February 21, 2010
Pirahã undermines Noam Chomsky's idea of a universal grammar
John Colapinto in The New Yorker:
Everett, who this past fall became the chairman of the Department of Languages, Literature, and Cultures at Illinois State University, has been publishing academic books and papers on the Pirahã (pronounced pee-da-HAN) for more than twenty-five years. But his work remained relatively obscure until early in 2005, when he posted on his Web site an article titled “Cultural Constraints on Grammar and Cognition in Pirahã,” which was published that fall in the journal Cultural Anthropology. The article described the extreme simplicity of the tribe’s living conditions and culture. The Pirahã, Everett wrote, have no numbers, no fixed color terms, no perfect tense, no deep memory, no tradition of art or drawing, and no words for “all,” “each,” “every,” “most,” or “few”—terms of quantification believed by some linguists to be among the common building blocks of human cognition. Everett’s most explosive claim, however, was that Pirahã displays no evidence of recursion, a linguistic operation that consists of inserting one phrase inside another of the same type, as when a speaker combines discrete thoughts (“the man is walking down the street,” “the man is wearing a top hat”) into a single sentence (“The man who is wearing a top hat is walking down the street”). Noam Chomsky, the influential linguistic theorist, has recently revised his theory of universal grammar, arguing that recursion is the cornerstone of all languages, and is possible because of a uniquely human cognitive ability.
Steven Pinker, the Harvard cognitive scientist, calls Everett’s paper “a bomb thrown into the party.” For months, it was the subject of passionate debate on social-science blogs and Listservs. Everett, once a devotee of Chomskyan linguistics, insists not only that Pirahã is a “severe counterexample” to the theory of universal grammar but also that it is not an isolated case.
More here. [Thanks to Lee Kottner.]
Posted by Abbas Raza at 01:44 PM | Permalink
Comments
Nice article.
Posted by: icastico | Feb 21, 2010 2:34:55 PM
The big headline is that Chomsky is wrong. But the Pinker comment
in this article suggests we take a deep breath before making that claim
Posted by: FRED LAPIDES | Feb 21, 2010 6:16:44 PM
Yes. A long pause. They supposedly don't have recursion in their spoken language. I really don't know if that means the theory of universal grammar and the larger biolinguistic approach is "wrong."
Posted by: Jonathan | Feb 21, 2010 6:25:26 PM
I don't think this article demonstrates anything conclusive. It's unclear, for example, that Everett himself actually secured a demonstrable mastery of the language of the Piraha. At best, one might conclude that this one particular language, which superficially seems to defy a feature intrinsic to every other natural language, namely, recursion, leaves unanswered important questions about the nature of language. It does not, however, definitively refute the idea of UG.
If UG is false, and language is instead shaped largely, if not entirely, by culture, then we have necessarily to explain how it is that the overwhelming majority of natural languages--all of them, apparently, but this one rare, isolated, and apparently poorly understood example--include the feature of recursion. The very scarcity of counterexamples seems to me to weigh very strongly in Chomsky’s favour.
I’m not saying this as a Chomsky devotee. I think there’s some merit to the idea that Chomsky has become a bit of an obdurate conservative who unnecessarily reacts to frank criticisms of his scientific work with reactionary arrogance and contempt. And, I think Pinker and Paul Bloom did us all a great service by effectively challenging Chomsky's inexplicable hostility to the idea of a Darwinian account of the origins of the language faculty.
Chomsky’s attitude notwithstanding, I don’t see any bombshell. This article begins with much hype and ends with much inconclusiveness. I guess that’s the most you can expect with a journalistic piece covering a complex debate in a complex discipline like linguistics.
Posted by: Belgian Beer | Feb 21, 2010 6:42:08 PM
I have been troubled by this fascinating article for a coupla years. The jury isn't in? I keep wondering: if the Chomsky theory is right, then it must not be language these folks are speaking. If the theory requires modification, on the other hand, then the members of this civilization conduct their affairs in language the same way they count without conceptualizing numbers higher than 5. Everybody in this article except the soon-to-be-ex-wife of Everett, seems to me to be pretty condescending about the Piraha, who at least understand each other.
