Ferris Jabr has recently written a highly illuminating article for The New York Times Magazine titled, Can Prairie Dogs Talk? (May 12, 2017), on the pioneering work of Dr. Con Slobodchikoff, an emeritus professor of biology at Norther Arizona University. Professor Slobodchikoff has been analyzing the sounds of prairie dogs for more than 30 years, and he thinks that they possess a form of genuine language. Specifically, he claims that when they give alarm calls for different kinds of predators, they identify not only the type of predator, but also its size, shape, color and speed. In other words, their messages do not consist merely of nouns; instead, they are more akin to descriptive phrases. In a follow-up interview with Professor Marc Bekoff (Psychology Today, May 14, 2017), Slobodchikoff argues that since the rate at which the alarm calls are produced tends to correlate with the speed of travel of the approaching predator (hawks, for example, elicit only a single bark because they are so swift), prairie dog talk also contains something analogous to a verb in human language. Most surprising of all, prairie dogs are capable of coming up with new alarm calls for abstract objects which they have never seen before, such as an oval, a triangle, a circle, and a square. And if that were not enough, it turns out that prairie dog calls, like human language, are composed of phonemes. Indeed, Slobodchikoff even declares that prairie dogs have the most complex language of any non-human animal.
Professor Slobodchikoff contends that it is only pure prejudice on the part of “human exceptionalists” (many of whom are linguists and philosophers) that prevents scientists from describing prairie dog calls as true language, rather than mere “communication.” In addition, many people’s thinking is still influenced by Aristotle’s Scala Naturae, which ranks humans at the top, followed by “higher” mammals such as apes and then “lower” mammals such as mice (and of course, prairie dogs), with birds, reptiles, amphibians and fish lower down in the pecking order, and with insects, worms and one-celled animals at the very bottom. Such a view, argues Slobodchikoff, is speciesist and profoundly primatocentric. It is time for scientists to cast aside their prejudices and recognize that humans are not the only animals that can talk.
Is Slobodchikoff right? In today’s post, I’d like to explain why I’m inclined to be skeptical of the claim that prairie dogs are capable of anything like language.
Some of Slobodchikoff’s claims remain contentious
First of all, Slobodchikoff’s scientific colleagues are divided as to the merits of his claim that prairie dogs communicate to one another about the color, shape and size of an approaching predator. To quote from Jabr’s article:
Some scientists worry that Slobodchikoff’s studies, especially the early ones, are too small and depend too much on unreliable techniques. “The statistical approach he uses can be treacherous,” says Julia Fischer of the German Primate Center in Göttingen. “It tends to pick up patterns that might not be there. If you redo his analysis with modern techniques, I’m not sure how strong it would be.”…
Other researchers counter that Slobodchikoff’s techniques are sound and widely used and that reluctance to embrace his research owes more to prejudice than empiricism. “It seems to me that some people don’t trust Con because what he has found is outside what they are willing to accept,” says James Hare, a biologist at the University of Manitoba who studies ground squirrels. “But when you look at it all scientifically, I can’t really pick apart his methods. He presents compelling evidence of fine-grained communication about color, shape and size. I think language is a perfectly reasonable thing to call it.” (Emphases mine – VJT.)
But the greatest problem with Slobodchikoff’s field work, as Fabr candidly acknowledges, is that there’s no evidence to date that prairie dogs actually use the information which is communicated to them about a predator’s color, size and shape:
Slobodchikoff’s playback experiments demonstrate that different predator-alarm calls trigger distinct escape responses, but so far he has not been able to link the acoustic variations that ostensibly encode color, shape and so on to any observable behavioral differences. Without such evidence, he cannot rule out the possibility that some of the discrepancies in the alarm calls are an inadvertent byproduct of prairie-dog physiology — an increased sensitivity to a certain color or shape invoking a more forceful rush of air through the vocal tract, for instance — and that the animals do not recognize such differences or use them to their advantage. Perhaps part of what Slobodchikoff deems prairie-dog language is just useless prattle. (Emphases mine – VJT.)
