Universal Grammar, Neanderthals, and the Evolution of Language—Response to Vyvyan Evans (Part 3)

Dec 1, 2015 by

[Many thanks to Martin W. Lewis and Matthew Jobin for insightful discussions of some of the issues considered in this post.]


The next set of objections to the Universal Grammar (UG) theory raised by Vyvyan Evans concerns its implications for human evolution. If UG is a species-specific innate biological capacity, it must be encoded in our genome, differentiating it from the genomes of other species. And, Evans writes, “if language is genetically hard-wired, then it self-evidently had to emerge at some point in our evolutionary lineage”. When the UG theory was originally being developed, language was generally assumed to be absent in other species—including our closest relatives, the Neanderthals—and its evolutionary emergence was dated variably between 200,000 and 50,000 years ago. The earlier date is generally assumed to correlate with the emergence of anatomically modern humans (AMHs), whose earliest fossils come from Omo (~200,000 years ago) and Herto (~160,000 years ago) in Ethiopia. (Some scholars now give a more recent date for AMHs, and consequently the date for the emergence of language would need to be understood as more recent as well.) Supporters of this earlier date for the birth of language argue that AMHs, with their lightly built skeleton, were viable only in the presence of modern behavior: our “smarts” made up for the deficiencies in our physique. According to advocates of the later date, circa 50,000 years ago, a sudden abundance of sophisticated man-made artifacts in the archeological records shows a leap in human cognitive and symbolic abilities, which is presumed to correlate with the emergence of language around that time.

Evans, however, maintains that this “sudden jump” theory (technical term: “saltationist theory”) is wrong. His objections can be summarized as follows:

  • The saltationist hypothesis is wrong as it “is at odds with the modern neo‑Darwinian synthesis, widely accepted as fact, which has no place for such large-scale and unprecedented leaps”.
  • Language ability, in evolutionary terms, is not limited to Homo sapiens and the Neanderthals must have had language too (the Neanderthals, Evans writes, “had a sophisticated material culture – including the ability to create cave engravings and produce sophisticated stone tools – not dissimilar to aspects of the human cultural explosion of 50,000 years ago”, and so they could not have “managed the complex learning and co-operation required for that if they didn’t have language”).*
  • Language could have—and did—emerge gradually “from many overlapping tendencies” revolving around cooperative behavior; essentially, as we got smarter and better at cooperating with each other, our communication system developed progressively until it became as sophisticated as it is.

Similar views are developed in Dediu & Levinson (2013).** They claim that the Neanderthals “were probably not very different biologically or cognitively from us, and that their linguistic capacities were closely similar to our own”; that “essentially modern language” was “already present in the common ancestor of these two lineages about half a million years ago (that is, five to ten times older than is often assumed)”; and that the emergence of language was “a gradual process of culture-gene co-evolution extending to the present day”.***

Before considering these issues in more detail, it is worth noting that we know relatively little about the biochemistry of language. As Hauser et al. (2014) point out, “the richness of ideas is accompanied by a poverty of evidence”, and as Dediu & Levinson readily admit, “the evidence is necessarily circumstantial”, “data […] give rise to conflicting interpretations”, and the relevant “literature […] crosses many fields and is complex, and moreover in continuous flux with the result that there are very few points of full consensus”. However, I disagree with Dediu & Levinson in that the evidence they bring up is convincing or conclusive; moreover, I think that most of what has been written about the evolution of language so far needs to be taken with a grain of salt. A high degree of uncertainty is simply unavoidable given the state of our knowledge at present and its rapid development, particularly, in the field of genetics.


