From ScienceDaily:
FOXP2, a gene implicated in affecting speech and language, is held up as a textbook example of positive selection on a human-specific trait. But in a paper published August 2 in the journal Cell, researchers challenge this finding. Their analysis of genetic data from a diverse sample of modern people and Neanderthals saw no evidence for recent, human-specific selection of FOXP2 and revises the history of how we think humans acquired language.
The paper is Elizabeth Grace Atkinson, Amanda Jane Audesse, Julia Adela Palacios, Dean Michael Bobo, Ashley Elizabeth Webb, Sohini Ramachandran, Brenna Mariah Henn, “No Evidence for Recent Selection at FOXP2 among Diverse Human Populations.” Thanks, Trond! (Trond’s comment in his e-mail: “First law of science: ‘It’s more complicated than that’.”)
The elephant in the room is the word “recent.” There’s a lot of contention between “recent,” meaning in the last few hundred thousand years, and ancient, meaning a million or more years ago. Since I’m firmly in the “language is a million or more years old” camp, this is not a surprising result.
Nothing to do with FOXP2 but the list of authors agrees with something I’ve often noticed: women group leaders tend to have a high proportion of women in their groups, and its corollary, young women tend to be in groups directed by women.
One other point: I note that Brenna Mariah Henn has worked with Luigi Luca Cavalli-Sforza, whose standing in linguistics is, I gather, much lower than it is in genetics.
Here’s an important paragraph from the Results section (surprisingly, the whole paper is free):
So FOXP2 may not be that important for language after all.
I knew it.
I don’t trust genetics, because as a science, it’s still very much in infancy state, so anything they say now will be disproved in a decade or two.
I expect language to be a highly polygenic trait and I doubt we’re going to find any Chomskian great leap forward events.
FOXP2 is in no reasonable sense of the words “a gene for language.” Nasty mutations therein interfere with speech (as do quite a lot of other things) and that is about all there is to it. Mark Liberman over at LL has addressed this favourite piece of journalistic mangling of fact on several occasions.
http://languagelog.ldc.upenn.edu/nll/?p=1896
The convention of saying “gene for X” when you mean “gene which if broken prevents X” is very misleading. There is no doubt a gene which if broken causes congenital blindness, but calling it “a gene for reading” or “a gene for watching TV”, as if those were genetically determined activities, is absurd.
David E.: FOXP2 is in no reasonable sense of the words “a gene for language.”
I know, but it keeps being touted as such, not at least because of the apparent recent selection event. It’s nice to have that firmly out of the way.
@Trond:
I knew you knew … (as, I suppose, does practically everyone else around here. Still, we here in Wales are never shy of telling people things they know perfectly well already. It’s a cultural thing.)
I presume that apart from the general “gene for X” baloney, the FOXP2 as gene-for-language thing got a boost from its convenient intersection with that somewhat similar piece of pseudoscience, Chomsky’s “language organ.”
Anyhow, you’re right – it’s good to see dubious evidence for bizarre misanalyses refuted. Though with the “language organ” it’ll take more than a simple stake through the heart to do the trick.
There is no doubt a gene which if broken causes congenital blindness
Indeed yes: literally hundreds of them. Like The Right Stuff, vision “can blow at any seam.” Every reason to suppose that speech can too.
One thing to keep in mind about ‘gene for X’ claims is that selection takes place in the ancient population and at the time that a genetic trait first emerges. So, e.g., while there may be a genetic component in mathematical ability, I doubt that access to reproductive success was determined by mathematical competition among australopithecines.
MattF,
“So, e.g., while there may be a genetic component in mathematical ability, I doubt that access to reproductive success was determined by mathematical competition among australopithecines.”
This is correct all over the genome. Mutations can produce a whole range of traits, some of them completely peripheral to survival. In domestic animals, both carnivores, perissodactyls (horses) and artiodactyls (cattle, etc,) the trait that makes an animal less timid around humans and allows for domestication also has the effect of changing the pattern of color to patches like a beagle displays. Piebald horses are an example.
In humans the mutation for red hair is basically a huge reduction melanin and that gives an adaptational advantage in dark climates. it also causes freckling as a random side effect, which is neither an advantage no disadvantage.
John,
” but calling it “a gene for reading” or “a gene for watching TV”, as if those were genetically determined activities, is absurd.”
This is exactly the case for the much sought after “gay gene”. It turns out that the causes are probably epigenetic.
