Mozart's Starling

by Lyanda Lynn Haupt

  Aiming to bolster his theory, Chomsky coauthored a 2002 paper published in Science with Harvard biologist and psychologist Marc Hauser and Tecumseh Fitch, a linguist at the University of St. Andrews in Scotland. In “The Faculty of Language: What Is It, Who Has It, and How Did It Evolve?,” the three researchers sought to identify the distinguishing features of human language and its development. They scoured the literature to compare aspects of human and animal communication, eliminating shared characteristics one by one. Memory, for example, is critical to language, because, as Hauser put it in an interview, “If you couldn’t keep in mind several pieces of a sentence, you couldn’t understand anything. But memory’s not specific to [human] language.” Some animals use memory in communication too. So memory was off the Unique to Human Language list. The researchers continued down their catalog of linguistic features until they were left with just one property that could be said to define human language alone: recursion.

  To test their theory, Hauser and Fitch created a study to see whether a nonhuman primate with complex vocal capacity could recognize recursive patterns. They selected the cotton-top tamarin, a tiny, beautiful New World monkey originating in Colombia. They chose the species for its vocal complexity, but if you have seen a cotton-top tamarin, you might wonder if it wasn’t chosen for its cuteness. They are among the earth’s smallest primates, with dark faces surrounded by long manes of white, feathery fur. Their gaze is intense, open, curious. Somehow, the look on their faces seems to say, Sure, let’s study recursion! Have at it! What do we do first?

  Hauser and Fitch started by creating a simple artificial language. The words were made up of short sounds pronounced by either men or women, sounds like li and mo. The difference in pitch between the male and female voices made it easy for anyone listening to tell the two groups apart. Then the scientists combined the sounds into patterns—short sentences, really—using two rules. The first rule was very simple: a female sound would always be followed by a male sound. With A and B representing the two sound groups, this simple rule would yield short sentences such as ABAB. The second, more complex rule embedded a female-male sound pair within one of the AB pairs, with a resulting A(AB)B.

  To create a baseline from which to judge the tamarins, the researchers first tested humans, innocent young Harvard undergrads, in fact, to see whether they could learn to recognize both patterns. After listening to thirty sentences, the students were tested on new ones. More than 80 percent could correctly recognize the recursive sentence patterns. Then Hauser tested the monkeys in his lab. The tamarins listened to recordings of the patterns over and over in the evening, and in the morning, the recordings were played for the monkeys again, along with the recordings that violated the patterns. When the monkeys recognized a novelty, they would turn toward the speaker and listen intently. According to Hauser, the monkeys noticed when the ABAB pattern was violated but were not able to pick out violations to the A(AB)B pattern. Chomsky was not the least bit surprised.

  In a disconcerting footnote, several years after the publication of the tamarin study, Hauser’s professional colleagues began to question his conclusions—and his research ethics. Hauser was a well-known tenured professor at Harvard in 2007 when the school began an internal investigation around possible scientific misconduct. By 2010, the allegations came to public light, and scandal erupted around him. While Hauser had honestly reported that the tamarins could not recognize recursive patterns, the U.S. Department of Health and Human Service’s Office of Research Integrity found him guilty of several counts of research fraud pertaining to other aspects of the tamarin studies—including fabrication of data and false description of research methods—in the published 2002 paper and in other, ongoing work. Hauser ended up resigning from his Harvard post. It might seem that the study served, in the end, only to sharpen the conclusion of linguists like Chomsky. But instead, it led to significant research with another animal that would take the debate much further.