Let’s now go back to Evans’ objection and consider each of them in turn, starting with how much of a “jump” the UG theory actually requires. When the UG theory was first developed, nothing was known about how language may be expressed genetically and if indeed it is expressed genetically at all. In 2002, geneticists discovered a gene, dubbed FOXP2, which is thought to be linked to language and speech. Its discovery resulted from a study of a British family with about half its members (15 people in 3 generations) affected by a strange disorder: although they are of near normal intelligence, they have certain difficulties with speech (both in production and comprehension) and especially with grammar. The familial pattern looked remarkably like an autosomal genetic disorder, and it was subsequently discovered that the affected members of the family had a “defective” variant of the FOXP2 gene (located on chromosome 7). Comparison of human and chimp genomes showed that the normal human FOXP2 gene differs from the corresponding region in chimps by just two “typos”: two single-nucleotide substitutions (or polymorphisms, or SNPs) at positions 911 and 977. Moreover, it has been shown that this gene is highly conserved amongst mammals, with only 3 non-synonymous SNPs separating humans from mice. Is this as huge a leap as Evans makes it sound? Consider this: the alignable sequences within genomes of humans and chimps differ by about 35 million SNPs. Allowing 2 out of 35,000,000 SNPs to “handle” language does not seem excessively generous to me.

One necessary caveat is that there may be other “language genes” besides FOXP2 that are undiscovered or little understood (but see Berwick, Hauser, & Tattersall 2013 for more details), and in all likelihood, there are additional differences in the regulatory DNA that are responsible for turning the “language gene(s)” on and off as necessary, a point to which I return below.

All in all, “the modern neo‑Darwinian synthesis” is not at odds with the possibility of a couple of SNPs mutating in the course of the ape evolution, with a side-effect of language emerging as a result. This linguistic ability of Homo sapiens could very well compensate for less heavily built bodies, allowing our ancestors to outsmart other ape species whose brains spent more energy on operating their more muscular bodies. (There is also some evidence that the Neanderthals had larger eye sockets, and perhaps better vision, and that more of their brain was devoted to processing visual information; see Pearce, Stringer, & Dunbar 2013.)

This brings us to the question of whether our closest evolutionary cousins, the Neanderthals, had language. Evans claims that they must have, as evidenced by their complex learning and co-operation. This is not a convincing argument, however. Members of the genus Homo, since its earliest days, have shown a slow increase in the complexity of tools; similarly, gradual increase in tool sophistication is observed in H. neanderthalensis. By the end of their existence, Neanderthals were using tools about as sophisticated as those of fully modern humans: flakes were struck from a core stone, and then carefully modified to hold a good edge. Moreover, some use of jewelry points to the existence of symbolic thinking among the Neanderthals. But many animal species show sophisticated co-operation, even rudimentary symbolic thinking, yet lack language in the sense of the human computational ability. For example, studies with chimps, such as Kanzi the bonobo and Nim Chimpsky, show that they can acquire a substantial “vocabulary” of symbols (over 250 forms, in the case of Kanzi, which corresponds to the vocabulary size of an average 19-month old human toddler). Yet, they fail to acquire anything like human syntax. Nim Chimpsky’s most sophisticated “sentence” is reported to have been “Give orange me give eat orange me eat orange give me eat orange give me you”, a far cry from human recursive embedding ability.

So did Neanderthals have syntax like us, or lacked it like chimps?

It is hard to know at present. Barceló-Coblijn (2011) argues that Neanderthals had the anatomical traits (particularly as concerns the larynx and the brain) for producing and perceiving vocalizations. A more recent study by D’Anastasio et al. (2013) confirms that internal architectures and micro-biomechanical behaviors of Neanderthal voice box were compatible with producing speech sounds. This ability to produce human-like speech sounds distinguished Neanderthals from chimps. Some scholars, and most journalists in the popular media, tend to jump from this fact to the conclusion that “Neanderthals could speak like modern humans” (as in the title of this article by BBC News science reporter, Melissa Hogenboom).

But human language ability goes beyond spoken sounds, as evidence by sign languages. As noted by Wendy Sandler and Diane Lillo-Martin, two prominent experts in the field, human sign languages “are bona fide linguistic systems, with structures and rules and the full range of expressive power that characterize spoken languages”, so much so that “the academic world is now convinced that sign languages are real languages in every sense of the term” (p. 533). Thus, as Berwick, Hauser, & Tattersall (2013) conclude, “the anatomical capacity for speech cannot by itself be taken as a proxy for language”. Our best hope, therefore, is to turn to the Neanderthal DNA.


