To the untrained listener, a bunch of babbling baboons may not sound like much. But sharp-eared experts have now found that our primate cousins can actually produce humanlike vowel sounds. The finding suggests the last common ancestor of humans and baboons may have possessed the vocal machinery for speech—hinting at a much earlier origin for language than previously thought.
Researchers from Grenoble Alpes University and Aix-Marseille University, both in France, and their colleagues recorded baboons in captivity, finding the animals were capable of producing five distinct sounds that have the same characteristic frequencies as human vowels. As reported today in PLoS ONE, the animals could make these sounds despite the fact that, as dissections later revealed, they possess high voice boxes, or larynxes, an anatomical feature long thought to be an impediment to speech. “This breaks a serious logjam” in the study of language, says study co-author Thomas Sawallis, a linguist at the University of Alabama. “Theories of language evolution have developed based on the idea that full speech was only available to anatomically modern Homo sapiens,” supporting an emergence of language 70,000 to 100,000 years ago, he says, but in fact, “we could have had the beginnings of speech 25 million years ago.”
The evolution of language is considered one of the hardest problems in science, because the process left no fossil evidence behind. One practical approach, however, is to study the mechanics of speech. Language consists roughly of different combinations of vowels and consonants. Notably, humans possess low larynxes, which makes it easier to produce a wide range of vowel sounds (and as Darwin observed, also makes it easier for us to choke on food). A foundational theory of speech production, developed by Brown University cognitive scientist Philip Lieberman in the 1960s, states the high larynxes and thus shorter vocal tracts of most nonhuman primates prevents them from producing vowel-like sounds. Yet recent research calls Lieberman’s hypothesis into question.
For one thing, scientists have discovered that a number of other species besides humans, including chimpanzees and some types of deer, possess low larynxes. In addition, babies can produce vowel sounds despite the fact their larynxes have not yet descended, and studies of Neandertals, who also have high larynxes, have suggested these ancient humans may have had the ability to produce such sounds as well. Moreover, scientists now think that the way the tongue muscle and lips constrict the airway is more important for producing speech than the position of the larynx.
In the present study, a team led by speech scientist Louis-Jean Boë recorded 15 guinea baboons (three male and 12 female) making spontaneous vocalizations, comprising grunts, “wahoos,” barks, yaks and copulation calls. They recorded the animals during the half hour before their feeding when they were most vocal. (They did not record the animals during feeding “to avoid potential distortion of the vocalizations due to chewing and full cheek pouches.”) Next they applied a technique known as linear predictive coding (LPC) analysis to determine the first two “formants”—frequencies that are emphasized by the vocal tract and define how a vowel sounds—as well as their fundamental frequency. “When you run sound through a resonating chamber, it emphasizes various parts along it—those are the formants,” Sawallis explains. The researchers detected the presence of five vowel-like formant configurations in the sounds they recorded from the baboons. In addition, the baboons used the same vowel-like sounds in different vocalizations, and strung two sounds together to produce “wahoos.”
The researchers also dissected the heads of a male and female baboon to measure the lengths of their vocal tracts and vocal folds as well as examine the tongue muscles. The monkeys’ vocal tracts were comparable with those of human children whereas their vocal folds were more similar in size to those of adult humans. And like us, the baboons appear to produce vowel-like sounds by moving their tongues horizontally and vertically.
Taken together, the results suggest the raw machinery for speech may have been present in the last common ancestor of humans and baboons. They bolster the findings of a recent study by evolutionary biologist Tecumseh Fitch of the University of Vienna and colleagues. Those researchers used x-rays of macaque vocal anatomy and computer models to show these animals were capable of producing humanlike speech sounds.
The two studies are complementary, says Fitch, who was not involved in the newer work. “They’re saying what baboons really do; we’re saying what [monkeys] could do.” His only “quibble” with the study was the argument that the baboons were producing vowel sounds despite their high larynxes; instead, he suggests that perhaps the animals are actually lowering their larynxes—but that doesn’t change the study’s conclusions. “The key question is simply, can a nonhuman vocal tract produce enough sounds that it could communicate linguistically? I think the conclusion of the Boë study and our study is, yes,” he notes.
One reason nonhuman primates were thought incapable of producing humanlike speech, some linguists believed, was every language had to have an “i” sound (pronounced “EE”). That sound, which monkey vocal tracts may be unable to manage, serves as a calibration that helps humans discriminate other vowels. But this may not, after all, be a necessary feature of language, says Ian Maddieson, a linguist at the University of New Mexico. “We now know quite a lot of languages that don’t have that sound,” he says.
“This study by Boë et al. is an elegant piece of research,” Adriano Lameira, an anthropologist at Durham University wrote in an e-mail. (Both he and Maddieson were not associated with this work.) “The results add to our understanding of how human speech—namely the vowel systems of the world’s languages—build upon primate precursors and how vowel systems operate within anatomical constrains that we also share with our primate cousins.”
Not everyone is as enthusiastic. Lieberman takes issue with the analysis technique. Rather than picking out formants, he argues, the LPC analysis is picking out harmonics, or overtones at multiples of the fundamental frequency. Sawallis acknowledges that confusing harmonics and formants is a risk with using this method but notes, “the trick is to identify the LPC setting that is least likely to misidentify the formants,” and says he and his colleagues took pains to do so.
Their result opens a whole slew of new questions about when and how language evolved. And scientists can now use this approach to look for this ability in other species, Sawallis says. “If the seeds of speech were planted 25 million years ago instead of 100,000 years ago,” he points out, “that’s an enormous difference.”