Um-- Slips, Stumbles, and Verbal Blunders, and What They Mean

by Erard, Michael.

  The converse issue is just as interesting: Why do we seem to be such lackadaisical, inattentive speakers? Why do we let such things slip past our lips? We do, in fact, monitor our speaking, and we often fix it, too. As a result, the failure to be fluent isn’t a failure of knowing how to speak. After all, Rudolf Meringer found that 75 percent of sound slips and 53 percent of word slips were caught by speakers. And as the conversational analysis pioneer Harvey Sacks and his colleagues noted, “Even casual inspection of talk interaction finds self-correction vastly more common than other-correction.” (We correct ourselves more than we correct one another.) It’s not at all uncommon to hear someone say, “Sure enough ten minutes later the bell r—the doorbell rang.”

  In the parlance of speech science, the fixing is called a “repair.” Willem Levelt provided a basic anatomy of the repair’s three parts: there’s the original sentence (1), interrupted by some moment of editing (2), then the repair itself (3). From adult speakers who were describing visual patterns, Levelt gathered examples like these:

  go from left again to

  uh

  pink again to blue

  1

  2

  3

  Levelt marked several crucial moments in the sentence with numbers. It’s easy to find examples in verbatim transcripts of prominent media figures we listen to all the time. Take this excerpt from a 2006 interview with Garrison Keillor and Robert Altman, when Altman’s movie A Prairie Home Companion had been released. The interviewer asked how Altman as a director had treated Keillor, who replied:

  No. I was alarmed at how—at how indulgent he—he is as—as a director. He’s very, very precise about—about camera movement and—and he knows exactly where—where he’s going at all times. But as far as, you know, the facial expressions, you know, of an actor and—and how you do something.

  The verbatim transcript makes Keillor seem disfluent, though anyone who’s heard him on the radio knows that his stumble has some charm. Keillor repeats himself and pauses, but until “of an actor and,” he hasn’t repaired. Until that point he’s doubtlessly been monitoring his speech. But at “of an actor and” he decides that he needs to repair. This is something we learn how to do when we are as young as two or three years old. The older we get, the more we repair, and the more complex the repairs become. Levelt found other reasons that people repaired. Sometimes it was because the information was incorrect; sometimes the speaker was trying to avoid social awkwardness.

  Of the many reasons for repairs, the greatest number occur because a speaker anticipates that he or she is about to make an error, say something ungrammatical, or perhaps make a slip of the tongue. A slightly less common reason is that the speaker realizes that the information needs to be less ambiguous. Finally, people interrupt themselves and start again because they want to present something in a different order. We can’t be certain that Keillor was about to make a speech error—more likely he was trying to keep his answer short, to make it fit into the conversation.

  In conversation and other types of spontaneous talking, people tend to stop speaking as soon as they’ve detected a mispronunciation, a slip of the tongue, something inappropriate, or any deviation from what they planned to say. Consider this exchange between Larry King and Charlie Rose (on Rose’s show) about Ted Turner and Jane Fonda:

  LARRY KING: Yeah. Ted would—Ted was—anybody who knows Ted would have to say they miss him. You have to miss him.

  CHARLIE ROSE: What happened to him and Jane?

  LARRY KING: You know, they’re the best of friends.

  CHARLIE ROSE: Yeah.

  In King’s first sentence, he starts and stops the sentence twice because he’s obviously having a difficult time selecting the most appropriate thing to say about Turner. Now, speakers usually finish the word that they’re saying (“Ted would—Ted was”), unless the word itself wasn’t intended. This happens so regularly that Levelt called it the “Main Interruption Rule.” It indicates that a speaker will “stop the flow of speech immediately upon detecting trouble.” This is what led Oprah Winfrey to say on her show, to one of the young men who had turned his father in for bank robbery, “I can see that, you know, I mean, obviously, all of you are up—upset about it, and you are still very much emotionally impacted by it.” (Levelt might note that Oprah didn’t need to interrupt “upset” if it was the word she wanted.)

