May 18, 2004

From Just So Stories to science, in biology and in pragmatics

Marc Moffett at Close Range suggests that Gould and Lewontin's complaint about adaptationism in biology -- that post hoc adaptationist stories are too easy to come up with -- also "infects pragmatic explanations in linguistics and the philosophy of language". He argues that this "is pernicious because the ready availability of pragmatic explanations ... allows one to preserve one's favorite semantic theory 'come what may'."

Moffett further points out the current "construction-based approach to grammar" makes it harder to restrain the proliferation of such stories, because it weakens Grice's principle of not multiplying meanings beyond necessity, which traditionally provides the core motivation for arguing that some meanings arise from pragmatic reasoning about the conversational context. At least I think that's his argument.

This is all a bit abstract. A concrete example might be the Gricean argument that sarcasm (e.g. "Yummy!" meaning "Disgusting!") arises from pragmatic reasoning in context rather than from a systematic multiplication of word senses. This idea is convincing mostly because of the parsimony argument, and not because the required pragmatic reasoning is obvious or compelling. For example, I had an exchange with David Beaver, Larry Horn, Ellen Prince and others last fall about why "reverse sarcasm" rarely works:

I described someone who comes home from a long hard day to find that the puppy has pooped on the rug, and says "oh, terrific!" (or "wonderful" or "great" or similar positively-evaluated adjective), meaning the opposite; and contrasted this with the same person finding a bouquet of roses, and saying "oh, disgusting!" (or "ugly" or "annoying" or similar negatively-evaluated adjective) to mean the opposite. I argued that the first is normal and the second is weird.

The best story about this seems to be David Beaver's: "you can sarcastically express a departure from a salient hope, not from a salient fear." But the whole discussion was very much an attempt to reason backwards to an explanation of an observed pattern; if we tried to predict the pattern without knowing in advance what it was -- say by running a theorem-prover on a set of conversational axioms -- I doubt that we would have gotten it right.

So to repeat, the conversational-implicature story is convincing here mostly because it's unattractive to systematically add a new, opposite sense to the meaning of every word and phrase that can be used to express a "salient hope".

How does"construction grammar" undermine this argument? Well, I guess that you could take the structure S(entence) to be (optionally) a "construction" whose meaning is created by negating its normal compositional semantics. Or you might say that copular sentences can be constructions with inverted meaning, and leave other sarcastic utterances in the pragmatic wastebasket. Or you could make some other division of explanatory labor. I'm not sure that this is what Moffett has in mind, but it seems that options of this kind do make it less clear when an account in terms of conversational implicature is called for.

Anyhow, I like the analogy between (the weakness of) explanations for conversational implicatures and (the weakness of) adaptationist arguments in biology. As a contribution to the discussion, I'll cite Russell Gray, Megan Heaney and Scott Fairhall's sharp but constructive critique entitled "Evolutionary Psychology and the challenge of adaptive explanation" (a chapter in a brand-new book).

This is the same Russell Gray who's been involved with the recent work on Indo-European dating. He and his co-authors are by no means gentle with the behavioral adaptationists:

... the impoverished view of evolution and psychology adopted by many Evolutionary Psychologists, and the weakness of their empirical science, is frankly rather embarrassing.

and they name names:

...our attack is confined to the specific program of Evolutionary Psychology associated with the “Santa Barbara church of psychology” ... a nativist approach to cognition that views the human mind as a collection of modules design by natural selection to solve the problems faced by our Pleistocene ancestors. This program was christened in the Adapted Mind book (Barkow et al, 1992), and proselytised to the lay public in Steven Pinker’s (1997) modestly titled book How the Mind Works. Its followers have applied EP doctrines to everything from social reasoning to preferences for green lawns and certain genres of erotic fiction...

They're specific about what they don't like:

Evolutionary Psychologists take current features of human cognition and posit that they are adaptive solutions shaped by natural selection to problems posed by life back in our Environment of Evolutionary Adaptedness (Tooby & Cosmides, 1992). This might be a good explanatory strategy if three criteria were commonly satisfied:
1. all traits were adaptations
2. the traits to be given an adaptive explanation could be easily characterized
3. plausible adaptive explanations were difficult to come by.

... The challenge of adaptive explanations is that all three of these criteria are frequently violated.

and they explain the difficulties in detail.

Even better, however, they not only debunk various instances of unwarranted speculation, they also give some examples of successful science. These include a case where a plausible adaptive hypothesis was disproved by careful research (the "promiscuous primate" theory of menstrual bleeding) and a case where an adaptive hypothesis about a behavioral trait is strongly supported by converging evidence that includes phylogenetic analysis of mtDNA data (development of wing-waving displays in Pelecaniforms from flight intention movements).

By analogy, in the area of pragmatic Just So stories, it would be nice to see a clear explanation of what hurdles such a story should have to get over in order to be accepted, along with at least one example of a plausible pragmatic explanation that was shown by careful empirical investigation to be false, contrasted with another that is strongly supported by converging evidence.

Posted by Mark Liberman at May 18, 2004 12:35 AM