Did Neanderthals Have Language? Part 2: The Anatomy of Speech

The Neanderthal language debate is old, actually quite old. Dating back almost fifty years, many of the researchers who initially began studying the topic are still heavily engaged in the subject today, and it continues to be one of the most fiery debates in paeloanthropology. Before the advent of genetics, our only way of answering the question of Neanderthal language was by looking at Neanderthal anatomy. The problem is, there are nearly a thousand ways to reconstruct a Neanderthal speech apparatus…


This is the second part of my series on Neanderthal language. If you have not read the introduction to the series and the first part on the language genes, you can follow the links above. The introduction also includes a terms glossary.

This post will probably be a lot easier for people who aren’t already familiar with the terms I’m introducing. This is my area of expertise, so I hope I do a good job of familiarizing you all with the wonderful debate on the subject.


Compared to the genetics, the question of Neanderthal language from the approach of anatomy is more contested. It appears that although Neanderthals possessed much of the relevant equipment we have to produce language, one’s interpretation of how this equipment works has drastic implications for one’s conclusions. For bioacousticians working in this area, what matters most is how different pieces of the speech apparatus were put together. To understand how most individual anatomical pieces relate to language, I will first have to give you a short background introduction to phonetics.

The bulk of the Neanderthal speech debate revolves around the question of whether they had the ability to produce what are known as a quantal vowels. In 1968, phoneticist Ken Stevens proposed a theory for speech which accounted for the limited variation we see in vowels around the world today1. His argument was that the most common vowels, /a/, /i/, and /u/, were the easiest to produce and thus the most critical in the evolution of speech. These three vowels are universally known among phoneticists as the quantal vowels. Ken Steven’s theory, now known as the Quantal Theory for speech, was revolutionary for phonetics as a field.

The same year that Stevens published on Quantal Theory, Johan Liljencrants and Björn Lindblom published the Adaptive Dispersion theory for speech, arguing that the use of these three vowels was a perceptual issue, rather than ease-of-production one2. As a matter of fact, if you look in the graph below, the vowels /a/ /i/ and /u/ all exist in the furthest possible phonetic spaces from one another and from other vowels. This will be important later.

This is a map of vowel windows for the English language. Each point represents a relationship between the F1 and F2 formants that an individual speaker has emitted in pronouncing a specific vowel. The quantal vowels happen to be furthest in phonetic space from all the other vowels.

Regardless of which theory one chooses, and others, such as the more recent Dispersion-Focalization theory uniting the two, the quantal vowels can be seen as the most important for language production in Homo sapiens3.

The Neanderthal Larynx Reconstruction

The debate on Neanderthal speech began as early as 1971, a full year before Quantal Theory was proposed. During this year, anthropologist Philip Lieberman and anatomist Edmund Crelin published a paper titled, “On the speech of Neanderthal man.” In their study, Lieberman and Crelin took measurements of the cranial base, palate, and mandible of a 60,000 year old Neanderthal skull known as La Chapelle-aux-Saints 1 from France and compared them to the corresponding measurements of a human infant and adult male. In doing so, they believed they were able to interpolate the location of the Neanderthal larynx4.

This would be a rather critical piece of the puzzle for making any judgement regarding Neanderthal language, as the larynx is the part of your throat which contains your vocal folds, which are critical for producing speech sounds. How is it that the larynx of a human adult and infant differ, though? Prior to the age of six or seven, your vocal tract is not fully developed and sits higher up in your throat closer to the base of your skull. As you begin to develop, your vocal tract expands with the your epiglottis (acting as an anchor point above your larynx) lengthening to accommodate the lengthening of the spine. This also pushes your larynx further down the throat into the neck. This process is known as the ‘descent of the larynx,’ both in developmental terms and in humans terms. While it is not necessarily rare in other mammals and happens to an extent in chimpanzees, it is greatly exaggerated in our species5.

Lieberman and Crelin used a comparative approach based on muscle attachment sites in human infants and adults to assess where the Neanderthal larynx would be.

