Speaking lasts 11 to 14 minutes and is a conversation, not a performance. The examiner is listening for four things at once: whether you keep going, whether your vocabulary is precise, whether your grammar holds up under pressure, and whether you are easy to understand. Fixing hesitation and under-developed answers usually moves the band more than learning new words.
You will stop losing marks to answers that end too early, to the paragraph you memorised, and to the pauses you fill with nothing.
The format at a glance
Length
11 to 14 minutes in total
Part 1
Introduction and familiar topics, 4 to 5 minutes
Part 2
Cue card: 1 minute to prepare, 1 to 2 minutes to speak, then one or two follow-up questions
Part 3
Two-way discussion of abstract ideas, 4 to 5 minutes
Criteria
Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, Pronunciation
Versions
Identical for Academic and General Training
The traps, worked through
Three of these four traps are structural rather than linguistic. They happen before vocabulary or grammar are even in play, because language you never produced cannot be marked.
The Part 1 answer that stops at the answer
Part 1 questions sound like small talk, so candidates answer them like small talk. A two-word answer gives the examiner nothing to assess.
On screen
“Do you prefer to study in the morning or in the evening?”
Most candidates write
“In the morning.”
Earns the mark
“Definitely the morning. My head is clearest before about eleven, so I do the reading that needs real concentration then, and leave the routine work — emails, tidying up notes — for the evening when I am running on empty.”
Why: The second answer is the same opinion plus a reason, a contrast and a natural idiom, which is three criteria's worth of evidence instead of none.
Rehearsed answers arrive with a flatter delivery and a register that does not match the question, and they answer the topic rather than the task. Memorised language is not credited.
On screen
“Describe a place you like to visit. Say where it is, how often you go, what you do there, and explain why you like it.”
Most candidates write
A polished learned paragraph about a favourite city, opening “Today I am going to talk about a place which is very near and dear to my heart”, that never reaches the fourth bullet.
Earns the mark
“It is a stretch of canal towpath about ten minutes from my flat. I walk it most Sunday mornings, earlier if it is warm, because by ten it fills up with cyclists …”
Why: The second answer is unremarkable in content and full of assessable language — tenses, comparatives, a reason — while the first offers a rehearsed opening the examiner is trained to set aside.
The preparation minute gets spent writing sentences instead of keywords, so the talk runs out the moment the written lines are used up.
On screen
The one-minute preparation note for that cue card.
Most candidates write
Three full sentences written out and then read aloud, finishing in about forty seconds with nothing left to say.
Earns the mark
Six keywords, one per bullet plus two spare: towpath / Sunday, 10am / walk, herons / quiet, no signal / after exams / would miss it.
Why: Keywords force you to build sentences live, which is what fluency is measured on, and the spare keyword is what keeps you going until the examiner stops you.
Part 3 asks about people in general. A personal story is not wrong, but on its own it leaves the abstract question unanswered and skips the conditionals and comparatives that show range.
On screen
“Should governments fund public transport rather than road building?”
Most candidates write
“Yes. In my city the buses are terrible — they come once an hour and last week I waited forty minutes in the rain.”
Earns the mark
“On balance yes, though it probably depends on how dense the place is. In a compact city a bus lane moves far more people than an extra carriageway would, whereas somewhere spread out you might get more from the road. My own city is a good example of the first case …”
Why: The second answer takes a position, hedges it, compares two cases and uses the anecdote as evidence rather than as the whole answer.
Eleven to fourteen minutes across three parts, and only one of them has a stopwatch you can see. Knowing the shape of each part stops you rationing your language in the wrong place.
Part 1 · about 4–5 minutes
Roughly a dozen short questions, so two or three sentences each: answer, then a reason or an example, then stop. Answers that run past thirty seconds here cost you the questions you would have answered better.
Part 2 · the preparation minute
Sixty seconds and four bullets is about ten seconds a bullet in keywords, leaving twenty for one concrete detail — a name, a time, a number, a smell. Specifics are what you can keep talking about.
Part 2 · the long turn
Plan roughly thirty seconds per bullet and keep a spare keyword for the last thirty. Aim to be interrupted: stopping early is a fluency signal pointing the wrong way.
Part 2 · the follow-up
One or two short questions close the part. Answer in a sentence or two — this is a bridge into Part 3, not a second long turn.
Part 3 · about 4–5 minutes
Five or six abstract questions is roughly forty seconds each. Opinion, reason, example, then stop and let the examiner ask the next one.
Part 1: extend every answer
Answer in two or three sentences. Give the answer, then a reason or an example. One-word answers starve the examiner of language to assess.
Do not memorise scripts. Rehearsed paragraphs are audible, sound flat, and are explicitly discounted.
Prepare topic vocabulary rather than answers: home, work, study, food, weather, transport, technology, hobbies, holidays.
If you mishear, ask once: 'Sorry, could you repeat that?' Guessing the topic damages coherence far more than asking.
Buy a moment with a natural filler such as 'That's an interesting one' rather than falling silent.