NeanderthalsAn influential study led by paleogeneticist Johannes Krause of the Max Planck Institute for Evolutionary Anthropology in Leipzig (reported in Krause et al. 2007) has shown that DNA from two partially sequenced Neanderthal bones found inside the El Sidrón Cave in northwestern Spain (labeled 1253 and 1351c) had the same bases in the crucial positions of the FOXP2 gene (A at position 911 and G at position 977) that modern humans do. The researchers offered three alternative interpretations for their findings, two of which they deemed unlikely.

Neanderthal FOXP2The first interpretation involves the Neanderthals getting the “human” version of the FOXP2 gene through interbreeding (aka sex) and gene flow from Homo sapiens. This scenario was reckoned to be improbable because in 2007 there was no other evidence of interbreeding between the two species (although Duarte et al. 1999 described the child discovered at the Abrigo do Lagar Velho, Portugal as a Neandertal-modern human hybrid resulting from “admixture between regional Neandertals and early modern humans dispersing into southern Iberia”).**** However, since then, evidence from whole-genome studies (cf. Green et al. 2010) showed that there must have been some interbreeding after all, as contemporary European men have somewhere between 1-9% Neanderthal DNA.***** Although Green et al. (2010) show that “all or almost all of the gene flow detected was from Neandertals into modern humans”, the possibility of reverse gene flow from modern humans into Neanderthals cannot be completely excluded. Thus, these recent findings put the possibility of the Neanderthals having acquired the human version of FOXP2 through gene flow back on the table.

The second possible interpretation of Krause et al.’s results is that mutations in positions 911 and 977 occurred in the common ancestor of both H. sapiens and H. neanderthalensis and were present in some individuals, but were subject to a selective sweep only in H. sapiens. Krause et al. consider this scenario highly unlikely as well because the mutations must have been frequent enough among the ancestral species to have the observed frequency among the Neanderthals, yet such high frequency among the ancestor would preclude the observed evidence of a selective sweep. However, it must be remembered that the these calculations are based on data from very few Neanderthal remains (possibly from related individuals), and rely as well on “inherently unreliable estimates” and “parameters not known with certainty”, as Krause et al. themselves admit.

Krause et al. (2007) favor the third interpretation, namely that the “modern human” version of FOXP2 was present and selected in the common ancestor of both H. sapiens and H. neanderthalensis, meaning potentially that not only did Neanderthals themselves have language not unlike that of modern humans, but that our common ancestor did as well. That pushed the emergence of language back potentially by as much as an order of magnitude, from 50,000 years to 500,000 years ago.

Another possibility, not considered by Krause et al. (2007) is that although Neanderthals and modern humans share the protein-coding sequences in the FOXP2 gene, some regulatory elements may be different—and thus the expression of the gene would have been different in Neanderthals. In other words, although the Neanderthals may have “had the same language gene”, they may not have had full‑fledged language as in modern humans. Differences in relevant regulatory DNA have been detected by Maricič et al. (2013), although they also show that the “Neandertal” genetic variant is still present in approximately 10% of today’s Africans, meaning that it is within the modern human variation. Here, I agree with Dediu & Levinson (2013) that “more research will help clarify what these […] differences entail” and that “regulatory changes can have hard-to-predict effects, making any inferences from the identity (or not) of gene sequences to speech and language necessarily tentative”. However, unlike Dediu & Levinson (2013), who conclude that the linguistic differences between Neanderthals and modern humans are quantitative in nature and concern such parameters as “perhaps range of speech sounds or rapidity of speech, complexity of syntax, size of vocabularies, or the like”, I do not exclude the possibility that the main gap between Neanderthals and modern humans was qualitative and concerned the ability to embed structures rather than simply string sequences.

