News and views about philosophy, the academic profession, academic freedom, intellectual culture, and other topics. The world’s most popular philosophy blog, since 2003.
I’d like to pose a question. Let’s be pessimistic for the moment, and assume AI *does* destroy the university, at…
Hear hear
I agree with all of this. The threat is really that stark. The only solution is indeed in-class essay exams,…
I’m not sure I’d yet go so far as to call LLMs an existential threat to universities. But I do…
My big problem with LLMs at the present time, apart from being potentially the epitome of Foucault’s panopticon & Big…
I’m also at a British university (in a law school) and my sentiments largely align with the author’s. I see…
If one is genuinely uninterested in engaging with non-human interlocutors, it is unclear why one continues to do so—especially while…
Several people have sent me the CHE write-up. I'd be grateful if some of the experts in the readership could share their perspective on all this.
Talk by Everett last week at the LSE: http://bit.ly/GRlOVJ
Brian, I'm no expert, but you (and other readers) could have a look at Mark Liberman's take at Language Log, or Geoff Pullum's in the Lingua Franca section of the CHE.
Hm, maybe no anchor tags allowed in typepad comments? Just in case:
Geoff Pullum's blog in The Chronicle has an interesting account of this "venomous dispute".See http://chronicle.com/blogs/linguafranca/2012/03/28/poisonous-dispute/
I'm a Brazilian PhD candidate in linguistics (and an avid reader of your blog, I must add) coming from a non-Chomskyan background and currently advised by a Chomskyan. I'm something of a rookie, but I believe I can throw my two cents as I've seen this debate from both sides and I've met at least one person from each camp mentioned in this article (Everett himself and Nevins, both of whom I admire), so this is all very familiar to me.
— The kerfuffle summarised —
Noam Chomsky's generative enterprise has been criticised for decades now (as the article says), and Daniel Everett's claims about the Pirahã are just something of a tipping point in what seemed to be constant duel between innatists and everyone else. Daniel Everett's ideas aren't exactly new, reason why he's able to cite loads of authors with a similar worldview. In a lecture I attended last year, he repeatedly mentioned Michael Tomasello, himself a critic of Chomsky's UG. The data (or lack thereof) that could falsify UG is, I believe, scattered across different languages. If you're interested, the article that says the claims from generativist quarters "are either empirically false, unfalsifiable, or misleading in that they refer to tendencies rather than strict universals" can be read in its entirety here:
http://www.mpi.nl/news/news-archive/the-myth-of-language-universals
What Everett brought to the table is just some more hard data that corroborates findings as the ones mentioned above, but they now come from a single language (being a novelty), and they bring down the apparent "last universal" that is considered sacrosanct. Or do they?
— My humble opinion —
Personally, I find the apparent tribalism appalling. Do you reckon Everett's methodology is a bit dodgy? Then let's all get over with the ad hominem attacks and get some more data from the Pirahã! If they point to the same direction, so what if Chomsky got it wrong? Less than 20% of all living languages have been described anyway, and we should all know something about the problem of induction by now. Maybe there are other languages that lack recursion! And, if there are, great, let's get back to the sketching board and cook up different explanations. The results of inquiry should not block the path to further inquiry!
Chomsky has made great contributions to the history of science, and for this even detractors must be grateful. We can't get everything right, specially with the scarcity of data linguists have at their disposal, so there's absolutely nothing wrong if Chomsky is mistaken.
Where to begin…
In May of 2009 I gave a talk at a conference on recursion in language at the University of Massachusetts, Amherst. I flew into Springfield MA, and shared a van ride to Amherst with a couple of other conference participants, including Dan Everett. I was only vaguely aware of the Pirahã controversy at the time, but one thing that became salient to me in that van ride is that Everett really totally hated Chomsky. I had to listen to him complain about Chomsky for the entire forty minute ride.
