Signal to Symbol.

The MIT Press Reader presents excerpts from a new book with a theory of how language evolved (yes, yet another one); here’s the introduction:

In their book “From Signal to Symbol,” Ronald Planer and Kim Sterelny propose a novel theory of language: that modern language is the product of a long series of increasingly rich protolanguages evolving over the last two million years. Arguing that language and cognition coevolved, they give a central role to archaeological evidence and attempt to infer cognitive capacities on the basis of that evidence, which they link in turn to communicative capacities.

If protolanguages began as largely gestural systems, Planer and Sterelny ask in the excerpt from the book featured below, why and how did vocalization become so important? They meet that challenge through the idea of a “firelight niche” — a term adapted from a phrase used by anthropologist Polly Wiessner in a 2014 article analyzing the fireside conversations of the Ju/’hoan (!Kung) Bushmen of South Africa — and the changed social and physical environments that came with the control of fire. In their view, selection for something like wordless singing and laughter led to improved vocal control. These behaviors helped to ease tensions and strengthen affiliative bonds as hominin social life became more complex and intense. With more vocal control available, the vocal channel offered various efficiencies, which were particularly salient at the fireside, in the firelight niche.
–The Editors

The excerpts themselves begin with an excursus on the contrasts between humans and great apes in feeding time (“Chimpanzees and orangutans, it is estimated, spend around 7 hours per day feeding, while gorillas spend some 8.8 hours per day on this activity”), then continues:

Perhaps unexpectedly, this excursion into the prehistory of cooking is relevant to the evolution of speech. First, a morphology ideal for chewing and swallowing tough food is not ideal for controlled articulate speech, and vice versa. Making food softer and easier to fragment eases constraints on the mouth, jaw, and vocal tract imposed by the need to chew through tough food. Once food is processed, there is less of a tradeoff between a system optimized for chewing and swallowing and one optimized for speaking, though there is still some tradeoff. […]

How significant for the evolution of speech was the availability of this extra time? Again, that is difficult to estimate. If hominin communication was brief and staccato, long feeding times would not inhibit the use of the vocal channel. If communication involved extended bouts, then the time freed looks more important. This issue becomes central in the next section, where we focus on the role of laughter, song, and talk in social cohesion. These are not brief and staccato uses of the mouth, and we suggest they have ancient roots, with the erectines. If and to the extent that such communication was central to social bonding and defusing conflict, the relaxation of the feeding-time constraint mattered.

Just as we argue in an early chapter of our book that the cognitive capacities that made elaborated gesture possible evolved to control skilled action, we believe the capacities that made speech possible — precise top-down control of vocalization, and the capacities to attend, recognize, and match others’ vocalizations — primarily evolved for song, or song’s evolutionary antecedents. They were then co-opted in the evolutionary transformation of multimodal communication, as the vocal component became more prominent and its role changed.

It is indeed possible that the breath control on which laughter now depends evolved for singing (or speech). But we find the suggestion that laughter was an early and important mechanism of social bonding plausible. […]

If laughter bonds more efficiently than grooming, singing is more efficient still. The idea that song was a precursor to human speech has a long history — Darwin himself proposed such a view. But it has typically been developed in the context of sexual selection, with males singing to impress females. Dunbar’s account is distinctive in suggesting that singing functioned primarily to promote group cohesion. (To the best of our knowledge, this theory dates back to 1993.) As Dunbar explains, singing also triggers the release of endorphins. But in addition, when performed in a group context, it tends to produce feelings of belongingness. Moreover, as the ethnographic record shows, people frequently sing in groups much larger than three — indeed, an entire band might sing together. Our view differs from both Dunbar and Darwin in that we think that while singing is a precursor to speech, it is not a precursor to language. Rather, singing helps explain the transition from gesture dominance to speech dominance. […]

Hominins who can sing, albeit wordlessly, and who can use a reasonably rich, largely gestural protolanguage, though with some vocal elements, are poised to make a transition to a primarily vocal system, if selection favors such a transition. To the extent that earlier systems are multimodal, they are already primed with the realization that sound can carry or modulate meaning. They have voluntary control over sound production that suffices for reasonably precise and extended sequences. They have the cognitive tools to recognize, respond to, and learn to reproduce others’ sound sequences. It is true that such a shift surrenders the advantages of iconicity. But gestures in regular use become stylized and conventionalized; even chimp gestures do. Once gestural protolanguages are well established, their stability does not depend on iconicity, though presumably it remained and remains an advantage in coining new signs.

