UC Studies : Structure Discovered in Chitchat

Imagine yourself walking down the street. You see a friend and call to him, but he apparently does not hear you. So you call out again. This time, he turns around and says, “Oh, hi. I didn’t hear you call the first time.”

If you thought about it, you might wonder how he knew you had hailed him more than once if he didn’t hear you in the first place. Had he made a lucky guess? Was he lying to you?

The answer, according to Emanuel A. Schegloff, a 49-year-old sociologist at UCLA, is neither of these. It is quite likely, Schegloff says, that your friend did not hear you the first time, but it is equally likely that he knew there had been a first time. Although he might not have been able to tell how he knew, something in your voice--its pitch, its volume, the pacing of your words or their placement--would have let him know that what he heard was not the first thing that had been said to him.

Real Conversations

Such changes in tone and inflection, the patterns of words and their timing are the subject of a relatively young scholarly discipline known as conversation analysis.

Founded by Schegloff and a colleague at the University of California about two decades ago, the field has focused on a subject long neglected by scholars: the talk that actually occurs in real conversations.

Linguists and rhetoricians have long theorized about the formal structure and use of language, but until the advent of conversation analysis, its practitioners contend, there was no systematic attempt to understand how people actually speak to one another--the elaborate exchanges of information and feelings, the give and take, the starts and stops, the often unintelligible utterances that make up ordinary conversation.

No Random Chitchat

In a soon-to-be published book, Schegloff explains some of the discipline’s early findings this way:

Conversation, even chitchat, is not as random or disorderly as it may seem. In most respects, it is a highly structured enterprise that relies on a complex and almost inviolate set of rules and highly regularized practices.

When people talk to one another, they ordinarily talk one at a time, one after the other. When their talk becomes disorganized--that is, when everyone tries to speak at once or when no one speaks at all--there is an immediate effort to restore “order.” Someone stops talking. Someone steps in to fill the silence.

Sometimes the order goes awry--a silence may continue for too long, or the simultaneous talk may not cease quickly enough. When this happens, the conversation takes on a special character. People begin to fidget; they look at the floor; they shift in their seats. Or they become competitive. Their voices become louder and higher pitched. These, Schegloff says, are “special states” of conversation that are invariably and quickly resolved in favor of “normality”: One speaker at a time, no more, no less.

These observations may hardly seem startling. Parents have long taught their children not to interrupt, not to talk too much and to answer questions when asked.

But how do people actually know what is too much? How do they know when their turn is up? What happens when people speak out of turn? Or when people fail to speak at all?

“To the extent that anyone thought about it, the assumption before was that talk was very personal, very idiosyncratic,” said Deirdre Boden, one of the organizers of a conference entitled “Talk and Social Structure” held recently at the University of California, Santa Barbara.

Now, Boden said, even expressions like “uh” and “well,” once thought to be individual errors, are no longer viewed as mistakes but as intentional expressions or manifestations of the rules of ordinary conversation.

“This is the kind of thing we have found out,” explained Boden, a recent graduate of UC Santa Barbara and a professor at Washington University in St. Louis. Referring to one of Schegloff’s as-yet-unpublished observations about the nature of telephone conversations, she said, “When somebody calls on the phone and says, ‘I’m just calling to say hi,’ there may be a few mundane exchanges. But then comes the real reason for the phone call. The caller himself may not even know what it is, but you can always tell when it’s coming because it’s invariably preceded by ‘ahhh.’ ”

Rules Are Not Taught

“What is so remarkable,” she added, “is that people follow these rules without ever being taught them--indeed, without even knowing . . . that they exist.”

According to a group of conversation analysts at the University of Texas, the rules for initiating, carrying out and extracting oneself from a conversation are not limited to English-speakers. “Virtually the same rules” seem to be followed by members of all societies, including “such far-flung language communities as Lue, Mayan, Tzoltzil and Tamil,” the Texas researchers wrote in a recent essay.

In addition to maintaining order, Schegloff has found, an important function of conversational rules is avoiding unpleasantness.

“That is what the introduction of ‘ah’ or ‘well’ is all about,” Schegloff said. “Think about what you do when you ask someone to do something for you or to go someplace with you. If you are afraid the answer may be no, you will fill your conversations with ‘ahhs’ and ‘hmms.’ You may stutter. You may even restart your sentences several times over before actually launching into what it is you really want to say.” What the speaker is really trying to do, he said, is avoid eliciting a response he does not want to hear.

Conversation analysis had its origins at UCLA. Although it has since spread to a few European universities and a handful of other U.S. institutions, it remains concentrated at the University of California, particularly at UCLA and the campuses at San Diego, Santa Barbara and Santa Cruz. The subject, according to Margaret L. Weeks, spokeswoman for UC Santa Barbara, “might be considered Southern California’s contribution to sociology.”

