Language Log

The main job of the girl brain

Louann Brizendine's book The Female Brain has "80 pages of notes and references supporting 190 pages of text", as Deborah Tannen observes in the Washington Post ("A Brain of One's Own", 8/20/2006). At least with respect to matters of speech, language and communication, however, the scientific support for Brizendine's picture of biological determinism seems to slip through the fingers as you try to grasp it.

I've discussed this is some earlier posts, starting with "Neuroscience in the service of sexual stereotypes" (see also "Leonard Sax on hearing" for additional discussion of a related issue.) This post will take up a small series of connected examples, in an attempt to illustrate the frustrating process of trying to use Brizendine's voluminous references to substantiate her claims.

Let me stress that I'm not cherry-picking problems here -- I'm just working through the material in Brizendine's book that deals with speech, language and communication. Let me also stress that I have no a priori beliefs or preferences about the existence of sex differences in cognition, nor about the roles of nature and nurture in creating those differences.

One of Brizendine's themes is that the female brain is "[a] machine ... built for connection." [p. 21] "That's the main job of the girl brain, and that's what it drives a female to do from birth. This is the result of millennia of genetic and evolutionary hardwiring... [S]ince you're smaller, you probably need to band with other females to fend off attacks from a ticked-off caveman..." On page 36, she reiterates this point with some modern numbers:

Many women find biological comfort in one another's company, and language is the glue that connects one female to another. No surprise, then, that some verbal areas of the brain are larger in women than in men and that women, on average, talk and listen a lot more than men. The numbers vary, but on average girls speak two to three times more words per day than boys. We know that young girls speak earlier and by the age of twenty months have double or triple the number of words in their vocabularies than do boys. Boys eventually catch up in their vocabulary but not in speed. Girls speak faster on average -- 250 words per minute versus 125 for typical males.

Her book doesn't have footnotes; instead, at the end, each chapter has a set of end-notes that are connected to the text by quoted fragments associated with schematic references. So turning to p. 195, we find that those five sentences are supported by five references:

36 ". . . talk a lot more than men.": Light 2005; Knaus 2004; Witelson 1995; Hyde 1988.
36 ". . . vocabularies than do boys.": Hyde 1988.
36 ". . . versus 125 for typical males.": Ryan 2000.

To resolve those references, we turn, in turn, to the bibliography (pp. 211-269).

1. The first of the five papers that she cites is Light, K.C., K.M. Grewen, et al. (2005). "Oxytocinergic activity is linked to lower blood pressure and vascular resistance during stress in post-menopausal women on estrogen replacement." Horm Behav 47(5): 540-48.

I think this reference must have slipped in by mistake. There's nothing in it about verbal areas of the brain, or about women finding comfort in one another's company; nor is there anything about amounts, rates or content of talking or listening by people of any sex or age. In fact, there's nothing much in it about speech or language or communication at all, except that one of the "stressors" that was used to raise blood pressure was "a Speech task that involved 2 min of silent preparation, then 3 min of giving a tape-recorded speech on a recent interpersonal interaction". This increased the subjects' blood pressure somewhat more than putting ice on the forehead did -- the "recent interpresonal interaction" was supposed to be one that made them angry -- and about as much as taking a cognitively difficult test on the computer did. Otherwise the paper is just about, well, oxytocinergic activity, blood pressure, estrogen replacement and so on, just like the title says. Here's the abstract, in case the link above doesn't work for you (increase the font size in your browser if you really want to read it):

