How OET Speaking is marked, in one paragraph
Your two role plays are recorded and the recording is marked later by two trained OET assessors, not by the interlocutor who played your patient. Each assessor scores you on nine criteria: four linguistic criteria marked out of 6 (intelligibility, fluency, appropriateness of language, resources of grammar and expression) and five clinical communication criteria marked out of 3 (relationship building, understanding and incorporating the patient's perspective, providing structure, information gathering, information giving). Per the OET results and scoring page, the scores for each criterion are combined and converted to the 0 to 500 scale, and 350 is Grade B, the grade the NMC, the NMBI, the GMC and ECFMG ask for in Speaking (AHPRA's table for tests from 23 April 2026 asks for 360; the GMC and ECFMG figures are from the GMC page on using your OET certificate and the ECFMG Pathways OET page, both read on 20 October 2026). The OET Ready speaking page says a Grade B candidate has "predominantly" 5 out of 6 on the linguistic criteria and 2 out of 3 on the clinical communication criteria. That is the whole system. Everything below explains what each of those nine lines means in a real five-minute consultation, and what the difference between a 4, a 5 and a 6 sounds like.
- Two role plays, five minutes each, three minutes of preparation before each one. About 20 minutes in total.
- Only the two role plays are scored. The ID check, the warm-up and the preparation minutes are not.
- Nine criteria: 4 linguistic (0 to 6 each) and 5 clinical communication (0 to 3 each).
- Assessed from the recording by two OET assessors (OET Ready speaking page). The interlocutor "does not assess you" (OET Speaking Platform User Guide, June 2026).
- Grade B = 350 on the 0 to 500 scale, reported in ten-point steps.
The four linguistic criteria (0 to 6)
These are the "English" half of the mark. OET's own descriptions, from its criteria overview post (5 November 2024) and its Speaking Guide Part 1 on the linguistic criteria, focus on one thing: the effect on the listener. Assessors are not counting errors. They are asking how much effort the patient has to make to understand you. The middle column below paraphrases OET; the right-hand column is our reading of where marks go.
| Criterion | What it measures (OET, paraphrased) | What costs marks in practice (TalkDrill) |
|---|---|---|
| Intelligibility | The impact of your pronunciation, intonation and accent on how clearly the listener can hear and understand you. | Stress on the wrong syllable in medical words, flat intonation that hides which word matters, swallowed word endings. An accent itself is not penalised; unclear sounds are. |
| Fluency | The impact of the speed and smoothness of your speech on the listener's understanding. | Long silent pauses mid-sentence, rushing through an explanation, filler ("basically", "actually") in every clause. Slowing down deliberately to let bad news land is good fluency, not poor fluency. |
| Appropriateness of language | The impact of your language, tone and professionalism on the listener's understanding and comfort. | Jargon to a lay patient ("hypertension", "NPO", "analgesia"), orders instead of requests, a tone that is too casual or too cold for the moment. |
| Resources of grammar and expression | The impact of your grammatical accuracy and vocabulary choices on the listener's understanding. | A narrow range of structures, tense slips that change the meaning ("you will take" vs "you took"), repeating the same three verbs. Range and accuracy both count. |
Two of these deserve a word for our readers. Intelligibility is, in our judgement, the criterion Malayalam, Tamil and Hindi speakers most often need to lift from 4 to 5, usually on word stress rather than individual sounds: "diaBEtes" and "HYpertension" are heard, "DIabetes" and "hyperTENsion" are not. Our word stress read-aloud drill is built for exactly that. Appropriateness is where Tagalog speakers can lose a point by being too deferential ("ma'am", "sir", "if it is okay with you") so that an instruction sounds like a suggestion. Both are fixable in weeks, which is why the 2025 data below shows Speaking is the sub-test our readers already pass most often.
The five clinical communication criteria (0 to 3)
This is the half that IELTS does not have and that candidates coming from IELTS preparation under-prepare. Each criterion is marked 0 to 3 and, per the OET results page, has three to five sub-indicators. The descriptions below are OET's, from the criteria overview post; the "sounds like" column is ours.
| Criterion | OET's description | What a 2 or 3 sounds like (TalkDrill) |
|---|---|---|
| Relationship building | Starting the conversation in a friendly and respectful way, listening carefully, not judging the patient, showing understanding for their feelings and situation. | "Good morning, Mrs Thomas. I'm Anu, the nurse looking after you today. I can see you're uncomfortable. Is it all right if we talk about the pain for a few minutes?" |
| Understanding and incorporating the patient's perspective | Asking about and understanding the patient's thoughts, worries and expectations, and responding to hints or changes in behaviour. | "You mentioned your neighbour had a bad reaction. What worries you most about starting this tablet?" and then using the answer in the explanation that follows. |
| Providing structure | Organising the conversation logically, clearly indicating when you change topic, using techniques to make explanations clear and organised. | "There are three things I'd like to cover: how the wound is healing, what to do at home, and when to come back. Shall we start with the wound?" |
| Information gathering | Open questions first, then specific ones; no confusing or leading questions; asking for more detail where needed; summarising to check understanding. | "Tell me about the pain." then "When is it worst?" then "So it's sharp when you stand, and better lying down. Have I got that right?" |
| Information giving | Finding out what the patient already knows, pausing to check understanding, encouraging reactions and questions, checking what more they need. | "What have you already been told about warfarin?" then information in small pieces, then "How does that sound to you?" rather than a three-minute monologue. |
Notice how little of this is about vocabulary. A candidate with modest English who asks an open question, listens, summarises and then explains in two or three short pieces will score 2 on most of these. A candidate with excellent English who delivers a flawless two-minute lecture on insulin will score 1 on information giving and 1 on the patient's perspective. In our judgement this is the most common reason a fluent doctor comes out with a C+: they prepared the English and not the consultation.
