IELTS Speaking Part 3: band 7 is a shape, not a vocabulary list
Aman Batth · 10 min read · · Updated
Part 3 is where band 7 is actually decided, and the thing that decides it is not the words you use. A band 6 answer states an opinion. A band 7 answer states it, gives a reason, and illustrates or qualifies it. That is a learnable shape.
Part 3 is the four to five minutes after the examiner takes your cue card back. Same examiner, same four criteria, and a completely different demand — which is why so many candidates come out of Parts 1 and 2 feeling fine and still land on 6.5.
Most Part 3 advice is a list of topics: education, technology, the environment, work. Those lists are accurate and they will not move your band, because the topic was never the problem. You are not being marked on what you know about education.
You are being marked on what you do with an opinion once you have stated one. That is the whole of Part 3, and it has a shape.
This is written from the marking side — we build the engine that scores speaking on Hilingo. For the criteria themselves, the source is IELTS, published by the test owners.
Why Part 3 is a different test from Parts 1 and 2
The four criteria do not change across the three parts: fluency and coherence, lexical resource, grammatical range and accuracy, pronunciation, weighted equally. What changes is what the questions force you to produce.
Parts 1 and 2 are, for most candidates, a production problem. Part 1 punishes answers that stop too early; Part 2 punishes running dry before the two minutes are up. In both, the fix is mostly more connected speech.
Part 3 is not that. The questions are abstract on purpose — they ask you to compare, to speculate, to explain a cause, to concede a point and then hold your ground. Those are not vocabulary demands. They are grammatical and organisational demands, and they load the two criteria that most candidates have never worked on directly.
If your speaking band has been stuck at 6.5 across several attempts, this part is the most likely reason, and it is the most fixable one.
The three-move answer
Here is the difference, on one question.
Question: Do you think governments should pay for higher education?
A band 6 answer:
Yes, I think the government should pay for it. Education is very important for a country. So I agree with this idea.
Nothing there is wrong. It is on topic, it is grammatical, it is clear. It also stops after the claim, repeats itself in different words, and hands the examiner nothing to assess.
A band 7 answer:
I'd say yes, mainly because the cost is what decides who applies, rather than the ability. In my city, plenty of students choose a shorter course simply because the fees for a longer one are out of reach — the talent is there, it's the funding that isn't. That said, I'd probably means-test it rather than make it free for everybody.
Three moves:
- Claim. "I'd say yes."
- Reason. "…because the cost is what decides who applies, rather than the ability."
- Illustration or qualification. The example from the speaker's city, then the concession about means-testing.
Now look at what is not in the second answer. No rare vocabulary. No idiom. The words are ordinary: cost, ability, fees, funding, talent. What arrived with the second and third moves were structures — a cleft ("the cost is what decides"), a contrast built across two clauses ("the talent is there, it's the funding that isn't"), a hedged modal ("I'd probably"), a concessive marker ("that said").
That is the honest mechanism. The band went up because developing the answer forced grammar that a one-sentence answer has no room for.
Being asked "why?" is not the examiner disagreeing with you
This derails more candidates than any vocabulary gap.
You give an opinion, the examiner says "why do you think that?", and it reads as a challenge — so you soften, reverse, or start agreeing with a position you do not hold. The answer gets shorter and less coherent exactly when it needed to get longer.
The examiner is not arguing. They are asking for moves two and three because you only gave move one. The follow-up is the examiner doing your job for you — and it costs you, because an answer you had to be prompted into is not the same as an answer you developed.
A useful diagnostic: if you are asked "why" after most of your Part 3 answers, that is not bad luck with the examiner. It is evidence that your answers are stopping at the claim.
What each question type is really asking for
Part 3 questions come in a small number of shapes, and each one forces a specific piece of grammar. This is the part worth rehearsing, because the structures are finite and the topics are not.
| Question shape | Typical stem | What it forces | The structure that carries it |
|---|---|---|---|
| Opinion | "Do you think…?" | A claim plus a reason | because / the main reason is that |
| Change over time | "How has X changed?" | Past set against present | used to / whereas now / tends to |
| Speculation | "What will X be like in future?" | A hypothetical | is likely to / if X happened, Y would |
| Cause | "Why do you think X happens?" | A causal chain | largely because / which in turn / leads to |
| Two-sided | "Some people say X. What do you think?" | Concession, then position | while it's true that… , I'd still argue |
| Comparison | "Which is more important, X or Y?" | A ranked comparison | far more / not nearly as / the bigger factor is |
Two things fall out of that table.
First, a candidate who answers every abstract question in the present simple has capped grammatical range, whatever their accuracy. Part 3 is the only part of the test that naturally asks for conditionals, perfect aspect and modals of speculation. If those never appear, half of that criterion — the range half — has nothing to score.
Second, the "two-sided" row is the one that rescues you when you have no opinion. You are allowed not to care about urban planning. You are not allowed to say nothing. Concede the other side, then take a position anyway: that produces a complete answer out of a subject you have never thought about.
Where the marks actually move in Part 3
Criterion by criterion, because "speak more fluently" is not an instruction anyone can act on.
