The “Language” Gene and Women’s Wagging Tongues
Aka, How to Twist Science to Reinforce Gender Stereotypes
Note: this article first appeared as a guest blog post in Scientific American. Not surprisingly, some were dissatisfied: primarily, those who still like to think that genes determine higher-order behavior and that “gender” differences are hardwired and extensive. Excerpt of an interminable pseudo-learned comment at the SciAm blog: “In fact, it can be argued that the differences between genders is far more distinct and pervasive than the differences between species.” Satoshi Kanazawa, is that you?
Stylized rendering of FOXP2 attached to DNA (Wikipedia, CCL)
Genes are subject to multiple layers of regulation. An early regulatory point is transcription. During this process, regulatory proteins bind to DNA regions (promoters and enhancers) that direct gene expression. These DNA/protein complexes attract the transcription apparatus, which docks next to the complex and proceeds linearly downstream, producing the heteronuclear (hn) RNA that is encoded by the gene linked to the promoter. The hnRNA is then spliced and either becomes structural/regulatory RNA or is translated into protein.
Transcription factors are members of large clans that arose from ancestral genes that went through successive duplications and then diverged to fit specific niches. One such family of about fifty members is called FOX. Their DNA binding portion is shaped like a butterfly, which has given this particular motif the monikers of forkhead box or winged helix. The activities of the FOX proteins extend widely in time and region. One of the FOX family members is FOXP2, as notorious as Fox News – except for different reasons: FOXP2 has become entrenched in popular consciousness as “the language gene”. As is the case with all such folklore, there is some truth in this; but as is the case with everything in biology, reality is far more complex.
FOXP2, the first gene found to “affect language” (more on this anon), was discovered in 2001 by several converging observations and techniques. The clincher was a large family (code name KE), some of whose members had severe articulation and grammatical deficits with no accompanying sensory or cognitive impairment. The inheritance is autosomal dominant: one copy of the mutated gene is sufficient to confer the trait. When the researchers definitively identified the FOXP2 gene, they found that the version of FOXP2 carried by the KE affected members has a single point mutation that alters an invariant residue in its forkhead domain, thereby influencing the protein’s binding to its DNA targets.
Like all transcription factors, FOXP2 regulates many promoters. The primary domains of FOXP2 influence are brain and lung development. Some of its downstream targets are themselves regulators of brain function (most prominently neurexin CNTNAP2). Not surprisingly, deleting or mutating both FOXP2 copies in mice results in early death, whereas doing so to one copy leads to decreased vocalization and slightly impaired motor learning. FOXP2 is broadly conserved across vertebrates, but its critical functional regions have tiny but telling differences even between humans and their closest ape relatives. Like other genes that influence human-specific attributes, human FOXP2 seems to have undergone positive selection during the broad intervals of crucial speciation events. Along related lines, Neanderthals and Denisovans apparently had the same FOXP2 allele as contemporary humans, and by this criterion were fully capable of the articulation that makes language possible.
Which brings us to the nub of the issue. What does FOXP2 do in brain? Genes don’t encode higher-order functions, let alone behavior. Also recall that the KE family members have a very circumscribed defect, despite its dramatic manifestation. Finally, keep firmly in mind that language in humans includes a complex genetic component that involves many loci and just as many environmental interactions. FOXP2 does not encode inherent language ability. Instead, the time and place of its expression as well as studies in cell systems and other organisms (zebra finches, rodents) indicate that FOXP2 may be involved in neuronal plasticity, which in turn modulates capacity for learning by forming new synaptic connections. FOXP2 may also be involved in regulation of motor neuron control in certain brain regions (cortical motor areas, cerebellum, striatum) that affect the ability to vocalize, sing and, in humans, form the complex sounds of language.
Given its connection, however over-interpreted, to “what makes a human” as well as its chromosomal location (in 7q31, which also harbors candidates for autism and dementia), it’s not surprising that FOXP2 has acquired quasi-mythic dimensions in the lay imagination. However, careful studies have shown that the genes on 7q31 responsible for autism and dementia are distinct from FOXP2. Also, as I said earlier, FOXP2 does not code for language ability – and even less for its culturally determined manifestations (many of which are a minefield of confirmation biases, unquestioned assumptions and simply sloppy work).
The latest round in the misrepresentation of FOXP2 is the gone-viral variation of “there’s more of this ‘language protein’ in the left hemisphere of 4-year girls and that’s why women are three times as talkative as men”. This came from the PR pitch of a research team who did a study primarily on rats (which confirmed the link between FOXP2 levels and vocalization) and then, perhaps attempting to latch onto a catchy soundbite, extended the gender link to humans based on… a single PCR amplification of ten Broca’s area cortices (from postmortem brains of 4-year olds, five from each sex; Broca’s area is involved in language processing).
To begin with, all studies conducted so far definitively show that women and men utter the same number of words by any metric chosen – and that in fact men talk more than women in mixed-gender conversations (to say nothing of the gender-linked ratio of interruptions). And whereas it’s true that girls develop vocal competence slightly earlier than boys and show higher linguistic skills during the early acquisition window, this difference is transient. Furthermore, the FOXP1 control that the authors of the study argue does not show a gender-correlated change (unlike FOXP2) in fact is on the verge of doing so, and the relative statistical significances might well change if a larger number of samples were tested. Finally, whereas decrease of FOXP2 reduces vocalization and increases pitch in male rat pups, it has the opposite effect in female rat pups. In other words, the correlation between FOXP2 levels and vocalization/pitch is not straightforward even in rats.