Note, however, that even if Krause et al.’s and Dediu & Levinson’s interpretations are correct in that language emerged in H. heidelbergensis, it could simply push the saltationist scenario one step back on the evolutionary tree. Perhaps we are not the first species to have a genetically-encoded language ability, but the common ancestors we share with the Neanderthals, H. heidelbergensis, was. But this possibility would still involve a sudden—if less recent—emergence of language due to a single genetic mutation, or a few correlated genetic mutations. This scenario is still a far cry from the scenario involving a gradual development of “co‑operative intelligence” over the course of more than 2 million years, which Evans and Dediu & Levinson advocate.

Moreover, adopting the hypothesis that full-fledged language developed in H. heidelbergensis and was inherited by us and our Neanderthal cousins, of course, reopens the question of why we ultimately survived and they did not. Even if some of the Neanderthal genes survived in modern humans, they did not survive as a species. Prior to the discovery that the Neanderthals had the “human” FOXP2, the scholarly consensus was that the Neanderthals died out shortly (in evolutionary terms) after H. sapiens appeared in the areas that they inhabited simply because we managed to outsmart them. Language was thought to be a key component of what allowed us to edge the Neanderthals out. But if we were of similar intelligence, but they were brawny and we were scrawny, how come we survived and they did not? As far as I can tell, this question remains open.

Equally unresolved are many other controversies that could have an impact on the question of language evolution, such as the proper interpretation of prehistoric remains that have been suggested to show signs of dual Neanderthal-Sapiens ancestry; Neanderthals’ cognitive development, brain plasticity, and infant maturation; the anatomy and physiology of their respiratory muscles, vocal tract, and hearing apparatus; and even the relationship between music and language (for some references on these issues see Dediu & Levinson 2013, and for a detailed discussion of music, language, and Neanderthals, see Mithen 2007). All in all, the jury is still out on when and how full-fledged modern language first appeared. While some of the abilities to map sound to meaning may have developed for hundreds of millennia, there is no unequivocal evidence that language-specific computational systems (syntax, semantics, phonology), based on hierarchical structures, could—or did—develop gradually.

Finally, the gradualist theory that involves a step-by-step development of cooperative behavior has no way of explaining why human language exists in so many mutually incomprehensible variants that we call “languages” (though some traditional “dialects” could fall into this category as well). If language “evolved for the purpose of communication”, as Evans indicates, why don’t we all have the same language so we could all communicate with each other? I will turn to this question in more detail in the following post.



* It should be noted that although Neanderthals left some evidence of symbolic behavior (art, burials), it appears that their behavior was not as sophisticated as that of modern humans (e.g. their material culture was relatively impoverished and changed very slowly).

**It is worth noting that, like Vyvyan Evans, Stephen C. Levinson is a renowned opponent of the UG theory (see, e.g., Evans & Levinson 2009).

***Elsewhere in the same article, Dediu & Levinson are careful to specify that what the Neanderthals shared with us is “something like modern speech and language” (boldface mine) and may not have included what Chomsky called “FLN (“faculty of language narrow”)” or “the operation of unbounded Merge”. As noted in Berwick, Hauser, & Tattersall (2013), because Dediu & Levinson’s definition of language is so vague, “it is difficult to test their evolutionary claims”.

****See also this interview given by Krause at the time.

*****It is thought that much of that gene flow may have involved genes that “conferred strong selective advantages in the out-of-Africa environment, especially in the immune system” (see Dediu & Levinson 2013 for details), which would make sense since the Neanderthals adapted to that environment for millennia before modern humans arrived.



Additional Sources:

Barceló-Coblijn, Lluís (2011) A Biolinguistic Approach to the Vocalizations of H. Neanderthalensis and the Genus Homo. Biolinguistics 5.4: 286–334.

Berwick, Robert C.; Marc D. Hauser, & Ian Tattersall (2013) Neanderthal language? Just-so stories take center stage. Frontiers in psychology 4.