  When University of California, San Diego, neurophysiologist Timothy Gentner reviewed the tamarin study (prior to the Hauser scandal), he immediately thought, My birds could do this. He had studied starlings for years, though he hadn’t come to them through an interest in birds or birdwatching or ornithological science. He’d come to starlings through gibbons. Gentner found himself unemployed the summer after he finished college, and took a volunteer job as a docent at the Woodland Park Zoo in Seattle. There, he heard the daily vocalizing of the siamangs—large, black-furred gibbons native to Malaysia, Sumatra, and Thailand. They’ve been popular primates at the zoo for decades, famous for their singing. The males and females join in ritualized vocal duets culminating in the males’ resonant boom, which can be heard, literally, a mile away, as those of us who have lived in the vicinity of the zoo can attest. Every day, the inhabitants of all the houses and apartments anywhere near the zoo have their dinner, their sleep, and their romantic tumbles interrupted (or enhanced) by the booming of the siamangs.

  When Gentner started as a graduate student in psychology at Johns Hopkins, he couldn’t get the siamangs out of his head; he wanted to study vocal patterning in animals and perhaps its implications for humans. His adviser suggested he start with the starlings they had in the lab, and Gentner never looked back. “The more I learned about these birds, the more interested I got,” he told me. “And they have never ceased to amaze me. They are so adaptive, so clever. Every question I throw at them, they are able to answer, one way or another.” He’d researched starlings for years by the time the tamarin study came around. He believed his starlings could recognize the patterns that the tamarins couldn’t; in fact, he didn’t think it would even be that hard for them. From his own observations, Gentner knew that both male and female starlings created wonderful and idiosyncratic songs by mingling an array of rattles, chirps, and warbles into uniquely patterned motifs and that they recognized one another as individuals based on these motifs. “Starlings are so tuned in to sequences. I knew they could do this. It was just a matter of figuring out how to test it.”

  Like Hauser, Gentner created a simple language, but his was based on starling-ish sounds—warbles and rattles. He tested eleven birds, and though it sometimes took thousands of trials, nine of the birds learned to recognize the recursive A(AB)B pattern more than 90 percent of the time, and with a high degree of proficiency, actually discerning the pattern when three new pairs of warbles and rattles were embedded between the original pair.

  The implications were ripe for the picking. “Our research is a refutation of the canonical position that what makes human language unique is a singular ability to comprehend these kinds of patterns,” Gentner declared. “If birds can learn these patterning rules, then their use does not explain the uniqueness of human language.” That is, if a songbird could recognize recursive syntax patterns, then the place of recursion in Chomsky’s linguistic views would have to be reconsidered.

  And the wild rumpus began. Linguists, psychologists, biologists, ornithologists, evangelicals. Everyone jumped into the fray initiated by these tattered little birds—birds so unwanted by the wider world that Gentner didn’t even need a wildlife permit for his research assistants to nab them from the trees. The not-yet-debunked Dr. Hauser’s response to Gentner’s study was seemingly open-minded. He pointed out that although the starlings could recognize recursive patterns, there was nothing to indicate that they could comprehend meaning in the patterns, which is true. (This, of course, would be much harder to study, since humans do not speak starling, or at least not yet.) But he recognized that Gentner’s study went far beyond what he was able to show with the tamarins, acknowledged it as an “important paper,” and claimed that he was inspired to think about ways he might re-create and improve his tamarin study by implementing some of Gentner’s methods, perhaps using tamarin rather than human sounds and giving the monkeys more tries at listening to the short sentences.

  Others, many of them linguists, disagreed strongly with Gentner’s conclusion. W
hen interviewed for the New York Times, Geoffrey Pullum, a linguist at the University of California, Santa Cruz, and coauthor of The Cambridge Grammar of the English Language, said dismissively, “I’m not buying it.” Pullum argued that the sentences involved in both the tamarin and starling studies were too simple to be of use in detecting cognitive abilities that had to do with grammar. It might have told us something about the abilities of starlings, but it said nothing about the more complex subject of human language. Yet as Gentner rightly pointed out on a UK radio program, “Primarily when humans are listening to acoustic patterns they’re perceiving language.” Here we are talking about the acoustic recognition of syntactical patterns, a linguistic ability, so if birds and humans can recognize the same patterns, why—and how—would we draw a line between them? And as a starting place for intellectual discourse, why would we? It seems far more expansive, more respectful of our complex intellect and the potential scientific secrets residing within the bodies and brains and lives of the creatures with whom we share the earth, to respond to such a study with our minds wide open, with a sense of curiosity and adventure.

