The story of the strange language of the Pirahã

June 18, 2007 9:10 PM

The story of the strange language of the Pirahãis just as much a story about the state of the field of linguistics. Professor Dan Everett of Illinois State University, who lived for decades with the Pirahã, first as a missionary, then as a linguist, believes Pirahã casts serious doubt upon Chomsky's theory of universal grammar. Chomskyites have started to fight back with a reassessment of Everett's famous paper on the Pirahã, where he claimed that the Pirahã "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." He also claims that it doesn't have recursion, a feature of language Chomsky recently claimed was the defining feature of human speech. Dan Everett has rebutted the Chomskyite reassessment of his work. Video interview with Professor Everett. [Pirahã previously covered on MetaFilter in 2004 and 2006]
posted by Kattullus (60 comments total)

Via grapefruitmoon via Wired Science Blog.
posted by Kattullus at 9:11 PM on June 18, 2007

previously, by Gyan.
posted by exlotuseater at 9:37 PM on June 18, 2007

Pirahã's lanuage is very strange and interesting, but I don't see how it refutes Chomsky.

You could argue that it's more a freaky broken thing than a real language.
posted by bhnyc at 9:40 PM on June 18, 2007 [1 favorite]

Sapir-Whorf inevitably gets brought up in these discussions.
posted by oaf at 9:41 PM on June 18, 2007 [1 favorite]

You could argue that it's more a freaky broken thing than a real language.

Moving the goalposts is so like Chomsky.
posted by oaf at 9:45 PM on June 18, 2007

You could argue that it's more a freaky broken thing than a real language.

So Chomsky is irrefutable because if it doesn't fit then it's not a language. I'm no linguist but those are the kinds of rules pseudo-science makes, not true science.
posted by vacapinta at 9:46 PM on June 18, 2007 [1 favorite]

Urgh, I don't know how I managed to miss Gyan's post. This is what I get for not paying attention.
posted by Kattullus at 9:46 PM on June 18, 2007

[Pirahã previously covered on MetaFilter in 2004 and 2006]

And Wednesday.

(But your writeup was much better, so good on ya!)
posted by bpm140 at 9:51 PM on June 18, 2007

Saphir-Whorf has little empirical support.
posted by k8t at 9:55 PM on June 18, 2007

So Chomsky is irrefutable because if it doesn't fit then it's not a language.

The massive amount of baseless Chomsky worship is holding linguistics back.
posted by oaf at 9:58 PM on June 18, 2007

The massive amount of baseless Chomsky worship is holding linguistics back.


How so?
posted by hifiparasol at 10:00 PM on June 18, 2007

Match only by the massive amount of BS by his by his non-linguistic detractors.
posted by RavinDave at 10:01 PM on June 18, 2007

How so?

I've seen enough people ignore data, or find horrifically convoluted explanations for it, just to "prove" that UG is a solid theory.
posted by oaf at 10:02 PM on June 18, 2007

Match only by the massive amount of BS by his by his non-linguistic detractors.

Who?
posted by oaf at 10:03 PM on June 18, 2007

Saphir-Whorf has little empirical support.
k8t

Everett and his supporters would say the Pirahã provide a good example of the theory in some form...
posted by Sangermaine at 10:11 PM on June 18, 2007

he claimed that the Pirahã have no numbers

I read this article previously. Now, I am not a linguist. But is it so hard to believe that Everett is simply a) mistaken about a cultural difference, not a language quirk, b) self-promotional to a fault or c) has been sold a bill of goods by some Indians with an odd sense of humor? Or all of the above.

No numbers. What do they call it when you hold up one finger? Two fingers? Three fingers? They may not have a name for the abstract notion of "one, two and three," but clearly they would need to distinguish individual units from among groups of related units encountered in nature, and communicate that to others. And those distinguishing terms would be numbers.

The village that I visited with Everett was typical: seven huts made by propping palm-frond roofs on top of four sticks.

So, without a means to distinguish between four and five sticks, how do you pass along the knowledge of how to build a hut?