How much of a difference does this actually make for speech? Lieberman and Crelin ran computer simulations on reconstructed air passageways between their three models and concluded that due to constraints on their anatomy, Neanderthals would not have been able to produce the vowels /a/, /i/, or /u/ or the consonants /g/ or /k/. The only consonants Neanderthals would be able to pronounce according to their model were labial and dental consonants such as /d/, /t/, and /b/, and the only vowels were higher vowels such as those in the words bit, bet, and bat.

If you will recall, /a/, /i/, and /u/ are the most common vowels in humans speech, and the absence of these in addition to archaeological evidence strongly making the case that Neanderthal culture developed rather slowly meant that Lieberman and Crelin were strongly compelled to conclude that the evolution of language must have been a gradual, rather than immediate, process in which Neanderthals represented something of a mid-point. In other words, it seemed they were saying Neanderthals did not possess language.

New Evidence Appears

Lieberman and Crelins’ reconstruction of the Neanderthal vocal apparatus was criticized by several authors on the basis that the position in which they placed the hyoid, a bone which anchors to the larynx, was too high. In response to their paper, paleoneurologist Dean Falk intuitively pointed out that if their reconstruction of the La-Chapelle-aux-Saints Neanderthal’s vocal tract was correct and if the Neanderthal hyoid were placed in a position similar to human infants, then due to the position of the tongue, Neanderthals would not have been able to swallow solid food!6

Most of the criticisms of Lieberman and Crelin’s reconstruction of Neanderthal vocal space refused to stick largely due to the absence of fossil evidence to the contrary, and as a result their conclusions indefinitely remained in the scientific canon. At that point no hyoid of a Neanderthal, or any other hominin for that matter, had ever been found. Given its fragility and position in the human body completely unattached from any other bone, the likelihood of ever finding one was close to zero. The lucky thing is that most finds in paeloanthropology come rather unexpectedly, and sometimes all it takes for the right evidence to appear is time. Almost eighteen years after Lieberman and Crelins’ original study a French-Israeli team announced that a Neanderthal hyoid had been found.7

The skeleton the hyoid came from was of a young adult found in a cave in Israel dating to about 60,000 years BCE and was labelled Kebara 2. To this date it is one of the most complete Neanderthal specimens ever discovered. Upon first look it was immediately clear that the Neanderthal hyoid and the human hyoid were virtually the same. Nearly all the measurements of the Kebara hyoid fell within the range of modern human variation. A comparative study by the French-Israeli team asserted that due to the angle of the horns on the hyoid, angles on the parts of the mandible, and the angle of both Neanderthal and human head posture, the hyoid of Neanderthals must have rested in the same position as found in humans. Although they were not able to reconstruct the Neanderthal phonetic space in the same manner as Lieberman and Crelin had, they concluded that since the vocal tract of Neanderthals and humans would have occupied a similar space, Neanderthals must have been anatomically capable of human speech.8

Lieberman responded four years later with a slew of criticisms directed at the team, mostly directed at anatomical features of their reconstruction. Many of these were rather indirect to the most obvious critiques Lieberman could provide- in fact, at one point he argued that that the Neanderthal hyoid was not derived compared to pigs, a fact which is completely irrelevant when you consider the fact that the hyoids of the great apes look nothing like the ones found in Neanderthals and humans. Most obviously though, Lieberman pointed out that despite all the effort his critics had put into debunking his reconstruction, they had not actually reconstructed a vocal tract, and they in fact had no idea what the acoustic space of Neanderthal speech might look like with this new information. Furthermore, Lieberman pointed out a fundamental misunderstanding that the team had relied on in their interpretation of the relationship between the vocal tract and speech.9

Click to listen. An annotated spectrogram of a vowel. The fundamental frequency (F0) is 144Hz, the first formant (F1) is 785Hz and the F2 is 1343Hz. Referring to the vowel chart above, you will find this is in the range for /a/.

