May 04, 2006

On the Internet Nobody Knows You are a Space Alien Lizard

(Guest post by Paul Postal)

David Beaver (DB) in his recent post about the linguistic abilities or lack thereof of animals cites the assertion by Noam Chomsky (NC) in (1):

(1) [...] if someone could show that other animals had the basic property of human language, it would be of very little interest to the biology of language, but would be a puzzle for general biology.

DB disagrees with this claim and ends up stating:

(2) Where I disagree with him is in the general principle he invokes, which seems to imply that even animals producing and comprehending grammatically correct English would be of no consequence for linguistics.  Such a conclusion would be ludicrous.

Now, while I have rarely over the last decades defended any claim of NC's and  have quarreled with many, the substance of (1) seems to me at least potentially sound, at least up to the but clause, which I ignore.  Moreover, one notes that in concluding as in (2), DB has a bit changed the terms of reference without warning.  NC's claim was about the biology of language and DB's is about linguistics.  Of coure, some, including NC, largely identify these two, but such an identification has never been justified and is, I would argue, nothing but a category mistake, confusing inter alia language and knowledge of language.

Here then is how I see (1) and (2).  (1) might be sound for uninteresting reasons.  If one finds an animal with the same linguistic competence of humans, one with NC's point of view would be free to ‘account’ for that in the same way he proposes to ‘account’ for human competence.  Namely, he could simply say that the innate linguistic organ he has posited for humans,  whose nature he has never specified in any biological terms, and which has no known physical properties, **** is present in the relevant animal as well.  If that account ultimately succeeds for humans, there is no reason why it would necessarily fail for the animal in question.  Under these assumptions, the animal simply provides another subject, no different in linguistic essence from a human, and that would yield a situation which is genuinely linguistically uninteresting, whatever thrill it might provide zoologists.  Of course, if NC's innateness views are wrong, then finding a linguistically competent animal might show something, e.g. that the general skill needed to learn a language is available more broadly than in humans. 

I don't care much about (1) because I do not believe linguistics, or more importantly, its subject matter, language, have anything much to do with biology.  But (2) which mentions linguistics is another matter and I consider it deeply wrong.  Two thought experiments can show why the issue of nonhuman linguistic competence is essentially irrelevant to the understanding of natural language.

First, let us briefly try to give a sense of a genuine linguistic issue, call it ISS.  The point will be that that ISS is such that no discovery about animal linguistic competence could bear on ISS in any way different than human linguistic competence does.  Take ISS to be an issue, which exists regardless of the specific theoretical assumptions cited to illustrate it. 

Seuren (1985) cited (3):
(3) John and every woman in the village want to get married.
About this, Seuren (1985: 22-23) claimed:
(4) In (10b) every woman cannot take scope over the whole remainder of the sentence; as a consequence it cannot mean that for every woman in the village John and that woman want to get married to each other.  Its only possible reading is the one in which John as well as every woman in the village want to get married.  This is explained in principle by the theory that quantifying into a co-ordinate structure is ruled out by  the Co-ordinate Structure Constraint (Ross, 1967).

So take ISS to be the issue of what principles of language determine that the universal quantifier phrase every woman in the village in (3) has the scope that it does.  According to Seuren, these principles included the coordinate structure constraint of Ross (1967) which in Seuren's framework or that of May (1985) would preclude a required quantifier lowering into/quantifier raising out of the coordinate phrase.  This claim, even if right, leaves things mysterious however, since it is then not obvious how every woman in the village can have any actual scope at all.  For its scope must certainly include material external to the coordinated subject, the predicate complex  want to get married.

OK, we are not going to solve  ISS today, since our interest is in nonhuman language possibilities and actualities.  Turn then to the two thought experiments.  One harks back to the bad 1983 science fiction television series V.  In this, large evil lizard like space aliens try to take over the Earth.  They don’t look like lizards though because of a fake outer coating which gives them human appearance.  Here is the thought experiment.  Suppose that one of the regular Language Log posters, say Geoff Pullum, is in fact a creature just like the V series space lizards (possibly  not a new idea).  One then must accept that nonhumans know and can use English just like real humans (assuming there are any).  And that tells us what about ISS?  Evidently, nothing.   Just how could  Geoff Pullum's being a space alien lizard instead of an Earth mammal offer any insight into ISS.  Nor would it matter if he were a raccoon, bluebird or triceratops in human form.