D’Anastasio, Ruggero; Stephen Wroe, Claudio Tuniz, Lucia Mancini, Deneb T. Cesana, Diego Dreossi,   Mayoorendra Ravichandiran, Marie Attard, William C. H. Parr, Anne Agur, & Luigi Capasso (2013) Micro-Biomechanics of the Kebara 2 Hyoid and Its Implications for Speech in Neanderthals. PLOS One.

Dediu, Dan & Stephen C. Levinson (2013) On the antiquity of language: the reinterpretation of Neandertal linguistic capacities and its consequences. Frontiers in psychology 4.

Duarte, Cidália; João Maurício, Paul B. Pettitt, Pedro Souto, Erik Trinkaus, Hans van der Plicht, and João Zilhão (1999) The early Upper Paleolithic human skeleton from the Abrigo do Lagar Velho (Portugal) and modern human emergence in Iberia. PNAS 96(13).

Evans, Nicholas, and Stephen C. Levinson (2009) The myth of language universals: Language diversity and its importance for cognitive science. Behavioral and brain sciences 32(5): 429-448.

Green, R. E., Krause, J., Briggs, A. W., Maricic, T., Stenzel, U., Kircher, M., et al. (2010) A draft sequence of the Neandertal genome. Science 328, 710–722.

Hauser, Marc D.; Charles Yang, Robert C. Berwick, Ian Tattersall, Michael J. Ryan, Jeffrey Watumull, Noam Chomsky, and Richard C. Lewontin (2014) The mystery of language evolution. Frontiers in psychology 5.

Krause, Johannes; Carles Lalueza-Fox, Ludovic Orlando, Wolfgang Enard, Richard E. Green, Hernán A. Burbano, Jean-Jacques Hublin, Catherine Hänni, Javier Fortea, Marco de la Rasilla, Jaume Bertranpetit, Antonio Rosas, and Svante Pääbo (2007) The Derived FOXP2 Variant of Modern Humans Was Shared with Neandertals. Current Biology 17: 1908–1912.

Maricič, Tomislav; Viola Günther, Oleg Georgiev, Sabine Gehre, Marija Ćurlin, Christiane Schreiweis, Ronald Naumann, Hernán A. Burbano, Matthias Meyer, Carles Lalueza-Fox, Marco de la Rasilla, Antonio Rosas, Srećko Gajović, Janet Kelso, Wolfgang Enard, Walter Schaffner, & Svante Pääbo (2013) A Recent Evolutionary Change Affects a Regulatory Element in the Human FOXP2 Gene. Molecular Biology and Evolution 30(4): 844-852.

Mithen, Steven (2007) The Singing Neanderthals: The Origins of Music, Language, Mind, and Body. Harvard University Press.

Pearce, Eiluned; Chris Stringer, & R. I. M. Dunbar (2013) New insights into differences in brain organization between Neanderthals and anatomically modern humans. Proceedings of the Royal Society B. 280(1758).

Sandler, Wendy & Diane Lillo-Martin (2001) Natural Sign Languages. In: M. Aronoff & J. Rees-Miller (eds.) Handbook of Linguistics. Pp. 533-562.

Related Posts

Subscribe For Updates

We would love to have you back on Languages Of The World in the future. If you would like to receive updates of our newest posts, feel free to do so using any of your favorite methods below:

  • Mary Ann Walter

    Thanks for this terrific overview. The mounting genetic data makes this question more and more important and interesting. To my mind, the Neanderthal burial customs, if genuine, are more significant than the tools, jewelry or even art, as they hint at an incipient theory of mind. I also wonder if the more developed visual system of Neanderthals is a bit of indirect evidence for the primacy of sign language systems over spoken ones.

    • I am not really a specialist in the burial systems, but it’s an interesting question. As is the issue of the primacy of sign language in Neanderthals: someone mentioned this in my class tonight, but it’s another open question as far as I can tell.

  • Milo Bem

    Really cool stuff Asya. Thanks a lot.