  Gentner described to me some of his new research that seeks to answer the question of why starlings might have the need for such advanced pattern-recognition capacities. In this work, he is focusing on the second part of the starling song, the one that mixes a starling’s individually collected motifs into a long sequence. This section of the song has two components: the variety of motifs, and the patterns they occur in. One thing about starlings that is unusual among songbirds is that they are open-ended learners. Most songbirds learn their species-specific calls and songs in the first year, then they’re done—they’ve successfully learned the “language” of their species. But starlings continue to learn new sounds, to bring in more sophisticated motifs and sequences throughout their lives. One morning recently I heard Carmen making an odd new sound. YEEeeeEEEK! Over and over. “Hi, honey.” I offered one of her familiar phrases, hoping to distract her from this unpleasant noise, but she was not to be deterred. I finally figured out what it was when I was walking past Carmen’s aviary into the dining room—she was perfectly mimicking the creak in our old oak floor.

  Gentner was studying the way starlings of different ages pattern their motifs into sequences, and he discovered that older birds were more predictable than younger birds. The sequences of young birds are mostly random, while the sequences of individual older birds (a one-year-old versus a four-year-old, for example) are more predetermined. Say you are listening to a four-year-old male starling. If you have heard this bird sing before and paid attention to the sequence of his repertoire, then you can predict fairly well what the eighth (and ninth and tenth) motif will be. Females, it turns out, prefer such predictability (preference is measured easily enough by observing which males the females choose as mates). Older males make better mates by all agreed-upon bird measures: they declare and defend better and bigger territories, claim better nest sites, and, together with females, build better nests. They provide more and better food for their egg-sitting mates and their eventual chicks. And—the real measure of evolutionary-biological success—they fledge more young. Thus, life-or-death decisions for the starling are based on their pattern-recognition ability (and the female’s leading part in mate choice feeds Gentner’s intuition that females are slightly better than males at pattern recognition—something he has not yet been able to measure definitively).

  But there is a point of diminishing returns—if a song is too predictable, it will become less attractive to the female. There appears to be a balance, a “sweet spot,” Gentner suggests, between novelty and habituation. Too much novelty is unsettling; too much habituation is boring. Listening to Gentner talk about all of this, my mind turned to music, to the use of set motifs and refrains within a changing musical landscape—enough of both to keep listeners charmed and involved but still excited and waiting to see what comes next.

  Just as writers use recursion to relieve the monotony of brick-by-brick sentences, musicians use recursion to allow their music to grow from within. The most famous example of recursion in beginning music-theory classes is Beethoven’s Fifth Symphony, where the familiar da-da-da-DUM opening can be traced easily via recurrence and variation throughout the evolution of the movements. Mozart uses musical recursion frequently. Good music, like a successful starling song, represents a bounded complexity, and a lot of similar basic aesthetics are built into the patterning of language, human music, birdsong, and natural sound.

  “I’m thinking of music,” I told Dr. Gentner as we talked about the balance of predictability and variation in starling songs. “Yeah,” he said, “me too,” and launched into an esoteric exposition of early versus late Coltrane, during which I completely spaced out. I was pondering Imogen Cooper at the piano, wandering through variations on a theme in Mozart’s Piano Concerto No. 17.

