IELTS pronunciation: how it's scored, band by band

A learner messages me, stuck at speaking 6.5 for the third sitting. Fluent enough, decent vocabulary, grammar fine on paper. The score sheet keeps marking pronunciation at 6, and every video they've watched gives the same useless note: "work on your accent." That advice has it backwards. The official band 8 descriptor says accent has "minimal effect on intelligibility." You can keep a heavy first-language accent and still score 8.

Pronunciation is the one speaking criterion with no writing equivalent, and the one most candidates misread. It isn't an accent test. Here is what it actually measures, in the examiner's own words, and what moves you from band 6 to band 8.

What pronunciation measures, and what it doesn't

Pronunciation is one of four criteria in IELTS speaking, each worth a flat 25%. The examiner scores all four, averages them, and rounds to the nearest half band, so the same averaging that builds your overall score runs inside speaking too. One weak axis drags the rest toward it.

The official name for what's assessed is "phonological features": stress, rhythm, intonation, the individual sounds, and how you join words together. The yardstick for all of it is one thing only — can the listener follow you without working too hard? The public Band Descriptors make this explicit. At band 6 you "can generally be understood throughout without much effort." At band 8 you "can be easily understood throughout," and "accent has minimal effect on intelligibility." Band 9 raises it to "accent has no effect on intelligibility."

Read those lines again. Not one of them mentions sounding British, American, or native. The test is intelligibility. A strong Japanese, Indian, or Vietnamese accent can sit comfortably at band 8, as long as the sounds, stress, and rhythm don't make the examiner stop and decode.

The band 6 wall, in the examiner's words

Most people reading this are stuck at 6 on this criterion, so start there. The official band 6 pronunciation descriptor reads:

Uses a range of phonological features, but control is variable. Chunking is generally appropriate, but rhythm may be affected by a lack of stress-timing and/or a rapid speech rate. Some effective use of intonation and stress, but this is not sustained. Individual words or phonemes may be mispronounced but this causes only occasional lack of clarity. Can generally be understood throughout without much effort.

That paragraph is a diagnosis. Take it apart line by line, because every phrase is a fault you can hear in your own recording:

• "Control is variable": you can produce the features, but not reliably. Good sentence, then a flat one.
• "Rhythm may be affected by a lack of stress-timing": English leans on stressed syllables at roughly even intervals and crushes the rest. Give every syllable equal weight and the rhythm goes machine-gun.
• "Some effective use of intonation and stress, but this is not sustained": your pitch moves on the first few sentences, then settles into a monotone when you get tired or nervous.
• "Individual words or phonemes may be mispronounced": a handful of sounds slip and the listener loses a word here and there.

Notice the repeated idea: the band-6 speaker does the right things sometimes. The control comes and goes.

What band 8 adds, and why band 7 has no checklist

Here is the band 8 descriptor, the thing band 6 is missing:

Uses a wide range of phonological features to convey precise and/or subtle meaning. Can sustain appropriate rhythm. Flexible use of stress and intonation across long utterances, despite occasional lapses. Can be easily understood throughout. Accent has minimal effect on intelligibility.

Put the two side by side and the lever is obvious. Band 6 says intonation and stress are "not sustained." Band 8 says rhythm and intonation are "sustained" "across long utterances." Same features. The difference is consistency over a two-minute answer, not a new bag of tricks.

So what about band 7? It has no feature list of its own. The descriptor defines it entirely by reference: "Displays all the positive features of band 6, and some, but not all, of the positive features of band 8." The detailed lists sit at bands 4, 6, and 8; bands 5 and 7 are scored as a blend of the band above and below. In plain terms, there's nothing to tick off at 7. You reach it by doing everything band 6 asks, reliably, and starting to sustain what band 8 asks. The bands you study are 6 and 8.

The features examiners listen for, decoded

The official wording hides five concrete features inside words like "phonological" and "chunking." Here they are in the order they tend to cost marks.

Word stress: one wrong syllable, one lost word

English words carry a fixed stress, and moving it can change the word or simply stop a listener cold. "Comfortable" is three syllables in natural speech (KUMF-tuh-bul), not four. "Photograph," "photographer," and "photographic" each shift the stress to a different syllable. Put the stress on the wrong one often enough and the examiner spends effort matching what you said to a word they know. That effort is exactly what the band 6 line calls "occasional lack of clarity."

Sentence stress and rhythm: English is stress-timed

This is the band-6 killer, and the descriptor names it directly: "a lack of stress-timing." In English, content words (nouns, main verbs, adjectives) get stressed and beat out the rhythm, while function words (to, of, a, the, for) shrink. "I was going to the SHOP to BUY some BREAD" has three strong beats; everything else compresses. Speakers from syllable-timed languages give each syllable equal length, which sounds even and clear in their own ear but flattens the English rhythm the examiner is listening for.

Intonation: the pitch that carries meaning

Intonation is the rise and fall of your voice. It signals a question, marks contrast, and tells the listener you've finished a thought. A band-6 answer often starts with lively pitch and slides into a monotone, which is the "not sustained" the descriptor flags. Falling pitch at the end of a statement, rising on a genuine question, a small jump on the word you want to emphasise — that movement, kept up across a long answer, is half the climb to band 8.

