Worried you say um too much? Check the rate for people your age

In 2001, adults in task conversations averaged 1.94 fillers per 100 words, from 1.54 for middle-aged speakers to 1.84 for older speakers. Language and Speech published the conversation analysis, which separated fillers from repeats and restarts in approximately 192,000 words. But the broader disfluency total was 5.97 per 100 words, so a normal meeting answer can sound worse in your head than the filler-only count.

The bigger number includes repeats and restarts

The 2001 paper by Bortfeld and colleagues examined task-oriented conversations, not scripted speeches. The corpus contained 48 pairs of people. It included 16 young pairs, 16 middle-aged pairs and 16 older pairs.

That source reported one combined disfluency measure for fillers, repeats and restarts together. On that broader measure, older speakers were reported at 6.65 per 100 words, middle-aged speakers at 5.69, and younger speakers at 5.55. The filler-only column is the tighter self-check if your worry is “um” or “uh.”

Hearing yourself hesitate changes the answer

The meeting problem is not only the sound itself. You hear yourself hesitate, then start monitoring every next word. In a review, interview or project update, that self-monitoring can pull attention away from the answer you were giving.

The Bortfeld paper is useful because it counted ordinary conversations while people were doing a task. If you want a broader frame for how listeners respond while you speak, audience feedback is the better lens than counting every hesitation as a failure.

Why does the age gap stay small?

The age effect was real in the paper, but modest. The authors wrote that older speakers produced higher overall disfluency rates than middle-aged and younger speakers, with a linear trend of t(82)=1.94 and p<.05.

For fillers alone, the rows are close. Young speakers were at 1.58 fillers per 100 words, middle-aged speakers at 1.54, and older speakers at 1.84. That spread is much narrower than the combined disfluency figures.

A 100-word answer makes the benchmark concrete

Take a 100-word project update in a meeting. At the overall filler benchmark, that answer would line up with 1.94 fillers. At the middle-aged row, it would line up with 1.54. At the older-speaker row, it would line up with 1.84.

That does not make every hesitation harmless in every room. A high-stakes interview answer can feel different from a task conversation. But it gives you a realistic scale before you decide that one audible “um” has ruined how prepared you sound.

Men and women split um and uh differently

A 2011 article in the University of Pennsylvania Working Papers in Linguistics looked at “um” versus “uh” in the Switchboard Corpus. Its age-by-gender table reported an um-to-uh ratio of 0.70 for women under 50, compared with 0.32 for men under 50.

The raw counts show why the ratio can feel different by speaker. Women under 50 had 3,232 ums and 4,611 uhs, while men under 50 had 3,262 ums and 10,141 uhs. For ages 70+, the reported ratios were 0.26 for women and 0.09 for men. That source counts which filler sound people use, not whether listeners judged it too many.

Every age row separates fillers from the bigger count

The cleanest table for your own age check is the filler-only column from the 2001 Language and Speech paper. The second measure keeps the broader disfluency comparison visible, because that larger number includes more than “um” and “uh.”

What this means for you

If you notice one “um” during a meeting, do not stop the sentence to punish yourself. Finish the thought. The filler-only benchmark is around 1.94 per 100 words overall, and the age rows in this paper sit between 1.54 and 1.84.

For practice, record a short answer and count only the fillers first. Then listen again for repeats and restarts. Those are different problems. If you lump them all together, the count can look like 5.97 when your actual filler rate is much lower.