A WOMAN asked another woman in her office if she would like to have lunch. The colleague said no, she was sorry, she had a report to finish. The woman repeated the invitation the next week. Again her colleague declined, saying she had not been feeling well.

The first woman was confused. So she asked her colleague what her refusals meant: Was she really just busy one week and ailing the next, or was she trying to say she simply didn't want to have lunch, so stop asking? The response only confused her more: "Well, um, sure, y' know, I really haven't been feeling well and last week really was difficult with that report which, by the way, was about a very interesting case. It was. . . ."

The woman was frustrated. She couldn't understand why her colleague didn't just say what she meant. But the other woman was frustrated too. She couldn't understand why she was being pushed to say no directly, when she had made perfectly clear that she was not interested in pursuing a friendship.

One woman was expecting directness; to her, indirectness is dishonest. The other was expecting her indirectness to be understood; to her, directness is rude, and being direct would mean being a sort of person that she finds unappealing. Both felt that their own ways of talking were obviously right. Neither realized that both systems can be right or wrong; each works well with other people who operate on the same system, and both fail with people who do not. They instinctively tried to dispel the tension by doing more of the same. Neither thought of adopting the other's system.

Many Americans believe that the only purpose of language is to convey information and that information should be stated outright. But there are many reasons why meaning should not be stated outright, why indirectness is useful and even necessary.

The study of indirectness and other politeness phenomena has received increasing attention in linguistic scholarship. This is a drastic departure from the trend dominant in linguistics in recent decades: formal representation of language not as it is used but as an abstract system. A linguist working in the latter tradition would be concerned with whether a given sentence is grammatical, regardless of whether it might actually be spoken by anyone, let alone how frequently it might be spoken. For linguists concerned with language as it is used in everyday life, sentences that are actually spoken -- and often spoken -- are the ones of interest, not those that are theoretically possible but never encountered. Keeping One's Verbal Distance

Using language to communicate requires balancing two conflicting needs: to be involved with others and to be independent. This duality has been identified by scholars in many fields. Psychologists write of the urge to merge and the urge to independence, and of the complementaryfears of separation and intimacy. Sociologist Emile Durkheim wrote of positive and negative religious rituals (such as prayers and taboos). Later, sociologist Erving Goffman showed that daily life is also a compendium of should-do presentational rituals (greet people, ask after someone's health and family, show concern, and show interest) and should-not-do avoidance rituals (invade another's personal terrain, ask nosy questions, touch too much, remark on embarrassing conditions).

Anthropologist Thomas Kochman, author of "Black and White Styles in Conflict," speaks of the rights of feelings (for example, the right to laugh loudly at a play, talk loudly in public or blast a radio) as compared to the rights of sensibilities (the right not to be disturbed by someone else's laughter, talking or radio).

Linguist Robin Lakoff, author of "Language and Woman's Place," suggests that in deciding what to say and how to say it, people apply different rules of politeness. A distant or deferent style of politeness applies the rules "Don't impose" and "Give options." A camaraderie style of politeness applies the rule "Be friendly." For example, in answer to an offer "Would you like to stay for lunch?" a distant response would be, "No, thank you, I just ate." A deferent response would be, "I don't want to put you to any trouble." And a camaraderie-motivated response would be, "Thank you, I'd love to." An even stronger dose of camaraderie might entail volunteering, "I'm starving| Have you got anything to eat for lunch?"

Penelope Brown and Stephen Levinson, an anthropologist and a linguist, use the terms positive politeness (for showing involvement) and negative politeness (for not imposing).

All of these systems for understanding human behavior reflect the universal human needs to be involved with each other and yet not to become engulfed or overwhelmed by involvement. Indirectness is a universal device for expressing ideas, opinions and desires -- that is, showing involvement -- in a way that does not impose on or offend others.

Furthermore, there is pleasure in being understood without stating explicity what one means. Everyone wants to get an appropriate birthday gift. But few come out and say what they want, because that would defeat its purpose: to show that the giver knows one well enough to choose an appropriate gift and cares enough to spend time getting it.

Differences in directness are a major source of confusion and dissatisfaction in communication. At weekly staff meetings, the director of a counseling agency never issued peremptory orders; decisions were reached after all staff members had expressed their opinions. Yet more often than not, the decisions reached were those the director thought best. One staff psychologist thought the director manipulative; if she knew what she thought best, she should just tell them so directly. But others appreciated the chance to express their opinions. They felt they had been part of the decision-making process, and if they happened to decide on a course the director preferred, they did so with an understanding of her reasons.

A real-estate appraiser complained to a colleague about a client who had called to say that she was leaving for vacation. His colleague knew immediately why the client had called: She was letting him know, indirectly, that she was impatient to receive her appraisal. The vacation provided an excuse to remind him.

The appraiser did not understand the indirect approach and didn't realize that the client wanted reassurance that her appraisal would be ready by the time she returned. He preferred the client who called and said, "Where the hell is my appraisal?" On the other hand, his less-direct colleague would have been shaken by such a call -- perceiving it not as direct but as nasty -- and therefore could not have assured the client that all was well.

A Greek woman explained that when she was growing up, she had to ask her father's permission for everything. If she asked if she could go to a dance and he said, "If you want, you can go," she knew that she should not go. If he really thought it was a good idea, he would say, "Yes. That's a good idea. Go." He never said no. But she understood by the way he said yes whether or not he meant it.

