Just perfect

It's been a long time since SC has felt like there's been a reason to roll out the cracker barrel, but thanks to a submission to HPSG-L, informing the community of a new article by Steve Pinker and Ray Jackendoff, we're happy to present:

The Faculty of Language: What's Special About It?

Unfortunately, Profs. Pinker and Jackendoff have chosen to use M$ Word as a publication format, so this won't be accesible to everyone. But it should be. Which is different from saying that SC endorses the paper wholeheartedly. Actually, what follows represents a purging of several years worth of unhealthy accumulation of bile in your host's mental spleen ([a-HA! He DOES believe in mental modules! -- ed.]), and is so unrelievedly snarky that he probably should have submitted it to NLLT for publication (disclaimer: NLLT is SC's favorite linguistics journal, precisely because they run this sort of thing).

Essentially, the goal of the paper is to argue against a philosophical view of Hauser, Chomsky and Fitch (the authors in properly credited order, referred to hereafter as HCF), that there exist two things called the Broad Language Faculty and the Narrow Language Faculty (inscrutably assigned the acronyms FLB and FLN), that the FLB includes a potpourri of generic primate abilities, and the FLN is a box containing the single trick of recursion, about which more shortly.

Now, it has to be said that SC was simply flabbergasted by the abstract, in which he learned that the views of HCF appeared in Science in 2002 (too bad SC doesn't read Science regularly; he would've had a field day with this one around Minimalists). In the abstract, we learn that HCF think language isn't "special" after all (which turns out to be necessary to keep it "perfect"), and that "the only such aspect is syntactic recursion, the rest of language being either specific to humans but not to language (e.g., words and concepts) or not specific to humans (e.g., speech perception)". The temperate response to reading this is: WOW. In order to illustrate exactly what sort of a reversal this represents, it would be as though Freud had published a paper late in life saying, "Maybe I'm really the only one obsessed with sex, and it doesn't explain much about the rest of you", or Darwin publishing something saying "To heck with Galapagos turtles, it's Genesis all the way down". How your host managed to miss this one is beyond him.

Readers wondering what exactly is at issue with describing the language faculty as "perfect" deserve an explanation here. HCF argue, following a line of Chomsky's since 1995 or thereabouts, that the mental apparatus concerned with language exclusively, the FLN in the above terminology, is basically just a little black box that does syntax, and everything else is not part of language. Semantics is dismissed to a "conceptual" unit which is now held to be general purpose, phonetics and phonology are just motor control devices for getting the vocal tract to spit out the conceptual content, and all of those are now held to be common primate capabilities. FLN not only is restricted to providing order and an interface between the other components, but recursion is the only human-specific ability claimed to be involved. This is held to be "perfect" (and "minimal") in the sense that it is the minimum possible adaptation for humans on top of their otherwise generic primate abilities that makes language possible.

To tie in another favorite theme of mine, financial columnist Jim Cramer often tosses variations on his favorite self-coined aphorism into his writing: "Irony isn't an investment thesis; it only looks that way". Similarly, irony appears to be the best explanation for how a couple of prominent cognitive scientists could now be complaining, "No! Wait! Language is special after all!". Actually, that misstates the point somewhat. In finally joining a number of linguists, neuroscientists, and psychologists which can be estimated at about...oh, all of them minus Chomsky on the subject of whether or not the language faculty makes use of general cognitive abilities, he's managed to go too far in the other direction. There are, after all, some things which are unique to humans. Pinker and Jackendoff lay out a few of them: the apparently unique ability of humans to discriminate among phonological features, the specialization of the human vocal apparatus for speech (compared to the more general sound-imitation abilities of other animals), the size of the lexicon compared to the symbolic inventories of animals. They also make some interesting points about how Chomsky and friends grossly overrate the importance of recursion in syntax, but their argument cannot be reasonably summarized in a shorter way than they do it, so go read the paper.

