Daniel Everett’s Recent Book is Mostly Not About Recursion and the Pirahã

Written by . Filed under past due reviews. Tagged , , . Bookmark the Permalink. Post a Comment. Leave a Trackback URL.

It’s odd that the release of this book served mainly as an occasion for the re-ignition of the “Universal Grammar (UG) is recursion/Pirahã has no recursion, so UG is wrong/no it isn’t/and also UG isn’t just recursion/oh and also starlings do recursion/etc.” debate, when the book itself is not centrally about either recursion or the Pirahã language. The initial press focused on this aspect of the text because it’s the most spectacularly controversial bit. I don’t want to wade into those waters, because I don’t think I can add will help clarify things, but see here, here, and here (and oh god the comments! gird yourself for the comments), oh yes, and this round-up of links and posts on the redoubtable Language Log, if you want a sense of what the fuss is about. Actually, if anyone in this discussion referred to Google’s lovely definition-by-example of the term — when you search for “recursion,” it says, “did you mean recursion?” – I missed it. So there. I have added something.

Language: The Cultural Tool turns out to be mostly about relationships between different aspects of language — vocabulary, discourse structure, unusual modes of transmission like whistling and humming — and the cultures in which they are situated. Some of what Everett says seems hard to argue with. If we take his very broad view of what the study of language should encompass, then there is a lot to say about the relationship between language and culture. There is a relatively uncontroversial version of the notion that language shapes our thinking. Call it the Orwell corollary of the Sapir-Whorf hypothesis, and I doubt we’ll find much daylight between UG theorists and functionalists: What should we suppose Noam Chomsky thinks about the name of the USA PATRIOT Act, or the Citizens United decision that money is speech? Obviously, how we use words reflects something about our thoughts and our (rapidly deteriorating) political culture.

Similarly, it is hard to argue with the notion that culture shapes language, again taking the broadest possible view of what constitutes “language.” Written discourse takes very different forms from spoken discourse. There are a lot of interesting things that happen when you start writing down a language, as Everett recounts: sentences get longer, there is less repetition of words and structures. Written and spoken language even have different standards for precision, so that, for example, legal (or legalistic) arguments are easier to make based on written than spoken agreements. Everett is coloring well outside the lines here, with respect to how boundaries are typically drawn between linguistics and adjacent disciplines like anthropology, psychology, and literary theory. By taking a holistic view of what constitutes language, Everett allows for a lot of very interesting questions that point to potential research directions, and he is usually reasonably careful to point out that the anecdotes he provides as preliminary evidence for his arguments are just that. One way to view Everett’s research program, then, is as an attempt to get away from the distinction between “core language,” i.e., grammar, and other aspects of language that are acknowledged to be “contaminated” by other forces, such as cognitive, communicative and cultural constraints.

Here, Everett is in the company of functionalists like Lakoff, Tomasello, Givòn, and many others. Some of the highlights of the book for me were sections in which he rehearsed the arguments of these other theorists, providing context for how their views diverged from more formalist approaches. These sections also take some of the urgency out of Everett’s central argument, however. Aren’t there, after all, rather a lot of linguists working on just the sorts of phenomena Everett is interested in?

What is really controversial here, then, is when Everett takes up the narrower case, that culture has an impact on grammar itself. Leaving aside the Pirahã lightning rod, he wants to claim that there are “lots of examples of culture having an influence on grammar,” and that therefore it is wrongheaded to study grammar as if it were sui generis. But he does not provide a very wide range of examples to back up this claim. Aside from the contested data from Pirahã, I found very few specific claims about culture influencing grammar in the text: a language (Amele) that has no word for “to give,” which is ascribed to the cultural centrality of giving in that culture; a language (Wari) in which the verb “to say” is omitted in many contexts because of the way indirect speech acts are used metaphorically in the language (e.g., “the sky says it is going to rain”); and a language (Banawà) in which gender agreement for first-person is always feminine, even if the person speaking is male. Everett straightforwardly admits to having no explanation for this last one, but he does provide a harrowing account of how women are sequestered just before — and ritually beaten just after — the onset of menses in this culture.

Examples like this seem to weaken the case for Everett’s proposed research program. It isn’t enough to observe that: A) this language does something really unusual with gender and B) the culture it is part of has some really unusual gender-related practices. There should be some way of systematizing those relationships across cultures. When you consider, in contrast, the wide variety of features of languages that can and do vary in ways that have no discernible relationship to culture, it’s hard to see the advantage of Everett’s approach. As it’s just come up, let’s focus a little more on grammatical gender as an example. As a native English speaker, the arbitrariness of decisions about whether, say, the moon is masculine or feminine is clear on its face. When we consider that even related languages can have different genders for the same concept (see some nice examples on Wikipedia) the arbitrariness of assigning genders to inanimate objects is clearer still.

One question we might ask, if we wanted to show an effect of culture on grammar, is whether societies in which the language has no grammatical gender are on average more (or less) egalitarian than those that in which it does? There doesn’t seem to be any evidence for this. There are some interesting experiments in which the particular gender of nouns seems to have a small impact on processing efficiency under some conditions, and that there may be some impact on semantic judgments, but these are contentious issues, and the experiments are very hard to get right and very easy to argue about. In searching for relationships between grammar and culture, Everett is looking for evidence of a very weak signal, against a background of near-total chaos. As someone who spends most of his time trying to eke out effect sizes of around 1% in a signal that fluctuates stochastically at much higher levels even when at rest, I am sympathetic, and certainly wouldn’t want to say it couldn’t be done.