  Levelt also found another pattern in repairs. It seems that speakers are more likely to detect trouble or make a change when it occurs toward the end of a phrase, clause, or other chunk of language. In “go left again to uh pink to blue,” the interruption occurs at the end of a chunk. Likewise, when Rush Limbaugh said of Democrats, “Some of them tried to couch it in—in, oh, disguised ways, but you strip all that away and they were clearly trying to make the case.” Here the repair occurs toward the end of the clause.

  In Levelt’s example,

  go from left again to

  uh

  pink again to blue

  1

  2

  3

  he calls moment 2 the “editing term.” This appeared in Tiger Woods’s response to a question about the origin of his personal discipline. He replied, “I get that—you know, people think I get that from my dad.” In Levelt’s parlance “uh” and “um” are editing terms. “You know,” “sort of,” “rather,” “well,” “I mean,” and “kind of” are, too. According to Levelt, these editing terms have several uses—after all, why not simply remain silent? A speaker may not want to be interrupted, so he prefers to keep a pause filled with sound. He may also use the editing term as an opportunity to make his meaning more clear, or to comment on the mistake itself. In Levelt’s sample of adult Dutch speakers, 42 percent of repairs had no editing term. When people repaired to make a message more appropriate, rather than to make the language more specific, they used “uhs,” “ums,” and other editing terms only 28 percent of the time. On the other hand, if a speaker repaired in order to correct a statement, 62 percent of the repairs used an editing term.

  In Levelt’s data, “uh” was the most frequent editing term, used in 30 percent of all repairs. The earlier that a speaker detected a problem, the more likely he or she was to use “uh.” Though the theories about the meanings of “uh” and “um” vary widely, Levelt is content to say that the pause filler basically means “I’ve temporarily forgotten X.” Searching for “uh” and “um” among transcripts on LexisNexis turns up very few of them—not surprisingly, given that transcription services like the Federal News Service routinely clean them up. However, there are a few that sneak in, such as this sentence from a Chinese expert speaking on National Public Radio: “And now people have their own, uh, some of them have their own independent income.”

  In Levelt’s sentence, the last moment in the anatomy of the repair is marked “3”:

  go from left again to

  uh

  pink again to blue

  1

  2

  3

  It is the repair proper, where the speaker picks up the sentence again.

  Interestingly, seeing actual examples of repairs allows one to see how strong the urge is to create grammatical sentences. Take this snippet of an interview with Diane Sawyer, in which she described the play Angels in America: “It is. It is. It’s a great play. And as we know, it really—it rocked Broadway. It really brought people into the theater who had never come before.” Even though the sentence looks disrupted by her repair, “it really—it rocked,” it follows another pattern in repairs: the words after the interruption have to give closure to the most important chunk of language before the interruption. Sawyer didn’t say, “It really—rocked Broadway.” Rather, to create this continuity, she backtracked to the beginning of the clause and repeated the entire thing. This drive to completion is what creates many instances of repetitions, whether of words or phrases.

  Unfortunately, the self-monitoring mechanism of the average person as well as the professional talker doe
sn’t block all the instances of disfluency or errors. It can’t. As Levelt puts it so colorfully, “The meshes of a speaker’s trouble net are too wide to catch all the queer fish in his own speech.” Depending on where in the utterance the problem is perceived, what the social situation is, what type of problem it is (a sound versus a word, for example), or whether it will be disruptive for the listener, a speaker’s monitor or self-editing mechanism catches things—more or less. About half of the errors in the speech samples that Levelt studied weren’t, in fact, caught or corrected at all. That’s not to say the speaker didn’t hear them; perhaps they just neglected to stop. Other speakers were inordinately fond of stopping. In about a quarter of all the repairs that Levelt studied, he couldn’t discern any reasons why the speaker interrupted him-or herself—some speakers are so quick to fix (or are self-monitoring so sensitively) that no queer fish are ever heard.