Apparently Chomsky never said he expected that to be a single part of the brain; it might be as discontinuous as the parts involved in any other activity. In general, from this paper (pdf) I’m getting the impression that Chomsky has only ever said surprisingly little…
Yeah. I don’t make eumelanin (black), only phaeomelanin (brown, slightly orangey).
it might be as discontinuous as the parts involved in any other activity.
That’s pretty much exactly the sort of thing about Chomskyanism I object to most. Invent a term with pretty specific implications in normal usage, but assign it a technical sense vague enough to protect against any conceivable experimental refutation (better yet, never really define it precisely at all.) Meantime, trade on the fact that your technical term sounds nicely concrete and biological and all, instead of being merely a linguistic reification of your unproved thesis, viz that all human linguistic ability is in some significant way a single language-specific capacity.
African albinos are redheads.
White people in fact have plenty of melanocytes. They just don’t actually do much.
If there weren’t so many white people in the world, Lazy Melanocyte Syndrome would undoubtedly figure in the textbooks as a congenital disease. Personally I’ve never felt that getting less rickets is enough compensation for the increased risk of skin cancer and blindness, but then I’m not a prehistoric hunter-gatherer or farmer trying to survive where the sun doesn’t shine. If I were, I dare say I would feel I was doing pretty well to live long enough for the downsides to become an issue.
Well said. See also (in the same paper): Universal Grammar not being a grammar…
Still, we here in Wales are never shy of telling people things they know perfectly well already. It’s a cultural thing.
“The genealogical trees at the end of the Red Book of Westmarch [the source document underlying The Lord of the Rings, as the Mabinogion underlies Evangeline Walton’s tetralogy] are a small book in themselves, and all but Hobbits would find them exceedingly dull. Hobbits delighted in such things, if they were accurate; they liked to have books filled with things that they already knew, set out fair and square with no contradictions.”
David E.: I knew you knew … (as, I suppose, does practically everyone else around here. Still, we here in Wales are never shy of telling people things they know perfectly well already. It’s a cultural thing.)
.. and your comment just served as a handy place to attach my “I know”, which was more to the thread in general.
@Trond:
So you knew that I knew that you knew.
Yes, sorry. I thought you knew.
I only suspected. I can’t really claim that I knew you knew I knew you knew.
My “First Law of Science” is of course The Second Law of Science.
A new page in the “language gene” saga: now it is the NOVA1 gene (regulating muscle function in many parts of the body, including where it’s important for speech).
https://www.nature.com/articles/s41467-025-56579-2
At one position of this gene, it’s different in the humans compared to other animals, or even Neanderthals.
One in several hundred thousand humans actually have the original old version of the gene, although nobody knows if those humans are any different, if there are any consequences. The authors report that the change in the humans makes NOVA1 protein more capable of regulating voice-related muscle cells. They also introduced a human version into mice, and reported that the mice squeaks changed.
Will it fizzle like FOXP2 once did?
I’m guessing yes, but I look forward to finding out!
Aaaaaaaaand the race is on.
I’m tempted to supply a quote from the extremely cheesy and chock full of stereotypes ’90s sci-fi game Sid Meyer’s Alpha Centauri. The quote is from of one the people you can play as, “Academician Prokhor Zakharov” (yes, really, including the spelling): “The genetic code does not, and cannot, specify the nature and position of every capillary in the body or every neuron in the brain. What it can do is describe the underlying fractal pattern which creates them.” The game supplies those when your faction unlocks a certain technology. Yes, all of the game is that hammy. And the voice actors are quite good.. “Zakharov” is apparently voiced by Yuri Nesterov — https://www.imdb.com/name/nm1031617/ EDIT:imbd credits his role as “proferssor” Zakharov, rather than academician.
I downloaded it from Good Old Games on a sale last New Year. Nostalgia is a… thing. Yes, I am playing it today for the first time in more than twenty years. Still good. All the Cheese and Ham of ’90s pop-sci-fi you want, in a strategy video game package. And I did just unlock that that tech, and that’s what reminded me of this thread.
I just realised this has nothing to do with the discussion except for the fact it maybe reflects on how genetics was perceived in sci-fi circles in the ’90s.
I remember the guy (easily my favorite faction, for multiple interlocking reasons only one of which was ethnic pride) as Provost Zakharov; for several years I thought that Provost was his name, before I found out that it’s a title (and, further years later, that his name was Prokhor).
TIL about his birthday: 16 September 1994 (over four years before the game came out). He’d be 31 this year.
January First-of-May Sorry, I had to off him. We had a pact, and he sent a probe team to my heaquarters to sabotage a project.
And a new chapter with the new hyped gene.
