IELTS Speaking pronunciation: the features to practise, and how
Quick answer
IELTS Pronunciation is marked on how easy you are to follow. The descriptors refer to a range of features, which this site groups as five: individual sounds, word stress, rhythm, intonation and connected speech. Practise them in that order of payoff, not in the order of a textbook: stress and phrasing first, then intonation, then only the sounds that make a listener misunderstand you. Shadow short recordings, record your own answers and listen back. An accent is not the target. IELTS publishes no list of sounds to learn and no score for any single feature.
In IELTS Mentor AI's Foundation Track, the Listen & Shadow step scores how closely you repeat a phrase and gives pronunciation feedback word by word.
iOS and Android
Paid app: what it costs · billing and refunds · support
On this page
Pronunciation is hard to practise because it is easy to confuse with accent. This guide turns it into a short list of things you can train. For the question of whether your accent costs marks, read the separate guide on accent and the Speaking score. For tools, see the roundup of pronunciation apps.
Key takeaways
- Pronunciation is one of four equally weighted Speaking criteria. The Speaking band is their average, reported as a whole or half band.
- It marks how easy you are to understand, not how close you sound to a native speaker.
- Train in this order: word stress, phrasing and sentence stress, intonation, then the individual sounds that cause misunderstanding.
- IELTS publishes no list of sounds to learn, and no band is promised for any single technique.
- Shadowing and recording yourself show you the problem directly.
What the criterion listens for
IELTS.org’s public Speaking band descriptors treat Pronunciation as one of four criteria of equal weight, alongside Fluency and Coherence, Lexical Resource, and Grammatical Range and Accuracy. The Speaking band is the average of the four, reported as a whole or half band. The descriptors do not name a long list of sounds. They describe a range of features and ask how many you control, how well, and for how long. This site’s paraphrase of each band is in the band descriptors explorer.
Read across the bands and the direction is clear. At the lowest, few features are in play and the listener has to work to follow you. In the middle, several are used with mixed control that does not last a whole answer. Near the top, a wide range is kept up and your first-language accent has little or no effect on how easily you are understood. The question throughout is how much effort the listener spends. The five features are individual sounds, word stress, rhythm, intonation and connected speech, and the sections below take them in order of usual payoff.
Individual sounds: only the ones that matter
Every learner has a long list of small differences from standard English, and most do not matter for an exam. A sound matters when a listener could hear a different word from the one you meant, or has to stop and work out which word you said.
IELTS publishes nothing about which sounds to practise, so test yourself. Record a minute of speech and ask a friend, a teacher or a transcription tool to write down what they heard. A word that comes back wrong is a real problem; one that comes back right, with an accent, is not.
Fix the problem sounds with minimal pairs, which are two words that differ in one sound, such as “ship” and “sheep”, or “light” and “right”. Use them only for the pairs that have actually caused confusion in your own recordings. Drilling pairs you can already hear and say wastes time you need for stress and rhythm.
Word stress
English stresses one syllable in each longer word: “PHOtograph”, “phoTOGraphy”, “photoGRAPHic”. The stressed syllable is longer, louder and clearer, and the others weaken. When the stress lands on the wrong syllable, a familiar word becomes hard to recognise even if every sound in it is correct. That is why stress usually gives more return than any single consonant.
Stress also moves between related words: “ecoNOmic”, “eCONomy”, or “IMport” as a noun and “imPORT” as a verb. Learn the stress with each new word, and with its family.
The word stress trainer on this site lets you hear a word and pick the stressed syllable. In your own recordings, listen for long academic words, where misplaced stress is most common.
Sentence stress, weak forms and rhythm
Words have stress, and so do sentences. In a natural English sentence you stress the words that carry meaning, which are usually nouns, main verbs, adjectives and adverbs, and you weaken the small grammar words around them. Compare these two:
- “I WANT to GO to the BEACH on SATurday.”
- “I WANT TO GO TO THE BEACH ON SATURDAY.”
The first has strong beats and quick, quiet words between them. The second gives every word the same weight and sounds robotic. Weak forms are the other half of the same habit. In “I can go”, the word “can” is usually short and quiet, close to “kn”. In “Yes, I can”, it is strong and clear. Saying “can” at full strength every time hides which word matters.
This is where rhythm comes from. English tends to place stressed beats at fairly even intervals, and it squeezes the unstressed words in between. If you stress the content words and let the small words shrink, the rhythm mostly follows.
Chunking: pausing at thought groups
Chunking means grouping words into short phrases and pausing between groups, not between single words. Compare:
- “I moved to Samarkand / when I was twelve / because my father / got a new job.”
- “I / moved / to / Samarkand / when / I / was / twelve.”
The first pauses where a thought ends. The second pauses after every word and loses the meaning between them. A good group is usually three to six words and ends at a natural break: after a time phrase, before a “because”, after the subject of a long sentence.