  Fifty years of research into speech disfluencies and the normal aspects of spontaneous human speaking have demonstrated that many of our norms for “good speaking” do not parallel the biological imperatives of language itself. Those standard-bearers who insist that speaking should be faultless and interruption free can’t account for the fact that we don’t perceive most verbal blunders, and they can’t explain the sheer numbers of faults, except to attribute carelessness, ignorance, lack of preparation, or some other deficit to speakers. Not only do the prescriptive tastes of society diverge from the biological reality of language. But language remains steadfast; norms, though, shift and evolve. This is clearly on display in the story of what may be the most noticed disfluency of all: “um.”

  5

  A Brief History of “Um”

  New members of the Canadian diplomatic corps play a ball game as they stand in a circle. Whoever holds the ball must talk, unrehearsed, on a random topic. As soon as the person says “uh” or “um,” he or she must throw the ball to someone else. The winner is the one who holds the ball the longest. (The long-running BBC game show, Just a Minute, is based on a similar premise; competitors attempt to speak on randomly assigned topics as long as possible without hesitating, repeating words, or switching topics. “Uh” is one sign of hesitating.) *31 New German foreign service officers play for multilingual stakes. They train to avoid pause fillers in the speeches they give in German (uh, um) as well as in English (“uh,” “um”) and French (euh).

  One could find many other scenes where people are trained to provide smooth, unbroken speech, and where “uh” and “um” are censored, derided, mocked, and erased—as are the people who say them. The attitude of the amateur public-speaking organization, Toastmasters, is not uncommon. Each club assigns a person, the “ah-counter,” to collect a nickel for each time a club member giving a speech says “uh” or “um.” (Most clubs cap the fine at fifty cents per speaker.) As Ralph Smedley, the Toastmaster’s founder, wrote, “Mannerisms are not limited to physical posture. Many enter into speech itself. My favorite aversion is the ‘grunt’—the ‘ah’ and the ‘er-r’ with which many speakers fill in the gaps between their words. It is a bad habit and should be broken by every speaker—even by every conversationalist.”

  Such distaste for “uh” and “um” isn’t merely a judgment about a person’s speech; it’s a deeper judgment about how much control he should have over his self-presentation and his identity. Not surprisingly, the battle against “uh” and “um” figures prominently in our lives with language at school and at work. My friend Sara, who’s now in graduate school in economics, had a second-grade teacher who refused to listen to her talk until she could start a sentence without “um.” It was an unattractive habit in such a gifted girl, the teacher said. Kristin felt a little embarrassed, but she now says that “more than anything I was frustrated—with myself, I think, not the teacher.” She had to pause “for a really long time” to prevent the forbidden word.

  Even among a more open-minded New Agey crowd, “um” isn’t acceptable. Lee Glickstein, a California-based public-speaking and presentation guru and founder of Speaking Circles International, used to tell students not to say “um.” But the admonition wasn’t effective. People kept saying “um.” And the proscription made them even more self-conscious and nervous. “So we don’t talk about ‘ums,’” Glickstein told me. He believes that when you’re fully present with the person you’re speaking with, you say “uh” or “um” much less. You don’t eliminate it entirely—Glickstein said “uh” several times when we were on the phone, though he noticed it immediately and chided himself. If you find yourself saying “um,” he suggested converting it into a slow “yum,” to make yourself more conscious of what you’re saying. “When you start hearing yourself go ‘um,’” Glickstein recommended, “start saying ‘yum,’ and turn your ‘ums’ to ‘yums, um, umm, yummm, mmmm, mmmm.’” By meditating on them, Glickstein said, a speaker should eventually be able to turn “mmmms” to what he calls “rich stillness.”

  Where does this aesthetic of umlessness come from? How did it become so ingrained that it seems to be a biblical commandment? Assuming that nothing short of archaeology was required, I began digging for its origin at the very root of Western ideas about language and eloquence, among the rhetoricians of ancient Greece and Rome. What did they have to say about “uh” and “um”?