“In a study published in Nature Communications, researchers in the lab of Rockefeller researcher Robert B. Darnell discovered that when they put this exclusively human variant of NOVA1—an RNA-binding protein in the brain known to be crucial to neural development—into mice, it altered their vocalizations as they called to each other.”
https://www.rockefeller.edu/news/37279-a-single-protein-may-have-helped-shape-the-emergence-of-spoken-language/
An interesting, and fairly convincing, sequel to the FOXP2 gene story.
The gene itself, while implicated in language anomalies, hasn’t seen evolutionary changes in the human lineage, and its broad variation isn’t linked to the human language abilities today.
But …
A new preprint makes a conclusion that it isn’t about the FOXP2 protein per se, but about this protein’s targets in the genomes. And targets of several similar genomic regulator proteins as well.
These targets didn’t just evolve; most of them – hundreds of them – emerged from scratch during the murky millions of years separating the split of our ancestors from the ancestors of the chimps, and our more recent split from the Neanderthals and Denisovans. These genomic targets typically came with the more recent waves of expansion of parasitic genomic elements in the genomes of our ancestors. It’s an emerging universal mechanism of changes in genomic regulation landscapes. Basically a jumping gene-parasite lands near an important gene of the host, and alters the patterns of regulation of the host gene, especially during the embryogenesis. And indeed the authors show that FOXP2 (and its likes’) targets affecting language skills are actually regulating genes in developing neurons, especially in the medium spiny neurons known to be involved in language processing.
The authors see a fast continuing evolution in the genomic components of general intelligence throughout most of the human evolution (selection coefficient in the recent millennia reaching a very high value of 0.088, and in the times immemorial, the Neanderthals had 3 standard deviations lower polygenic scores than us). But there is virtually no ongoing evolution of language abilities! And the Neanderthals actually score a little higher than us on the authors’ polygenic score from the language abilities. Why? The authors note that the markers of language abilities also have abnormally high heterozygocity, a hallmark of balancing selection, where too much and too little are both undesirable outcomes. Presumably it’s the classic dilemma of bigger brains / dangerous childbirth.
The paper is complicated, and uses many language-skill metrics and many evolutionary partitions of the genome, rising the risk of spurious statistical associations. But it looks fairly convincing overall. They even show that very similar genetic elements are at work in non-primate species with complex vocalizations. If it holds, then we must conclude that speech is an ancient skill, quite separate from intellect and developed independently in different mammalian lineages. And that in our ancestors, it ended up developed to its logical extremes, where even more of the supposedly good thing would hurt. So it stopped evolving a long time ago.
https://www.biorxiv.org/content/10.1101/2025.03.07.641231v2.full.pdf
Interesting!
Very interesting, thanks! This is obviously not the final word on the matter, but it’s a very suggestive finding that Neanderthals and Denisovans had more of these complex-language enabling mutations than modern humans. It actually seems that we may have traded away some of our ancestral language capabilities for less risky childbirths, allowing more healthy births per woman and eventually letting us crowd our closest relatives out of their biotopes.
It also gives reason to ponder the plights of our female ancestors.
So the brutish apemen actually supplanted true humanity, by outbreeding them …
Perhaps the upper palaeolithic revolution was the outcome of the linguistically-challenged trying to replicate the beauty of an oral literature they could no longer comprehend by using material artefacts. Like a forgetful person tying knots in a handkerchief … a whole culture of aide-mémoires.
Our grandmothers evidently confronted a kind of choice of Achilles: do you want clever children, or more children with less danger?
I didn’t read the paper, but… It is obvious that there are tradeoffs everywhere. Humans are not 10 meter tall, etc. Of course, there should be tradeoffs for larger brains. Childbirths or caloric needs or maybe something else. The interesting thing is if language ability is nothing more than the brain size. But that can be checked on modern humans, no? Turgenev reportedly had a very large brain. I’ve recently heard that Hunter’s sketches are not about hunting, but about horrors of serfdom. This is hardly an improvement!
Anatole France famously had a very small brain.
https://www.scientificamerican.com/article/half-brained-schemes/
The ghastly TESCREAL crowd tend to obsess about women’s brains being generally smaller than men’s. Their own brains may be large, but they don’t seem to function very well.
Man with tiny brain shocks doctors:
I have seen this myself, in my days as a teenage neurosurgeon. (OK, neurosurgery resident.)
The patient I saw came to light because he was being investigated for headaches. (He was doing a maths degree at Cambridge.)
@ D.O.
