Decades ago, while Philip H. Lieberman was soaking in a bathtub and listening to the radio, he heard anthropologist Loren Eiseley ponder an evolutionary puzzle: Why couldn't monkeys talk? Like us, they're social primates, intelligent and certainly not quiet. Rhesus macaques grunt, coo, screech and scream. Infant macaques make sounds known as geckers. Despite the grunting and geckering, though, no other primates — not even the chimpanzees and bonobos, our nearest ape relatives — can make the vowel and consonant sounds we know as speech.
Scientists figured there were two likely sticking points. Either the brain was not wired for speech in nonhuman primates, or their windpipes were shaped the wrong way.
Lieberman, now a professor emeritus of anthropology and cognitive and linguistic sciences at Brown University in Rhode Island, got out of the tub and took the puzzle with him. In groundbreaking experiments with rhesus macaques in the late 1960s and early 1970s, Lieberman and his colleagues pinned the problem to monkey throats. They concluded that macaques lacked a sufficient supralaryngeal vocal tract, the space in humans that begins in the mouth and follows the hump of the tongue into the throat. Even if a monkey brain had the correct wiring for speech, the monkey vocal tract simply couldn't produce adequate sounds to talk.
This vocal tract explanation caught on, appearing in textbooks and even a science comic book. “Among experts in the evolution of speech, this idea was common but not widespread,” said W. Tecumseh Fitch, a professor of cognitive biology at the University of Vienna, in an email. “But among biologists, anthropologists, and psychologists it was, and still is, very widespread.”
Recently, what seems like a whimsical question — Why can't monkeys talk? — has turned into a serious and heated debate among former collaborators. In December, Fitch and colleagues published a paper in Science Advances that announced its counterpunch in the title: “Monkey vocal tracts are speech-ready.” On Friday, a pair of reports published in the same journal advanced the match to rounds three and four, with Lieberman's reply and a response to his reply by Fitch and the other authors of the December study.
For Lieberman's earlier study of rhesus macaques, published in 1969 in the journal Science, “We took a rhesus monkey and the started to see what the anatomical limits were,” he said. The researchers made a plaster cast of a monkey throat from a macaque that died naturally. With a live but sedated monkey, the researchers manipulated the animal's tongue and documented the positions it could make. Using this information, they estimated the range of monkey speech sounds. It was much smaller than a human's, and indicated that a macaque could not produce vowels, such as the long E, common to most languages.
The researchers followed this work with X-ray videos of infant humans, whose tongues resemble those of monkeys at birth but shift toward the throat as they grow. Additionally, Lieberman posited that Neanderthal vocal tracts resembled those of infant humans, too. And though Neanderthals must have been capable of limited speech, Lieberman said, they would not have spoken with the clarity of an adult human.
Since the 1980s, Lieberman has focused on primate brains; as early as 1968, he said, his work demonstrated that primates other than humans didn’t have the brainpower for complex speech.
In December 2016, a team of cognitive biologists and anthropologists replicated Lieberman's macaque research using more advanced techniques. Among the study's authors was Fitch, who once was a graduate student of Lieberman's. The key difference in the more recent work was that Fitch and his colleagues took X-ray videos of live macaques as they made noise or chewed.
From the new research, “we had much more data” than the plaster or silicone casts of decades prior, said Bart de Boer, an author of the 2016 study and an expert on the evolution of speech at the Vrije University in Brussels. “It turned out that that represented a much bigger range of possible sounds.”
Using 99 images of rhesus macaque vocal tracts from the X-rays, these researchers simulated the three-dimensional space in the monkey mouth and throat. By mapping the flow of air through this space, the scientists generated a hypothetical range of speech sounds that monkeys could produce. The “phonetic potential” of a macaque, they concluded, was eight times larger than estimated in 1969.
