Several people have sent me links to Tom Bartlett’s Chronicle Review piece on Daniel Everett’s attempt to demolish Noam Chomsky’s hegemonic linguistic theory and the messy academic battle that has ensued. I wrote about Everett here, and anyone who’s been reading LH for a while will know that I root for anyone going up against Chomsky and his minions, but I must confess that my acquaintance with the current state of linguistics is so scanty that I do not have an informed opinion on the details of the argument. I have seen it said that Everett is attacking a long-abandoned form of the theory, that nobody any longer believes what Chomsky used to say about recursion, etc. But I will quote a section that illustrates why I despise what Chomsky has done to a once collegial field:

Critics haven’t just accused Everett of inaccurate analysis. He’s the sole authority on a language that he says changes everything. If he wanted to, they suggest, he could lie about his findings without getting caught. Some were willing to declare him essentially a fraud. That’s what one of the authors of the 2009 paper, Andrew Nevins, now at University College London, seems to believe. When I requested an interview with Nevins, his reply read, “I may be being glib, but it seems you’ve already analyzed this kind of case!” Below his message was a link to an article I had written about a Dutch social psychologist who had admitted to fabricating results, including creating data from studies that were never conducted. In another e-mail, after declining to expand on his apparent accusation, Nevins wrote that the “world does not need another article about Dan Everett.”
In 2007, Everett heard reports of a letter signed by Cilene Rodrigues, who is Brazilian, and who co-wrote the paper with Pesetsky and Nevins, that accuses him of racism. According to Everett, he got a call from a source informing him that Rodrigues, an honorary research fellow at University College London, had sent a letter to the organization in Brazil that grants permission for researchers to visit indigenous groups like the Pirahã. He then discovered that the organization, called FUNAI, the National Indian Foundation, would no longer grant him permission to visit the Pirahã, whom he had known for most of his adult life and who remain the focus of his research.
He still hasn’t been able to return.

Chomsky has remained magisterially in the background and refused to comment, but his minions are behaving in a way more appropriate to a down-and-dirty political campaign than to an academic disagreement. In a sense, the facts of the language are irrelevant; the way the dispute is carried on speaks volumes.
Update. See now Geoff Pullum’s excellent summary of the case.


  1. Anonymiss says:

    Say what you will, but it’s not as if Chomsky is calling Andrew & Cilene up to tell them what to say (they are a married couple, by the way). I read in another article that FUNAI requested the letter and I’ve heard it said (from someone in the other group who went to visit them) that the Pirahã really did not want Everett back, but I don’t know how much faith to put in either of these statements.

  2. marie-lucie says:

    This is nasty. The Pirahã and other indigenous groups have to walk a fine line about outsiders wanting to study their language and/or customs. What choice to they have if they are told that Everett is no good (for whatever reason) and they should instead welcome Dr. XYZ who will be so much better for them?

  3. My current favorite thing about Pirahã is how much it has become like Sumerian, a poorly described and poorly understood ink-blot onto which people ignorant of the core of the Universal Grammar debate, or linguistics in general, can project all manner of theorizing.

  4. There’s a Brasilian reporter who got confirmation from FUNAI that Everett had done his research in Piraha territory without legal authorisation,. The article is linked from the Chronicle of Higher Education comments page. Someone else there points out that this removes the need to look for a Chomsky conspiracy as explanation for his banning.

  5. marie-lucie says:

    Everett first went to Pirahã territory as a missionary. He spent years there with his family, learning the language and losing in faith. He went back as a linguistic researcher, but was not an outsider about to begin fieldwork. Perhaps he did not apply for legal authorization because he had already lived among the people and did not think that he needed to apply for such authorization. According to the article, another team of linguists wanted to check Everett’s data and only got permission for a week’s work – a ridiculously small amount of time for any linguist to get even a passing acquaintance with the language.

  6. The one thing I don’t get is why journalists – and occasionally his critics, especially from the right – go out of their way to compar Chomsky’s political activism with his work in linguistics, as if the two were the same thing or at least two sides of the same coin. And even if they were, the example Bartlett gives perfectly illustrates what’s wrong with Chomsky – if Chomsky kicked Buckley’s ass, it’s because he could argue facts. This is not the case with Chomsky’s linguistics.
    Recently there’s been a rise in so-called corpus linguistics
    Nevermind the ‘so-called’ bit, which is bad enough, it’s the way he ties corpus linguistics into brain imaging that’s the real story: That [corpus linguistics], along with the brain-scanning technology that linguists are increasingly making use of, may be able to help resolve questions about how much of the structure of language is innate[…]. Is that lazy writing or is Bartlett really that clueless about what corpus linguistics does?

  7. Thesauro says:

    Bulbul, don’t cherrypick your accusations of cluelessness against Bartlett, though. The whole article’s like that, including what he says about Chomskyan linguistics.

  8. Thesauro,
    please provide some evidence for your claim.

  9. Asya Pereltsvaig at Languages of the World has a couple of posts critical of the Chronicle article here and here, while Lane Greene wrote a short review of Everett’s book in the Economist.

  10. Be sure to read the comments, they’re a goldmine – I particularly like the suspiciously well coordinated response of the UG folks. The exchange between Asya Pereltsvaig and tyroneslothrop* was especially enlightening. This comment by Asya is telling in more way than one. The way she doubles down a little later is even more so:

    I am not saying that “exotic” languages contribute nothing to our understanding of human language in general, only that there is no a priori reason that the interesting things must necessarily be discovered in such exotic languages.

    That’s your problem, right there: ‘a priori’. You won’t even bother to check, all the while it just so happens that it’s languages from non-air-conditioned parts of the world that exhibit features which you guys until recently argued were not possible (e.g.).
    And when she’s called on it:

    I don’t say that doing fieldwork in an exotic locale isn’t challenging. But I don’t see how challeges of such practical, logistical sort can add *scientific value* to such work.

    So suddenly it’s not about the languages involved and their features, it’s about the practicalities of fieldwork.
    I’d love to hear what the folks who do field linguistics (Claire, Lameen, m-l et al.) have to say about that.
    * Whoever you are, you’re my hero. Or at least I owe you a drink.

