Talking, feeling, staying together

Twenty thousand words

In 2006 a best-selling book, Louann Brizendine’s The Female Brain, told its readers that a woman uses about 20,000 words a day and a man about 7,000. The number spread everywhere. But the linguist Mark Liberman, writing on the blog Language Log that August, went looking for the study behind it and found none. What he found instead was a trail of earlier figures, each different: 25,000 against 12,000 in a marriage-counseling pamphlet from 1993, 50,000 against 25,000 in a 1993 self-help book, 20,000 to 24,000 against 7,000 to 10,000 in a 2004 television interview, and a joke website. None of them said that anyone had counted. The author later said the figure had come from unreliable secondary sources and asked her publisher to take it out.

Someone did count. A team led by the psychologist Matthias Mehl at the University of Arizona gave 396 university students in the United States and Mexico a small recorder, clipped to a belt or kept in a pocket, that switched on for 30 seconds every 12.5 minutes for several days. The wearers never knew when it was on, and the recorder caught under 5 percent of their day. Transcribers counted the words, and the totals were scaled up to the waking day. The result, published in Science in 2007, was that women spoke on average 16,215 words a day and men 15,669. The difference, 546 words, was not statistically significant. Individual people varied enormously, from fewer than 700 words to more than 47,000.

Fair is fair: the study was small, and all its volunteers were students. In 2025 Mehl’s group and others reported a much larger test, with 2,197 people in 22 samples from four countries (the United States, Switzerland, Serbia, and Australia), ages 10 to 94, and 631,030 recordings. Women spoke 13,349 words a day and men 11,950: a small difference of 1,073 words (d = 0.13). It depended on age. It was about 500 words among teenagers, about 3,300 among people aged 25 to 64, and reversed, with men a little ahead, among people over 65. A 2007 meta-analysis of the older literature by Campbell Leaper and Melanie Ayres found, if anything, that men talked slightly more (d = −0.14, men), with differences that were “statistically significant but negligible.” Try the numbers below.

Two languages?

If talk does not differ much in amount, perhaps it differs in kind. In 1990 the linguist Deborah Tannen argued in You Just Don’t Understand, a book that spent years on the best-seller lists, that men and women grow up in different “cultures” of talk: women use it to build rapport, men to report information and establish status, and so they misread each other. The claim fits many people’s experience. A husband hears “I’m fine” and takes it at face value; a wife says “go ahead” and means the opposite. Tests of it have been less kind. A review of the evidence found problems in how Tannen had used the research, and a study with a title that says it all, “The myth of gender cultures,” found that similarities outweighed differences. Interruptions show the typical size: across 53 studies men interrupted slightly more, by 0.15 standard deviations, and by 0.33 for the intrusive kind.

What about “men do things side by side, women talk face to face”? A meta-analysis of 37 studies of friendship by Judith Hall found that women expect more closeness and self-disclosure from friends (d = 0.39), but expect about as much shared activity as men do (0.03), while men put more weight on status and fitness (by 0.34 the other way). The “go ahead” problem is real but not male or female: it is indirect speech, in which a meaning is carried by tone and context. All people use it, and some cultures use it more. In one study Koreans were more likely than Americans to speak indirectly, and people who used indirect speech were quicker to read it; in another, a person’s own sense of self predicted how direct they were better than their nationality did. The lesson for a marriage is the same in both languages: when a meaning matters, say it out loud, or write it down.

Reading faces

Are women better at reading other people, the quality sometimes called women’s intuition? It depends, once more, on how you ask. When people are asked how empathic they are, the difference is large, favoring women: in an online survey of more than 630,000 volunteers, the gap on the empathy questionnaire of Simon Baron-Cohen was 0.39. Yet when the same people are tested on how well they read others, or are measured through the body, it nearly vanishes. A 1983 review by Nancy Eisenberg and Randy Lennon found a large difference on questionnaires and none on physiological or unobtrusive measures. On tests of reading emotion from the eyes, the advantage for women is about 0.2 in a large study of 89,553 people; across 215 samples on recognizing emotion in faces and voices it is 0.19. Those are small, with a woman ahead in about 55 or 56 pairs out of 100.

Part of the difference seems to be about motivation and role. In a 2001 study by Kristi Klein and Sara Hodges, women were better at inferring a stranger’s thoughts, but only after a sympathy task; when participants were paid for accuracy, both sexes did better and the gap disappeared. In a boss-and-worker experiment by the psychologist Sarah Snodgrass, subordinates, whatever their sex, were more sensitive to how the leader felt about them than leaders were to their staff, which suggests that “women’s intuition” may partly be the intuition of the less powerful. Not all studies agree: another found that men’s accuracy never overtook women’s even when the task was framed in men’s favor. The popular reason offered in some accounts, that women have a wider field of vision, we could not verify at all.

The strongest claim of this kind comes from Simon Baron-Cohen, who proposes that on average women are more strongly drawn to empathizing and men to systemizing, to understanding rule-governed systems. In his group’s 2018 survey of 671,606 volunteers on a website, the gaps were 0.39 for empathy (women higher) and 0.31 for systemizing (men higher). But only about 40 percent of the women and 24 percent of the men fit the “empathizing” type, and 40 percent of men and 26 percent of women the “systemizing” type: most people, of both sexes, did not fit the pattern. Critics point out that the volunteers chose themselves, the measures were self-reports, and the popular test of “reading the mind in the eyes” may measure vocabulary and emotion recognition more than empathy.

Sadness and memory

Women are about twice as likely as men to be diagnosed with depression: across 65 studies and 90 countries the odds ratio was 1.95. The gap opens at about age 12, peaks at ages 13 to 15, then narrows. Women also ruminate a little more (a gap of 0.24). In one early PET study, 10 men and 10 women recalled sad events, and the women’s brains showed activity over a wider part of the limbic system. In 1997 a small PET study of 14 people reported that men made serotonin 52 percent faster, which was widely retold as an explanation. A second study with the same tracer a year later, of 5 men and 5 women, found women’s values 10 to 20 percent higher, and studies of the serotonin transporter and receptors disagree with one another. The evidence is weak and conflicting, so we would not state the 52 percent as fact.

The same goes for the idea that a wife remembers every wrong. There is evidence that women recall emotional material and personal events in more detail. In an experiment where married couples taped their descriptions of their first date, a vacation, and an argument, wives gave more vivid accounts and more feeling, and outside readers of the transcripts agreed. In a meta-analysis of 617 studies and 1.2 million people, women’s edge in remembering words and prose was 0.28, and for all memory 0.19, while men were ahead for routes and abstract shapes. But the gap in verbal memory was smaller in Asian samples than in European, North American, and South American ones, a hint that culture shapes it.

In the love lab

What keeps a marriage together? Course 9 (Lesson 4) told how passionate love gives way to companionship. This lesson asks about the daily craft of staying together, and the person who has done most to study it by watching is the psychologist John Gottman. In 1983 in Bloomington, Indiana, he and Robert Levenson recruited 79 couples through newspaper advertisements, who had not talked to each other for at least eight hours, to a laboratory where two cameras behind darkened glass filmed them and sensors measured their heart rate, finger pulse, and sweat. Each couple talked for 15 minutes about a problem they kept returning to. Days later each partner watched the tape and turned a dial from “extremely negative” to “extremely positive.” Gottman later built a studio apartment laboratory at the University of Washington (the Gottman Institute dates it to 1986) where couples lived for a day and were filmed.

The first results, published in 1992, tracked 73 of the couples for four years: nine had divorced, 18 had separated. Couples whose conflicts were more negative and more physiologically stirred up were less satisfied, in poorer health, and at greater risk of breaking up. In a later study, of 130 newlyweds in the Seattle area followed for three to six years, 17 divorced. The team had expected that “active listening,” in which a listener paraphrases what the speaker has said, would mark the happy couples. They were surprised: it hardly ever happened (about 4 seconds in 900 of conflict talk) and did not matter. What predicted divorce instead was a husband who brushed aside his wife’s influence, a wife whose complaints began harshly (“start-up”), and a husband who did not calm after small negativity. Gottman had already proposed, in 1993, a cascade of four behaviors, later known as the “four horsemen”: criticism, contempt, defensiveness, and stonewalling, each leading to the next. In his statistics defensiveness predicted divorce most strongly, though the popular books name contempt. He also found a “magic ratio”: in conflict talks, couples who stayed together showed about five positive moments for every negative one, while unstable couples showed 0.8.

How much of this stands? The prediction figures, such as the 83 percent for newlyweds, came from groups that were chosen after the fact (17 divorced couples and 40 stable ones, 57 in all), and they were never tested on fresh couples. When the psychologists Richard Heyman and Amy Slep tried that discipline on a different set of marriages, an equation that classified 90 percent of couples correctly in the group it was built on got only 69 percent right in a second group, and only 29 percent of the couples it called doomed actually divorced. A team that tried to repeat Gottman’s 1998 findings with 85 young, low-income couples reported that the major findings “failed to replicate.” In a study of 431 low-income newlyweds in Los Angeles, followed for about two years, the way couples spoke to each other predicted their later satisfaction only weakly, and their satisfaction predicted how they spoke about as well or better. The psychologists Benjamin Karney and Thomas Bradbury summed up in 2020 that negative communication “does not always predict distress,” is hard to change, and may work differently for couples with less money. What survives is modest: negativity and contempt go with unhappiness, and a long, careful record of marriages is more useful than a formula for predicting a particular one.

Who demands, who withdraws

The standing joke is that wives nag and husbands sulk. Research on the “demand and withdraw” pattern finds that it is real and unhealthy, but it follows the topic more than the sex. In a study by Andrew Christensen and Christopher Heavey, 31 couples discussed two problems, one where the husband wanted change and one where the wife did. The wife-demands, husband-withdraws pattern appeared mostly when the wife wanted the change; each spouse demanded more when they wanted change and withdrew more when the other did. A meta-analysis of 74 studies and 14,255 people found that the pattern went with worse marriages whichever partner demanded (a correlation of 0.36). Another, of 64 studies, found that women showed somewhat more hostility and distress and men more withdrawal, but “of a small magnitude.” The pattern also seems to track who gets more of what they want from the marriage. In the United States in 2025, women spent about 2.4 hours a day on household activities and men about 1.6, with 2.8 hours of paid work for women against 3.8 for men (all adults, not only couples). In one study of dual-earner couples, a sense of unfairness in who does the housework lowered both spouses’ happiness and raised the chance of divorce among wives.

Conflict also gets under the skin. In a hospital study in 2005, 42 healthy married couples, aged 22 to 77, came in twice for 24 hours; during one visit they discussed a marital disagreement, and during the other they were given social support. Small blister wounds healed more slowly after the disagreement, and couples who were more hostile healed at about 60 percent of the speed of those who were less so.