Posted by: Elatia Harris | Feb 21, 2010 6:51:07 PM
Elatia, why do you think Everett is condescending to the Pirahã?
Posted by: Vicki Baker | Feb 21, 2010 7:22:22 PM
Vicki, he quite clearly likes them and cares about them, but it seemed to me when I originally read it that he had not learned their language well enough to theorize like that, while his soon-to-be ex-wife, who had a missionary interest in saving their souls, had a more highly developed ability to communicate with them. Maybe I'm just seeing an O. Henry ending where there is none?
Posted by: Elatia Harris | Feb 21, 2010 7:34:31 PM
I agree with everything Belgian says up the the last two paragraphs. That is, I think the argument that the Paratha don't have a recursive syntax is not made by this article and does not conclusively show that Chomsky is "wrong," if by that one means to say that language is learned and not grown, that culture determines the rules of grammar not an innate faculty, and so on. I don't see how Chomsky is inflexible or conservative about any of this. I do think, however, that few outside of linguistics and cognitive psychology know that much about what his theory entails. So he's easy to take pot-shots at. Bloom and Pinker were simply lucky to come around at the beginning of the evolutionary psychology bandwagon. Even so, their criticism of Chomsky is far from Everett's. They're just as innatist and universalist as Chomsky. The disagreement turns on whether language was a target of selection (was "selected for") or whether it was an excaption of other innate faculties, like mathematical computation. Chomsky's position is hardly outlandish. It just isn't amenable to the adaptationist hoo-ha that has swept the academy over the past fifteen years (with folks like Pinker leading the parade).
Posted by: Jonathan | Feb 21, 2010 11:06:55 PM
Jonathan,
Just to clarify: I didn’t state or imply that Pinker and Bloom work within a theoretical paradigm different from the one Chomsky established. Pinker happily works within a Chomskyan tradition. That much is obvious and I never suggested anything to the contrary. However, Pinker has very legitimate disagreements with Chomsky, disagreements *internal* to that paradigm. One can affirm a biological faculty for language, while also affirming a Darwinian account of the origins of that faculty.
Chomsky has never made a convincing argument that language is a spandrel. In fact, he acknowledges that his view is a minority one, even among those who work within his paradigm. He seems content declaring language a spandrel and then declaring that there’s no good reason to believe otherwise. Moreover, he gets downright impatient and contemptuous when anyone even tries to provide a counterargument, even when it’s made by a sincere and intelligent colleague. Pinker and Bloom have provided very good reasons to believe that language is an adaptation designed to facilitate communication. Moreover, Pinker has called Chomsky on his arrogant and contemptuous attitude towards those who disagree with him, and rightly so.
I’m somewhat sympathetic to the idea neo-Darwinists are rigid and uncompromising. However, in this case, I think it’s the other way around.
I’m happy to hear a contrary opinion.
Posted by: Belgian Beer | Feb 22, 2010 12:29:37 AM
I read this article some time ago and it was engrossing. The main point of the article is that if Everett's observations are correct then Chomsky's universal grammar theory is "wrong". However, the crux is the validity of Everett's observations. The article is more of a story about his studies it seemed than a statement about linguistics.
Posted by: Alex@TheAstronomist | Feb 22, 2010 1:53:15 AM
A few months after this well written New Yorker article, a video statement by Everett appeared on Edge where he nicely frames the issue for general audiences, followed by an exchange between several researchers. I just read these and found them quite instructive.
More formal publications appeared last year in response to Everett's 2005 paper in the following order: first, second, third. To my non-specialist eye, the debate seems to be far from over.
Posted by: Namit | Feb 22, 2010 2:03:42 AM
Belgian,
I don't think we're disagreeing about that much. I do find Chomsky's resistance to adaptationist accounts of language to be persuasive, at least in two senses: first that it is very hard to make a knock down argument in support of the selection of higher cognitive faculties without it seeming circular (e.g. language was selected for communication because communication was adaptive) and second, that, given both what we know about the evolutionary time line and about the uniform features of language, its plausible it was excapted from computational faculties selected for other purposes. You might be right that the former position is just asserted, but the latter (as I'm sure you know) has been developed over several papers with Fitch and Hauser, which have been responded to in turn by Pinker and Jackendoff.