If you will recall, Lieberman’s original analyses of Neanderthal language rested on the fact that they could not produce the quantal vowels /a/, /i/, or /u/. But what is it that makes the difference between one vowel and another? Vowels in humans are produced in a rather complex manner related to different resonances controlled by adjustments made in the vocal tract. Sounds emitted through human speech are typically measured and perceived on the basis of pitch, measured by scientists in Hertz (Hz). The lowest of these frequencies, known as the fundamental, is typically quite important for a number of reasons as it is usually a measure of the sound initially produced in the body. As these sounds travel from the diaphragm to the mouth, they are modified by the larynx and the mouth which produce a series of harmonics. Each step in this series is known as a formant frequency. The way humans perceive different vowels is by unconsciously measuring the difference between these formants. For example, in figure 1, the difference between vowels is being calculated by the relationship between the first formant (F1) and the second formant (F2). This relationship is evidently quite important in humans, as babies learn to discriminate vowels in this manner even before their first six months of life10.

Lieberman pointed out that even if the team’s anatomical analyses were correct and the Neanderthal hyoid had rested in the same position as modern humans, this would have almost no bearing on what sounds Neanderthals could make. The fact that Neanderthal vocal tracts were of a similar size was irrelevant to whether they could produce these vowels. What really mattered was the shape of the vocal tract, and this would spark one of biological anthropology’s fiercest debates.

A New Reconstruction
vowel space
The vowel space of Neanderthals constructed by Boe et al 2002. The points in asterisks are those previously proposed by Lieberman and Crelin.

In 2002, a team lead by French phoneticist Louis-Jean Boe challenged Lieberman by reattempting the same study he had conducted 30 years prior on the La Chapelle-aux-Saints skull, this time with a different reconstruction which included the newly discovered hyoid bone. Compared to Lieberman and Chelin’s construction, Boe’s team had produced a radically different interpretation of Neanderthal speech. In their simulation, the Neanderthal vocal tract was similar in nearly every proportion to the one found in humans except for in height of the larynx. Although this parameter is critical for the formation of different vowels, the authors argued that other mechanisms, such as tongue gestures, would have allowed Neanderthals to make the same sounds of human speech. Boe’s team admitted that although Neanderthals may have had more difficulty in making a perfect version of the vowel /a/ without some sort of compensatory mechanism, they drew attention to the fact that different pharynx sizes between modern men and women do not seems to make too large of a difference in vowel production.11

Lieberman was not happy with this reconstruction as his view of the larynx as central to producing vowels had been challenged on the basis of a compensatory mechanism which had nothing to do with the larynx. What was the point of the part of the vocal tract, then? Lieberman and Boe had the chance to respond to one another directly in the Journal of Phonetics. In a letter to the editor, Lieberman stated that Boe et al. had completely ignored a critical factor he had been accused of ignoring by Dean Falk in 1974. Namely, how was the Neanderthal to swallow with this construction? Contrary to Falk, Lieberman stated that his reconstruction was correct because chimpanzees could swallow, Neanderthals must have been able to, as well (again, rather irrelevant considering that the chimpanzee larynx is anatomically lower than Lieberman’s Neanderthal larynx). Lieberman also noted that Boe had ignored the size of the tongue. If a Neanderthal vocal tract had the exact same 1:1 vertical and horizontal ratio as humans, Neanderthals, having a more extended vocal cavity than humans, would have had a larynx placed all the way in the chest behind the sternum. In this case the larynx would not have been able to move at all! Lieberman reasserted his position that his reconstruction had been accurate and that perhaps Neanderthals had some sort primitive speech, but not human language.12

Boe and his team responded in the same issue stating that Lieberman was biased by bringing up language in the first place and that he actually didn’t understand their reconstruction. They then pointed out that the horizontal-vertical vocal tract ratio in their Neanderthal was not intended to be 1:1, which Lieberman saw as essential, but actually 0.80. His argument about the sternum was therefore irrelevant. Furthermore, there was no reason to assume that a 1:1 ratio was even necessary to produce the quantal vowels given the potential capacity for Neanderthals to articulate using other means. Boe further emphasized that not only was Lieberman’s reconstruction of the Neanderthal vowel space wrong, but so was his original reconstruction of that for human infants. As it turned out, Boe found that infants most likely could produce these vowels, but it was neurological constraints which kept them from doing so.13