"Bring me sticks to build a hut."
"Here you go."
"No, that's not correct. I need enough to build a hut with."

And so, "four" is "a sufficient supply of sticks to build a hut." Shazam, that's a number.

Any real linguist want to explain this to me?
posted by Cool Papa Bell at 10:25 PM on June 18, 2007

My question, I guess, is this: Do these people not have a concept of what "each" or "every" is? Is there not an "all" to them, on any level? I can't really understand that. Would they, for example, when seeing all of their family or entire tribe together in one place, not make note of it internally?

With no deep memory and lack of the basics, I guess what I'm wondering is, in what ways this is different than dogs barking or whales communicating, as opposed to what we normally consider human speech?

Interested, now, and the googles do nothing.
posted by zerolives at 10:25 PM on June 18, 2007

Cool Papa Bell and I seem to be having the same thoughts on this. Someone set us straight.
posted by zerolives at 10:27 PM on June 18, 2007

>>The massive amount of baseless Chomsky worship is holding
>>linguistics back.
> How so?

Theoretical linguistics has been the focus of grant funding since Chomsky's role in the cognitive revolution roundabouts 1954, to the detriment of researchers still doing applied linguistics, field research. Chomskyan methods and models of syntax and phonology have been dominant since his first publications, and even his detractors tend to use introspection rather than empiricism to support claims about the validity of their models. In fact, as in many of the social sciences, much of what is considered settled theory is nothing more than hypothesis. It's a very frustrating environment for an empiricist.
posted by litfit at 10:29 PM on June 18, 2007 [3 favorites]

What color is the sky in your world?
posted by BeerFilter at 10:32 PM on June 18, 2007

Would they, for example, when seeing all of their family or entire tribe together in one place, not make note of it internally?

I see that question and raise you ... if I assembled the entire family in front of them except one person, would they not notice someone missing? And wouldn't that recognition of one person missing be an understanding that the state of "all" exists?
posted by Cool Papa Bell at 10:34 PM on June 18, 2007

And so, "four" is "a sufficient supply of sticks to build a hut." Shazam, that's a number.

Any real linguist want to explain this to me?

Any lexicon tends to capture the necessary and most-valued objects and concepts of a given language community. I'd guarantee that Pirahã speakers have an ad-hoc means of specifying mass and count, even if it's only a gesture. I wouldn't guarantee that there's a need for a formal system in the language.
posted by litfit at 10:40 PM on June 18, 2007

"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."

I'm with Cool Papa Bell on this. Language doesn't need to have numbers because Chomsky defined it so, it seems like one of the most basic requirements for human communication.

No numbers? So in other words, these people have the same mental capacity (and apparent need) to communicate as a 1 year old. I ask my daughter how many fingers I am holding up and she replies "duck." I ask her what color the cherries are and she says "duck." And then she points to the ducks and says "duck" and I say "look she's got language!"
posted by three blind mice at 10:47 PM on June 18, 2007

"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."

I'm with Cool Papa Bell on this. Language doesn't need to have numbers because Chomsky defined it so, it seems like one of the most basic requirements for human communication.

No numbers? So in other words, these people have the same mental capacity (and apparent need) to communicate as a 1 year old. I ask my daughter how many fingers I am holding up and she replies "duck." I ask her what color the cherries are and she says "duck." And then she points to the ducks and says "duck" and I say "look she's got language!"
posted by three blind mice at 10:48 PM on June 18, 2007

I ask my daughter how many fingers I am holding up and she replies "duck." I ask her what color the cherries are and she says "duck." And then she points to the ducks and says "duck" and I say "look she's got language!"

Sounds like she's overqualified for participating in political debates here.
posted by oaf at 10:50 PM on June 18, 2007

I'd guarantee that Pirahã speakers have an ad-hoc means of specifying mass and count, even if it's only a gesture.

