Copy Me

A structure for pairs with three steps, the first of which happens only once. First the class forms pairs and agrees who is A and who is B; both take pen and paper. Then A makes something (a drawing, a construction) and says step by step what A is doing, while B, turned away, makes the same thing from the words alone (ninety seconds). Then the two put their work side by side and point to one instruction that was exactly right and one that was too vague, and say how it could have been better (one minute). The next round the roles swap. Four rounds is the standard, so everyone narrates twice and copies twice. In the preparation you choose whether B may ask back, and you give a task per round, or let A invent one.

Because the eye fills in everything the words skip. As long as B can watch, nobody notices that "over there" and "about this big" say nothing: B simply sees it. Back to back, B has only the sentence A just said, and it has to be good enough. That is exactly the situation of a written explanation, a route description or a work instruction: the other person can neither look nor ask back. The structure practises that in ninety seconds, and the comparison afterwards shows which word was missing. It is also why A's work stays flat on the table until the comparison.

With the default settings the whole structure takes ± 14 minutes: 4 rounds, briefing and look-back included.

One round takes two and a half minutes with the default times: ninety seconds of making and narrating, sixty seconds of putting the work side by side. Four rounds are therefore ten minutes, plus a minute to form pairs and a minute for paper and pens: a quarter of an hour at most. Six rounds are fifteen to twenty minutes. Whoever prepares the structure sets the time per step and picks the number of rounds, from one to twelve, and picks an even number, so A and B narrate equally often. If the pairs are already sitting together, set the first step to zero and the player skips it.

Something of three to five steps that may be done roughly and that B does not already know by heart: a figure of three shapes that touch, a graph with a kink, a map with a river and a town, a cell with three parts, a tower of eight blocks that does not go straight up. The more position, size and direction the task demands, the more there is to say, and the more the comparison teaches. Type one task per line in the preparation; the player shows it to A. No time? Under ‘Set up & start’ the AI writes them for you: choose ‘AI writes them’, give your topic and year group, and you get short drawing or building tasks, which you read through before you start. Prefer Microsoft Copilot? Ask: ‘Come up with 6 short drawing or building tasks about [topic] for pupils aged [age]. Each task should take about ninety seconds to make and be explainable step by step to someone who is not allowed to look, so with shapes, positions and sizes. One task per line, no numbering.’ Paste the answer into the Tasks field; every line becomes a round.

Both are fine, and it is a choice per lesson. With a task per round the class practises subject content (today's diagram, the graph, the map), and you decide which words are needed. Without a task A invents something; the player says so on screen and you only pick the number of rounds. That is the energiser version, and also the best first time: the class learns the structure without the content getting in the way. So an empty field is not half a preparation; it is the other way of filling it.

Listening only if you want A to hear which words were missing; asking back if you want B to learn to ask the right question. In the one-way version B may ask nothing, and in the comparison A sees exactly which instruction was too vague. That is the strictest practice in saying things precisely, and it is the default. With asking back switched on, B may ask "how big?" or "left or right?": easier for A, and B practises the question that closes the gap. For a class doing this for the first time: start with listening only. In a language lesson asking back is often the goal itself. You choose it in the preparation; the steps stay the same, and even with asking back A still may not point or look round.

Three things carry the structure. One: the seating. Back to back or turned away, and A's work flat on the table. Check it in the first round, because whoever can see does not listen. Two: narrating while making. Walk around and listen for step language ("first…, then…, to the left of that…"); have a word with any A drawing in silence. Three: the comparison. Steer towards the third bullet of that step: not "close enough", but which word was missing and how it could have been said. If you hear the same gap in several pairs ("big" with no measure, "there" with no place), that is your whole-class minute after the last round. The player does the rest: the timer, the role swap and the task on screen.

One group of three: one narrator and two copiers, both turned away. At the comparison there are three pieces of work side by side, and two copiers who missed the same thing point out together, more sharply, which word was not said. At the swap the role moves along, so in the three as well everyone gets a turn at narrating. It is better than pairing a student with you, because then you are the copier and you cannot see the rest of the class.

Both work in pairs with rotating roles, but the direction is reversed. In pair coach one solves an exercise and the other asks questions without giving the answer away; it is about practising, and the coach watches along. In Copy Me one makes something and narrates it, and the other copies without looking; it is about putting things into words, and watching along is exactly what is forbidden. Rule of thumb: pair coach when the material needs practice, Copy Me when the language about it needs practice, and for a diagram or a graph, often one after the other.

Read and Ask works with a text that already exists: A reads aloud, B listens without reading along, and A asks a question. Copy Me works with language A has to produce while doing something. The first practises listening to a text and answering a question about it; the second practises putting things precisely yourself and hearing where your words fall short. They share the back-to-back idea (whoever can see does not listen) and complement each other: Read and Ask for understanding, Copy Me for explaining.

Yes, and online the seating is even easier: B's camera off, or B does not share their screen. Pairs in breakout rooms, A draws on paper or in a drawing program and narrates, B draws along without seeing anything. At the comparison both hold their work up to the camera. The player shared in the main room, or on everyone's screen through the live link, gives the step, the timer and the task.

The structure belongs to the family of cooperative pair structures in which two learners take turns in a role and check each other: the structures described by Spencer Kagan. The back-to-back drawing itself is older and broader: in language teaching it is known as an information gap or a barrier game (one person knows something the other can only get through language), and in communication training it circulates as "describe without showing". What makes the structure work in a classroom (narrating while making, not looking, and the comparison as a step of its own with a clock of its own) is recorded here as the player runs it.