I see Chomsky in this respect, along with the even more controversial Jerry Fodor, as the last in the line of Gouldians. I'm not so sure that the Gould/Lewontin spandrel/punctual equilibrium model really lost out to the Pinker/Dawkins/Wilson pan-adaptationist model for the right reasons. It seems as much to do with which model caught on with media hype as what provided the best explanatory theory of the data.
Posted by: Jonathan | Feb 22, 2010 5:54:26 AM
Folks,
There are so many new papers on Piraha that I cannot list them all here. But you might want to check out my website, where I post them as fast as I write them. You might also be interested in seeing the 'trailer' from the new documentary in progress about Piraha: http://www.vimeo.com/7877204
Dan
Posted by: Dan Everett | Feb 22, 2010 7:33:21 AM
To spare others a google, here's Dan Everett's homepage:
http://llc.illinoisstate.edu/dlevere/
Thanks for stopping by, Professor Everett. Looking forward to your new book!
It's interesting to me how people will take seriously an "evolutionary psychologist" who's spent literally weeks analyzing the results of a questionnaire but are ready to doubt that a linguist who's spent 7 aggregate years in the field with a language knows his stuff.
Posted by: Vicki Baker | Feb 22, 2010 10:09:07 AM
Not that it's necessarily the same people in the cases I mentioned above, of course.
Posted by: Vicki Baker | Feb 22, 2010 10:09:54 AM
I don't think anyone takes Joseph Carroll seriously. I know I don't. Plus, as the above conversation indicates, Chomsky is not an evolutionary psychologist, quite the contrary. Having said all that, yes, very cool that Dan Everett stopped by this lowly thread. Who knows who lurks out there ...
Posted by: Jonathan | Feb 22, 2010 11:25:31 AM
Yes, real science can read like a thriller with red herrings, false leads, conflicts, mystery, conspiracies, and egos. Real science can be messy and combative.
It requires the stamina of a long distance runner, the tracking skills of a Native American hunter, the intellectual discipline of a Talmudic scholar, and the endurance of an arctic explorer through months of privations and darkness.
It's hard work over a long period of time - even a life time.
Posted by: Norman Costa | Feb 22, 2010 12:11:44 PM
Jonathan,
Maybe we’re not disagreeing at all.
I think the debate over the origins of the language faculty hinge on what, exactly, is meant by “communication,” a point correctly raised by Fitch, Hauser, and Chomsky in their 2005 response to Pinker and Jackendoff. To be fair, I don’t think PJ do a very good job of making sense of communication as a natural phenomenon, whether in the animal world or in the human world. They have lots to say about syntax and the use of syntax, but not specifically linguistic communication, which entails more than the use of mere syntax. They don’t theorize very well about pragmatics. Pinker describes the function of communication as “expressive,” which I think is somewhat ambiguous. Had he used the term “inferential,” I think he would have been on the right track.
I agree that PJ haven’t offered a knock-down argument. However, they need simply to supplement their argument with a more substantial definition of “communication”. Pinker and Bloom were correct to focus on the expression of propositional statements, but propositional statements cannot themselves be understood or expressed apart from a larger, implicit background system of inferential relations. That background system is necessarily social. Communication, in this sense, can be understood as a social-inferential process, and I think language arose to facilitate that process. Chomsky insists that a computational or problem-solving faculty arose first to facilitate private, individual thought, and that language developed as a lucky and convenient consequence, which we today exploit for the purpose of communication. He therefore resists the idea that “the essence of language is communication”.
There is a way out, one that involves redefining communication as a social-inferential process—a process that need not be fundamentally different from the computational, problem-solving process Chomsky originally had in mind. I think the work of people like Ruth Millikan and Robert Brandom might be the key to settling the debate between FHC and PJ. There’s a paper waiting to be written on this.
But, this is all going way off topic, which is Dan Everett’s work on the Piraha. I’d be curious to know from him whether the language of the Piraha contains the use of conditionals, one of the cornerstones of inferential articulation. I’d be surprised if they had no “If…then…” in their language.