Agreed. Which is why I'm (cynically) thinking that trumpeting "they have no numbers" is code for "someone give me a Nobel Prize, quick!"
posted by Cool Papa Bell at 10:50 PM on June 18, 2007

This will prove to be bullshit.
posted by tkchrist at 10:57 PM on June 18, 2007

I think Cool Papa Bell and some others are laboring under some misunderstandings. From what I have read, the Piraha have a "one-two-many" counting structure. There are other Amazonian and Aboriginal languages supposedly have the same counting structure and that has been reported independent from anything Everett is doing, so all these doubts about Everett or insinuations that he is making it up for personal glory or being tricked by the tribe don't make sense to me. What was supposed to be weird was that the Piraha cannot or will not learn how to count in another language, which is what separates them from the other "one-two-many" languages.

I'm not a linguist, so I don't know how true any of that is, but that is what I believe is reported.
posted by Falconetti at 10:59 PM on June 18, 2007

Quite right Cool Papa Bell. I'm not impressed by Everett's paper linked above - it seems to mostly consist of anecdotes along the lines of "heehee! check out these wacky tribespeople who couldn't learn to count!". And when he says things like:

what I say about Piraha˜ semantics is largely unreplicable unless the “replication” linguist learns to speak the language. (fn. 12)

...it makes me wonder what he thinks he's doing: surely the whole point of linguistics is to describe the language under study in such a way that other linguists *don't* have to learn the language!
posted by nomis at 11:01 PM on June 18, 2007

Also, Everett has said that the "one" and "two" of Piraha refer not to the numbers 1 and 2, but rather to "small amount" and "larger amount."
posted by Falconetti at 11:02 PM on June 18, 2007

Cool Papa Bell: I've just finished reading the thing and they sort of brushed on this. In essence, if you hold up one finger, they will say one word. If you hold up two, it's the same word with different intonation. However, the first word for 'one' will sometimes be used for two. So it's sort of like a word for 'very few', and the next one is for 'a little more than very few'. They did not go into this but I imagine it may depend on context, like if you're talking about seeds which are small in themselves, they may use a word for 'one' for a few of them, but if you're talking about a large airplane, they'd use the word that they normally use for a few to describe just one of them. In regard to using 4 sticks to build a hut - they probably don't have to be very exact, as five would work too, and you can just keep adding sticks until they hold up.

They ran an experiment where the natives could not reliably match a bunch of stones if the bunch had more than 3 of them. If a number of stones were put in a box and take out one by one they could only predict the box to be empty when original amount was not more than 3. There are other languages where you only have words for one, two and many, but the difference is that only in this tribe natives were either unable or unwilling to learn counting in other languages. But they don't appear to be lacking in intelligence. It was interesting that the article mentioned that they are more efficient in the forest than all other indians in that area.
posted by rainy at 11:04 PM on June 18, 2007 [2 favorites]

Falconetti: From what I have read, the Piraha have a "one-two-many" counting structure.

See Everett's paper linked above:

Piraha is the only language known without number, numerals, or a concept of counting.
posted by nomis at 11:05 PM on June 18, 2007

nomis: in my opinion linguistics' goal is to learn the way languages work, rather than describe how they work. If you can learn the working of a language but are unable to describe it, that's still valuable to you. If we suppose for a second that there is a language, semantics of which cannot be described in another language, that would teach us something about linguistics. Unless he's lying. I guess what I'm trying to say is that if you only limit yourself to learning things that can be described, you might miss something interesting. I don't know..
posted by rainy at 11:14 PM on June 18, 2007

I'm not a linguist, so I don't know how true any of that is, but that is what I believe is reported.

One of the implications is that it's Sapir-Whorf revisited. People tend to have strong opinions about relativism, but in linguistics and cognitive psychology it's more or less a settled matter. The other is the theory of universal grammar. Linguists don't really talk about "deep structure" anymore (even Chomsky has done a lot of backpedaling). The implication here is that if a language group demonstrably lacks a grammatical feature that was considered universal to human language, it somehow undermines the notion that language is innate to humans in some predictable way, or that features of language are instinctual.