At some stage in our evolutionary history there was a transition to a primarily vocal mode. After all, talking offers some obvious advantages over signing. It allows us to freely use our hands for other tasks while still communicating. That is particularly important when we are coordinating effort in some collective manual task, like shifting a heavy or awkward object through a cluttered environment. It enables us to communicate over longer distances and in the dark. It is immediately attention-grabbing. It is less physically demanding, and it allows us to fully visually attend to and act on our environment while listening and talking. It is thus no surprise that humans everywhere primarily use speech rather than gesture to linguistically communicate unless forced to do otherwise. (In hunting, for example, silence is often important.) However, it is not enough to merely list those advantages. We need to detail a context in which those advantages were salient enough to drive a transition. We have already hinted at our pick for that context: the control of fire. […] In our view, the control of fire was a major — if not the major — driver of the gesture-speech transition.

There is a great deal more about the fireside, but you get the general idea. Then they discuss the archaeological and genetic evidence:

Given this timeline, our account of the gesture-speech transition predicts that mid-Pleistocene Homo experienced selection for enhanced vocal communication. In particular, we would expect to see changes in Homo heidelbergensis. How does this prediction fare? In assessing it, we have two main sources of evidence to go on: fossil and genetic. There is mixed opinion regarding the potential for fossil evidence to tell us about the evolution of speech. Historically, proposed indicators have included cranial shape and dimensions, the size of the hypoglossal canal, the angle of the base of the skull, the morphology of the hyoid bone, and the width of the thoracic vertebral canal. Virtually all contemporary discussion revolves around the last two, the others having been dismissed as too unreliable. […]

In sum, while the genomes of modern humans and Neanderthals are strikingly similar, a number of changes have occurred in our line since our split with the Neanderthals some half a million years ago. Of these changes, a small number may be implicated in the development of human speech capacities. But on the whole, there is little to suggest a major architectural difference between human and Neanderthal vocal tracts or neuroanatomical structures relevant to speech. Rather, these differences appear more like “finishing touches” to an already sophisticated system supporting speech production and perception capacities, a conclusion further supported by the fossil evidence mentioned above. Moreover, differences in sapiens speech compared to that of Neanderthals need not mean that Neanderthal capacities were limited or impaired compared to those of sapiens. All things considered, the evidence suggests that the capacities for speech were present in the mid-Pleistocene ancestor of humans and Neanderthals. The capacity for speech was probably available to these hominins, the heidelbergensians, as they settled into the fireside niche.

Longtime readers will not be surprised to hear that I view all such suggestions with the deepest skepticism, but they are certainly interesting. Thanks, Martin!


  1. David Eddyshaw says

    Moreover, differences in sapiens speech compared to that of Neanderthals

    We have no evidence that there were any at all. It’s all pure, unadulterated speculation.

  2. David Marjanović says

    I think these unspecified differences in speech are inferred from “these differences” in the preceding sentence, which are unspecified but presumably measurable anatomical differences that are, as the sentence before that assures us, not “a major architectural difference”.

    Bad writing. It looks like the inventor of the terms “the erectines” and “the heidelbergensians” felt under pressure to shorten the text and did it the wrong way.

  3. I didn’t like “erectines” at all.

  4. Stu Clayton says

    … the heidelbergensians, as they settled into the fireside niche.

    When I lived in Heidelberg at the end of the 60s, people heated with coal and oil stoves, which get very hot. Settling in next to one was suicidal.

    Meals were taken at a table, and the tv did most of the talking. I never saw a fireside there, must be a status symbol.

  5. Next from the authors, a sadly under-stimulating book about where the Homo erectus line fell flat – Erectine Disfunction in Early Hominids.

    They explain the lack of hard facts for their theory on the rarity in the fossil record of soft tissue.

  6. David Eddyshaw says

    Despite the heading, the entire collection of essays “The cradle of language” is available at this link. (There is probably no real need to read Bonny Sands’ and Tom Güldemann’s chapter on click languages to guess their conclusion, giving that they actually know what they are talking about.)'t_tell_us_about_language_origins

    Chapter 6, “Fossil cues to the evolution of speech” convincingly rubbishes a lot of the purported anatomical evidence for the probable age of modern-human-style Language.

  7. Excellent, I love a good rubbishing!

  8. I’m kinda over breathless headlines about archaeological discoveries which show that “our” ancestors could make tools, or draw, or navigate a boat, or have fun 50,000 years ago. It’s nice to see, but it’s not surprising anymore.

