IELTS Speaking: The Three Parts and Four Criteria

What happens, minute by minute

The structure is fixed, the same in the Academic and General Training versions, and the same whether you sat the written papers on paper or on a computer. Knowing the shape removes most of the anxiety, because nothing in the room will be a surprise.

The three parts of the Speaking test.

The examiner’s manner is deliberately neutral, and this unsettles people who read it as disapproval. It is not feedback. Examiners are trained to keep the conditions identical for every candidate, which means no encouragement, no reactions and no help — and it means nothing about how you are doing.

The four criteria, and what they actually reward

Speaking is marked on four published criteria, each worth a quarter of the score. As with Writing, the descriptors themselves are public and detailed; what follows is an explanation of what each rewards in practice.

• Fluency and Coherence

  Can you keep going, and does what you say hold together? Pausing to search for a word is normal and is not heavily penalised. Pausing repeatedly, restarting sentences and losing the thread is. Coherence is the other half: answers that develop rather than jump.

  Speed is not fluency. A measured speaker who never stalls scores well.

• Lexical Resource

  Range, precision and naturalness — including whether you can paraphrase when the exact word will not come. That last ability is worth more than a stock of rare vocabulary, because it is what keeps the answer moving.

  Not knowing a word is fine. Stopping because of it is not.

• Grammatical Range and Accuracy

  The same balance as in Writing: complex structures that stay correct. Speaking exclusively in short simple sentences caps the band, and so does attempting complexity that collapses.

  Part 3 is where complex grammar has somewhere natural to go.

• Pronunciation

  Assessed on how easily you can be understood, and on control of stress, rhythm and intonation. It is not assessed on sounding British or American, and having a first-language accent does not cost you marks.

  Word stress in the wrong place costs more than an accent ever does.

Nothing in that list is about content. There are no marks for a clever answer, a true answer or an answer the examiner agrees with — and no marks lost for inventing a detail to keep a story going. If you have never been to a wedding and the card asks about one, describe a wedding you have seen. You are being assessed on the English, and only on the English.

Part 2: using the minute properly

The long turn is the part that can be prepared for most directly, because the format never varies: a topic, three or four prompts to cover, one minute with paper and pencil, then one to two minutes of uninterrupted speech.

The minute is not for writing sentences. Sentences written down get read aloud, and reading aloud is audible, flat and penalised under Fluency. Write single words — five or six of them, one per prompt point — and let them cue you.

1. Seconds 0–10: choose your subject. Pick whatever gives you the most to say, not what is most true. Hesitating over which memory to use burns the whole minute.
2. Seconds 10–40: one word per prompt point. The card’s bullets are a ready-made structure. Use them in order and you cannot get lost.
3. Seconds 40–55: plan the ending. Decide now how you will finish — a reflection, a consequence, why it mattered. Trailing off is the commonest way Part 2 ends badly.
4. Seconds 55–60: fix the first sentence. Having the opening ready removes the stall that otherwise starts the long turn.

Filling the time without padding

Most candidates who run dry do so because they are reporting rather than narrating. Detail is what fills two minutes, and detail comes from the ordinary journalistic questions: when, where, who else was there, what it looked like, how you felt at the time, how you feel about it now. Adding tense variety at the same time is free marks under Grammatical Range.

• Set the scene in the past: It was about three years ago, when I was still living in my home town…
• Add a background action: I’d been meaning to go for months, so when a friend suggested it I said yes immediately.
• Reflect with a conditional: If I hadn’t gone that weekend, I’m not sure I ever would have.
• Close on the present: Looking back, it’s probably the reason I ended up studying what I study now.

If you finish early, the examiner will simply stop you or ask a follow-up — no marks are lost for a long turn that runs slightly short, and none are gained by talking past the point where you have anything left to say. Do not stop after thirty seconds, though: a genuinely short turn gives the examiner too little language to assess and pulls every criterion down.

Part 3: where the higher bands are decided

Part 3 moves from the personal to the general — from your experience to society, causes, consequences and comparisons. It is where a candidate who is comfortable at band 6 either climbs or does not, because abstract questions require exactly the structures that distinguish B2 from C1.

The most common failure is answering a general question personally. Asked whether public transport should be free, a weaker answer describes the candidate’s own bus journey. A stronger one generalises, considers a counter-argument, and hedges appropriately — and hedging is a grammatical skill as much as a rhetorical one.

• Generalise: In most cities I’m familiar with, people tend to use the car simply because the alternative is unreliable.
• Concede, then counter: It’s often argued that it would be too expensive, although the cost of congestion is rarely counted on the other side.
• Hedge a claim: It may well be that the effect is smaller than people assume, particularly in smaller towns.
• Speculate: Had that policy been introduced ten years earlier, the outcome might have been very different.
• Compare across time: A generation ago that would have been unthinkable, whereas now it barely gets commented on.

Those five moves — generalising, conceding, hedging, speculating and comparing — cover most Part 3 questions between them, and each one naturally carries complex grammar with it. Practising the moves is far more productive than practising topics, because the topics are unpredictable and the moves are not.

Advice that lowers scores

Speaking attracts more bad advice than any other paper, largely because it is the least transparent. Four pieces of it are actively harmful.

• “Memorise answers for common topics.” Examiners hear memorised answers constantly and are trained to identify them. Rehearsed language is discounted, the delivery gives it away, and a follow-up question outside the script exposes it immediately.
• “Use as many idioms and advanced words as possible.” Lexical Resource rewards precision and naturalness, not rarity. A misplaced idiom is more damaging than a plain word used correctly, because it signals language that has been collected rather than acquired.
• “Speak as fast as you can to sound fluent.” Fluency is continuity and coherence, not speed. Rushing degrades pronunciation and grammatical accuracy at once, so it costs marks under three criteria to gain them under none.
• “You must lose your accent.” Pronunciation is assessed on intelligibility and on stress, rhythm and intonation. A first-language accent is not penalised. Stress placed on the wrong syllable is — which is why individual word stress is worth far more study than accent reduction.

One further point, since it is the fear behind most of the others: the examiner is not trying to catch you out. They are following a script, timing the parts and applying published criteria, and they have no interest in whether your opinions are unusual or your story is true.

Practising speaking alone

Speaking is the hardest paper to prepare for without a partner, and the one most improved by a small amount of the right practice.

• Record yourself and listen back

  Uncomfortable and unusually effective. You will hear your own hesitations, repeated fillers and the exact points where a sentence collapses — none of which you notice while speaking.

  One two-minute recording a day beats an hour of silent worry.

• Answer Part 3 questions aloud, not in your head

  Thinking an answer is not producing one. The gap between what you can plan and what you can say under time pressure is the gap the test measures.

  Set a timer. Speak until it stops.

• Drill word stress, not accent

  Mis-stressed words are the most common cause of genuine misunderstanding. Learning the stress of your topic vocabulary is far more valuable than trying to change your accent.

  Vocabulary lists with audio

• Build the grammar Part 3 needs

  Hedging, speculation and conditional reasoning are grammar before they are style. They are also exactly what separates a band 6 answer from a band 7 one.

  Modals of deduction · Third conditional · Wishes and regrets