This sounds to many Americans like hyprocrisy: He got her to do what he wanted without stating it directly. But indirectness could have advantages for both of them. The father could feel that his daughter did the proper thing of her own free will rather than simply obeying. The daughter could feel that she was choosing to please her father rather than following orders. And if she did go against his wishes, they would both save face: He did not go on record as forbidding her, so her going would not be openly defiant. "Hypocrisy|" "Dishonesty|"

The indirect system causes misunderstandings when it is not shared. If an American cousin who spoke Greek visited the family, she might take her uncle's hedged approval literally. Then, if she went to the party, he would be angry at her for going and she would be angry at him for his inconsistency. An indirect message is crystal clear to those who know the system but opaque to those who don't.

American businessmen have similar problems communicating with their Japanese counterparts. An American journalist at a trade fair in Japan asked if he could see a particular robot. His Japanese host answered, "That might be possible." To the American, this meant "yes," so he later insisted that he had been told he could see the robot. To the Japanese, "No" is too face-threatening to be used. "Maybe" means "no," and only an unqualified "yes" means "yes." The Japanese host felt he had made his refusal clear and could not understand the American's dishonesty in claiming to have been misled.

Since all speakers tend to take their own system of communication as self-evident, talking with someone who operates on a different system frequently results in mutual accusations of dishonesty (not meaning what was obviously said) and hypocrisy (not saying what was obviously meant.

Many misunderstandings are caused by unstated assumptions. For example, a telephone conversation made less and less sense until it emerged that one party assumed erroneously that the other was calling from home. The confusion could have been prevented by the caller stating where he was. But it would be absurd for all callers to begin by announcing their location. Every utterance is based on innumerable assumptions that cause problems only when they are not shared, and no one can predict which of all their assumptions will turn out not to be shared.

Understanding how language is used is the focus of two sub-fields of linguistics, discourse analysis and sociolinguistics. Analyzing language as it is used in communication immediately points up the pervasiveness and necessity of indirectness. Among the many reasons:

Deciding to tell the truth leaves open the question of which aspects of the truth to tell. For example, everyone resents being told the obvious; it seems to imply criticism or condescension. However, what is obvious to one person is not obvious to another; it may even be unimagined.

Social requirements are real: Stating the truth in no uncertain terms may hurt the feelings of others. For example, a woman called a friend and backed out of a dinner engagement, saying she was tired. The friend did not doubt that this was true; but she was hurt because simply being tired was so slight a reason to let her down that it implied small regard for the friendship. Had the caller invented a better excuse, such as having gotten ill, she would have accomplished her goal without implying carelessness about the friendship.

A difference of opinion stated directly is more difficult to rescind than one that has only been hinted. In fact, one may not be sure what one wants or thinks until one has a sense of what the other wants or thinks. This need not be seen as lack of conviction. It may simply be that one has a slight but not a strong preference.

Conversational style -- including joking, irony and figures of speech -- is a basic part of language and provides creativity, pleasure and the basis for a sense of community and shared style. Stylized language is open to misunderstanding because it does not state meaning directly, but being explicit would defeat the social purpose of language and rob individuals of the means to express their personalities. Language as We Live It

Ignoring the social and psychological functions of language is at the heart of most demands for directness, and also of those scholarly approaches which treat language solely as a grammatical system. Neurologist and essayist Oliver Sacks, author of "The Man Who Mistook His Wife for a Hat," wrote a recent article in The New York Review of Books about Tourette's, a neurological syndrome that causes multiple convulsive tics. Sacks noted that advances in modern medicine have resulted in "a real gain of knowledge but a real loss of understanding" because neuroanatomy "became compartmentalize . . . seeing the motor, the intellectual and the affective in quite separate and noncommunicating compartments of the brain." The results were "efforts. . . to 'physicalize' or 'mentalize' {the syndrome}, to make it one or the other, when it is so manifestly both. . . . By the turn of the century a split had occurred, into a soulless neurology and a bodiless psychology, and with this any full understanding of Tourette's disappeared. . . . What Tourette's is really like -- this has been forgotten, and we can only recapture it if we listen minutely to our patients, and observe them, everything about them, with a comprehensive eye."

The developments in linguistics discussed here parallel Sacks' account of neuroanatomy. The compartmentalization he describes is analogous to modern linguistics' separation of language into autonomous parts: phonology (the sounds), morphology (the bits that make up words) and syntax (the sequence in which words are strung together in sentences).The field of sociolinguistics arose to bridge the gap between a sociology that ignored the structure of langauge and a structural linguistics that ignored the social and psychological forces at play when people use language.

Sacks' concern with describing "what Tourette's is really like" parallels the concern of many in linguistics today with describing what language "is really like" -- not only as a grammatical system, but also as a part of people's lives. Even the method he recommends -- listening minutely to how people talk, making use of videotaping and slow motion playback -- parallels methods being used by linguists trying to understand the "full character, connection and meaning" of language.

Sacks calls for "a neurology of living experience." The approach to linguistics I have been describing amounts to a "linguistics of living language," reflecting the reality of our experience using language: That we often can't say what we mean. Clarity vs. Color

ONE DANGER of indirectness is lack of understanding. An indirect person may assume that meaning is clear when it is not.

One person said to another, "What kind of salad dressing should I make?" The other answered, "Oil and vinegar, what else?" This was meant ironically: "Oh, you know me. I'm unimaginative. I always make oil and vinegar. So don't pay attention to me. Make whatever you like."

In this case, the irony was missed. "Oil and vinegar, what else?" was heard as a demand. And furthermore, it sounded like an implied criticism: "You should have known."

But while irony is always open to misunderstanding, banishing it and other forms of indirectness would rob speech of most of its creativity, character and expressive potential.

Deborah Tannen is associate professor of lingustics at Georgetown University and author of "That's Not What I Meant|"