Two assessors, the recording and the 0 to 500 scale
Here is the process as OET describes it across its results page, its Ready speaking page, the Test Handbook 2026 and the Speaking Platform User Guide (June 2026).
- The session is recorded; the warm-up is not scored. The interlocutor checks your ID (on the Speaking Platform this includes a camera check of you and the room) and asks a few warm-up questions. OET's Speaking Platform User Guide says this part of the conversation "is not assessed"; the handbook allows about two to three minutes for it.
- Two role plays, two separate chances. You get three minutes with each card (you may make notes on it and refer to it at any time during the role play, per the OET Ready speaking page; the Test Regulations say you will not be given any explanation of the test questions), then about five minutes of role play. OET says a role play usually comes to a natural end at around the five-minute mark, and that the interlocutor signals clearly if it does not.
- Two assessors mark the audio. They work from the recording, not from the interlocutor, who OET says "does not assess you". The Ready speaking page says two official OET assessors; the criteria overview post says at least two.
- Nine criterion scores become one scaled score. OET combines the criterion scores and converts them to the 0 to 500 scale, reported in ten-point increments. OET does not publish the conversion table, so no coaching institute's "350 calculator" is official. What OET does publish is the Grade B profile: predominantly 5 of 6 and 2 of 3.
- The scaled score becomes a letter grade. OET reports a score and a grade per sub-test and no overall grade (OET test statistics page).
| Scaled score | Grade | Who asks for it in Speaking |
|---|---|---|
| 450 to 500 | A | Nobody requires it; it is reported. |
| 350 to 440 | B | NMC, NMBI, GMC and ECFMG. AHPRA asks for 360 in Speaking for tests from 23 April 2026 (see regulator requirements). |
| 300 to 340 | C+ | Some regulators accept C+ in one sub-test, usually Writing, not Speaking. |
| 200 to 290 | C | Below every major regulator's requirement. |
| 100 to 190 | D | Well below. |
| 0 to 90 | E | Lowest band. |
The band boundaries are those printed in the grade table of the OET Test Handbook 2026. If you want to understand release times, re-marking (within 72 hours of receiving results, per the OET results and scoring page) and how to share results with a regulator, our results and grades page covers it.
What a C, a B and an A opening sound like
The first thirty seconds set the relationship-building score and the structure score, and they colour how the assessors hear everything after. Below is one nursing card and three openings. All three are written by TalkDrill as teaching examples; they are not transcripts of real candidates. The card is in the style of an OET role play; it is not an official OET card. The criterion scores in the headings are our estimates of how the criteria would apply; OET does not publish a conversion table, so no scaled score is attached.
The card (nursing, hospital ward)
Setting: Surgical ward. Nurse: You are looking after a 68-year-old patient who had a hip replacement two days ago. The physiotherapist wants the patient to walk to the corridor and back today. The patient has refused, saying the pain is too bad and the hip might "come out". Task: Find out the patient's concerns about walking. Explain why early walking matters after this surgery (blood clots, stiffness, recovery time). Reassure the patient about the hip and the pain relief available. Agree a plan for today.
A Grade C opening (roughly 3 to 4 on the linguistic criteria, 1 on relationship building and structure)
"Hello sir. I am nurse. Today physiotherapist is telling you have to walk, but you are refusing, why? Walking is very important after the hip replacement surgery because otherwise there is chance of deep vein thrombosis and also the joint will become stiff. So you should walk. The pain we can manage with analgesics. Okay? So shall we go now?"
What the assessors hear: no name, no permission sought, a closed accusatory question ("why?"), jargon ("deep vein thrombosis", "analgesics"), an instruction ("you should walk") before any listening, and the whole card delivered in forty seconds. The English is understandable, so intelligibility might reach 4, but the patient has not been asked a single open question. Expect 1 on relationship building, information gathering and the patient's perspective, and nothing like the Grade B profile.
A Grade B opening (predominantly 5 and 2)
"Good morning, Mr Joseph. I'm Priya, one of the nurses on the ward today. How are you feeling this morning? ... I understand the physio came to see you about walking a little, and that you'd rather not. That's quite common after this operation, so I'd like to hear what's worrying you, and then I can explain why we encourage people to get moving. Would that be all right? ... So what is it about walking that concerns you most?"
