Hesitation, Fillers and the Long Pause
Filler density and the longest silence in your answer are both taken from the audio and both carry ceilings. Hesitation is a planning problem far more often than a language problem.
What the recording shows about hesitation
Two things about hesitation are measured directly from the audio rather than inferred from what you said. The first is filler density: how many times per hundred words you produce um, uh, er, mm or one of their relatives. The second is your longest single silence, measured as the gap between one word ending and the next beginning.
The thresholds are specific. Under two fillers per hundred words is unobtrusive and costs nothing. Over six per hundred words caps Listenability at 6. Separately, any single silence over four seconds inside an answer caps Listenability at 7. Those two ceilings are independent, so a candidate can trigger both in the same sixty seconds.
Six per hundred words sounds generous and is not. At a comfortable pace a sixty second answer runs to roughly 140 words, so nine fillers crosses the line. Nine is easy to produce: three at the start while you find your footing, three at each join between ideas, and you are there without ever feeling that you struggled.
Why hesitation happens, and it is not usually vocabulary
The instinctive explanation is that you paused because you did not know a word. Sometimes that is true. Far more often, the thing you were retrieving was not a word but an idea, and the pause fell at exactly the point where your plan ran out.
The evidence for this is in where the gaps land. Vocabulary hesitation happens mid phrase, after the or a, with the sentence left hanging open. Content hesitation happens at sentence boundaries, after a full stop, when the previous thought is finished and the next one does not exist yet. Listen back to your own recordings and count which kind you produce. For most candidates it is overwhelmingly the second.
That distinction matters because the two have completely different fixes. Content hesitation is cured in the preparation window by deciding on three ideas instead of one. Vocabulary hesitation is cured by building the habit of going around a missing word rather than waiting for it. Practising fluency drills to fix content hesitation is the most common wasted effort in speaking preparation.
The opening three seconds
The recorder opens on its own when preparation ends, and a large share of all the filler on a test is produced in the first three seconds of each answer. Eight tasks multiplied by two or three fillers at the top of each is twenty fillers before any of the difficult parts.
The fix is mechanical and takes one line of your preparation window: write your opening sentence out in full. Not the whole answer, just the first ten or twelve words. When the microphone opens you speak something you have already composed, and the part of your brain that would have been improvising is instead getting your second sentence ready.
It helps to make that opening sentence carry content rather than ceremony. Well, in this picture I can see that there are many things happening is a full sentence of nothing. This looks like a Saturday morning at a neighbourhood bakery, and it is completely packed is the same length and has already established a setting, a time and an inference.
Drill one: the paraphrase reflex
The goal is to remove the four second search. Pick twenty everyday objects that are hard to name under pressure: a whisk, a dustpan, a spirit level, a hole punch, a bicycle pump, a parking meter. For each one, say out loud, within two seconds, a description that would let somebody identify it without the noun.
The thing you beat eggs with, like a bundle of wire loops on a handle. The flat pan on a stick you sweep dirt into. The tool with the little bubble in it that tells you if a shelf is straight. None of those needs the word, and all of them are more interesting to listen to than the word would have been.
Two minutes a day for a week converts this from an emergency measure into a reflex. The measurable outcome is that your longest pause stops crossing four seconds, because the search never starts. The secondary outcome is that Vocabulary improves too: circumlocution under pressure is precisely what Task 8 is built to test, and the habit transfers straight into it.
Count your own fillers on one recording before you do anything else. Over six per hundred words caps Listenability at 6, and most candidates have no idea which side of that line they sit on until they count.
Drill two: three anchors, no script
Take any prompt. Give yourself thirty seconds and write exactly three words, one per idea, nothing else. Then record sixty seconds. The constraint is what makes this work: with only three words on the page you cannot read, so every sentence is improvised, but you can never run out, because there is always a next anchor waiting.
Listen back specifically for pauses at sentence boundaries. If there are still long ones, your anchors were too close together and the second one was really a restatement of the first. Three genuinely separate ideas produce three genuinely separate stretches of speech.
Do this on the same prompt three times across a week. The interesting result is that the answer gets fluent without getting memorised, because the wording changes every time while the structure holds. That is the state you want on test day: a plan that is firm and language that is live.
Drill three: the deliberate pause
This one is counter-intuitive. Fast, filler heavy speakers usually try to eliminate pausing altogether, and fillers are exactly what rushes in to fill the gaps that are left. The alternative is to schedule silence rather than banish it.
Record a sixty second answer with one rule: at each join between your three ideas, stop completely for one full second before starting the next sentence. It feels enormous from the inside and it is inaudible from the outside. A one second break at a structural boundary reads as confidence and gives the listener a beat to catch up.
What the measurements show after a few sessions of this is the pattern you are aiming for: filler density down, because the fillers had been doing the job the pauses now do, and longest pause still well under four seconds, because a planned one second break never grows into a five second search. Hesitation is not the absence of silence. It is silence in the wrong place, wearing sound.
Schedule one full second of silence at each join between ideas. Fillers appear in unplanned gaps, so giving the gap a purpose removes the thing that was filling it, and a one second break at a clause boundary sounds deliberate rather than stuck.
Practise it
Five questions on hesitation, what it costs and where it actually comes from.