Here’s how the pendulum swings:
- The original question about language origins, back when such speculation was taboo in respectable circles, asked how people had ever come to agree on the meaning of a particular word. The question assumed language was a cultural artifact.
- The question became more respectable when it was turned around to ask how humans ever evolved the ability to use language. Now the assumption was that language was a biological product.
- Then the pendulum swung back a bit, finding the midpoint between the two extremes. Language was said to be both a cultural artifact and biological product. It co-evolved. Like a bee adapting to an orchid and the orchid adapting to the bee, linguistic coevolution imagines the brain adapting to use language and language adapting to fit the brain. For a time it looked as though co-evolution was taking first place (e.g., see Co-Evolution Idea Won Big in Barcelona) in theories of linguistic origins.
- Now with Christiansen and Chater’s book the pendulum returns to its starting point. Language is a cultural product that, over time, has changed to be easily usable by the brain.
Experiments (e.g., Language Structure is Cultural, Not Genetic) showed that if you introduce a single learner to a simple, unstructured language and have that person teach the language to another, and that person then teaches another, and so on for ten generations of learners, you end up with a structured, easy-to-learn language.
Other researchers, most notably Christiansen and Chater, went on, however, to show the great problem of linguistic coevolution. Cultural change is so fast that biological processes cannot fix the changes speedily enough. Certainly words change too fast to be fixed, and nobody suggests any words are built into language. But grammatical rules change as well. Latin diversified into French, Italian, Spanish, Catalan, Romanian, Romansh etc. in just a few centuries. Thus, the arbitrary rules of language cannot co-evolve, and, what do you know, all the rules of Chomsky’s Universal Grammar [UG] are arbitrary. They cannot have evolved.
The authors admit that there is still one kind of possible evolution. They state that “specific cognitive mechanisms may have coevolved with language” [p. 29] if the mechanisms play a functional role such as making changes that allow for handling a larger vocabulary, remembering longer utterances, speaking phonemes more exactly, etc. These abilities persist even as language changes. The authors are not very fond of this possibility and play it down, but they grant it, and more power to them for their intellectual integrity. Their openness makes their book much more valuable than it would be if they were dogmatic about all their ideas. You do not have to agree with them about everything to still find the work useful and instructive.
This blog stresses the likelihood that powers of attention, sharing, and working memory all improved significantly after the first words were spoken. It is likely too that the mental ability to use metaphors and follow other figures of speech also evolved. It is nice not to have to choose between those ideas and the ones in this book.
The really important contribution is the authors’ insistence on pushing the linguistic evolution side of the story as far as it can be taken. What is the “geometry” of the mind that makes some sentences more learnable than others? Why does the effort to invent a logical language so often result in something unlearnable? Are there techniques writers and speakers could use to make their language more comprehensible and memorable? Answering these questions would be both useful and interesting.
Meanwhile, these days when Chomsky speaks on language he tends to insist that natural languages do not evolve. Oh, sure. They change, but they change within a constant framework. They might go from expressing case through morphemes to relying on word order, but cases remain the same. Well, you might ask, what about the evolution of articles in Romance languages? Latin did not have them; French, Spanish and Italian all do. But say the generativists, articles are not part of the UG. How about the loss of noun genders that occurred between Old English and today’s English? Again, not part of the UG.
Denying that languages evolve turns out to be much less of a claim than it might at first seem. Languages can fluctuate between being tonal to atonal, between declining nouns and not declining them, between categorizing nouns according to gender or class and not categorizing them and none of that counts as evolution. It’s like saying that all quadrupeds share a universal body plan, so quadrupeds do not evolve. The truth or falsity of the claim is less important than the way it serves as a blockade to learning about the ways the body can and cannot change. Likewise, an insistence on focusing on a UG bars the way from exploring what is malleable and what is stable in language. Generative linguistics has been stuck for a long time, refining its procedures to a minimum without actually adding new facts. Christiansen and Chater are offering a way of escaping the mud and their path needs to be considered seriously.
In the next post I will discuss the question of how literally we should take the word evolution.