How to practise OET Speaking online: partners, tutors, and AI simulated patients

Four ingredients make a practice role play resemble the real thing: a card you have not seen before, a patient who raises their own concerns rather than simply agreeing, real timing (about three minutes of preparation and about five minutes of role play), and feedback against the published criteria rather than a general impression.

Those criteria come in two families: four linguistic criteria (intelligibility, fluency, appropriateness of language, and resources of grammar and expression) and five clinical communication criteria (relationship building, understanding and incorporating the patient's perspective, providing structure, information gathering, and information giving). Practice that never touches the second family misses about half of what is scored.

A fellow candidate from a Facebook or Telegram study group costs nothing and gives you a real human voice and real unpredictability. The limits are feedback and reliability: your partner is not a trained assessor, their comments are usually impressions rather than criterion judgements, and scheduling across shifts and time zones is hard to sustain.

A tutor brings trained human judgement, hears your pronunciation directly, and can coach the clinical communication habits in real time. The trade-offs are cost per session and availability, which usually limits how many role plays you can do in the weeks before the exam.

You can ask a general assistant such as ChatGPT to play a patient and run a role play by voice or text. It is available at any hour and is useful for rehearsing structure and building fluency. Its limits are worth knowing: unless you instruct it carefully, it tends to be cooperative rather than raising concerns the way OET role play patients are briefed to, a typed conversation does not practise speaking, and its comments are general feedback rather than marking against the OET Speaking criteria.

Purpose-built tools combine the patient and the marking. MedEngly's Speaking practice is a live, hands free, two way role play with a simulated patient that keeps to the scenario on its own card, followed by marking from the transcript against the published linguistic and clinical communication criteria. The first marked role play is free. The limits are stated plainly: a simulated patient keeping to its card is less likely to surprise you than a real interlocutor, and because marking works from a transcript, judgements that depend on how you sound, such as pronunciation, are approximate, and every grade is practice feedback rather than an official evaluation.

A practical mix for most candidates: use a simulator or chatbot for frequent role plays and criterion feedback, use a partner or tutor now and then for a trained human ear, and record yourself every time. Listening back to your own recording remains the single highest value habit; most candidates immediately hear interruptions, jargon, and missing check-ins they never noticed while speaking.