How to Solve Mysteries
The reaction to our paper, somewhat predictably, devolved into a debate over centrality (or irrelevance) of discrete infinity.
But our paper starts not with discrete infinity but with “How to study the evolution of a trait”. One of the Seven Samurai is the evolutionary biologist Mike Ryan, whose day job is to study the evolution of communication. I hope you didn't skip that section.
Before raising the question of how tungara frog calls evolved, a lot had been known about their communicative system (whine only vs. whine plus chucks), the fitness costs and benefits (calls attract females as well as predators), the phylogenetic distributions of the traits (restricted to tungara and its sister species), and the underlying mechanisms in call production, perception and behavioral response (e.g., playback experiments, and I’m guessing more than a few have given their lives for the advancement of science). But its evolutionary history remained difficult to reconstruct. The obvious hypothesis, that the female evolved perceptual preference for lower frequency calls (and bigger males), turns out to be incorrect. The perceptual ability had been around in the clade for a very long time, but was recruited only after the males evolved large larynges that made chucks possible.
A lot of work, for a relatively simple system. In fact, whine and chucks are not unlike the two letter “simple language” that Alex complained about. But so far as I know, no one decided that the chucks—the species specific FLN for tunguras—were too narrow of a conception of the call system and started looking at other bodily noises instead. No one dug up fossils of Engystomops neanderthalensis and declared that these guys had chucks also (as if that solved the problem of evolution). And no one taught the calls to common toads so they could settle on a consensus through multi-generational breeding and transmission.
I'm not against EVOLANG. Looking at the program from the Vienna meeting, many presentations especially those in the poster session are straight up behavioral, neurological and computational studies of language. They would not have been out of place at the LSA, Cognitive Neuroscience Society meeting or the BU conference on language development, and these are very interesting and informative studies. Many of the research methodologies, including those in the studies we criticize, are well established and may inform us about language and cognition. If slapping on the sticker of evolution helps us get language research done, I’m all for it. (Like sex, evolution sells too, probably because they are the same thing.)
But suppose you are really after the Big Question. Suppose further that discrete infinity is anathema to you and you fancy cooperation, complexity, efficiency, social contracts, sound symbolism … instead as the key question for language. Fine. But by all means study it like Mike Ryan: "a clear specification of the target phenotype, empirical evidence linking details of trait design features to fitness consequences, an understanding of the comparative landscape in terms of homologous and analogous traits, and tests that distinguish adaptive from non-adaptive explanations for trait diversification.” We need empirical fitness measures and we need comparative evidence, which are a bit harder to obtain for isolates like us and our language. As the frog call study shows, adaptive logic may be sound but like any scientific hypothesis, is guilty until proven innocent.
I will soon return to the regularly scheduled programming, with a post on Baker’s Paradox, thereby continuing our tribute to the 1980s.