Accusations of primatocentrism are bogus
Professor Slobodchikoff contends that it is primatocentrism that largely accounts for the reluctance of scientists to accept that prairie dogs are capable of genuine language. As someone whose Ph.D. thesis was on the topic of animal minds, I have to say that’s simply not true. Indeed, Fabr’s own article refutes Slobodchikoff’s claim:
In the South Pacific, biologists have shown that humpback-whale songs are neither random nor innate: rather, migrating pods of humpback whales learn one another’s songs, which evolve over time and spread through the ocean in waves of “cultural revolution.” And baby bottlenose dolphins develop “signature whistles” that serve as their names in a kind of roll call among kin.
With the help of human tutors, some captive animals have developed especially impressive linguistic prowess. Dolphins have learned to mimic computer-generated whistles and use them as labels for objects like hoops and balls.
Dolphins, which belong to the order Cetacea, are only distantly related to primates, but scientists working in the field take claims that dolphins are capable of language very seriously. I don’t know of any researcher in the field who would argue that just because they’re not primates, they’re incapable of language.
Fabr goes on to describe the impressive abilities of Alex, the African gray parrot which “learned to identify seven colors, five shapes, quantities up to eight and more than 50 objects.” So much for primatocentrism.
I would also like to address Slobodchikoff’s claim that the thinking of many linguists and philosophers is still colored by Aristotle’s Scala Naturae. Frankly, I don’t buy it, for three reasons.
First, it’s not “speciesist” to assert that some animals have vastly superior linguistic abilities to others, any more than it is to assert that some animals have a vastly superior sense of smell, vision, or echolocation, to that of other animals.
Second, it simply doesn’t follow from the fact that the genetic differences between humans and other mammals are relatively slight – mice, for instance, are 85% similar to humans in their protein-coding genes – that the psychological differences differences between humans and other mammals must be differences purely of degree. Such an argument assumes, fallaciously, that small changes can never give rise to large-scale effects or to qualitative (as opposed to quantitative) changes.
Finally, there are solid scientific reasons why great apes, cetaceans and elephants would be expected to possess a high degree of intelligence, making them at least prima facie candidates for the possession of language, while other mammals would not be expected to possess this trait. To see why, I would invite readers to have a look at this table here, from Harvard Medical School’s Kirschner Lab of Systems Biology, showing brain weight, encephalization quotient and number of cortical neurons for various kinds of mammals. (The figures for the number of cortical neurons are a little out-of-date, and Wikipedia has some more up-to-date numbers, but the overall trend is clear enough. By any measure, humans and other hominids, as well as dolphins and some whales, score very highly, while rats and mice do very poorly. There are no figures for prairie dogs, but it’s worth noting that they belong to the same family as squirrels, which are right near the bottom of the Harvard Medical School’s table. A squirrel’s brain weighs just 7 grams, and its E.Q. is just 1.1. And while it’s true that birds also have small brains, scientists now believe that their densely packed brain cells are what explains the highly sophisticated cognitive abilities of the more intelligent birds, such as songbirds and parrots. A parrot brain, for instance, may have as many neurons as in a mid-size primate. The prairie dog, on the other hand, is a mammal, so its brain would not be expected to possess the dense packing of neurons that we see in birds.
The straw man: language and consciousness
In his article, Fabr suggests that the human exceptionalist claim that language is unique to human beings is tied to the view, defended by some philosophers, that language is a prerequisite for consciousness. However, argues Fabr, “there is now a consensus that numerous species, including birds and mammals, as well as octopuses and honeybees, have some degree of consciousness, that is, a subjective experience of the world.” Let’s leave aside the fact that the ascription of consciousness to octopuses, let alone honeybees, remains highly controversial. What I object to most in Fabr’s argument is the assimilation of the view that language is unique to humans to the Cartesian view that animals are unconscious robots – a view which the vast majority of human exceptionalists would categorically reject.
What Fabr overlooks is the possibility that while language is not a prerequisite for consciousness, it is a prerequisite for what neuroscientists call higher-order consciousness, which is thought to be confined to only a few species of mammals (and perhaps birds), with some researchers still maintaining that it is unique to humans. In his paper, The Neurobehavioral Nature of Fishes and the Question of Awareness and Pain (Reviews in Fisheries Science, 10(1): 1–38, 2002), Dr. James Rose defines this kind of consciousness as follows: “Higher-order consciousness includes awareness of one’s self as an entity that exists separately from other entities; it has an autobiographical dimension, including a memory of past life events; an awareness of facts, such as one’s language vocabulary; and a capacity for planning and anticipation of the future.” It is surely arguable that language is a prerequisite for this type of consciousness.
