Thursday, April 19, 2007

Finally: Everett on Pirahã

Over a week ago, I noted that NPR had done a piece on Dan Everett and his, ahem, striking claims about the Pirahã language. This stuff has been plowed through repeatedly on blogs and in the news, and even the new (April 16) article in the New Yorker has already been worked over pretty well. I finally got a chance to read that, if only quickly, and for those who won't read this piece and the scholarly papers the controversy has sprung from, let me make a few casual observations.

First, though, I agree with what I've read several times: The reporter, John Colapinto, did a solid job. Like a lot of the best journalistic work, he seems to have a pretty good sense of how to sketch the landscape in linguistic theory — Chomsky, Pinker, Pesetsky (who's in Madison right now), Whorf and Sapir are key players in this story. It's a shame that the Linguistics Wars motif still plays so prominently, though, since most of us have moved beyond the good versus evil view of these particular theoretical differences.

In fact, Everett's personality ends up being directly relevant: He is pictured as he's talked about in the field, as a 'bomb thrower'. The thing is, he's changed sides enough times that you may not be sure exactly who he's out to destroy at any given point: Is he a foot soldier for SIL (the missionary Summer Institute for Linguistics)? Or is he hellbent on destroying their reputation? Is he going to war for Chomsky? Or at war with Chomsky? He's been all those things and more, and a true believer on all counts, it seems. This has led some people to conclude that he's more about slinging crap than moving the field forward. (I don't know him personally, by the way, in any relevant way, and can't judge.)

But the actual papers on this topic are worth looking at — Everett's earlier one in Current Anthropology and responses there, plus now a response by Nevins, Pesetsky & Rodriguez and a counterresponse by Everett (the last two available from LingBuzz). The core principle Everett proposes is this:
IMMEDIACY OF EXPERIENCE PRINCIPLE (IEP) IN PIRAHÃ: Declarative Pirahã utterances contain only assertions related directly to the moment of speech, either experienced (i.e. seen, overheard, deduced, etc. – as per the range of Pirahã evidentials, as in Everett (1986, 289)) by the speaker or as witnessed by someone alive during the lifetime of the speaker).
In a world where most linguists don't believe that culture shapes the fundamental patterns of grammar, that's a big claim, and it makes a lot of predictions, including that Pirahã has no phrase structure and no counting. That these limits are directly cultural is crucial, since Everett's arguing that these people lack things that are widely regarded as reflecting core human capacities: It can sound alarmingly close to saying they aren't 'normal' humans, and he tries to dodge that, obviously. Peter Gordon (author of an earlier paper in Science on the lack of numbers in Pirahã) is quoted in the New Yorker on this …
if there was some kind of Appalachian inbreeding or retardation going on, you'd see it in hairlines, facial features, motor ability. It bleeds over. They [the Pirahã] don't show any of that.
Boy, am I reassured. And I know our friends and colleagues from the Southern highlands will be pleased by this scientist's characterization of them. At any rate, Nevins et al. pit Dan Everett today ("CA", for the piece in Current Anthropology) against the earlier Dan Everett ("Diss", etc.). Here's their punchline:
CA asserts ... that the embedded clauses amply documented and described in the earlier work are not actually embedded clauses, but offers no account or even acknowledgment of the numerous facts that argue in favor of the old view over the new. Similarly, CA offers as an argument for the new view the absence of long-distance wh-movement, but offers no new account of the data that in earlier work motivated the claim that Pirahã has no overt wh-movement of any kind. Likewise, as we have seen, CA asserts that Pirahã lacks quantifiers, but offers no coherent evidence against the proposal that the words described as quantifiers in the earlier work were described wrongly. …

CA simply asserts that Pirahã grammar has properties that, if true, would place it outside the pale of grammar and culture as we know it and would demand a special explanation for Pirahã's seeming uniqueness.
They argue that the structures in the language, as analyzed by Everett himself, are thoroughly consistent with what we find in other languages. Everett's response? From the abstract:
I … argue that the methods traditionally used in the Generative Grammar tradition are flawed and do not meet the scientific standards accepted across different fields in science, thereby making it difficult to assess the validity of the claims made by Nevins et al.
His actual argumentation about language structure looks muddy in many places, in part because the data are complex and unfamiliar to me. My hunch is that some of the material presented would indeed count as showing more 'grammar' than Everett is trying to posit, but I am not sure that the current state of knowledge allows any firm conclusion yet. After all, the basic analysis of a lot of, say, German grammar is still difficult and controversial, there's not exactly such a descriptive tradition for Pirahã.

Everett is definitely right that the "fieldwork in the library" school of linguistics is fatally flawed. And if you don't have firm control of your data, you are very likely to screw up the analysis: garbage in, garbage out. Most of us understand all of this, but he pretty much says that you can't draw any conclusions about a language by reading the published literature on it without doing fieldwork on the language. This gets close to undermining the purpose of publishing a lot of scholarly work, and it looks like a tool for discrediting Nevins et al.'s interpretations of his work — some of those strike me as fair and pretty uncontroversial, others are hard to judge, and he's no doubt right on some points.

