Mason Inman - science journalist

2007-04-03, 18:40:54 

I just went to a Q&A with Noam Chomsky at MIT, hosted by the Knight Science Journalism Fellowship program.

The first question someone asked was the main thing I've wondered about Chomsky's ideas about the evolution of language. He's become sort of infamous for supposedly saying that language does not seem to be a product of natural selection, or that there's no evidence that human language is a product of natural selection.

I was never quite sure what he meant by this, even after attending a couple long talks by him on language and the mind at UC Berkeley a few years ago.

Finally, now, I know what he thinks. And I owe it to my bladder.

Chomsky's talk was split into two halves—first he talked about language, then about politics. When we got to the halfway point, he had to pee, and so did I. I was kind of nervous about following him into the bathroom, but when we were in there, he was nice and he started up small talk with me, asking who I write for. Like many people, he asked, "So, you're able to make a living with freelancing?" and sounding skeptical that anyone ever could. But in any case, this opening gave me the chance to ask him more about his ideas on the evolution of language.

Most people would assume that human language is so complex and so central to our lives that it must be something that evolved through natural selection. But Chomsky's view—assuming I understood it correctly—is that when most scientists talk about natural selection and evolution, they are being very sloppy and jumping to conclusions. This is as true when talking about the evolution of human language as for the evolution of the human eye and tons of other traits. Let me explain why he takes this stance.

When asked about what role natural selection may have played in the development or evolution of language, he told the Knight fellows that the more we learn about evolution, the more we find that the mutations and random variations that are viable and helpful are actually highly constrained by physical and chemical laws. So evolution and natural selection are not synonymous. Natural selection is one process that helps direct evolution—that is, the change over time of groups of organisms—but physical and chemical laws also channel evolution.

This I agree with in general, and it's not a new idea, although until recently very few evolutionary biologists talked about this aspect of evolution. It's something that Stephen Jay Gould talked a lot about, and before him Alan Turing and D'Arcy Thompson, as Chomsky noted.

Chomsky also pointed out how there's basically no evidence for natural selection of human language. We don't have recordings from back then of people talking, or fossils showing soft tissue such as vocal cords, so we can't really say much about it. He said that most scientists observe organisms' traits and see that they're well fitted to their environments and lifestyles, and then make the leap to assuming that natural selection was responsible for this.

This is the same criticism that Gould made often when he railed against "adaptationism," the assumption that any identifiable trait exists and has its particular properties—say, why animals have legs and why they're the lengths, color, texture that they are—because those traits were shaped by natural selection to make the organism well adapted to its environment.

In very few cases do researchers actually have really solid evidence for natural selection having acted to create a trait. Philosopher of science Elizabeth Lloyd describes the various kinds of evidence for natural selection that researchers can get in her book The Case of the Female Orgasm. She argues against the idea that females—especially female humans, aka women—evolved through natural selection to have orgasms. Instead, she argues that women's orgasms are a by-product of selection on men for having orgasms.

(It's like the question "why do men have nipples?" They don't serve any obvious purpose for men, but seem to be there as a by-product, because it's obviously useful for women to have nipples. Imagine there were some kind of genetic switch that would flip one way for female fetuses, leading to nipple growth, and that would flip the other way in men, leaving them with no nipples. This switch could sometimes go awry, leaving women with no nipples—and before the invention of baby formula, these women would have had a hard time raising babies unless they were able to recruit wet nurses. Now, there's clearly a lot of genetic controls that makes men's and women's bodies different—we're not all hermophrodites, and men don't grow whole breasts. But it does seem safe to say that men's nipples are a by-product of natural selection for women having nipples.)

I won't go into Lloyd's arguments about the female orgasm, because it's more complicated than the case of nipples. But I will say I found Lloyd's book, and her arguments that the female orgasm is an evolutionary by-product, totally convincing. Note, though, that she's not against using natural selection to explain traits; she shows how the evidence we have about female orgasms doesn't fit what we'd expect if that trait was created or shaped by natural selection. Many other traits do fit what we'd expect if they were products of natural selection (although that's not the same as having direct evidence that they were shaped by natural selection).

Chomsky takes things further, it seems, ruling out the possibility of talking about natural selection in many cases. Natural selection might have played a role in most of our traits; everyone except Christian fundamentalists accepts that natural selection is at work on organisms, Chomsky said during the discussion with the Knight fellows. It's just that there are other forces, such as physical and chemical law, at work, and also we have limited knowledge of what happened in the past. This means, he argues, that there's little we can say about how natural selection may have contributed to a trait.

However, I didn't feel my understanding of his position, from the discussion with the Knight fellows, was concrete or blunt enough. So during a break in between the discussions on language and on politics, I followed Chomsky to the bathroom. (I also had to pee, so this wasn't totally calculated.) He was nice and started talking to me when we were washing our hands, and I asked him more about natural selection and language.

I asked him to clarify what he was saying. Would his criticisms of people inferring that language is a product of natural selection also apply to pretty much any trait of any organism? That's right, he said. Then I asked him whether he thought that most researchers, when they talk about natural selection having effects, are being sloppy. He said, "extremely sloppy."

So it seems that rather than disagreeing on the facts, Chomsky disagrees with others about what counts as science, how researchers should proceed—and where they have to remain silent. I admire this in a way, since it's rigorous. Also it makes sense given his background in linguistics, in which a lot of things were presumed to be true, like the idea that we learn language by imitating what we hear, and then we gradually learn the rules. But he helped start a revolution that showed that many of those ideas were false, or that we just can't say how things work. This line of work has shown that as early as we can tell, infants and toddlers have a pretty much fully formed sense of the grammar of their language, even if they can't speak well themselves.

So Chomsky's view is respectible and understandable. But if we took Chomsky's approach to all of science, it seems to me that science would be a lot poorer. It would be much more limited to sets of facts and abstract descriptions of how these facts are related—such as the Chomsky hierarchy of formal grammars. Taking Chomsky's approach to science would severely limit our ability to tell a story about how things work, and to jump across gaps in our knowledge, to try to get at the underlying reality. When we paper over these gaps, we inevitably get things wrong—but we hope that we can eventually fix these mistakes as we gather more data and try to make it all fit together in some framework.

I guess Chomsky would rather have us live with these gaps in our understanding and acknowledge that they're there—and where we can say nothing, then we should remain silent. Now that I think about it, if I understood correctly, that was the view of the young Ludwig Wittgenstein and other logical positivists. ("Whereof one cannot speak, thereof one must be silent," Wittgenstein wrote.) And if I remember correctly from Chomsky's book Language and Problems of Knowledge, he thinks along these lines, too. I think it's too restrictive of an approach, but I can see how it can be a good rebuff to a lot of crappy ideas, sort of like bringing up the idea of Occam's razor.