Pausing in the wrong place sounds like hesitation, so chunking overlaps with fluency. The answer on how to improve Speaking fluency covers that criterion on its own terms, though the criteria are marked separately. Two sibling guides cover the fluency side of the same habit: fillers, pauses and self-correction and linking words for Speaking. Practise on a written text first: mark the slashes where you would pause, then read it aloud.
Intonation: lists, questions and finished points
Intonation is the rise and fall of your voice. Three patterns come up constantly in the Speaking test.
Lists. In a list, the voice rises slightly on each item and falls on the last: “I like cycling, swimming, and walking.” A falling pitch on every item sounds as if you have finished too soon, and a rising pitch on the last one sounds as if you are still going.
Questions. Yes/no questions usually rise at the end, and “what”, “why” or “how” questions usually fall. “Could you repeat that?” rises.
Finished and unfinished points. A falling pitch at the end of a sentence tells the listener you have finished. A level or slightly rising pitch says there is more to come. If your voice falls after every clause, the examiner may think you have stopped and wait for you to continue. If it never falls, you sound unsure. Make the voice fall on the last word of a point you have really finished.
Flat intonation makes a good story sound recited. If you speak in one pitch, overdo the rises and falls on a recording, then reduce them until they feel natural.
Connected speech, in moderation
In natural speech, words run into each other. “Want to” becomes close to “wanna”. The last consonant of one word links to the vowel that starts the next, so “pick it up” sounds like “pi-ki-tup”. Some sounds disappear in fast speech, such as the “t” in “next day”.
The descriptors list connected speech among the features, but you do not need to copy every contraction. Two habits are enough: link a final consonant to a following vowel, and let weak words shrink as above. Do not force fast, blurred speech. Speaking at native speed before your sounds are secure costs the clarity you need.
Clarity is the goal, and an accent is not the problem
The descriptors for the upper bands say a first-language accent has little effect on intelligibility. This site’s guide to accent and the Speaking score explains the wording in detail. In practice, do not spend months trying to remove an accent. Spend them on the features above, which change how easily a stranger follows you. A speaker with a strong accent, clear stress and good phrasing is easy to understand; one with a mild accent and flat stress may not be.
A practice routine
This routine takes about 20 minutes a day.
- Shadow a short clip. Choose 30 to 60 seconds of clear speech, such as a news item, a talk or a Listening recording from a practice paper. Play one sentence, pause, and repeat it with the same stress, rhythm and tune.
- Record yourself on one topic. Use a Part 1 question or a Part 2 cue card from the cue card hub. Speak for one to two minutes, as in Part 2. Keep the recording.
- Listen back for one thing. Do not fix everything. On Monday listen only for word stress, on Tuesday only for phrasing, on Wednesday only for intonation. Note three words or phrases to correct.
- Fix and re-record. Say the three items in isolation, then in a sentence, then record the answer again. Compare the two recordings.
- Test sounds weekly. Have someone transcribe a minute of your speech, and add any sound that produced a wrong word to your minimal-pair list.
What to expect from tools
Check any app’s feedback against the descriptors yourself: a tool that marks a clear sound as an error because it differs from a reference accent is measuring similarity, not clarity. The apps roundup sets out what products claim in their own listings. IELTS makes no promise that any method raises your band, and neither does this page.
Frequently asked questions
Which pronunciation features does IELTS Speaking assess?
IELTS.org's public Speaking descriptors for Pronunciation refer to a range of features, which this site reads as individual sounds, stress, rhythm, intonation and connected speech. The descriptors scale with how many of these you control and how long you keep that control. They do not list particular sounds or words.
Do I need to fix every sound I pronounce differently?
No. IELTS publishes no list of sounds to practise, and the criterion is about being understood. Work on a sound only when you notice that it makes a listener hear the wrong word. A difference that never causes confusion is part of your accent and is not what is being marked.
Is word stress or individual sounds more important?
For most candidates word stress gives more return, because a misplaced stress changes how recognisable a whole word is, while one slightly different consonant often leaves the word clear. This is a practical judgement about speech, not an IELTS rule, so check it against your own recordings.
Does shadowing really help pronunciation?
It trains rhythm, stress and phrasing because you copy a whole stretch of natural speech at its own speed. It works best with short recordings that you repeat several times and compare with your own voice. It is practice material, and IELTS makes no claim that it raises a band.
Should I speak faster to sound fluent?
No. Speed that blurs your sounds or removes your pauses makes you harder to follow, which is the opposite of what this criterion rewards. A steady pace with clear stress and short, sensible phrases is easier to understand and easier to keep up for the whole test.
Practise this with instant AI feedback
Record your answer in the app and get your grammar, vocabulary, fluency and pronunciation analysed in seconds.
Paid app: what it costs · billing and refunds · support