harry lerner wrote:
> I'm a newcomer to this list and I thought I would put this out there for
> some feedback.
Hello Harry, and greetings from Calgary. It's nice to see an increase
in the "Canadian content"!
> Dunnell writes on pg. 51:
>
> "What constitutes analysis and what constitutes classification can be
> defined only in the context of a [given] classification. Analysis (etic)
> and classification (emic) are relative in a general context" (in square
> brackets added).
This must be the greatest dilemma facing anyone who would try to
interpret material culture in any meaningful way. The definition of
any class is usually arbitrary, it is an opinion that is based upon
the observation of an incomplete, and sometimes not even
representational record. Additionally, the details selected for
definition owe more to convenience in the present than in the
reflection of the intent of the original creators of the object in the
past. I speak of this from the viewpoint of one that has spent many
years wrestling with the problems of classification, and one that has
accomplished a major reclassification of a Celtic coinage
(Coriosolite).
> He further states on pg. 56:
>
> "Quantification articulates with classification in using, not defining,
> classes."
>
Again, this is very true. If the sets do not reflect original intent,
then any subsequent analysis will be flawed, regardless of how
carefully one conducts their statistical analyses. This can be
demonstrated with Colbert de Beaulieu's classification of Coriosolite
coins. He said that no distribution pattern was evident in the hoards
of these coins. It was a circular argument. His classification system
was flawed in that his divisions were arbitrary, based on too few
observations, and a mistaken idea of what defined an issue. He also
confounded factors of time and geography, effectively obfuscating the
true pattern. My reclassification of the coins corrected all of this
and a very clear distribution pattern emerged. See for yourself at:
http://www.writer2001.com/hoards.htm
> The second quote places quantification firmly in the phenomenological
> (emic) realm according to Dunnell. If classification is emic in nature
> relative to analysis which is etic in nature, as Dunnell suggests in the
> first quote, then could not quantification be a logical basis for defining
> classes? Perhaps not individual values of a given measurement, for example,
> as it would lead to an infinite number classes, but certain statistically
> evaluated pairs or combinations of such measurements, specifically the
> distribution of their respective values in either time or space, may
> potentially prove useful in teasing out certain patterns in material culture
> and therefore serve as a reasonable foundation for class definitions.
>
> Am I being too optimistic about the utility of quantification in the process
> of classification, or is classificatory reality as interpretatively
> arbitrary, and therefore in a sense limited, as Dunnell envisages it?
You are not being too optimistic at all. In fact, this is the soundest
approach generally, in that one must allow the evidence to speak for
itself by providing an environment where this communication can be
understood. Rather than picking a few identifiers (and believe me, I
tried to do this for months, unsuccessfully), I took all observable
data into consideration and plotted the changes on flow charts. The
key to this was that, in matters of design, it is more likely that
major changes are made arbitrarily, and very minor changes are evolved
through repetition of insignificant detail (one develops a sort of
"shorthand"). This turns the common method of classification on its
head.
Essentially, I did as you said: I made each object its own class, and
then observed the inherent patterns over hundreds of design elements
to determine where the actual breaks were. When I saw most features
suddenly change, I took this to signify that everything before that
point was a true class. The internal changes said other things.
What Colbert de Beaulieu had defined as six classes, one following the
other, I redefined as three series, almost running concurrently in
different geographical areas. Subsequent testing has confirmed the
classification, and newly discovered specimens have all fitted into
the chronology perfectly. It is entirely objective, and the chances of
errors are minuscule where the chronology is defined.
Instruction in my method can be found at:
http://www.writer2001.com/analysis.htm
There is no reason that this method could not be adapted to other
classes of artifact, but it does require a large number of related
specimens to be workable, and these specimens must show considerable
variation in design.
It would be interesting to hear from others that have reclassified
artifacts, or have undertaken an original classification.
Regards,
John Hooker
--
Visit our Website at http://www.writer2001.com
Coriosolite Expert System...Animation...Poetry...Books
Hooker & Perron, Total Project Coordination
Technical Writing...Graphics...Maps...Colour Suites...Expert Systems
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|