  11. Say what you will, but it’s not as if Chomsky is calling Andrew & Cilene up to tell them what to say
    Heh. Yes, I was using “minions” as a rhetorical jab; after fifty years of Chomskyan dominance, he doesn’t have to tell anyone what to say. They know.
    The one thing I don’t get is why journalists – and occasionally his critics, especially from the right – go out of their way to compare Chomsky’s political activism with his work in linguistics, as if the two were the same thing or at least two sides of the same coin.
    Very true.
    Asya Pereltsvaig at Languages of the World has a couple of posts critical of the Chronicle article
    Yes, I was sorry to see those; she’s circling the wagons against those who dare criticize the Truth, and descending to such deplorable rhetorical tricks as “There is a strong assumption in both articles that conducting actual fieldwork and getting data from some obscure language spoken by a small group in an area without air conditioning or running water is somehow more valuable for the linguistic enterprise than working with a language spoken by millions in a more civilized part of the world. By why would data from Pirahã be more valuable in principle than data from, say, Russian? Are Russian minds less human?” There are all sorts of nasty insinuations going on there; I’ll just mention that the point (as Pereltsvaig knows perfectly well) is not that an “obscure language” is more valuable than “a language spoken by millions” because it’s “spoken by a small group in an area without air conditioning or running water” but because by definition we already know pretty much everything about the major languages that have been studied for centuries and on which the soi-disant universals are based, and we need to check those “universals” against as much data as we can find, which will necessarily come from smaller, less-studied languages. And yes, if some miserable obscure language spoken by people without air conditioning or running water has some feature that destroys an alleged universal, then it trumps all the famous well-known languages lazy linguists prefer to study, since (according to them) it doesn’t matter anyway, because all languages are the same under the skin. Which is precisely the point at issue.

  12. Thesauro says:

    Bulbul asks me to substantiate my claim that the article gets Chomskyan linguistics wrong.
    The article repeats the Everettism that Chomsky claimed that “recursion was the most important feature of langage”, that this claim appears in the 2002 Science paper, that if Piraha doesn”t make use of recursion Chomsky is disproved, and that to say otherwise is to “back away” from this supposed claim of Chomsky’s. Not a piece of this is true. If there’s any “well coordinated” campaign going on, it’s coming from the Everetts and the Evans & Levinson types, who can’t win against their actual linguist opponents, so they make up fictional versions of them to strike blows against instead.

  13. marie-lucie says:

    There are also interesting readers’ comments below the Bartlett article, including those by Asya Pereltsvaig who repeats or anticipates the ones in his two responses, and by David Pesetsky who is one of the co-authors with Nevins of an article critical of Everett.
    I am not personally familiar with Everett’s work, so I can’t comment on it, but Pereltsvaig and Pesetsky are Chomskyan linguists, and to my mind, some of their claims or objections arise from their faith in the Chomskyan model and are not necessarily justified.
    I just want to comment on one particular aspect. Some readers take Chomsky and his disciples to task for neglecting little-known, exotic languages. The reply is that after Ken Hale (a genius linguist and wonderful all-around person by all accounts) was hired at MIT (perhaps because of such criticisms by other linguists), students and scholars there started to do fieldwork and write about such languages. True, but the “theoretically-informed” results often consist in looking for (and of course finding) “Universal Grammar” features in those languages, to the detriment of other potentially interesting features which are ignored by UG (which is mostly concerned with syntax). There is also the problem that such research involves conducting fieldwork through the medium of another language (Portuguese in the case of Pirahã): the form of the sentences of that language often influences what the speaker will produce as a translation, so that sentences “elicited” by a linguist may not be representative of what native speakers actually produce spontaneously, and can even be quite incorrect from the point of view of how the language normally works. Here a person such as Everett, whose first interest in learning the language was communication rather than theoretical research, is more likely to have learned the language as actually used rather than as deliberately produced in response to a linguist’s “elicitation techniques”, especially if those are designed to “push the envelope” in order to see how far the language can be “pushed”.

  14. Thesauro,
    thank you.
    Everettism that Chomsky claimed that “recursion was the most important feature of langage”, that this claim appears in the 2002 Science paper
    Just to be clear: the article does not suggest that those words appeared in the 2002 Science paper, it how Bartless chooses to summarize the paper’s conclusions. And he prefaces them with “a paper, published in 2002, that said (or seemed to say)”. As for the claim itself, let me quote the version on Chomsky’s site (p. 1571 of the printed edition):

    All approaches agree that a core property of FLN [faculty of language in the narrow sense] is recursion.

    Can you please explain to me how Bartlett’s summary of this is, to use your words, not true?

  15. J.W. Brewer says:

    Yes, but in this case it’s not Piraha v. Russian, it’s apparently Piraha v. English, Russian, Basque, Ainu, plus every exotic tongue spoken by a small non-air-conditioned community in New Guinea, the Kalahari, the Caucasus, etc. So if there’s a theory that appears to account satisfactorily (which I am NOT other than purely arguendo assuming that anything Chomsky-derived in fact does) for 5999 out of 6000 (or whatever numbers you like) attested human languages but doesn’t account for a single outlier, one wants to make extra sure that the data from the outlier is rock-solid and not being misunderstood. If the outlier is a language that’s only really been studied by one guy from the outside scholarly world, that’s a big big problem even without any need to speculate about intentional falsification. (If that sole data collector is also the scholar theorizing about the data’s revolutionary implications, one should be even more worried.) The solution is presumably to encourage others to study Piraha, NOT to ban the first guy from returning to the field, of course. It’s like the recent controversy over faster-than-light neutrinos – either our entire discipline’s model needs to be seriously tweaked or you guys screwed up your experiment and ended up with contaminated/misinterpreted data – aggressively pursuing the latter hypothesis is not necessarily some sort of horrible conspiracy being run out of the Vatican by albino assassins.
    I remember seemingly endless discussions in a windowless seminar room circa ’86 as to whether (the details are getting a little fuzzy here . . .) the way ergativity (?) worked in Dyirbal (?) falsified some fashionable Chomskyan hypothesis of the day, and even at the time I was thinking, you know, we’re pretty much all at the mercy here of what Dixon (if I’ve got the language right) is claiming about what Dyirbal actually does, so he damn well better be right.

  16. No dog in this hunt, but I will point out that when there are two well-defined camps in opposition, the arguments of each side, which are based on a shared intellectual/theoretical foundation, can look like a “well-coordinated attack” to the other side, because each side is blind to their opponent’s common theoretical foundation, which their own theoretical stance causes them to dismiss as nonsense.
    In other words, the ideological uniformity of a group’s arguments actually springs from the fact that they share similar views, not because they’re coordinating their arguments, but to someone who doesn’t share those views, the only possible explanation for the uniformity is some kind of suspicious coordination going on in the background.

  17. One question I have is the problem of falsifiability. Let’s assume that someone completes a Universal Grammar that accounts for absolutely every human language in a consistent way.
    Languages are dying all the time. How can the UG camp ever prove that their system accounts for all *possible* human languages? Won’t there always be a lingering doubt that maybe there did exist a language that had [UG-demolishing feature], but it died before it could be studied?

  18. Won’t there always be a lingering doubt that maybe there did exist a language that had [UG-demolishing feature], but it died before it could be studied?
    One day we might have the technology to resuscitate it, though, so not all hope is spent. A zombie language would be the speculative equivalent, for linguists, of a brain in a vat.
    Has anyone noticed the heads in vats often featured in Futurama, Nixon among them ? I like the way Matt Groening thinks.

  19. marie-lucie says:

    I started to write my comment after reading Stan’s earlier post.
    I totally agree with bulbul and LH.
    Thesauro, like Pereltsvaig and Pesetsky, apparently thinks that only Chomskyans are “actual linguists”. I wonder what that makes me!