Of course Chomsky haters are not rare, even among linguists. But usually, in the linguistic case, the Chomsky hater offers facts about a language that someone else has access to, and offers a theory that we can evaluate. The problem in this case is that the Chomsky hater is talking about a language that no one else at the moment has access too, and is not claiming to have found some phenomenon that refutes something, but is claiming *not* to have found a particular phenomenon. And the question that became salient to me in the course of the ride was this: Forget about how hard he was looking, does Everett even know what he is supposed to be looking for? Does he even know what recursion is and the role it is supposed to play in contemporary linguistic theory?
Let’s start with that. In recent work Chomsky (e.g. Hauser, Chomsky, and Fitch) has argued that the language faculty developed very quickly, and involved a mutation that led to the wiring up of the perceptual/articulatory system with the conceptual/intentional system, thereby allowing us to generate form-meaning pairs of arbitrary complexity. On Chomsky’s hypothesis, this did not involve a long period of gradual evolution, but a form of “punctuated equilibrium” and this was possible because wiring solution was driven largely by low-level biophysical processes. The idea is that because it is largely a product of physical and mathematical constraints it is not surprising that we see recursive structures in the product (just as we do throughout nature – in the growth of shells, etc). That is, recursion is evidence for the idea that the evolution of language was driven by the physical channel rather than by gradual evolutionary tinkering.
I reiterate that the main point here is that recursion is involved in the construction of form-meaning pairs in human language, and nothing is immediately predicted about where the recursive structure (or recursive interpretation rules) are located. In fact, it turns out that there is lots of variation among known languages about the kinds of recursive structures that languages evince on the surface. For example, the point has been made that in Pirahã you can’t recursively nest possessives (John’s mother’s cousin’s dentist), but neither can you do this in French. Tom Roeper, who was an organizer of the UMass recursion conference has a very interesting book about the wide cross-linguistic variation in recursive structures.
So is it consistent with Chomsky’s theory that there could be a language without recursion and if so could Pirahã be such a language? Some people have argued that there is no prediction that recursion should appear universally – the interesting point, they argue, is that we have a capacity which allows us to acquire a recursively structured language if exposed to the right input. That is correct, I think. Furthermore, recursive structure is still evidence for the punctuated equilibrium and physical channel thesis, even if you don’t find it everywhere. It would hardly undermine a similar thesis about shell growth if you found shells that didn’t evince recursive structure. However I personally seriously doubt that Pirahã or any language is recursion free in the relevant sense.
Let’s start with a well-studied language that, like Pirahã, seems not to have recursive structure. As Geoff Pullum pointed out in The Chronicle blog, many years ago Ken Hale observed that the Australian aboriginal language Walbiri (like Pirahã) does not appear to have relative clause recursion. That is, you can’t say something like (1).
(1) John saw the cat that ate the rat that ate the cheese.
Does this mean that Walbiri doesn’t have recursion? Everett would probably say that this is what it shows, but in this case we are lucky enough to have someone around who knew what to look for. In the early 80’s Richard Larson wrote his dissertation on Walbiri under the supervision of Jon Barwise and Robin Cooper. His analysis was that Walbiri did have something like relative clauses, it’s just that they are not embedded in the noun phrase. So the way you say (2),
(2) John saw the cat that ate the rat
Would be something like this.
(2w) Jangala saw the cat. This one ate the rat. (or you could use variants of ‘that one’ like ‘that rat’ or ‘that animal’)
You can think of there being a cross-sentential event variable or situation variable linking ‘the cat’ and ‘ate the rat’. The relative clause has simply been displaced. How would you say something like (1)? Larson (pc) thinks it would be as follows:
(1w). Jangala saw the cat, that one chased the rat, that one ate the cheese
Again, the clauses are not embedded in the noun phrase, but they are exported, as it were, and they are chained together by cross-sentential variables. The resulting structure is certainly recursive – it’s just that the recursive structure is not contained in a single sentence but is in a discourse structure. All this is consistent with the core Chomsky observation, which is that the construction of form-meaning pairs seems to involve recursion. It also gives rise to the observation that when you take recursion away somewhere it seems to come back somewhere else.
The key point is that the questions are complicated and subtle, and I don’t see any evidence that Dan Everett is equipped to answer these questions in the case of Pirahã.
