OET Speaking is a 20-minute, profession-specific interview with a trained interlocutor who plays a patient or carer. It is not a test of clinical knowledge — it tests whether you can communicate clearly, appropriately and safely in the kind of exchange your profession has every day. Most candidates lose marks not because their English is weak overall, but because they don't understand what the two role-plays are actually assessing or how the criteria map onto the three minutes of preparation they're given. This guide covers how the sub-test is structured, exactly what the assessment criteria mean in practice, and a routine you can repeat with any role-play card to build the habit before test day.
How the OET Speaking role-plays work
After an ID check, the interview opens with a short warm-up conversation about your background and profession. This part is not scored — it exists to settle your nerves and let the interlocutor calibrate to your natural speaking pace before recording begins.
You then complete two role-plays, one at a time. For each, you get a printed card describing the scenario, the patient's situation, and your task as the health professional. You have 3 minutes to prepare, during which you can write notes directly on the card, and you keep the card in front of you throughout the role-play — nothing needs to be memorised.
The interlocutor has their own confidential card describing how the 'patient' will behave: what they're worried about, what they might resist, what questions they'll ask. This means the role-play is semi-scripted but genuinely interactive — you have to listen and adapt, not deliver a monologue.
Each role-play runs for around five minutes and is audio-recorded. The whole interview, including the unscored warm-up, is later marked by two independent assessors against the same criteria, and your final Speaking grade combines both role-plays.
Two kinds of criteria: linguistic and clinical communication
OET assesses Speaking against two separate criteria groups, and it's the interaction between them that trips candidates up if they only prepare for one.
The linguistic criteria judge your English as language: intelligibility (can you be understood without undue effort), fluency (do you speak at a natural pace without excessive hesitation), appropriateness of language (register, tone, and technical language suited to a patient rather than a colleague), and resources of grammar and expression (range and accuracy of structures and vocabulary).
The clinical communication criteria judge how you manage the interaction itself, independent of grammar: relationship building (a natural, professional opening and rapport throughout), understanding the patient's perspective (picking up on and responding to their concerns, not just their symptoms), providing structure (signposting what you're doing and why, e.g. 'Can I ask you a few questions first?'), information gathering (asking relevant, well-sequenced questions), information giving (explaining clearly, checking understanding, avoiding jargon dumps), and dealing with the patient's concerns.
A candidate can have very accurate grammar and still score poorly on clinical communication by ignoring what the 'patient' says and ploughing through a prepared script. Equally, someone with minor grammatical slips but strong rapport and clear explanations can score well overall. Both criteria groups count, so preparation has to address both.
A practical practice routine
You don't need a partner for every session, but you do need to rehearse the interactive parts, not just plan what you'll say. Use this routine with any two role-play cards from your profession:
Read the card once for the situation, once for your task, and once to note the patient's likely concern in the margin — most cards hint at it.
In your 3 minutes, jot 4–6 short prompts, not full sentences. Full sentences tempt you to read rather than talk.
Say your opening line out loud before you start the timer. Relationship building is scored from the very first sentence.
Record yourself doing the role-play with a study partner, a tutor, or by voicing both sides. Listening back is the fastest way to hear filler words, rushed pacing, or jargon you didn't notice while speaking.
After each attempt, check off one linguistic criterion and one clinical communication criterion you specifically improved — trying to fix everything at once slows progress.
Repeat with a card you haven't seen before under real time pressure at least once a week in the run-up to the test.
Task by task
Warm-up interview
A few minutes, unscored Not scored — settles nerves and lets the interlocutor calibrate to your speech
How to do it
1Answer briefly and naturally about your professional background and experience.
2Treat it as a real conversation, not a test — there is no advantage to over-preparing this part.
3Use it to settle into a natural speaking pace before the recorded role-plays begin.
Common mistakes
Over-rehearsing a speech about your career that sounds memorised.
Assuming this section is being scored and becoming visibly tense.
Role-play 1: information gathering
3 min prepare · ≈5 min role-play Linguistic criteria + clinical communication, weighted towards building rapport and gathering the patient's history
How to do it
1Open with a natural greeting and a brief explanation of what will happen in the conversation.
2Ask open questions first to let the patient explain in their own words, then narrow with specific questions.
3Listen for and acknowledge the concern behind the symptoms, not just the symptoms themselves.
4Keep your notes brief so you're looking at the patient (or interlocutor) rather than reading from the card.
Common mistakes
Firing a checklist of closed questions without reacting to the answers.
Skipping an opening and launching straight into questions.
Missing an emotional cue the interlocutor gives, e.g. hesitation or worry.
Role-play 2: explaining & reassuring
3 min prepare · ≈5 min role-play Linguistic criteria + clinical communication, weighted towards explaining a plan and handling concerns
How to do it
1Structure the explanation: what the finding or plan is, why, and what happens next.
2Use plain language for anything a patient wouldn't be expected to know as a professional term.
3Pause to check understanding and invite questions rather than delivering one long block of speech.
4Address any concern directly and specifically instead of a generic reassurance.
Common mistakes
Using unexplained clinical terminology as if speaking to a colleague.
Talking continuously without checking the patient has understood.
Dismissing a stated worry instead of responding to it.
Common questions
Do I need clinical knowledge to do well in OET Speaking?
No. The role-play cards give you the clinical information you need. OET Speaking assesses communication skill, not medical accuracy, so prior subject knowledge cannot substitute for language ability and a lack of it will not cost you marks.
Can I really keep the role-play card during the conversation?
Yes. You may write on the card during the 3-minute preparation time and keep referring to it throughout the role-play, so nothing needs to be memorised word for word.
Is the warm-up conversation at the start scored?
No. The opening conversation about your background is unscored. It's there to check your identity, put you at ease and let the interlocutor adjust to your speech before the two role-plays are recorded and marked.
What's the difference between the linguistic criteria and the clinical communication criteria?
The linguistic criteria (intelligibility, fluency, appropriateness of language, resources of grammar and expression) assess your English as language. The clinical communication criteria (relationship building, understanding the patient's perspective, providing structure, information gathering, information giving) assess how well you manage the interaction itself. Both are scored, so strong grammar alone is not enough if you don't respond to the patient's concerns.
How long is each role-play and how much time do I get to prepare?
Each role-play card gives you 3 minutes to prepare, followed by a role-play of around 5 minutes with the interlocutor. There are two role-plays in total, making up the profession-specific, recorded portion of the Speaking sub-test.
Now rehearse the role-play
Criteria make sense once you speak them. Linguently gives you healthcare role-play practice with AI feedback on clarity, structure and empathy, plus partners to rehearse with.