Why Conversations Go Wrong: Tannen and Levine

Two old friends are walking around a lake. One is talking through a worry. A duck paddles by with a line of ducklings behind her. The other friend points and says, "Oh, look at that duck. Isn't that charming?" Her friend stops. "You're not listening to me," she says. "Of course I'm listening to you," comes the reply. "Why would you think I'm not listening to you?" "Well, why are you talking about the ducks?" (17:08). Nobody in that exchange said anything false. Nobody meant any harm. And yet something real had gone wrong.

That small, ordinary rupture is the subject of this episode of Hidden Brain, built around two researchers who have spent careers studying the gap between what people say and what people hear. Deborah Tannen, a linguist at Georgetown University, has spent decades showing that most misunderstandings are not about bad intentions. They are about mismatched styles: differences in pace, directness, and timing that we never notice until they collide. Emma Levine, a psychologist at the University of Chicago's Booth School of Business, studies a related gap: the space between the literal truth and the lies people tell to protect each other. Put together, their work suggests that conversation is less like a transparent pipe carrying meaning from one head to another and more like a negotiation over things nobody agreed to negotiate.

Tannen's method for studying this began, fittingly, with a family dinner. One Thanksgiving, she recorded a three-hour conversation among friends: two Californians, a woman from the UK, and three New Yorkers, herself included. It took her two and a half months to transcribe it, timing every pause with a stopwatch (09:10). What she found was that the Californians and the British guest could barely get a word in. It was not that the New Yorkers were rude. It was that different groups carry different internal clocks for how long a pause should last before it is someone else's turn to speak (08:10). The New Yorkers' sense of a normal gap was shorter, so when a Californian was still gathering thoughts, a New Yorker heard silence and jumped in to fill it.

Something similar happened with overlapping speech. When two New Yorkers talked over each other, Tannen realized, it was often a sign of engagement, not interruption. "Well, sometimes we talk along as a way of showing enthusiasm," she says. "We're so interested, we don't wait for you to stop" (10:29). But if you believe only one voice should go at a time, that same overlap reads as rudeness. One of the New Yorkers at that dinner tried seven times before he got a turn to make his point; a Californian who was ignored once would simply give up (11:57). Both reactions felt natural to the person having them. Neither person thought they were doing anything wrong.

Tannen calls the payoff of shared style a kind of relief. "A perfectly tuned conversation is the vision of sanity, a reassurance that you're a right sort of person and all's right with the world," she says (15:04). When style clashes, the opposite happens: people do not think, "our pause lengths differ." They think, "you don't respect me," or "you only want to hear yourself talk."

Indirectness works the same way. Tannen's own marriage, to a Greek husband, ran into it constantly. She would ask if he wanted to go to a party; he'd say "okay," which to her was not an answer. To him, her follow-up question, "are you sure you want to go," read as evidence she couldn't make up her mind (05:59). She later tested this with Greek respondents, who mostly agreed with her husband's read: a wife wouldn't ask twice unless she was unhappy with the plan. Tannen calls this the birthday present logic. You could simply tell someone what gift to buy you, and you would get the object you wanted. But you would lose the other thing you wanted: proof that the person loves you enough to know you without being told (23:28).

This is where Tannen introduces a term she borrows from the anthropologist Gregory Bateson: the meta-message. The message is the plain meaning of the words. The meta-message is what saying those words, in that tone, in that context, says about the relationship (26:22). A colleague once greeted an older male colleague on a cold day with, "Hi, where's your coat?" He shot back, "Thanks, Mom." She meant warmth. He heard condescension (26:57). Tannen argues that most domestic arguments are fights about the meta-message dressed up as fights about the message. Couples argue about whether he should have come home fifteen minutes ago; they are really arguing about whether she felt cared for.

