How Language Began

by Daniel L. Everett

  Our results support the hypothesis that hominin reliance on stone tool-making generated selection for teaching and language and imply that (i) low-fidelity social transmission, such as imitation/emulation, may have contributed to the ~700,000 year stasis of the Oldowan technocomplex and (ii) teaching or proto-language may have been pre-requisites for the appearance of Acheulean technology. This work supports a gradual evolution of language, with simple symbolic communication preceding behavioural modernity by hundreds of thousands of years.10

  This is a growing area of research, linking toolmaking to language evolution via brain development. Thus the presence of tools in a society, because they can be interpreted as symbols themselves, offers evidence that the toolmakers had achieved a form of symbolic representation, though it sometimes exaggerates the degree of closeness between tool complexity and linguistic complexity.

  In discussing tools relative to language, one also looks to the qualities that both language and tools illustrate of culture, shared intentions and the ability to match form and function. This is the conceptual basis of symbols. Olduwan tools are the earliest known. They were used from roughly 2.6 million years ago. Such tools, whose uses included chopping, scraping and pounding, were probably invented by Homo habilis (if one accepts this name as a separate, non-erectus species within the genus Homo), or possibly by australopithecines, but the tools at the Olduvai Gorge were clearly transported and manufactured by erectus. The Olduwan tool kit in Figure 13 shows stones crudely shaped to work as weapons and tools, like a hammer or a hand-axe. They would have not been precision instruments comparable to later tools, but they represent a step forward in hominin technology, serving perhaps as precursors to culture.

  To produce an Olduwan tool, a ‘core rock’ is struck on its edge by a round ‘hammerstone’. The striking produces a sharp, thin flake, leaving conchoidal fractures on the core rock, as seen in the image. The flakes are often reworked for other purposes.

  The manufacture of tools requires planning, imagination (having an image of what the final tools should look like) and, at least eventually, communication of some sort for instructing others in how to make tools. The sequential operations call upon the prefrontal cortex and produce cultural selectional pressure for more cortical horsepower, more smarts. However, this pressure might have worked, the larger prefrontal cortex of earlier Homo toolmakers, relative to australopithecines, may be a response to it. Not surprisingly, therefore, about 1.76 million years ago, roughly 300,000 years after the rise of Homo erectus, Olduwan tools were joined by other erectus-manufactured tools, in particular a new type called the Acheulean (Figure 14). Many people suggest that this long, mysterious period without innovation, longer by far than (yet still reminiscent of) the Dark Ages of Europe, is due to ‘low fidelity social transmission’. In other words, because erectus lacked language. But this is not a necessary inference. Cultural conservatism is a powerful and common force. It is always easier to imitate than to innovate, especially if a culture discourages innovation, as is still common throughout the world.

  Figure 13: Olduwan tool kit

  If erectus indeed possessed language, then is there a problem for my theory of erectus linguistic achievements, namely that it took hundreds of thousands of years for erectus to develop the technological advance of Acheulean tools? It is possible that this is correct, that although erectus invented tools and symbols during the first cognitive, and in all likelihood linguistic, revolution, about 1.9 million years ago, it took an additional 600,000 years, roughly, of evolution followed by invention to achieve language. Acclaimed palaeontologist Ian Tattersall makes the same suggestion in several works. Nevertheless, such a pessimistic conclusion does not follow.

  It is known that human cultures, even in the twenty-first century, are resistant to change. Imitation is favoured strongly above innovation when what is being imitated still works fine, as several anthropologists have claimed.11 The lag might have resulted from a lack of cognitive or linguistic development. But it might also result from the nearly universal principle of ‘satisficing’, in other words, nature tends to be satisfied with ‘good enough’, not striving for the best.12 Or religious conservatism. It is indeed a surprisingly long time. But this ‘innovation gap’ does not wear its explanation on its sleeve. And in light of all other evidence, it does not alter the hypothesis that Homo erectus invented language.

  Regardless of a long delay, erectus did eventually improve on its Olduwan tool kit. Though Olduwan and Acheulean tools overlapped in their use by earlier hominins, Acheulean tools were more advanced. They were carried from Africa to Europe by Homo erectus, with Spain being their earliest European destination, about 900,000 years ago. Acheulean tools were not created exclusively by striking stone upon stone as were Olduwan tools. They also involved shaping after flaking with bone, antlers, wood and other tools, which provided more control for the toolmaker. Also, Acheulean toolmakers preferred to use the cores over the flakes as the primary tools. So they were an advance over, but also as a complement to, Olduwan tools.

  Building on Acheulean technology, erectus added other innovative improvements to develop the more advanced Levallois technique (ca. 500,000 years ago). In the spread of all of these tools, however, we see communication, if not in explicit instruction or linguistically, then in the revelation of the tools themselves to other hominins, as they spread and their use and design became known and emulated.

  The Levallois technique required fine work along the edges of a core, followed by a final blow that lifted the flake, presharpened by the earlier striking. These tools were often made of flint, a more workable material, and thus had finer edges, as seen in Figure 15.