This would not be the end of the debate between Lieberman and Boe. In the end, other reconstructions were made which vindicated the anatomical reconstruction made by Boe’s team. These included reconstructions of a vocal tract using the cervical vertebrae of humans which likewise found Neanderthals would have difficulty in pronouncing the /a/ vowel, as well as one of the Sima de Los Huesos Hominins from Spain which date nearly 350,000 years before the Chapelle-aux-Saints individual.14,15

The Hard Evidence from Soft Anatomy

Although the disagreement on how to construct a Neanderthal vocal tract went back and forth between Boe and Lieberman, other phoneticists and comparative anatomists studying the evolution of language made significant contributions to the debate.

Perhaps the most important outsiders to comment were Bart de Boer and Tecumseh Fitch in 2010. In a paper critiquing a plethora of computer simulations of phonetic vowel spaces, Boe’s reconstruction of the Neanderthal vowel space was singled out. Although they agreed that the lowering of the larynx was not the key factor in producing vowels, as Lieberman had argued, they drew attention to the near 90° angle of the human tongue produced by the lowering of its root at the anchoring point near the epiglottis. For example, the creation of the vowel /u/ requires constriction of the tongue, rather than the larynx. Boer and Fitch claimed that Boe’s simulation was broken, and that in using a human tongue model, it had implictly assumed that the Neanderthal tongue was shaped at an angle like the human tongue, which to them was the most critical component of human speech use.16

Boer and Fitch threw another ringer in the debate by saying that most of this was non-pertinent to the question of hominin language or speech, as most animals could produce their own vowel windows. A proto-language could have hypothetically been formed by a human ancestor with a chimpanzee vocal tract, and this nitpicking over the anatomy did not matter. In 2016, Fitch and Boer proved this point, much to the shock of bioacousticians around the world by showing it was so easy a monkey could do it. In a study of macaque vowels, Fitch and Boer took a number of x-ray still images from a young macaque named Emiliano producing various types of calls, facial expressions, and chewing on a number of different foods and used these images to create a hypothetical vowel space of rhesus macaques- all based on the soft anatomy. They then used the recordings acquired from Emiliano in conjunction with a model to simulate him saying a single phrase: “Will you marry me?” Emiliano sounded eerily human.17

Ironically, Lieberman found himself vindicated in this new study- arguing that their data showed that monkey vocal tracts were not in fact ‘speech-ready’ as they had claimed, but rather restricted. He asserted that both the development of the brain and the vocal tract were necessary for human speech, neither of which Neanderthals had. Fitch and Boer disagreed. They had found that monkeys could produce at last five vowels, which is the average number of vowels used in language around around the world. They also pointed out that quantal theory didn’t extend just to the human quantal vowels- it simply required that vowels be quantal in nature to necessitate speech. If a monkey had the brain requiring language, they could utilize language.18,19

Anatomical drawings of the human and chimpanzee hyoids. The bulla in the chimpanzee hyoid helps accomodate an air sac system found in most primates.

Additionally, just several years prior Boer had examined the differences between the hyoid of great apes and humans and found that the loss of the bulla in the human and Neanderthal lineage likely had direct implications for speech.20 A recent finding of a chimp-like hyoid associated with an Australopithecus afarensis in Kenya indicated this must have happened in the last 3 million years.21 It had also been previously found that the thoracic vertebral canal in humans and Neanderthals had expanded suggesting both species had better manual control of their breathing.22 Furthermore, a study in 2013 on the internal architecture of the Kebara Neanderthal hyoid indicated that it likely underwent the same functional strains as the human hyoid does, associating Neanderthals with speech.23 It seemed the evidence was stacking up for some form of Neanderthal language.

In closing out their contribution to the debate, Fitch and Boer stated, “We find that neither Boë and colleagues’ strong positive arguments nor Lieberman’s negative ones are compelling.” In many ways, these small nuances of human and Neantheral anatomy which are less obvious, are more important to the central question. If Neanderthals had so many shared features associated with language that humans had, it would be rather parsimonious to assume they also carried the end result.


