August 06, 2006

Sex-linked lexical budgets

[Update 7/3/2007: a list of links to other relevant posts can be found here. ]

This morning, I spent a fruitless hour trying to track down the source of Louann Brizendine's assertion that "A woman uses about 20,000 words per day while a man uses about 7,000". I found many similar assertions, with estimates of the male lexical allowance varying from 2,000 to 25,000, while assertions about the female daily word budget ranged from 7,000 to 50,000. But nowhere could I find any evidence that anyone has ever supported these assertions by actually counting words or measuring talking times. My current best guess is that a marriage counselor invented this particular meme about 15 years ago, as a sort of parable for couples with certain communication problems, and others have picked it up and spread it, while modulating the numbers to suit their tastes. This is what happened in the case that Geoff Pullum called The Great Eskimo Vocabulary Hoax, discussed here. If I'm wrong, and you know a source for Brizendine's numbers that isn't just passing along someone else's story, please tell me.

Here's what I've found so far --

We can start with this joke at

A husband looking through the paper came upon a study that said women use more words than men.
Excited to prove to his wife that he had been right all along when he accused her of talking too much, he showed her the study results. It read "Men use about 15,000 words per day, but women use 30,000".
The wife thought for a while, then finally she said to her husband "It's because we have to repeat everything we say."
The husband said "What?"

Some of my favorite jokes are well supplied with footnotes, but this one isn't.

A compilation of averages on a trivia page at says that "On average women say 7,000 words per day. Men manage just over 2000." There are no footnotes here either, needless to say.

The same numbers appear in a 2004 article by Hara Estroff Marano from Psychology Today, "Secrets of Married Men", which quotes "Rhode Island psychiatrist Scott Haltzman, M.D." as follows:

The average woman uses 7,000 words a day and five tones of speech, he points out. The average man uses 2,000 words and three tones. "Men are talk-impaired, relatively speaking," he says.

Haltzman has also written a book, also (according to the blurb at "emphasizing the biological differences between men and women", but if he's published in the scientific literature on this topic, Google Scholar doesn't know about it.

The 7,000/2,000 numbers also feature in Kevin Burke's one-man show "Defending the Caveman", according to a review by Adrienne Broaddus:

Did you know that men generally speak about 2,000 words a day and women 7,000? Men, Burke explains, bond and communicate by sharing long periods of silence and occasional name-calling, whereas women bond by gossiping, processing things and sharing emotional insights. That explains why men are never able to tell women details about their night out with the guys - they don't talk.

A 2004 CBS News segment quotes Kate White, the Editor-in-Chief of Cosmopolitan, citing numbers in the same range:

"We make the mistake as women sometimes of thinking that, because (men are) different, there is something wrong. The average guy speaks 2,000 to 4,000 words a day and the average woman 6,000 to 8,000. So we're different, but it doesn't mean there is something the matter with the relationship."

No footnotes, alas.

An undated article by Debbie Waitkus on the website of the magazine Ladies Golf Journey, "To Speak or Not to Speak, That is the Question", tells us that

Women, for the most part, love to talk. In fact, women speak an average of 30,000 words a day. Compare this with an average of 12,000 a day for men. Women can talk about anything, anywhere, anytime – golf course included.

A 1999 article in The Nationalist tells us:

Research shows that women can speak 20,000 to 25,000 words a day compared to men’s paltry 7,000 to 10,000.

Again no source is cited, though the article discusses a book by Allan and Barbara Pease where the meme does occur, though with different numbers. The cited book (Allan and Barbara Pease in "Why Men Don't Listen & Women Can't Read Maps" p. 80-81) not only gives different totals, but also breaks the totals down in an unusual way:

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.

The referents of these multiple counts ("vocal sounds", "body language signals") are kind of vague, to say the least; and all of the counts seem roughly as scientific as the count of months in the quoted joke.

Another variant of the Pease-derived version of this meme can be found in a compilation of sex-stereotype wisdom by Dan Bova from Redbook:

Why doesn't he want to talk about his day when he gets home?
"I just want to leave all the annoying crap of the day behind me and think about nothing for a while," says Jim, 31, a father of two from Beacon, NY. At the end of the day, men are tired of thinking, and, more important, we're tired of talking. "Studies show that women use 8,000 to 9,000 words a day. Men use 2,000 to 4,000 words a day on average," explains communication expert Allan Pease. "By the time they come home from work, they've used up their words. And women have 5,000 left to go."

And here's another Allan and Barbara Pease bit, in the transcript of a 2004 CNN interview plugging their book "Why men don't have a clue and Women Always Need More Shoes" (yes, they've got a million of 'em, it seems -- they're serial stereotypers):

MCINTYRE: All right, so help us out. Help us clueless men out. How do we crack the code?
A. PEASE: Well, first of all, to understand that the brain scans that we talk about in our book show very clearly that men and women are different. Now, it's politically correct, I know, to go around pretending that men and women are exactly the same now. But the reality is we're not. And everybody knows this. We're scientifically different. And so, first of all, is to accept that.
Secondly, to understand that women can speak 20,000 to 24,000 words a day versus a man's top end of 7,000 to 10,000. And where this becomes apparent is at the early evening when you're having dinner, because most men have done their 10,000, right? She might still have 15,000 to go, and someone's got to hear them.

