Part 1 — Introduction and interview (4–5 minutes)
Answer each question aloud in 2–4 natural sentences. Press 🔊 to hear the examiner ask it. Don't script — react.
Part 2 — Individual long turn (3–4 minutes)
You have one minute to prepare (make notes!) and should then speak for one to two minutes. Use the timers — the discipline is the practice.
Describe a conversation that changed the way you think about something.
You should say:
- who you had this conversation with
- where and when it took place
- what the conversation was about
- what you believed before it happened
and explain how this conversation changed your thinking, and whether that change has lasted.
You will have one minute to prepare and should speak for one to two minutes.
Part 3 — Two-way discussion (4–5 minutes)
These get harder as they go. Build answers with: Answer → Explain → Example → Alternative view → Evaluation. Open the follow-ups only after you've answered — that's what the examiner asks when an answer is thin.
Examiner follow-ups
- Can you give an example of the kind of conversation that tends to stick?
- Is it the ideas that make them memorable, or the person and the moment?
Examiner follow-ups
- What other reasons might there be?
- Would that be true for important beliefs as well as small preferences?
Examiner follow-ups
- Why do you think that happens?
- Is it worse in online discussions than face to face?
Examiner follow-ups
- What could be the consequences of that difference?
- Does anonymity make this better or worse?
Examiner follow-ups
- Can you give an example?
- Is that influence stronger when we're younger?
Examiner follow-ups
- What would that actually look like in a classroom?
- Who else, apart from schools, has a responsibility here?
Examiner follow-ups
- What role does technology play in that?
- How might that differ between generations?
Examiner follow-ups
- What might make societies more tolerant of people changing their positions?
- Can you think of a situation where refusing to change your mind is actually the right thing?
Examiner follow-ups
- Who, if anyone, should set limits on persuasion — say, in advertising or politics?
- Would that be true in every situation?
Examiner follow-ups
- Would that change be mostly positive or negative, in your view?
- How could people protect their own judgement?
What a Band 8–9 performance should demonstrate
Judge your recording against the four official criteria. An accent is not required — clarity is.
Fluency & Coherence
- Speaks at length about a single conversation without drying up — the Part 2 story flows from context to change of mind without long unnatural pauses.
- Uses discourse markers naturally ("funnily enough", "looking back", "to be fair") rather than mechanical ones ("firstly", "secondly") in a personal story.
- Self-corrects without losing the thread — a brief "or rather…" and straight back into the idea.
- In Part 3, extends answers beyond one sentence unprompted — the examiner should rarely need the follow-up prompts.
- Nothing sounds recited: answers respond to the actual question about opinions and persuasion, not a rehearsed speech about 'communication'.
Lexical Resource
- Precise vocabulary for the topic: persuade, reconsider, assumption, perspective, echo chamber, come round to, talk someone into/out of.
- Natural collocations — "changed my mind completely", "see it from her point of view", "a heated discussion" — not invented pairings.
- Paraphrases flexibly when a word is missing instead of stopping to hunt for it.
- Idiomatic language used sparingly and accurately ("it stopped me in my tracks") — one natural idiom beats five forced ones.
- Can move from the personal story to abstract Part 3 language (influence, bias, accountability, discourse) without the register wobbling.
Grammatical Range & Accuracy
- Past narrative under control in Part 2: past simple, past continuous and past perfect woven together ("I had always assumed… until she pointed out…").
- Hypotheticals and conditionals in Part 3 ("If people only ever heard opinions they agreed with, they would…").
- Concessive structures for balanced answers: "Although evidence matters, most of us…", "Even so…".
- Relative clauses that add density without derailing: "the friend I mentioned, who I'd known since school, …".
- Errors are rare and self-corrected quickly; accuracy never comes at the price of fluency.
Pronunciation
- Clearly intelligible throughout — an accent is completely fine; swallowed word-endings are not.
- Sentence stress lands on the contrast words this test invites: "I used to think X — now I think Y."
- Intonation separates the story (lively, varied) from analysis (measured) in a natural way.
- Connected speech sounds natural ("changed_my_mind", "sort_of") rather than word-by-word delivery.
- Multi-syllable topic words — persuasion, deliberately, controversial — are stressed on the right syllable.
How to develop a Band 9 answer for this test
- Opening / context — one or two sentences placing the conversation: who, where, when, and why you remember it.
- Main description — what was said, in enough detail that the listener can follow the actual disagreement or discovery.
- Specific moment — the exact sentence or point that landed; a line of reported speech ("she asked me why I'd never actually tried it") lifts the story.
- Personal reaction — what you felt and thought at the time, including resistance if there was any.
- Reflection / significance — what you believed before versus after, and how the change showed up in what you later did.
- Concluding sentence — one line on whether the change lasted, answering the cue card's final question explicitly.
“Do you think people generally change their opinions because of arguments and evidence, or for other reasons?”
- Answer: take a clear position — e.g. evidence matters less than we like to admit.
- Explain: give the mechanism — opinions are tangled up with identity and belonging, so facts that threaten the group are resisted.
- Example: one concrete illustration — people rarely change political views after losing a debate, but often drift after moving city or changing jobs.
- Alternative view: concede the other side genuinely — in technical questions with low emotional stakes, evidence does change minds quickly.
- Evaluate: land the judgement — so persuasion through relationships and experience tends to beat persuasion through argument, which is uncomfortable but useful to know.
Common mistakes for this specific test
The cue card asks for a conversation. Candidates drift into a portrait of the friend or relative and never say what was actually discussed — which means the deeper part of the card (how it changed your thinking) never gets answered.
The fourth bullet — what you believed before — is what makes the change meaningful. Without it, "it changed my thinking" is an empty claim the examiner can't credit.
"Do you prefer maps or asking directions?" answered with "Maps." forces the examiner to drag reasons out of you. Two to four sentences with a because is the natural length.
Several Part 3 questions touch on phones and online discussion, and memorised technology essays surface fast. The questions here are about opinions and persuasion — a generic "technology has advantages and disadvantages" answer is off-target.
"My friend once…" works in Part 2, but Part 3 asks about people, generations and societies. Band 8–9 answers generalise — 'many people', 'in most workplaces' — and use a personal case only as illustration.
The topic invites it. Vary the language: reconsider, come round to, shift my position, see it differently, be talked out of it.
"Umm, you know, like…" before every Part 3 answer costs fluency marks. Buy time in English instead: "That's not something I'd thought about, but I'd say…".
Questions like "easier or harder to hear different views?" need a stance. "It depends" with no follow-through is a Band 6 pattern; "It depends — mainly on X, and here's why" is fine.
The abstract later questions push nervous candidates into a sprint, and pronunciation features collapse first. Slowing down slightly reads as thoughtfulness, not weakness.
The final question is about human judgement, not machine specs. Stacking half-understood jargon ("neural networks, algorithms…") hurts Lexical Resource more than it helps.