  20. Bulbul quotes from the Science paper that “a core property of FLN [faculty of language in the narrow sense] is recursion” and asks what’s not true in Bartlett’s summary.  Happy to do that, but with some friendly frustration, since this has been explained in countless responses to Everett’s contentions already. So here’s one mo’ time.
    In his discussion of recursion, Chomsky is talking not about human behavior but about a human CAPACITY.  That’s what the word “faculty” in “faculty of language” means.  Let’s imagine Everett is right about Piraha and that Piraha speakers don’t use that capacity for some reason.  They still have the capacity, which is why they can and do learn Portuguese if they’re brought up in town.
    What is special about this capacity?  In your quote, Chomsky calls it “a core property of FLN”.  That’s pretty different from “the single most important feature of human language” (quote from the Chronicle article) and wildly different from “all languages are based on recursion” (Everett’s new book, p. 281). The distortion is pretty shocking.
    Chomsky’s actual proposal was the idea the capacity for human language is a bunch of evolutionarily pre-existing cognitive systems that we share with other primates, plus a capacity for recursive phrase structure that is uniquely human, so FLN is nothing but that capacity. That’s pretty out-on-a-limb, and probably wrong.  But nothing in this idea predicts the impossibility of even Everett’s version of Piraha, and nothing there implies that recursion is more “important” than any of the other pieces of human language.
    Have I helped? I’m not sure I can keep up this pace of replying, though.

  21. m-l,
    Thesauro, like Pereltsvaig and Pesetsky, apparently thinks that only Chomskyans are “actual linguists”.
    To be fair, I think Thesauro was merely making a contrast between Everett arguing with people vs. arguing with straw men.

  22. marie-lucie says:

    If the outlier is a language that’s only really been studied by one guy from the outside scholarly world, that’s a big big problem
    I have no idea whether Everett is correct or not in his description or conclusions, but it is not as if he was untrained in linguistics. Not all linguists (even some former Chomsky disciples) subscribe to the Chomskyan framework). According to his Wiki page, he was first trained by SIL (an organization which is very efficient at providing practical linguistics training for missionaries) and did an MA and PhD at Brazilian institutions (of course, viewed from the heights of MIT, these must be way beyond the pale of the scholarly world).
    There are many examples of excellent older grammars of “exotic” languages produced by people who were natural linguists although trained only in Latinate grammars (such as many Spanish missionaries in Latin America in past centuries) or aware of the fact that each language had its own logic (such as Boas or Sapir in more recent times, but before the advent of modern linguistics). And of course, the classical example of generative grammar (the theory started by Chomsky) is that of Panini for Sanskrit, hundreds of years ago.

  23. michael farris says:

    I’m too busy to spend too much time here but boy, does this bring up mixed emotions.
    On the one hand I have some skepticism about Everett’s work, including, but not limited to
    – his apparent misunderstanding of Hockett’s design feature “interchangeability”(sp?) in one article it seems he confuses it with some idea of ‘translatability’.
    – his “killing the jaguar” “text” doesn’t seem to be a text at all. I’m not sure what it is, but the idea that that could be a text in Piraha culture is weirder than all his linguistic stuff put together
    – some of his claims seem under-supported and more a question of terminology than anything else.
    There are cases that (to me) look like they could be called subordination but he goes out of his way to argue they aren’t without explaining why. Also, lack of subordination doesn’t seem like a big deal. Once while talking with my former boss (a Finno-Ugric specialist) about weird Hungarian subordination said boss let drop with “(ugric language I’ve heard of but can’t remember which one) doesn’t use subordination, just coordination”. Allowances have to be made for US/European terminological differences but no subordination doesn’t strike me as that unusual.
    But his fiercest opponents don’t seem any more credible to me with their “Piraha is just another boring langauge” shtick .
    I also know how frustrating it can be for someone with first-hand experience trying to explain real differences to those who don’t believe in the concept. I often have the feeling Everett is trying to explain qualitative differences through quantitative means (I know that’s not exactly what’s going on – it’s a metaphor).
    Finally (for this post) part of the reason that I’m Poland and not South America now is not wanting to have to deal with organizations like FUNAI (and the third hand experience I had with linguistic fieldwork there where as a colleague put it “everybody hates everybody”).
    The constant tension and sabotage between missionaries and secular linguists and nationals and foreigners, not to mention between different schools of linguists and the unenviable position of the natives caught in the crossfire made almost anything else seem appealing in comparison.

  24. BTW, these are the slides accompanying Gibson’s talk at 2012 LSA that Barlett refers to (“A computational analysis of Pirahã grammar”). All of it is pretty interesting, but check out slides 2-7.

  25. J.W. Brewer says:

    m-l,I certainly didn’t mean to suggest that Everett was incompetent or untrained. My point is only that even well-trained and subjectively sincere researchers can and do get things wrong, often badly wrong (without being either grossly incompetent or intentionally dishonest), so it’s always good when multiple independent researchers come up with approximately the same data for the same phenomenon.
    Fwiw, I found Thesauro’s last point about the possibility of a language that for whatever contingent reason happens not to employ the full range of available cognitive capacity to be helpful.
    As to the falsifiability point . . . a lot of interesting work in typology etc. deals with statistical tendencies not absolute existence/non-existence. So on the one hand it’s a great anecdote that one day in the ’70’s Geoff Pullum was passing on in a lecture the then-conventional wisdom that there were no default OVS languages when Desmond Derbyshire politely raised his hand and said “well, the language I’ve doing fieldwork on for the last decade [how close do the Hixkaryana speakers live to the Piraha speakers? my sense of internal Amazon geography is minimal] is actually the counterexample you didn’t think existed.” On the other hand it’s interesting and perhaps meaningful that OVS languages, while (we now know) attested, seem exceedingly rare.

  26. how close do the Hixkaryana speakers live to the Piraha speakers?
    At least 700 km apart, maybe more.

  27. David Eddyshaw says:

    From the article: “Two linguists, Nicholas Evans and Stephen Levinson, published a paper in 2009 titled “The Myth of Language Universals”
    This paper is easily found in pdf via Google and does (I reckon) exactly what it says on the tin. Well worth reading.
    Nicholas Evans has himself described two Australian languages, Bininj Gun-Wok and (especially) the deeply weird Kayardild, which are pretty good ammunition for demolishing Chomskyite overstatements about the alleged deep similarity of all languages. By the way, according to the paper, recursion, though possible, is in fact rare in polysythetic languages like BGW, and is actually limited in Kayardild to just one round of embedding.

