IELTS Speaking Band Descriptors Explained

The four criteria and equal weighting

IELTS Speaking uses four assessment criteria: Fluency and Coherence, Lexical Resource, Grammatical Range and Accuracy, and Pronunciation. IELTS states that the four criteria are equally weighted. That immediately kills one popular study plan: collecting spectacular vocabulary while hoping it will compensate for answers that are hard to follow or hard to understand.

The official IELTS Speaking Key Assessment Criteria explains the features inside each criterion. The official Speaking Band Descriptors shows how those features change across bands. IELTS also notes on the descriptor sheet that candidates are rated on their average performance across all parts of the test. So a single beautiful answer—or a single ugly one—is not a trustworthy summary of your Speaking performance.

One short answer, four different lenses

Here is one original IELTS-style Part 1 answer. It is deliberately ordinary, because the rubric is about language control, not having an extraordinary life.

Question: Do you enjoy cooking?

Answer: “Yes, especially at weekends. During the week I usually make something quick because I get home quite late, but on Saturday I like trying one new recipe. Last weekend I made a spicy noodle soup, and it took longer than I expected, but I’d definitely make it again because it was much better than takeaway.”

Do not assign this text a band. A transcript cannot show its actual pauses, stress, rhythm, intonation or sound clarity, and one short answer cannot represent average performance across the test. Instead, annotate it four ways:

That is the central habit for this page: collect evidence for each criterion separately before you decide what to practise.

Fluency and coherence

Fluency is not “talk as fast as possible,” and coherence is not “say firstly, secondly, moreover every twelve seconds.” The official criteria describe fluency in terms of continuity, rate and effort, while coherence includes logical sequencing, relevance and the way spoken ideas are linked.

Two hesitations that look similar but mean different things

Compare these original versions of the same thought:

• Language-search hesitation: “I prefer small towns because they are… um… more… what’s the word… not stressful… less stressful, and the transport is… um…”
• Content-planning hesitation: “I prefer small towns because they feel less stressful. [brief pause] The main reason, I think, is that I can do most things without a long commute.”

In the first, the speaker repeatedly stalls inside the clause while searching for accessible language. In the second, the speaker completes one meaning unit, pauses to decide what to say next, then continues. The IELTS descriptors explicitly distinguish hesitation used to prepare content from hesitation caused by searching for words or grammar at the higher bands. That does not mean every content pause is “good” or every word search is “bad.” Frequency, disruption and the overall performance matter.

A useful self-review therefore marks where a pause happens. Pauses between ideas can help listeners. Repeated mid-phrase searching, false starts and backtracking may reveal retrieval or sentence-building trouble. If this is your main bottleneck, the broader How to Speak English Fluently hub gives you fluency-specific practice beyond IELTS.

Coherence is the route, not the number of linking words

Suppose the question is, “Why do some people prefer working from home?” A coherent answer might make one point, explain the reason, give a consequence, then stop. A less coherent answer might mention commuting, children, pollution, housing prices and technology without developing any one line. More ideas do not automatically create more coherence.

If you often know several ideas but cannot decide what comes first, use Organise Ideas While Speaking English. For IELTS-specific answer mechanics, practise the fast familiar responses in IELTS Speaking Part 1 Questions and Natural Answers, build a route during the preparation minute in IELTS Speaking Part 2, and develop one line of reasoning in IELTS Speaking Part 3.

A practical fluency-and-coherence check

• Can you identify the answer’s main point in one sentence?
• Do pauses mostly fall between meaning groups, or do they repeatedly split short phrases?
• Do restarts help you repair the message, or create a second unfinished version?
• Does each new detail develop the question, or simply add another topic?
• Can you answer a changed follow-up without returning to a memorised opening?

Lexical resource

Lexical Resource is not a scavenger hunt for the rarest synonym. The official criteria focus on range and precision, appropriacy, collocation, style and paraphrase. That is why a simple exact phrase can be better evidence of control than an idiom dragged into the answer by force.

Precise everyday vocabulary beats decorative vocabulary

Imagine the question is, “What do you like about your neighbourhood?”

• Precise everyday wording: “It’s quiet at night, the supermarket is within walking distance, and there’s a small park where my daughter can ride her bike.”
• Forced idiom version: “My neighbourhood is a blessing in disguise and the park is the best thing since sliced bread.”

