Surely there can be language without thought. Why not thought without language? Where is the division between them?
For example: in the process by which one word recalls others of similar meaning or sound, it is not necessary for the triggering word to occur consciously. Often things I see and do bring to my mind thoughts of other things which sound like the names of these sight or actions, or which occur together with those names—even in contexts where the name of the thing is equivocal and means something else.
This could, of course, be explained as an instance in which words function as things within a thought in itself nonverbal. That is not so strange a phenomenon; it must happen in writing poetry. But if we adopt this explanation, where do we stop? We could do away with the notion of specifically verbal thinking altogether; which is absurd. We know that many things are true about things of which we have no other experience other than hearing their names. If we dismiss these as thoughts about reports, we must explain how our knowledge holds when we come to experience them.
There are forms of thinking prior to language—clearly, the thoughts of animals—and there are forms of human thought in which language is not only unnecessary, but a hindrance. Efficient mental calculation requires the omission of intervening words, even such as times and equals, instead hearing only numbers and rendering a result by a process that feels less like reckoning than recognition. You draw best what you cannot name, or whose name you have forgotten for the purpose. In any game, or any system of rules yielding winners and losers, though words are of use beforehand in study and planning, they distract when you come to it—not because they are slow or awkward, but because they involve too much. Words, by which we compass the world, always drag the world in; but to play the game well, you must enworld yourself in it. To play with words in your head keeps the rules from sinking in, keeps your thoughts off the rails. You think of the game itself, its origin, use, nature; of the wordings of the rules, of the form of the strategies—how they resemble strategems of nature or of the strategems of other games.
This kind of thinking canot help you win by rule: the only use of words in games is in cheating. And this is good. Intelligence is for cheating. We cannot win any of nature's games by nature's rules. We are not strong enough, or fast enough. We are too big to hide and too small to shake off attack. But we can change the rules: levers for strength; shoes, boats, tame horses for speed; blinds and camouflauge for hiding; walls and armor for bulk. All invention is cheating; and cheating is made possible by language.
Some animals, of course, can do what may be called cheating—chimpanzees, dolphins, ravens, &c.—the list is long and growing. Some of these can be discounted—complex acts, yes, but only very complex moves within the rules. Yet some are cheaters for sure; do they have language, then?
There may be a kind of thinking in between nonverbal thought and language which is not (in the old phrase) sub-vocalized speech, but rather sub-verbal language; or, if you prefer, language may be able to exist without words.
Language is the means of stepping outside of the rules, of recognizing rules as being only rules and to be broken. All this requires is a faculty of association. In human beings this faculty is untyped, consistent, and social. By "untyped" I mean that we can associate things without logical connection: when we say the sky is angry, we do not mean sky:?::person:anger nor that the change in the appearance of the sky is isomorphic to the change in the appearance of a person becoming angry. We are simply associating the ideas sky and angry.
Intelligent animals are intelligent because they too are capable of untyped association. I would say that a chimpanzee that fishes for ants with a stick has an association between ant and stick. I do not think that the chimpanzee has any thought in this process corresponding to tool. Having an association between ant and stick, he tries to keep them both in mind, to direct his attention to both at once—which means, for a forgetful animal, first that he picks up the stick and stays near the anthill; then that he touches them together; &c. The stick and the anthill each have only so many degrees of freedom; the chimpanzee need only iterate the act of holding the association in mind to have a stick covered with delicious ants.
Untyped associations are illogical; so if untyped association is a necessity of thought, then so is illogic. We retain a naive habit of thinking of logic as a function of intelligence; but living the age of the computer, we have no excuse for this mistake. Logic is embedded in nature: transistors and electrons, chutes and marbles, can reason. Illogic, consequently, is never real; but it takes the highest complexity to pretend to it.
That is one out of three for animals. The other two—consistency and sociality—are only ours. Any idea is an act of differentiation of the chaos of experience. When I say stick I isolate a segment of a continuum that runs from splinter and sliver through branch to tree and forest. And sticks are always stick to me. So when I say that a monkey picks up a stick, I impute to the chimpanzee's thought a quality of my own thought—the stability of stickness. But the chimpanzee's thought may in this instance be closer to tree, and the stick he ends up with the limit of what his strength and dexterity could manage. Or he might have been thinking closer to branch, and again been defeated; or to splinter, and been unable to break the wood; or just of wood, and picked up the stick off the forest floor. On each occasion that I observe him with a stick his thought of what he carries may be anywhere on that continuum.
If his association has no stability, he cannot teach another monkey what a stick is, because he cannot, as we can, associate two consistent ideas—one that must be named (stick), and one that names itself (the sound stick)—so that another can imitate the association. Thus his associations are not transferable—they may be imitated in a broad sense, but the ideas associated to the same result may not match up at all.
Some of the above is probably reinvention. And certainly all of this could be formulated more precisely in the terms of neurology, semiotics, linguistics, &c. I take the risk because I know of no work on the continuity of human and animal thought but arguments about whether it could be—never about what that continuity might be.