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

by Erard, Michael.

  Among linguists, Fromkin is best known for her work on slips of the tongue. In 1971, she published a landmark paper of the slip renaissance, “The Non-anomalous Nature of Anomalous Utterances.” It was the first paper to argue—and in grand fashion—why slips (such as “I’m going to buy a new broof keese today” or “take it out to the porch—uh, veranda”) were relevant to linguists. Though Fromkin was a fan of Chomsky, her title was a challenge to his view of errors: if slips fell into discernible patterns, how could they be random? And if they weren’t random, what accounted for the patterns? Here she also presented a model of how speech actually occurred, which she called an “utterance generator.” It showed how a meaning that a speaker wanted to express began with an intended idea and ended with commands that muscles of the mouth move to produce sound. The utterance generator itself is a graphic of boxes, triangles, and arrows that appeared in an academic article. Unlike other models, however, this one was derived entirely from the predictable dynamics that slips of the tongue revealed.*49

  In 1973, she published a collection of classic articles on slips from Freud, Wells, and others, putting her 1971 paper last in the book, suggesting that the science of slips had a venerable history (and perhaps that she was its heir). She sponsored a conference in Vienna on the state of the science and spoke at New College, Oxford University, about Spooner. At scientific meetings, people who lectured after her blundered more than they normally would. “This produced amusement and strange looks directed toward me,” she once wrote, “as if I had special psychic powers manipulating the audience to provide support for my contentions.”

  She also encouraged people to collect slips of the tongue, teaching them how to do it. By 1971, Fromkin already had her own stash of six hundred slips, such as “lumber sparty” for “slumber party,” “smart smoking” instead of “start smoking,” and “moptimal number” for “optimal number.” In a file cabinet in a back office in the UCLA linguistics department is her collection, built over thirty years. One morning, Carson Schütze, a tall, lanky German psycholinguist who became its caretaker when Fromkin died in 2000, showed it to me. I had thought five thousand slips (most in English, but also in French, Italian, German, and Spanish) would have occupied more space. But the collection is remarkably compact, filling four narrow metal trays that look like big recipe boxes.

  Each slip that Fromkin collected is written in pencil on an index card in the scientific phonetic alphabet that linguists use; each card is numbered, and they’re arranged by the date when Fromkin heard them. Schutze pulled out a few cards and fanned them between his fingers. In the late 1980s, this collection was moved to a searchable database and is available on the Web site of the Max Planck Institute for Psycholinguistics.*50 As Meringer had done, Fromkin attached names to some samples. Jimmy Carter is there (“that is expectable”); so are Noam Chomsky (“are part, simply part of his innate intellectual equipment”) and Joe Garagiola (“he knew every crook and nanny”), as well as random stewardesses, gas station attendants, students, district attorneys, cheerleaders, the head of the UCLA medical school, copious colleagues, her husband, and generous appearances of her own blundering self.

  One of her favorite slips was “Rosa always date shranks.” Her friend Anne Cutler, a psycholinguist, called it “perhaps the most famous verb-inflection error of all.” At first glance, English has only one way to mark the past tense of verbs, as the slip “she wash upped the dishes” shows. But they also become the past tense through a vowel change (“sing-sang”). How are the two related? The slip “date shranks” suggests that underlying both “-ed” and the vowel change is a sort of past-tense essence that’s as real as it is mysteriously shy. The speaker of “date shranks” meant to attach this past tense essence onto “date” (to make it “dated”) but blundered and moved it to “shrinks.” Somewhere along the line, “shrinks” as a noun is mistaken for “shrinks,” the verb. And that’s where the past-tense essence floats to the surface, making “shranks.”

  Part of the success of slip science came from the fact that collecting slips gave linguists something to do. Like many who get pleasure from puzzling over very small, oft-overlooked corners of the world, linguists can be socially awkward, especially if they have to spend time among others who don’t share their fascination. Collecting, an all-weather, all-season activity, was (and is) a convenient social mask. “It can give the collector a feeling of doing some useful work while on holiday, at a dinner party, or watching a television interview,” Anne Cutler once wrote. The well-mannered Archibald Hill bemoaned that he was too sensitive to interrupt conversations to collect slips. As a result, he said he never accumulated the extensive collection he longed for.

