I generally agree with him that our unit of analysis needs to be
both interpersonal and activity-specific, and hence so will our
model of a communicative activity be.
My analysis in a previous posting, based on earlier work in this
area, used a model in which there were different 'frames' or
productive and interpretive thematic-semantic formations
(patterns, expected regularities) hypothesized for student and
teacher. It should be noted though that the basic notion of a
thematic formation, or frame (which is not my usual term, just
convenient here), is an _intertextual_ one: patterns that recur
from text to text, and not specifically tied to individuals as
units of analysis.
As Eugene notes, the analyst constructs the model of the frame(s)
or patterns in use. If we have dialogue data, we will find that
some elements of the frame(s) seem to be shared (up to a point,
at some level of abstraction, not necessarily in any absolute
sense), and some elements seem to be in use in T's discourse, or
S's, only. It is usually not possible in practice to fully model
any of the following: the patterns of the T-S dialogue; the
separate utterances of T; those of S. This is because generally
the patterns also depend on other, unheard texts (intertexts),
such as: yesterday's lesson, a paragraph in the textbook, what a
friend told me, etc. The methods of analysis that I use often
reveal the existence of a gap, but not how to fill it (though
good guesses can often be made and later confirmed).
In practice, with this kind of data, it often does make sense to
model the interpersonal phenomenon, the T-S dialogue as a whole.
Especially in cases where there is a lot of overlap in frames, or
where, as in IRF triadic dialogue patterns, a complete semantic
unit is jointly constructed by T and S (but not completely
constructed by either's utterances alone). Note that this can
include proleptic phenomena, or not, and that it can also produce
an illusory sense of agreement that would _not_ be confirmed by
comparing extended T- and S-monologues on the same topic.
It also happen however that the same 'slot' in the semantic
pattern is filled by two thematically inequivalent elements, and
these must be indexed by speaker in order to model the data in
detail. It is precisely this kind of data that _does_ motivate
'individuals' as units of analysis. But it remains, as Eugene
notes, still activity-specific. There is no necessity that the
'same' speaker will use the same patterns in a different
activity. If s/he does, this is part of our grounds for defining
a notion of social person across activities. Often, s/he does
not. We need to recognize in general that the trans-activity
grounds for individuals as units of analysis may not be the same
as the intra-activity grounds. We need to set aside our cultural
ideology's insistence that individuals are universal, 'real'
units, if we are to understand the role of our constructed units
at various points in our modeling efforts. (This is more or less
the theme of chap 5 in _Textual Politics_).
Epistemologically, too, I agree with Eugene's proposal. These
matters fit very well with the quantum 'complementarity'
principle of Bohr's interpretation of Heisenberg's
_Unbestimmtheitsprinzip_, as I discussed on Bohr's own home
grounds in my Denmark paper a few months ago. The existence,
identity, and characteristics of units in interaction are always
functions of the specific conditions of the interaction-
situation. There are no universal descriptions valid across all
conditions, at least not ones specific enough to be mapped onto
observations. As always, our ingenuity can fashion abstract
representations to cover all cases, but then we cannot know how
to 'raise' these to the level of the 'concrete' specifics of
given situations without putting in additional, situation-
specific information, 'by-hand', which is not given by the
general theory. All generalizations are really of this kind; the
quantum case simply represents an extreme that forces us to give
up the traditional hand-waving (and ideological certainties) that
obscured the problem.
JAY.
--------------
JAY LEMKE.
City University of New York.
BITNET: JLLBC who-is-at CUNYVM
INTERNET: JLLBC who-is-at CUNYVM.CUNY.EDU