Speaking is what they came for and what gets cut

It is also the hardest thing to manage in a full room.

Ask students what they most want and they say speaking. Ask them what they got least of and they usually say the same. The reason is structural: speaking practice is difficult to manage with twelve to eighteen people, hard to assess, noisy, and less visible as evidence of teaching than a completed worksheet.

The arithmetic is worth confronting. In a fifty-minute lesson with sixteen students, whole-class question and answer gives each student well under a minute of production. Pair and small group work is the only structure that changes that number materially, and it requires accepting a room where the teacher cannot hear everything.

The common objection is that unmonitored pair work produces errors that go uncorrected. That is true and it is a smaller cost than the alternative, which is a room where nobody speaks. Sampling pairs, noting recurring errors, and addressing them to the class afterwards captures most of the benefit.

Fluency and accuracy work need to be separated explicitly, and students told which is happening. A student corrected mid-sentence during a fluency activity learns to stop speaking, and adult learners in particular will comply permanently.

It is worth adding that mixed ability is partly a timetabling artefact rather than an inevitability. Schools running many levels can place more precisely than schools running three, and the number of levels is a commercial decision about class sizes. Teachers frequently absorb the consequences of that decision without it ever being named as its cause.