Microteaching is a scaled-down teaching encounter, typically 5–15 minutes, in which a trainee teacher practises a specific skill or technique with a small group of peers, followed by immediate feedback. Developed by Dwight Allen and colleagues at Stanford University in the 1960s, it remains a core component of pre-service teacher training programmes worldwide, including CELTA, CertTESOL, and university TEFL/TESOL courses.
The power of microteaching lies in its constraints: by reducing the time, the audience, and the scope, it isolates specific teaching behaviours for focused practice and feedback. A trainee is not trying to teach a whole lesson; they are practising one thing: giving instructions, eliciting vocabulary, setting up a role-play, or managing error correction.
The reteach cycle (teach → feedback → reteach) is what distinguishes microteaching from simply "doing a short demo." It operationalises the principle that teaching skills are developed through deliberate practice with feedback, not through observation or reading alone.
| Skill | Example micro-task |
|---|---|
| Giving instructions | Explain a jigsaw reading task in under 60 seconds |
| Concept checking | Teach the meaning of "used to" and check understanding without asking "Do you understand?" |
| Error correction | Respond to spoken errors during a fluency activity |
| Board work | Present and organise vocabulary on the board |
| Eliciting | Get students to produce target language rather than simply presenting it |
| Setting up pair/group work | Organise a group discussion with clear roles |
| Staging | Transition smoothly between lesson phases |
On the CELTA course, teaching practice (TP) is longer than classic microteaching (typically 40–60 minutes) and involves real learners rather than peers. However, the CELTA retains microteaching principles: focused skills, structured observation, immediate written and oral feedback, and iterative improvement across the course's six assessed TP sessions.
Many CELTA centres also use shorter peer-teaching activities in input sessions, which are closer to the original Stanford model.