Open the app

Beginning of language: from Homo erectus to Homo sapiens

Beginning of language: from Homo erectus to Homo sapiens

The origin of language — the question of when, why and how our ancestors first began to speak — is among the oldest questions humans have asked about themselves, and also one of the hardest to answer. Language leaves no fossils. Words do not survive in rock the way bones and stone tools do, so there is no moment we can point to and say “here speech began”. Everything we claim about the beginning of language is an inference drawn from indirect traces: the shape of an ancient throat, a gene shared with Neanderthals, a scratch of ochre on a cave wall. This makes the field unusually speculative, and it is worth saying so at the outset.

The caution is not new. In 1866 the Société de Linguistique de Paris — the Linguistic Society of Paris — famously wrote into its founding statutes that it would accept no papers on the origin of language at all. The topic had generated so much untestable speculation that serious scholars treated it as a dead end, and the ban (however overstated its later legend became) helped keep the question out of respectable science for roughly a century. Only from the 1970s onward, as genetics, archaeology and comparative anatomy gave researchers something firmer than armchair guesswork, did the study of language evolution revive. What follows should be read in that spirit: a survey of evidence and competing hypotheses, not a settled story.

The indirect evidence: anatomy, genes and archaeology

Because speech itself cannot be dug up, researchers look for its preconditions and its side effects — the anatomy that makes complex vocalisation possible, the genes that build the speech-and-language system, and the archaeological signs of the symbolic mind that language expresses. Each line of evidence is suggestive; none is decisive.

Anatomy. Modern speech depends on a low larynx and a flexible vocal tract, and on fine motor control of the tongue and breath. Two much-discussed clues are the hyoid bone and the hypoglossal canal. A remarkably complete Neanderthal hyoid — the small U-shaped bone that anchors the tongue muscles — was recovered at Kebara Cave in Israel and dated to roughly 60,000 years ago; its size and shape are almost indistinguishable from a modern human’s, which shows Neanderthals were at least anatomically equipped for speech-like sound. But similarity of shape does not prove similarity of use, and the case is not closed. The hypoglossal canal (which carries the nerve controlling the tongue) was once proposed as a marker of speech, on the argument that speakers need a thicker nerve; DeGusta and colleagues (1999) then showed that canal size varies too much — overlapping with apes and australopithecines — to serve as a reliable signal. The lesson of the anatomy debate is that the hardware needed for speech seems to have accumulated gradually, and probably existed well before any date we can pin down.

Genes. The gene most associated with language is FOXP2. It came to attention through the “KE family”, roughly half of whose members inherited a mutation in it and, with it, severe difficulty controlling the mouth and face for speech (Lai et al., 2001). Crucially, Neanderthals turned out to share the same two human-specific changes in FOXP2, implying the variant predates our split from them several hundred thousand years ago. But FOXP2 is emphatically not “the language gene”: it is a regulatory gene active in many tissues and species, and the early story of a sharp, recent selective sweep on it has since been questioned. It is one contributing piece of a genetic system we still only partly understand — a reminder of how far the evidence is from a simple answer.

Archaeology. The most tangible traces come not from speech but from symbolism — the behaviour language is bound up with. At Blombos Cave in South Africa, pieces of ochre engraved with deliberate cross-hatched patterns date to around 77,000–73,000 years ago, and shell beads and pigment use from the same site point to a mind that traded in symbols. Since a symbol is an arbitrary, agreed-upon link between a form and a meaning — exactly what a word is — many researchers read such artefacts as indirect evidence that fully symbolic language was in use, at least among Homo sapiens, by this period. Here too the inference is a bridge, not a proof: symbolic art implies a symbolic mind, and a symbolic mind is the kind of mind that has language.

When did language begin? Homo erectus or Homo sapiens?

This is where honesty about uncertainty matters most, because the honest answer is a range of hundreds of thousands — possibly nearly two million — of years, and the endpoints belong to different arguments.

The cautious mainstream view ties fully modern, syntactically rich language to Homo sapiens and the last few hundred thousand years, with the symbolic explosion of the Middle and Upper Stone Age (Blombos and after) as its clearest signature. The shared FOXP2 variant and the modern-looking Neanderthal hyoid push at least some speech capacity back further — plausibly to the common ancestor of humans and Neanderthals, perhaps 500,000–700,000 years ago.

A bolder minority view reaches back much further, to Homo erectus, a species that lived from roughly 1.9 million to a few hundred thousand years ago. The linguist Daniel Everett has argued that Homo erectus already had a simple form of language — his case rests not on fossils of the throat but on behaviour: erectus made sophisticated tools, controlled fire, cooperated in hunting, and appears to have crossed open water to islands, feats Everett contends would be hard to organise and transmit without some symbolic communication. On his account, once a hominin could treat one thing as standing for another — a mark for an idea — it was “just a short hop” to language, and culture did the rest. It is a stimulating hypothesis, and a genuinely contested one: many researchers regard the behavioural evidence as compatible with sophisticated non-linguistic cognition, and the deep dating as unprovable on current data. It should be read as one serious position in an open debate, not as an established fact — and the same caution applies to the round “one million years” figure often attached to it.

How did it start? Gesture and voice

Alongside when, there is the question of in what medium language first emerged — and here the main division is between gesture-first and vocal accounts.