Posted by: Belgian Beer | Feb 22, 2010 12:37:39 PM
This great article reminded me of the experiences of a developmental psychologist, Joseph Glick, who studied cross-cultural aspects of intelligence. He was investigating the 'universality' of Jean Piaget's stages of cognitive development. His research focused on children in North America, Europe, and Liberia.
According to Piaget, the child develops grouping concepts along functional lines, before the ability to develop classifications along abstract lines. Putting a knife with an orange is organizing along functional concepts. Putting table utensils together, and fruit in a separate group would represent organizing by abstract concepts. Children would sort a collection of utensils and fruit into functional categories at an earlier age than sorting by abstract category, which comes at a later age. The universality theory says that all children are functional before they are abstract; and all children pass through these stages of cognitive development at about the same age.
Well, Glick found that Piaget's theories of universality held for North American and European children. However, Liberian children were still doing functional categorizing when their age mates in North America and Europe were categorizing by abstract concepts. So, he and his colleagues discussed the results with the European missionary running the Liberian school for children. The missionary said, “Of course.” The Liberian children were not as smart, not as intelligent, as their white age mates.
Glick and his crew didn't like the attitude of the Missionary so they decided to change the design of the experiments to see if they could find conceptual categorization in Liberian children at the same age as white children. Earlier, they asked the children to put things that belonged together from a collection of utensils and fruit. The Liberian children paired a knife with an orange – functional grouping. This time, however, they asked the children, what would a wise man do, if asked to put things together. They said a wise man would pair a knife with an orange. Then they asked, what would a stupid man do, if asked to do the same thing. They said a stupid man would put all the utensils together and all the fruit together.
There were two lessons to be learned from this study. First, it appeared that Liberian children were developing the capacity for abstract concepts at the same time as their white counterparts. The second lesson is a more subtle and profound issue for scientific researchers: When do you stop asking questions? When do you stop collecting data?
Posted by: Norman Costa | Feb 22, 2010 12:44:57 PM
Jonathan - I didn't mean to imply that Chomsky (or anyone serious about language and evolution) is aligned with Carroll. I'm just very curious why you would be doubtful of Everett's command of the Pirahã language. What leads you to infer this?
Posted by: Vicki Baker | Feb 22, 2010 12:46:58 PM
I don't think anyone takes Joseph Carroll seriously.
And yet he holds the highest ranking professorship at Mizzou. Giving Carroll tenure in an English Dept. is like putting a young earth creationist in charge of the AMNH. But it happened; and now his students will take him seriously as well.
I confess to not understanding Chomsky's logic about recursion being the sine qua non of human language, and I'm inclined to say he's overreaching. We could easily enforce a no-recursion rule experimentally (say, by restricting communication to writing, which was vetted by a third party). The result would be something that no bird, primate, or cetacean could ever aspire to do. (Well, maybe the jury is still out on cetaceans).
Recursion gives language a great robustness and flexibility, but this is not the same as saying (human) language can't exist without it. The Greeks didn't know any algebra, but they were still pretty good at math.
A much better candidate for what makes language language is symbolic thought. The "language" of birds, insects, etc., is a manipulation of signals, not symbols. (With primates it's less clear.) As much good as Chomsky has done to clear behaviorist thinking from the playing field, he has led psycho-linguistics seriously out to sea with his emphasis on structure divorced from meaning. (The fact that tamarin monkeys can't recognize "mistakes" in meaningless recursive sentences has little to do with whether they would understand the sentences [had they not been gibberish] whether or not they employed recursion.)
Belgian,
In your view, is the computational problem-solving model something we have good reason to articulate? It seems to me to rest upon a lot of dicey presumptions about the nature of communication, and it seems especially vulnerable to the risk of defining humanity down by means of a Turing Test.
Posted by: Chris Schoen | Feb 22, 2010 2:20:08 PM
Hi Chris,
I’m not saying that communication ought to be reduced to a mere computational process; I’m saying that it ought to be understood as a fundamentally inferential process, one that need not be totally and completely different from the original problem-solving faculty that Chomsky had in mind.