As it turns out, much time was spent stressing the point of quantal vowels and the proportions of the Neanderthal vocal tract when these were only one critical part of the equation. Of course that doesn’t mean this information must go to waste because the question remains regarding what Neanderthal language sounded like. Was Lieberman’s original thought correct in assuming that the ability to use language better than Neanderthals a contributing factor to their downfall? These forms of linguistic determinism have showed up now and again in the anthropological literature. The archaeologist V. Gordon Childe even thought that perhaps the reason Indo-European culture came to dominate Western Eurasia was due to their superior language.24

There is enough evidence provided by both Lieberman and Boe to sway young paleoanthropologists either way. What is obvious about this part of the debate are several factors: the original reconstruction of the La Chapelle-aux-Saints-1 vocal tract was incorrect and the new reconstruction may have its own set of issues (the discussion around this is still slowly ongoing). Fitch and Boer made a good point when they argued that much of this does not matter to language, but I still believe it is essential. I also would also wonder about rates of constrictions due to changes in musculature granted to Neanderthals and humans in determining which vowels are quantal, rather than simply signing off the anatomical debate with the finding that monkeys hypothetically possess the ability to emit vowels of a generally quantal nature.

It may also be that the quantals used by humans are produced by a very innate feature unique to humans or humans and Neanderthals. Remember our discussion regarding the Dispersion Theory of Vowels? Compared to Quantal Theory, its assumptions are radically different. Quantal theory, being based on the vocal abilities of a species, may be rather subjective. Dispersion theory on the other hand, being based on the inherent acoustic components of phonemes carries an entirely different set of predictions. Much of the distance between these two theories have been recovered largely due to a contribution to phonetic theory which includes Boe on the authorship line that argues both quantal and dispersal theories are linked. In this hypothesis, phonemes are spaced acoustically as far apart as possible in order to highlight articulatory features. What this means for Neanderthal language is not much, as per the original predictions of quantal theory, these articulatory features could result in any nouns. It is evident that continuing research in this direction in necessary3,19

Boer and Fitch’s research showed us that human quantals or not, Neanderthals could speak. Usually in a sentence like the previous one, I would be more cautious and say that, “Neanderthals could probably speak,” but given the evidence from experiments with Emiliano the macaque and the other recent findings in hominin anatomy, I’m going to call it a day and say that Neanderthals could produce something similar to language.

Of course, production is only half of the story we can get from anatomy. If it was the case that Neanderthals could not pick up and understand speech, what would be the point of all this tedious research on the vocal tract? Any system of communication is dependent on two actors, the sender and the receiver; but how can we know if Neanderthals could understand language? Next week please join me again when I talk about consonants, the Neanderthal hearing system, and ask whether Neanderthals had the ability to comprehend speech.

Like my blog and want to support my writing? Please consider supporting me on Patreon.


1Stevens, K.N., 1972. The quantal nature of speech: Evidence from articulatory-acoustic data.

2Liljencrants, J. and Lindblom, B., 1972. Numerical simulation of vowel quality systems: The role of perceptual contrast. Language, pp.839-862.

3Schwartz, J.L., Boë, L.J., Vallée, N. and Abry, C., 1997. The dispersion-focalization theory of vowel systems. Journal of Phonetics, 25(3), pp.255-286.

4Lieberman, P. and Crelin, E.S., 1971. On the speech of Neanderthal man. Linguistic Inquiry, 2(2), pp.203-222.

5Nishimura, T., Mikami, A., Suzuki, J. and Matsuzawa, T., 2003. Descent of the larynx in chimpanzee infants. Proceedings of the National Academy of Sciences, 100(12), pp.6930-6933.

6Falk, D., 1975. Comparative anatomy of the larynx in man and the chimpanzee: implications for language in Neanderthal. American Journal of Physical Anthropology, 43(1), pp.123-132.