To give you one more example of the complexity here, consider something simple like adjectival recursion. I for one would be shocked if Pirahã didn’t allow you to say things like
(4) That’s a big hairy rat
Can you say something like (4) in Walbiri? Yes, but it’s complicated, because in Walbiri there is no difference between (4) and (5).
(5) That’s a hairy big rat
Both sentences are ambiguous (both have the meanings we attach to (4) *and* (5)). Does this mean that the structures are not recursive? Well, it could mean that the nesting possibilities are not fixed by surface order, and that there are two possible logical forms (LFs) associated with (4), or it could mean that the noun phrase structure is flat and that the semantics can interpret it in two different ways. However even if this is the case, the fact that an ambiguity is recognized suggests that at a minimum the semantics is picking an order in which to evaluate the adjectives and this just means that the recursion has been moved into the semantics (for example the successive application of a recursive truth predicate in the case of a Davidsonian semantics).
The really annoying thing about all this is how the popular media is hyping Everett. Suppose someone in physics conducted an experiment that could not be successfully reproduced and the physicist claimed that in his experiment he failed to detect a particle (that every other laboratory in the country had successfully detected) and he concluded that therefore the dominant theory had been refuted. Now suppose the physicist was known to despise the author of the dominant theory. Finally suppose that there is no reason at all to believe that the physicist had the necessary expertise to conduct the experiment correctly and did not have the theoretical knowledge necessary to even know what to look for. Would the popular media be swooning over such a physicist and hyping his work or would they just laugh it off? I suspect the latter.
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.
So are some linguists pissed off about the Everett hype machine? You bet they are. And they should be.
I enjoyed reading Marcus's comments, which I agree with. Let me just give a bit of history. In 2005 I published a paper arguing that Piraha lacked embedding and any other form of recursion. I also discussed lack of numbers, simplicity of kinship in Piraha, etc. I concluded that culture is causally implicated in grammar. I further suggested that theoreticians should should consider "whole" languages and not merely bits and pieces, what I referred to as the "smorgasbord approach" to fieldwork. The editor of Current Anthropology, as they often do, told me, after the fact, that he had asked the University of Chicago Press to issue a press release on this article because he thought it was important. I felt honored. I had no idea what was coming.
As a result of that paper, I was invited to give a talk at MIT's Brain and Cognitive Sciences Department. Before that talk, extremely vicious emails about me exploiting the Pirahas for my own fame were circulated to the mailing lists for that talk. This shocked me and hurt my feelings. At the talk, Marc Hauser sat on the front row, and the three authors of a long Language article on the subject (then in progress) were in the audience. I had offered to meet with them prior to the talk to discuss anything they'd like, but they refused. And I saw emails in which they suggested I was racist.
This not only hurt my feelings, it made me *angry*. And there were times in those first few months that I did not control my emotions, especially my anger, and I said intemperate things and allowed my first responses to be overly emotional. I apologize for this here, as I have apologized elsewhere.
I do not hate Chomsky. I owe him a great deal. He wrote a letter of recommendation for me for my first job at the U of Pittsburgh. And he wrote a letter for my tenure case. I was for years, prior to the internet, when I was living in Brazil, on his mailing list and it was always a matter of pride to receive his mss. He inspired me tremendously in my career. And though reporters these days ask me to say negative things, I no longer feel that anger I once did and I do not. And, again, I am sorry for angry things I have said.
I have made all of my data, a huge amount, available to researchers at MIT's Brain and Cog Sci department. And I made a large amount collected by the missionary who preceded me to the Pirahas available to Ray Jackendoff and Eva Wittenburgh at Tufts (they have a chapter in a forthcoming book about Piraha recursion, essentially agreeing with the conclusions of my last paper in Language).
My new book has nothing to do with this controversy – though it may start controversies of its own. I think that universal grammar is one of the least fruitful ideas anyone has had, though historically I can understand why it seemed good to many of us thought otherwise.