  Merrill Garrett, another early slip researcher, told me that it is difficult to stop collecting slips once you’re accustomed to listening for them and writing them down. When your ears are attuned to slips, it’s hard not to listen that way all the time. Garrett used to set a time period during which he’d listen assiduously. People remember that he was always flipping open his notebook to write down something someone said. He doesn’t study slips exclusively anymore, but he still listens for them.

  Slip science was also successful because of the valuable insights that scrutinizing a slip produced. One of Garrett’s favorite slips of the tongue is “the skreeky gwease gets the wheel.” This slip has two parts. One is a sound exchange, the other is a word exchange. Which one happens first?

  If you try the sound exchange before the word exchange, you couldn’t get the slip that Garrett heard. You’d get something like, “the sweeky kweel gets the grease.” On the other hand, if you posit the word exchange before the sound exchange, you get the right answer. That is, the word exchange (“wheel” for “grease”) gets you “the squeaky grease gets the wheel.” Then the exchange of sounds between “squeaky” and “grease” gets you “skreeky gwease.” The implication? That words are selected in the brain as soundless frames and placed in slots in the sentence, before the individual sounds of those words are located and imported into those frames. That is, language processing works at a word level before it works at a sound level. Such suppositions were grounded in evidence from slips long before other experimental methods were invented.

  Just as slips in English follow the underlying structure of the English language, slips in other languages fall in line with the rules of those languages. In Arabic, for instance, slips never occur in combinations of sounds that a native speaker couldn’t—or wouldn’t—say. And when an Arabic speaker selects the wrong word, he reaches for a noun when he means to say a noun, and he reaches for a verb when he means to use a verb. These patterns reflect basic properties of human language that are shared among all languages.

  Slips also take a specific language’s shape. In English and other Germanic languages, words are made up out of pieces of meaning that are connected contiguously. The word “destabilizing” is composed of four of the smallest possible units of word meaning (called “morphemes”), “de-,” “stable,” “-ize” and “-ing.” In Arabic, however, the morphemes that make up words do not appear continuously. There’s a root of the word, its stem, which is a sequence of consonants with spaces between them. The stem for all things related to writing is k-t-b. To make nouns, verbs, and other words, you add other morphemes, which appear as a sequence of vowels that slot between the consonants. Thus, to make the verb “to write,” you add the morpheme a-a-a to the morpheme k-t-b. The resulting word is kataba. Other vowel patterns produce other words. “Book” is kitab. “Library” is maktaba.

  Slips of the tongue in Arabic follow this word structure. In some slips two words exchange their consonants, leaving the vowels in place. Here’s one example, which I’ve taken from Hassan Abd-El-Jawad and Issam Abu-Salim, two Arab linguists. They had a Jordanian speaker try to say sakta qalbiyya (heart attack), but it came out as qalba saktiyya (which doesn’t mean anything). In other words, the vowel morphemes (a-a, a-i-a) stayed put, while s-k-t and q-l-b danced arou
nd them. There’s no English analog to this—you might exchange vowels between words, but those vowels aren’t morphemes, as in Arabic.

  Getting words in order in time is also an issue for those using sign language, who can say, “Seymour sliced the knife with the salami” or “a Tanadian from Toronto.” In the 1970s, slips of the hand gave researchers Edward Klima and Ursula Bellugi evidence that signed languages were, in fact, languages and not, as some supposed, mere mimicry or a word-to-word translation of spoken languages.

  As with the features of spoken languages, slips of the hand showed how signs can be broken down into smaller segments that can be combined and repeated in near-infinite sequences. The equivalent of features in sign languages are hand shapes, where the sign is made on the body, and how the sign moves. Klima and Bellugi collected 131 slips of the hand, most of which consisted of an incorrect hand shape. For instance, in ASL one signs “be” with a flat hand slicing out away from the face, but a typical error replaced the full hand with only two fingers in a victory sign. These were “knows better” errors, not made by children or new learners but by otherwise fluent signers of the language.