But for a guy claiming to be so invested in defending "scientific method" and "standards", parts of this are too much of a rambling rant, with a section on ad hominem attacks by Nevins et al. and a big section on all the people he's gotten to do work on Pirahã language and culture. More to the point, he doesn't lay out clear, explicit arguments about lots of key claims he makes. Let me conclude with one tiny example:

Nevins et al. note that some analyze German as not allowing prenominal possessive structures to embed other possessives, unlike English (Everett, p. 9):
a. [John's car's] motor (English)
b. *[Hans-ens Auto]-s Motor (German)
He quotes an email from Manfred Krifka showing that German does have recursion here, with the familiar restriction of prenominal possessives in German to names and kinship terms. From this, he concludes that:
they get the German facts wrong. More interestingly, the German restriction might in fact be cultural, which would offer further support to the central thesis of Everett (2005b).
He repeats the first assertion a couple of times, giving the impression that Nevins et al. somehow made this up or missed obvious information. In fact, I believe that they are taking a pretty common view, even if it's not the entire story, i.e., that it's possible with certain types of nouns. [See the comment on this post by A.S.] On the second point, that the restriction to names and kinship terms is "cultural", he provides no additional account of how this follows. As I was mulling this over, I told someone about this and their response was: If the German restriction is cultural, it sounds like it means that English speakers somehow believe that inanimate objects can possess other objects. It's a smallish point about German, but highlights how cautious we should be about languages, especially when we have precious little information.

In the end, I'm pretty discouraged to see this as the latest example of linguistics in the media. Everett, in his own paper, gets too close to sounding like a preacher on a street corner, not enough like a scientist.

Update, April 23, 4:54: New piece up at on this.

(Image from here. It's the Maici River where the Pirahã live.)


A.S. said...

Whether or not Everett's claims are true, he is not doing serious researchers on linguistic relativism a favor. The main problem for me is that he simply doesn't give any evidence for a link between language and culture -- he sees (or imagines) a strange property of Pirahã grammar, he sees (or imagines) a strange property of Pirahã culture, and he posits a link. But positing is not proving, as serious relativism researchers like John Lucy or the Levinson group at the MPI in Nijmegen have argued many times, which is why they (and others) have resorted to sophisticated psycholinguistic experiments that allowed them to demonstrate relativity effects (much weaker and much more subtle than Everett's claims, incidentally).

Everett's IMMEDIACY OF EXPERIENCE PRINCIPLE is a good example of bad relativism research. The Pirahã evidentiality system seems to be an obligatory part of the grammar, so obviously every clause will be marked for evidentiality. From this he concludes that the Pirahã can only talk about things for which evidence in the sense of the evidentiality system exists. But that is akin to arguing that, say, Germans can only talk about things that have happened in the past, are happening in the present or will happen in the future, because every German clause is marked for tense. However, I have often talked about things that never happened and never will, and still I‘ve used the tense system in talking about them.

He's right about one thing though: German quite definitely has recursion with prenominal possessives: as long as you adhere to the general restrictions on these (they must be pronouns, proper names or kinship terms), you can embed as many of them as you like. Our understanding of German grammar may be limited, but not in this respect.

Mr. Verb said...

Just got an email from someone saying basically, "Surely he did not actually assert that this language has "no phase structure". On p. 4 of the current LingBuzz version, Everett writes:

"This means, in effect, that Pirahã grammar has no phrase structure. I think that this is right."

Mr. Verb said...

Thanks, A.S. I don't know anything to speak of about most of the work like that by the Levinson folks, but am not surprised at all that they are looking are far smaller things and doing it with more careful methods -- this is probably part of the irony of Everett's posturing about 'science'.

On the prenominal possessor point, these basic patterns are certainly well known, and surely to Krause and Roeper & Snyder, but the restriction gets lost in the shuffle. But I find it very hard to imagine seeing that restriction as 'culturally determined'.

A.S. said...

But I find it very hard to imagine seeing that restriction as 'culturally determined'.

Absolutely. I wouldn't even know where to look for cultural patterns that would determine this, especially since the semantically equivalent postnominal possessives show no such restrictions.

Victor Hackett said...

Forgive an amateur's observation, but Everett says that Piraha is a very HARD language to learn and it also includes a musical singing/whistling version that seems to parallel the segmental spoken language. Is there not some chance that Everett has not learned Piraha very well himself? And who is to cross-check him? Are there any bilingual and bicultural speakers who can confirm his data (with respect to the garbage in/out principle)? Perhaps his children who grew up the Piraha?

Anonymous said...

I wouldn't be surprised if Everett was right about the Pirahã but the IEP is clearly already an endangered species if not extinct. The experiments created to test the IEP have destroyed it by changing the environment to one in which the IEP is no longer sufficient - the Pirahã have deftly adapted to the conversations with scientists by creating innovations in tone use and such.

Anonymous said...

The Pirahã language is changing as they deal with ours. They seem to be reveling in it and I am, too.