How do dogs understand words?
I imagine that few dog owners haven’t wondered, as they watched their pet sniff with profound absorption at a patch of grass, how a dog might explain the attractions of the olfactory world if he had the gift of speech. Do dogs translate smells into stories with a past and future tense? Could they teach us words for smells that we didn’t know to name? Certainly, fiction has often given talking animals a rich vocabulary for smell, just as the Inuit were once believed to have a hundred words for snow. (In a sequence of the comic strip “Calvin and Hobbes,” a boy interrogates his friend, who happens to be a tiger, about words for smells. “What’s the word for how wet leaves smell?” the boy asks. “Snippid,” the tiger replies. “What’s the word for how I smell?” “Terrible.”)
According to Alexandra Horowitz, an expert on canine cognition who recently published a book on how dogs use their sense of smell, we can only speculate loosely, based on “what dogs spend a lot of time smelling,” about what a canine vocabulary of smell might be like. Dogs would likely create words for olfactory features that mark individuals: for example, hormones that distinguish males from females, old from young, this old, healthy neutered male from that old, healthy neutered male. They might name a spectrum of smells belonging to food, a spectrum of smells belonging to other species, and a spectrum of smells belonging to places where other species live. And because dogs might use olfactory cues to “anticipate future events or someone who is arriving,” they might augment their language of smell with linguistic markers that express future or past activity. (Perhaps suffixes to express past and future tenses: snippid, the smell of wet leaves; snippidae, the smell of leaves that were recently, but are not now, wet; snippida, the smell of leaves that will soon be wet, as the air smells like rain.) They might create markers to convey that a smell is freshly here, has been here for a long time, or is not here any longer.
The problem with this line of thought, of course, is that dogs lack the capacity to create words. Humans have linguistic protocols embedded in the brain that allow us to create words, that is, lexical units of meaning that obey morphological rules; dogs, which lack those protocols, don’t create mental “words” as we would recognize them.
We know that dogs can learn—to an extent—the meaning of human words. A border collie named Chaser has famously learned the names of more than 1000 toys; she can fetch any of the toys when given its name, and even perform actions in response to commands made up of nuanced sentences. Moreover, we know that dogs analyze separately, as humans do, the meaning of a word and the tone in which it is spoken; a study published in Science last year showed that the brains of dogs, scanned as the dogs lay still in an MRI machine (they were very good dogs), separately registered the tone and meaning of spoken words, and responded powerfully when the tone fit the meaning.
Nevertheless, dog owners tend to believe that their pets understand language better than they actually do. This leads to common mistakes in communication—for example, when we repeat a command in different words, as though to explain what we mean. “Synonymy is not a useful concept for dogs,” Horowitz says. “You have to use the same language in some way. Then even if they don’t understand something in the sentence, they have an audio cue that they can translate into a meaning that works for them.”
Another common mistake, Horowitz says, is to use a dog’s name as we would a human name, for instance adding it as a tag to a sentence. Dogs may not understand their names as we understand ours; rather than as something that belongs to them, they may see a name simply as a finger pointed in their direction—“a cue that we’re talking about them.” We make better use of a dog’s name when speaking it in isolation than when adding it to a sentence.
In short, dogs do not understand words in the complex way that humans understand them. Still, one feature of human morphology may overlap surprisingly well with dog communication. This is the phenomenon that linguists call phonetic symbolism.
We know that dogs interpret high and low-pitched sounds as meaningfully different. The meanings they interpret concern social relationships, emotional states, and possible threats. Unlike the meanings that we derive from the arbitrary symbols of our words, tonal meanings are embedded in the brains of dogs and indeed most animals; they enable dogs to comprehend each other as well as other species. “For instance,” writes the animal psychologist Stanley Coren, “low-pitched sounds (such as a dog’s growl) usually indicate threats, anger, and the possibility of aggression. Low-pitched sounds basically mean ‘Stay away from me.’ In general, high-pitched sounds mean the opposite. They indicate, ‘I’m no threat,’ ‘It’s safe to come closer,’ or ask, ‘May I come closer to you?’”
Humans are animals, too, of course, and our own responsiveness to high and low-pitched sounds comes out in many aspects of our cultures. The distinctive speech that parents in many cultures use with infants, which used to be called motherese (now caretaker language), is high-pitched, as though to reassure hearers of their safety; as Horowitz notes, babies and dogs alike take special interest in this kind of speech. Horror movies, in the meanwhile, tend to use soundtracks that alternate two kinds of sounds to create a chilling effect: extremely deep sounds, which give hearers the uneasy sense of a looming threat, and high, sharp sounds, which resemble the screaming of wounded animals.
Phonetic symbolism, which the linguist Steven Pinker calls “a quaint curiosity of English and many other languages,” may represent a different realization of this general pattern. The term refers to our tendency to think small when we hear vowels pronounced in a part of the mouth that amplifies higher frequencies (beet, bit), and to think large when we hear vowels that are pronounced in a part of the mouth that amplifies lower frequencies (father, core, cot). “Thus mice are teeny and squeak,” Pinker writes, “but elephants are humongous and roar. Audio speakers have small tweeters for the high sounds and large woofers for the low ones. English speakers correctly guess that in Chinese ch’ing means light and ch’ung means heavy.” In the slang of audio technicians, adjusting the knobs on an audio panel on a large scale is frobbing, on a medium scale is twiddling, and on a minute scale is tweaking: “The ob, id, and eak sounds perfectly follow the large-to-small continuum of phonetic symbolism.”
Some linguists suggest that phonetic symbolism reflects our subliminal awareness of the shape of the mouth cavity: a small cavity for tweak, a large cavity for frob. However, a simpler explanation may be that phonetic symbolism participates in the tonal consciousness that we share with so many other animals. The ability to create words is uniquely human. But perhaps here, as elsewhere, the tonal meanings that run beneath our words sound the depths of our animal nature.