  For the past few months we have had a quiet visitor in our household. Carmen’s Aunt Trileigh and Uncle Rob (of the Guppy Incident) generously loaned us their Gnome Chomsky, the garden Noam statuette. Garden Gnome Chomsky was the first sculpture created by Steve Herrington for his small Portland, Oregon, company Just Say Gnome! Their mission: “To create and market garden gnomes… that will bring a bit of both humor and peace to people’s lives and will hopefully also inspire deeper political, environmental, and spiritual awareness and reflection.” We have loved having Gnome Noam here on the living-room mantel, emanating an aura of combined impishness and wisdom that captures the real Chomsky well. Like any good gnome statue, Garden Noam wears a fine red-pointed hat and brown boots, but instead of standing in a ring of mushrooms, he poses next to a table topped with scholarly books, wears glasses, and smiles to himself.

  I have harbored a fantasy that Carmen would provide an irresistible photo op by perching on Garden Noam’s little red hat. Dr. Chomsky is famous for refusing to make personal comments on certain issues, and I hoped that such an image might inspire him. I would send him the photos along with my linguistic queries, and he would be charmed into responding. Instead, Carmen was deathly afraid of Gnome Chomsky and wouldn’t go anywhere near him for three days; after that, she’d tiptoe over to investigate only after I put dollops of her favorite peanut-butter-with-bits-of-arugula on his boots. Eventually she warmed to him, and as long as I stayed close by, she’d visit Garden Gnome and give his glasses little exploratory gapes. Contrary as ever, she has refused to sit on his hat for my imagined photo, but Tom managed to get a few nice shots, one of which I did send to Dr. Chomsky, who apparently remained unmoved. I never heard from him and must join the cadre of writers who are forced to say, “Noam Chomsky declined to comment.” I love the photo anyway.

  Carmen and Gnome Chomsky, the Garden Noam. (Photograph by Tom Furtwangler)

  As it happened, Chomsky’s rejection of the suggestion that Gentner’s study might invite some re-thinking of his own work was immediate, complete, and rather truculent. For Chomsky, all the study showed was starling memory tricks, rattle counting, and number storage (all of which are rather interesting in a bird, let’s not forget). “It has nothing remotely to do with language—probably just short-term memory,” Chomsky said in his terse response to the New York Times. (Later research by Gentner uses more diversity of patterns and fewer repetitions to counter the possibility that the starlings are memorizing sounds rather than recognizing patterns.)

  It’s a bit of a sore subject for Chomsky. The starlings are not the first challenge to his theory that recursion is both unique to humans and universal to human language. There is a tiny tribe, the Pirahã (pronounced “pee-da-ha”), living at the mouth of the Amazon. There are only about three hundred and fifty Pirahã people in twenty or thirty small villages scattered along the Maici and Marmelos Rivers. They speak no outside language, and their own tribal tongue—a confounding mixture of clicks, rattles, lip flicks, air intakes, and birdsong-like bouts of prosody—have befuddled nearly every linguist to encounter the group, no matter how
many years are given to its study. Surface stats on the language would lead to the conclusion that it ought to be simple to learn; there are only eight consonants (females use just seven) and three vowels. But the need for consonants falls away as tribal members slip into humming, whistling, and a complex rainbow of tones, stresses, and variations of syllabic length.

  Daniel Everett, a linguist at Bentley University, is the only non-Pirahã in the world to speak Pirahã. He originally came to the Pirahã in the 1970s as an evangelist working for the Summer Institute of Linguistics, or SIL, an international group that trains language specialists to translate the Bible into the tongues of remote tribal groups. The organization doesn’t attempt a lot of direct evangelization. Their belief is that once isolated communities get their hands on the Bible in their own language, their conversions will naturally follow.

  In his decades of working with the Pirahã, Everett has slowly come to understand that as far as he can discern, they have no collective memory, no original creation mythology, and an insistent denial of complexity regarding numbers and amounts (the Pirahã recognize only one, two, and many, and attempts to teach them to count have failed). There is little grasp of or attention to the past or the future and little concept of the lives people are living when they are not standing there in the flesh, at least not in the way most of us typically understand such things. A person who walks into the woods to go hunting is simply “out of experience,” says Everett. There is a profound “ethos of the present moment,” as he calls it.