How Language Began

by Daniel L. Everett

Culture is abstract because it cannot be touched, or seen, or smelled – it is not directly observable. However, the products of culture, such as art, libraries, political roles, food, literature, science, religion, style, architecture and tolerance or intolerance, are non-abstract, visible and tangible. Culture as a dynamic force is found only in the individuals of a society. Members of any society share a culture when they agree on a range of values and the relative priority of all the values that they hold. Members of a culture in turn share knowledge and social roles. One observes values and knowledge applied or examples of expectations from different social roles in action through individual members of the society. This is culture in action.

  Each modern human, as did every Homo erectus, learns their place in society, what is more or less important as members of that society, as well as the knowledge common to all the members. And they teach these things by word and example to their offspring. All humans, past and present, learn these things. As do other creatures.

  Nowadays, a very different theory of the origin of language than what I am urging here is popular among some. This is the idea that language is a disembodied object, along the lines of a mathematical formula. In this view, language is little more than a particular kind of grammar. If that kind of grammar, a hierarchical recursive grammar, is not found in a communication system, then that form of communication is not a language. Proponents of this idea also maintain that grammar ‘popped’ into being some 50–65,000 years ago via a mutation. This suggestion, even though very widely accepted, has surprisingly little evidence in its favour and turns out to be a poorer fit with the facts than the idea that language was invented, but subsequently changed gradually through all Homo species, to fit different cultures.

  Though language is best understood as an invention, the mutation proposal is very influential. The theory comes from the work of Noam Chomsky, who began publishing in the late fifties and is, according to some, now the leading linguist in the world. But Chomsky’s view that language is a recursive grammar, nothing more nor less, is a highly peculiar one. Already in 1972, a review in the New York Review of Books by American philosopher John Searle noted how strange Chomsky’s conception of language actually is.2

  This view is unusual because we know that languages need not have intricate grammatical structures. Some might instead merely juxtapose words and simple phrases, allowing context to guide their interpretation, as in the examples that begin this chapter. The main problem with the idea that language is grammar boils down to a lack of appreciation for the source and role of meaning in language. The view here, to the contrary, is that grammar is helpful in languages, but that different levels of complexity are to be expected among the languages of the world, including the extinct languages of Homo erectus. Moreover, complexity can vary tremendously from language to language. In other words, language is not merely a synonym for grammar. It is a combination of meaning, form, gestures and pitch. Grammar aids language. It is not itself language.

  Language, whatever its biological basis, is shaped by psychology, history and culture. I try to show what this means in Figure 7.

  Figure 7: Language is a nexus

  In order to get down to the nuts and bolts of how language itself actually evolved, there are two alternative views of development that must be distinguished. These are uniformitarianism vs catastrophism.

  Uniformitarianism is the idea that the way things work now is the way they worked in the past. That is, the forces that operate in the world today are the same forces that have shaped the world since it began. Uniformitarianism does not deny the possibility of cataclysmic or catastrophic events playing roles in history and evolution. After all, uniformitarian scientists accept that there was a great dinosaur extinction event around 65 million years ago, when an asteroid crashed into the Yucatan. But it says that catastrophic change is not the main driver of evolutionary theory and that catastrophes should not be proposed as explanations without very clear evidence.

  Catastrophism, on the other hand, appeals to major upheavals, such as Noah’s flood or elevated rates of mutations, as frequent explanations for the origin and development of life on earth. Niles Eldredge and Stephen Jay Gould proposed that a great deal of evolutionary change is brought about by sudden macroevolutionary jumps that they called ‘saltations’. Saltational models might be accurate for some examples of evolutionary change. But they always require additional evidence.

  Uniformitarianism rather than catastrophism is taken to be a foundational truth in most scientific disciplines. In physics, few question the assumption of uniformitarianism. Physical laws show no evidence of having changed during the universe’s natural history at least subsequent to the ‘Big Bang’. And in geology, Charles Lyell’s 1833 work, Principles of Geology, is known in part for its advocacy of uniformitarianism in earth history studies. By assuming uniformitarianism, a model of natural selection is expected to account for the transformation of ancient life forms into modern life forms via gradual, homeopathic, ‘baby steps’.

  In the case of language evolution, there are good reasons to reject catastrophism-based views such as Chomsky’s. Reasons include its poor account of the genetics involved and its failure to account for the influence of culture on language emergence. Moreover, this catastrophism view fails to account for the fact that mutations for language are superfluous because language evolution can be explained without them. Invoking mutations without independent evidence is unhelpful. In fact, the idea of language as a mutation simply offers no insights at all that help to understand the evolution of language. That is to say that language evolution can be explained without mutations, based instead on gradual, uniformitarianist assumptions, rendering superfluous proposals of language-specific genes or language-specific mutations.