But none of the evidence Everett presents on this score is really compelling. Nor does he seem to have the right tools. In fMRI research, we can pick out the tiny signals we’re looking for because we know roughly what signal ought to look like when it’s there, and we can collect tons of data that allow us to compare our results to what would be expected by chance, and we can repeat experiments multiple times. How would you establish that any relationship between how gender is used in a language, and cultural practices around gender is not just a coincidence predicted by chance, when you don’t know a priori what this relationship ought to look like, you rarely have all the data you need, and it’s hard to say what would properly count as a “replication” of your observations?

The worst case scenario here would look something like Michael Lewis running afoul of, well lots of people with his extended riff on the claim that Germans are fascinated by filth, based on the supposedly unusual productivity of words for excrement in the language. This piece was as popular as a turd in the punchbowl, revealing as it did that the author didn’t know shit from shinola about the prevalence of scatological terminology across languages, probably because he’d done crappy research. And it does not seem to be what Everett is advocating, but at times his descriptions are so loose and impressionistic that it’s hard to describe exactly how his research program would distinguish itself from this kind of anecdote collection. Here’s a frustrating example. Everett points out that “There is an obvious cultural question raised by the fact that in Wari’ wife and vagina are the same word.” Indeed! But he then goes on to punt on that question entirely, continuing:

“Why would the Wari’ make this generalization?…Some outsiders might jump to the facile conclusion that this is a crude and demeaning comparison, showing that Wari’ men hold women in low esteem. But another possible conclusion is that we cannot know what the Wari’ mean until we understand the culture that produces their meanings. Perhaps to the Wari’ reproduction and the family are such important values that they honor the wife and the vagina as the source of life. Is this a possible conclusion? Yes. Is it the right one? I don’t know. No one can know unless they undertake a systematic analysis of Wari’ culture.”

OK, but what evidence is there that a systematic analysis of Wari’ culture is going to solve this particular riddle? As John McWhorter points out in his review in the Times, “In French, the verb sortir is used to describe leaving, sticking out your tongue and being pulled out of a hole. Are the French somehow less culturally sensitive to the differences among those things…?” Unfortunately, there is not much here to demonstrate what such a systematic analysis would look like, or how one would identify candidate topics of investigation, beyond the frisson one gets from using the word “vagina” in scholarly discourse (which was apparently enough to keep Naomi Wolf going for close to four hundred pages, so, you know, there’s that).

This example is also indicative of how little of Language: The Cultural Tool is actually at odds with the the target of Everett’s argument. Chomsky’s prose style is famously dense and difficult (thus the chomskybot) but this recent article is actually pretty clear about why he thinks that communication is only a secondary function of language, and thus why he’s skeptical of claims that the core elements of language could be shaped by culture:

“One illustration is Mark Baker’s demonstration, in his book Atoms of Language (Baker 2001), that languages that appear on the surface to be about as different as can be imagined (in his case Mohawk and English) turn out to be remarkably similar when we abstract from the effects of a few choices of values for parameters within a hierarchic organization that he argues to be universal, hence the outcome of evolution of language.”

Facts about about the referential extension of the word for “vagina,” are explicitly irrelevant on this view:

“Complexity, variety, effects of historical accident, and so on, are overwhelmingly restricted to morphology and phonology, the mapping to the sensorimotor interface. That’s why these are virtually the only topics investigated in traditional linguistics, or that enter into language teaching. They are idiosyncrasies, so are noticed, and have to be learned.”

So for Universal Grammar theory, the stuff that differs between languages, and that proponents of studying linguistic diversity like Everett advocate studying, is happening at the interface between the logically elegant “core language capacity,” and the messy, kludgey cognitive and motor apparatus that translates its output into external signals. As defined in the Chomskyan paradigm, most of Everett’s book is not really about “language.”

I should say, in case it isn’t obvious, that I share a lot of Everett’s misgivings about this position, mainly for reasons that have to do with how I do my own work. One problem raised by making a bright line division between “core language” and “the mapping to the sensorimotor interface” is that it prescribes certain goals for things like speech perception and word recognition. If Chomsky is right, the most interesting questions someone like me should ask are of the form: “How do people’s brains identify the abstract symbolic structures at the core of language when they can take on an infinite variety of surface forms because of this messy interface?” There are a lot of arguments to suggest that that approach is not particularly fruitful when studying the levels of description I’m most interested in. So it makes life easier for me to just proceed as if there were no such thing as UG, and study how people learn to use words and speech sounds without assuming that their goal in doing so is to identify abstract symbolic units.

This is different, though, from the positive claim that UG doesn’t exist, which is what Everett tries to show. He could have stopped short of this claim and still produced a really fascinating book, which would have been met with less brouhaha. When he writes about different languages as great storehouses of human knowledge, when he writes (movingly) of how the decline in linguistic diversity impacts us as a species, and when he recounts anecdotes of life in cultures few of us will ever have direct contact with, Everett is a delight to read. He is also, as popular science writers go, unusually scrupulous about warning us when he is “musing” rather than arguing, and when arguing he is typically very good about acknowledging where other researchers would disagree with key premises of his arguments. If only he had argued much more effectively, or much less.

Image credit: Snapshot from a trip to work with the Pirahã, from Michael C. Frank, photo by Ted Gibson.

Post a Comment

Your email is never published nor shared. Required fields are marked *


You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>


Get every new post delivered to your Inbox

Join other followers: