EXPERTS TELL US the Eskimos have about four dozen words for snow. Or is it 200? Or seven? Or maybe four? Here's a hint: It's roughly the same number as in English. And here's another hint: Most of the people who throw Eskimo snow-word numbers around don't know anything about it, and haven't bothered to look it up.
Here's one example: Over the last 15 years, a series of books and articles have told us that women talk a lot more than men do. According to Dr. Scott Halzman in Psychology Today, women use about 7,000 words a day, and men use about 2,000. On the other hand, Ruth E. Masters, in her book ``Counseling Criminal Justice Offenders," tells us that ``Females use an estimated 25,000 words per day and males use an estimated 12,000 words per day." And according to James Dobson's book ``Love for a Lifetime," ``research tells us" that God gives a woman 50,000 words a day, while her husband only gets 25,000.
A bit of Googling easily turns up at least nine different versions of this claim, ranging from 50,000 vs. 25,000 down to 5,000 vs. 2,500. But a bit of deeper research reveals that none of the authors of these claims actually seems to have counted, and none cites anyone who seems to have counted either.
The most recent to join the chorus is Dr. Louann Brizendine, clinical professor of psychiatry at the University of California, San Francisco. In her current best-seller, ``The Female Brain" (Morgan Road), Brizendine tells us that ``A woman uses about 20,000 words per day while a man uses about 7,000."
``The Female Brain" has made quite a splash since its publication last month,
and this word-count claim is one of the most striking facts supporting her
argument that the female brain is ``a lean, mean communicating machine." The
20,000 vs. 7,000 numbers have been cited in reviews all over the world, from The
Since Brizendine is the director of a clinic at UCSF, one of the world's most important biomedical research institutions, and her book provides 90 pages of endnotes and references to back up 180 pages of text, I hoped it would finally give me a reliable source for this statistic.
The book's endnotes appear to attribute the numbers to a 1997 self-help book by Allan Pease and Allan Garner, ``Talk Language: How to Use Conversation for Profit and Pleasure." But Pease himself has presented several different word count numbers in other sources. In 2000, he published ``Why Men Don't Listen and Women Can't Read Maps" (with Barbara Pease), which attributes to women ``6,000-8,000 words," while men get ``just 2,000-4,000 words." (They also offer daily counts for women's and men's ``vocal sounds" and ``facial expressions, head movements, and other body language signals"-but don't provide a source for any of the counts.) In a 2004 CNN interview, Allan Pease said that ``women can speak 20,000 to 24,000 words a day versus a man's top end of 7,000 to 10,000."
Allan Pease is a prolific writer, and a sampling of his other recent titles gives a sense of his men-are-from-Mars, women-are-from-Venus philosophy: ``Why Men Don't Have a Clue and Women Always Need More Shoes"; ``Why Men Lie and Women Cry"; ``Why Men Can Only Do One Thing at a Time and Women Never Stop Talking."
As it happens, there are many scientific studies that count the words used by females and males in a variety of same-sex and mixed-sex interactions: phone conversations, interviews, group discussions, and so on. These are always time-limited situations-a few minutes to a few hours of talking-not recordings across the whole range of people's daily activities. But together, these studies involve thousands of speakers of many ages, regions, languages, and cultures.
The findings? According to a 1993 review of the scientific literature by researchers Deborah James and Janice Drakich, ``Most studies reported either that men talked more than women, either overall or in some circumstances, or that there was no difference between the genders in amount of talk." The research since that review, including counts from my own research, follows the same pattern.
I haven't been able to find any scientific studies that reliably count the entire daily word usage of a reasonable sample of men and women. But based on the research I've read and conducted, I'm willing to make a bet about what such a study would show. Whatever the average female vs. male difference turns out to be, it will be small compared to the variation among women and among men; and there will also be big differences, for any given individual, from one social setting to another.
Unfortunately, this is just one of several cases in recent books on sex and neuroscience where striking numbers turn out to be without apparent empirical support. On page 36 of ``The Female Brain," Brizendine writes that ``Girls speak faster on average-250 words per minute versus 125 for typical males." In support of this assertion, her endnotes cite Bruce P. Ryan, ``Speaking rate, conversational speech acts, interruption, and linguistic complexity of 20 pre-school stuttering and non-stuttering children and their mothers," Clinical Linguistics & Phonetics, 14(1), pp. 25-51 (2000). Alas, in Ryan's paper, you won't find the 250 vs. 125 numbers, and in fact, he gives no data at all that breaks down speaking rates by sex.
The truth is out there, however, in many studies over the years that do give figures for speaking rates of females and males of various ages. The most recent data comes from a paper presented at a conference this month, in which Jiahong Yuan, Chris Cieri, and I looked at various measures of speaking rate in thousands of English and Chinese telephone conversations. We found that in both languages, the males spoke about 2 percent faster, on average, than the females. This effect was small compared to the variation among female or male speakers, and it was also small relative to the effect of situational factors. For example, people talking with family or friends spoke about 10 percent faster than people talking with strangers.
These numbers might be unrepresentative or otherwise mistaken, but we've documented the procedures we used and the data we analyzed. And we used conversations that have been published as digital audio, along with time-aligned transcripts and demographic data for the speakers, so others can check our work if they want to.
This ability to check or replicate research is central to scientific progress. It doesn't stop people from disagreeing about facts and theories, but it helps organize the arguments and keep them on track.
The authors of self-help works, as a group, don't seem to have any particular standards of accuracy. Journalists, meanwhile, generally take them at their word in reviews and interviews, and publishers are happy as long as the books sell well.
It's a shame to see this approach to the facts spreading into the growing genre of books about the neuroscience of sex differences, where the facts can have real consequences.
Mark Liberman is Trustee Professor of Phonetics at the University of Pennsylvania.