Some earlier discussions of this theme here have begun from very
_practical_ problems of research method and technique. Peter Smagorinsky
notes that the status of case studies (in many fields) is problematic from
methodological viewpoints that dismiss them as "N=1" research. In those
views, typicality derives only from frequency, as if instances had no
connections to larger systems of which they are both parts and instances,
and so typicality is merely a happenstance, to be determined by random
sampling. What is missing in such views is often a more substantial role
for theory than simply as a generator of hypotheses. Theory ought to tell
us what kinds can be, and what logical, material, and semiotic
relationships among these kinds are expected. If linguistics tells us in
what sense texts are instances of the system of a language or dialect, then
we can know how they are typical and look to them for evidence about the
system. So also in the famous anthropologist's quip to his sociologist
critic, that his account of funeral practices based on witnessing one
funeral was buttressed in its validity by the fact that none of the dozens
of indigenous participants 'was acting as if anything unusual was going on'
-- hence N was the number of witnesses, not the number of funerals!
Bill Barowy wondered if ANT includes abiotic factors, and in principle it
does; in fact the point on which it is most often criticized is its
insistence that the social and the natural cannot be disentangled and that
the relevant networks equally relate social and abiotic actants. The
difficulties perhaps come in deciding what is an actant, esp. when the
semantics of our natural languages bias us toward relatively isolable,
localized, causative agents ... relegating many other participants in
activities to the status of 'circumstantials' ... so hurricanes have human
names, but soils do not.
I think the practical issue, whether we talk of "units of analysis" or
"wholes in synthesis", is where we start in each investigation and how we
link what we have learned in one to another. If we begin from scratch and
define new units and new phenomena for each case, what kind of theory is
worth anything to us at all? Geertz has occasionally pressed this point in
relation to hermeneutic vs. "positivist" studies. Or if we study each case
in its own terms (as phenomenology recommends), how then do we connect one
case to another? We can say, if we like, that all 'larger scale systems'
are illusions, the results of our analytical will and desire, but we must
be cautious that we do not thereby privilege one scale, that of simple
human action and interaction (e.g. conversation) as the only real one. I
strongly believe that the study of most phenomena, natural and social,
makes more sense if we recognize that multiple scales can be posited, and
with them cross-scale relationships that provide a theoretically grounded
basis for relating not just events on different scales, but those on a
common scale, to one another. Just as it is too extreme to presume that
entities and units/wholes (atoms, persons, cultures, communities) can be
defined or known as Platonic ideals independent of our agendas and
practices, or independent of the particularities of each case, so it also
seems to me too extreme to imagine that each is so completely contextually
specific and sui generis as to have no grounded relations to any other.
Many of these issues, I think, are easier to talk about if we do free
ourselves up a bit from the semantics of entities and agents. The primary
alternative is a semantics of process, which implicates and defines agent
roles, but does not make them ontologically or discursively logically
prior. It also brings the bonus, over the semantics of relations, say, of
not privileging the static over the dynamic when looking for cross-instance
invariants. In this perspective, a unit of analysis can be itself a
process, or a system-in-process, a something-becoming rather than a
thing-being. It can indeed, as Bill imagined for a developing Jay, proceed
from being less to more well-defined in various respects, as well as
changing in those properties that are comparable from one Jay to another.
New Jay-properties come to be in the course of development, and they may
even do so very gradually (rather than all-at-once). But we are very far
from having an adequate logic of relative "vagueness" to deal with the
study of systems/processes on many extensional and durational scales which
develop new kinds of properties, not just in the sense of new for all
systems and all times, but even just in the sense of newly meaningfully
applicable to themselves. I am currently discussing such issues with a
group of developmental and evolutionary biologists interested in the study
of complex dynamical systems. JAY.
---------------------------
JAY L. LEMKE
PROFESSOR OF EDUCATION
CITY UNIVERSITY OF NEW YORK
JLLBC who-is-at CUNYVM.CUNY.EDU
<http://academic.brooklyn.cuny.edu/education/jlemke/index.htm>
---------------------------