Individual sounds: the phonemes that blur

English has around 44 phonemes, and a few reliably cause trouble. The "th" in think and this gets swapped for /s/ or /d/. Long and short vowels collapse, so ship and sheep, full and fool stop being distinct. None of these are about accent — they're about whether two different words still sound different. Fix the handful that change meaning and leave the rest.

Connected speech: how words join

Native speakers don't say words one at a time. They link them: "an apple" becomes "a-napple," "did you" becomes "dija," and weak forms shrink to down to /tə/. You don't need to force this, and overdoing it backfires. But the band 9 line, "flexible use of features of connected speech is sustained throughout," shows where the top of the scale lives. For most candidates aiming at 7, getting rhythm and stress right matters far more than chasing connected speech.

Your accent is not the problem. Clarity is.

Worth repeating, because it's the single most common worry. The descriptor rewards intelligibility, full stop. So the useful question isn't "how do I lose my accent" — it's "which features of my first language are actually reducing clarity." A few patterns show up often enough to name, and they're worth checking against your own speech:

• Japanese first language: inserting a small vowel after final consonants ("and" becoming "ando"), and merging /r/ and /l/. Both can blur word boundaries and swap words, which costs clarity.
• Hindi and other Indian languages: syllable-timed rhythm that flattens English stress-timing, plus /v/ and /w/ overlapping. The rhythm point maps straight onto the band 6 "lack of stress-timing" line.
• Vietnamese first language: dropping final consonants, so like and light or back and bad lose their endings. Final sounds carry grammar and meaning in English, so this one hits intelligibility hard.

None of these mean "sound less like yourself." They mean fix the specific habits that make a listener guess, and keep the accent that doesn't.

Diagnose your pronunciation in one recording

You can't fix what you can't hear. Record yourself answering a part 2 cue card for the full two minutes, then play it back and listen once for each band-6 marker:

1. Does your pitch flatten into a monotone after the first 30 seconds? (intonation not sustained)
2. Does every syllable get equal weight, with no strong beats? (lack of stress-timing)
3. Do certain words make you replay the clip to catch them? (mispronounced phonemes)
4. Do word endings disappear? (dropped final consonants)

Whichever marker is loudest is your target. Don't try to fix all four. The whole point of the band descriptor is that you climb one feature at a time, and pronunciation is the same diagnose-then-target logic as the four-axis self-check for the 6.5 ceiling.

Five minutes a day on one feature

A full mock speaking test spreads your attention across all four criteria, so pronunciation rarely gets isolated work. You do another mock, get "band 6," and still can't tell whether it was rhythm or sounds. Targeted drills fix that, and they're short:

• Shadowing for rhythm. Take 30 seconds of a podcast or the part 1 sample answers, play one sentence, and copy it immediately — same stress, same speed, same melody. Match the beats, not the accent.
• Stress marking. Take a sentence you'll likely use, mark the three or four words that carry the beat, and say it hitting only those. "I'd RATHER work from HOME because it SAVES TIME."
• Minimal pairs for one sound. If /θ/ is your blur, drill think/sink, three/tree, both/boss for a few minutes until the contrast is automatic.

Five minutes, one feature, recorded so you can compare. That's the idea behind IELTS Drills: feedback is tagged to a single Band Descriptor, so when pronunciation is the axis holding you at 6.5, you drill rhythm or sounds on their own instead of running another full test and hoping it sorts itself out.

Frequently asked questions

Does my accent lower my IELTS pronunciation score? No. The band 8 descriptor allows a first-language accent that has "minimal effect on intelligibility," and band 9 allows one with "no effect." You're scored on how easy you are to understand, not on how native you sound.

Do I need a British or American accent for IELTS? No. Either is fine, and so is your own, as long as it's clear. Examiners are trained on a wide range of accents and assess intelligibility, not a target accent.

Is there a band 7 pronunciation checklist? No. The descriptor defines band 7 by reference: all the positive features of band 6, plus some of band 8. The feature lists exist at bands 4, 6, and 8, so you study 6 and 8 and aim to sustain what's currently intermittent.

How much is pronunciation worth in IELTS speaking? A flat 25%. It's one of four equally weighted criteria, alongside fluency and coherence, lexical resource, and grammatical range and accuracy. The four are averaged for your speaking band.

Can weak pronunciation cap my whole speaking score? Yes. Because the four criteria are averaged, your lowest one pulls the band down. Three 7s and a 5 average to 6.5, not 7. A single weak axis sets your ceiling.

Is reading aloud good pronunciation practice? It helps with sounds and stress, but record it and compare to a model, because reading aloud without feedback just reinforces the habits you already have. Shadowing a native sample beats reading cold.

The bottom line

IELTS pronunciation isn't graded on your accent. It's graded on intelligibility: clear sounds, English stress-timing, intonation that keeps moving, and word endings that survive. Band 6 does these things some of the time; band 8 sustains them across a long answer. Band 7 sits between, with no checklist of its own.

Record yourself once, find the one feature that makes the examiner work, and spend five minutes a day on it. That's the difference between trying to erase an accent you were never marked down for and fixing the clarity that's quietly costing you 0.5.