  28. J.W. Brewer: Don’t know if this counts, OVS is pretty common in spoken Japanese.
    Eigo shabereru yo, boku. (“I can speak English.)
    Aitsu shiran yo, ore. (“I don’t know him.”)
    Gakkou made no michi nara shittemasu yo, watashi. (“I know the way to the school.”)
    Dare ni nani o agetara ii ka tte iu setsumei o shite kureta hito no namae gurai wa wakatteru hazu darou, kimi! (“You ought to at least know the name of the person who explained to you what to give to who.”) (This one thrown in just to show that OVS structure can be found in long sentences, too.)
    etc. etc. etc. And the subject isn’t optional (demonstrated by the fact that Japanese speakers use it). Granted, the subject canonically comes at the beginning of the sentence in these cases, but the fact remains that Japanese speakers choose this OVS structure quite commonly.

  29. J.W. Brewer says:

    marc, from GKP’s account: “A short report appeared in the Sunday Times on the fact that an OVS language had been discovered (we still thought it was the first one known to linguists, though as it turned out we were wrong), and the next week, to my great irritation, a letter appeared from someone who said that this was nothing new, it occurred in German. The point was apparently not an easy one to get across to the public at large: it was not about what was permitted, but about what was typical. German has some stylistic inversion possibilities, e.g. in relative clauses; but no one could think it was normal in German to tell your sweetheart “Dich liebe ich” rather than “Ich liebe dich.” In Hixkaryána, OVS was the basic, ordinary order of phrases, not just a sometimes permissible stylistic variation. So Des and I both learned a little about the difficulty of presenting linguistics to the public at large.” Somewhat less colorfully but more precisely, he phrases his prior claim that Derbyshire subsequently rebutted as “There were no languages, I had decided, in which fixed constituent order rather than case was the key indicator of grammatical relations in simple clauses, and the normal order, in contexts not conditioned by special discourse factors, was OSV or OVS.”

  30. @marc: A kind of focus/topic inversion, is what I think. Like the previous sentence (instead of “I think it’s a kind of focus/topic inversion”). “Nani miterunda, kimi wa” puts a kind of emphasis in kimi that’s absent from “kimi wa nani miterunda”. Subject-last in Japanese has this discursive value precisely because it goes contrary to the expected syntactical order. In Portuguese we can use almost any word order to various effects, but it’s my understanding that generativists are talking about the unmarked order, the one that’s perceived as neutral (in our case plain old SVO). Thus, Brewer pointed that until Hixkaryana they believed there was no default OVS language.

  31. marie-lucie says:

    JWB, sorry for the misunderstanding, thank you for the explanation. I basically agree with you, although I thought your initial wording was a little ambiguous.
    marc, I don’t know much Japanese, but even though your sentence types might be common, they are not necessarily basic or unmarked, as shown by the comma you use to isolate and thereby emphasize the last word (presumably the subject).
    The issue of basic word order in a language (or as a universal) is tricky. A full sentence is assumed to be composed of two nouns and a transitive verb, but these are not the only possibilities for a simple sentence (intransitive verb, verb with adverbial complements, copula with noun or adjective, etc). Even using the same order, let’s say the typically European SVO, the pragmatic or discourse roles of the two nouns may not follow the same order (eg which one provies new or background information – an important point to keep in mind when attempting translation of connected texts). This means that the linguist beginning the study of a language where SVO exists but is highly marked might take it for the basic order, rather than a marked one, especially when relying on an intermediate language like English or Portuguese where SVO is basic and native consultants do their best to follow the lead of the linguist.

  32. Has anybody read Dixon’s “Basic Linguistic Theory”? He esentially argues against the idea of default word order or at the very least against a rigid classification by word order. The thing is, for some languages, it is quite difficult to determine what exactly is typical or unmarked.
    What Marc describes really does seem like topicalization by left dislocation which is common enough. There is usually something else going on in those cases, like a particular intonational pattern.

  33. That would be a nasty trick indeed, if that’s what happened – and would need to be exposed. As for the linguistic controversy, though: until a much larger corpus is published, there’s no point in discussing Piraha. With the limited data available, I don’t see how any of the people involved expect to convince anyone who doesn’t already want to believe. Merely asserting that Piraha has no recursion (and that the examples that look recursive aren’t) certainly doesn’t do the trick; neither does assuming that a recursive analysis is correct there because it works in other languages.

  34. marie-lucie says:

    Word order continued:
    There is also the fact that O and S can be represented by pronouns instead of nouns, and in many languages the word order for pronouns is not the same as that for nouns. For instance, French, Spanish, Italian usually have SVO if S and O are nouns, but SOV if S and O are pronouns, an order which reflects the older Latin basic word order applying to both nouns and pronouns. Historically then, for these languages SOV came first! So word order is variable, not only synchronically for stylistic purposes within a language, but also during the life of a language.

  35. Thesauro is absolutely right that Chomsky’s proposal is about a universal human capacity, rather than about universals which are necessarily instantiated in each language. But in the particular case of Piraha, the evidence that recursion is absent seems to me to be pretty weak.
    A lot of the controversy is about the right way to understand the structure of sentences that one might be tempted to give translations like ‘Hoagaixoxai said she is not giving birth’–that is, sentences that appear to involve clausal embedding, and hence recursion. The Gibson slides that bulbul cited above just assume that Everett is right in his most recent way of representing these sentences, as pairs of independent clauses (so in this particular case, “Hoagixoxai spoke. She is not giving birth”). This is one heck of an assumption, especially since Everett has to be right about this if he’s going to be able to claim that Piraha lacks recursion. So it would be nice to see evidence.
    The paper by Nevins et al (which I think has been mentioned above) is one place to look for a critical review of Everett’s evidence for this claim:
    And Everett has a reply here:
    There has been further back-and-forth on the LingBuzz site, which I won’t bother to link here.
    (just in case my biases aren’t clear from the above, I’m a Chomskyan, and a fieldworker–one of many such people, as I and others noted in the comments to the Chronicle piece).

  36. ” The point was apparently not an easy one to get across to the public at large: it was not about what was permitted, but about what was typical. German has some stylistic inversion possibilities, e.g. in relative clauses; but no one could think it was normal in German to tell your sweetheart “Dich liebe ich” rather than “Ich liebe dich.”
    Sorry, I think that’s a cop-out. Sentences like “den Typ kenn ich”,”Gottschalk mag ich nicht” “Tee trink ich gerne” are perfectly normal and common in spoken German as stand alone utterances. Yes, they are “marked” as topic fronting, but I really wonder to what extent the existence of “marking” and “neutrality” is a creation of grammarians rather than the way speakers think. Certainly in other languages with “free” word order and topic fronting, like Russian, OVS sounds much more awkward (e.g. not “того парня знаю я” but “того парня я знаю”) and does not occur as regularly in speech. So while German is not primarily OVS, the fact that OVS occurs fairly often and easily in spoken German seems like something one should not dismiss out of hand.

  37. Bathrobe says:

    The same goes for English. “This I know” is perfectly natural, more natural in some contexts than “I know this”. And yet that doesn’t seem to be totally good grounds for chucking SVO out. Essentially SVO is a kind of starting point (a kind of canonical norm) for describing variation.

