OET Speaking sub-test format and scoring

The Speaking sub-test is the part of OET candidates worry about most, and rightly so. It is the only sub-test where a real human looks back at you, you cannot scroll back, and you have five minutes per role-play to demonstrate skills that take real working clinicians years to develop.

This guide breaks down exactly what happens in the room (or on Zoom), what the role-cards look like, who the interlocutor is, and how the 9 OET Speaking criteria actually map to your score.

How long is the Speaking sub-test?

About 20 minutes total, split into:

| Stage | Duration | What happens | |---|---|---| | Warm-up | 2–3 min | Interlocutor chats with you about your professional background to settle nerves | | Role-play 1 prep | 3 min | You read role-card 1, can take notes, may ask the interlocutor questions | | Role-play 1 | 5 min | You play the healthcare professional; interlocutor plays the patient/family member | | Role-play 2 prep | 3 min | Same as role-play 1 prep, with a different scenario | | Role-play 2 | 5 min | Same as role-play 1, different scenario |

The warm-up is not scored. Use it to calm down — that's its purpose. Everything else is scored, including off-card moments like greetings, transitions, and closings.

What does a role-card look like?

A role-card is a one-page brief that tells you:

• The setting — e.g. "You are a nurse in a community clinic"
• The patient — name, age, presenting concern
• The patient's background — relevant medical history, social context
• The tasks — 4 or 5 numbered cues like "Address the patient's concern about...", "Explain the next steps", "Provide information about..."

The tasks are the spine of the assessment. Every numbered task is a graded cue. If you don't cover a task, you don't earn that mark, even if everything else you say is brilliant.

In OET Live, every case in our 11,000+ bank follows this same task structure. We verify each task against your transcript and tell you hit / partial / miss with the exact quote you used as evidence. The official exam does the same — without the transcript and without the explicit hit/miss feedback.

Who plays the patient?

A trained interlocutor plays the patient or patient's family member. They follow a script for what to say and how to react. They are not the examiner — they are an actor. Their job is to give you something realistic to respond to.

The examiner is a separate person, usually a recording-based marker who scores your role-play after the fact (and a second marker does the same independently). The interlocutor stays neutral about your performance during the role-play itself — they will not nod approvingly when you say something good. Do not read warmth or coldness from them as a signal about how you are doing.

The 9 OET Speaking criteria

Every role-play is scored on 9 criteria, grouped into two clusters.

Linguistic criteria (4)

| Criterion | What it measures | |---|---| | Intelligibility | How easy you are to understand at the sound and stress level. Accent is fine; unintelligibility is not. | | Fluency | Speed, rhythm, hesitations, fillers. Are you keeping a real conversational tempo? | | Appropriateness of language | Vocabulary, register, hedging. Does your language fit a real clinical consultation? | | Resources of grammar and expression | Range and accuracy of grammatical structures. Can you reach for the right structure? |

Clinical communication criteria (5)

| Criterion | What it measures | |---|---| | Relationship building | Greeting, rapport, empathy markers, closing. Did you treat them as a person? | | Understanding the patient perspective | Eliciting concerns, ideas, expectations, fears. Did you listen for what they actually came in worried about? | | Providing structure | Signposting, transitions, agenda-setting. Did the patient know what was about to happen? | | Information gathering | Open and closed questions, screening, summarising. Did you get the information you needed? | | Information giving | Chunks, checks, jargon control. Did you explain things in a way the patient could understand? |

Each criterion is scored independently 0–5 by each of two examiners. The aggregate is what becomes your 0–500 score. The OET band scores explained guide covers how the numeric score maps to the band letter.

What candidates most often lose marks on

In our calibration sets — public role-plays scored by examiners and re-scored by our model — the criteria most commonly under-performed are, in rough order:

1. Understanding the patient perspective — most candidates over-deliver information and under-elicit concerns
2. Providing structure — transitions and signposting feel "extra" to non-native speakers and get skipped
3. Information giving — too much jargon, too little chunking, too few comprehension checks
4. Fluency — usually a knock-on of nervousness rather than language ability
5. Resources of grammar and expression — repeated structures, narrow vocabulary

If you want to lift your band, working on the top three of those is usually the highest ROI. The linguistic criteria are slower to improve but degrade least under exam stress.

How to practice realistically

The hardest thing about OET Speaking practice is finding a partner who:

• Will stay in character as a patient
• Knows the 9 criteria well enough to give per-criterion feedback
• Is available when you have time to practice
• Doesn't cost £50/hour

This is exactly the gap OET Live was built to fill. The AI patient stays in character. The scoring engine grades every criterion. It's available the moment you open the app.

Next steps

• Want to understand the broader test? See What is OET?
• Studying for the first sitting? See How to prepare for OET
• Trying to choose between OET and IELTS? See OET vs IELTS

When you're ready to practice the format described in this guide with full per-criterion feedback, join the OET Live waitlist.