms up.
• 10–25 Language focus drawn from the pre-task responses. The teacher is presenting learner language, not slides, so the input is evidence-based and the learners recognise themselves in it.
• 25–55 Small-group task with a defined product posted to a shared board - a written summary, a ranked list, a recorded 60-second decision. Because breakout rooms fail for three, run an alternative pathway: paired phone or voice call plus a shared document. Audio-only participation is legitimate.
• 55–70 Gallery review. Products are read aloud or displayed; the teacher feeds back on language. This is the accountability that makes the group task real.
• 70–85 Second iteration, harder. Same task type, one added constraint. Fluency comes from repetition, not from new content.
• 85–90 Close and set the next pre-task.

Why this works with the constraints: the three with unreliable breakout still contribute by phone and document; mobile users can use audio and text; the teacher's talk share falls as a by-product of structure rather than willpower.

Indicators that test the actual claim. The course is called 'Conversation and Fluency', so evaluate talk, not attendance.

1. Share of learner talk in the live hour, sampled from two recorded sessions per month and estimated, not transcribed. Baseline is roughly 30%; target above 60%.
2. Number of distinct contributors in audio and in chat per session, tracked by name. This catches the pattern where six people do everything.
3. Pre-task completion rate - if it falls below about 70%, the live session will revert to input delivery.
4. Task performance sample - a recorded 90-second spoken turn at weeks 1, 5 and 10, rated on fluency and interaction criteria by the teacher blind to timepoint if possible.
5. Retention week to week, plus a three-item mid-course survey and exit interviews with two leavers.
6. Learner-reported confidence using a simple repeated scale; weak evidence alone, useful alongside the rest.

Report honestly: small numbers, one teacher, no control group. Say 'consistent with improvement', not 'proven'.

Camera conflict. Withdraw the mandate and replace it with a written rationale plus functional necessity. Explain that the requirement was not creating interaction; then design a jigsaw task in which each participant holds information the others need, so presence becomes useful rather than policed. Offer alternatives - virtual background, audio-only with active chat - and check whether the three with connectivity problems stay in the course. Monitor whether the two who complained disengage. A rule that produces complaints and no learning gain is a rule to drop, and the evaluation should be able to show whether dropping it helped.