Table of Contents |
|
Back to Part 1 of this article V.
Problems in Topic Domain Methodology |
6) Schneirla is the only one with
non-zero entries for all 7 categories, which suggests, for his work, a
high
degree of integration into the literature. Skinner, Beach, and
Bitterman each
crossed 5 categories, while Mackintosh is specialized to 3.
Because I am not familiar with the
work of the group I cannot personally evaluate these findings
concerning the
relationship of these five men. I invite the reader familiar with the
work of
these writers and their reputation to judge the outcome. Are Technical
Innovations and Consciousness indeed areas of low contribution by these
men? Is
it reasonable to say that all of them are heavy contributors in
Theoretical
Contributions, and that they share a solid interest in Behavioral
Engineering?
Is it fair to say that Mackintosh and Bitterman, unlike the others,
specialize
in Discrimination, while Beach unlike any of the others specializes in
Central
Nervous System? And is it true that Schneirla is the only specialist in
Observed Behavior? It would of course be possible to check these
comparisons
systematically against the judgments of knowledgeable people.
An
immediate issue to be considered is the psychological nature of the
phenomenon
of topic domain classification. Have I uncovered a hidden, talent I had
in
historical prophesy...or have I more practically stumbled upon a
heretofore
unsuspected relationship between common behavioral skills (e.g. making
up
titles, summarizing what one has talked about ant referring to parts of
it---see my (unpublished) lecture notes on ethnosemantics)? These
common
everyday discourse skills are the same that the scientist uses in
writing his
text and titling it. That titling and indexing of technical materials
do not
need extensive training in the field is demonstrated by the fact that
librarians and publishers have always been the chief producers of
indices and
abstracts that give modern science its basic structural
interrelatedness. There
is thus the possibility that historical processes in the development of
organized knowledge can be investigated through topic domain
methodology.
(ii) Special instructions attempting
to explain the nature of the categorizing task; see questionnaire in
Appendix 2
for the specific way in which I attempted this. Table 2
gives
the results. Shown are percentage scores for each article which were
computed
as means for the responses of 22 students. I used the following
rationale for
the quantification of the classifications I assigned a score of 1
whenever an
article was assigned to a category; and a score of O whenever a
category was
not used for a particular article. In this manner I obtained a mean
score
(n=22) for each topic domain category as applied against each of the 48
articles.
Having consulted several reputed wise
men in statistics; I was so hopelessly confused that I ended up
devising my
own, for better or for worse. (At the same time vowing to pursue this
in the
future). I argued that I could obtain an indication of the efficiency
value of
the topic domain methodology by counting the number of agreements of
subject
classifications and the independently selected articles, which I had
done to
equalize the distribution (see above). My selection yielded the
following
number of Bitterman's articles in each of the five categories
respectively: 18,
18, 15, 15, 9. In Table 2,
I
have marked by an asterisk the independent selection manipulation for
each
article. Because I used a procedure of multiple entries (see discussion
below)
two criteria of validation are possible. The first which we might call
the
looser criterion, consists of computing the percentage agreement of the
subject's responses with my independent selections. As documented in
Table 2,
there were 45 articles which received a mean classification score for
the
independently chosen topic domain that was higher than the mean
classification
score for the other topic domains. This is a corroboration of 94
percent and in
the majority of cases the margin is quite large as can be seen in Table 2.
I am
quite sure that probability theory would show this result to be
extremely
unlikely through chance factors.
What are some
effects of the single versus multiple classification treatment of topic
domain
methodology?
This is an interesting empirical issue
that relates to theoretical mechanisms one could examine to account for
the
relationship between common discourse skills and the historical process
of
topic domain development. It certainly deserves special study. For the
moment
however, I can glean some indication by contrasting the 94 percent
corroberation figure of the looser criterion involving single
classification to
a more stringent criteria which would require that multiple entry
selections
from the independent treatment be fully replicated by the judgments. If
this is
done in Table
2 I obtain 34 of the 48 articles which is a 71 percent
corroboration rate.
I would guess that the difference between the 94 and the 71 figures for
the
looser and tighter criteria respectively, is significant. If so, the
findings
would suggest that the structural parameters of topic domains are
dynamically
affected by the common practice in indexing and cataloguing of multiple
entries
called cross-referencing. My prediction would be that the integrative
function
of cross-referencing produces divergent topic domain networks not
unlike what I
understand by the notion of dispersion in evolutionary
cross-fertilization. In
my grand conception I envisage topic domain methodology as a biological
phenomenon. This notion comes to me from my studies of ethnosemantics
(see
Jakobovits and Gordon 1976). In other words, the commonplace everyday
behavioral manifestation of what may be called "the topicalization
work" of a speech community, whether spanning the centuries through the
archives or contemporary and personal, proceeds according to biological
laws
that are describable through mathematical models. I shall illustrate
how the
data in Table
2 lends itself to the production of theoretical hypotheses
concerning the
historical process in the scientific literature, viz. the
topicalization work
that is represented by a bibliographical profile such as my
instructor,
Dr. M. E. Bitterman.