  38. marie-lucie says:

    The ‘basic’ word order of a language is a neutral one without overtones of emotion or insistence. But identifying the basic word order in a strange language is not always obvious, since non-basic orders may be very common.
    “You are here” – “Here you are!”
    The first is neutral and informative on a map of the shopping mall, or said by a helpful person orienting you on your map of a strange city. The second is not meant to inform you of your own whereabouts right next to the speaker but is an emotional statement, the sigh of relief of a person who has been waiting for you for longer than expected and perhaps worrying about you.

  39. It can also be used when giving something to someone, in which case it can’t be put in ‘neutral’ word order at all!

  40. marie-lucie says:

    Yes, because then the meaning is totally linked to the situational context of interaction between two persons in specific roles, not to the literal meaning of the words as in the neutral order.

  41. I see. Thanks for the explanation. These forms are definitely “expressive,” for lack of a better word.

  42. michael farris says:

    I thought most linguists didn’t use the old SVO/SOV typology much anymore.
    I find the general distinction between head final and head initial (each of which has a set of characteristics that usually co-occur) to be more useful even while recognizing that some languages (like Mandarain or Hungarian) mix head initial and head final structures.

  43. I thought most linguists didn’t use the old SVO/SOV typology much anymore.
    That would make sense. Any system that simply classifies English, Italian, and German as equally “SVO” (as I remember from my undergrad days) would seem to lack a lot explanatory power.

  44. German is SOV underlyingly, with V2 in finite clauses.

  45. marie-lucie says:

    Yes, SOV like basic Latin. Opinions seem to be divided about PIE word order, but SOV seems to have been at least very prominent even though not the only ordering possibility.
    SVO, SOV, etc are useful abbreviations in certain contexts, but since few languages rigidly follow one type or the other, using “basic word order” as a major diagnostic tool for typological classification is an oversimplification. English seems to be mostly SVO, more so than its neighbours: French or Spanish (among others) use VSO much more than does English, and Spanish in turn uses inversion more than does French.

  46. J.W. Brewer says:

    The possible limited utility of basic word order as a typological tool (or perhaps accepting that it’s only meaningful for a particular subset of languages – e.g. word order can be less significant in a language that uses morphological inflection or something similar like the Japanese postposed particles to unambiguously distinguish grammatical role) need not detract from the observation that it is striking that O-initial word order is so rare as to have been thought non-existent by reputable scholars within the quite recent past and that it provides a good example of the dangers that would have accompanied any attempt to theorize from apparent non-existence to impossibility-in-principle (since such a theory would have been highly vulnerable to a counterexample turning up in the Amazon). FWIW, on Pullum’s account, when Derbyshire was doing his fieldwork (and his translation of the New Testament into Hixkaryana) he had no idea that the word order was completely unattested in the modern typological literature, so he didn’t know how exciting a discovery he’d made. This is a distinction from Everett’s situation (which, again, is not to accuse Everett of bad faith or incompetence, just to observe that subjective awareness of how high the stakes are can sometimes be a suboptimal circumstance for data collection/interpretation).

  47. Bathrobe says:

    Since the comment function at The Invention of Diagramming seems to be broken, I’ll mention it here: Did diagramming lead to phrase structure grammar, thence to our beloved transformational grammar?

  48. OK it looks like the discussion of Grammar Diagramming moved here? Nothing is possible anymore without invoking Chomsky!
    I wondered if the Diagrams are perceived as un-British (since we studied Queen’s English, and used copious amounts of parsing). And – if there are manuals for applying the art to foreign languages, especially Russian? We spend inordinate amount of time trying to teach American kids about the proper gender matching between nouns and verbs or adjectives, and perhaps, just perhaps, a diagram might bring some clarity into the way these things are connected? It’s most definitely not taught to the (less confused) native speakers.

  49. Strangely, just last night via Wikipedia I found this academic article on Interlinear Morphemic Glossing. Perhaps a combination of the two would help…

  50. I have no idea how I managed to close off comments when posting, but I’ve reopened them now, so if y’all could repeat your comments there, that would be great (and keep the new post from feeling lonely and unwanted).

  51. marie-lucie says:

    Update on Everett v. Chomsky: Today Mark Liberman posted what seems to be a comprehensive list of articles and Language Log posts about the topic. David Pesetsky does not seem to be amused.

  52. marie-lucie says:

    and more: In the same post, Mark Liberman links to an article by Geoff Pullum in Lingua Franca which is short and to the point. In turn, Pullum links from this to a previous post of his on Language Log, which is also well worth reading about the sources of suspicion against Everett.

  53. I’m afraid I’m temperamentally inclined to side with LH. The UG people aren’t interested in language; they’re interested in highly abstract modelling that happens to take language as its subject matter. There is nothing wrong with a search for universals and the deeper truths of the human mind, but for this approach to marginalise all other approaches to language is not only ridiculous, it’s also pretty sterile.

  54. And one more, with even more comments by Everett and Pesetsky.
    I just realized I spent the entire week reading on this debate, and on the “ideograph debate” (about Chinese writing; Hansen vs. Unger etc). I think I probably enjoy this kind of telenovelas a bit too much.

  55. Ah, an âme-soeur here! I spent my formative years compulsively browsing the Wikipedia talk pages of Gdanzig, Lw*v, Szczettin and every other imaginable locus of contentiousness.

  56. marie-lucie says:

    Thanks for the link, leoboiko. The comments are by principal actors in the debate, they are civil, and fairly long but not too numerous nor too technical.

  57. I wouldn’t call Peter Ludlow particularly civil:

    It is hard to avoid thinking that the media rush to lionize Everett is really just a way to try and cut Chomsky down to size. I seriously doubt that anyone would bother hyping Everett if Chomsky was not the target here. It is also hard to avoid thinking that this hyping of Everett is of a piece with the climate change deniers and evolution deniers. It is the hyping of bad science out of political motives.

    But yes, the rest of the discussion is pretty civil, and quite interesting.

  58. marie-lucie says:

    You are right on both counts. I forgot PL’s contribution by the time I got to the end of the thread.

  59. Responding innocently to Peter Ludlow (this is one aspect of linguistics I haven’t boned up on): has UG born any fruit in the way historical linguistics (reconstructed languages) and biology (vaccines) have? If not, it’s hard to lump the UG skeptics in with the plain ol’ stupid skeptics.

  60. Not as far as I’m concerned, but I’ve been out of the loop for decades. Norvin up there would disagree.

  61. jamessal says:

    Thesauro is absolutely right that Chomsky’s proposal is about a universal human capacity, rather than about universals which are necessarily instantiated in each language.
    That seems like much more of a philosophical than a linguistic statement to these lay eyes. Shouldn’t Chomsky or his followers be fighting with Peter Millican rather than Geoff Pullum?

