LINGUIST List 12.1796

Thu Jul 12 2001

Disc: On Nonobjects of Syntactic Study

Editor for this issue: Lydia Grebenyova <lydialinguistlist.org>


Directory

  1. Jose-Luis Mendivil Giro, Re: 12.1785, Disc: On Nonobjects of Syntactic Study
  2. Chris Johnson, Re: 12.1785, Disc: On Nonobjects of Syntactic Study

Message 1: Re: 12.1785, Disc: On Nonobjects of Syntactic Study

Date: Wed, 11 Jul 2001 22:20:51 +0200
From: Jose-Luis Mendivil Giro <jlmendiposta.unizar.es>
Subject: Re: 12.1785, Disc: On Nonobjects of Syntactic Study

>From: "Dan Everett" <Dan.Everettman.ac.uk>
>
>I hope that what follows below in this posting will provoke discussion on
>objects of study in linguistics. I am posting to both
>LINGUIST List and FUNKNET, so I apologize for multiple receptions of the
>letter. This is a very condensed form of a thesis I am currently developing
>in book form, so feedback would be personally useful, in addition to what
>I believe the benefits of such a discussion would be for linguists more
>generally.
>
>The basic thesis is that in a Chomskyan/Cartesian linguistics there is in
>principle no object of study. Alternatively, there is in-principle no
>way at getting at that object, however clear it may sound conceptually.


>Here goes: Chomsky claims that the object of study in syntax is I-language
>or, to use an older term, speaker competence. What is this supposed to be?
>It is an internal *grammar* (not language - whether of the 'I-' or 'E-'
>variety - of any type widely accepted in Linguistics). Such a grammar is
>necessarily a Cartesian construct based on assumptions about the mind, e.g.
>that there is a mind and that it is inside the head (instead of, for
>example, between members of a society).

Nope.
The object of study of what you have inadequately called Chomskyan/Cartesian 
linguistics is a ***natural object***: i.e., a state, a property of the human 
mind/brain.

Here is what Chomsky actually says, in an appropriate context (To make things 
easy I quote only from Chomsky's recent 'New Horizons in the Study of Language
and Mind' , Cambridge University Press, 2000):

"The faculty of language can be regarded as a 'language organ' in the sense in
which scientists speak of the visual system, or immune system, or circulatory 
system, as organs of the body" (...) "We assume further that the language 
organ is like others in that its basic character is an expression of the 
genes. How that happens remains a distant prospect for inquiry, but we can 
investigate the genetically-determined 'initial state' of the language faculty
in other ways. Evidently, each language is the result of the interplay of two 
factors: the initial state and the course of experience. We can think of the 
initial state as a 'language acquisition device' that takes experience as 
'input' and gives the language as an 'output' - an 'output' that is internally
 represented in the mind/brain" (p. 4).

> What could count as evidence for
>this Cartesian construct/grammar? All and only phenomena which have no
>nongrammatical explanation.

No. As Chomsky goes on saying, both the 'input' and the 'output' are open to 
examination:

"we can study the course of experience [the process of language acquisition, 
JLM] and the properties of the languages that are acquired. What is learned in
this way can tell us quite a lot about the initial state that mediates between
them" (p. 4).

Then, there is no limit to the different types of data that may be useful in 
such inquiry (grammatical, psychological, neurological, etc.). The only 
condition being falsability.

>What sorts of phenomena will have this property?
>Just those linguistic-like phenomena with no explanation in terms of history,
>function, sociolinguistics, phonetics, semantics, culture, sex, baldness,
>etc. (this list is ultimately 'everything but grammar'). How do we recognize
>which phenomena are grammar-only in this sense? We do not. We have not.
>We will not. We cannot.

I agree. We cannot know what is pure-grammatical a priori, but the notion of 
'pure grammar' does not appear in Chomsky's writings, to my knowledge.

>And the problem of recognition here is not merely hard. It is in-principle
>impossible. This is because to know that this or that fact is 'pure grammar',
>uncontaminated by nongrammatical factors, would require knowledge of
>everything about that fact, i.e. just everything.

Again I agree, but it is not the matter.
You are constructing your thesis on the assumption that the object of study 
of Chomkyan linguistics (let's dispense with Descartes by now) is an ideal 
pure grammar, but it isn't.

Let's see again what 'the Master' says about I-language and grammar:

"The approach I have been outlining is concerned with the faculty of language:
its initial state, and the states it assumes. Suppose Peter's language organ 
is in state L. We can think of L as Peter's 'internalized language'" (p. 5).

When Chomsky uses 'I' he is meaning internal, individual, and intensional 
(the characterisation of a function in intension), not 'purely grammatical' 
nor 'uncontamined by nongrammatical factors':

"When I speak of language here, that is what I mean. So understood, a 
language is something like 'the way we speak and understand', one traditional 
conception of grammar. Adapting a traditional term to a new framework, we can 
call the theory of Peter's language the 'grammar' of his language. 
Peter's language determines an infinite array of expressions, each with its 
sound and meaning. In technical terms, Peter's language generates the 
expressions of his language. The theory of this language is therefore called 
a generative grammar" (p. 5).