Two facts about Pinker and Jackendoff's paper irk SC greatly, and that's without even addressing the flaming hypocrisy of Steve Pinker resorting to the lexicon without also using a typographical convention that indicates he spat after every invocation of the idea. First is the insistence on taking the background pomposity of Minimalism seriously. Never mind the footnotes in the Minimalist Program that clearly state: "Merge and Spell-Out are defined as the two least possible activities of any biologically based computational system". I'm kidding; that's not really in there. But it might as well be considering that Chomsky actually has written "But in any useful sense of the term, communication is not the function of language, and may even be of no unique significance for understanding the functions and nature of language" and also has asked "how closely human language approaches an optimal solution to design conditions that the system must meet to be usable at all" (citations for both of which can be found in Pinker & Jackendoff's paper). If the language faculty has little or nothing to do with communication, then what design conditions could it possibly have needed to meet? I don't believe that anyone searching the archives of this blog will find anything at all that has been said with the intention of being kind to Minimalism, but it is deeply insulting to Minimalists to suggest that just because Chomsky continues to pursue a 50-year-old neurosis about the autonomy of syntax, they felt obliged to abandon the computationally intractable device of Move Alpha for the highly implausible notions of derivational phases and feature checking (defined as LFG or HPSG attribute-value matrices without the computational grounding). Oh wait, that's exactly what they (approvingly) cite Shalom Lappin, Bob Levine and David Johnson for saying first. Dismissing Minimalism on account of pretheoretical speculation is a lot like dismissing logarithms simply because John Napier also wrote another book where he mispredicted the date of the end of the world, or dismissing heliocentrism because Copernicus' explicit goal in advancing the theory was to produce better tables for astrological divination. In fairness to Pinker and Jackendoff, their issue is primarily with the speculative claims, not with the architecture of Minimalism, but they repeatedly appeal to the suspect validity of Minimalism to explain why the speculative work is invalid. It's not entirely cricket. But it is fun if you like this sort of thing (and your host does, very much).

The second thing that gets SC's goat (click here to hear it being dragged off) is that having decided to systematically dismember HCF's claims about the language faculty, Pinker and Jackendoff wrap up an otherwise successful paper by saying "Nyaah! Nyaah! Everything you claimed for your views...is actually true about ours!". Chomsky has handed them an unusually large elephant gun, and placed his ducks in an unusually small barrel -- he really does argue that "current biology must be revamped to accommodate the findings of Minimalist linguistics" (P&J's phrasing, but the longer following quote isn't different), while having also offered the disclaimers that Minimalism is “still just an ‘approach’” or “a conjecture about how language works”. It's too easy, it really is. But all musings about the evolutionary goals of language are necessarily speculative, and the necessary corrective to excessively boastful claims is a solid dose of humility, not more of same. After adducing as much evidence as they have, including so many embarrassing contradictions, Pinker and Jackendoff are entitled to a rhetorical victory lap. It's just that it's a state track meet that's been won, not the Olympics.

SC can't stop there without acknowledging the not entirely disinterested motives that: 1) brought this to his attention, and 2) led him to write about it. As has been written here before, the HPSG mailing list is prone to a certain amount of reflection and even defensiveness, which is perhaps inevitable when you're in a minority position in any debate. But the tone of the announcement is less that of scholarly debate than open war: "It seems to me that we sometimes exaggerate the strength of minimalism. It is coming under attack by various people who were once quite close to Chomskyan syntax". It's because your host tends to share these feelings, even though his stake in the argument is just about nil these days, that he can't help writing about it. Cracker barrel philosophizing about the ultimate goals of linguistic theory is enormously fun, but with a bit of sober reflection, it's a bit disturbing to think about the real results that Profs. Pinker and Jackendoff -- and SC -- could have been producing instead.

Comments

[Neel Krishnaswami]

Hello, I'm a programming language semanticist, and I periodically read computational linguistics blogs just because it looks like a peek into a weirdly-altered parallel dimension.

That said, I want to ask: does FLN even exist? Even if you grant that recursion is what distinguishes human language from animal vocalization, it seems like you can make a pretty good argument that recursion is a general cognitive facility for humans.

In particular, humans are the only animal that are secondary tool-users. Chimps and otters use tools, but only humans make tools to make tools with, and so on.

Also, humans are the only animals that engage in intermediary trade. Chimpanzees understand the concept of trade: Frans de Waal wrote that if you leave a broom in the chimpanzee cage, you can get it back by getting a chimp's attention, pointing to the broom, and then showing it an apple or banana. But humans are the only animal that engage in intermediary trade: we trade for things that we don't need, but which we can use to trade with other people.

Add language to this list, and to me it looks like recursion is a cognitive facility that humans developed and them applied to the whole range of general primate stuff.

[speedwell]

The office suite "OpenOffice.org" is a free and poweful download, well worth the downloading time and insignificant learning curve (if you already know MSOffense). It will open MSWord files just fine.

I'd give you a link to the site but I don't want to insult anyone's intelligence.