What "Um" Tells Your Audience
By Mira Lindqvist · Voice & Pace
How do you stop saying um and uh? Do not aim for zero. Listeners use fillers to get ready for the next word, so the cost is not confusion but sounding unsure. Replace the filler with a silent beat, keep fillers away from the sentences that carry your key claims, finish each phrase cleanly, and count them on a recording afterwards, not while you speak.
Most speaking advice treats "um" and "uh" as an editing problem: noise a good speaker eventually learns to cut. A body of psycholinguistic research has spent decades asking a more specific question. Rather than "should speakers say these?", it asks "what do listeners do with them, in real time?" The answer is that a filled pause is not silence with a sound stapled to it. It carries information, and some of that information helps the listener while some of it costs the speaker.
Listeners use "uh" to prepare
In a 2001 study, psycholinguist Jean Fox Tree tested how listeners respond to filled pauses in recorded speech.1 Listeners were asked to respond as soon as they heard a particular word, and the recordings either kept a filler just before that word or had it removed. After "uh", listeners recognised the following word faster. "Um" did not produce the same speed-up, and it did not slow them down either.
The interpretation is that "uh" works as a signal that a short delay is coming, and the listener's comprehension system uses the signal to get ready rather than simply waiting. The two fillers are not interchangeable, at least in this respect.
Fillers can even help memory
A 2011 study by Scott Fraundorf and Duane Watson played listeners a story with or without filled pauses scattered through it.2 The listeners who heard the fillers remembered the plot better. The benefit was specific to the fillers: coughs placed at the same points did not help. A break in fluency seems to tell the listener that something worth attention is about to arrive.
None of this is a case for adding "um" to a talk. It is a case against the idea that fillers only ever confuse an audience. For comprehension, they are mostly harmless and occasionally useful.
A filled pause and a silent one don't say the same thing
The cost lies somewhere else. In a 1995 study, Susan Brennan and Maurice Williams asked listeners to judge how sure a speaker was of the answers they gave to quiz questions.3 The speakers' answers came after delays of different lengths, sometimes silent and sometimes filled. Answers that came after a long delay were heard as less certain, and so were answers preceded by a filler. Rising intonation lowered the judgement too.
The two kinds of pause are read as different signals. A silent pause can sound like composure. A filled one sounds like searching. Same delay, same missing word, different verdict on how sure the speaker is.
Confidence lives in the sound as well as the words
That listeners read confidence from the type of pause fits a broader finding. A 2017 study by Xiaoming Jiang and Marc Pell found that listeners reliably tell a confident statement from a doubtful one using vocal cues alone, such as pitch, loudness and tempo, independent of what the sentence says.4 The filled-pause effect is one instance of that general pattern, not a quirk of "um". Much of how sure a speaker sounds comes from the delivery layered on top of the words.
Zero fillers is the wrong goal
This collection has already argued the case for speaking badly in public that chasing a filler-free talk is the wrong target on its own terms. The research adds a sharper reason. Fillers do real work for the listener, so the problem with them is not comprehension. The problem is what a run of them suggests about the speaker's certainty, especially when they land just before a claim the speaker wants believed.
There is also a practical trap. Speakers who try to suppress every filler can trade them for something worse: rushed sentences with no pauses at all, or a strained delivery that sounds rehearsed. The aim is not to remove the delay. It is to change what the room hears during it.
How the Coach counts them
Exprea's Coach treats fillers and silences as separate signals. Fillers are counted as a rate per hundred words, so speaking faster does not change the figure; how that list works is set out in filler rate. Silences are read by where they fall: a pause of 0.6 to 1.5 seconds at the end of a clause or sentence counts as a deliberate pause, while a pause of half a second or more inside a clause counts as a hesitation. The scoring follows the research above: the same delay is treated differently depending on whether it marks structure or a search for words.
What this means in practice
Keep the pause, change its sound. Where a pause is going to happen anyway, because you are retrieving a word, a number or a name, a silent beat reads as more composed than the same delay filled with "um", at no cost to the listener's understanding.
Protect your key claims. A filler in the middle of an anecdote costs almost nothing. A filler immediately before your main recommendation is the one that makes it sound unsure. If you rehearse anything, rehearse the first few words of the sentences that matter most.
Finish the phrase, then stop the sound. Many fillers are the voice running on while the mind catches up. Ending the phrase cleanly and letting the breath go, rather than holding a sound in reserve, leaves a silence where the filler would have been.
Count on a recording, not in the moment. You cannot monitor your fillers while also thinking about your content. Listen back afterwards, notice where they cluster, and work on those spots.
Don't chase zero. A few fillers in a ten-minute talk are normal speech. The research suggests your audience barely notices them, and may even be helped by them. The ones worth removing are the ones standing in front of a claim you need believed.
Silence is the alternative to all of this, and it is harder than it sounds. The pause is the point covers how long to hold one and where.
References
- Fox Tree, J. E. (2001). Listeners’ uses of um and uh in speech comprehension. Memory & Cognition, 29(2), 320–326. https://doi.org/10.3758/bf03194926
- Fraundorf, S. H., & Watson, D. G. (2011). The disfluent discourse: Effects of filled pauses on recall. Journal of Memory and Language, 65(2), 161–175. https://doi.org/10.1016/j.jml.2011.03.004
- Brennan, S. E., & Williams, M. (1995). The feeling of another’s knowing: prosody and filled pauses as cues to listeners about the metacognitive states of speakers. Journal of Memory and Language, 34(3), 383–398. https://doi.org/10.1006/jmla.1995.1017
- Jiang, X., & Pell, M. D. (2017). The sound of confidence and doubt. Speech Communication, 88, 106–126. https://doi.org/10.1016/j.specom.2017.01.011