So, the object of study is not 'grammar' but some system of knowledge that 
uncontroversially lives in our minds, in our brains. A 'grammar' is a theory 
(idealised as all theories about something in nature) of that state of the 
mind/brain.

>Therefore, there is not, nor could there be, an object of study for an
>Cartesian-Chomskyan research program. There are only aspects of study
>(hence the appropriateness of the title of a certain syntax book from 1965).
>What could syntacticians study, then, if not a Cartesian or mental grammar?

As I have shown a 'mental grammar' may be considered a natural object and 
then a licit object for scientific inquiry. But this is only true if we use 
the word 'mental' in a proper sense (i.e. a non-dualistic, non-metaphysical 
sense).

As Chomsky again puts it:

"The approach is mentalistic, but in what should be an uncontroversial sense. 
It is concerned with 'mental aspects of the world', which stand alongside its 
mechanical, chemical, optical, and other aspects. It undertakes to study a 
real object in the natural world -the brain, its states, and its functions - 
and thus to move the study of mind towards eventual integration with the 
biological sciences" (p.6).

What Chomsky is saying here is that the adjective 'mental' must be used just 
as we use 'chemical', 'optical', etc. When we say that some events or 
processes, etc. are chemical or electrical we are not including a metaphysical
division. We are just grouping various aspects of the world for the purpose of
investigation. Chomsky uses the terms 'mental' and 'mind' in the same way:

"By 'mind', I mean the mental aspects of the world, with no concern for 
defining the notion more closely and no expectation that we will find some 
interesting kind of unity or boundaries, any more than elsewhere; no one 
cares to sharpen the boundaries of 'the chemical'" (p. 75).

>That answer is easy: whatever we find useful to study. Ergo, the guiding
>principles for linguistic theory are more likely to be found in Pragmatism
>(James, Peirce, Dewey, CI Lewis, Rorty, Quine, Putnam, Wittgenstein), not in
>Cartesianism, especially as developed in Chomskyan linguistics.

Non sequitur. The premise is fine, but the conclusion is untenable. All the 
great thinkers you have mentioned have been interested in the use of language,
in the ways language relates to society, truth, logic, and the world, but they
have not said a word about the structure of language (i.e., about the 
properties of the mind/brain that make it not only possible, but just like 
it is).

Jose-Luis Mendivil
Universidad de Zaragoza
Spain
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue

Message 2: Re: 12.1785, Disc: On Nonobjects of Syntactic Study

Date: Wed, 11 Jul 2001 23:02:24 -0400
From: Chris Johnson <chrajohnindiana.edu>
Subject: Re: 12.1785, Disc: On Nonobjects of Syntactic Study

At 2:25 +0000 7/11/01, Dan Everett wrote:

>Here goes: Chomsky claims that the object of study in syntax is I-language
>or, to use an older term, speaker competence. What is this supposed to be?
>It is an internal *grammar* (not language - whether of the 'I-' or 'E-'
>variety - of any type widely accepted in Linguistics). Such a grammar is
>necessarily a Cartesian construct based on assumptions about the mind, e.g.
>that there is a mind and that it is inside the head (instead of, for
>example, between members of a society). What could count as evidence for
>this Cartesian construct/grammar? All and only phenomena which have no
>nongrammatical explanation. What sorts of phenomena will have this property?
>Just those linguistic-like phenomena with no explanation in terms of history,
>function, sociolinguistics, phonetics, semantics, culture, sex, baldness,
>etc. (this list is ultimately 'everything but grammar'). How do we recognize
>which phenomena are grammar-only in this sense? We do not. We have not.
>We will not. We cannot.


This is an extremely interesting issue. I've been struggling a bit 
with related issues myself lately, so I've got some (muddled) ideas. 
Basically, I guess I see language as a social thing, but to 
understand how people participate in this social institution we have 
to understand how social convention is reflected in the cognitive 
abilities, systems and behaviors of the individual speaker. 
Linguistic competence is a kind of bridge between social conventions 
(existing out there in some nebulous, abstract social space) and 
what's in the head.

I just read Howard Wettstein's book "Has Semantics Rested on a 
Mistake? and Other Essays" (1991, Stanford University Press). The 
basic point of the last few essays is the contrast the Fregean 
approach to linguistics semantics, where language is considered 
thought externalized, from a more Wittgensteinian approach to 
language as a set of institutionalized conventions, as a social 
practice.

Imagine a Martian showed up at your doorstep speaking English. This 
Martian has a radically different sort of brain-like thing than we 
do. Nevertheless, it's able to use language in exactly the same way 
that we do: it would be able to tell that "John saw the badger" is 
grammatical but "John badger saw the" is not, it could (in principle, 
given enough knowledge about the world) recognize when "John saw the 
badger was true and when it was false, it would know that if "Every 
man saw the badger" is true then so is "John saw the badger", it 
would know whether it was appropriate in a particular context to say 
"a badger" or "the badger" to communicate about a specific 
individual, etc.