Some of the sharpest meta-message flashpoints, Tannen has found, sit between mothers and adult daughters, clustered around what she calls the big three: hair, clothes, and weight (46:41). A mother's comment about a daughter's haircut is, to the mother, an act of love: she is the one person who will tell you the truth because she cares the most. To the daughter, the same comment lands as proof she is flawed in the eyes of the person whose opinion matters most. "Suggestions for improvement really are implied criticism and are also a sign of caring," Tannen says (47:42). Both things are true at once, which is exactly why the conflict is so hard to resolve by simply being right.

Tannen's proposed fix is not a script. It is a shift in default assumption. "If you frame everything as a fight, it's corrosive to the human spirit. Everyone feels less safe and more vulnerable," she says (54:15). Her suggestion is to treat most misunderstandings as accidents of style rather than acts of malice, on the theory that you will be right more often than you think, and that even when you are wrong, assuming good faith costs you very little.

The Lies That Hold Relationships Together

If Tannen studies the messages we send without meaning to, Emma Levine studies the messages we withhold on purpose. Levine's research asks a blunt question: when do people think lying is not just acceptable but right? Her answer complicates the tidy rule that honesty is always the best policy. People endorse deception most, she finds, when the truth cannot lead to any improvement or long-term benefit for the person hearing it (60:30). Telling a friend her three-year novel manuscript has serious structural problems might be honest, but if there is nothing left to fix and no decision left to make, some listeners think a kind word does more good than an accurate one.

Levine calls these pro-social lies: lies intended to help, not to exploit. A family-medicine physician told the show about a dementia patient who smoked. Rather than repeatedly explain why he could no longer smoke in memory care, staff and family simply told him he had decided to quit. He accepted it, stopped asking within a couple of months, and came off his emphysema medication (63:20). It worked, by the measure of his comfort. But Levine is careful not to treat this as a template. "Paternalistic lies tend to backfire even if they're told with pro-social intentions," she says, because taking away someone's ability to choose for themselves, even gently, can breed resentment (66:00). The dementia case felt more acceptable to listeners partly because the patient's autonomy was already limited by his illness, not because the lie itself was categorically fine.

Levine also studies the middle ground people think they have found: saying nothing at all. A listener wrote in to ask whether omission counts as lying if you never state an untruth. Levine's research suggests the speaker and the listener disagree sharply here. The person staying quiet about an ugly baby photo feels they have found a moral compromise, better than lying, kinder than blunt honesty (79:41). But the listener usually sees through it, and the silence denies them two things at once: the truth, and the comfort of a warm answer. "It's actually not a compromise," Levine says. "Sometimes it's the worst of both" (80:30).

The clearest warning in the episode concerns lies that outlive their original purpose. One listener, Mia, described her family's decision to hide her brother's terminal cancer diagnosis, and later his death, from her aging mother, at his request. For five months, relatives invented cover stories: he was sick with a germ, he was undergoing tests, he had traveled to South Korea for treatment. The mother sensed something was wrong and quietly stopped asking (93:46). Levine's read is not that the initial choice was unreasonable. It is that protective lies are judged kindly when they are temporary, and judged harshly the longer they run, because withheld truth compounds: it blocks grieving, blocks goodbye, and erodes trust between the people keeping the secret and the person being protected from it (102:16). A second listener, Kate, hid a sexual assault from her grandmother for five years to spare her the pain of knowing. The lie worked in the narrow sense that her grandmother was never burdened with the truth. But Kate says it cost her the ability to ever again be fully honest with the person she loved (101:02).

Levine's broader point connects back to Tannen's, even though the two researchers are working on different problems. Both are describing conversation as a place where the literal content of what is said is often not the point. Tannen's speakers misjudge each other's pauses and think they have learned something about character. Levine's liars withhold a fact to protect a feeling and end up damaging the very closeness they were trying to preserve. "Difficult conversations are the ones that both cause pain and promote long-term benefits like intimacy, growth, understanding," Levine says (105:28). The instinct to avoid that pain, whether through a mismatched pause or a merciful lie, is understandable. The research from both scholars suggests the avoidance usually costs more than the discomfort it was meant to prevent.