Last spring Tecumseh Fitch published a textbook that every regular on this blog should have to hand, The Evolution of Language. I call it a textbook, but it's more like one of those classic 19th century science texts, Lyell's Principles of Geology is a good example, that surveyed a field in terms of some particular theory. Even if you disagreed with the theory, the book was a must-have because of the vast amount of data presented. So any criticisms I have to offer—and I've got 'em—have to be understood in the context of saying, but of course you have to check out this book.
While I'm on the subject of this book as an invaluable resource, let me protest the skimpy indexes—one for authors, another for subjects—that grossly reduce the book's convenience. As one instance, I immediately determined that B.F. Skinner is in the bibliography but not in the index of Authors. Does that mean he's not even mentioned in the text? No; I quickly came across a dismissive reference on page 20. Tsk-tsk. Are they trying to make us depend on Google Books for our quick lookups?
I was going to title this post "Chomsky's Bulldog" because the theory that Fitch uses to hold his book together is straight out of Chomsky, but then I realized that Chomsky is the bulldog and Fitch the quieter one. I always get a kick out of Chomsky's outrageous dismissal of rival points of view. Agree or disagree, you know where the man stands. Fitch, on the other hand, claims to be reasonable, open to all sides, and a reader ignorant of the battle lines might be fooled into thinking Fitch is just a referee calling fouls and goals as he sees them.
Early on Fitch quotes Susanne Langer as saying that scholars are unlikely to spout pure nonsense, so when something they say is dismissed as tommyrot, probably the argument being dismissed is but a gross caricature of what they really said. To which, of course, there can be only one reply, "Nonsense!!" And I don't say that just because there is no better example than Chomsky of a scholar who triumphed through the derisive dismissal of a leading rival.
Scholars, with their insatiable need for something new to say, tend inevitably toward nonsense. The discipline that keeps them away from the yellow brick path to the wizard of hokum is science with its constant demand for some kind of empirical evidence. Without evidence, metaphysics can take easy root.
One technique Fitch uses to smooth over quarrels is to say that people who quarrel about language are often using the same word (language) to mean different things, and, he suggests, the quarreling would go away if we only realized that we are using one word in different ways. Chomsky, for example, would be less controversial if we realized that instead of using 'language' to mean "language in toto," we realized that he used the word to "denote…the computational mechanisms central to syntax" . Well, that sounds reasonable. If Chomsky wants to look at the origins of syntax while I'm looking at the origins of the speech triangle, why not live and let live?
Actually, there is more at stake than a quarrel over a pun. If we are trying to understand the origins of language, there are a couple of possible scenarios:
- Speakers and listeners began to communicate about topics, and as they became more fluent at the task their language became more syntactically structured.
- Thinkers began to use syntax to organize their thoughts and as they did so they came to understand the benefits of cooperation and began to give voice to their thoughts on particular topics.
So here we have two a priori speculations about where language might have come from. They cannot both be right, even if 1 and 2 use language in different senses. The only scientific way to select between the scenarios is to see where the evidence points. What kind of evidence should we look for?
One clear difference between the scenarios is in the role of the individual in relation to language. Language is somehow built into the brain in Chomsky's thought-first scenario, while it is learned from others in the topics-first approach. Empiricists, like Morten Christiansen and Nicholas Chater, see language as 'out there' to be learned while nativists, like Fitch and Chomsky, say there is an internal, I-language, and the language out there is merely the sum of all those little I-languages. How to settle the dispute? Look for factual evidence.
For example, a paper published in 2007 surprised everybody. This was a study of genes and tonal languages by Daniel Dediu and Robert Ladd (abstract) reporting that a feature of many languages (tonality) did not reflect the genes of individual speakers but of the population as a whole (see: Was The First Language Tonal? and Do Genes Bring Up the Rear?). The implication of that study was that there are genetic differences underlying at least one linguistic feature, but the difference is in the speaking population as a whole, not in the individual speakers. That finding seems to contradict the idea that the language of a population is the sum of all the I-languages, although it is not decisive and I don't know of anybody from either side of the empiricist/nativist divide who predicted it.
Since nobody expected this finding, both schools had to jump. It was not so demanding on the part of empiricists, however, who already focused on group teaching, joint attention, and some kind of conformity bias that might encourage the use of tonality even if it is not in one's genetic disposition.
Nativists were going to have to work a little harder to absorb the finding. Here is how Fitch handles it:
There is no evidence that populations of humans are genetically predisposed to learn the language of their community more than any other (though see Dediu and Ladd, 2007). 
I don't know about you, but I feel that a finding contrary to the assertion just made should be highlighted a bit more forcefully. Not quite 270 pages later, Fitch does add a few sentences on the Dediu/Ladd work, concluding, "The causal significance, if any, of this gene/language correlation remains unknown" . It is true, of course, but some kind of hint at the possible importance of a previously denied gene/language correlation should be given readers.
Instead, Fitch is content to assert that the language of a population (what Chomsky calls E-language) , "is simply an aggregate epiphenomenon, no more than the output of a set of I-languages" . But a population's language is not an "aggregate epiphenomenon" if its internal realizations (however that might be) came from the population rather than being inborn. Nor is language an "aggregate epiphenomenon" if it reflects the structure of the population as a whole rather than the sum of the individual structures.
Nevertheless, Fitch concludes, "I concur with Chomsky's argument that scientists interested in the genetic and neural mechanisms underlying language need to focus on I-language, as instantiated in individuals' brains. I-language and the capacity to acquire it are the core systems we seek to understand biologically" .
Well, you can wish he were a little more open to rival approaches, or at least to admitting that there are controversies more profound than how to define language, but we are going to have to take this fact-laden book as it is.
More on this fat book next week.