Listenability, one of the four areas every speaking answer is scored on, has four factors. Two are about how your sentences are built, and they are on the Listenability page. This page is about the other two:
Rhythm, pronunciation, and intonation
Pauses, interjections, and self-correction
These are about delivery: what your answer sounds like when it reaches the listener.
Think about what the pronunciation factor is really measuring. It is not whether you sound Canadian, or British, or like anybody in particular.
Nothing in CELPIP's published criteria asks for a particular accent. The question is always the same: could the listener understand you without effort? A strong accent that is clear passes that test. A neutral accent that swallows its word endings does not.
This matters because a lot of people spend months trying to change their accent. The notes suggest that time is better spent on clarity: whole words, clear endings, and a pace the listener can follow.
A typical Level 8 answer can already have a steady rhythm and good intonation. What holds it at 8 is usually that every sentence is a simple one, which is the other half of Listenability.
So be realistic about this page. Clear delivery is the floor, not the ceiling. If your pronunciation causes the listener problems, or long pauses break your answer up, it will hold you down. If your delivery is already clear, polishing it further is not where your next level is. The sentence work on the Listenability page is.
A quick pace can make even a Level 9 answer hard to follow. Many people speed up on test day because they think fluent means fast, or because they are worried about finishing. Both cost clarity.
The right pace is one a stranger could follow on the first hearing. That is slower than most nervous speakers go, and a little slower than you speak to friends.
Rushing also creates the next problem on this page. Speakers who go fast run out of planned material early, and then they pause.
🎥 What a natural pace sounds like (13:26)
Two full ninety-second answers, performed at normal speaking pace against a real clock. Listen to the speed and the pauses, not the content.
Jump to: 1:00 the first full answer at exam speed · 10:25 the same question answered again
Long pauses break your rhythm, and so do pauses filled with um and uh. That is the Listenability cost.
There is a second one. A long pause is not only a delivery problem. It is time not spent answering, and completeness is scored in a different area, under Task Fulfillment. One silence, two scores.
Clear speech with no long pauses is exactly what a stronger answer sounds like.
Where long pauses come from
Almost never from pronunciation. Long pauses come from three places:
Not knowing what comes next. You finished your first idea and you have not decided the second one. This is a planning problem, and the fix is in your preparation time: write the next thing down, as a noun, before you start speaking. Numbered notes are what bring you back when you forget what comes next at forty seconds.
Searching for a word. You know what you mean and the English word will not come. The fix is to stop searching and describe it in easy words. A wooden frame that plants grow up. It takes two seconds and nobody minds. The full method is on Vocabulary for Speaking.
Starting a sentence you cannot finish. You begin a structure, lose it halfway, and stop. The fix is to finish the sentence simply, in any correct form, and start the next one fresh.
Short pauses are good
Not all pauses are problems. A short pause between two ideas helps the listener, the same way a comma helps a reader. The notes mark down long pauses, and pauses that interrupt rhythm. A one-beat pause at the end of a thought is what a clear speaker does.
🎥 Pauses are the result of mistakes in the preparation time
CELPIP Speaking: 18 Mistakes That Cost You Points on All 8 Tasks (17:37)
Jump to 15:50: preparation time, and why you must write notes rather than sentences.
CELPIP Speaking Task 8: The 3-Line Note That Fills 60 Seconds (11:19)
Jump to 4:12: why you number the parts in your notes, and what that does at forty seconds.
Interjections are the small sounds and words that fill silence: um, uh, like, you know, so yeah.
Self-correction is going back to fix something you just said: They was... they were late.
A few of each are normal, even in a Level 9 answer. The problem starts when there are enough of them to interfere with flow. So the goal is not zero. The goal is few enough that the listener does not lose the thread.
How to self-correct well
When you make a mistake that changes the meaning, fix it quickly and keep going:
I waited twenty... ten minutes, and then the bus came.
Two words and you are back in the sentence. Do not apologise, do not restart the whole sentence, and do not explain the mistake.
When the mistake does not change the meaning, leave it. He go to work every day is understood. Going back to fix it costs flow and gains nothing the listener needed. Small grammar slips are common even in Level 9 answers.
How to cut interjections
Most ums fill the moment where you are deciding what to say next. They are a sign of a planning gap, not a speaking one. A short silent pause is better than an um. Silence for one beat sounds like thinking. Um, uh, like repeated sounds like searching.