The second thought experiment appeals to the known phrase "On the Internet nobody knows you are a dog", which comes from a New Yorker cartoon found on page 61 of the July 5, 1993 issue (available at  The cartoon shows a dog sitting at a computer terminal with another dog in attendance and the phrase is the caption.

So simply assume the cartoon is realistic...suppose that all the messages on some website, say Daily Kos, have in fact been written by canines.  That might have political implications, but as far as ISS is concerned, it means nothing.  One learns and can learn in principle nothing more about ISS by the discovery that there are English knowing dogs than by the discovery that there are English knowing space alien lizards.  It just doesn't matter.

There is one case of putative linguistic ability in nonhuman animals not covered by the thought experiments.  Suppose one finds an animal  who is shown somehow to know some variety of some hitherto unknown natural language.  That would be of more interest, exactly as much as finding a hitherto unknown language known by some humans.  Maybe this could contribute to linguistics, but if so, it won't be because the language is known by a nonhuman, but because the language itself can teach us something.

However, lets face it, with thousands of known languages available for study, we haven't really been able to understand that much.  Why would just finding one more be likely to change things, regardless of whether that language is known by space alien lizards, canines or simply a group of people not previously contacted or known.

For what linguistics lacks is not languages to study but insight into them.  And that won't be  provided in any clear way by further linguistic creatures, regardless of whether they are people, gorillas or roaches.

Underlying these remarks is the view that language is entirely distinct from biology, just as mathematics, set theory and logic are.  If we discover a crocodile with the same knowledge off mathematics as the best human mathematician, it  won't inherently help determine whether Goldbach's Conjecture is true, and a set theoretical expert gerbil will not thereby provide any insight into the truth of the Continuuum Hypothesis.

***It is an odd organ, to say the least, that has no specifiable physical properties.  But worse, NC's assumptions do not permit the hypothetical organ to have physical properties, since he claims that a human language is a state of the innate organ and is discretely infinite.  Nothing  infinite can be a physical organ or a state of such, on the naturalist assumption, which NC of course makes, that human bodies are physical.  Hence the claim that the posited organ has something to do with biology is not serious.  Real organs, e.g. livers, are all too finite.  Moreover, real organs can't produce an infinite number of, or amount of,  anything, e.g. bile.  The bottom line is that NC's position  that natural language is both to be taken as an organ state and as discretely infinite is simply incoherent.

- Paul Postal, NYU Department of Linguistics

[Postscript April 4, 06

DB (email sent in reaction to the post above):
I should make clear that the reason I think animal results could in principle have anything to do with linguistics is not because they *ought* to have anything to do with linguistics. It's because the lamentable history of our field is choc-a-bloc with people making unwarranted claims about language organs, their distinctive functional ability, their biological uniqueness in humanity etc. This is why I mockingly suggested at the end of my previous post that the new data suggests birds have a language organ but we don't.

It is because ridiculous and un-evidenced claims about biology are strewn about the field like acne on a friend's face that real genuine evidence about what animals can or cannot do *is* relevant to our field. To deny this is to be an idealist, to pretend that linguistics is unsullied by all the weird things that linguists say. But it's nice to be an idealist.

PP (return email):
We are essentially in total agreement about unwarranted biological claims in if all that is involved in citing animal results is debunking the relevant posturing, I am all for it.

My view is though that at a deeper level, biology has to be irrelevant, for the same reason it is irrelevant to e.g. Goldbach's Conjecture.

I doubt if anyone will care about the post....some dogmas are too deeply embedded in widespread thought to be confrontable with fact or argument. But one will see.]

Posted by David Beaver at May 4, 2006 01:40 AM