The first answer gives exact, relevant meaning. The second uses idioms, but they do not fit naturally: a “blessing in disguise” normally refers to something that first seemed bad, and “the best thing since sliced bread” is exaggerated and stylistically odd here. The problem is not that idioms are forbidden. The official descriptors do mention idiomatic language at higher levels. The problem is treating idioms as collectible score tokens rather than language that must be accurate, appropriate and naturally placed.

Paraphrase is a recovery skill

A missing word should not end the answer. If “colander” disappears, you can say “the bowl with holes that you use to drain pasta.” If “commute” disappears, you can say “the journey between home and work.” That ability to get around a gap is explicitly part of the official lexical criteria.

Train paraphrase under speaking pressure, not only as a vocabulary worksheet. Pick five ordinary objects or ideas, hide the target words, and explain each one in one sentence. Then put the word back and continue the answer. You are training continuity and lexical flexibility at the same time.

What to log after a recording

• One vague word you repeated too often: good, bad, nice, thing, stuff, important.
• One more precise phrase you already knew but did not retrieve quickly.
• One collocation that sounded translated or unnatural.
• One moment where you successfully paraphrased instead of freezing.
• One “advanced” expression you should remove because it did not fit the meaning or style.

Grammatical range and accuracy

This criterion contains a built-in tension: you need enough range to express relationships and complex propositions, but range that collapses into constant errors is not useful control. The official key criteria explicitly include error density and the communicative effect of errors alongside structural range.

Range means grammar doing useful work

Consider the question, “Should cities invest more in public transport?”

Useful range: “Yes, especially in areas where traffic is getting worse. If buses were more frequent, some people who currently drive might leave their cars at home, which could reduce congestion during rush hour.”

The answer uses a relative clause, a conditional relationship and another relative clause because the meaning needs them. Nothing is there just to prove the speaker has studied conditionals.

High error density can erase the benefit of complexity

Now imagine a speaker tries to make every sentence complex:

“If the government will invest, people would using buses which is reduce traffic because cars was too many.”

There is ambition here, but several errors arrive in a small amount of speech and the relationships become harder to recover. A better practice repair is not “avoid complex grammar forever.” It is to rebuild one relationship accurately, then vary it:

• “If buses were more frequent, more people might use them.”
• “If trains were cheaper, some commuters might stop driving.”
• “If the network improved, the city could reduce congestion.”

That turns one structure into usable spoken grammar. For a broader system built around speaking jobs rather than rule collecting, use Grammar for Speaking English.

Review patterns, not isolated slips

One self-correction is not the same as a repeated error pattern. After a recording, choose a short extract and mark only errors that repeat or change the message. If articles are occasionally missing but meaning remains clear, that may be a lower-priority repair than unstable tense that makes the timeline confusing. Your aim is a smaller error pattern you can actually fix—not a red-pen massacre of every spoken sentence.

Pronunciation

Pronunciation is where candidates most often import the wrong goal: “I need a British accent.” The official IELTS criteria do not make accent identity itself the target. They look at phonological features and, crucially, the effect of accent on intelligibility—how successfully the listener can understand the message.

Accent and intelligibility are not the same thing

You can have a clearly identifiable Iranian, Spanish, Korean, Polish or Brazilian accent and still be easy to understand. Accent becomes relevant to the IELTS pronunciation criterion when its features affect intelligibility. The band descriptors themselves move toward language such as accent having less effect on intelligibility at higher levels. That is very different from “sound native.”

A useful question is therefore: Where did the listener need extra effort? Was a key vowel or consonant unclear? Did the word stress make a familiar word difficult to recognise? Did every word receive equal stress so the sentence focus disappeared?

Prosodic control matters beyond individual sounds

The official criteria include chunking, rhythm, stress timing, connected speech, stress and intonation. These are prosodic features—the patterns across phrases and sentences, not just single consonants and vowels.

Take this sentence: “I thought the new schedule was actually easier.” If every word is delivered separately—“I / thought / the / new / schedule / was / actually / easier”—the listener must assemble the message piece by piece. A more useful grouping might be “I thought the new schedule / was actually easier,” with prominence on new or easier depending on what you are contrasting.