Counter-evidence from Slobodchikoff’s own research findings
In his interview with Professor Marc Bekoff, Slobodchikoff makes much of the fact that prairie dogs are capable of producing novel alarm calls in response to stimuli that had never been seen before:
However, one of the most surprising things was that prairie dogs were able to come up with alarm calls for abstract objects that they had never seen before, such as an oval, a triangle, a circle, and a square. This shows a level of abstraction that people did not expect from a ground squirrel.
This sounds very impressive, since the ability to create new signals is regarded as one of the hallmarks of true language (it’s referred to as productivity, in linguistic jargon). However, in his article, Fabr includes some extra information which undercuts Slobodchikoff’s claim that prairie dogs are using language. In the passage below, he describes an experiment designed to test the animals’ reaction to a novel stimulus:
First [Slobodchikoff and his colleagues] built plywood silhouettes of a coyote and a skunk, as well as a plywood oval (to confront the prairie dogs with something foreign), and painted the three shapes black. Then they strung a nylon cord between a tree and an observation tower, attached the plywood figures to slotted wheels on the cord and pulled them across the colony like pieces of laundry. Despite their lack of familiarity with these props, the prairie dogs did not respond to the cutouts with a single generalized “unknown threat” call. Rather, their warnings differed depending on the attributes of the object. They unanimously produced one alarm call for the coyote silhouette; a distinct warning for the skunk; and a third, entirely novel call for the oval. And in a follow-up study, prairie dogs consistently barked in distinct ways at small and large cardboard squares strung above the colony. (Emphases mine – VJT.)
Why, one might ask, did the prairie dogs give exactly the same calls for an object that they’d never seen before? After all, would you expect a group of English speakers to all come up with exactly the same word to describe an object that they’d never seen before? The behavior of the prairie dogs in this case certainly doesn’t sound like language use to me.
I decided to check Slobodchikoff’s own work, as I did not want to be accused of misrepreesenting him. I happen to possess a copy of The Cognitive Animal (edited by Marc Bekoff, Collin Allen and Gordon Burghardt; MIT Press, 2002), which contains an article by Professor Slobodchikoff titled, “Cognition and Communication in Prairie Dogs” (pp. 257-264). The article describes Slobodchikoff’s experiment with novel stimuli in considerable detail. Allow me to quote a brief excerpt:
All of the [painted plywood] models elicited alarm calls. The calls were recorded and each prairie dog’s alarm call in response to each model was noted. For each type of model, all of the prairie dogs consistently called the same for that model, using the same frequency and time components. Although none of the prairie dogs had ever seen a large black oval before, they all had a call that corresponded to the presentation of the black oval, and this calls was significantly different from their calls for either the coyote silhouette or the skunk silhouette. (2002, p. 262. Emphases mine – VJT.)
Again I have to ask: why the unanimity, in response to a novel stimulus? After all, it’s not as if the prairie dogs had any time to confer among themselves, and say to one another: “Let’s call it a click-clack.” What made them give the new stimulus the same name? What the unanimity suggests to me is that it wasn’t a name at all, but a physiological response to a stimulus – which would mean that it is not language.
Language: what’s it all about?
Whatever reservations one might have about Slobodchikoff’s claims that prairie dogs can actually talk to one another, it is undeniable that the communication system employed by these animals is a very sophisticated one. So it is reasonable to ask: why not call it a language, of sorts? In his essay, Fabr makes a powerful plea along these lines. He notes that we are perfectly willing to ascribe tool use to animals, despite the fact that the tools we make are far more advanced than those of any non-human animal. Why, then, are we so reluctant to attribute language use to animals?
What this argument overlooks is the radical, qualitative difference between human language and animal communication systems. One of Slobodchikoff’s critics who was interviewed by Fabr was Yale University linguist Stephen Anderson who argued that the defining feature of language is the ability to systematically combine symbols into an infinite array of sentences. Fabr then asked Anderson what an animal would have to say, before it could be said to possess true language. Anderson replied that if you showed a parrot a fruit that it had never seen before – say, a pineapple – and it said, “My, that looks spiky, so I don’t think I want to eat it,” that might qualify as language.
