December 11, 2006

Sex differences in "communication events" per day?

Louann Brizendine's book The Female Brain, published last August, featured a number of striking quantitative assertions about sex differences in communication. The jacket blurb claimed "A woman uses about 20,000 words per day while a man uses about 7,000", while the text (p. 14) gave the same numbers in the other order: "Men use about seven thousand words per day. Women use about twenty thousand." Dr. Brizendine gives a set of references in her end-notes, but none of them support those numbers. In fact, no study of any sort has ever measured any numbers at all like these, as far as I've been able to find.

What are the facts about sex and talkativeness? There's an enormous amount of individual variation, and each individual talks more or less depending on mood and context. Against this background of variation, many studies have measured how much women talk, on average, compared to how much men talk, on average. The differences that they find between men and women as groups have always been small compared to the differences among men as individuals or among women as individuals. And more often than not, these small group differences actually show men talking a bit more than women do. For additional details, see the links at the end of this post.

Earlier this month, it seemed that Dr. Brizendine had been persuaded to withdraw the word-count numbers (Stephen Moss, "Do women really talk more?" The Guardian, 11/27/2006):

When I reach Brizendine, just as she is crossing the Golden Gate bridge, she tells me that she has accepted the criticism of the numbers quoted in the book - on both volume of words and rate of speech - and will be deleting them from future editions. Nor will they appear in the UK edition, to be published by Bantam in April. "I understand Mark Liberman's point and I am grateful to him," she says. "He felt I was passing on data that was not nailed down, and thus perpetuating a myth, so it will be taken out in future editions." She admits language is not her specialism, and she had been reliant on the advice of others.

But another interview, published last weekend, offers a re-interpretation rather than a retraction (Deborah Solomon, "He thought, she thought", NYT Magazine 12/10/2006):

Q: Your book cites a study claiming that women use about 20,000 words a day, while men use about 7,000.

A: The real phraseology of that should have been that a woman has many more communication events a day — gestures, words, raising of your eyebrows.

Now, given the relatively slow pace of the magazine business, it's quite possible that Ms. Solomon interviewed Dr. Brizendine before Mr. Moss did. Thus the NYT magazine Q&A may not reflect her current position. But the point is worth following up in any case.

An executive summary of the conclusions: the claim about "communication events" seems to have essentially the same status as the claim about word counts. I can't find any studies that yield numbers at all like those in the re-interpreted claim, which would be something like "women use about 20,000 communication events a day, while men use about 7,000". The studies I've been able to find that count something like "events" of nonverbal communication yield the same sorts of results as studies that count words: there's a lot of individual variation; individuals vary a lot depending on context and mood; male and female averages are not consistent with a large sex difference in counts of overall "communication events". (Though there can be large sex differences in the case of particular nonverbal signals: in a study discusssed below, women laughed about 43% more than men, while men produced 623% more "chin thrusts" than women did.)

Given the references cited in the end-notes to p. 14 of Brizendine's book, it seems that she got her "communication event" numbers from one of the many books by Allan Pease and his various co-authors. For example, Allan and Barbara Pease, Why Men Don't Listen & Women Can't Read Maps, p. 80-81:

A woman can effortlessly speak an average of 6,000-8,000 words a day. She uses an additional 2,000-3,000 vocal sounds to communicate, as well as 8,000-10,000 facial expressions, head movements, and other body language signals. This gives her a daily average of more than 20,000 communications. That explains why the British Medical Association recently reported that women are four times more likely to suffer from jaw problems.

"Once I didn't talk to my wife
for six months," said the comedian.
"I didn't want to interrupt."

Contract a woman's daily "chatter" to that of a man. He utters just 2,000-4,000 words and 1,000-2,000 vocal sounds, and makes a mere 2,000-3,000 body language signals. His daily average adds up to around 7,000 communication "words" -- just over a third the output of a woman.

They give no citation for any research backing up these numbers. I've spend a fair amount of time searching fruitlessly for published studies that might support such assertions. If you know of any, please tell me. [The work that Brizendine actually cites is Pease. A. and A. Garner (1997) Talk Language: How to use conversation for profit and pleasure. I don't have a copy of this work, but I'm assuming that it offers the same numbers, and the same lack of documentation, as the one that I've quoted from -- and in both cases, it seems that the numbers represent the type of pseudo-scientific bible story that has become regrettably common in popular works on psychology, as well as in the system of folk beliefs that Arnold Zwicky calls "bizlore". (Update 12/22/2006 -- I've bought a copy of Talk Language, and in fact it contains no information whatever about any counts of words, eyebrow movements or any other "communication events".)]

As you can read in the links at the end of this post, I'm quite confident that the word-count part of this assertion (6000-8000 words a day for an average woman, 2,000-4,000 words a day for an average man) is not consistent with the numbers that have really been measured in many published studies. What about the "additional ... vocal sounds" and the "facial expressions, head movements and other body language signals"?

One study that will give us an idea of how this is likely to work out is John F. Dovidio, Clifford E. Brown, Karen Heltman, Steve L. Ellyson, Caroline F. Keating, "Power displays between women and men in discussions of gender-linked tasks: A multichannel study", Journal of Personality and Social Psychology, 55(4), 580-587, 1988.