  Slips of the tongue—from well-known ones like “the skreeky gwease gets the wheel” and “Rosa always date shranks” to thousands of others—became so widely accepted in discussions about language that it is virtually impossible to have an empirical or theoretical discussion about language or speech without addressing Fromkin and mentioning Meringer, even if to dispute in passing the details of their analyses. Willem Levelt’s exhaustive 1989 work, Speaking, discusses speech errors on nearly every page. Using slips, researchers have devised at least seven different models of how speakers find words in memory and say them. Slip science showed how words and sounds are organized in the brain and how they interact; it even opened ways of studying other activities, such as musical performances.*51 A collection of speech errors at MIT contains 6,000 English slips; there are 900 in Dutch at Utrecht University in the Netherlands, 4,000 at the University of Toronto, 7,200 at the University of California, San Diego, 6,000 at Technical University in Braunschweig, Germany, 3,612 at the University of Oviedo in Spain, and 2,400 at the University of Lyon. Another mark of the popularity of slips is in how many languages the phenomenon has been observed. Though slip research began in German and English, linguists have looked at slips in Greek, Latin, Arabic, Croatian, Spanish, Japanese, Dutch, Hindi, Finnish, Korean, Breton, Chinese, Choctaw, Hungarian, Portuguese, Swedish, Thai, Welsh, and American Sign Language.

  While the enthusiasm for slips has never waned, the science of slips eventually began to slow, because of a single problem. In general, we utter more slips than we notice—as nearly as anyone can estimate, people make about one or two slips every 1,000 words. If we assume that someone giving a lecture says about 150 words per minute, this means a lecturer might slip once or twice every seven minutes. (Greater time spans will separate slips in conversations, which unfold more slowly.) However, people report noticing about one slip a week. Moreover, people tend to hear some sorts of slips more than others. In one experiment, Rosa Ferber, a Swiss linguist, found that listeners detected only one-third of slips she gave them; of these, one-half were perceived incorrectly. Another test by the Australians Jan Tent and John Clark showed that people tended to mishear slips involving individual sounds, but they caught ones involving syllables, parts of words, and whole words.

  Was it possible that these discrepancies influenced the science of slips as well? After all, people built theories using slips that they had heard and written down. This implied that celebrated collections of slips of the tongue like Meringer’s and Fromkin’s, built by listening to real-life conversations, contained more of some types of slips than others. Thus, to use Goffman’s terms, they weren’t comprehensive lists of technical faults. Rather, they were incomplete lists of perceptible faults. This has hampered efforts to compare the frequency of exchanges (such as Reverend Spooner’s) to that of other types of sound slips. As a result, most linguists qualify their comparisons or don’t make them at all.

  Listeners can be prone to mishearing certain sounds, even if they’re spoken correctly, what are called “slips of the ear.” People tend to mishear consonants more frequently than vowels, and the misheard sound or syllable is usually unstressed and comes in the middle of a word. For this reason it’s easy for people to mishear “It’s about time Robert May was here” as “It’s about time to drop my brassiere.” One type of slip of the ear has been labeled the “mondegreen,” a term coined in 1954 by Sylvia Wright, an American writer, to describe her childhood mishearing of the line of poetry “and laid him on the green” as “and Lady Mondegreen.” Song lyrics often become mondegreens (I never fail to hear the line from Handel’s Messiah “for we, like sheep, have gone astray” as the more alarming “for we like sheep”), and the bumper sticker “visualize whirled peas” depends on a deliberately misheard “visualize world peace.”

  For linguists, relying on casual slip collecting turned out to be problematic. One could listen to taped conversations, but as Anne Cutler puts it, “It means living your life twice.” In the 1980s, she invented a more efficient method: keep a microcassette recorder on all the time. Then, when you hear a slip, immediately stop the tape and pop in a new tape. With this method, you would always know that the slip occurred close to the end of the recording.