  62. jamessal says:
  63. J.W. Brewer says:

    Yesterday on the train home from work I browsed the Evans & Levinson paper commended by David Eddyshaw way upthread, and thought it was fairly interesting although the back and forth with critics provided some ways to think about the plausibility of various possible intermediate positions.
    Unpacking a box of very miscellaneous books last week after moving offices, I was somewhat surprised to come across the texts I was required to buy for an intro syntax class I took in ’85 (the textbooks themselves were copyrighted ’79 and ’81). I expect the state-of-the-art Chomskyan/UG/etc. position has probably shifted sufficiently over the intervening decades that dipping back into them would not necessarily be a productive way to understand one of the factions here.

  64. marie-lucie says:

    jamessal: has UG born any fruit in the way historical linguistics (reconstructed languages) …
    Chomsky is totally uninterested in the aspects of language which tie it to actual use, such as sociolinguistics and historical linguistics. The object of his research is an abstract view of language and languages, devoid of the richness of idiosyncratic details which make each language what it is, and independent of actual use.
    For instance, one of his first studies concerned the formation of yes-no questions (answerable by yes or no only) from basic sentences: eg Will you have coffee? from the basic (though pragmatically inappropriate) You will have coffee, through movement of the auxiliary will to the left of the subject you. But the formal study of question-formation is quite removed from the study of the reason for using a special syntactic form for questions in the first place. For instance, it is possible to question someone without using a formal question, as in You will have coffee, I suppose? Chomsky is not concerned with the circumstances under which formal or implied questions are appropriate, only with the syntactic form and the change in word order.
    Comparative-historical linguistis is my own true love and specialty in linguistics: you have to know not only how languages are put together (sounds, words, etc) and how they are likely to change (based on what changes have been observed in long-attested languages and extrapolating from there) but also something of how they are (or were) used, by whom and under what circumstances. In turn, each language carries within itself features which link it to its past, and therefore to the lives of the people that used to speak it, including social mores, contacts with other peoples, and major upheavals such as revolution or conquest. It also carries features linking it with others which have evolved from a common ancestor. Present-day languages are the visible “tip of the iceberg” represented by their long history, and they continue to evolve along with the lives of their speakers. Of course, in most cases this “tip” is the only thing that is known about a language (eg Pirahã), but treating it as the only thing worth studying leaves aside a major portion of that constitutes a language. Detailed study of the language itself will eventually give at least hints of its history and its relatives in the same language family.
    “Reconstructing” past stages of a language or the common ancestor of a language family is not at all easy: after more than two hundred years of research by dozens of scholars, most of them speaking or at least reading several languages of the Indo-European family, the study of Proto-Indo-European still has its share of unanswered questions. I personally am wary of declarations such as “Proto-XYZ had the following consonant inventory”, for little-known language families where only a few scholars (sometimes even just one) have worked on the problem (and sometimes the “XYZ” group turns out to be invalid as well). This is not a problem that UG is concerned with at all.

  65. marie-lucie says:

    jamessal I don’t think I actually answered your question, but with your mention of historical linguistics I could not resist putting in my own two cents.
    JWB: I expect the state-of-the-art Chomskyan/UG/etc. position has probably shifted sufficiently over the intervening decades that dipping back into them would not necessarily be a productive way to understand one of the factions here.

  66. jamessal says:

    Thanks for that, M.L.
    an abstract view of language and languages
    That doesn’t even sound like linguistics to me; that sounds like language philosophy, like Wittgenstein did (though doubtless the Chomskyans have access to better data), and Peter Millican does now. Then again, as you and Hat keep graciously pointing out, you’re both neither au currant nor disinterested. So could some Chomskyans come forth and explain the fruits of their labor, please?
    you have to know not only how languages are put together (sounds, words, etc) and how they are likely to change (based on what changes have been observed in long-attested languages and extrapolating from there)
    As I just commented in the other post, Charles Barber in his The English Language: A Historical Introduction gives a great sense of this. I’m loving the book, though I very much doubt it has anything to teach you.

  67. marie-lucie says:

    Thanks, jamessal, but there is always more to learn! I have several books on the history of English, and I am pretty sure this one is among them (I haven’t looked at them in a while and things are rather messy around here right now, so I can’t find it).
    David Crystal is one linguist who is very good at explaining in non-technical terms (as far as possible) what linguistics (in all its aspects) is all about, especially as concerns English.

  68. Yes, I like Crystal a lot, and just for that reason (especially his Lynne Truss send-up); but I also like Barber (from whose book I just pasted a gigantic section in the Hollinghurst thread) because he doesn’t shy from technical terms, but rather works them in gradually, such that anyone but a total novice could follow along and learn.

  69. There’s a short interview with Everett on Nightwaves right now. The name sounds nothing like what I thought.

  70. Thesauro says:

    Just listened to that interview. Oh my, how awful. No, Mr. Everett, Chomsky never wrote that all languages must have subprdinate clauses. That’s right, Mr. Everet, he didn’t. No, Mr. Everett, not even a tiny bit. Not even in jest.

  71. You should have said you listened to the intro, because nowhere in the interview does Everett make that claim. That’s the announcer’s interpretation.

  72. Heh.

  73. Thesauro says:

    Hmmm… Here’s a transcript I just made, you can check it yourself.
    (36:13) “In English, if we want to make – or any European language, and most languages, if you want to express a complex idea as a single sentence, it’s quite easy to do that. So I can say “The man who came in here yesterday is here again today”, and that “who came in here yesterday” is a relative clause, so it’s a sentence inside another sentence.  That property is called recursion, and it’s so common that Noam Chomsky said it was the unique aspect of human language.
    (Interviewer and Everett identify who Chomsky is.)
    “So it turns out that if you look at all the structures in the language (Piraha), you see no evidence for recursion. You see a couple of things that could be – they’re reminiscent – but on closer examination they’re not recursion. This means that all the Piraha sentences sound relatively simple.  So how would a Piraha say “I think John is here”, because that’s recursion in English.  “I think” is one sentence, “John is here” is another.  They would actually just have a suffix on the verb to get this across.  “John is here” and then a little suffix thatr means “think”…
    (Interviewer asks about language shaped by culture, supposedly challenging Chomsky and Pinker. How is Piraha a counterexample?  Isn’t it an exception that proves the rule, since 99% of languages obey the rule?)
    “It doesn’t seem that Piraha is the only language like this. But this is equivalent to saying that what makes birds distinct is their ability to fly. All birds fly.  Now you discover penguins, and you say “whoops, is this the exception that proves the rule or is it simply false to say that all birds fly”.
    (Interviewer: “but the penguin is still a bird”)
    “…but it’s a bird that doesn’t fly.  And Piraha is still a language, but it’s a language that doesn’t have recursion.  And this means that if I’m correct, that can’t be the universal basis for human language.
    (Interviewer asks whether this is a politically correct view, citing Pinker. Is it Eurocentric condescention, testing Piraha against European languages?)
    (39:12) “That would mean that Chomsky is the Eurocentric person, because Chomsky’s the one – it’s not me that said recursion is found in all languages, it’s Chomsky that said they’re found in all languages.

