Well, yes, we are all 'structuralists' or something, in the sense
that a theory is meant to be more abstract than anything it
claims as an instance of its generalizations. What we often do
not probe, however, is what _kind_ of generalization we are
making, that is: (1) what is the model, the meta-discursive
template, for our model or theory? and (2) what kinds of
relationship precisely do we posit between general statements and
particular 'instances'?
Gordon Wells mentions, in this same connection I think, the
relevance of genre theories. One could mention also inquiries
into the nature of Narrative as a mode of
abstraction/generalization/representation of experience, into
Flow-chart (if-then branching dynamic option models)
representations, Constituency structure models (whole-part
embeddings, with many levels), Paradigmatic models (static option
models with inter-option dependencies), Expository models
(discourse which taxonomizes processes and chains them by
abstract relations such as causality, sequence, etc.), etc. etc.
Genre analysis, or Foucault's program of discourse formation
analysis, are directed at still more specific variants of such
organizing strategies (all of which are semiotic, most
discursive-linguistic, all culturally and historically specific
and historically antecedented, etc.)
Michael Halliday has noted that the grammars of natural languages
present us with models of activity, experience, process, and
relation that are probably inescapable (once they become
unconscious mediators of the rest of our thinking; ala Whorf),
and also represent one of, if not the most successful kinds of
such models in human history. His work can be read as an
explication of what the grammar of English has to say about the
structure of experience, activity, reality. This sort of inquiry
is also, I would say, a theoretical prerequisite to any
particular kind of modeling or representational activity, because
it will deeply influence it, and such influences need to be
explicit for the theorist. (Naturally any interpretation of the
grammar, any 'grammatics', must also be made in some historically
and culturally specific set of discourse formations.)
Language represents a critical case I think for notions of
'scripts' or AI approaches to modeling or generating real-world
human activity. While it is probably true that we bring certain
expectation templates to our experiencing, including our
production and interpretation of sentences, utterances, 'texts',
and that some of these have static forms, and that some of these
static forms are structured like 'scripts' (and others like many
of the other model-types above), we all know that human activity
has a dynamic, unpredictable, (Keith's 'improvisational') aspect
not captured by most of these sorts of models. This is doubtless
because our linguistic and other activity is epiphenomenal to our
dynamics as subsystems of larger self-organizing systems and
networks of interactivity. There are, so far as I know, no
generative or general models of such systems; there are at best
only limited simulations of them. They probably cannot be
'represented' in any sense we are accustomed to. This does not
mean that the 'texts' they generate over any finite period cannot
be modeled statically, or that there are no statistical
regularities in behavior even over longish intervals. But such
systems are profoundly _individuals_. What matters to us about
such systems are the phenomena which are unique to them as
individuals. It is these linked aspects of being open-endedly
dynamic and individual which, for me, accounts for the unclosable
gap between general representational models such as scripts,
genres, etc. and the detailed trajectories of real systems. JAY.
--------------
JAY LEMKE.
City University of New York.
BITNET: JLLBC who-is-at CUNYVM
INTERNET: JLLBC who-is-at CUNYVM.CUNY.EDU