Can Dr. Bitterman's
topicalization work in his behavior of giving a title to his published
works be
characterized in terms of the integrative function they have of tying
in his
work into the scientific literature?
As indicated earlier, what is called
the body of scientific literature can be usefully conceptualized as
being
composed of interrelated fragments which I refer to as topic domains. I
have
shown the way in which the internal organization of topic domain can be
measured or quantified. I argued for the central role of titling in
this
historically dynamic process. I have presented an empirical methodology
for the
study of how this important integrative function of titling affects
readers or
consumers of the works in their categorization behavior. This behavior
is not a
special one but overlaps with the common everyday use of discourse. In
this
manner it is possible to view and understand history, the topical
history of a
discipline, as being produced by common everyday discourse mechanisms.
The data in Table 1
documents the role of topic domains in the production of what might be
called
"overlapping interests of psychologists" in this case a group of
contemporary workers in the general field of Learning. The amount of
dispersion
in subject's categorization behaviors reflects boundary areas between
small
sub-groups of scientists we might refer to as "cells". Thus, the five
psychologists analyzed in Table 1 represent a cell. Undoubtedly cells
touch
each other with some key figures straddling two topic domains (perhaps
we could
call them the "cross-fertilizers"). These dynamic processes of
historical reconstruction are exciting possibilities opened up for
future
studies inclined towards the practical scientific theory of science
making or
science makers--surely a worthwhile subject matter for the behavioral
sciences.3 To illustrate some possible uses of data in the form given
in Table
2, one
may note that the patterns of distributions indicate for each article a
profile
type.
How can article
clusters arranged by profile type indicate a degree of integration into
the
topic domains in the literature?
Profile types may be characterized in
part by their modal distributions across the 5 categories. Articles
with unimodal
distributions i.e. viewed by the judges as belonging in only one
category (see
profile type A below), indicate that an article is specialized to a
single
topic domain. Bimodal or trimodal distributions show
dense
clusters in multiple topic domains, which indicates a higher degree of
integration into the topic domains in the literature. Multiple
classification
profiles may perhaps be thought of as the "cross-fertilization
divergents". Another feature of profile types is their arrangement in
clusters by category types within the modal framework. For example,
articles
which characterize Profile type A, (17, 18, 33, 34, 37, 38, 43,
44),
have a unimodal distribution, indicating specialization to a single
topic
domain. Topic domain IV, Technical Innovations, has the largest number
(8) of
articles with unimodal profiles. It also may be noted that, for this
category,
scores in the four remaining topic domains were judged far below the
one third,
cut off point, which establishes a prominent topic domain for an
article. A
second profile type may be gleaned which I will call profile type
B.
Articles which characterize profile type B (7, 8, 28, 32, 42, 46) have
a
trimodal distribution indicating a high degree of topical divergence
(cross-fertilization). These articles straddle topic domains I
Theoretical
Contributions, II Behavioral Engineering, and III Discrimination.
Judges
avoided the two remaining categories uniformly.
Bitterman's validation of the profiles
is a crucial test of the validity of classification by title. I invite
Bitterman to corroborate these rather surface analyses (I intend to do
a more refined
analysis such as cluster analysis in pursuing this). This of course
applies not
just to Bitterman's personal reactions and evaluations of this report,
but to
his reputation and Are opinion of his peers. Who, like the judges of my
experiment, use topic domain categories wherein to assign Bitterman's
works, as
well as the works of others. Without this validation the empirical
methodology
that I suggest would be suspect.
What further
studies can be proposed to illustrate the validity of topic domain
classification?
I have urged that historical processes
in the development of organized knowledge can be investigated through
topic
domain methodology. I have piloted one such investigation involving the
classification of titles by students specialized to the general field
of
psychology into topic domains derived from Bitterman's bibliography. A
second
validation check would involve having a sample of the same population
of
students ant faculty members read a sample of Bitterman's articles in
full and
then classify the articles. A third level validation could involve an
informal
survey of people familiar with the area "comparative learning" in
which they would classify Bitterman's work by their knowledge of his
reputation.
A validation check which provides an
academic spectrum would contrast other groups (historically) distant
from
psychology in varying degrees, to some pre-established criterion on the
sample,
then compare their distance scores with their classification scores.
This study
would yield a statistical measure of the degree to which each group is
standardized or assimilated into the topics in the area of interest,
psychology
broadly, and comparative learning specifically as a discipline in the
field.
There are many other possibilities to be investigated to demonstrate
the
usefulness and validity of topic domain methodology.
Topic
domain exists as an integrating mechanism of the readers, a sample of
which I
used here called judges. The study has revealed a basic mechanism of
understanding a discipline, or the way the regulars of a
discipline
understand their own activities. My approach views topic domain as a
natural
and biological phenomenon, rather than a mental one. Topic domain is
given,
evolutionary, and dynamic.