Here's their description of the study:

In preliminary testing at the beginning of the term, 88 introductory psychology students rated their familiarity (0 = no familiarity, 10 = a great deal of familiarity) with the materials, steps, and potential problems of 14 activities (e.g., washing and waxing a car, writing a research paper) that varied in their association with masculine and feminine gender roles. On the basis of those anonymous ratings, we selected three tasks as discussion topics: automotive oil changing, for which men showed greater familiarity than did women (M s = 7.0 vs. 2.4, p <.001); pattern sewing, for which women showed greater familiarity than did men (M s = 6.4 vs. 1.4, p <.001); and vegetable gardening, for which men and women indicated equal familiarity (M s = 6.5). We drew the 24 men and 24 women who participated in our study from this pool of students. Each mixed-sex dyad discussed the masculine topic (oil changing), the feminine topic (sewing), and the non-gender-linked topic (gardening). [...]

We randomly selected 24 male and 24 female undergraduates from a pool of 50 male and 38 female students in an introductory psychology class at a midwestern liberal arts college. We randomly paired the subjects, who were not previously consociated, in mixed-sex dyads. [...]

Each dyad discussed all three topics. "The order of the three discussion tasks (oil changing, sewing, and gardening) was counterbalanced, and one male and one female experimenter ran two dyads in each order."

The (three-minute-long) conversations were videotaped, and coded as follows:

Two coders recorded the verbal and nonverbal behaviors from the videotapes. The verbal measures were the number of speech initiations by each participant (Rosa & Mazur, 1979) and the percent of the total interaction time that each subject spoke (Berger et al., 1985). The nonverbal measures were (a) looking while speaking, the percent of time that the subject looked at his or her partner while the subject spoke (Dovidio & Ellyson, 1985); (b) looking while listening, the percent of time the subject looked at his or her partner while listening to the partner speak (Dovidio & Ellyson, 1985); (c) rate of gesturing, the number of expressive hand movements (not in contact with one's own body) that occurred per second while speaking (Dittman, 1972; Henley, 1977); (d) frequency of chin thrusts (Camras, 1980; Henley, 1977); (e) frequency of smiling (Henley, 1977); (f) frequency of self-touching, hand movements in contact with part of one's own body; and (g) frequency of laughing (Henley, 1977; Waxer, 1977).

The results?

There was an obvious, and interesting, effect of topic. How people communicate does depend on the interaction between who they are and what they're communicating about! But averaging over topics so as to focus on the sex differences, we find (if I've done the arithmetic correctly):

Male Female
Time speaking
Speech initiations
Looking while speaking
Looking while listening
Rate of gesturing
Frequency of chin thrusts
Frequency of smiling
Frequency of self-touching
Frequency of laughing

So the guys did more of the talking, as is often the case -- 43% more, this time, which is a bigger difference than one usually sees. What about non-verbal signals? Well, the guys did 80% more gesturing, and produced 623% more chin thrusts. The gals did 28% more smiling, 7% more self-touching, and 46% more laughing. Dovidio et al. didn't count eyebrow motions, it's true. But there's certainly no support here for the view that women produce about three times more "communication events" on average than men do.

[In more detail: Suppose we insist, rather against common sense, on an overall count of "communication events" in this data. We assume (without any basis) that each hand-gesture, each word, each laugh, etc. is a single "communication event". Then the average male spoke for 40% of three minutes, which is 1.2 minutes; since this calculation eliminates all pauses, let's assume 200 wpm (it was probably more than this), yielding 240 words; the average female totaled .28*3 = 0.84 minutes, which yields 168 words. I'm not sure how to count the gaze percentages, so let's leave them out for a first approximation. The "rate of gesturing" is measured per second, so we get .09*180 = 16.2 for the males, and .05*180 = 9 for the females. The other measures are all (I think) counts for the whole conversation. So for the males, we get 240+16.2+1.62+10.6+6.1+4.1 = 278.62 "communication events". For the females, we get 203.36 "communication events'. These sums are a preposterous farrago of category errors -- it's like adding up cars, bicycles, shoes and socks as transportation modalities -- but some such sums seem to be required by the Pease/Brizendine theory, and I don't see any way to get them to come out in a way that's consistent with the females-are-three-times-more-communicative theory.]

I'm hoping that the Guardian interview really was later than the NYT magazine interview. I'm afraid, though, that whether or not Dr. Brizendine retracts the word-count claims or just re-interprets them as "communication-event" claims, what the world will remember is the "scientific proof" that women are three times talkier.

Other posts on Louann Brizendine's The Female Brain:

Neuroscience in the service of sexual stereotypes (8/6/2006)
Sex-linked lexical budgets (8/6/2006)
Sex and speaking rate (8/7/2006)
Yet another sex-n-wordcount sighting (8/14/2006)
The main job of the girl brain (9/2/2006)
The superior cunning of women (9/2/2006)
The laconic rapist in the womb (9/4/2006)
Open-access sex stereotypes (9/10/2006)
David Brooks, neuroendocrinologist (9/17/2006)
Sex on the brain (Boston Globe, 9/24/2006)
Gabby guys: the effect size (9/25/2006)
"Every 52 seconds": wrong by 23,736 percent? (10/13/2006)
Two new reviews of Brizendine (10/30/2006)
Word counts (11/28/2006)

More on the spread of these ideas in the media:

Regression to the mean in British journalism (11/28/2006)
Censorship at the Daily Mail(11/29/2006)
Contagious misinformation(12/1/2006)
Femail again(12/2/2006)
Bible Science stories(12/2/2006)
Fabricated but true?(12/3/2006)

Posted by Mark Liberman at December 11, 2006 09:57 AM