  Of course, one is free to propose mutations or anything else in order to construct a theory. And mutations are indeed among the drivers of evolution. But the rule of thumb in proposing mutations in the evolutionary record must be that ‘in the absence of evidence, do not assume miracles’. And the proposal of a mutation as a crucial link in human language evolution must come with a full story of what evolutionary forces were at work at the time of the mutation that led to the spread of the mutation; otherwise, it is just invoking a miracle. Even if language ability were a mutation, it could only have produced an ability to learn language at a time when, ironically, there was no language. Apart from this lack of synchrony between need and mutation, someone proposing such a mutation must also explain what the survival advantage of a particular trait, such as language, was at the time of the mutation. And such an explanation must go beyond speculation to evidence. It could not merely be that ‘language or grammar made thought clearer’. That is in all likelihood correct, but it doesn’t speak to how or when language or grammar came into being. Nor does it offer any details on how it spread, either genetically or culturally. Otherwise, bandying about the word ‘mutation’ is unwarranted and speculative. This is the major weakness of the saltationist or what one might refer to as the ‘X-Men’ theory of language origin. Moreover, such a conjecture is unnecessary. Good old-fashioned Darwinian natural selection offers a more scientifically grounded story.

  The sudden-emergence theory of language origin, echoed by palaeontologist Ian Tattersall in several works, also depends heavily on arguments based on absence of evidence. In this case, imagine a picture of a hawk in the sky accompanied by the caption ‘There are hawks here.’ Barring photo-altering software, this picture is pretty good evidence for the caption’s veracity. On the other hand, a picture of a clear sky with the caption ‘There are no hawks near where I live,’ is much more problematic. The latter photo shows only an absence of evidence. It does not show solid evidence of the absence of hawks. It could be nothing more than a coincidence that the photograph failed to capture a hawk in a sky otherwise frequently populated by hawks. What is needed in such a case are more data, such as the lack of flora and fauna that hawks prefer along with the proper
climate to sustain hawks. And identical reasoning applies to the unwarranted claims that language originated as a mutation, or that Homo erectus lacked symbolic representations. If they were right, of course, and erectus lacked symbols, then it would make sense to deny language to erectus. But in fact all that one is entitled to claim is that no one had noticed any evidence for such representations to date. But such evidence does exist in the record of the erectus cognitive explosion marked by their migration from Africa.

  Moreover, there are languages spoken in the world today whose grammars have aspects reminiscent of what Homo erectus languages might have been like – namely symbols ordered according to cultural conventions. In this kind of case symbols follow an order agreed upon by members of a particular society. For example, Americans and Britons prefer to say ‘red, white and blue’ rather than ‘white, red and blue’ when discussing their national flags. Symbols and ordering can sometimes be vague and ambiguous, and therefore erectus would have needed the ability to use context and culture to interpret fully what others said.

  To elaborate the nature of the absence of evidence argument, consider again what is known today about Amazonian tongues, which are unquestionably full human languages. But what record would exist of these languages if all their speakers died out and archaeologists discovered their speakers’ bones 500,000 years hence? Forgetting for now that linguists and anthropologists have published grammars, dictionaries and other studies of Amazonian communication and culture, would these cultures and languages bequeath any material evidence that they were capable of language or symbolic reasoning thousands of years hence? Likely not. As stated earlier they would leave even less than what is found for neanderthalensis culture, aside from the few cultures that make ceramics, such as the famous Marajoara culture discovered on the Switzerland-sized delta at the mouth of the Amazon. It would be nearly impossible to find direct evidence that they had language – just as is the case for many ancient hunter-gatherer groups.

  We are also unable to prove that the Homo sapiens who originally left Africa, or neanderthalensis, or Denisovans, or erectus had language, though it would be astounding if they did not, based on the cultural evidence. Therefore, one must be careful not to conclude that earlier hominins lacked language, merely based on the absence of evidence of artwork or what are commonly recognised as symbols in the prehistoric record. In the absence of evidence, the simplest idea about language evolution is that language gradually appeared via natural, incremental processes, following the invention of symbols, which in turn were made possible by the gradual evolution of the human brain and culture. This means that the burden of proof is on those who propose a sudden mutation for language, not on those who analyse the evolution of language as a gradual, uniformitarian process, fitting in with the rest of what we know about human evolution.

  In recent years several palaeoanthropologists have inferred links between toolmaking and language evolution.3 These researchers do not employ the ‘absence of evidence’ argument. That said, these studies appear to be based on an unusual conception of language as largely grammar and words, failing to consider the overall role and origins of symbols from abstract culture. They examine the growing complexity of tool use and relate this to a purported increase in language complexity, based on the assumption that the syntax of modern languages will always include complex syntactic devices for combining symbols such as hierarchy and recursion. In addition, these researchers discuss the absence of symbols among earlier Homo species, contrasting this with the widespread use of symbols among sapiens.