Mundane Details

Scholars in the field say the formal analysis of conversation grew out of ethnomethodology, a small and relatively esoteric area of sociology created in the early 1960s by a UCLA scholar named Harold Garfinkel. Rather than focusing on large, abstract assumptions about how societies work, Garfinkel’s approach to studying human behavior was to observe and record the mundane details of how people actually behave in particular situations--whether it is buying clothes or carrying out scientific experiments.

As Garfinkel was developing his theories, a young graduate student at UC Berkeley named Harvey Sacks learned of them and began applying the Garfinkel approach to a study of the behavior of people on telephone hot lines at a suicide prevention center in Los Angeles. As he worked on his doctoral thesis, Sacks began brainstorming about Garfinkel’s ideas with Schegloff, a fellow graduate student at Berkeley.

Focusing ever more closely on what his would-be suicide victims and counselors were actually doing on the hot line, rather than what sociologists had theorized they would do, Sacks became intrigued with the process of the talk itself. What made the conversations work? What caused them to fail? What rules were being followed? How did people know what the rules were?

Schegloff also became fascinated with the structure rather than the substance of phone conversations, so the two young scholars began a scrutiny of hundreds of hours of recorded conversations, ranging from emergency calls to the police to group counseling sessions.

Elementary Conclusions

Their work led them to reach certain elementary conclusions about how conversations work. They found that conversations invariably proceed in turns and that turns fall into pairs--that is, certain remarks require certain responses. When there is a failure or a confusion in the structure, when the unspoken rules of conversation are not followed and the system breaks down, there is an immediate attempt to repair it.

Sacks, who got his degree from Berkeley in 1965, continued his research at UC Irvine until his death in an auto accident in 1975. Before he died, he and Schegloff, who went on to teach at UCLA, had set up what one scholar called “the basic conceptual scaffolding for subsequent developments in the field.” They also had begun to attract a handful of other scholars to their enterprise.

Today, the field is still small--probably fewer than 30 full-time scholars--and a basic textbook has yet to be written on it.

What exist are hundreds and perhaps thousands of research papers on a variety of topics, many of them esoteric and a few of them surprising.

For example, Sacks found in his studies that even the telling of jokes and stories is dictated by the rules of conversation. In a series of lectures delivered in 1971 but published only recently, Sacks said that jokes, even lewd ones, are told with far greater frequency than stories, even though the two are similar in purpose (entertainment, primarily) and format (description of an event followed by a punch line).

Joke Need Not Fit

It’s not that there are fewer stories to be told or that they are inherently less entertaining than jokes, Sacks said. The difference in frequency occurs because a story “needs to be fitted into” a conversation, while a joke does not.

In other words, a story needs to be introduced with a phrase like, “That reminds me. . .” A joke, however, can be told by anybody at any time. A momentary lull is all that is required in most cases. “If nobody is talking, one can say ‘I gotta joke,’ tell the joke, and it has no bearing on the conversation so far or thereafter,” Sacks said.

In studying jokes and other forms of conversation, conversation analysts do not rely on “clean” transcripts, the kind a writer might create for a TV script or that a court reporter might reproduce from a trial. As they study a conversation--or, often, only a fragment of a conversation--they try to note every detail, which often means listening to a recording hundreds and even thousands of times.

To catch every nuance, every sound, every hesitation, an elaborate system of transcription had to be developed. No order of detail was thought to be insignificant or irrelevant. A pause of any more than about one-tenth of a second, for example, appears to be quite significant in most conversations and is therefore noted, as are points of laughter and their duration, stutters, mispronunciations and simultaneous utterances.

The system of transcription that is used by most conversation analysts, although it has been adapted and altered by individual researchers, was developed in the 1960s by Gail Jefferson, who began her work while an undergraduate at UCLA.

‘Much More Meaning’

“What we think we have heard is quite often not, in fact, what has been said. And, conversely, what has been said often conveys much more meaning than what we think we heard,” said Jefferson, an independent researcher in England.

Because of the enormous attention to detail and the ultimate complexity of their work, many conversation analysts in the early years of their research confined themselves to telephone conversations. That way, they did not have to take into account nonverbal exchanges and physical movements that are clearly an important part of face-to-face encounters.

As the discipline has become more sophisticated and as video equipment has become more affordable, however, many researchers have begun to record and study other conversations--classroom discussions between students and teachers, interviews between politicians and reporters, exchanges between mothers and daughters in shopping centers and encounters between parents and children around the dinner table.

One recent study, which has been particularly controversial, focused on the differences in conversation patterns between men and women.