Whatever. Everett is doing good work. It's journalists and colleagues creating the tempest in a teapot. I've read his work, it's solid. I'd like to study with him.
posted by litfit at 11:16 PM on June 18, 2007 [1 favorite]

I'm having a hard time imagining this. So the difference between, say, two legs, and zero legs, isn't a very big deal to them, linguistically?

Steve in that hut has "very few" legs?
posted by zerolives at 11:21 PM on June 18, 2007

nomis: I guess "counting structure" is the wrong phrase, I just made it up. But the "one-two-many" structure is also in his work and it helps assuage some of the difficulty people were having in how a language without the sort of counting we are used to could function, which was my point in mentioning it.
posted by Falconetti at 11:24 PM on June 18, 2007

Everett on "all"

Although there is no word “all” in Pirahã, it could be countered that perhaps it is the construction itself that produces the universal quantifier reading. Superficially this is appealing, but I think that it is another manifestation of the translation fallacy. Even though there is a certain “quantificational smell” here, the truth conditions are not the same for generics as for quantificational readings (see, e.g., Krifka et al. 1995). In fact, I and others who have visited the Piraha˜ have misunderstood statements like these and/or their literal translations because we do translate them into Western languages as generic, universal quantification. These never mean that all beings with blood, for example, fail to inspire fear. That there are always exceptions is understood by the utterer and the hearer. It seems, though, that such sets conform to the postulate of cultural constraint on grammar and living because they are bounded by immediate experience (e.g., “evil spirits I know about”) and thus are not fully intensional. Rather, each member of the set has to be inspected to see whether it is an evil spirit or being with blood and, if so, whether it is like other such beings.

The paper is fairly long, but also fascinating and easy to read (IIRC, I read it a year or two ago).
posted by Kattullus at 11:32 PM on June 18, 2007

One of the implications is that it's Sapir-Whorf revisited. People tend to have strong opinions about relativism, but in linguistics and cognitive psychology it's more or less a settled matter.
litfit

I know you didn't volunteer yourself as a guest lecturer, but would you mind expanding on this a bit? I'm no linguist, and from what I gather (and from some comments in this thread) people seem dismissive of Sapir-Whorf. Why? The Wikipedia article's criticism section wasn't very helpful in explaining why that is. A curious mind thanks you in advance.
posted by Sangermaine at 11:38 PM on June 18, 2007

rainy: I agree that the individual linguist's goal is to learn the way languages work, but the goal of linguistics as an academic discipline is to have those findings communicated to other linguists.

Learning a language is not the same as doing linguistics: if you learn a language and understand how to use its structures, but are unable to describe those structures in terms that linguists can understand, then you're not doing linguistics.
posted by nomis at 11:56 PM on June 18, 2007

One of the funniest things about this discussion is the way it has Chomskian linguists scrabbling. They have, in effect, come out witht the claim that recursion is what really makes us human. Now we find a language without recursion, and they are in a very uncomfortable position. Their theory is dear to them, but its a little unseemly to be caught in a position where some perfectly good humans suddenly appear less than human. Did anyone link to Pinker's response?
posted by fcummins at 11:57 PM on June 18, 2007

Here are the many things the experts at Language Log have had to say about this stuff, over several years.
posted by LobsterMitten at 12:00 AM on June 19, 2007

Falconetti: my understanding of Everett's claim is that the Pirahã don't even have terms for 'one', 'two' and 'many' in the numerical sense. So the word(s) used to distinguish 'one fish' from 'two fish' is also used to distinguish 'one small fish' from 'one big fish'.
posted by nomis at 12:00 AM on June 19, 2007

Good link LobsterMitten, and this Language Log article discusses the point I was trying to make just then about numbers.
posted by nomis at 12:03 AM on June 19, 2007

And their response to the most recent flare-up.
posted by LobsterMitten at 12:06 AM on June 19, 2007

Er, that is, I agree with nomis.
posted by LobsterMitten at 12:07 AM on June 19, 2007

ohh we studied them in high school sociology at the beginning of the year!
I remember we were examining different cultures and we read about the Piraha and their lack of words for numbers, colors, etc, and how this was because of their ethnocentric belief that they were better than the rest of the people around them and thus didn't mix in with the Brazilians or whatever.
posted by alona at 3:36 AM on June 19, 2007