We can also see here the introduction of "brain scans" into this world. The Peases have clearly recycled this meme many times over the years, and it's possible that they're actually its source, though so far I don't have any references that antedate the citations coming up next.

James Dobson wrote in his Focus on the Family column, June 2004:

Research makes it clear that little girls are blessed with greater linguistic ability than little boys, and it remains a lifelong talent. Simply stated, she talks more than he. As an adult, she typically expresses her feelings and thoughts far better than her husband and is often irritated by his reticence. God may have given her 50,000 words per day and her husband only 25,000. He comes home from work with 24,975 used up and merely grunts his way through the evening. He may descend into Monday Night Football, while his wife is dying to expend her remaining 25,000 words.

Apparently this was recycled material, since Christianity Today quotes it from a book first published in 1993:

Dr. James Dobson hits on the wiring problem in his book, Love for a Lifetime. He writes: "Research makes it clear that little girls are blessed with greater linguistic ability than little boys, and it remains a lifelong talent. Simply stated, she talks more than he." Dobson suggests that God may have given Mrs. Cell Phone 50,000 words per day while Mr. Computer may average 25,000.

I don't have access to the 1993 edition of this book (though ads for the current version feature the same quote), so I'm going to be tentative in naming Dobson as one of the two earliest publications of the sex-linked vocabulary allowance idea. And Dobson says that "God may have given" particular lexical allowances to a hypothetical (if prototypical) woman and man, which is fair enough, so we can't blame him for the fact that the quantities are unsourced.

Dobson also brings out one of the interesting secondary features of (some versions of) this meme: the notion that we are dealing not with a behavioral average (like, say, the number of steps someone takes in the course of a day), but rather with a sort of behavioral budget. You get X number of words, and once you've said that many, you've got to stop. Sorry, too bad if you've got more to say, your word tank is empty. Try again tomorrow. (This is in contrast to a version in which women talk more just because they're more verbally adept or more verbally inclined than men are.)

In the obviously serious volume Counseling Criminal Justice Offenders, Ruth E. Masters wote "Females use an estimated 25,000 words per day and males use an estimated 12,000 words per day". And wonder of wonders, she gives a reference: (Smalley, 1993)!

Alas, (Smalley, 1993) turns out to be a little pamphlet made to be handed out by marriage counselors -- Gary Smalley, "Connecting with your husband", p. 18-19:

Studies show that the average male uses about 12,000 words a day, the entire day, and most of those are spent relating to people while on the job. Remember, most men are aggressive and driven. They will talk at length in the workplace in order to successfully complete an assignment, project, or task.

A woman, on the other hand, averages 25,000 words per day. Now these aren't just any words but words that must connect with people or emotions. In other words, when a woman spends her day iin the workplace, generally there are few opportuntities for her to realy dig in and use her allotment of words.

Here's the problem. At the end of the day -- whether the woman works in an office of in the home -- there is huge difference between the man's word count and the woman's. A man has spent nearly all his words. He comes home tired and drained, looking for a place to recharge for the next day's battle at the office.

A woman, however, is just warming up. She has thousands of words left to speak, and since her husband's word count is depleted, the conversations often wind up sounding like nothing more than question-and-answer sessions.

If this passage is in the 1993 edition (the one that I was able to search online is more recent), then Smalley is more or less tied with Dobson for priority on this idea. Smalley also features the "lexical budget" or "word tank" version of this meme. This is such a bizarre idea that it probably has a common source. Either Dobson got it from Smalley, or Smalley got it from Dobson, or both of them got it from some third party. The question is, did Louann Brizendine really just pick the idea up from this demi-monde of pseudo-scientific urban legends in order to deploy it in her pop neuroscience, or is there actually some sort of half-way respectable research in its history somewhere?

Looking through Brizendine's book, I couldn't find this factoid other than in the jacket copy. That's too bad, since in general she gives references for the assertions in the book. [update 9/2/2006: Brizendine makes the assertion on p. 14 of the book, and offers a list of references that includes Pease. A. and A. Garner (1997) Talk Language: How to use conversation for profit and pleasure, which is presumably where she found it.] The word-budget business comes up in this passage from one of her online lectures:

And some of you may know this interesting fact. By female- adulthood, females speak on average about twenty thousand words a day, and ... My sixteen-year-old son will sometimes say to me, "Mom, I've already said my seven thousand for the day. Call up your girlfriend!" [laughs]

[I should say, if it's not already obvious, that I'm skeptical about the truth of this claim. For either sex, the variance of daily word-production counts will be enormous, and even for a particular individual, the count will depend massively on variable aspects of life circumstances. At best you'd be able to say that in a certain range of comparable circumstances, there were different mean values for men and women; and even in that sort of controlled comparison, I'd be very surprised if the within-group variation wasn't much larger than the across-group difference. So far, I haven't found any evidence that there is any empirical warrant for saying even that much.]