It is often said that “when a woman talks, one can’t get a word in edgewise,” but the fact is that when men talk, women find it quite difficult to participate in the conversation on an equal footing, according to the authors of the study, Don H. Zimmerman, a professor at UC Santa Barbara, and Candace West, an associate professor at UC Santa Cruz.

Men Interrupt

Zimmerman and West reached their conclusions after analyses of many hours of conversations between male and female students in controlled laboratory settings. In the conversations they studied, males were responsible for 75% of the interruptions that occurred within what were, apart from the laboratory setting, normal conversations.

Although it was quite obvious why such research might provoke controversy outside the discipline, the reason it is causing turmoil within the field is more difficult to understand but is directly tied to scholars’ perceptions of both the purposes of conversation analysis and its status as an emerging discipline.

Schegloff, who continues to be considered one of the dominant forces in the field, strenuously objects to his colleagues’ efforts to use sex, occupation or status as a way to explain why conversations work the way they do. Who the participants are or where they are, he argues, is often irrelevant to the kind of close analysis of what he believes should be the basic focus of conversation analysis: the starts and stops, overlapping speech and turn-taking.

“If, for example, we say that one speaker is interrupting another speaker just because he is male and she is female, then we may miss the fine details of what happened in the conversation that may have caused the interruption to occur.”

Like scholars in other fields who want to keep their discipline unsullied by practical concerns that may distort fundamental findings, Schegloff questions whether the discipline is ready for as much applied work as some of his colleagues would like. “There is still a great deal of basic information about talk that we simply do not yet understand,” he said.

Practical Applications

Nonetheless, increasing numbers of well-regarded conversation analysts seem to be taking on questions with obvious practical applications.

Lucy Suchman is a researcher who is applying some of the fundamental discoveries of conversational analysis to business. Working at Xerox Corp., Suchman is trying to make machines “understand” enough basic conversational patterns so that they can respond to the human voice.

Turning the art of old-fashioned rhetoric into a science, a number of conversation analysts have demonstrated precisely how politicians manipulate the rules of conversation to cause audiences to applaud at specific moments. Other researchers have shown scientifically how such manipulation can cause even experienced news reporters to ask certain kinds of questions of politicians and steer clear of others.

What is behind the concerns of many of these researchers, they say, is an important and intriguing question: Who dominates talk and holds power in a conversation.

Laughter, for example, has been a recent subject of study in this regard and is proving to be surprisingly significant in determining who is the dominant partner in any given exchange, Boden, the Washington University researcher, noted.

Laughter is a “shared experience” in a conversation, in that when a speaker says something humorous, he in effect invites the other person to laugh. But when that other person declines to laugh, he has wrested power from the person who has tried to initiate the laughter, Boden said.

Doctors Take Control

An illustration can be found in conversations between doctors and patients, UC Santa Cruz’s Candace West says in a recently published book. One of her findings was that patients frequently say things to initiate laughter--perhaps, she theorizes, as a way to put themselves on an equal footing with doctors. Doctors are inclined not to laugh, she said. It is not because they do not have senses of humor, she theorizes, but because they are trying to distance themselves from their patients and thus maintain their dominant position in the relationship.

Perhaps the most vivid use--or misuse--of power in conversations was evident in the interrogations of alleged Communists by the House Committee on Un-American Activities in the 1950s, according to a study written jointly by Boden while she was a postdoctoral fellow at Stanford and by Harvey L. Molotch, chairman of UC Santa Barbara’s sociology department.

The people accused of being Communists often took refuge in the Fifth Amendment simply because of the structure of the question “Are you now or have you ever been a member of the Communist Party?” The answer many of the alleged Communists wanted to give was “Yes, but. . .” and then explain the circumstances, Boden said. But because the questions were persistently--and intentionally--framed in such a way as to elicit yes or no answers, the accused were left with only one option: to say nothing, Boden said.

Used by Trial Lawyers

So convincing and potentially useful are assessments such as these that in recent years conversation analysts have come to be in demand by trial lawyers who are eager for expert assistance in directing the questions of witnesses and in making certain that their own clients can talk around or through the questions of opposing counsel.

While the debate continues within the discipline about the relative merits of applied research versus more fundamental inquiries, most conversation analysts are not at all eager to defend their work outside the discipline on the basis of its relevancy to modern-day problems.

“The truth is, no area of sociology is going to build a better mousetrap,” Zimmerman said in a recent interview. “It’s interesting to note that these questions (of relevance) just aren’t asked of the physical sciences, at least not nearly as much. We somehow ‘know’ that what physicists or chemists do is important, even if we don’t understand it. Somehow, they have won the right to ask fundamental questions--to do basic research--unmolested in a way that the social sciences have not.

“Yet,” Zimmerman said, “I believe that fundamental knowledge about how the world works--and how people work--is good. . . . And certainly there is nothing more fundamental than how it is that we talk to one another.”