Atheist mathematician Jeffrey Shallit takes issue with my recent post in which I pointed out that animals are incapable of logic or language. He makes several points. I’ll address animal language in this post.

Shallit:

… there is an area of science that is actively interested in testing [the claim that animals have language], although you’d never know it from reading Egnor. It is a branch of ethology, which is the science of animal behavior. Contrary to Egnor’s claims, the evidence for animal language is quite strong, although of course there are doubters. Animal language exists in many different animals, including bees, elephants, dolphins, baboons, and whales. (I am not an ethologist by any means, but I can recommend the eye-opening books of primatologist Frans de Waal.)

Language in animals has never been demonstrated, because animals are incapable of language. Claims of animal language have been made by some ethologists, but those claims are mistaken. A closer look at language and animal behavior helps to clarify the misunderstanding.

We should begin with an examination of what we mean by language. A meaningful notation or event is a sign. A sign presents to the mind an object other than itself for attention. The word “cat” is a sign. A fire alarm is a sign. Darkening clouds that mean rain is a sign.

There are, for our purposes, two kinds of signs — signals and designators. Signals are objects or events that draw attention to something else in physical or temporal proximity. Fire alarms, smoke (signifying fire), dinner bells, and such are signals. Signals are linked intimately to the object to which they point — the fire alarm sounds loudly in the vicinity of the fire, the smoke rises from the fire, and dinner bells are rung in the dining room.

Designators are words that refer to the objects that they name. Designators are arbitrary, in the sense that there is no intimate linkage between the designator and the object designated. “Cat” is a designator of the animal of that name, although the letters c-a-t have no intimate physical link with the animal designated, aside from the linguistic link. Designators are abstract.

Signals are not language. Human language is an abstract process that relates designators in grammatical relations to objects designated. Language has a variety of properties, including semantics, arbitrariness between the designator and the object of the designation, unlimited productivity (an infinite number of sentences can be produced via a grammar that governs relations of designators), reference to particulars or to abstract concepts denoted by the designator that are not physical objects in the vicinity, among others.

Some signs can function as signals or as designators. The shout of “FIRE!” in a crowded theater is both a signal and a designator. “FIRE!” as a signal draws attention to the emergency and implies the need to exit immediately, and “fire” as a designator draws attention to the combustion itself.

The confusion between signals and designators is at the root of ethologists’ misunderstanding about animal “language.” The sounds and gestures that wild and domesticated animals use to express emotions and desires are signals, not designators, and these expressions are not language. Natural animal signals have no grammar, have no arbitrariness between sign and object, have limited productivity, don’t express concepts abstracted from particulars, etc.

In captivity, under laboratory conditions, animals can be trained to use signals (which they use naturally) in patterns that mimic designators and mimic language. But none of it is actual use of abstract designators to produce genuine language. Animals never use signals to designate any object that is not in the animal’s perceptual realm. Animal signals refer to objects of perception and emotional states and the like. Animals do not signal abstract concepts.

Further evidence that all animal signs are signals and not designators is that there are no examples of animals who acquire signs via pure collocation of signs (such as by textbook), without any direct perceptual acquaintance with the object named. Learning by textbook is characteristic of genuine language — the user of the language can generate an infinite variety of new designators using linguistically based learning without any necessary perceptual acquaintance with the object designated. If one has language, one can learn history from a book without personally experiencing the events described. Without language, learning can occur only by actual perception of the object to be learned.

Animals are incapable of learning by textbook without perceptual acquaintance with the object learned. This restriction of animal learning to perceptual acquaintance is characteristic of learning via signals (i.e., training), and is not characteristic of learning via language. There has never been an animal that learned the meaning of a sign via inference from designators that point to its meaning, absent direct contact by the animal with the perceptual object so designated. Animals learn only by perception, not by textbook.

Animals communicate by signals, not by language. Some animals, after sufficient training by human researchers, use signals that mimic language. But mimicry is not language itself, any more than training a parrot to mimic the Preamble to the Constitution is training the parrot in constitutional law.

It is easy for gullible or biased researchers (a regrettable predilection of ethologists) to be duped into believing that an animal that they have spent years training has acquired the use of language. The prototypical example of this error is the case of Clever Hans, a horse who appeared to understand language and think abstractly but was subsequently found to merely be responding to subliminal training on the part of his investigators. The error is so common in ethology that it is called the Clever Hans Effect. It haunts research in animal behavior and haunts efforts to demonstrate animal language and abstract thought.

Animals do not naturally use designators, and they do not use designators as language even in laboratory situations. All animal signs are signals. Animals can be trained to use signals that mimic designators and language, but animals have no language. Only humans have language.

Shallit, again:

Contrary to Egnor’s claims, the evidence for animal language is quite strong, although of course there are doubters. Animal language exists in many different animals, including bees, elephants, dolphins, baboons, and whales. (I am not an ethologist by any means, but I can recommend the eye-opening books of primatologist Frans de Waal.)

Shallit mischaracterizes de Waal’s work. De Waal is a pioneer in the study of animal emotion and moral behavior. De Waal’s views on the link between thought and language are nuanced and are not views I share, but his view on animal language is worth noting.

De Waal:

You won’t often hear me say something like this, but I consider humans the only linguistic species. We honestly have no evidence for symbolic communication, equally rich and multifunctional as ours, outside our species.

Perhaps, in addition to recommending de Waal’s books, Shallit should read them.