  Efforts to explain language based on the archaeological record are admirable. Unfortunately, they often borrow bad ideas from linguistics to make their case. The principal bad idea is that quotidian objects are not themselves symbolic. Tools are symbols when they are the product of a culture. If one finds tools in conjunction with evidence for shared values and knowledge in a society, there is no need to look any further than the tools themselves for symbols. Tools may not be symbols for chimps, but they would have been symbols for erectus. Second, grammar does not require complex syntax and, therefore, neither does language. There are many groups today that have well-functioning, completely adequate languages but lack the kind of complex syntax that palaeoanthropologists sometimes assume in attempts to correlate complex tools with complex language. But such cultures also use complex tools, an unexpected state of affairs for theories proposing steady parallel growth of complexity of language and tools.

  The understanding of natural language evolution must incorporate also the fact that language is a cultural tool for community building. More intricate syntactic structures of the kind found in many modern languages, such as subordinate clauses, complex noun phrases, word-compounding and others, are not crucial for language and are later additions made for cultural reasons. ‘Expression of thought,’ proposed by some as the raison d’être of language, is also a secondary feature of language. The evidence on hand from contemporary languages and the evolutionary record count against this in favour of communication as the primary function of language. At the same time, there is no doubt that the uses of language for thinking and communication are dependent upon one another. Each enhances the other.

  If communication is the basic function of language, however, then human languages are not quite so unlike the communication of other creatures as some linguists, philosophers and neuroscientists assume. Communication is, after all, pervasive in the animal kingdom. Humans are simply the best communicators, not the only ones. But exactly in this quality lies the distinctiveness of human languages.

  Alternative hypotheses should not, of course, be rejected out of hand, but only if the evidence is ultimately against them. The following quote is typical:

  [C]ommunication, a particular use of externalised language, cannot be the primary function of language, a defining property of the language faculty, suggesting that a traditional conception of language as an instrument of thought might be more appropriate. At a minimum, then, each language incorporates via its syntax computational procedures satisfying this basic property … We take the property of structure dependence of grammatical rules to be central.4

  Why do many researchers, such as those responsible for this quote, claim that communication is not the primary function of language? This idea flies in the face of what most people would regard as intuitive. Of course, the fact that a scientific idea doesn’t match common sense, does not automatically make it wrong, since scientific judgement frequently differs from the average person’s opinion. However, the reasoning here seems to be that humans are not all that good at communication. Take two well-known examples: ‘Flying planes can be dangerous’ and ‘Visiting relatives can be a nuisance’. Both of these are ambiguous. The former example could mean ‘Planes in flight can be dangerous because one might land on you’ or it could mean ‘Piloting a plane is occasionally risky’. Likewise for the latter example, which can mean ‘Relatives who come to visit are not always welcome’ or something like ‘A visit to my grandparents can prove tiresome’. For some, it follows that although ambiguity is found in communication, it is not found in thinking. Thus language works less well for exchanging ideas and information. If that were true it would then suggest that the principal purpose of language is to aid thinking, not communication.

  This, however, doesn’t follow. First of all, there are studies from the Department of Brain and Cognitive Sciences at MIT that explain why ambiguity is to be expected in a communication system.5 It is produced by the need to keep the amount that must be memorised low while maintaining efficient communication. Therefore, if one says, ‘I want two,’ you know I mean ‘two’ and not ‘too’ or ‘to’ depending on the context. (This clearly is an English-only example, but homophones seem to occur in all languages.) Second, ambiguity and vagueness are rarely problems because context usually enables the hearer to pick out the meaning the speaker intended. If one says, ‘He came into the room,’ the pronoun ‘he’ is vague. One can only inte
rpret ‘he’ if one shares enough information with their interlocutors to know who this particular ‘he’ refers to. Third, ambiguity in writing and speaking is not inherently problematic for language. Rather, ambiguity often is the result of bad planning. Thus, if someone began to cross a classroom and walked into student desks on the way to the door, the judgement would not be that this is evidence that walking is not for locomotion. Rather, one might conclude that walkers should watch where they are going. And the same goes for language – most ambiguity, vagueness and other shortcomings of speech can be avoided by planning and thinking before speaking or putting pen to paper. Planning for communication, like planning in most activities, is helpful.

  There are other problems with the idea that language is not for communication. Evolution never designs perfect systems. Rather it builds jury-rigged devices piece by piece, using what it already has in place. Language, like everything else about natural life, is imperfect. Communication breaks down. But so does thinking! The assertion that one’s thoughts are unambiguous to oneself is just that, an assertion. It needs to be tested. Another issue to note is that it is by no means clear that all people always or even most of the time think in language. Many people, such as biologist Frans de Waal and author Temple Grandlin, claim that they think in pictures, not in words. Experiments to check such purported patterns of thinking are called for.

  In the earlier quote on language and communication, the authors claim that communication is a ‘particular use’ of ‘externalised language’. This makes it sound rare indeed. They believe that the only kind of language that can actually be studied is so-called internal-language, or I-language. An I-Language is just what the speaker knows in order to produce their spoken external or E-language. French, English and Spanish are E-languages, but their speakers’ knowledge of what underlies languages are their I-languages.