Speaker Hub uses necessary cookies for secure accounts. With your permission, first party analytics help us improve signup and onboarding. We do not use advertising or cross site tracking cookies. Read the Cookie Policy.
How to stop saying um, and why the research says not to try too hard · Speaker Hub
How to stop saying um, and why the research says not to try too hard
Two psychologists spent years studying um and uh and concluded they are words, not mistakes. Here is what to cut instead, and the one filler that actually costs you.
•6 min read
Almost every list of public speaking tips opens the same way. Stop saying um. Count them, punish them, drill them out.
The two researchers who studied um most closely reached a different conclusion, and it is worth knowing before you spend a month training a habit out of your mouth.
Um and uh are words
In 2002, Herbert Clark and Jean Fox Tree published Using uh and um in spontaneous speaking in the journal Cognition. Their argument is in the title. Not noises, not errors. Words.
Their claim is that speakers use them deliberately, if not consciously, to announce a delay. Uh signals a short one. Um signals a longer one. A listener hearing um adjusts, because they have just been told something is coming and it is going to take a second.
Fox Tree's earlier paper, Listeners' uses of um and uh in speech comprehension in Memory and Cognition, tested what that does to the person listening. Fillers helped. Listeners recognised the word after a filler faster than the same word without one.
So the honest version of the advice is not "never say um." It is closer to "um is a signal, and signals stop working when you send them constantly."
The number is not the problem
Counting fillers is satisfying because it produces a number, and a number feels like progress. It also misleads, because it treats every um as identical.
Ask instead whether the filler is carrying a load:
One before a difficult idea is doing what Clark and Fox Tree described. Leave it.
Three in a row while you hunt for a word is a stall, and the room can hear you hunting.
One at the front of every sentence is a tic. It has stopped meaning anything, which is exactly why it grates.
The tic is the version worth working on, and it is a smaller problem than "eliminate all fillers."
The filler that actually costs you
There is a category worse than um, and nobody counts it because it is made of real words.
"Sort of." "Kind of." "I guess." "Does that make sense?" "This might be obvious, but."
These are hedges, and unlike um they change what you said. Um marks a pause. "I guess" tells a room full of people who came to hear an expert that the expert is not certain. You can deliver a flawless talk with zero fillers and still give away every ounce of authority in it through hedging.
If you have time to fix one habit, fix this one. It is doing more damage and almost nobody is watching for it.
Replace the filler with the pause
Here is the mechanism. A filler is not a random noise. It fills a gap you are not comfortable leaving open, usually because silence on a stage feels ten times longer than it is.
So the work is not deletion. It is tolerating the gap.
Record five minutes of yourself talking about your topic without notes. Phone audio is fine.
Play it back and mark every filler. Do not fix anything yet. Just notice where they cluster, because they will cluster.
Speak the same passage again, and everywhere you marked a filler, close your mouth instead.
It will feel wrong. That feeling is the whole exercise. The gap that feels like a catastrophe from the inside reads as composure from the fifteenth row.
Where the clusters usually are
Fillers gather in predictable places, and knowing them saves you a lot of recording.
At the seam between two ideas. You have finished a thought and have not decided which one comes next, so you hold the floor with a noise. Fix the running order, not the noise.
Right after a question from the audience. You are buying thinking time in public. This one is legitimate, and a pause serves you better because it looks like consideration.
In the first thirty seconds. Adrenaline is highest and your material is least warmed up. This is the strongest argument for knowing your opening cold, which is a separate job covered in how to open a speech.
What to do about a talk you have already given
If you have a recording, you have data, and it is more reliable than your memory of how it felt.
Listen once for fillers, and separately for hedges. Write down the sentence around each one. The pattern usually shows up in three or four passes, and it is almost always tied to a specific part of the talk rather than spread evenly.
That is the part to rehearse. Not the whole thing.
Reading about it only gets you so far, though. The speech analysis tool marks fillers, hedges and pace against the actual recording, which is faster than transcribing yourself and less forgiving than listening back once.
Not in the way most speaking advice claims. Herbert Clark and Jean Fox Tree argued in Cognition in 2002 that uh and um are conventional English words that signal a coming delay, uh a short one and um a longer one. Fox Tree's earlier work in Memory and Cognition found that listeners recognised the following word faster when a filler preceded it. Fillers become a problem at volume, when they replace pauses rather than mark them.
What is a normal number of filler words?
There is no official threshold, and any specific number you see quoted is someone's rule of thumb rather than a research finding. A more useful test than counting is asking whether a filler is doing a job. One before a hard idea buys the room a beat. Six in one sentence is a habit running without you.
How do I actually reduce filler words?
Replace them with silence rather than trying to delete them. A filler fills a gap you are uncomfortable leaving open, so the fix is getting comfortable with the gap. Record yourself, mark every filler, and practise the same passage pausing where the filler was. The habit weakens when the silence stops feeling like a mistake.
The Speech Analyzer pairs your recording with a transcript and an estimated response timeline, and none of that is a measurement of how the room actually felt. Here is what the number is built from, what to check before you act on it, and why Compare matters more than any single score.
