Could Neanderthals Talk? Speech in Other Human Species

In 1983, excavators at Kebara Cave on Mount Carmel in Israel uncovered the skeleton of a Neanderthal man, about 60,000 years old. Known as Kebara 2, he was missing his skull and legs, but one small, fragile bone had survived: the hyoid, a U-shaped bone in the throat that anchors the tongue and the muscles of the voice box.

When Baruch Arensburg and colleagues described it in Nature in 1989, it looked almost identical to a modern human hyoid. That find opened a long debate. Could Neanderthals talk? Language does not fossilise, so researchers look for indirect clues: the anatomy of speaking and hearing, the genes involved in language, and the behaviour that complex communication makes possible.

The problem with sound

Spoken language is invisible to archaeology. Writing appears only about 5,000 years ago. Everything earlier must be inferred from proxies, and each proxy has limits. A modern-looking throat shows that speech sounds were possible, not that they formed a language.

A bone for speaking

The hyoid matters because its shape and position relate to how the tongue and larynx move. Chimpanzee hyoids are quite different from ours, with a large cup-shaped body linked to air sacs.

Kebara 2's hyoid is modern in shape. In 2013, Ruggero D'Anastasio and colleagues used micro-CT scans to study its internal structure, which reflects the stresses a bone experiences in life. They found a pattern similar to modern humans, suggesting it was used in much the same way.

Earlier hominins show the contrast. A 3.3-million-year-old hyoid from the Australopithecus afarensis child at Dikika, Ethiopia, is ape-like. In 2022, Takeshi Nishimura and colleagues showed that humans lost the vocal membranes and air sacs that other apes have, which makes our voices more stable. That loss probably happened somewhere along the line between those two fossils.

Ears tuned for talk

Hearing matters as much as speaking. Each species' ear is tuned to the sounds that matter to it.

At the Sima de los Huesos in Spain, skulls about 430,000 years old preserve the tiny bones and canals of the middle and outer ear. In 2004, Ignacio Martínez and colleagues modelled their hearing. These early Neanderthals heard best across a range similar to ours, unlike chimpanzees, which are more sensitive at lower and higher frequencies.

In 2021, Mercedes Conde-Valverde and colleagues extended the analysis to later Neanderthals. Neanderthal hearing was most sensitive across a band reaching up to about 4–5 kilohertz, similar to ours and slightly wider than in the Sima de los Huesos fossils. The upper part of that band carries many consonants, such as "s", "f" and "t". The authors argue that Neanderthals had a communication system as efficient as human speech.

Breathing for speech

Speaking requires fine control of breathing, so that air can be released steadily over a sentence. The nerves for this run through the thoracic part of the spinal cord.

In 1999, Ann MacLarnon and Gwen Hewitt showed that modern humans and Neanderthals have a wider thoracic spinal canal than other primates. The 1.5-million-year-old Turkana Boy (Homo erectus) had a narrower canal, which they read as limited breath control. Later work on Homo erectus vertebrae from Dmanisi in Georgia found canals within the modern range, and some researchers think Turkana Boy may have had a spinal disorder. The erectus picture is therefore uncertain.

A cautionary tale: the hypoglossal canal

In 1998, Richard Kay and colleagues proposed that the hypoglossal canal, the opening at the skull base for the nerve that controls the tongue, was larger in Neanderthals and modern humans than in apes. They suggested it signalled speech.

A year later, David DeGusta, Henry Gilbert and Scott Turner measured more skulls and found that many non-human primates have canals as large as ours, relative to body size. Canal size does not predict speech. The episode is a reminder to test single-feature claims carefully.

The FOXP2 gene

FOXP2 came to public attention through a British family, known as KE, in which a mutation in the gene causes severe difficulty with speech and grammar. It is sometimes called the "language gene", which is misleading. FOXP2 controls many other genes and is active in the brain, lungs and gut. There is no single gene for language.

In 2007, Johannes Krause and colleagues found that Neanderthals carried the same two amino-acid changes in the FOXP2 protein that distinguish humans from chimpanzees. Denisovans share them too. In 2013, Tomislav Maricic and colleagues found a difference in a regulatory region of the gene between modern humans and Neanderthals, so the gene may have been controlled differently.

An earlier claim that FOXP2 went through a recent selective sweep in modern humans was not confirmed by a larger 2018 study. The overall lesson: FOXP2 shows that Neanderthals shared some of the genetic machinery for speech, but it cannot settle the question alone. You can explore the protein in our 3D molecule viewer.

The old vocal tract debate

In 1971, Philip Lieberman and Edmund Crelin reconstructed the vocal tract of the La Chapelle-aux-Saints Neanderthal and argued that it could not produce the full range of human vowels. The reconstruction was widely criticised, partly because the skull base was incomplete and distorted. Later acoustic modelling, including work by Louis-Jean Boë and colleagues in 2002, suggested a Neanderthal vocal tract could produce the main vowels. Most researchers no longer see vowel range as a barrier.

Could Neanderthals talk? What behaviour implies

Neanderthals did things that are hard to coordinate or teach without rich communication:

• At Bruniquel Cave in France, about 176,000 years ago, they broke off around 400 stalagmite pieces and arranged them into rings deep underground.
• They made birch tar adhesive, a multi-step process, to haft tools.
• They buried their dead, used pigments and wore eagle-talon ornaments.
• They hunted large and dangerous animals in groups.

None of this proves grammar, but it makes some form of complex vocal communication very likely.

What about earlier species?

For Homo erectus, the evidence is thin. Its brain was smaller, its thoracic canal is debated, and no hyoid or ear bones of the right kind have been studied in detail. Some researchers argue that its long-distance dispersal, standardised handaxes and possible sea crossings to Flores required a form of language. Others think a simpler protolanguage, or gesture, would have been enough. This remains speculation.

Why it matters

If Neanderthals could talk, language is older than our species, and probably goes back to a shared ancestor more than half a million years ago. That changes how we think about what makes us unique. The fossils whisper rather than shout, but for Neanderthals they increasingly say yes. Compare the two species in Homo sapiens vs Neanderthals, or explore the interactive family tree.