  This method didn’t eliminate some other confounding factors, such as the frequency of a word. Some words (like “some” and “words”) show up fairly frequently in normal English usage; other words (such as “dehisce” and “merisma”) hardly show up at all. Which words more frequently go awry? Joseph Stemberger, a linguist who collected slips from his children for years but now studies slips exclusively in a laboratory, showed that the more infrequent the word, the more prone you are to slip with it. Why? It was basically a matter of practice—the neural connections for more frequently spoken words were more durable.

  The word-frequency problem was solved in the 1970s by Bernard Baars, a cognitive scientist, who devised a way to make people slip in laboratories. Rather than waiting for slips to occur, he would set up experiments with lists of words, types of sounds, and various sorts of distractions. The setup worked like this: a person sat in front of a computer that flashed words (or sometimes sentences) about a second apart. The person was told she was being tested on her memory of the pairs, so she should read them silently. In some cases, a buzzer sounded, prompting the person to read the pair out loud. The person did so correctly—unless some of the preselected pairs biased her toward certain patterns. If the pairs were “bill deal,” “bark dog,” and “bang doll,” then the prediction was that when she reached “darn bore,” she would say “barn door.” This happened 10 to 20 percent of the time. If the pair was “Do I have to put my seatbelt on?” and “Put on your seatbelt,” the person would tend to produce what is known as a “phrasal blend,” such as “Do I have to put on my seatbelt on?”

  In other experiments, the buzzer accompanied other instructions, which produced other types of slips. For instance, asking the subject to reverse the order of words or to repeat a reversed order would lead to spoonerisms, blends, and word substitutions. This method was indeed reliable enough—speakers would slip up about 10 percent of the time—which eliminated the need to wait for people to blunder, thus allowing Baars to test a hypothesis about the origins of slips of the tongue. He called it the “competing plans hypothesis.” The lab situation created slips, he posited, because speakers had to navigate conflicting goals. In some situations the speaker had to be fast—but she also had to be accurate. In other situations, she had to pronounce what was in front of her—yet she had also been prepared to say some sounds, not others. The two mental plans competed for cognitive resources that could be taxed only so far. When the competition resolved itself, a slip was produced. Most often the resolution favored an automatic or habitual process. This explains in part why slips of the tongue hew to the language’s
basic shape. And recalling Spooner, it also explains why transpositions of sounds tend to create actual words—when the words are real, the speaker lets them pass, but when they aren’t real, the speaker notices what he is about to say, then repairs.

  The competing plans hypothesis was a clear step forward. It was also an opportunity to reappraise Sigmund Freud’s ideas about Fehlleistung, or “faulty performances.” Hadn’t he posited that a slip resulted from a repression? Wasn’t a repression simply a competing plan? Baars wondered if the repression was simply another competing goal. To test this formulation, he rigged a situation in which people who admitted to eating problems would have to produce food-related spoonerisms. The speakers were primed with nonsense word pairs, like “rork poast,” “froocy joot,” or “shanilla vake.” To make the experimental subjects even more food-conscious, he placed a bowl of candy on the table. If Freud had been right, Baars hypothesized, the food-sensitive subjects should have made more food-related spoonerisms than people who had no eating issues. That is, if emotionally laden and repressive thoughts were a kind of competing plan, then people would tend to make spoonerisms that communicated this repression. In the end, the correlation between their admitted eating problems and their food-related slips was too weak to establish Freud’s hypothesis incontrovertibly.

  Yet in one sense, Freud was right: a plan to say something could be knocked off course by a rival plan. He’d looked for gaps between what we know and what we actually do. Even Freud’s critic Sebastiano Timpanaro was willing to accede the existence of true Freudian slips that arose because of some intruding emotion—for instance, when a Viennese politician said that people should follow the emperor’s orders spinelessly (rückgratlos), not unreservedly (rückhaltlos). It may be that because Freud minimized alternative explanations for slips, he delayed a more comprehensive consideration of their meaning for some sixty years.