Estrogen administration results in increased release of the oxytocin (OT) prohormone reflected by increases in oxytocin intermediate peptide (OT Int) in both animal models and humans, and sequential treatment of ovariectomized rats with estrogen/progesterone then progesterone withdrawal leads to increased hypothalamic OT mRNA. Blood pressure (BP) reductions have been related to increased exogenous and endogenous OT in rats and to higher endogenous OT activity in premenopausal women, but not previously in postmenopausal women. Thus, we used plasma obtained at rest and during a speech stressor from 54 postmenopausal women who participated in a 6-month randomized trial of oral conjugated estrogens vs. placebo to examine effects of estrogen replacement therapy (ERT) on plasma OT and OT Int levels and their relationships to changes in BP during the trial. ERT alone and with progesterone (but not placebo) led to significant increases in plasma levels of OT Int, but no change in plasma OT levels. Women showing greater increases in OT Int during treatment showed greater decreases in BP and total vascular resistance during a series of behavioral stressors compared to women with moderate or no increases in OT Int, even after controlling for effects related to treatment condition or to changes in plasma estradiol. The findings suggest that enhanced oxytocinergic activity may contribute to BP decreases associated with ERT in more responsive postmenopausal women.

2. The second paper that Brizendine cites is Knaus, T.A., A. M. Bollich, et al. (2004) "Sex-linked differences in the anatomy of the perisylvian language cortex: A volumetric MRI study of gray matter volumes." Neuropsychology 18(4):738-47. This one is relevant to one of Brizendine's claims -- it's about the relative size in men and women of some language-related brain areas. Unfortunately, its findings directly contradict her assertion that "some verbal areas of the brain are larger in women than in men".

Knaus et al. looked at 8 language-associated brain areas (the left- and right-hemisphere instances of Heschl's gyrus (HG), the planum temporale (PT), the posterior superior temporal gyrus (pSTG), and the posterior ascending ramus (PAR)). They found that all 8 areas were substantially larger, on average, in their 12 male subjects than in their 12 female subjects. Because total brain volume was about 26% larger on average in their male subjects than in their female subjects, they adjusted the brain-region measurements with respect to each subject's total brain volume. (The magnitude of the TBV difference in their study seems surprisingly large -- I would have expected something more like 10%. Their table of raw and adjusted results is here.) In the adjusted measurements, 6 of the 8 areas were on average larger in their male subjects, while 2 of the 8 areas were on average larger in their female subjects. (Those two areas were Heschel's gyrus in the left hemisphere, and the posterior ascending ramus in the right hemisphere.)

So Brizendine's statement "some verbal areas of the brain are larger in women than in men" is simply false, at least with respect to the data in this study. To make it true, her clause would have to be amended to read something like "all verbal areas of the brain are larger on average in men than in women, but when area sizes are adjusted for total brain volume, women came out ahead in two out of eight areas examined in one study". (A tabular survey of the -- variable and inconclusive -- results of other studies of sex differences in language-related brain areas, reproduced from the Knaus et al. paper, can be found here.)

3. The third paper that Brizendine cites is Witelson, S.F. (1995). "Women have greater density of neurons in posterior temporal cortex." J Neurosci 15(5, Pt. 1):3418:28. Since this study is mainly about neuron density, about which Brizendine says nothing, it doesn't bear directly on anything that she says. The most relevant thing in this study is "Due to the small sample size and the homogeneity of the cases studied, generalizability of the results requires replication by other studies." (The sample size was 5 women and 4 men, with a mean age of 50.) The second most relevant thing, in my opinion, is a comment on the cortex-wide difference in neuron density reported in other studies: "The magnitude of 11% for the sex difference in N_V for the total cortex corresponds closely to the 10% sex difference in brain size (Dekaban and Sadowsky, 1978). Such a result raises the hypothesis that one possible cause of the greater N_V in women is a simple mechanical compression or geometric consequence of a smaller brain." (N_V is the number of neurons per unit volume.) The details of the Witelson study are extremely complicated. I'll just say that given the tiny sample, the large amount of individual anatomical variation, and the large number of different measurements made, it's probably premature to try to put much meaning into the fact that some parts of posterior temporal cortex seem to show sex differences (in various directions!) in depth, size and neuron density, while other parts don't.