  The complexity and uniformity of Levalloisian tools leads some to argue that language is implicated in their manufacture, in order to account for the error-correction assumed to have been necessary. But speaking is not absolutely required. Learning is often a matter of observation, followed by trial and error under a watchful eye, with very little verbal communication required even in modern societies. However, some form of advanced communication does seem to be necessary for feedback, even in language-minimal training. Moreover, there is no doubt that making tools together and the correction of flawed techniques by learners would have favoured language development for instruction. And this was occurring with the first hominins, to actually produce intentionally iconic and geometrical art. The idea that erectus was capable of some sort of sophisticated communication, such as at least a G1 language, is supported not only by their art and tools but also by their travels. It stretches credulity to believe that they travelled over land and sea and developed the settlement patterns they did without symbolic communication.

  Figure 14: Acheulean tools (https://en.wikipedia.org/wiki/Acheulean#/media/File:Biface_Cintegabelle_MHNT_PRE_2009.0.201.1_V2.jpg)

  Some researchers suggest that icons like the Makapansgat pebble and the Erfoud manuport may have led to new neural pathways for recognising that one thing may stand for another. The evidence seems too sparse to justify this statement, though I do agree that new ways of thinking can lead to new evolutionary pressures on the brain. These can then enhance the ability to comprehend representations more complex than mere indexes. More importantly, the iconic manuports might have led to cultural change, far before cerebral change.

  An interesting further bit of evidence in the evolution of the symbolic comes from early art, such as the 250,000-year-old Venus of Berekhat Ram (Figure 16). Some deny that this is art, claiming it is nothing more than a rock bearing a human resemblance, the same as the Makapansgat example. However, to some experts, it shows evidence, upon careful examination, of having been manually altered to take on more of a ‘Venus likeness’. And there is some suggestion of red ochre added to the stone as a form of decoration. Though it may not have been a completely ex-nihilo objet d’art in this sense, the evidence suggests strongly that this is the oldest extant work of art in the world, either because it is carved from scratch or becau
se it is a human-modified natural formation.

  Figure 15: Levalloisian tools (https://en.wikipedia.org/wiki/Levallois_technique#/media/File:Pointe_levallois_Beuzeville_MHNT_PRE.2009.0.203.2.fond.jpg)

  The jump from index or icon to symbol is a relative baby step in conceptual development, though huge in language evolution. In my very first encounter with the rain forest, I was constantly on the lookout for snakes. Every tumescent root that ‘slithered’ across my path, partially covered in leaves, appeared to me first as a writhing, threatening serpent and only secondarily as inert flora. Perhaps in recounting similar experiences, two erectus buddies could have come to reinterpret the roots as icons for snakes, even eventually as snake symbols. (A similar evolution is seen in comparing the earliest with the latest of the Egyptian hieroglyphics or the Chinese writing system – icons become symbols, that is, more arbitrary, with the passing of time.)

  Symbols arise naturally within minds embedded in cultures, able to learn, retain and integrate knowledge into a sense of personal and group identity. One example, just given, is how the mind makes use of errors, perhaps moving from misperceptions to icons to symbols, one image ‘standing for’ another.

  Figure 16: Venus of Berekhat Ram

  But they also arise from adaptation of the natural to the conventional in culture. One treatment of this route towards symbolisation is proposed by anthropologist Greg Urban. In his work on ritualised lamentation in Gê languages of Brazil, Urban argues that natural crying was transformed culturally into ritualised crying. This would illustrate a transformation of natural emotionally reactive sounds into a form of ‘strategic vocal manipulation’, a form of iconic representation of the emotional state of sadness. Further, he claims that ‘strategic vocal deceptions in non-human primates are possible precursors of true socially constructed, socially shared metasignals, which in turn may be ancestors of modern human language’. Though newly created icons are insufficient for symbolic language, they do seem to offer a distinct and natural source of developing representations and thus perhaps a source of symbol invention.13

  Another area in which symbols arise is in tracking social relationships. Most primates, among many other creatures, have elaborate social organising principles, through kinship, such as polyandry, polygyny, dominance relations, cross-cousins and parallel cousins. These concepts are learned via interactions, based initially on physical opposition, such as male vs female, strong vs weak, malleable vs non-malleable, or mother vs child. As people use concepts they come to understand them. So one can say, accurately, that even without language, many animals use something like concepts as they negotiate their ways through social relationships. Keeping track of such relationships would have increased the cultural and cognitive selectional pressures for symbols as some anthropologists in fact claim.

  Numerous researchers have written on the evolution of the symbolic. However, as illuminating as these discussions are, they share a common lacuna, the connection of symbol-evolution and grammar to a well-developed theory of culture. It has been claimed that status symbols (such as expensive athletic shoes) have little to do with linguistic symbols. If correct, this would mean that a culture’s use of status symbols is unrelated to whether they have symbols such as words. The most parsimonious interpretation of personal ornaments found in erectus burial sites is that these are nothing more than status markers. Part of the reason that researchers reject the potential linguistic significance of status symbols is because they claim that such symbols lack ‘displacement’ – a reference to something absent from the immediate context. Since we regularly talk about things that may or may not exist in other times or places, displacement is a fundamental feature of human language.