The film, Grammar of Happiness, to appear from the Smithsonian Channel in May, does give a lot of space to the controversy. I had no input into the script. I simply answered questions for about a hundred hours.
I know that a lot of the publicity has come about because Chomsky's name is involved. No doubt about that. But that is the person whose ideas I am criticizing. Now, Chomsky has called me a charlatan and a liar in print and insinuates this in the film. But I have tried to avoid responding in kind.
Have I said things in anger that I wish I could unsay? Yes I have. I am trying to be a better person.
Dan Everett
I should add that I am familiar with all of the arguments that would distinguish the Warlbiri claims and the Piraha claims that Ludlow raises. I have done considerable work on recursion over the years, with lots of experience with logicians, computer scientists, and mathematical linguists. Three of the first graduates of the program in computational linguistics from Carnegie Mellon University were my students. It is correct of Ludlow to point out that there are always other possible analyses. I have tried to consider the alternatives and appreciate his raising additional questions of his own. I organized the very first conference on recursion ever held, at Illinois State University in 2007, which inspired the U Mass folks to organize one of their own.
Have I vented about Chomsky? As I said in my earlier posting, yes I have. I apologized to Ludlow off-line for him having to sit through this. I had been called a charlatan and a liar and a racist in print not long before that, the first two by Chomsky. So I wasn't in the best of moods.
The issue is to evaluate the results. Here is the link to the paper in progress with Gibson et al:http://tedlab.mit.edu/tedlab_website/researchpapers/Piantadosi_et_al_2012_LSAtalk_Piraha.pdf
Interested folks might also contact Ray Jackendoff at Tufts, who has seen a great deal of unpublished data on Piraha and who has his own paper on Piraha to appear.
But these days, I am much less interested in the Piraha controversy than the arguments in my new book, Language: The CUltural Tool. I was disappointed that both the Chronicle and the TImes ignored this. But I understand that the Times actually has a review to appear in their book review magazine that will, shock, focus on the book.
Controversies and disputes arise in science. The press can whip these up. I have been in the press for more than 7 years now. In just about every major newspaper in the world. But I have never asked to be "covered". And anyone who knows anything about the press knows that one doesn't just ask to be covered.
In any case, although the David vs. Goliath aspect has played a role in the coverage, there is also no denying that many scientists, philosophers, and reporters think that the chomskyan paradigm is on its last legs and that cases like Piraha are serious empirical problems. This is not all hype folks. There are serious issues for Chomskyan theory.
But the most serious ones are not recursion. As I point out in the new book.
I'm not going to add to the long comments here with one of my own, since I've replied extensively to both the original Chronicle article and to Pullum's blog posting in their comments sections. The current hubbub about Pirahã is a revival by the press of a dispute that was never "settled" but was otherwise dormant (since there seemed to be nothing new to say) for the last three or four years – prompted by the publication of a new book by Everett. Since there's nothing new going on, the best sources of information are the old ones. One good way to get up to speed, I think, is to read my contribution to the discussion here – http://edge.org/conversation/recursion-and-human-thought . Scroll down below Everett's statement to read replies by Steven Pinker, the linguist Robert Van Valin, and myself. I give a brief history of our response to Everett, discussing how we happened to write it, what we said, and what reaction we received.
The current spin on the story – that we overstepped the bounds by turning an academic discussion into a personal attack – is curious (and curiously data free – Chomsky, not us, called him a "charlatan" in a press interview), especially in light of the complaint I voice in the edge.org discussion about the ad hominem nature of Everett's response at that time. I mention this only because conduct, rather than content, appears to be the issue that obsesses both the Chronicle reporter and Pullum.
If you would like to read our published paper, Everett's published reply, and our reply to that, you can find these materials here: http://ling.auf.net/lingBuzz/000411 . If you are a *real* enthusiast and want to see the unpublished previous version of our paper, which prompted the never-published reply by Everett that I complained about on edge.org, you can find a link to our first draft at the same as the published verson, and you can find Everett's unpublished reply here: http://ling.auf.net/lingBuzz/000427
Leave a Reply