[Update -- Clay Beckner reminded me of the review by Deborah James and Janice Drakich, "Understanding Gender Differences in Amount of Talk: A Critical Review of Research", in Deborah Tannen, ed., Gender and Conversational Interaction, Oxford U Press, 281-312 (1993); and sent the abstract:

It is shown that the widely held belief that women talk more than men is unsupported in the literature. Of the studies reviewed that examined mixed-sex interaction, the majority found either that men talked more than women, or that there was no difference between men & women in amount of talk. Approaches to understanding these findings are explored, with one theory - status characteristics theory - highlighted as most helpful in understanding gender differences in amount of talk. The effect of the research activity on the amount of talk in each study is explored, with studies divided into those that used formal task activities, informal task & nontask activities, & formal nontask activities. 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. In each of these contexts, the findings are explored in light of the status characteristics theory. It is concluded that rather than viewing the overwhelming tendency of males to talk more than females as further evidence of domination & exploitation of power over women, the different goals for interaction, to which both men & women are socialized, should be considered in the context of social structure.

The cited research is not quite a refutation of the Dobson/Smalley/Pease/Brizendine/etc. claim, since it monitors what happens during a period devoted in some sense to interaction. It's still possible that when not forced into socializing, men uniformly clam up (e.g. in front of football games or behind newspapers) while women uniformly seek out someone to talk with, so that average daily word counts would still favor women by the nearly three-to-one margin that Brizendine claims. But I doubt it.

And in fact there is a body of material that tests the sexual word budget directly. Clay also reminded me of Rayson, P., Leech, G., and Hodges, M., "Social differentiation in the use of English vocabulary: some analyses of the conversational component of the British National Corpus". International Journal of Corpus Linguistics. 2(1) pp 133-152 (1997).

This data comes from 153 demographically balanced speakers, 75 female and 73 male, who "were equipped with a high-quality Walkman sound recorder, and recorded any linguistic transactions in which they engaged during a period of two days", obtaining permissions from their interlocutors as well. The total number of speakers recorded by this method included 561 females and 536 males.

TABLE 1 Distribution of the Conversational Corpus between Female and Male Speakers
Female Speakers Male Speakers
Number of Speakers 561 536
Number of turns 250,955 179,844
Number of words spoken 2,593,452 1,714,443
Number of turns per speaker 447.33 335.53
Number of words per turn 10.33 9.53

Unfortunately, this particular table covers all the contributions by all those involved in any of the conversations, not just the people who were recording all of their output during two days. However, we can look at the results in a couple of ways that bear on the sexual word budget question.

1. Average words per day per speaker (both produced and heard):
    1/2 days * 4,552,555 words / 153 speakers = 14,878
2. Average words spoken per day, assuming equal conversational shares:
    14,878/2 = 7,439
3. Ratio of words per female speaker to words per male speaker (recorder-wearers or not):
    (2593452/561)/(1714443/536) = 1.45    (i.e. 45% more per speaker for females)
4. Inferred female and male daily word totals:
    male = 14,878/2.45 = 6073 words per day.
    female = 14,878-6073 = 8805 words per day.
[Note that this is a doubly indirect estimate -- but the data is out there, so we should be able to get exact counts, including an estimate of the overlap in the female vs. male distributions!]

The table also indicates that the female participants took an average of 33% more turns (per speaker) than the male participants, and used 8.4% more words per turn, as well as using 45% more words. I'm not sure why this result is different from those of the studies surveyed by James and Drakich (where males generally took more turns and used more words), though perhaps a clue can be found in the fact that the eight words "most characteristic of male speech" (by χ2 value) in the BNC conversational corpus were fucking, er, the, yeah, aye, right, hundred, fuck, while the corresponding list for the women was she, her, said, n't, I, and, to, cos. In other words, I suspect a difference in formality, and probably also in class, between the BNC collection and the (much smaller) studies surveyed by James and Drakich. (I should say also that these counts seem a bit low to me. At overall conversational speech rates (from all parties in a given conversation) of about 200 WPM, 15,000 words would be only about an hour and fifteen minutes of talking time. Unless the participants led rather lonely lives, perhaps they didn't have their Walkmen turned on all the time, or didn't get permission from all their interlocutors?

In any case, 6,073 vs. 8,805 is a far cry from 7,000 vs. 20,000 -- and if we looked into the details further, I suspect we'd find a very large overlap between male and female daily word counts in the BNC conversational corpus. (This will be easy enough to check, as mentioned above.) But in any case, the differences between this result and the studies surveyed by James and Drakich reminds us that these numbers represent not the automatic consequence of chromosomes and hormones, but rather contingent actions of complex individuals in varied contexts of interaction. Who are influenced by their genes and hormones, sure enough. But still.]

[Update: more on this here ; and a list of links to other relevant posts can be found here.]

[Update 2/24/2007 -- Jeff Allen writes:

There is quote by James Dobson, from a June 2004 column, and then back to his 1993 book. I recall hearing James Dobson use the same example as the cited June 2004 text in a radio broadcast sometime during the period of 1984-1988.


Posted by Mark Liberman at August 6, 2006 09:22 AM