Fourteen minutes, four parts, two candidates. Nothing on this page is marked, because speaking is not marked by a page. Record yourself on each part, play it back against the model, and say it better. The second recording is where the improvement is — not the first.
The examiner asks you and the other candidate questions in turn. Nobody is trying to catch you out: this part exists to settle you and to let the examiner hear ordinary, unrehearsed English. Answer, then extend — about twenty to thirty seconds each time.
What the examiner is listening for in Part 1: that you answer the question that was asked, that you go past one sentence without being asked twice, and that you stop. Talking for ninety seconds here costs you nothing in marks and costs you time you will want in Part 3.
A strong twenty-five-second answer to “What do you use English for outside your lessons?”:
“Mostly listening, actually. There's a podcast I've got into where two journalists argue about something British every week — queues, the BBC, the weather — and they interrupt each other constantly, which is the part I find useful. I understand the textbook English perfectly well. It's the interrupting I can't do yet.”
Why it works: it answers in the first three words; it gives one concrete example rather than a category; it uses a present perfect (I've got into) naturally rather than displayed; and it ends on a short sentence, which is how you signal you have finished without trailing off.
Three phrases worth owning for Part 1: “Mostly …, actually.” · “I suppose the honest answer is …” · “Not really — what I do instead is …”
Each candidate speaks alone for one minute about two photographs, and then the other candidate answers a short question about the same pictures for about thirty seconds. You are not describing the photographs. You are comparing them and answering the question printed above them — and the question is where the marks are.
The shape of a one-minute long turn. Fifteen seconds of comparison, thirty-five seconds on the question, ten seconds of speculation to finish. If you spend the whole minute describing what is in the pictures you can still only reach the middle band, however accurate your English is — because you have not answered the question.
A strong long turn for the two museums:
“Both of these are people visiting something, but the experience looks completely different. In the first one there must be two hundred people in the room, and they're all photographing the same object over each other's heads, so I doubt anybody is actually reading anything. The second one is the opposite — one visitor, one volunteer, and a conversation.
As for why they've come: the people in the big museum have probably come for the famous thing itself. It's on the list, they're in the city for two days, and they want to have stood in front of it. Whereas the woman in the second photograph may well have come because it is her own town — those look like local photographs, and she might even be in one of them. I'd guess she's not really visiting a museum at all; she's checking something she half remembers.
Which would explain the kettle, I suppose. You don't put a kettle out for tourists.”
Why it works: the comparison is over in two sentences; the answer to the question is the body of the turn; and the whole thing is speculative — must be, I doubt, probably, may well have, might even, I'd guess. Photographs cannot be known, only guessed at, and modals of deduction are the grammar the task was built to collect. Say they are tourists and you have thrown the item away.
Phrases worth owning: comparing — “Both of these …, but …” · “Whereas in the second one …” · “The obvious difference is …, though what strikes me more is …”. Speculating — “He must be …” · “She can't be enjoying …” · “They may well have …” · “I'd imagine …”. Running out of ideas — “It's hard to say, but if I had to guess …”, which buys you four seconds and sounds like fluency rather than panic.
You are given a question and five written prompts. Talk to each other for about two minutes; then the examiner asks you to make a decision together and you have about one more. This is the part where candidates lose most marks, and almost never for language: they lose them for taking turns instead of having a conversation.
What is actually being marked here is Interactive Communication: whether you invite the other person in, react to what they said rather than to your own next idea, disagree without stopping the conversation, and build something together. A candidate who speaks beautifully for two minutes while their partner waits is marked down, not up.
Thirty seconds of a strong Part 3:
A: Shall we start with the buses? That's the one I keep coming back to.
B: Go on — why that one?
A: Because everything else on the list assumes you can get to the town centre. If you're sixteen and you live four miles out, a free sports pitch may as well be in another country.
B: That's a fair point, although — doesn't it depend on how big the town is? If everyone can walk in, the fares don't matter and I'd rather have the library open late.
A: True. So maybe it's not either-or; it's whichever one removes the reason people aren't coming. What would you say stops people, the distance or having nowhere to be?
B: Nowhere to be, honestly. Which is why I like the café with nothing to buy — though I can't see a council paying for it.
Look at what they are doing: A opens by inviting B in rather than by stating a view. B asks a question instead of taking a turn. B disagrees with a softener (that's a fair point, although) and gives a condition rather than a contradiction. A concedes (true) and then reframes, which is the move that lifts a Part 3 out of the middle band. Nobody has made a speech, and both have said something the other could not have said alone.
Phrases worth owning. Inviting: “What do you think — would that work where you live?” · “Go on, why that one?” Agreeing and adding: “Yes, and the other thing is …” Disagreeing safely: “I see what you mean, but doesn't that depend on …?” · “I'm not sure — I'd have said the opposite, actually.” Deciding: “Shall we say those two, then?” · “Can we agree on the buses and leave the second one open?”
The trap: being so polite that neither of you ever disagrees. Two minutes of “yes, I agree, and also…” gives the examiner nothing to mark. Disagreement is the material.
The examiner asks broader questions arising from Part 3, sometimes to one of you, sometimes to both. Answers here are longer than in Part 1 — thirty to forty-five seconds — and they want an opinion with reasons, not a description. If the examiner asks “What do you think?” after your partner has spoken, they are asking you to respond, not to start again.
What lifts Part 4: a position, a reason, an example, and one honest sentence against yourself. Four moves, forty seconds. Candidates who reach the top band here are almost never the ones with the widest vocabulary — they are the ones who organise forty seconds.
A strong Part 4 answer to “Why do you think so many public places now cost money to enter?”:
“Partly it's simply that they cost more to run than they used to — heating a big old building is not what it was five years ago. But I think the bigger reason is a change in how these places are described. Once somewhere is called an attraction rather than a service, charging for it sounds completely normal, and nobody has to argue for it.
I'd add that free entry isn't automatically fair, though. If a museum is free but it's two hours away and you're paying for the train, it isn't free to you — it's just free to people who already live nearby. So I'm in favour of free entry, but I don't think it settles the question on its own.”
Why it works: two causes, ordered and signposted (partly … but the bigger reason is); a concrete example that costs nothing to invent; and a self-correction in the second paragraph that shows the candidate can hold two ideas at once. Notice how much of it is ordinary vocabulary used precisely. Nothing here is a rare word.
Phrases worth owning: “Partly …, but I think the bigger reason is …” · “I'd add that …, though.” · “It depends what you mean by …” · “I'd have said the opposite five years ago, actually.” · handing over: “…that's roughly where I stand. What about you?”
Finally — the recording. Record all four parts, then listen back once for content and a second time with your eyes shut, listening only for the gaps. Where you stopped is more useful to you than what you said. Then do it again. The second recording is always better, and the exam is a second recording.
Do it twice. Once alone with the recorder, listening back for the four things the examiner listens for; then once with a partner, in real time, with a clock running. The partner version is the only one that rehearses Part 3, because Part 3 is a conversation and you cannot have one with yourself.
The four criteria are Grammar and Vocabulary · Discourse Management · Pronunciation · Interactive Communication. Three of the four are about how you talk, not what you know — which is why practising alone in your head does almost nothing for them.