What the assessors hear: a greeting with names, an open check-in, a non-judgemental summary of the situation, a signpost of what will happen next ("hear what's worrying you, and then explain"), permission sought, and an open question that invites the patient's perspective. The grammar is accurate and the vocabulary is plain. This is a 2 on relationship building, structure and information gathering, and a 5 on appropriateness. It is not fancy. It is what the B profile looks like.
A Grade A opening (6 and 3)
"Good morning, Mr Joseph. I'm Priya, I'll be looking after you until this evening. May I sit down? ... I can see from the way you're holding yourself that the hip is quite sore today. ... The physio mentioned you weren't keen to walk just yet, and I completely understand; two days after an operation like this the idea can feel frightening. Before I say anything about walking, I'd really like to understand how it feels from your side. Could you tell me what goes through your mind when you think about standing up?"
What distinguishes this from the B: the nurse picks up a non-verbal cue (how the patient is holding himself), names the emotion before asking about it, explicitly delays information giving until the patient has spoken, and the question is open, warm and specific. Intonation does the work: the voice falls and slows on "frightening". This earns 3 on relationship building and the patient's perspective. Note that the words are still simple; the A comes from timing and empathy, not vocabulary.
Try the same card yourself with an AI patient in our role-play practice, record it, and mark your own opening against the three above. More nursing cards with full model dialogues are on the nurses' role-play page; medicine cards, including breaking bad news, are on the doctors' page.
What the 2025 data says about Speaking
Coaching adverts tell nurses that Speaking is the sub-test to fear. OET's own numbers say the opposite for our readers. The table is taken from the OET test statistics page, 2025 demographic data PDF (first-language and nationality tables; "B or above" is Grade A plus Grade B, rounded to one decimal place as printed).
| Group (self-reported, 2025) | Speaking B or above | Listening B or above | Reading B or above |
|---|---|---|---|
| Malayalam first language | 85.0% | 27.8% | 24.0% |
| Hindi first language | 89.1% | 58.2% | 48.9% |
| Tagalog/Filipino first language | 85.4% | 67.4% | 43.9% |
| Arabic first language | 91.1% | 66.1% | 69.7% |
| Indian nationality | 85.9% | 37.3% | 32.0% |
| Nursing (all candidates) | 82.2% | 32.3% | 24.9% |
Read it honestly in both directions. Speaking is the sub-test most of our readers already clear, so the right amount of Speaking preparation is enough to hold a B under nerves, not months of drilling while Reading sits at C. But roughly 15 per cent of Malayalam and Tagalog speakers still finished below B in Speaking in 2025, and the NMC, AHPRA and the GMC do not accept a C+ in Speaking. If you are in that 15 per cent, it is almost always the clinical communication half, which is quicker to fix than pronunciation. Our Listening and Reading page is where the data says most of the preparation time should go.
A self-score sheet you can use tonight
Record one five-minute role play (a phone voice memo is fine, or the recording from a TalkDrill AI role play). Listen back once per column. Give yourself the mark the evidence supports, not the mark you hope for. The questions are ours; the criteria and scales are OET's.
| Criterion (scale) | Ask yourself | Common mark loss (our judgement) |
|---|---|---|
| Intelligibility (0 to 6) | Could a stranger write down every medicine name I said? Did my voice rise and fall, or stay flat? | Word stress in multi-syllable medical words; final consonants dropped. |
| Fluency (0 to 6) | Any silence over three seconds? Did I speed up when explaining? | Pauses while searching for a word; rushing the explanation. |
| Appropriateness (0 to 6) | Any word a patient would not know? Any order where a request would do? | Jargon; "you have to" instead of "I'd suggest"; over-formal "kindly". |
| Grammar and expression (0 to 6) | Did tenses match the timeline? Did I use more than one way to say "important"? | Progressive for states ("I am having"); articles dropped; narrow range. |
| Relationship building (0 to 3) | Names? Permission? One sentence that showed I noticed how they felt? | Straight into the task; no empathy statement before advice. |
| Patient's perspective (0 to 3) | Did I ask what they think and worry about, and then use the answer? | Asked, then ignored the answer and gave the planned speech. |
| Providing structure (0 to 3) | Did I say what we would cover? Did I mark each topic change? | No signposting; topics in card order with no transitions. |
| Information gathering (0 to 3) | Open question first? A summary to check I understood? | Closed and leading questions; no summary. |
| Information giving (0 to 3) | Did I ask what they already knew? Pause every two sentences to check? | A long monologue; "Okay?" as the only check. |
If you are at 5 and 2 on most rows, you are at the B profile OET describes. If a row sits at 4 or 1, that is your week's work. For Malayalam, Tagalog, Hindi and Arabic speakers, the specific habits behind each low row, and the drills, are on the first-language pages. The official criteria PDFs, the Linguistic and Clinical Communication guides and the sample role plays are free on the OET Ready speaking page; use those, not a coaching centre's paraphrase.