Monday, June 11, 2007

Recursion in grammar

OK, the Everett-Chomsky thing has moved too far beyond what I know about the history of our field. A day after I did, Language Log picked up the Tribune story about this, here, with a post by Mark Liberman. Like me, he focuses on the appalling stupidity of the article, but he provides some really crucial background on the debate over recursion, arguing that its centrality to defining human language is very recent:
This began to change with Chomsky's "mimimalist program" in 1995, which jettisoned the idea of any well-defined layer of syntactic structure in between meaning ("logical form") and sound ("phonological form"), and therefore also discarded many of the parametric switches and knobs available in his earlier theories. And the change became important in the early oughts, when he decided that the human language faculty "only includes recursion".
The important piece was that Ken Hale had long ago argued that Walpiri has 'adjoining relative clauses' rather than recursive structures.

That this is the key moment in the "only recursion" stuff I'll buy, but is it really true that recursion wasn't assumed to be a core part of grammar earlier in linguistic theory? I recall recursion being central through lots of early generative grammar in ways that would make it an extraordinary property for any language to lack. And poking around quickly in a few early works on generative syntax, I sure see a lot made of the notion. But I don't know the history here, so consulted Newmeyer's Linguistic Theory in America (the 1st edition, from 1980). He talks about ways of capturing "the recursive property of human language" (p. 27), and then (p. 36, emphasis in original):
By the early 1950s many logicians simply ASSUMED that a natural language was defined by a set of recursive rules.
So, what am I missing here? Hasn't recursion long been central enough to make Everett's claims outrageous even before minimalism? I can see that this is the perfect moment for him to trot out this stuff, in terms of attacking generative theory, but isn't it a broader attack on all modern formal approaches to language? Who has a view of language where recursion is something we can use or not use, like marking an illative case? (These are real questions — this is too far from my ken.)

Gotta run ...

Image from here.


Cassidy Rasmussen said...

As with Mr. Verb, this is beyond my linguistic history but an aspect of what Ken Hale was saying about Walpiri syntax that is not presented by Liberman is whether Ken was trying to argue that Walpiri had NO RECURSION at all or whether just this particular construction was not recursive. This is a major distinction because there is a difference between a particular clause type being adjoined in a non-recursive manner (Ken's claim for Walpiri) vs. a language completely lacking recursion in the syntax (Everett's claim for Piraha).

Within Chomskyian linguistics there are grammatical components that are known to be relatively (or possibly completely) 'recursion free' such as the phonology module or phonetics. So, I don't buy the conspiracy aspect of the current Piraha Brouhaha that Liberman is pushing in that post where he suggests that there is something fishy now because recursion is being defended as a core feature of human language.

My guess is that the difference in reaction is based on the purposes or goals of the persons making the non-recursive claims. I only met Ken Hale a few times but know many who have extremely high respect for the man and I'm unaware of any axe that Ken was grinding regardless of the claims he was making. I haven't met Everett but let's just say that Chomskyian grammar appears to be one huge whetstone for the man...

Mr. Verb said...

Actually, I was wondering the same thing, and figured the talk about parallels between Pirahã and Australian languages suggested that the answer might be 'yes'.

Seems like we all agree that personalities play a huge role in this thing.