Part 2: use the preparation minute properly
Write keywords against each bullet, not full sentences. You cannot read a script aloud naturally in two minutes.
Cover every bullet on the card. The examiner is following them and a skipped bullet breaks your structure.
Aim to still be talking when the examiner stops you. Finishing at 45 seconds is a fluency signal in the wrong direction.
Front-load the specifics: names, places, dates, what happened, how you felt. Specific detail is easier to keep talking about than generalities.
Watch your tenses. Most cue cards ask about a past experience, so past simple and past continuous need to be reliable.
If you finish the bullets early, add why it mattered to you or whether you would do it again.
The examiner closes Part 2 with one or two short questions on the same topic. Answer them in a sentence or two; they are a bridge into Part 3, not a new long turn.
Part 3: develop, compare and speculate
Use the shape opinion, reason, example. It prevents abstract questions from turning into rambling.
Answer about people in general, not only about yourself. Part 3 is deliberately broader than your own experience.
Show range with speculative and hedging language: 'I suppose', 'it tends to be', 'on balance', 'that would probably depend on'.
Narrow an enormous question aloud rather than freezing: 'That varies a lot, but for most working adults I would say...'
Disagreeing with the examiner is fine and often produces better language. There are no right opinions.
Compare, contrast and predict. Conditionals and comparatives are what distinguish a developed Part 3 answer.
Fluency habits that raise the band
Hesitation is only penalised when it is searching for language, not when it is thinking about content. Signal the difference: 'Let me think about that for a second.'
Self-correct once and move on. Repeatedly restarting a sentence costs more than the original error.
Speak at a natural conversational pace. Speed is not fluency, and rushing damages pronunciation.
Do not translate in your head. Say the simpler idea you can express accurately now.
Practise talking for two minutes on random everyday topics with a timer. Sustained speech is the specific skill Part 2 tests.
Pronunciation without an accent change
Your accent is not marked. Intelligibility, word stress, sentence stress and intonation are.
Word stress carries meaning: PHOtograph, phoTOGrapher, photoGRAPHic. Getting it wrong can make a known word unrecognisable.
Stress the content words in a sentence and reduce the function words. Flat, evenly stressed speech is hard to follow.
Pronounce word endings, especially plural -s and past -ed. Dropping them also reads as a grammar error.
Record yourself for two minutes and listen back once a week. Most learners hear their own repeated issues immediately.
Advice to ignore
Five things candidates are told about Speaking that are the wrong way round.
“Learn idioms and drop them in — they lift Lexical Resource.”
Idiom used out of place lowers it. “Last but not least” bolted onto an unordered list tells the examiner the phrase arrived before the meaning did.
“Speak as fast as you can so you sound fluent.”
Fluency is judged on continuity and coherence, not speed. Rushing damages pronunciation, which is a quarter of the mark, and buys nothing on the other three.
“Never admit you have not thought about something.”
“I have honestly never thought about that — I suppose it would depend on …” is fluent, coherent English. Thinking aloud is not penalised; silence is.
“You need to sound British or American.”
Accent is not marked. Pronunciation is assessed on how easily you are understood, together with word stress, sentence stress and intonation, all of which improve without changing your accent.
“Give the opinion the examiner will agree with.”
There are no correct opinions and the examiner's own views play no part in the marking. Take whichever position you can develop with more examples.
Common mistakes and how to fix them
Giving one-word answers in Part 1.
Fix: Add a reason or an example to every answer, every time.
Delivering a memorised paragraph regardless of the question.
Fix: Prepare vocabulary and ideas, not scripts. Examiners are trained to recognise memorised language.
Writing full sentences during the Part 2 preparation minute.
Fix: Write four or five keywords. You will speak more naturally and cover more.
Dropping in linkers such as 'last but not least' where they do not fit.
Fix: Use connectors only where the logic requires them. Misused idiom lowers Lexical Resource.
Answering Part 3 questions with a personal anecdote only.
Fix: Address the general issue first, then use your experience as the example.
Going silent while searching for a word.
Fix: Paraphrase around it. 'The thing you use to...' shows strategic competence, silence shows none.
Question type by question type
The single mistake that costs most marks on each task. Each guide then covers what the task looks like on screen, how it is marked, four strategies and the two traps specific to it.
Is the Speaking test the same for Academic and General Training?
Yes. The format, timing and marking criteria are identical.
Does my accent affect my score?
No. Pronunciation is marked on how easily you can be understood, along with stress and intonation, not on sounding like any particular variety of English.
What happens if I run out of things to say in Part 2?
Return to the last bullet and add why it mattered, how you felt, or whether you would repeat the experience. Keep going until the examiner stops you.
Can I ask the examiner to repeat a question?
Yes, in Parts 1 and 3. In Part 2 the cue card stays in front of you, so you can re-read it instead.
Do I need a formal register?
No. Speaking is a conversation. Natural, appropriate everyday English scores better than stiff, over-formal phrasing.
Put these tips under exam conditions with a full timed section.
Record a two-minute long turn from keywords only, then listen back once for a single thing: the moment you ran out of material. That moment is the bullet you under-prepared.