  In other words, clothes and jewellery don’t represent anything other than the taste and status of a person in the immediate context. But a minute’s reflection reveals that this statement is incorrect. It lacks an adequate appreciation for culture. Status is inherent neither in ornaments nor in individuals, nor do ornaments bestow status. If someone finds the royal crown, putting it on their head not only does not make them a monarch, it makes them subject to the status-lowering charge of being an imposter. Status derives from culture. Status symbols are social signs. They are signs dependent for their meaning on abstract, displaced, cultural values. Thus although it is correct to say that status symbols are not linguistic symbols, both linguistic and status symbols are arbitrary, socially indexical and displaced. Therefore, they are conceptual kin. To have one is related to having the other. They would be expected to occur together in the same society at the relevant level of conceptual complexity or simplicity.

  Displacement, the element which some claim to be missing in status symbols, is itself subject to cultural constraints. The crucial components for developing symbols are not displacement so much as arbitrariness and intentionality. But displacement is present for both status symbols and tools. Both kinds of artefacts reference abstract entities including the cultural values, social roles and structured knowledge that are present in the minds of all members of a culture.

  What is the general evolutionary path to the development of symbols? To refer back to the example of roots across the jungle floor from earlier, when I see branches or roots at times as I am walking through the jungle, I jump back if I haven’t had a good look, worried that one might be a snake. This mistaken association of one thing for another can lead later on to the intentional use of the object of the false impression to represent the thing it was falsely associated with. One could draw a root to mean a snake, or use the word for ‘root’ to mean ‘snake’. As early hieroglyphic writing systems from different parts of the world show, this use of representations based on resemblance can further evolve so that all resemblance is gone, thus leading from an intentional use to an intentionally arbitrary symbol. As this happens, a drawing changes from an icon to a symbol. And just as this has happened in writing systems, so it is likely that it has happened in spoken systems, with different sound combinations becoming conventions, associated with a particular meaning.

  In the research of evolutionary psychologists and anthropologists, one finds arguments to the effect that the development of kinship relationships would have created concepts in need of forms. That is, kinship relationships exert pressure on humans to go beyond icons to invent symbols. Concepts go looking for forms to serve as cultural exchange. I have a father. How should I communicate that to you? How should I say ‘father’? But if, as many researchers believe, non-human animals too have concepts, then why don’t animals develop symbols? One could retort that animals lack the language gene, but this is not terribly insightful, merely pushing the explanation back one level to the evolution of the gene rather than the evolution of symbols.

  While there is no evidence for a specific language gene (the oft-cited FOXP2 gene is certainly not one, though it is sometimes claimed to be so), a great deal is known about the evolution of human intelligence and it is clear that humans are more intelligent than non-symbol-using creatures.14 Thus a richer array of concepts requiring symbols and a richer, more inventive intelligence would have both been under pressure to find a joint solution to concept communication. Linguistic symbols arise to satisfy needs as cultures develop and they may emerge from status symbols, burial symbols and the like.

  Anthropologist Michael Silverstein analyses the recursive properties of human thinking as applied to the use of language in representing cultural meaning, at multiple levels simultaneously. Another person exploring similar themes, explicitly linked to the recursive thinking (thinking about thinking or thoughts within thoughts) that underlies human cognition is Stephen C. Levinson.

  Peirce anticipated both Levinson and Silverstein, however, in proposing that symbols are constructed of other symbols. In Peirce’s writings, the phrase ‘infinite semiosis’ means that there is no limit to the number of symbols available to humans for languages. This in turn is based on the view that signs are multifunctional. Each sign determines an inter
pretant, but an interpretant is also a sign, so every sign embodies a second sign. This is a kind of conceptual recursion, concepts within concepts, and represents a huge step forward in human communication. It means that a string of signs always contains other signs. According to Peirce, this can be understood when we see infinity even in a simple sequence like:

  Sign1/Interpretant1 Sign2/Interpretant2 … Signn

  This representation looks finite until we realise that Signn cannot be the end because if it lacks an interpretant it is not a sign. Likewise, Sign1 cannot really be the beginning, because by definition it is connected to an interpretant of an earlier sign. So there is no beginning or end to symbols and signs. The process that creates them is infinite because it is recursive. Any random sign is always partially composed from another sign.

  The origin and composition of symbols we have been discussing highlights the fact that, like other biological functions, human language is not simple. Language arises from interaction of meaning (semantics), conditions on how it is used (pragmatics), the physical properties of its inventory of sounds (phonetics), a grammar, phonology (its sound structure), morphology (the way the language creates words, such as by suffixes or prefixes or no additions at all) and the organisation of its stories and conversation. Yet, even after all this, there is something else. Language as a whole is greater than the sum of its parts. When we hear our native language we do not hear grammar or particular sounds or meanings, we hear and instantly understand what is being said as a whole, individually and together in a conversation or story.

  Grammar not only is important for language to fulfil its culture-building function but also helps us to think more clearly. Yet in spite of the focus of many linguists on grammar as synonymous with language, grammar itself is no more important than any other component of language.