    Speaking of which, a sad and fond farewell to George Booth, creator of Ip Gissa Gul and many nervous dogs, who has recently left us.

  9. If protolanguages began as largely gestural systems, Planer and Sterelny ask in the excerpt from the book featured below, why and how did vocalization become so important?

    Oh, this again.

  10. And I thought that singing with guitar near campfire was the apex of human evolution, not the beginning.

  11. PlasticPaddy says

    Prehistoric beer cans are an unexplained gap in the archaeological record, they were clearly a primary driver in metallurgy. Current theory is that they were recycled by eco-conscious hominins or erectins.

  12. giving that they actually know what they are talking about.” – No, they do not. They know the modern situation of Khoisan langauges:-)

    The suggestion that proto-World could contain click consonants is fine. It is similar to “it could countain a sound /a/”, but /a/ does not seem unusual, while click consonants are exotic. It is good if people here learn more about Khoisan.

    As for accuracy… What such claims in newspapers mean in terms of probability, and what probability you would assign to a claim that proto-World if it exists contained clicks? 1/2? 1/10? 9/10? Does not matter, not 1/100. The argument that such consonants are “rare” and thus as likely to be in use in some language 120 kiloyears ago as any rarity on the one hand or as in any of modern languages on the other hand is cancelled by these considerations:
    – The only alternative to normal consonants (apart of whistling lnaguages maybe)
    – The clicking areal contains the genetically divergent population
    – The areal also contains high genetic diversity according to Knight et al 2003 (the article to which Güldemann and Sands are objecting).

    What it means is that when you are fantacising about proto-World you can’t exclude click consonants from your list of options based on rarity.

  13. David Eddyshaw says

    You should read the actual article, drasvi. They pretty much agree with you.

  14. @DE, I read some (not all of it). I think I’ll read the rest.

    But it is mostly dedicated to Knight et al 2003 – while I here rather speak about naïve claims in newspapers: I think they are not so naïve.

  15. Click languages of Africa are often portrayed as ancient languages, with clicks being viewed as probable relics of an ancestral mother tongue. This view espouses a monogenetic origin of clicks, and assumes a linguistic conservatism particular to the languages referred to as ‘Khoisan’ (or ‘Khoesan’).

    are often portrayed … being viewed …

    They specify by whom below.
    1. van Ginneken 1939; Chatterji 1939; Stopa 1972 (Structure of Bushman and its traces in Indo-European)
    2. newspapers, including one that cites both Knight and Sands.
    3. Knight et al. , African Y Chromosome and mtDNA Divergence Provides Insight into the History of Click Languages, Current Biology, 2003

    Their article is an answer to Knight et al. Knight et al. 2003 believe that if Hadza and San are distantly related, then the best explanation for clicks in their languages is that they are inherited from when they formed one population.
    Which as they suggest could be a very distant time (possibly predating other splits).

    What do I like about this? I like the title “Structure of Bushman and its traces in Indo-European”.:))))

  16. “a phrase used by anthropologist Polly Wiessner in a 2014 article analyzing the fireside conversations of the Ju/’hoan (!Kung) Bushmen of South Africa”
    (from LH’s original post)

    I think I am going to read it too, but I wish audio recordings were included:)

  17. More input form geneticists:

    Collectively, autosomal, Y, and mitochondrial DNA support early divergence of Hadza ancestry (Knight et al. 2003; Tishkoff et al. 2007). Lachance et al.’s (2012) conclusion of late divergence was based on a neighbor-joining tree; the assumption of treeness or bifurcation is violated by admixture and gene flow, thus invalidating their conclusion. An early divergence of Hadza and Khoe-San peoples is consistent with the grouping of Hadza and Khoisan languages (Knight et al. 2003). On the other hand, evidence for gene flow between 7.5 and 20 thousand years ago is consistent with the hypothesis that click sounds are a recent addition to Hadza and Sandawe languages (Rito et al. 2013). Our results suggest a third possibility. The semi-supervised analysis revealed that Hadza ancestry is closer to Omotic ancestry than to Khoisan ancestry. Also, Omotic ancestry does not cluster with Arabian, Berber, or Cushitic ancestries, consistent with the hypothesis that Omotic languages are not part of the Afroasiatic language family (Theil 2006). Taken together, our genetic findings support a phylolinguistic hypothesis that Omotic and Hadza languages form a language family (Elderkin 1982). Furthermore, if both Cushitic and Niger-Congo ancestries in the Sandawe sample are comparatively recently acquired, then the core of the Sandawe sample is predominantly Omotic, supporting a phylolinguistic hypothesis that the Sandawe language also belongs with Omotic and Hadza languages. The hypothesis that Hadza and Sandawe peoples are not Khoe-San peoples is supported by previous osteological and serogenetic studies (Morris 2002).