Fluency and coherence — development, not speed. This criterion is about keeping going coherently, not about talking fast. A pause while you think is far cheaper than a full stop after one sentence. Candidates rush Part 3 because the questions feel hard, and rushing produces restarts, abandoned clauses and self-corrections — which cost more than the silence they were trying to avoid.
Grammatical range and accuracy — range is the half you can plan. Accuracy improves slowly. Range is a decision you can make before you walk in: this part will contain a conditional, a contrast between past and present, and at least one modal of speculation, because the questions are built to invite them.
Lexical resource — precision beats rarity. An unfamiliar word used slightly wrong damages this criterion more than a plain word used correctly. "The government should regulate it" is worth more than an idiom deployed at the wrong angle. Reach for the precise ordinary word, not the impressive one.
Pronunciation — stress carries the argument. Accent is not assessed; intelligibility is. In Part 3 specifically, contrast is heard through stress: in "the talent is there, it's the funding that isn't", the stress is what makes the concession audible as a concession. Flat delivery makes a well-built argument land as a list.
Four things that quietly cost marks here
The memorised marker stack. "First and foremost", "last but not least", "at the end of the day" — pre-learned phrases are recognisable as pre-learned, and they signpost rather than connect. Coherence is whether your ideas follow from each other. A marker notices when the linking is mechanical.
Answering Part 3 as though it were Part 1. Two or three sentences about yourself is the right size for Part 1 and too small here. Part 3 asks about people in general, not about you — a personal example is move three, not the whole answer.
Generalising with no example. "In my country, people are very traditional" is a claim with no reason and no illustration attached. It is move one wearing a hat.
Reversing without signalling it. Changing your position mid-answer is fine and quite natural — "actually, thinking about it, the opposite is probably closer to the truth" is a coherent sentence. An unsignalled reversal, where the second half of your answer contradicts the first with no marker, reads as incoherence.
How to practise this so the band actually moves
Not "read the news daily". That is a six-month intervention and you probably do not have six months.
- Record yourself answering six Part 3 questions, on the real clock, with no notes.
- Transcribe them — or use anything that gives you back the words you actually said, rather than the ones you meant.
- Label every sentence C, R or I: claim, reason, illustration or qualification.
- Count the answers that are C only. That number is your Part 3 problem, expressed as a number.
- Redo only those answers, adding a reason and an example. Do not re-record the ones that already had all three moves.
This works because the thing you are fixing is countable. "Be more fluent" is not. "Four of my six answers stopped after the claim" is.
See your Part 3 answers written down
The hard part of step two is that you cannot hear your own answer accurately. You know what you meant, so it sounds developed to you.
Hilingo gives you one full scored IELTS mock free, with no card. Speaking runs across Parts 1 to 3 on the real clock, and each answer comes back with the transcript of what you actually said next to it, banded on all four criteria separately rather than as one number — so you can see whether Part 3 cost you fluency and coherence or grammatical range, which are different problems with different fixes. Writing and speaking bands are labelled as estimates, because they are produced by a model rather than a certified examiner, and we would rather say so than sell you a number.
- IELTS Speaking Part 1 — why short answers cost more than they look
- IELTS Speaking Part 2 — the long turn, and what to do when you run dry
- The IELTS mistakes you cannot see yourself — why one criterion is usually a band below the rest
- IELTS band score chart — band to CLB per skill
- The last week before your IELTS — what moves your band late
Frequently asked questions
How long is IELTS Speaking Part 3?
Four to five minutes, immediately after the Part 2 long turn, with the same examiner. It is the longest stretch of continuous discussion in the test, which is why it carries so much of the evidence the examiner is working from.
What is the difference between a band 6 and a band 7 answer in Part 3?
Usually development rather than vocabulary. A band 6 answer states an opinion and stops. A band 7 answer states it, gives a reason, and then illustrates or qualifies it. The extra moves are what create room for the conditionals, contrasts and modals that the grammatical range criterion is looking for.
Are IELTS Speaking Part 3 questions related to the cue card?
Yes. Part 3 develops the Part 2 topic into general and abstract territory — if your cue card was about a journey you enjoyed, Part 3 might ask why people travel, or how tourism has changed. The link is in the theme, not in the specifics of your own story.
Does the examiner disagree with me when they ask "why do you think that?"
No. Examiners assess how you express an opinion, not whether they share it. A "why" follow-up almost always means your answer stopped at the claim, so treat it as a prompt to add the reason and the example you had not yet given.
Do I need advanced vocabulary for IELTS Speaking Part 3?
No, and reaching for it is risky. An unfamiliar word used slightly wrong lowers lexical resource more than a plain word used correctly. Part 3 rewards precision and range within words you genuinely control, plus the grammatical structures that developing an argument requires.
What if I have no opinion on a Part 3 question?
Say what each side would argue, then take a position anyway — "while it's true that…, I'd still argue…". You are not being marked on holding sincere views about urban planning. You are being marked on producing a developed, coherent answer, and a two-sided structure gets you one from a topic you have never considered.
Can I ask the examiner to repeat a Part 3 question?
Yes, and asking naturally is far better than answering a question you misheard. A drifting, off-topic answer costs you coherence; one clear request for repetition costs you nothing.