  74. I see nothing about subordinate clauses there.

  75. Thesauro says:

    He’s talking about nothing else. I don’t understand the comment.

  76. He’s talking about recursion, using subordinate (relative) clauses as an example, the same one that Chomsky, and everyone else, most often gives. By pretending that he, Everett, thinks recursion only means subordinate clauses and that therefore Chomsky is saying that all languages have subordinate clauses, you make him sound stupid.
    Chomsky, of course, would never say such a thing. It would be far too easy to refute. Better to talk of recursion, and narrow or broaden the meaning as needed.

  77. Thesauro says:

    “so it’s a sentence inside another sentence.  That property is called recursion, and it’s so common that Noam Chomsky said it was the unique aspect of human language…it’s not me that said recursion is found in all languages, it’s Chomsky that said they’re found in all languages.”
    Chomsky did not say they’re found in all languages, but Everett says he did, Right there. Right above. What is there to argue about?

  78. Bathrobe says:

    That property is called recursion doesn’t necessarily mean only relative clauses are recursive. You’ve got to remember he’s not writing an article, he’s talking to a radio journalist. He’s just trying to make the concept as easily understood as possible. If you want to trip someone up for what they say to a journalist on BBC 3, it shows pretty clearly that you’re just trying to score cheap points.

  79. Thesauro says:

    This feels like arguing with climate change deniers or creationists. Not a good use of time, even by time-wasting-on-the-Internet standards. The text is there for all to read. Believe what you want.

  80. Thesauro says:

    This is silly. The text is there for all to read. Believe what you want.

  81. Bathrobe says:

    I assume your point is that Chomsky did not say all birds fly; he is saying that all birds have wings, even if they don’t use them to fly. Point granted, but I think you should save your Internet time-wasting for academic articles, not BBC interviews.

  82. This feels like arguing with climate change deniers or creationists.
    That’s exactly what it feels like from this end, too.

  83. Thesauro says:

    Have a nice day.

  84. marie-lucie says:

    Since this discussion started I have been uneasy about the use of the word “recursion”. My syntactic training must be quite dated by now: I always thought the word “recursion’ was related to “recur, recurrent” and referred to the way a sentence can be added to and made longer by adding extra material, especially by embedding sentences or clauses of a similar type, as in This is the cat that ate the rat that lived in the house that Jack built, a sentence which is made to grow by adding more material similar to what is already in the sentence, by inserting new relative clauses after This is. I did not think that adding a single subordinate clause as a complement to a verb, as in You think [that] you’re right, was an instance of recursion: what is recurring? Recursion would create something like I know that you think that you’re right. I taught introductory linguistics for many years, including the basics of generative syntax, and this is the type of thing that my textbooks (regularly updated) were stating. Here is an example of recursion, not of embedded clauses, but of recursively embedded noun-phrases, taken from a 2004 textbook:

    the cat in the hat on the table by the chair in the corner of the kitchen in the house under the tree…

    (although “in the hat” here may be a mistake as it suggests the cat is actually within a hat, not wearing one like Dr Seuss’s character).
    When did “recursion” become a synonym for “clause embedding”?

  85. Bathrobe says:

    From an amateur:
    I was under the impression that one problem in the interview is that Everett is equating clause embedding and recursion. The other example of ‘recursion’ that has come up in the Everett discussions is ‘Tom’s mother’s house’.

  86. Bathrobe says:

    Also, you only need one embedded sentence to talk about recursion — anything deeper than a flat syntactic structure (even one layer of embedding) is already recursion.
    Again from an amateur.

  87. Marie-lucie, you’re quite right; ‘recursion’ is not a synonym for ‘clause embedding’.
    When Chomsky, Hauser, and Fitch were writing a popular science article for non-linguists, they set themselves the task of trying to explain what the difference was between human language and animal communication systems. They suggested that what was special about human language was the fact that we can combine words into larger grammatical structures. There are animals (like vervet monkeys) that have alarm calls that you can call ‘words’ if you want to; the vervet monkey vocabulary, as far as I know, consists entirely of nouns, mainly names for predators. So a vervet monkey can effectively yell ‘Leopard!’, or whatever, so that his fellow vervet monkeys can take evasive action. But he can’t combine the various alarm calls into a sentence or into any larger grammatical unit; there’s no reason to posit a vervet monkey equivalent of a verb phrase, for example. Human beings, by contrast, can take words and put them together to form phrases, and keep adding phrases and words to the structure until they form a sentence.
    That’s it. That’s what ‘recursion’ means, in Chomsky, Hauser, and Fitch. ‘Recursion’, in this context, means that the process of creating grammatical structure can take its own output as an input; you can take the phrase you’ve created, and add more stuff to create a bigger phrase, to which you can add more stuff to create a yet bigger phrase, and so on. In this sense of ‘recursion’, there really is no human language that lacks it. Everett’s right, in a sense, that “I think John is here” involves recursion–but so does “John is here”.
    A consequence of the universal human faculty of recursion is that sentences have no arbitrary bound on their length. The easiest way to show that is with clausal embedding, or with recursive noun-phrase embedding, which I think is where a lot of the confusion comes from, since these are often the examples of recursion that people use. But they are not what Chomsky, Hauser, and Fitch meant by recursion.
    The recursive process of sentence construction obviously has to be constrained; the proposal is not that every language can combine any kind of grammatical object with any other kind of grammatical object. To take one concrete example, English prepositions cannot take declarative clauses as their complements; we cannot say things in English like ‘They are talking about that it is raining’. In some other languages, by contrast (Spanish, for example), it’s possible to do this. Everett is claiming that Pirahã is unlike English in that its verbs are like English prepositions; they cannot take clausal complements. He also thinks that this would be a problem for Universal Grammar, about which I hope I have shown him to be mistaken. His arguments that Pirahã verbs cannot take clausal complements have also been challenged, I think pretty effectively.

  88. marie-lucie says:

    Tom’s mother’s house is indeed another example of recursion. Somewhere in the context of the Chomsky-Everett controversy I read that someone said that German did not have recursion in such a case, as you could only have one genitive: eg Tom’s mother or Tom’s house, but not two (I think this is because you can only use the preposed genitive with names, but I could be wrong). But you can use recursive noun-phrases for the same meaning, so German does have recursion, just not the same recurring genitives as in English. I don’t want to try to create a complex German NP, but in French, where there is no genitive, you would say la maison de la mère de Tom, another example of recursion (as in “the cat on the table …” above).
    It seems to me that “recursion” is becoming a grab-all, different people using it with different meanings. This is not helping to solve the controversy.

  89. marie-lucie says:

    Norvin, thank you for your detailed explanation, which you posted while I was writing my next comment. I still think that “recursion” is not the right word to use in this very general sense: if any sentence, even the simplest one, exhibits “recursion”, the word seems to mean only “syntax”!

