[Corpora-List] Looking for linguistic principles

[Mike Maxwell]

Stefan Bordag wrote:
> And I might add a little further up in the same section of Finchs 
> dissertation:
> 
> This [structuralist] paradigm was criticised by Chomsky (57) for failing 
> to properly dissociate the definition of what structure existed in 
> natural language from the procedures which allowed that structure to be 
> found, and of being too ambitious in any case, there not being enough 
> information in a corpus of a natural language to define its structure.

What Chomsky was arguing against was the prevailing American Structuralist 
theory of the time, which was indeed very much concerned with the issue of 
procedures for discovering generalizations, what were termed "discovery 
procedures".  This was particularly true in phonology.  It was harder to 
come up with discovery procedures in morphology, although people tried 
valiently to come up with s.t. like the notion of minimal pairs--Longacre, 
long after most American Structuralists had retired, wrote a book which 
(IIRC) was called "Discovery Procedures", and proposed a notion of minimal 
pairs in grammar.  As for syntax, there was virtually no work done on that 
among the American Structuralists (except for the aforementioned work by 
Longacre, and a grammar of English by Nida, IIRC).

As for the dissociation between structure and procedures, I think that is a 
very apt characterization of Chomsky's view.  What he was saying was 
basically, look, here is the obvious phonological structure of language X; 
but the discovery procedures don't allow you to capture that, they can only 
get you this inferior analysis that misses an obvious generalization. So 
get rid of the discovery procedures, so we can get on with real 
linguistics.  (The most widely used example of the failure of discovery 
procedures came from Russian phonology, I believe; the analysis was done by 
Halle, and I would guess that Chomsky simply took that over as a good 
example case.  Chomsky and Halle of course famously teamed up together.)

> However, as another set of important citations from Roy Harris' book 
> 'Saussure and his interpreters' (01) shows:
> 
> There seems to be no indication that Noam Chomsky, founder of modern 
> generative linguistics, had ever read or paid attention to the work of 
> Saussure until the appearance, in 1959, of the first English translation 
> of the Course de Linguistique Generale (Baskin 59).

I have no idea whether Chomsky knew about Saussure (although I can't 
imagine he didn't).  But arguing over Saussure's theory in the 1950s would 
be rather like arguing over the theory of SPE today.

In general, the European Structuralists were doing, I think, very different 
sorts of things from the American Structuralists, and so far as I know, 
Chomsky had very little to say about them, either.  Of course the 
phonological feature system that Halle developed (and which was used in 
SPE--Chomsky and Halle 1968) were developed out of European work.

BTW, there was (at least) one American phonologist who had already argued 
against the validity of Discovery Procedures, although he otherwise 
retained the concept of phoneme that the American Structuralists had.  And 
that linguist was Ken Pike.

 > ...
> So it looks that he was attacking behaviorism (which indeed cannot 
> provide generalizations), not structuralism (and therefore the 
> distributional methods which are based on structuralism)...

As I say, I think what he was doing was attacking the _American_ 
Structuralists, who (apart from Pike) were indeed wedded to behaviorism. 
He was not, perhaps, attacking Structuralism if you include both American 
and European Structuralists under the same umbrella.  (Although to go out 
on a limb, I'm not sure that the two camps had any similarities beyond the 
name.)

 > ...which he at
> that time did not really know and later simply failed to understand (or 
> to acknowledge) properly because he was trying to narrow language to a 
> purely generativist point of view...


Hmm.  I would say he knew American Structuralism very well.  And far from 
trying to _narrow_ language to a generativist view, I would say he was 
trying to _widen_ language to a generativist view.  That is, generativist 
theories were (and are) open to a much wider range of evidence than the 
American Structuralists ever were.  (For example, discovery procedures in 
phonology didn't allow reference to meaning, so phonologists couldn't 
"discover" morphophonemic alternations, even though it was clear they 
existed.)

Furthermore , the generativist view has always been that language learning 
is a matter of psychology (something real), whereas AS (sorry, I'm getting 
tired of writing it out) more or less treated linguistics as a way of 
cataloging data, and certainly not s.t. that might have some sort of 
psychological reality.

So on both counts, I think Chomsky's view of language was wider than that 
of the AS.

 > ...and to make prominent his
 > distinction between performance and competence...

I'm pretty sure this distinction didn't come out until later, in the early 
1960s.  The first clear discussion of it that I know about is in Aspects of 
the Theory of Syntax (1965), although I'm sure it was around before that 
(as was the ms of Aspects).  For all I know, maybe it was in Chomsky's PhD 
dissertation all along...but even if that is true, I don't think much was 
made of it in public debates until later.

> Of course this is also due to the fact that currently only very basic 
> notions such as the distinction between word classes can be extracted 
> fully automatic from raw text whereas generativists are currently 
> occupied with a lot more intricate questions as titles of talks of 
> recent conferences show. This might indicate that indeed, generality is 
> not much possible with such methods. But I would also say that the field 
> is approaching the moment where 'generality', e.g. grammars of a given 
> language can be extracted in a fully automatic way from raw text, 
> without any introspection. 

The issue for generative linguistics is not how the linguist does it 
(distributional methods or introspection or...), but how the child does it. 
  It still amazes me: hundreds (if not thousands) of smart linguists have 
been analyzing languages for decades, and we still can't come up with 
complete and accurate descriptions of these languages.  And more recently, 
computers have been analyzing huge corpora, and they can't come up with 
complete and accurate descriptions either.  Yet virtually every child, 
smart or not, and without exposure to a huge corpus, in a few years has a 
complete understanding of a language.  How is that possible?

-- 
	Mike Maxwell
	maxwell at ldc.upenn.edu
         (actually, I'm not at LDC any more, I'm at CASL;
	gotta change my subscription...)