4. The fourth paper that Brizendine cites is Janet Shibley Hyde and Marcia C. Linn, "Gender Differences in Verbal Ability: A Meta-Analysis", Psychological Bulletin, 104:1 53-69 (1988). This paper doesn't discuss brain size, nor does it discuss whether "women, on average, talk and listen a lot more than men". However, it does relate to Brizendine's general perspective that the female brain is "a machine built for connection" (in a way that the male brain isn't), or at least to the aspect of this hypothesis that is engaged by psychological measures of verbal ability. Here's the start of Hyde and Linn's abstract:

Many regard gender differences in verbal ability to be one of the well-established findings in psychology. To reassess this belief, we located 165 studies that reported data on gender differences in verbal ability. The weighted mean effect size (d) was +0.11, indicating a slight female superiority in performance. The difference is so small that we argue that gender differences in verbal ability no longer exist. [emphasis added]

Hyde and Linn are using "effect size" to mean the difference in group averages, normalized by the group standard deviations. (To be more precise, I think that their "weighted mean effect size" is what the wikipedia entry on effect size calls "Hedges' ĝ"). As the wikipedia article explains, "0.2 is indicative of a small effect, 0.5 a medium and 0.8 a large effect size". In a later post, I plan to take this up in more detail, comparing the effect sizes for sex differences in cognition, sex differences in height and weight, and looking at some recent research on the effect sizes for cognitive differences associated with socio-economic status. [Update: that discussion is here.]

Brizendine's end-notes for p. 36 cite "Hyde 1988" twice:

36 ". . . talk a lot more than men.": Light 2005; Knaus 2004; Witelson 1995; Hyde 1988.
36 ". . . vocabularies than do boys.": Hyde 1988.

The second citation apparently covers the stretch of her text that reads "The numbers vary, but on average girls speak two to three times more words per day than boys. We know that young girls speak earlier and by the age of twenty months have double or triple the number of words in their vocabularies than do boys." But there's nothing in Hyde and Linn 1988 about how many words anyone speaks per day, nor is there anything to support that assertion that "young girls ... by the age of twenty months have double or triple the number of words in their vocabularies". The only thing that Hyde and Linn say about developmental trends is this:

Meta-analysis is capable of detecting age trends in the magnitude of gender differences. For example, Hyde (1984) found that gender differences in aggression were twice as large for preschoolers (d = .58) as for college students (d = .27). The present analysis, however, found little evidence of age trends in the magnitude of gender differences, either when considering the evidence from all measures of verbal ability combined or when considering two particularly frequently studied aspects of verbal ability, vocabulary and reading comprehension (see Table 6). The majority of the effect sizes are .11 or less. The largest value is −.26 for vocabulary measures with 6- to 10-year-olds, based on nine studies. This finding of male superiority can be traced to four studies (Buswell, 1980; Corah, 1965; France, 1973; and Rebecca, 1974), all of which found moderate to large differences favoring males. Of those studies, two of the four were unpublished, and three of the four had rather small sample sizes (N s = 36, 60, and 40). It is difficult to say whether there is sufficient evidence of this effect to warrant pursuing it with further research.

The only other paper by Hyde in Brizendine's bibliography is that 1984 study on the development of aggression, which has nothing in it that bears on the claims about speech and language in the passage we're discussing.

5. The fifth paper that Brizendine cites

36 ". . . versus 125 for typical males.": Ryan 2000.

is 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). I discussed this case at (excessive) length in an earlier post ("Sex and speaking rate"). You can read more about it there than you probably want to -- let's just say here that there's nothing whatsoever in that paper comparing speaking rates of boys and girls. I concluded that

I'm at a loss to see how Prof. B. can interpret anything in this paper as support for the view that "girls speak ... 250 words per minute versus 125 for typical males". Perhaps her research assistant pulled the wrong index card for that talking point?

So let's review the connections between Brizendine's five sentences and five references:

Many women find biological comfort in one another's company, and language is the glue that connects one female to another.

This is certainly true for many women, but it's also true for many men. It's even true for mixed-sex groups. Brizendine offers no references to support this generalization -- not that any are needed, except to the extent that she argues by implication that men find less comfort in company, and make less use of language as social glue.

No surprise, then, that some verbal areas of the brain are larger in women than in men

This appears to be false. Some verbal areas of the brain are less smaller in women than the brain as a whole is. Some aren't. What this means, if anything, is unclear.

and that women, on average, talk and listen a lot more than men.