The scientists generated the phrase “Will you marry me?” as if spoken by macaque. The words are odd and clipped, but the authors argued it sounds as comprehensible as English with a slight “foreign accent.” (Less charitably it sounds like Gollum, from “The Lord of the Rings,” trying to speak through a chokehold.) Crucially, the models showed the macaques were capable of vowel sounds, as in “bat,” “bet,” “bit,” “but” and “bought.”
“It certainly showed that we had been underestimating the range of sounds that they can make,” de Boer said. “We should focus our research on the evolution of cognition if we really want to understand the origins of speech.”
The reaction to this research among experts studying the evolution of speech, “was actually very positive and supportive,” said de Boer. (With the exception of Lieberman, he added.) In the years leading up to this paper, he said, “the idea that cognition is more important than vocal anatomy was already gaining traction” among evolutionary specialists.
At the time of the 2016 paper's publication, John Esling, a linguist at Canada's University of Victoria who was not involved with this work, told Science Magazine that, “This certainly shows that the macaque vocal tract is capable of a lot more than has previously been assumed."
Anna Barney, a University of Southampton in England biomedical acoustic engineer who also was not involved with the research, told the New York Times in December that the new research was convincing but raised questions, such as a lack of macaque consonants. “What they’ve shown is that monkeys are vowel-ready, not speech-ready.”
To Lieberman, the 2016 study was not a refutation but a confirmation. “There’s nothing really new in the Fitch paper except a series of misrepresentations with the intent to deceive,” he said. What he called “monkey-speech” still could not produce the long E vowel sound, as in see or beet. “They still have no ‘ee.’ They barely have an approximation of an ‘oo’ and an ‘aa.’ ”
The long E has special status among the vowels, he said — in fact, research indicates it is possible to estimate the size of a person based on the sound of her or his long E. It is the most easily identified vowel sound, he said. There's really only one way for our vocal tracts to make it, he said, but “for virtually any other vowel there are different ways to skin the cat.”
To produce a long E, humans pull their tongues upward and forward, creating a large back cavity above the larynx while constricting the forward area, creating a 10-to-one difference in volume from back to front. “You can’t do this with a nonhuman primate. The tongue is almost flat in a monkey,” he said. “You’d have to take a knife to cut the tongue in half. It’s simply impossible.”
Listening to someone speak a long E “is an optimal signal that allows listeners to estimate the length of speakers' vocal tracts — a necessary step in determining what a person intended to say,” he said.
De Boer acknowledged that the long E was special to humans as it “sits in a corner of the [human] acoustic space.” But, he countered, apes might have a different acoustic corner that helps estimate size. “The fact that macaques cannot produce 'ee' is therefore a bit of a red herring,” he said.
“The importance of the vowel remains debated,” Fitch said. “Lieberman argues it is important and we are unconvinced.” In fact, Fitch and de Boer described the long E as “mythical” in their latest report, a phrase that Lieberman found “offensive and demeaning” and likened to a “Trump tweet.”
Barney, in an email to The Washington Post said, "My own view regarding [long E] is in agreement with theirs – it is not in a unique category in terms of determining the size of an individual’s vocal tract."
In his new comment on Fitch and colleagues' paper, Lieberman cited Charles Darwin, who reflected on the perilous position of our trachea in 1859: Food has to pass over our windpipe. Other primates do not have this arrangement. This improves our range of sounds, Lieberman said, but the opening also increases the chances we will choke on food. In his view, the selective advantage of speech explains this choking hazard. (That humans are more likely to choke on food than other primates is an intuitive idea but remains unproven, Fitch said.)
“We are not saying that the human vocal tract is not fine-tuned for speech,” de Boer said. He said he believed it is, though noted that “there is still a lively debate about this.” But, he said, their point was that a monkeylike vocal tract does not necessarily preclude primitive speech. “Speech could therefore have started to evolve even in a situation where our ancestors still had a monkeylike vocal tract.”
Despite decades of observation, however, monkeys have not yet begun to speak.