A seminar at a British university, ten or so students around a table. The tutor asks a question and nobody answers it. The silence runs longer than silences usually run in that setting, long enough that everybody in the room registers that it is happening.
Then the students sitting near one particular student turn and point at her. There is no whispering first. Nobody says her name. They point, and she answers. The seminar carries on.
We were told this by the student who was pointed at. She is British, she is a friend of ours, and as far as she could tell she was the only person in that group whose first language was English. That last fact matters less than it looks. Academic English is learned rather than inherited, and academic literacy is plural and specific to a discipline rather than a single transferable skill somebody either has or lacks (Lea and Street, 1998), so our friend was acquiring it at that table like everybody else around it. Her English was not the thing under strain in that room, and that is the reason we have the account at all. It is also the thing that limits it. We will come back to that.
What the room had already decided
What we keep returning to is the speed. Whatever happened in that room happened in about two seconds, without anybody proposing it and without anybody agreeing to it out loud. There was no negotiation, because a negotiation would have required somebody to speak, and the whole difficulty was that speaking was the thing under pressure. Several people arrived at the same solution at the same moment and carried it out silently. That only works if the solution was already understood before the question was asked. The pointing did not invent the arrangement. It made an existing arrangement briefly visible.
The ordinary way to describe seminar silence puts the problem inside one person. A student does not have the words, so the student says nothing. It is an account of an individual coming up short, and it leads naturally to a picture of what would fix it, which is more words for that individual. What we were told about is a different kind of event. It is a room working out how it is going to handle a language problem, arriving at an answer, and putting that answer into effect without any of the usual machinery of a group decision. Everyone in that seminar understood who was going to speak. The arrangement was social before it was linguistic.
The two diagnoses we have run together
We should say what the account does to a position we have stated elsewhere. Asked partway through this project whether most seminar silence is about risk rather than vocabulary, we agreed, and then in the same answer replaced risk with something else: students have not had anywhere to practise speaking. That substitution is not a small one. Risk is social, and it is about what it costs you in front of other people to produce a sentence badly. Practice is a skill problem, and it is about whether the sentence comes out at all. They are different diagnoses, they imply different treatments, and the practice diagnosis happens to be the one our product already addresses. Elsewhere again, asked directly whether seminar silence is a language problem, we answered yes with no hedging at all.
Our own best material argues against us. Nothing in the pointing story is a story about words being unavailable to somebody. It is a story about a group that had already settled the question of who talks. A person who cannot produce a sentence and a group that has decided in advance which sentences come from where are not the same problem, and we have been describing the first while the evidence we are proudest of is evidence of the second. We have not resolved this. We are recording it because a company that only reports the evidence agreeing with its product is not worth reading on the subject.
Now the limit, and it is a real one. We have the behaviour and we do not have the experience. We know what the room did. We do not know what it was like to be one of the students who pointed, whether it was a relief or a humiliation or nothing much at all. Our account comes from the one person in that seminar whose English was not under strain. She could see the arrangement precisely because she was outside it, which makes her a good witness to the mechanics and a poor one to the feeling.
No first-hand account of the silence
Neither of us has ever been the person in a room who could not follow. We are a maths student and an engineering student. We have friends who have lived in other countries, and that is not the same thing, which we said before anybody put the objection to us.
So we asked our users. Two wrote back. One had spent about two days with the app and sent several paragraphs. She named the roleplay as the most useful part for her, and explained why in a sentence better than anything we have written about it: for Chinese students, we only remember and spell but don't know how to use it. She also asked for one-click flashcard creation and for points that unlock themed decorations. The other wrote three sentences. She had uploaded one of her lecture slide decks and said the questions it generated helped her spot points she might have missed before.
Neither of them mentions seminars. Neither mentions speaking up, or how it feels to sit through a question you cannot answer. We went looking for that account and we did not have one to find. So the part of this product that carries the most weight in how we talk about it has no first-hand testimony behind it at all.
The origin is proximity, not experience
It is worth saying where the company came from, because it is the same limit in an earlier form. UniFluent exists because we were undergraduates sitting in tutorials next to international students who were visibly struggling, and struggling more at the start of the year than later. There was no market research, no theory, and no prior view between us about what university-level English even was. We talked to the people it was happening to, we concluded quickly that the problem was real, and we could not find anything that was specific to the language of a student's own course.
That origin is the honest one and we would not trade it for a tidier story. It is also proximity rather than experience, and it took writing this document to notice how much weight we had been resting on the difference between the two.
What we have built for that room is narrow, and we would rather understate it than have it read as a solution to the seminar. The roleplay offers the least explicit hint first and escalates only as far as it needs to. We did not know until this year that this has a name and a literature behind it. It is what Poehner and Lantolf call dynamic assessment, where the amount of help a learner needs is itself the measurement, and we arrived at it because it seemed like the decent way to treat somebody mid-sentence rather than because we had read anything. That much is built and we can describe it. We cannot tell you what it produces, because nobody has used UniFluent for a hundred hours and there is no efficacy data of any kind.
We should also be exact about a limitation that bears directly on the seminar in question. Uploaded course material drives vocabulary extraction only. It does not feed the voice tutor and it does not feed the writing tutor. A student can practise speaking with us and practise none of the actual content of the seminar they went quiet in. That is buildable and it is not built, and until it is, the thing we point to when people ask about speaking is not connected to the thing we point to when people ask about their course. Anyone reading our marketing would reasonably assume otherwise, which is our fault rather than theirs.
A room we do not enter
Before any question about method there is a question about purpose, which is what a piece of teaching is meant to help a student become. We can answer that about the app. We cannot answer it about the seminar, because the seminar is a room we do not enter and have no means of entering. It belongs to the tutor who asked the question and to the people around that table. The most we have ever been willing to claim is that our job is to get them ready for those social interactions, and even that splits into a part we can work on and a part that belongs entirely to other people. The readiness is ours. The interaction is not.
So the position we are left holding is that we have built a product around a feeling neither of us has had, on the word of somebody who was not the one having it.
The question we cannot answer is what the pointing meant to the people who did it. Our friend did not know either. She was the one being pointed at, and she told us the story as something that had happened to her, which is the only way she could have told it. We do not know who else was in that room and there is no way now to go back and ask them. Everything we have inferred about seminar silence rests on two seconds we did not witness, reported by the one person present for whom those two seconds were not the problem.