The basic methodological issue of the
approach I'm proposing is; when does the reader have sufficient
standardized
background in a discipline so that the reader's template of topic
domains,
through which text is processed, is sufficiently activated by
bibliographical
citations informations alone. Another way of asking this question is
how much
and which parts of the text of an article needs to be looked at by the
reader
before the topic domain integration is sufficiently activated by the
text the
reader exposes himself to: What's in a title? What's in a reference?
What's in
the introduction? Etc. Clearly, there are relationships here to be
studied
empirically in terms of the basic mechanisms, structures, and functions
of this
living monster called the scientific literature. Not to be overlooked
are the
implications for training scientists in this kind of empirical work in
topicalization within a discipline. Since, obviously the training
program must
succeed in transmitting the standardized topic domain grids in
sufficient depth
and breadth.
Operationally defined boundaries of
topic domains is the task of classifying, or the operation of
"readership", the text into what it counts for i.e. what it
implies for what. Science involves reading text and putting it
somewhere,
as a function of the science-making process. Topic domain methodology
simulates
this normal process by having the judges formally classify the articles
into
the categories. Topic domain is carried in the titles and allows
mapping the
coherence of fields, disciplines, and arguments.
After all, once a work is published it
starts its own life within topic domain space, and neither the author
himself
nor the readers of a particular decade can predict its life course.
Thus, the
notion of "readership" obviously assumes localized proportions in
that both time (in decades and centuries) and cultural area (countries
and
specialties) are determining factors in how text will be integrated
into the
contemporary topic domain template or network. Out of this notion one
can
elaborate the empirical meaning of "schools" and systems within
disciplines, since the latter involve the tracing of evolutionary
changes in
topic domain networks within the living literature space. The function
of
scientific history then will be to trace the history of topic as
carried in the
titles of Publication units.
1. Nor is it the case
that the empirical approach I adopted here is less of a chore than
other things
I might have done, since this report took approximately 450 hours to
complete
to the present stage, and this not counting additional and extensive
raw data
not presented here by which suggest the empirical richness and promise
of such
an approach.
2. I wish to thank Sue Austin for
suggesting directions and for providing some of the materials.
3. I am planning graduate study which
would allow me to pursue the study of the history of topic by going
into
ethnosemantics, library science, psychology, and history.
4. This does not imply that titles
should be read as a substitute for
the content of the text since the
issue is rather, what information is carried
in the title itself.
Beach, Frank A. The
Snark was a Boojum. in T. F. McGill, Readings in Animal Behavior.,
1965,
3-15.
Bitterman, M. E. Flavor aversion studies.
Science,
192, 1976, 266-267.
Bitterman, M. E. Incentive contrast in
honeybees. Science., 192, 1976, 380-382.
Bitterman, M. E. Intradimensional
versus extradimensional transfer in the discriminative learning of
goldfish and
pigeons. Learn. Motiv., 4, 1976 197-203.
Bitterman, M. E. Lecture notes for
History of Psychology, 423, Fall 1976, University of Hawaii.
Heidbreder, Edith. Seven
Psychologies. Englewood Cliffs, N.J.: PrenticeHall Inc., 1933.
James, Leon. Notes on ethnosemantics.
(Unpublished manuscript), 1975, TEC, University of Hawaii.
James, Leon. Lecture notes for Seminar
on William James, psy. 705H, Fall 1976, University of Hawaii.
James, William. The Principles of
Psychology., New York: Dover Publications, Inc., 1890.
Miley, A. D. Structural dimensions of
4 behavior therapy journals. Ph.D. dissertation, 1975, University of
Hawaii.
The American Psychological
Association, APA Thesaurus. Washington D.C.: The American
Psychological
Association, Inc., 1974.
The American Psychological
Association, Psychological Abstracts., Washington D.C.: The
American
Psychological Association. Inc.
Xhignesse, L. V. and Osgood, C. E.
Biographical citation characteristics of the psychological journal
network in
1950 and in 1960. Amer. Psychologist., 22, 1967, 778-791.
Ziman, J. M. Public knowledge: An
essae concerning the social dimensions of science., Cambridge,
England:
Cambridge University Press, 1968.
Deese, James and
Hulse, S. H. The Psychology of Learning., New York: McGraw-Hill
Book
Company, Third Ed., 1967.
Herrnstein, R. J. and Boring,.E. G.
(eds.) A Source Book in the History of Psychology., Cambridge,
Mass.:
Harvard University Press, 1965.
Rychlak, J.: Introduction to
Personality and Psychotherapy., Boston, Mass.: Houghton Mifflin
Co., 1973.
Wertheimer, Michael, Fundamental
Issues in Psychology., New York: Holt, Rinehart and Winston, 1972.
Wilkening, H. E. The Psychology
Almanac., Monterey, Calif.: Brooks/Cole Publishing Co., 1973.