  18. David Eddyshaw says

    Taken together, our genetic findings support a phylolinguistic hypothesis that Omotic and Hadza languages form a language family

    This is pure, unadulterated pseudoscience.

  19. Two issues here:

    – the degree of confidence implied by “findings support a hypothesis”.

    – an UFO lands and aliens say: ‘we have been observing your species for a million years, and Hadza is the closest (outside Omotic) relative of most Omotic languages’ and you believe them.
    Linguists have never found themselves in this situation, when a degree of relatedness is known in absence of any reconstructions. (African Romance is a different situation: we don’t know the African Romance rather than how to connect it to Latin)

    Of course geneticists can’t say anything like this. But their data affects the likelihood of the event described by aliens.

    Anyway, what I mean is: D&S 2009 objected by Knight et al 2003, but it seems now we have a whole club of those geneticists making claims about language classification that look outlandish to linguists.

  20. Trond Engen says

    Just imagine what debased, adulterated pseudoscience could achieve.

    But with a little more sympathetic interpretation: The quoted parahraph seems to say that a linguistic hypothesis connecting Hadza or Sandawe with Khoi-San languages will find no extralinguistic support in genetics. That’s fair enough. It also says that one hypothesis of common origin that might be supported would be that of Omotic (as suggested for Hadza by Elderkin 1982). I don’t read that as making a linguistic claim based on genetics, but as pointing out a pre-existing linguistic hypothesis that would better fit the extralinguistic evidence.

  21. AAAAAA!!!!!!!!!!!

    “making a linguistic claim based on genetics,”

    I recently learned the name of my freind’s son. The friend is Uyghur/Uzbek and the son’s name is Einar. And my brain all the time tried to interpret “Einar” as Turkic but failed, because it is not Ai- and not -nur.

    So I wanted to make a linguistic claim (that Einar is now aquiring Russian langauge) and realised that the name is Norse. Idiot.

  22. David Eddyshaw says

    But with a little more sympathetic interpretation

    Fair enough. Moreover, ’tis the season of goodwill.

    Nɛ ya bʋriyasʋŋ, Zupibigdima!

  23. “as pointing out a pre-existing linguistic hypothesis that would better fit the extralinguistic evidence.”

    I make a different distinction.
    A “family” is
    (1) A structure made of forms and sound laws.
    (2) a historical hypothesis about groups of people transmitting language to each other.

    The second object is not “lingustic” in any other sense than “about speakers and what they do”. Such links can be known with full confidence without linguistics, as is the case with Einar (or historical accounts like “people from A understand people from B”).

    The first object in turn can be constructed without any historical knowlege. Linguists work with it, and I don’t see how genetics can contribute in this work.

    But linguists believe that the first object is associated with the second object. That we can draw a tree linking communities who transmit language to each other – and the roots of the two trees will be a some cluster of similar dialects… and tribes.

    Now, links in the second tree can be probabilistic: “languages of Kush and Meroë are similar” 1/2, “Hadza and Omotic have a MRCA within X years”, 0.3. New information can modify these figures.

    Sometimes genetical information can increase the probability. Sometimes a linguistic proposal does NOT (and is not even meant to) increase our expectations regarding possibility of some link: a linguist has an idea about how we can compare two langauges IF they are related (assuming that some slight possibility of relationship existed anyway, because why not).
    It is absolutely not true that a geneticist’ss or historian’s suggestions are worse in this respect than linguist’s suggestion. Linguists can sometimes rely on suggestions of ancient historians.
    But it is true that a geneticist does not suggest anything about the object one.

    Now, for a linguist the object one is what she works for. For some of them our “classification” ideally must approach this object (still allowing tentative links), for others like DE it must represent it.

    So when a linguist reads such text, she is shocked: she reads it as a claim about the object one.
    Again the two objects are of course related.

  24. ” Linguists work with it, and I don’t see how genetics can contribute in this work.”

    I mean, other than suggesting directions. Like “these two people are genetically related’ is a reason to look if we can find similarities in their languages.

    But in the second graph it means: the probability assigned to a certain link jumped from 1/10 to 1/5.

Speak Your Mind