The motivating idea behind an inferentialist model of communication is rationality linguistically conceived; rationality as a responsiveness to reasons – reasons being propositional statements that can serve as either the premise or conclusion of an inference. I don’t see why there should be any problem in defining rationality—our distinctive feature—in such a way that it could conceivably be reproduced in the form of artificial intelligence. Brandom has addressed this point in his recent Locke lectures at Oxford:
* http://www.philosophy.ox.ac.uk/__data/assets/pdf_file/0013/904/LL3_Text.pdf
I agree with you that syntax alone cannot account for meaning. I don’t think that Chomsky offers a semantic theory, at least not a convincing one. I think the most fruitful semantic theory is the inferentialist model developed by Wilfred Sellars and Robert Brandom. Jaroslav Peregrin has an essay comparing Chomsky and Brandom on semantics:
* http://www.ingentaconnect.com/content/jbp/pc/2005/00000013/00000001/art00004
Curious to hear your thoughts.
Posted by: Belgian Beer | Feb 22, 2010 7:13:47 PM
This is three years old. Why is 3qd referencing it now?
Posted by: Chandan | Feb 22, 2010 11:32:32 PM
I didn't notice that either sending it on, but I think that it might stand in contrast to the posted upcoming review of A. Gopnik's book in TNYROB:
"Even infants are sensitive to statistical patterns. The learning of language in its earliest stages involves the statistical prediction of which sounds are most likely to follow one another—an unconscious exercise in probability theory. Gopnik argues that this ability to detect probability patterns extends beyond language—to musical tones in eight-month-olds, for instance—and isn't limited to a specialized part of the brain as Noam Chomsky and others believe."
Posted by: Whoa | Feb 23, 2010 2:06:22 AM
Belgian,
OK, I've stayed up too late reading those lectures, with the following results:
1. I like Brandom's beard very much.
2. My understanding of inferentialism is still very shallow.
In the context of AI (or "synthetic sapience"), I think it's interesting that a theory like Brandom's based on the study of normative rules will have difficulty accounting for the variation, novelty and ambiguity that we know is integral to language, and which are, after all, the features that make AI so challenging. But perhaps he addresses this issue elsewhere.
Let me respond instead to your comment that:
The problem--or one problem--as I see it, would be that in defining rationality in such a way it ceases to be "our distinctive feature." I suppose it hinges on what you mean by "conceivably," but since in actuality AI reduces at present to basically being really good at chess, we don't gain much by talking about what it "conceivably" can do. Compared to the real thing the answer is still: not much.
Susanne Langer makes the observation that people commonly confuse the general term "reason" with the specific form "discursive reason," and I think you so do here. There are also non-discursive logical forms (think of the arts) and the use of and engagement with these forms is every bit a part of what "makes us human" as language. It is not completely clear that thinking is even a separate activity than feeling (Langer's late work was mostly dedicated to rehabilitating the concept of "feeling" as a more apt foundation of Mind, biologically, than "thought.")
I'll say this tentatively, but Brandom's inferentialism, by ascribing meaning to "rules," seems to elide the role of personal interest. Why did whoever made the rules make them? Why do we sometimes abide them and sometimes bridle at them? I think we are still too close here to a computational hermeneutics where cognition is essentially a dispassionate intellectual activity, when in fact it almost never is. But I haven't thought very long and hard about it, and the best I can give you is this rather flip response.
Posted by: Chris Schoen | Feb 23, 2010 2:34:50 AM
What is new since then ?
Posted by: Anil Thakuria MD | Feb 24, 2010 12:36:35 PM
To what extend Chomsky's theory of UG is accurate or useful is debatable, but there is definitely a universal process: Children input a linguistic environment and from that learn a language. Apparently the input can be such that the language learned is not recursive. But it seems to me that this process of acquisition is what linguists would do best to focus on, and universal or near universal features of syntax should be taken as hints as to what that process is.
Posted by: Justin | Mar 5, 2010 10:30:35 PM
At last, fresh findings about the Pirahã, this time a phonetic analysis of experimental data. Yes, they have embedded clauses after all.
http://ling.auf.net/lingBuzz/001095
Posted by: Luc | Sep 8, 2010 11:21:34 PM
Post a comment