The gesture-first hypothesis, associated with Michael Corballis among others, holds that early hominins communicated symbolically with the hands before the voice took over, and points to the ease with which humans still gesture, the sign languages that arise spontaneously wherever deaf people gather, and the manual skills of other apes. Its hardest problem is the handover: explaining why and how a gestural system migrated to the vocal channel. Vocal-continuity accounts run the other way, deriving speech from the calls and cries of primate ancestors, while others look to song and shared music as a bridge. None of these can yet be decided, and they need not be mutually exclusive — early communication was very likely multimodal, hands and voice and face together.

A widely used organising idea across these debates is Derek Bickerton’s protolanguage: the proposal that before full grammar there was a simpler stage of meaningful words strung together without much syntax — content without the combinatorial machinery. On this picture the first “languages” would have been short sequences of words, closer to “eat … fire … come” than to a modern sentence, with complex grammar arriving later as the pressures of an increasingly complex social world selected for it. Complex grammar, on this view, is neither the first thing language had nor a strict requirement for it — a claim that connects to debates over what grammar even is, from the Chomsky hierarchy to what we mean by the grammar of a language.

Slow adaptation or sudden leap? Pinker & Bloom versus Chomsky

Even researchers who agree that the capacity for language is built into human biology disagree sharply about how it got there — and this is the deepest fault line in the field.

On one side is the adaptationist case, argued most influentially by Steven Pinker and Paul Bloom in their 1990 paper Natural Language and Natural Selection. Grammar, they contend, shows exactly the kind of intricate, many-parts-for-a-function complexity that in biology is the fingerprint of gradual natural selection and of nothing else; language must therefore have evolved slowly, like the eye, because it helped our ancestors communicate and so survive and reproduce. This is the view developed at length in Pinker’s account of language as an instinct and the evolution of the language faculty, and the 1990 paper is widely credited with making language evolution a respectable scientific question again after its long exile.

On the other side, Noam Chomsky has long been sceptical that natural selection explains the core of language. In an influential 2002 paper with Marc Hauser and Tecumseh Fitch, the narrow language faculty was identified with recursion — the operation (often called Merge) that lets finite units combine into unlimited nested structures — and the suggestion was that this single capacity might have appeared relatively suddenly, as a by-product of some other change in the brain, rather than being assembled piece by piece by selection for communication. On this view much of the “design” of language is not an adaptation for talking at all. The disagreement is not about whether language is biological — both camps accept that it is — but about gradual accumulation versus a comparatively abrupt origin, and it remains genuinely unresolved.

What this means for learning a language

Draw the threads together and one thing survives every version of the debate, however the dating and mechanism are eventually settled: language is not an incidental cultural gadget bolted onto a general-purpose brain but part of the biological equipment of our species. Whether the capacity is 200,000 or a million-plus years old, whether it arrived by slow selection or a sudden reorganisation, the human brain comes ready for it in a way no other animal’s does. That is the core of the nativist view that language is innate, and it is a genuinely encouraging fact for anyone learning a second language: you are not attempting something alien to your nature — you are re-using machinery you were born with and already ran to fluency once, as a child.

Two practical hints follow. First, the deep evidence suggests language began with meaningful units in use — words and short utterances doing real communicative work — long before it had elaborate grammar, and children still acquire it from a flood of such utterances rather than from rules recited in the abstract. That is a strong argument for building an adult’s learning out of whole, meaningful sentences met and produced in context, with grammar as scaffolding rather than the starting point. Second, if the capacity really is standard-issue biology, then the job of learning is less to install something new than to feed and exercise what is already there — through repeated, spaced, active use until the patterns run by themselves. That is exactly the reasoning behind sentence-based learning with active recall and spaced repetition: work with the grain of a mind that evolution already built for language.

FAQ

When did human language begin?

Nobody knows for certain, and any single date should be treated with suspicion. A cautious mainstream reading ties fully modern, grammatically rich language to Homo sapiens over the last few hundred thousand years, with symbolic artefacts such as the Blombos ochre engravings (about 77,000–73,000 years ago) as its clearest sign. Shared genetics and anatomy with Neanderthals push some speech capacity back perhaps half a million years or more. A bolder minority, including Daniel Everett, argues that a simple language existed already in Homo erectus, up to one to two million years ago — a stimulating but contested claim, not an established fact.

If language leaves no fossils, how can we study its origin at all?

Only indirectly. Researchers infer language from three kinds of trace: anatomy (bones like the Neanderthal hyoid from Kebara Cave, which show the physical capacity for speech), genetics (genes such as FOXP2 involved in the speech-and-language system, though FOXP2 is not simply “the language gene”), and archaeology (symbolic objects like engraved ochre and beads, which imply the symbol-using mind that language requires). Each is suggestive rather than conclusive, which is why the field remains genuinely uncertain — and why the Paris Linguistic Society once banned the topic outright in 1866.

Did language evolve gradually or appear suddenly?

This is the central open dispute. Steven Pinker and Paul Bloom (1990) argue that the intricate design of grammar is the signature of slow natural selection, so language evolved gradually like any complex adaptation. Noam Chomsky, with Hauser and Fitch (2002), locates the core of language in recursion and suggests it may have arisen comparatively suddenly, as a by-product of another change in the brain. Both sides agree language is part of human biology; they disagree about whether it was assembled piece by piece or arrived in a leap.

Sources

Related