I like Everett's response to the latest Language Log post (LobsterMitten, I think you linked an older post):

I would simply like to emphasize that the popular media's reporting on this controversy has been extremely hit or miss. The New Yorker did a reasonably good job, but I am really weary of pieces like the Chicago Tribune's that puts this all in personal terms. It is bad for me, bad for the field, and potentially bad for the Pirahas, because it makes them look somewhat freakish. I wrote a paper and published it in a peer-reviewed journal. Looking past all of the publicity and vitriol, there is really only one appropriate way to deal with these claims: design experiments, go to the field, and test them.
posted by mediareport at 5:05 AM on June 19, 2007

Someone else needs to go and study the Pirahã thoroughly. At the moment Everett appears to be the only person who actually knows the language and the people in sufficient depth. In trying to refute him, his opponents have to cite his own research, which is a pretty desperate position (and I think it shows in some cases, notably that dismal effort by Pinker).
posted by Phanx at 5:35 AM on June 19, 2007

I'm having a hard time imagining this. So the difference between, say, two legs, and zero legs, isn't a very big deal to them, linguistically?

Steve in that hut has "very few" legs?

You might want to read this. 0 as a number is a relatively new invention.

Also you need to remember that numbers and quantity are not the same thing. Sure, you can represent quantity with numbers, but the concept of "lots" vs "less" is much more basic than actual numbers and the implication of maths that goes along with it.
posted by public at 6:08 AM on June 19, 2007

The story of the strange language of the Greeks:

In his writings Homer surprises us by his use of color. His color descriptive palate was limited to metallic colors, black, white, yellowish green and purplish red, and those colors he often used oddly, leaving us with some questions as to his actual ability to see colors properly (1). He calls the sky "bronze" and the sea and sheep as the color of wine, he applies the adjective chloros (meaning green with our understanding) to honey, and a nightingale (2). Chloros is not the only color that Homer uses in this unusual way. He also uses kyanos oddly, "Hector was dragged, his kyanos hair was falling about him" (3). Here it would seem, to our understanding, that Hector's hair was blue as we associate the term kyanos with the semi-precious stone lapis lazuli, in our thinking kyanos means cyan.

posted by asok at 6:35 AM on June 19, 2007

At first glance, I thought this said "The story of the strange language of the Piranha", and I was very worried.
posted by kafziel at 6:46 AM on June 19, 2007

asok,
That article was really interesting, especially this part:
"It also seems likely that Ancient Greek perception of color was influenced by the qualities that they associated with colors, for instance the different temperaments being associated with colors probably affected the way they applied color descriptions to things. They didn't simply see color as a surface, they saw it as a spirited thing and the word to describe it was often fittingly applied as an adjective meaning something related to the color itself but different from the simplicity of a refined color. "

That sounds a lot like what Everett describes the Pirahã's non-color "color" system as, referring to the attributes of a substance instead of the surface appearance we think of as color. Maybe this is something all languages go through at some point, but the Pirahã just had no reason to develop the abstract concept of color as independent of any one object?

Also, if language really can cause people's brains to limit their color perception, how do new colors (or at least the concepts and words for new colors) arise?
posted by Sangermaine at 7:34 AM on June 19, 2007

Moving the goalposts is so like Chomsky.

Heh. Well said.

Thanks, litfit, for providing the linguistic explanations; I'm glad I don't have to do all the heavy lifting around here.

It always amazes me to see how people who have never given this stuff a moment's thought can click on a MeFi post, skim an article or two, and confidently decide that someone like Everett, a professional linguist who's devoted decades of his life to this, must be full of shit. You rock, armchair generals of science!
posted by languagehat at 7:47 AM on June 19, 2007 [3 favorites]

from litfit: Linguists don't really talk about "deep structure" anymore (even Chomsky has done a lot of backpedaling).

They (or we, I suppose I should say) still have something closely akin to deep structure, though - LF in syntax, and underlying representations in phonology.