In the larger context of expression and reception of vocalizations, the difference is not how much women talk, but how welcome and/or valued their input is. Even trivial zomboid blathering is given higher value if it’s culturally coded as masculine (examples: sport newscasters; most congressmen). In fairness to the researchers of the study that caused all this rehashing of kneejerk stereotypes and evopsycho Tarzanism, here is the concluding paragraph of their paper. It states something both measured and, frankly, obvious:
“Gender is a purely human construct consisting of both self and others’ perception of one’s sex and is arguably the first and most salient of all phenotypic variables. Sex differences in how language is received and processed and how speech is produced has the potential to influence gender both within and external to an individual. Whether human sex differences in FOXP2, and possibly FOXP1 as well, contribute to gender variation in language is a question for future research.”
Relevant publications and links:
Lai CS, Fisher SE, Hurst JA, Vargha-Khadem F, Monaco AP (2001). A forkhead-domain gene is mutated in a severe speech and language disorder. Nature 413(6855):519-23.
White SA, Fisher SE, Geschwind DH, Scharff C, Holy TE (2006). Singing mice, songbirds, and more: models for FOXP2 function and dysfunction in human speech and language. J. Neurosci. 26(41):10376-9.
Bowers JM, Perez-Pouchoulen M, Edwards NS, McCarthy MM (2013). FOXP2 mediates sex differences in ultrasonic vocalization by rat pups and directs order of maternal retrieval. J. Neurosci. 33(8):3276-83.
Mark Liberman. Gabby Guys: The Effect size (Language Log, Sept. 23, 2006)
Mark Liberman. An Invented Statistic Returns (Language Log, Feb. 22, 2013)
Athena Andreadis. Eldorado Desperadoes: Of Mice and Men (Starship Reckless, July 18, 2009)
Athena Andreadis. Miranda Wrongs: Reading Too Much into the Genome (Starship Reckless, June 10, 2011)
An eloquent debunking AA.
Thanks AC!
We did a pretty thorough analysis of the KE case in my class last semester, and oh the loathing our professor had for the moniker “the language gene.” It might be pertinent to mention that even within the case study the FOXP2 deleterious effects were somewhat randomized, based on where in the promoter/inhibitor/gene sequence the mutation expressed itself. That, of course, did not stop several students continuing from that class to my Animal Behavior class with the idea that if an organism has the FOXP2 gene it should be able to speak human language.
In short, there really is never an easy answer in genetics, and anything that sounds so simple as “Gene X controls Development Y” should be immediately suspect.
Yes, such invalid linkages drive scientists crazy.
“…if an organism has the FOXP2 gene it should be able to speak human language.” Such “not even wrong” ideas are embedded in transhumorist concepts of gene function and animal uplift (and of course their Hollywood depictions — for example, the Planet of the Apes reboot).
Interestingly, I had read and enjoyed this piece (over on SA) before you mentioned it on twitter, *without* happening to notice that it was yours. (Don’t ask me how I managed to miss your name!) It was before the troll invasion, too, I think.
Those trolls were just sad. But I did enjoy the bit where you linked the tantrum statue!
They also showcase that men talk at least as much as women — to say nothing of the quality of the product!
Eloquent debunking article, Athena. Relatively simple observations of TV stations and internet chats prove that meaningless blathering is both non-gender-specific and widespread. *laughs*
I find it a bit disturbing that scientific researchers will stoop to misrepresenting their research in hopes of gaining more publicity. When you simplify (read: distort) science for media appeal, it becomes meaningless.
In fairness to the researchers, they didn’t do the extension to “women talk more” — the media did that. But their discussion section in the paper is annoying because it implies far more than their results warrant.
You raise a good point about labeling. As the old joke goes, “a biologist would study a radio by pulling out its parts.” Inevitably that leads to description like “a volume control part”, even if that is part of a system that results in the change in volume. Its why so many genes have names that are meaningless except in their original context, and even then are just convenient labels.
I was listening to a neuro talk just recently where the researcher was talking about “reward” and “punishment” neurons in the frontal cortex, even though it is believed that these neurons change their role as new circuits form for new training. While this may be a shorthand, it seemed mislabeling to me.
I’m not sure what the answer is when dealing with complex systems, especially brains, but it might be helpful for future research if naming parts was done more carefully.
Labeling is important, I fully agree. Also important is not to say something blatantly incorrect in the service of “simplification” because soundbites tend to lodge in people’s minds.
“because soundbites tend to lodge in people’s minds.”
That is the beauty of memes. But it results in a problem. Do you try to state complex truths that do not replicate well, or simple part-truths that do? While scientists should try to promulgate truth, the competition for funding may well result in untruthful memes being spread. We can see how effective that is with corporate sponsored spreading of untruthful memes in a number of scientific areas.
Sentence 1: “During transcription, DNA is copied into mRNA.”
Sentence 2: “During transcription, DNA is copied into RNA.”
Equally short and equally memorable, except that #1 is completely wrong: mRNA requires splicing, and only ~5% of RNA is mRNA. I agree that complexity of any kind, not just of science, is not readily amenable to soundbites. However, if people can figure out racing odds, they can figure out scientific concepts without being spoon-fed simplistic (not the same as simplified) half-truths.