  These writers said plenty about good speaking that’s rhythmic and smooth. But they didn’t mention pause fillers at all. In On Rhetoric, Aristotle urges an orator to choose words well, to craft beautiful metaphors, and to use the right rhythm and speak grammatically. In his day, most oratory was extemporaneous, not prepared ahead of time. Around 400 BCE, if one sat long enough in the Athenian marketplace, the Agora, or in the assembly on Pnyx Hill, or in the Theater of Dionysus, one was bound to hear people say “uh” and “um.” Yet Aristotle doesn’t instruct speakers to avoid a hesitant style. Nor does he mention “uh” or “um” (or whatever the ancient Greek equivalent would have been).*32

  Almost four hundred years later, ancient Roman teachers of oratory described their likes and dislikes in terms of boldness and confidence. In De Oratore, one of Cicero’s works of oratorical instruction, Crassus, a character in the dialogue, describes a good speech as intelligible and perspicacious, graceful, and suited to the topic. Those who speak the best, he explains, appear to have lost their sense of shame about speaking. This allows them to begin their speeches without timidity. Not everyone can be an accomplished speaker, Crassus warns. Some individuals are “so hesitating in their speech, so inharmonious in their tone of voice,” he says, that no amount of oratorical training can redeem them from their own incompetence. Cicero himself was known for starting a speech weakly. In a passage that Cicero’s biographer Anthony Everitt takes as Cicero’s description of himself, Crassus says, “Indeed, what I often observe in you I very frequently experience in myself, that I turn pale in the outset of my speech, and feel a tremor through my whole thoughts, as it were, and limbs.”

  Trembling, pale, confused, some of the orators of Athens and Rome stumbled through their speeches—and were noticed and criticized for doing so. Later in De Oratore, another character, Sulpicius, criticizes Antonius’s performance at a trial for starting too slowly. “What timidity was there! What distrust! What a degree of hesitation and slowness of speech!” Sulpicius himself is accused of speaking too slowly, “owing to his genius.” However, he is still “somewhat too redundant.” It’s safe to say that Cicero, reflecting the aesthetic of his time, was attuned to speed and fluidity in oratory. Yet he doesn’t record particular faults. And he doesn’t mention pause fillers.

  Even Quintilian, a Roman rhetorician who stressed the value of the delivery of a speech, seems not to have heard the Latin equivalent of “uh” or “um.” In his oratorical textbook, Institutes of Oratory, he urged orators to practice delivering pleasing-sounding speeches. To Quintilian, “pronouncing well” consisted of a varied tone and an equal rhythm, “so that our speech may not proceed by starts…as a person halts in walking from havi
ng legs of unequal length.” He noticed bad speakers who suck air through their teeth, pant, cough, spit on their audience and speak through their nostrils, and he believed that orators should not be “dull sounding, gross, bawling, hard, stiff, inefficient, thick, or, on the contrary, thin, weak, squeaking, small, soft, effeminate.”

  To Quintilian’s ears, a speaker’s worst fault was to speak in a singing tone. “I know not,” he moaned, “whether it is more to be condemned for its absurdity or for its offensiveness.” Such preferences were Quintilian’s as much as they belonged to the tastes of his era. But he didn’t say a thing about the beauty of the umless.

  It would be a fallacy to think that the ancient Greeks and Romans didn’t say “um” simply because they appear not to have recorded it—the evidence that they heard and wrote it down may have been lost. They likely said “um” for all the reasons that we do: they had to delay when they explained difficult or new things, or they had other demands put on planning what to say. The neural substrate of the brain that leads to the cycles between planning to act and executing an action have been in place for a long time, perhaps millions of years. Meanwhile, the structure of human language has been around for at least one hundred thousand years. Some people may think that “uh” and “um” are creatures of the twentieth century, the products of young sex-addled minds corroded by drugs and hip-hop. More accurately, it’s the twentieth century that made us call “uh” and “um” ugly. (They’ve been there all along.)

  It’s not that the ancients were insensitive to deviant or disfluent speaking—but in their wisdom, they had other reasons for not noting “uh” and “um.” They did notice abnormal talking, particularly if it had a medical cause. According to the historian Ynez Violé O’Neill, the earliest description of speechlessness appears in an Egyptian surgical text, written on papyrus, that dates from 3000 to 2500 BCE. A list of forty-eight case histories, it describes several individuals with head injuries who couldn’t talk.