That does not mean there is one approved melody for every sentence. The goal is controlled pronunciation that supports meaning and reduces listener effort. For targeted work on intelligibility, stress, rhythm and connected speech, use the English Pronunciation hub or go straight to English pronunciation practice.

What changes between adjacent bands

The most useful way to read adjacent descriptors is not “How many idioms do I need for the next band?” It is: what becomes more reliable, flexible, precise or easy to understand? The official scale is holistic within each criterion, and some bands are described partly in relation to the positive features of neighbouring bands. Do not turn the table below into a checklist where ticking three boxes magically creates a score.

Example: the difference between “can do it sometimes” and “can sustain it”

Adjacent-band movement is often about reliability. A speaker may produce one elegant complex sentence, one beautifully stressed phrase or one successful paraphrase. The next question is whether that control survives the rest of the response, unfamiliar topics and spontaneous follow-ups.

For fluency, this can mean a shift from frequent language-search interruptions toward longer stretches where pauses are mainly about planning content. For pronunciation, it can mean moving from occasional effective stress and intonation toward using those features more consistently across longer utterances. For grammar, it can mean complex forms moving from risky experiments to structures that are used more flexibly with a higher proportion of accurate sentences.

Diagnose the limiting criterion

“My speaking is bad” is useless data. “I lose the sentence whenever I try to explain a condition” is a practice target.

Here, limiting criterion means the criterion that most consistently limits your ability to communicate in your current practice recordings. It is not a claim that you have discovered your official lowest criterion score. Because IELTS evaluates average performance across the test, diagnose from a small set of comparable recordings rather than one answer.

Use symptom → evidence → repair

Do not let one recording bully your study plan

Record at least a small spread: several Part 1 answers, one Part 2 long turn and several Part 3 answers on a connected theme. Look for a symptom that appears more than once. If the same failure survives different topics, you have stronger evidence that it deserves priority.

Then change one thing and repeat. If you repair vocabulary but the answer is still incoherent, vocabulary was not the main bottleneck. If you simplify one grammar pattern and fluency immediately improves, grammar processing may have been overloading delivery. Diagnosis should lead to an experiment, not a personality judgment.

Build a criterion-specific practice week

You do not need four separate personalities for the four criteria. You need one week in which each criterion gets a deliberate turn, followed by mixed speaking where the skill must survive real-time pressure.

Keep each drill narrow enough to learn from

If you try to fix pausing, vocabulary, grammar, pronunciation and idea quality in the same 60-second answer, you will not know what caused the change. Narrow practice first; mixed practice later. That is why the week separates the criteria before bringing them back together.

Use the live general speaking practice when you want a fresh speaking rep. If intelligibility is the main problem, use English pronunciation practice. The dedicated IELTS recording route is /practice/ielts-speaking/.

What automated feedback can and cannot assess

Automated tools can be useful because they can make a recording less slippery: they can transcribe it, count repeated words, show pauses, flag possible grammar patterns, or visualise acoustic features. But useful measurement is not the same thing as IELTS examiner judgment.

Any automated result should be treated only as practice feedback. It is not an IELTS band, official score, examiner judgment or examiner-equivalent assessment.

Use automation to generate questions, not verdicts

A productive workflow sounds like this: “The tool marked four long pauses inside clauses. Were those language searches? Let me listen.” Or: “The transcript shows I used important six times. Which two places need a more precise phrase?” Or: “The pitch trace is flatter than my model. Does that actually make the contrast harder to hear?”

An unproductive workflow sounds like: “The app says 7.0, therefore I am 7.0.” That claim skips examiner training, test-wide evidence, criterion interpretation and the limits of automated speech processing.

The safest practice loop

1. Record a test-like answer without stopping.
2. Use automated feedback, a transcript or a listener to surface one observable issue.
3. Check the issue against the relevant official criterion, not against a made-up universal rule.
4. Repair one thing.
5. Re-record with the same communicative goal but fresh wording.
6. Compare the two performances and keep the evidence—not a fake band.

If you want to go back to the test parts after diagnosing the rubric, use the Part 1 guide, the Part 2 preparation guide, or the Part 3 development guide. The rubric should tell you what to listen for; your practice should tell you what to repair next.