By contrast, prairie dog communication is confined to a very limited number of nouns (for different kinds of predators), a modest collection of adjectives (used to describe the color, size and shape of a predator), and perhaps two verbs, for fast and slow motion respectively. (Here, I’m assuming for the sake of argument that Slobodchikoff’s characterization of the structure of prairie dog calls is as sophisticated as he claims it is.) With a vocabulary like that, how many sentences can you make? Not very many, and certainly not an infinite number. Why, then, would anyone be tempted to describe such a communication system as language?
What about tools, then? We are willing to allow that animals can make primitive tools, so why not primitive language? First of all, it appears that the impressive tool-making feats of animals such as Betty the crow have been over-rated: recent research suggests that she didn’t use insight to fashion wire into hooks after all, as bending is actually part of a crow’s natural repertoire (Was Betty the crow a genius—or a robot? by Virginia Morrell, Science, August 9, 2016). What’s more, crows that have been hand-reared in isolation turn out to display the same kind of tool-making behavior as that of crows in the wild, leading some researchers to conclude that the tool-making behavior of crows is largely innate, although it is worth noting that crows raised in isolation never went on to display the more sophisticated behaviors found in crows in the wild, which were exposed to models and competitors (Kenward B, Rutz C, Wiet AAS, et al. 2006. “Development of tool use in New Caledonian crows: inherited action patterns and social influences.” Animal Behavior 72:1329–1343).
Second, one could argue that the difference between, say, a hook fashioned by a crow and a simple human tool – say, an Acheulean handaxe – is one of degree rather than kind. Indeed, it has been recently argued that the manufacture of the Acheulean handaxe was largely under genetic control, and that the changes in tool-making seen seen in the late Acheulean period (300,000 to 200,000 years ago) are the result of a shift from genetic transmission to greater reliance on cultural learning (Corbey R, Jagich A, Vaesen K, and Collard M. 2016. The acheulean handaxe: More like a bird’s song than a beatles’ tune?. Evolutionary Anthropology: Issues, News, and Reviews 25(1):6-19). With language, on the other hand, the fundamental unit of meaning is not the word but the sentence, as Wittgenstein so convincingly argued in his Philosophical Investigations. A sentence is not like a tool but an assortment of tools all working together – in other words, a machine. And machines, like the sentences of a language, are capable of displaying an infinite variety of forms. Finally, machines, like the sentences of a language, are entirely unique to human beings. In likening language to mere tools rather than mechanisms, Fabr has been employing a mistaken analogy.
What do prairie dogs talk about? The universality of language
Suppose I told you of a remote mountain tribe with a highly complex language, unlike anything you’d seen before. You might ask me to show you a dictionary of this tribe’s language. Suppose I were to show you one, and you found that all of the words in the dictionary were about the tribe’s enemies. What’s more, there are only two verbs in the dictionary, meaning “fast attack” and “slow attack.” The adjectives are all used to describe invading enemies, too. What would you think if I told you that this was a complete dictionary of the tribe’s language? I’m quite sure that you wouldn’t believe me. You would impatiently retort: “Don’t these people ever say, ‘It’s a beautiful day,’ ‘I love you,’ ‘That’s not fair!’ or ‘What are you doing tonight?’ What’s wrong with them? Why do they talk about nothing other than their enemies?”
Prairie dogs are like the strange tribe I described above. All they seem to talk about, as far as Slobodchikoff can tell, is predators. Their communication system might be complex, but its range is narrow.
Language, on the other hand, is a universal tool,as we saw above. It is not designed for talking about enemies: indeed, it is not designed for talking about anything in particular. That’s because its open structure enables language users to talk about everything. Consequently, if we observe an animal communication system in which users send messages relating to only a single topic, then we should rightly be skeptical of claims that it qualifies as a language.
Before I sign off, I’d like to make one more observation. Remarkably, the prairie dog communication system, despite its complexity, lacks two little phrases that we’d expect to find in any language: “That’s right” and “That’s not right.” I leave it to readers to draw their own conclusions.