I wouldn’t expect an individual-level correlation between high language ability and brain size, for three main reasons:
1. both of speech ability and fetal brain size are posited to be targets of thousands of small genetic effects, rather than anything like cause and effect
2. there are numerous other ways in which mother-fetal conflict for nutrient access is resolved to restrict fetal brain growth (which takes an outsize share of fetal calorie intake in the humans). It’s actually very close to my current field of study (gestational diabetes, preeclampsia, premature birth are all related to the battle between placenta and mother’s body for calories). Human placenta is unusually aggressive because of the heightened needs of the fetal brain, but mother’s immune system keeps in check by destroying too aggressively growing placental tentacles.
3. in the SPARK cohort, the biggest one where the authors checked links between genetics and language skills, their polygenic score was strongly linked with language-impairement diagnoses, but has beta of only 0.03 per standard deviation unit for general quantitatively measured language skills (implying that the top “genetic” decile is only ~5% more skilled than average) – a fairly unremarkable effect
On brain size (but see above!) and childbirth, I wonder what difference head shape makes – the unusually tall foreheads and short skulls of H. s. sapiens in this case.
@DE & Trond,
technically, the Neanderthals’ genetic “speech score” was higher but not significantly higher (scoring on average better than ~20% of today’s humans). Their “general intelligence” genetic score was quite significantly lower than ours, though.
Of course it doesn’t, by itself, mean that they were better (slightly) on speech or worse (significantly) on the smarts than us. That’s because the scores measure only the impactful genetic variation which continues to exist among us today. The Neanderthals may have had plenty of their own genetic variants which don’t exist in our gene pool anymore, or exists in too small quantities or in too faraway corners of the globe to associate with the mental abilities. Because of these unknown genetic effects, their actual propensities might have been different from what we can estimate from only the effects known to us.
And it also doesn’t mean that our ancestors who met Neanderthals ~47 thousand years ago were as smart as we are, on average, today, because our genetic components of intelligence were evolving quite fast in the more recent millennia.
Some previous posts here about missing chunks of brain:
No Brain? No Problem, April 13, 2022
The Hole in Her Head, September 8, 2022
@DP:
Indeed: I was really just fantasising on the theme.
But I do like the idea of definitively breaking the too-readily-assumed link between behavioural modernity and the genesis of Language as we currently have it. (The suggestion that any correlation might actually be negative was too delightful to ignore.)
But this is probably a straw man nowadays. I think only Chomskyite Old Believers still really subscribe to that particular Just So Story.
Dmitry Pruss, thank you for the reply. I was half-joking. Thank you for focusing on the other half.
On your number 1, of course, but it is not clear why it would make difference for inter- vs intra-species comparisons. It is interesting how different evolutionary pressures may lead to different outcomes within species compared to closely related species. I really need to read the paper…
If polygenic score for language ability in Neanderthals was larger than in humans, it might mean that speech is not all that important for the humans and evolution decided to max out on our silent contemplation abilities and left mere outward expression to deteriorate. Which would confirm Chomsky’s view on language (please, I am writing it only to annoy DE).
I decline to rise to the bait.
@Dmitry: I know, but I enjoyed the idea. We should keep reminding ourselves that modern humans don’t constitute some ideal endpoint of evolution, but is a niche adaptation that may or may not be sustainable. What may be special with us (but probably not, if we start looking) is how easily we adapt culturally to new conditions, how that cultural adaptivity can take on a life on its own and create different societies even in seemingly similar natural conditions, and how culture can set the conditions for natural selection. Intelligence itself is a culture-specific bag of abilities. They probably evolved in tandem with human culture through constant trade-offs, shedding mental (and physiological) capabilities that became less important.
Also, if what I learned decades ago still holds, (at least some) paleolithic Homo Sapiens had bigger brains than us, which can be interpreted to mean that there’s selection for more efficient wiring, i.e. doing more with less.
Re the well-functioning people with tiny brains – so either the 50s SF trope of people not using 80% of their brains is true or those persons must be impaired in some way that doesn’t show easily…
@Hans: Let’s say evolution has found a complicated way to develop all this new functionality in the brain, but it comes at the cost of expensive and inefficient brainmass. Optimizing the brainmass through evolutionary mechanisms turns out to be difficult and takes a long time. That doesn’t mean a more efficient brain can’t be stumbled upon accidentally once in a while. Which could be just what evolution is waiting for, come to think of it — if the reason is genetic, and it comes without forbidding costs elsewhere.
For instance, the big brain might be scaffolding for development and the finished building might be able to stand without it.
I like that image.
The 50s SF trope and 00s woo trope of people not using 90% of their brains is a misunderstanding of the fact that we don’t use most of the brain at the same time – when we do, that’s apparently an epileptic seizure.