Now, due to its radically different mental functioning, the Martian's 
ability to use the English language is implemented in a very 
different way than yours. It uses a different sort of parser, it 
chooses between two allophones by consciously following a set of 
if-then rules rather than by using whatever possibly 
constraint-based, non-conscious phonological mechanisms we uses, etc. 
Despite these difference, we would be inclined to say that the 
Martian was competent user of English.

For a more real world example, think of speakers of English who 
acquired it as a second language as adults. It's generally accepted 
by linguists that there are major differences between the way the 
mind of a native speaker comprehends and produces utterances in a 
language and the way the mind of a second language speaker does. Yet 
if the second language speaker's usage conforms enough to that of 
native speakers, we say that that speaker is a competent speaker of 
the language.

The Chomskyan competence-performance distinction is basically the 
same as the mathematical distinction between a function and an 
algorithm.*

Consider the function n^2 ("n squared"). The function itself is 
simply an infinite set of ordered pairs: {<1,1>, <2,4>, <3,9>, 
<4,16>,...}. There are lots of algorithms that compute this 
function. For instance, "multiply n by itself" or "add up the first n 
odd numbers".

One might say that both the human English speaker and the Martian 
English speaker are able to compute the same linguistic function, but 
they do it with radically different algorithms. So, one idea I 
briefly entertained was that Chomky's "competence models" were simply 
accounts of the social conventions that made up a particular 
language. That doesn't actually work, however. This 
function/algorithm view doesn't exactly capture the relationship 
between a language as a social institution and a speaker's ability to 
use that language.

The reason has to do with the sorts of arguments proponents of the 
"new theory of reference" like Putnam and Kripke came up with. Take 
Putnam's "elm" vs. "beech" example. Presumably, social convention 
maps the word "elm" to one sort of tree and "beech" to a different 
sort. A sentence "That is an elm" is true if the indicated tree 
really is of the type the speech community associates with "elm". 
The individual speaker, however, doesn't have to know anything at all 
about the type of tree "elm" is mapped to in order to use the word 
competently. He/she might not be able to distinguish elms from 
beeches. Similarly, one doesn't have to know which guy linguistic 
convention associates with "Aristotle" in order to competently use 
and understand the name. So it seems to me that linguistic 
competence is a weaker notion than social convention.

Thus, it seems to me that there are three sorts of accounts of language here:

(1) an account of linguistic convention (e.g., a set of mappings like 
{..., <"Aristotle", the actual flesh and blood guy in ancient 
Greece>,...})

(2) an account of linguistic competence (e.g., Maybe a function that 
maps "Aristotle" to the descriptive meaning "the individual which the 
speaker's speech community calls 'Aristotle'"... I don't particularly 
like this formulation. Maybe the competence function maps the word 
"Aristotle" to a rule associating conditions with objects in the 
world: {<community calls OBJ_123 "Aristotle, OBJ_123>, <community 
calls OBJ_7 "Aristotle", OBJ_7>,...}.)

(3) an account of linguistic performance/processing (either in one 
particular speaker or generally in human beings) (e.g., maybe 
something *very* roughly like "When the person hears sounds the 
phonological recognizer recognizes as /eirIstOtl/, the mind searches 
the lexicon for that phonological string; if it finds it, the 
associated representation of an individual is activated. If it 
doesn't a new representation of a discourse referent is created and 
attached to this phonological representation, as well as some default 
assumptions about animacy and humanness.... etc., etc.)

(1) is essentially E-Language, while (2) and (3) are the competence 
and performance aspects of a study of I-Language.

There are obviously complex relationships between these different 
levels. Linguistic competence is something like the rules and 
conditions an individual speaker has to conform to for their 
linguistic practice to count as part of the social institution we 
call the English language. Linguistic performance is related to 
linguistic competence as an algorithm is to a function. Linguistic 
convention is strongly influenced by our cognitive machinery, since 
most of the members of a speech community have essentially the same 
sorts of perceptual categories, conceptual systems, language specific 
abilities, etc.

Thus, one object of study for linguists is something like the way a 
social institution (a language) is reflected in the head of a native 
speaker - both what a speaker has to know (or how they have to 
behave, if you don't like describing linguistic competence as 
"knowing that") to participate in the social practice of that 
language (competence) and how that speaker's brain actually performs 
the various tasks involved in language use.

* Actually, things are a bit more complex. If a function is an 
infinite, as n^2 and the function mapping sentences to their truth 
conditions are, it can't be specified by simply listing the set of 
ordered pairs. Thus, one way of specifying a function is to use an 
algorithm that computes it. Many generativists and truth-conditional 
semanticists hope that the algorithms they use to specify the 
linguistic competence functions they're trying to account for will 
also suggest something about what a processing model of language 
would look like. (Example: Many supporters of Discourse 
Representation Theory see the intermediate representations in their 
theory, discourse representation structures, as being in some way 
related to the mental representations we use while 
comprehending/producing discourse. One can try to flesh this out in 
a bunch of ways.)
- 
Chris Johnson 	 chrajohnindiana.edu
Mail to author|Respond to list|Read more issues|LINGUIST home page|Top of issue