There is no evidence relevant to this in the five references that Brizendine provides for this passage.

We know that young girls speak earlier and by the age of twenty months have double or triple the number of words in their vocabularies than do boys.

There is no evidence relevant to this in the five references that Brizendine provides for this passage.

Boys eventually catch up in their vocabulary but not in speed. Girls speak faster on average -- 250 words per minute versus 125 for typical males.

There is no evidence relevant to this in the five references that Brizendine provides for this passage.

Frankly, I feel a little silly devoting so much verbiage to investigating these five little sentences and five little references. However, the issues about language and gender that Brizendine raises are serious and important ones, and the evidence that she presents all seems to be as illusive as these examples have turned out to be. In that situation, it seems wrong to simply dismiss what she says without trying to engage it seriously -- but it's hard to do that briefly.

So let's sum up. Of the five references that Brizendine provides for this passage, two (Light 2005 and Ryan 2000) have no connection whatever to its content; one (Witelson 1995) is vaguely relevant but does not directly address anything that she says, and is a small study with very complex and hard-to-interpret results; and two (Knaus 2004 and Hyde 1988) directly contradict either the letter or the spirit of her assertions.

Coming at it from the other side, let's list the seven assertions in this short passage that are susceptible to empirical test:

1. some verbal areas of the brain are larger in women than in men
2. women, on average, talk and listen a lot more than men
3. on average girls speak two to three times more words per day than boys
4. young girls speak earlier and by the age of twenty months have double or triple the number of words in their vocabularies than do boys
5. Boys eventually catch up in their vocabulary
6. but not in speed.
7. Girls speak faster on average -- 250 words per minute versus 125 for typical males.

Point #1 is contradicted by one of the references she cites, and none of the other assertions are addressed one way or the other. I'm quite certain, based on results from other studies discussed here, that point #7 is spectacularly false, and point #6 as well; and I'm also fairly certain that points #2 and #3 are false.

As for point #4, none of the citations offered for this passage addresses it, but it is more or less true, at least of average values. For example, according to Svetlana Lutchmaya, Simon Baron-Cohen and Peter Raggat, "Foetal testosterone and vocabulary size in 18- and 24-month infants", Infant Behavior and Development, 24:418-424 (2002), in a sample of 18-month-olds,

For boys, vocabulary size ranged from 0.0 to 222.0, M = 41.8 (SD = 50.1). For girls vocabulary size ranged from 2.0 to 318.0, M = 86.8 (SD = 83.2).

while at 24 months,

For boys, vocabulary size ranged from 0.0 to 414.0, M = 196.8 (SD = 126.8). For girls vocabulary size ranged from 15.0 to 415.0, M = 275.1 (SD = 121.6).

Thus at 18 months, the girls' average vocabulary was roughly double the boys' average. Given the cited variation, if we were to pick a girl and boy at random from their sample, we would expect the girl to have a larger vocabulary about 68 times out of hundred, while the boy would have the larger vocabulary about 32 times. At 24 months, the girls' average vocabulary was only about 40% greater. The betting odds haven't changed much, though -- a random girl would have a larger vocabulary than a random boy about 67 times out of a hundred.

And point #5 (that boys catch up in vocabulary) is also true, as Table 6 from Hyde 1988 shows:

Overall, none of the seven factual assertions in this passage is supported by the references that she provides for it. One of the assertions is contradicted by a reference she gives, and four seem to be contradicted by studies she doesn't cite. Two of the seven assertions seem to be true, based on research that she doesn't cite (at least not in support of this passage).

This pattern of disconnection between assertions and cited science is not what I expect from a book with "80 pages of notes and references supporting 190 pages of text", written by a professor at UCSF, one of the world's foremost biomedical research institutions. But I'm afraid it seems to be all too common in what Leonard Sax calls "the emerging science of sex differences".

[More discussion and links here.]

Posted by Mark Liberman at September 2, 2006 07:49 AM