The examiners do not publish a method for this section, so what follows is ours: standard pronunciation teaching, aimed at the one thing the notes ask for, which is being easy to understand.
English has a beat. In each phrase, one or two words carry the meaning, and those get the stress: a little louder, a little longer. The small words around them, a, the, to, of, and, for, stay light and quick.
I'd take the early train, because the later one is always full.
If every word has the same weight, the listener cannot hear which ones matter, and the rhythm sounds flat even when every word is correct.
The tap test. Tap the table once on each stressed word as you say a sentence. If you are tapping on every word, you are stressing too many.
A flat voice costs you in two places: intonation sits under Listenability, and tone sits under Task Fulfillment. On a persuasion task, an answer that sounds flat also sounds less convinced than it should.
These are two sides of one thing. A flat voice makes a persuasive answer sound like it does not care. Tone is scored under Task Fulfillment, and intonation is how the listener hears it.
What to practise, again ours rather than CELPIP's:
- Rise and fall. A statement usually falls at the end. A voice that stays level to the end sounds unfinished, or bored.
- Lift the word that matters. In I really think this one is better, the lift on better is what makes it persuasive.
- Match the person. Warm and a little brighter for a friend in Task 1. Steady and calm for difficult news in Task 6. Lively for something you are arguing for in Task 5.
CELPIP publishes no list of sounds. This list is ours, and it is here because these are the sounds that most often make a word hard to recognise, which is what the pronunciation notes are about.
Word endings: -ed and -s. The most common problem, and the most important, because the ending often carries the meaning. I walk home and I walked home are different times. The bus and the buses are different numbers.
Practise: She missed two trains and walked the rest of the way.
/θ/ and /ð/, the th sounds. Tongue lightly between the teeth. Three is not tree. Then is not den.
V and W. V: top teeth on the bottom lip, with voice. W: round lips, no teeth. Very is not wary.
Long and short vowels. Ship and sheep. Full and fool. Hold the long ones a little longer.
Consonant groups at the end of words. Asked, texts, months, sixths. Do not drop the middle sound, and do not add a vowel after the last one.
You do not need all of these. Record yourself once and listen for which one makes your words unclear. Work on that one.
The same Task 1 advice, delivered two ways. The words in brackets are what a listener hears.
Before:
I think you should, um, take the... [3-second pause] ...the evening class, because it's, sorry, it is cheaper and, uh, you know, you can, you can still work in the... [2-second pause] ...the day.
After:
I'd take the evening class. It's about half the price, and you can keep your day job while you do it.
The second version is not faster. It is planned. The speaker knew the three things they were going to say before they started, so there was nothing to search for. Every pause, um and repeated word in the first version came from deciding while speaking.
Speeding up to sound fluent. Pace that is too quick makes even a strong answer hard to follow.
Long pauses while you decide what to say. They cost Listenability and completeness. Decide in your preparation time.
Searching for the perfect word. Describe it in simple words and keep going.
Correcting every small slip. Fix mistakes that change the meaning. Leave the rest.
Filling silence with um and like. A short silent pause sounds better.
Swallowing word endings. -ed and -s often carry the meaning.
A flat voice. Intonation carries tone, and tone is scored.
Working on your accent instead of your clarity. None of the notes asks for a particular accent.
1. Mark the pauses (5 min). Record any sixty-second answer. Play it back and write down the second of every pause longer than about two seconds. Then ask, for each one: was I searching for a word, or deciding what came next? The answer tells you which fix you need.
2. Slower by a little (3 min). Record the same answer twice, the second time about ten percent slower. Listen to both. Most people find the slower one sounds more confident, not less.
3. Shadowing (5 min). Play one of the full answers in the videos above. Speak along with it, matching the pace, the pauses and the rise and fall of the voice, not just the words.
4. Endings only (2 min). Read ten sentences aloud with -ed and -s endings and make every ending clear. She walked. He texts. They waited. Two buses.
5. The warm-up that matters. On test day, the unscored practice task is your chance to set your pace and volume. In practice sessions, record one short sentence before you start and listen to it once.
Related: Listenability for grammar and sentence structure, the other half of this score · Vocabulary for Speaking for describing a word you cannot find · Fast Planning for Tasks for the notes that prevent pauses · Scoring and Performance Standards for all sixteen factors.