  90. The point about Tom’s mother’s house is in the paper by Nevins, Pesetsky, and Rodrigues, which I linked above; here it is again:
    Their point is that we often have to declare that languages lack recursion in one or another position (in the German case, pronominal possessors are non-recursive, as they are in Pirahã), so to find a language that lacks recursion in several places shouldn’t surprise us.
    Everett’s reply is here:

  91. Marie-Lucie–
    I agree! Chomsky, Hauser, and Fitch should have used a different word–would had spared everybody a lot of suffering.

  92. The definition of recursion is simple and straightforward:
    Recursion: see recursion.

  93. marie-lucie says:

    Norvin: Their point is that we often have to declare that languages lack recursion in one or another position (in the German case, pronominal possessors are non-recursive, as they are in Pirahã), so to find a language that lacks recursion in several places shouldn’t surprise us.
    But here the definition of recursion is the old one: the possibility of nestled or embedded structures, within similar structures.

  94. marie-lucie says:

    Perfect definition, JC!

  95. Thanks very much for your helpful and collegial comments, Norvin.

  96. Thesauro says:

    As opposed to comments like “heh”.

  97. Computer scientists use the term tail recursion for recursion at the end of a construct, which can be arbitrarily deep without processing difficulties. The same source defines it thus:
    Tail recursion: If you aren’t sick of it already, see tail recursion.

  98. marie-lucie says:

    Were you plagiarizing this, JC?

  99. I was, though I prefer to think of it as “unattributed sharing”.

  100. marie-lucie says:


  101. On investigation, this definition of recursion doesn’t seem to have a clearly established original source: it is basically computer-science folklore. There is a lot of that around, and there is a whole tradition of CS scholars writing “folklore papers” so that some well-known fact, result, or method can get into the literature and be properly citable. For example, “The Folklore of Sorting Algorithms” by Khamitkar et al. (2009) appears to be such a paper.
    A 1993 paper by John Reynolds, “The Discoveries of Continuations”, explains how the concept of a “continuation” (roughly, the formalization of the informal notion of “everything that is left to be done in a computation past a certain point”) was discovered independently at least seven times: once in 1964, repeatedly in 1970-71, and several times thereafter. What is more, two of the discoverers were both named “Morris”, and are distantly related, though they were working quite independently.
    I also found an alternative definition, Recursion: If you still don’t get it, see recursion, which has the advantage that the definition will in fact terminate if the reader ever understands it.

  102. Thesauro says:


  103. I didn’t do well in programming, but isn’t what Chomskyïsm is talking about just concatenation – stringing words together?

  104. marie-lucie says:

    Sili: In syntax, concatenation occurs if you link words or sentences by and or but. Remove those linking words and you get individual words or sentences. Recursion involves a hierarchy of elements within larger and larger wholes, as in the full development of This is … the house that Jack built. But the normal meaning of the word seems to have been broadened to such an extent in some contexts that it is practically meaningless.

  105. The House That Jack Built is indeed an example of tail recursion, which is why we can understand it even at depth 10 (whereas non-tail recursion becomes unintelligible to humans at about depth 3):
    This is the horse and the hound and the horn
    (That belonged to the farmer sowing his corn
    (That kept the cock that crowed in the morn
    (That woke the priest all shaven and shorn
    (That married the man all tattered and torn
    (That kissed the maiden all forlorn
    (That milked the cow with the crumpled horn
    (That tossed the dog that worried the cat
    (That killed the rat that ate the malt
    (That lay in the house (that Jack built)))))))))).
    Having all the right parentheses at the end is the mark of tail recursion. We need not keep track where we are in the nested structures, because we are going to abandon all of them at once anyway.

  106. Daniel Everett is getting overrated. I debated for a long while whether or not I should comment on this incredibly dull (and I am keeping the language civil here) post, and some of the even duller comments. Much of Everett’s work is mere descriptivism that, at best, reports some irregularities which are well permitted within the premise of UG. Everett, and his rag tag band of wanna be-debunkers, motivated primarily by an intrinsic jealousy of the Chomskyan paradigm and its dominance, propose half-witted arguments against concepts that they do not fully understand. If you want to talk about UG, for or against it, you need first of all to understand UG. Recursion, in UG, is an uniquely human ‘ability’. It is an OPTION, NOT a COMPULSION. One language, imposing pragmatic barriers against manifesting possessive recursion does not falsify UG. Everett’s own data shows that Piraha uses recursion, albeit to a limited extent. It’s simply an atypical variation. Nothing more. Everett does not understand the concept of recursion, and is simply a descriptive linguist reporting a few aberration which, if anything, makes the case for UG and FLN much stronger.

  107. David Marjanović says:

    motivated primarily by


    Everett’s own data shows that Piraha uses recursion, albeit to a limited extent.

    Please explain that, then.

    (I’m not being sarcastic. I’m not qualified to form an opinion on that matter.)

    reporting a few aberration which, if anything, makes the case for UG and FLN much stronger.

    How so?

  108. John Cowan says:

    Gotta love that “mere descriptivism”; a merely scientific activity.

    It turns out that the use of English that, ancestrally a demonstrative, to introduce embedded sentences is precisely a diachronic change from concatenation to recursion: in pre-Old English, people said things like “He said that. The man bit the dog”, where that is a demonstrative pointing to the following independent sentence. By OE times, that had become a (tail) recursive sentence “He said that (the man bit the dog)”. CGEL indeed calls that a subordinator here rather than a relative pronoun.

  109. marie-lucie says:

    I guess I would be described by some people as “a mere descriptivist”. I wonder what adjective is appropriate to fill the blank in “a ____ generativist”. Sure, I have seen some poor language descriptions, but a generativist is not necessarily going to do a better job than a descriptivist. Trying to fit the facts of a language into a ready-made generative framework is not necessarily intellectually superior to setting up a specific framework that fits the facts of a language.

    JC, I was about to make the same point with English “that”, which went from being a demonstrative pronoun to a subordinator in certain cases, something that presumably should not happen if the Chomskyan model is right.

  110. John Cowan says:

    Well, it doesn’t violate the current Hauser-Fitch-Chomsky model of human language (whatever claims have been made in the past), which simply says that languages with recursion are unique to human beings (and the computers they have invented). This limited claim is essentially indefeasible, unless a form of non-human communication is discovered that requires a Chomsky type 2 (context-free) grammar to parse it.

  111. Trond Engen says:

    What little I have understood of the whole brouhaha:

    1) Everett made some claims about Pirahã that sound outlandish and may or may not be correct. If they are correct, they provide counterevidence to UG as he understands it.

    2) Recent formulations (or refinements) of UG, however, are nothing of the sort. In fact they seem to claim so little that it’s hard for this layman to understand why the concept matters at all and what five decades of linguistic trenched warfare has been about.

    3) This should be a good time for both sides to claim victory, go home, pick up their tools were they left them and do some honest work. Or it would be, if it hadn’t been for the fact that everybody did so years ago anyway, only still shouting at eachother from across the hall at gatherings.

Speak Your Mind