Don’t Be Silly
I have recently plowed through the new book by Berwick & Chomsky summarizing their long held views of language. One point they never tire of mentioning is what they call language’s Basic Property, complete with capital letters: “language is a finite computational system yielding an infinity of expressions.” For the moment, I’m going to accept that claim, because the important thing to notice is not whether the definition is true or false but that it is ridiculous.
Let’s list a few other “computational systems” that have the same property:
- Vision is a finite computational system yielding an infinity of images.
- A biological cell is a finite computational system yielding an infinity of organisms.
- The equation X = n+1 is a finite computational system yielding an infinity of numbers.
There are perhaps an infinity of examples I could generate, but only the algebraic example gets at the essence of whatever I am trying to define. In the other cases it is distracting to call them “computational systems.” Even if we are trying to model vision, or cells, or language on a computer—and thus reduce it to something computable—we still need to remember that the computational part is not the essential property. If we are trying to construct a visual system, for example, we are trying to create something that can use light to recognize its environment. If it works well, it will automatically be able to generate the limitless number of images that it may detect. Don’t mistake the side-effect for the goal.
Visual cortex, cells, languages all do something that results in something. It is the result itself that matters, not the number of possible results.
Ridiculous definitions lead to ridiculous conclusions. If you take the basic property of language to be the ability to yield an infinity of expressions, you find yourself taking seriously a sentence like: How many cars did you tell your friends that they should tell their friends … that they should tell the mechanics to fix? [page 123] Let’s just say that if you find yourself producing sentences like this one for any reason other than making a point about unbounded length, you need to go to a hospital.
It is no wonder then that the authors end up in the land beyond the looking glass. They have gone off on the wrong foot.
Finding the Right Foot
What characterizes a linguistic result? Many of us would point to a communicative result, but Berwick and Chomsky are more interested in our ability to think of a verbal proposition. Instead of quarreling with them, lets give them their due. One of the results of using language is the ability to think quietly in words. I’m very verbal myself, and am well acquainted with the experience of lying in bed quietly mulling something over before hopping up and acting on my thoughts. So plainly, private thinking is one of the results of knowing a language.
Since the authors are writing about the origin of language, we have to ask another question. Was private thinking the first result of language usage? The authors say yes, but I do not believe them. When I lie in bed pondering what I will spend the day writing, I ponder in English. Where did that English come from? I learned it from the English speakers and writers around me. Pondering is a secondary result that arises from previously having lived in a cultural setting. So that is something we can say about language: it is a cultural artifact.
Yet something is wrong there. Cultural artifacts depend on tradition and can be lost by breaking access to a tradition. The slave trade broke many traditions as people were captured in Africa, taken to the Americas and mixed with people who knew nothing of the old traditions. But the descendants of these captives created a variety of creole languages. Similarly, deaf children placed in a new school for the deaf will create a sign language. So something inborn seems to be at work as well. This mix of instinct and culture is peculiarly human. So I am going to call language a human artifact.
The Real Basic Definition
Berwick and Chomsky argue that speech is externalized thought (which they call internal-language or I-language), and it is true that not all thoughts depend on culture. Einstein, for example, claimed that little or none of his thought used language. He imagined and manipulated images. I don’t think like Einstein, but I do know what it is to have internal, non-linguistic knowledge. For example, when driving an automobile along a highway, I pay attention to the road. When rounding a curve at high speed I press the accelerator to maintain speed (identified by feel) and am alert to the cars in nearby lanes, adjusting my car's position as I go to minimize the risk of collision. This process is non-verbal and has a very old history, older indeed than human, or even Homo-, history. All mammals perceive their world and sometimes get beyond simple reflex actions by paying attention to their perceptions.
Of course Berwick and Chomsky do not accept the idea that I-language is based on perception. They argue that we manipulate abstract concepts which are mysteriously available only to humans; however, there is a great deal of evidence to support the proposition that language externalizes the perceptual process.
- Language is about something specific. That is to say it draws attention and focuses on something. This trait, of course, is fundamental to perception. We do not just perceive; we perceive something.
- What is more, we always perceive something in terms of ourselves. We don’t just see the highway. We see it in relation to our own position. Likewise, language always expresses a point-of-view. Thus, the same event can be described in different ways depending on the speaker’s point of view: e.g., I sold/bought a shirt today. The sentence, Two people exchanged a shirt for cash on Jan. 31, 2016, is a third way of describing the event, this time from an outsider’s viewpoint.
- Perception can be narrow or wide. That is, while driving down the highway I can focus on what is directly ahead of me, or expand my vision a bit to include what is in the other lanes. Language too has a variable scope. Thus, the same event just cited can be described with an expanded scope as I bought a shirt today on a busy street corner.
- Finally, perception imposes spatial and active relations between things perceived. Space relations cover things like near, on, in, etc. Active relations establish a doer and a done to. Any action requires a doer, even if one has to be made up. It rained all night has a dummy doer (it). We can change that to Rain fell all night, but then the rain becomes the doer and the action becomes falling.
It is important to grasp that these four features are found in all natural languages, and nowhere else. Not computer languages, which might say something like BUBBLESORT STRINGY. Like all sentences in a computer language, the example is a control that makes the computer do something. The sentence focuses on nothing, expresses no point of view, and has no scope, spatial or case relations. Equations assert computational relationships rather than perceptual relations. 2 + 2 = 4 is basic math, but it asserts a relationship without a focus, point of view, scope or perceptual space.
I do accept a long standing argument repeated endlessly by Berwick and Chomsky in their latest book: speech is an externalization of a way of understanding the world. But I say our understanding is sensual, not an abstract mentalese. The authors also insist that there are a great many ways that our understanding can be externalized, and I buy that; hence there are many differences between languages, although all languages bear the four marks of their perceptual origins.
So here is the definition of language I propose: A language is a human artifact used for contemplating or communicating topics as though they were perceptions.
Note: In my next post I will show how this definition of language explains syntactic puzzles proposed by Berwick and Chomsky.