etc.

[Tom Givon]

                          1-17-97

Dear FUNKfriends,

Now that the traffic has subsided somewhat, I want to take the opportunity
and tell you how valuable I thought the last burst of discussion--thanks to
Phil Bralich who, rather unintentially, I suspect, wound up starting it--
really. I saw it as a beautiful example of communal thinking I have always
thought we started FUNKNET just for that (rather than for book and conference
announcements, however useful those may be). So to me this discussion
demonstrated that FUNKNET can serve its intended purpose -- even if
it does it only once a year. The following comments are not intended thus
as grabbing the last word, but rathe as part of this progressive refinement
of our communal thinking.

I thought Matt Dryer and Liz Bates defined the two poles of our discussion
most succintly. What I would like to suggest here that the two poles of our
practice of linguistics -- theory and methodology -- are indeed intimately
connected.

Matt suggested two "theses" of our approach to structure:

     (a) STRONG:  "grammatical structure strongly correlates
                   to semantic and pragmatic functions"
     (b) WEAKER:  "grammatical structure exists"

It might perhaps be useful to point out that **logically** a belief
in A entails a belief in B. That is, if (a) is asserted, (b) must be
presupposed. But, at the same breath, (c) must also be presupposed:

     (c)  "semantic and pragmatic functions exist"

Edith Moravcsik's latest comments indeed pursued this logic: If you
believe in (a), then you must define **both** structure and function
independently of each other. That is, in my terms --by different methods.
Otherwise, all you are left with is *a tautology**.

On the methodological end, I think Liz Bates (and Lise Menn) expressed
our need for multiple methodology rather elegantly. But notice that, among
other reasons, the logic of (b) and (c) above being presupposed by our
strong belief in (a) already points at the need of multiple methodologies.
We obviously need methods that probe into structure QUA structure. And the
traditional -- Bloomfieldian, Chomskian -- methods of analyzing clause
structure and morphology come in handy precisely for this reason. Indeed,
I cannot imagine studying and describing the grammar of a new language I
work on **without** such methods. Have you tried recently to go **directly**
to studying discourse-pragmatic functions lately? And are your results
yielding form-function correlations?

For people like Fritz Newmeyer and Dave Pesetsky, whose contribution to
our discussion was truly valuable, the terrain might look like this
(and do forgive me for the hypothetical nature of (1)-(4) below):

(1) We certainly see some correlations of the (a) type;
(2) But, they are either sporadic or never 100%.
(3) Therefore, to be really rigorous and not go on a (frail) limb,
    we cannot abide by the strong assertion (a); we will therefore
    confine our investigation of syntax to what is obvious -- obvious
    from using **only** the traditional clause-level methodology.
(4) So, we will only describe structures, and develope and independent
    theory of syntactic structures.

Now, notice that the vast majority of communicative functions do not
reveal themselves, in any obvious/intuitive way, if you confine yourself
to the traditional methodology. i.e. to the study of isolated clauses
outside their discourse context. So much of the doubt expressed by Fritz
and David about the partiality and non-systematicity of form- function
pairing (i.e. Matt's principle (a)) must indeed be traced to their reluc-
tance to go beyond the traditional clause-level methodology.

This is in no way a **logical** necessity, but rather a pragmatic
methodological consequence. Just as you cannot get at structure without
the appropriate methods, so you cannot get at communicative-pragmatic
functions without the appropriate methothology; that is, without
studying what grammar does in actual communication.

What has always baffeled me, suppose--ever since reading and heartly
approving of Chomsky's (and Postal/Katz' and Fillmore') drift, between
1962 and 1965, to **semantically-relevant** (and thus more astract) deep
"syntactic" structure--is the seeming reluctance of generative linguists
to take the rather obvious next plunge. Propositional semantics was licensed
by Aspects (1965) as being strongly correlated to syntax, i.e. to "deep
structure". So why not take the obvious next plunge and admit that the
"stylistic transformations", those Joe Emonds characterized in his disser-
tation as "root transformations", are just as relevant to syntax (and syntax
relevant to them) as the "triggered" transformations (Joe's "structure-
preserving" transformations)?

In other words, if you've already opened the doors of syntax to semantics,
why don't you open it further to pragmatics?

Here I think is where, inadvertently, implicitly, methodology rears
its sweet head. If you don't practice the methodology of looking
for what syntactic structures do in communicative context, then
pragmatic function remains rather invisible to you. You sense its existence,
but it remains mysterious, unwieldy and highly suspec. You approach it with
the same an inborn skepticism that Bloomfield and Carnap and the Positivists
did, as "stylistic" intuition that cannot be captured **rigorously** by
science.

On reflection then, what we've got here is a fairly transparent case,
leastwise to me, of the methodological tail has continuing to wag the
theoretical dog.

With apologies for the long-windedness, TG