The implication here is that if a language group demonstrably lacks a grammatical feature that was considered universal to human language, it somehow undermines the notion that language is innate to humans in some predictable way, or that features of language are instinctual.

I can't tell if you support this idea or not, litfit. The 'somehow' seems to indicate you don't agree. In any case, it seems to me that, if a language truly lacks a putative universal, it just means that they hypothesis about that particular structure's universality is wrong. This clearly has important implications for people that think that recursion is the sum total of the human linguistic endowment, but it doesn't destroy nativist (or more generally, generativist) linguistics by any stretch of the imagination.

Whatever. Everett is doing good work.

On some level, this may be true, but he also says some seriously goofy things with regard to his work. For example, on p. 29 of his reply to the Nevins, et al., critique of his work [pdf], he claims that seeking generalizations is inappopriate in a 'pragmatic linguistics', yet he makes ample use of generalizations in his own research.

It's journalists and colleagues creating the tempest in a teapot.

Not completely. Everett is prone to overstate the importance of his findings. For example, he likes to claim that his findings (i.e., his reanalysis of his own data) signal the end of generative linguistics.
posted by noahpoah at 11:09 AM on June 19, 2007

"So in other words, these people have the same mental capacity (and apparent need) to communicate as a 1 year old. I ask my daughter how many fingers I am holding up and she replies "duck." I ask her what color the cherries are and she says "duck." And then she points to the ducks and says "duck" and I say "look she's got language!""

I think you are coming at this from an entrenched cultural position (to put it kindly). You can't imagine not needing or using numbers - and you use that as a meter of your daughter's intellectual progress. To say that because your daughter doesn't yet understand numbers, and the Pirahã don't use them then they are on the same level of intellectual development is bad logic. It also highlights the difficulty of understanding a truly new concept.
posted by The Light Fantastic at 11:55 AM on June 19, 2007

1) A lot of the criticism of Everett posted here re numbers and such is dealt with in the last link (to an Edge interview that includes an exchange with Singer).
Aside from linguistics it is interesting to hear Everett speak about his loss of faith. The Piraha may or may not be culturally superior but they won this guy as a religious convert.
2) Everett's work does not infer Whorf-Sapir, it inverts it. Whorf-Sapir says language shapes culture, Everett says culture shapes language. (None of these three would be so doctrinaire as to say that either language or culture is always dominant.)
posted by CCBC at 2:51 PM on June 19, 2007 [1 favorite]

"Whorf-Sapir says language shapes culture, Everett says culture shapes language. (None of these three would be so doctrinaire as to say that either language or culture is always dominant.)"

It's a fascinating argument, but I would think it would be very difficult to prove conclusively - especially when taking into account that very few languages exist in a vacuum of cultural isolation. Most common languages on the planet have evolved through a process of cross-pollination.
posted by The Light Fantastic at 3:31 PM on June 19, 2007

They are both wrong. Next.
posted by MapGuy at 3:34 PM on June 19, 2007

Light Fantastic: I would think it would be very difficult to prove conclusively

You can prove either proposition in small ways. Whorf noticed that fuel drums designated as "Empty" were handled carelessly by workers even though the drums held fuel vapor that was potentially explosive -- something the workers knew from their training but disregarded because of the language used to describe the drums. And here we have Everett showing that culture can bar liguistic elements not related to direct experience. So, both notions can be demonstrated and probably both operate constantly as language(s) develop(s). But, either way, there may or may not be a Chomskyian Universal Grammar.
posted by CCBC at 4:49 PM on June 19, 2007

I'm not sure how I feel about Everett's claims, but I like the way he's gone about making them. He's putting his data online, he's encouraging other researchers to learn Pirahã and do fieldwork in Brazil — he's trying to draw attention to an interesting phenomenon, not to himself or his theories. That's awesome.

(And even if he turns out to be wrong, it suggests that he believes what he's saying. Frauds and bullshit artists don't like having their data scrutinized.)
posted by nebulawindphone at 10:43 AM on June 20, 2007