On a summer evening years ago, I dined with a group of friends at a rural Midwest restaurant where the parking lot was a patch of rough ground without marked bays. We came out to find a Cadillac parked close in beside our car. Edging into the gap between the vehicles (the other side was also tight), we did our best to get the doors far enough open to slide in without dinging the Cadillac. Our close approach triggered the Cadillac’s motion-sensitive theft alarm. A loud synthesized voice told us: “You are standing too close to the car! Step! Away! From the car!”
We had to gain access to our vehicle, so we ignored it. The alarm system consequently repeated its vocal warning several times. Eventually it decided that we were incorrigible car thieves, and went into a paroxysm of sirens and horn honking and flashing lights. By then we had with great difficulty managed to squeeze in and get our seatbelts on. The Cadillac was still screaming and honking, but before the owner could abandon his baby back ribs and rush outside with his handgun, we had pulled out of the parking lot and made our getaway.
Now, the question I want to raise is this: Does a Cadillac, when its theft alarm is set, have language?
This apparently frivolous question relates to a New York Times Magazine article that Ferris Jabr published a few weeks ago on the work of Con Slobodchikoff, an emeritus professor of biology at Northern Arizona University. The title asks a question: “Can Prairie Dogs Talk?”
I briefly answered that question here, but some readers have thought my treatment inadequate. It is time for me to answer in greater depth.
Slobodchikoff thinks the answer is yes. He has found that a prairie dog, on spotting a potential predator approaching the colony, will emit a sequence of yips, using different yips for different kinds of potential predators (eagles, dogs, snakes, biology graduate students, or whatever). Moreover, as Jabr reports it, he found that “Beyond identifying the type of predator, prairie-dog calls also specified its size, shape, color and speed; the animals could even combine the structural elements of their calls in novel ways to describe something they had never seen before.”
Let me concede all of that for the sake of argument, and ask you to concede one thing to permit a Gedankenexperiment. Imagine Cadillac has upped its game on theft alarms, and cars can now adjust their siren and horn performances to different patterns depending on the height, shape, clothing, and approach speed of a person coming their way. (This is actually not implausible given current technology.)
I argue that a Cadillac, despite the clear communicative nature of its behavior, still does not have language.
First, the car does not address itself to any particular addressee: Its audience will be simply whatever organisms happen to be in the vicinity (possibly none). There is no communicative goal.
Second, intruder approach is the only thing it can communicate about. It is limited to producing messages like “Warning! Tall skinny human, blue shirt, approaching swiftly!” or “Warning! Short fat human, dark coat, approaching slowly!” (Notice that structural elements like SHORT, SKINNY, BLUE, SLOWLY, etc., can be combined to “describe” something the car has not necessarily ever encountered before.)
Third, the behavior is nonvolitional: The car cannot choose whether to produce the siren-sounding, horn-blasting, and light-flashing suitable for the situation. Its compulsion to warn is wired in.
Of course, errors and malfunctions may occur. My Saab has developed an unfortunate habit of random theft-alarm activation when no one is anywhere near, and this is proving hard to fix. A modern car’s nervous system is complicated and not entirely modular. And Saabs are not manufactured any more, so my mechanic’s electrician cannot consult the makers.
The situation with prairie dogs seems fully analogous (right down to our being unable to consult their creator). Even conceding that the creatures behave in ways that can warn fellow colony members differently for large swift-moving aerial threats, small slow-moving ground-traveling ones, etc., a wired-in compulsion to engage in such behaviors comes nowhere near constituting a capacity for language.
Humans talk voluntarily, often without any connection to present conditions. They can chat about hypothetical or long-past situations. Or choose not to.
Cars cannot so choose; and I suspect the same is true of prairie dogs confronted with a potential threat. Freedom from stimulus control is of the very essence when it comes to language of the sort that humans possess. (Notice, the elephant-in-the-room topic of syntax is not mentioned above; most linguists would point to it immediately, but for my argument here I don’t need to.)