Imagine standing up to give a speech in front of a critical audience. As you do your best to wax eloquent, someone in the room uses a clicker to conspicuously count your every stumble, hesitation, um and uh; once you’ve finished, this person loudly announces how many of these blemishes have marred your presentation.
This is exactly the tactic used by the Toastmasters public-speaking club, in which a designated “Ah Counter” is charged with tallying up the speaker’s slip-ups as part of the training regimen. The goal is total eradication. The club’s punitive measures may be extreme, but they reflect the folk wisdom that ums and uhs betray a speaker as weak, nervous, ignorant, and sloppy, and should be avoided at all costs, even in spontaneous conversation.
Many scientists, though, think that our cultural fixation with stamping out what they call “disfluencies” is deeply misguided. Saying um is no character flaw, but an organic feature of speech; far from distracting listeners, there’s evidence that it focuses their attention in ways that enhance comprehension.
Disfluencies arise mainly because of the time pressures inherent in speaking. Speakers don’t pre-plan an entire sentence and then mentally press “play” to begin unspooling it. If they did, they’d probably need to pause for several seconds between each sentence as they assembled it, and it’s doubtful that they could hold a long, complex sentence in working memory. Instead, speakers talk and think at the same time, launching into speech with only a vague sense of how the sentence will unfold, taking it on faith that by the time they’ve finished uttering the earlier portions of the sentence, they’ll have worked out exactly what to say in the later portions. Mostly, the timing works out, but occasionally it takes longer than expected to find the right phrase. Saying “um” is the speaker’s way of signaling that processing is ongoing, the verbal equivalent of a computer’s spinning circle. People sometimes have more disfluencies while speaking in public, ironically, because they are trying hard not to misspeak.
Since disfluencies show that a speaker is thinking carefully about what she is about to say, they provide useful information to listeners, cueing them to focus attention on upcoming content that’s likely to be meaty. One famous example comes from the movie Jurassic Park. When Jeff Goldblum’s character is asked whether a group of only female animals can breed, he replies, “No, I’m, I’m simply saying that life, uh…finds a way.” The disfluencies emphasize that he’s coming to grips with something not easy to explain—an idea that turns out to be a key part of the movie.
Experiments with ums or uhs spliced in or out of speech show that when words are preceded by disfluencies, listeners recognize them faster and remember them more accurately. In some cases, disfluencies allow listeners to make useful predictions about what they’re about to hear. In one study, for example, listeners correctly inferred that speakers’ stumbles meant that they were describing complicated conglomerations of shapes rather than to simple single shapes.
In fact, designers of synthesized voice systems have begun experimenting with the insertion of naturalistic disfluencies into artificial speech.
Disfluencies can also improve our comprehension of longer pieces of content. Psychologists Scott Fraundorf and Duane Watson tinkered with recordings of a speaker’s retellings of passages from Alice’s Adventures in Wonderland and compared how well listeners remembered versions that were purged of all disfluencies as opposed to ones that contained an average number of ums and uhs (about two instances out of every 100 words). They found that hearers remembered plot points better after listening to the disfluent versions, with enhanced memory apparent even for plot points that weren’t preceded by a disfluency. Stripping a speech of ums and uhs, as Toastmasters are intent on doing, appears to be doing listeners no favors.
Moreover, there’s reason to question the implicit assumption that disfluencies reveal a speaker’s lack of knowledge. In a study led by Kathryn Womack, experienced physicians and residents in training looked at images of various dermatological conditions while talking their way to a diagnosis. Not surprisingly, the expert doctors were more accurate in their diagnoses than the residents. They also produced more complex sentences—and a greater number of disfluencies, giving lie to the notion that disfluencies reflect a lack of control over one’s material. On the contrary, the study’s authors suggest that the seasoned doctors had more disfluent speech because they were sifting through a larger body of knowledge and constructing more detailed explanations while planning their speech.
If disfluencies appear to generally help communication more than they hinder it, why are they so stigmatized? Writer and linguist Michael Erard argues in his book Um… that historically, public speakers have been blissfully unconcerned with disinfecting their speech of disfluencies until about the 20th Century—possibly because neither hearers nor speakers consciously noticed them until it became possible to record and replay spoken language in all its circuitous and halting glory. The aversion to disfluencies may well have arisen from speakers’ horror at hearing their own recorded voices. Erard suggests that the modern repugnance for disfluencies is less an assessment of a person’s speech than it is a “deeper judgment about how much control he should have over his self-presentation and his identity.” In truth, disfluencies appear to distract mainly those who have been trained to revile them.
Perhaps there’s an argument to be made that public speaking is different from day-to-day communication, that it’s a performance in which the artist is meant to demonstrate almost superhuman mastery over speech and make verbal virtuosity look easy precisely because of the absence of cues that reveal its complexity. Maybe so. But the prohibition of ums should be recognized for what it is—a display focused on presenting the speaker in a flattering light—and not mistaken for courtesy directed at the listener.
In fact, designers of synthesized voice systems, who often are rather solicitous when it comes to the hearer’s ease and comfort, have begun experimenting with the insertion of naturalistic disfluencies into artificial speech (though it’s too soon to tell whether listeners respond to these as they do to human disfluencies). It’s an irony of our age that robots, unconcerned with ego, may be busy putting disfluencies into their speech just as humans, preoccupied with their self-images, are submitting to strenuous training to take them out.
Julie Sedivy has taught linguistics and psychology at Brown University and the University of Calgary. She is the co-author of Sold on Language: How Advertisers Talk to You and What This Says About You and more recently, the author of Language in Mind: An Introduction to Psycholinguistics.