    Asking for a gene of language is a bit like asking for a gene of walking. We may identify some mutation that makes it impossible for a person to walk but it doesn’t make it “the” gene of walking, because there’s plenty of ways of become a cripple, genetic or otherwise. Similarly there must be many genes cooperating in giving us the ability to talk but not one of them is more important than the other. Loosing either means loosing some or all of the language. Different genes for language might have existed in different population of proto-humans for millennia until they converged in one small population giving it the actual speaking ability.

    We also need to remember that the whole talk of “gene of X” and “gene of Q” are very simplified picture. Genes may have side effects completely unrelated to their original role, as can any tissue, organ or even the limb. The lungs evolved from swim bladder. The many genes of language might have some other use originally, that we may never find out. Finally, first speaking population need not have all the language ability we have now. If I remember correctly it’s Guy Deutscher (“The Unfolding Of Language”) who suggested there may be even hints for early written languages being deficient in recursion, so why not allow early humans and all neanderthals to be lacking the mental capacity of forming any complex grammar despite being able to communicate?

    Yes, walking and talking are not the same thing. One may also say that you either can talk or not but that’s clearly not the case because the are many ways of being barely communicative.

    I understand it’s a bit more controversial to compare an unfortunate individual loosing some of their speaking ability to language emerging in early human or prehuman but I would expect even very limited “language” consisting of six different grunts to be an advantage to making no sounds at all. One hundred grunts is better than six. Simple “grammar” of only dozen verbs and fingerpointing at nearby “objects” with no other parts of speech is better than unstructured set of grunts. Finally, grammar without recursion is even better, although it wouldn’t allow good translation of Twilight saga.

    I’m not suggesting that’s what had happened. I’m not a linguist. But I don’t see a reason why it couldn’t.

    • “so why not allow early humans and all neanderthals to be lacking the
      mental capacity of forming any complex grammar despite being able to
      communicate?” — I don’t see that anyone says that…

      • Milo Bem

        You seem to be arguing against what you call gradualist theory. I assumed I understood your point but I guess I didn’t. It’s quite late in my timezone, maybe I need to read the whole cycle again in daylight when I have some time.

        I admit, I don’t fully comprehend the concept of Universal Grammar as argued by professional linguists, though, even when I’m rested:). I’m an evolutionary biochemist by training and a programmer by trade.

        In my uninformed opinion, whether the evolution of mental capacity for communication was gradual or not doesn’t seem relevant to the question of why it allows so many different realization of language. Having elements of universal grammar doesn’t need to limit the number of particular implementations, we have analogical examples everywhere from biology to software.

        I guess it depends on how much of Universal Grammar is there in actual languages’ grammar rules. I imagined it to be some general framework of linguistic features for actual grammars to build upon with relative freedom. This kind of UG could be innate and still allow multitude of languages with widely differing grammars. First, simple languages could be built upon exactly the same hypothetical UG we still use, but we became more creative since the times of our interactions with neanderthals. But then I didn’t understand a single sentence from the few papers on grammar by Chomsky I tried reading during my undergrad course in computer science, long time ago.

        Only now it dawned on me that I may be arguing with a chimera of my own making of Evans’ position and yours. I noticed some, including myself, had similar problem with your earlier critique of Demoule. Maybe you could try identifying different authors in different colours so the lay readers like me have a chance of following the exciting argument:)

        • “whether the evolution of mental capacity for communication was gradual
          or not doesn’t seem relevant to the question of why it allows so many
          different realization of language.” — conceptually, probably not, but anti-generativists (like Evans) use the existence of multiple languages (which generative linguists neither deny nor disregard) as an argument against generative grammar, so that’s why it is relevant.

  • tstev

    The imprinted code for universal grammar and language looks to be universal even across other species .

    As the level of sophistication in application are required, the same type of wiring in typical mammalian brain produce similar patterns which is basically called grammar.

    “”Prairie Dogs: America’s Meerkats – Language””