December 31, 2006

Busy tongues

The latest instance of "Everyone already knows this" commentary on The Female Brain comes from Cinnamon Stillwell in the SF Examiner, 12/29/2006, under the headline "Experts Discover Men And Women Are Different!":

When it was revealed that scientific studies published in the new book "The Female Brain" demonstrate that women talk more than men, many of us responded with a collective shrug. Anyone who has ever been in a relationship with a member of the opposite sex -- whether romantic, familial or friendly -- knows that women talk more than men. A lot more.

"The Female Brain" indicates that not only do women talk three times as much as men, but they also get a chemical rush in their brains from hearing their own voices. This may explain why women describe "feeling better" after talking about problems or issues in their lives, beyond the mere relief of getting it off their chest.

The "revelation" behind the hyperlink is the Daily Mail article that I discussed around Thanksgiving ("Regression to the mean in British journalism", 11/28/2006). Brief recap: Louann Brizendine neither did nor cited any "scientific studies" about sex and talkativeness, but just invented some numbers out of thin air -- or maybe quoted someone else who invented the numbers. She's semi-retracted the claim. And the "chemical rush" business is apparently just as bogus -- see the links collected here for some discussion.

But let's light a scientific candle instead of cursing the journalistic darkness.

In an earlier post ("Gabby guys: the effect size", 9/23/2006), I discussed some data from the Fisher English Corpus Part 1 (FECP1), a collection of 5,850 telephone conversations lasting up to 10 minutes each, recorded in 2003. 10,950 of 11,700 conversational sides involved native speakers of American English for whom information about sex, age and years of education is available, I've posted a summary of data from those calls here. Each line presents information about one conversational side, laid out like this:

sex   no. of turns  no. of words  total time  words per min.   ID   age  years of edu
 m       149           773          208.5        222.446      2602   34     16
 f       138           876          212.26       247.621      1790   24     16 

One simple thing we can do with this data is to fit a linear regression model. A script for the free software statistics package R, which will read in the data and fit such a model, is here. Type the R expression

summary(M1 <- lm(words ~ sex + age + edu + sex:age + sex:edu - 1))

(it's in the script as well) and you'll learn:

            Estimate Std. Error t value Pr(>|t|) 
   sexf     731.1017    21.7214 33.658   < 2e-16 ***
   sexm     808.0448    26.2205 30.817   < 2e-16 ***
   age        2.5811     0.2945  8.765   < 2e-16 ***
   edu        4.3731     1.2402  3.526  0.000424 ***
   sexm:age  -0.2608     0.4285  -0.609 0.542853 
   sexm:edu  -1.6424     2.0538  -0.800 0.423915 
   Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 
Residual standard error: 288.2 on 10944 degrees of freedom
   Multiple R-Squared: 0.909, Adjusted R-squared: 0.909 
   F-statistic: 1.822e+04 on 6 and 10944 DF, p-value: < 2.2e-16 

The effects of sex, age and education were highly "significant" in the technical statistical sense. Whether these effects were significant in the ordinary language sense, you can judge for yourself. The sex-age and sex-education interactions were not significant.

Your basic modeled male put out about 77 more words per conversation than your basic modeled female did -- 808 vs. 731, or about 10% more. And independent of sex, each additional year of age was worth about 2.6 additional words per conversation, while each additional year of formal education was worth about 4.4 additional words. A light-hearted way to put this would be that being male is worth about 30 years of experience (76.9/2.58 = 29.8) or 18 years of formal education (76.9/4.37 = 17.6 ). In terms of word count in a 10-minute phone conversation, that is... [Emily Bender has written to warn me that irony is dangerous, and I risk having my numbers quoted by the BBC or the AP, along the lines of "From the point of view of verbal facility, being male is worth 30 years of practice or 18 years of formal education, according to research published this month". We'll see...]

Once again, young men talk like old women. But we already knew that, right?

Here's a boxplot to help you judge the size of the sex and education effects. The top and bottom of the box are the 75th and 25th percentiles; the whiskers extend out to the edges of the range, once some statistical outliers have been trimmed. The four boxes show males and females over the age of 25 with a high school education or less, vs. with a college degree or more. (The script also includes the code to generate this plot -- you can make your own plot for the sex and age effects...)

[Update -- several people, including Geoff Pullum, have copied me on email sent to Cinnamon Stillwell, clueing her in to the non-existence of the "science" confirming what she thinks everyone "knows", and to the contrary results of such studies as do actually exist. She hasn't responded yet, but in fact Steven Colbert has already scripted the response that I expect:

And on this show, on this show your voice will be heard... in the form of my voice. 'Cause you're looking at a straight-shooter, America. I tell it like it is. I calls 'em like I sees 'em. I will speak to you in plain simple English.

And that brings us to tonight's word: truthiness.

Now I'm sure some of the Word Police, the wordanistas over at Webster's, are gonna say, "Hey, that's not a word." Well, anybody who knows me knows that I'm no fan of dictionaries or reference books. They're elitist. Constantly telling us what is or isn't true, or what did or didn't happen. Who's Britannica to tell me the Panama Canal was finished in 1914? If I wanna say it happened in 1941, that's my right. I don't trust books. They're all fact, no heart.

And in their hearts, everybody knows that "women talk more than men. A lot more." And blacks are lazy, jews are avaricious, celts are drunks, southerners are stupid... You can't fight truthiness -- or the pop psychology books that promote it. Don't be fooled by the long list of scientific-looking references -- The Female Brain and its ilk might be books that invoke the authority of science, but they're all heart, no fact.]

Posted by Mark Liberman at December 31, 2006 09:50 AM