7Arensburg, B., Tillier, A.M., Vandermeersch, B., Duday, H., Schepartz, L.A. and Rak, Y., 1989. A Middle Palaeolithic human hyoid bone. Nature, 338(6218), p.758.

8Arensburg, B., Schepartz, L.A., Tillier, A.M., Vandermeersch, B. and Rak, Y., 1990. A reappraisal of the anatomical basis for speech in Middle Palaeolithic hominids. American Journal of Physical Anthropology, 83(2), pp.137-146.

9Lieberman, P., 1994. Hyoid bone position and speech: reply to Dr. Arensburg et al.(1990). American Journal of Physical Anthropology, 94(2), pp.275-278.

10Swoboda, P.J., Morse, P.A. and Leavitt, L.A., 1976. Continuous vowel discrimination in normal and at risk infants. Child Development, pp.459-465.

11Boë, L.J., Heim, J.L., Honda, K. and Maeda, S., 2002. The potential Neandertal vowel space was as large as that of modern humans. Journal of Phonetics, 30(3), pp.465-484.

12Lieberman, P., 2007. Current views on Neanderthal speech capabilities: A reply to Boe et al.(2002). Journal of Phonetics, 4(35), pp.552-563.

13Boë, L.J., Heim, J.L., Honda, K., Maeda, S., Badin, P. and Abry, C., 2007. The vocal tract of newborn humans and Neanderthals: Acoustic capabilities and consequences for the debate on the origin of language. A reply to Lieberman (2007a). Journal of Phonetics, 35(4), pp.564-581.

14Barney, A., Martelli, S., Serrurier, A. and Steele, J., 2012. Articulatory capacity of Neanderthals, a very recent and human-like fossil hominin. Phil. Trans. R. Soc. B, 367(1585), pp.88-102.

15Martínez, I., Rosa, M., Quam, R., Jarabo, P., Lorenzo, C., Bonmatí, A., Gómez-Olivencia, A., Gracia, A. and Arsuaga, J.L., 2013. Communicative capacities in Middle Pleistocene humans from the Sierra de Atapuerca in Spain. Quaternary International, 295, pp.94-101.

16De Boer, B. and Tecumseh Fitch, W., 2010. Computer models of vocal tract evolution: An overview and critique. Adaptive Behavior, 18(1), pp.36-47.

17Fitch, W.T., de Boer, B., Mathur, N. and Ghazanfar, A.A., 2016. Monkey vocal tracts are speech-ready. Science Advances, 2(12), p.e1600723.

18Lieberman, P., 2017. Comment on “Monkey vocal tracts are speech-ready”. Science Advances, 3(7), p.e1700442.

19Fitch, W.T., de Boer, B., Mathur, N. and Ghazanfar, A.A., 2017. Response to Lieberman on “Monkey vocal tracts are speech-ready”. Science Advances, 3(7), p.e1701859.

20de Boer, B., 2012. Loss of air sacs improved hominin speech abilities. Journal of Human Evolution, 62(1), pp.1-6.

21Alemseged, Z., Spoor, F., Kimbel, W.H., Bobe, R., Geraads, D., Reed, D. and Wynn, J.G., 2006. A juvenile early hominin skeleton from Dikika, Ethiopia. Nature, 443(7109), p.296.

22MacLarnon, A.M. and Hewitt, G.P., 1999. The evolution of human speech: The role of enhanced breathing control. American Journal of Physical Anthropology: The Official Publication of the American Association of Physical Anthropologists, 109(3), pp.341-363.

23D’Anastasio, R., Wroe, S., Tuniz, C., Mancini, L., Cesana, D.T., Dreossi, D., Ravichandiran, M., Attard, M., Parr, W.C., Agur, A. and Capasso, L., 2013. Micro-biomechanics of the Kebara 2 hyoid and its implications for speech in Neanderthals. PLoS One, 8(12), p.e82261.

23Trigger, B.G., 1989. A history of archaeological thought. Cambridge University Press.

One thought on “Did Neanderthals Have Language? Part 2: The Anatomy of Speech”

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: