A number of the messages that have been posted to the MODS list over the
past week seem to deal with the granularity issue.  Most of us are
aware of the level of granularity in MARC and the relative absence of it
in Dublin Core.  MODS appears to be establishing, or at least suggesting,
a new level of granularity somewhere in between.  Whether someone is
talking about names, titles, or subjects, as Karen Coyle pointed out in
her last posting to the list, every community is going to have it's own
notions and requirements about granularity.  Even the library community is
not in total agreement.  AACR divides information one way, ISBD does it
slightly differently, and MARC too takes a sometimes unique approach
although tries to accommodate both other standards.  I would hate to
see MODS get bogged down by this issue.  It seems clear that MODS is
attempting to eliminate granularity that has not proven useful.  In the
case of headings, for example, this might mean the loss of subfielding.
I hope we can avoid making MODS as granular as MARC or AACR.  Since the
ISBDs have not, up to now, dealt with headings, we are still a bit freer
to design things for headings, although MARC seems to be pushing us toward
greater granularity there.

After considering the granularity issue I have come to the conclusion that
very little of it (in MARC at least) is actually put to much good
use.  What seems to be of greatest importance in MARC is actually the
useful redundancy provided.  The fact that there are distinct data
elements for an inverted (and highly granular form of name) and a
trascription of the name as it appeared on a bibliographic item seems to
provide the most for later use.  The granularity of the encoding of that
data doesn't seem to matter much in most systems.  It makes me wonder why
we do it at all in MARC, and certainly dissuades me from having any of
that granularity in MODS.

Thus, I am not fond of any of the proposals I have read that suggest
additional granularity in specific MODS elements.  I can only see
usefulness in providing some redundant (ungranular!) elements to
support different uses of the data.  This does raise the questions, will
people ACTUALLY use these elements?  We have fought with this for years in
MARC with nonlibrarians who find it so hard to understand why we have a
transcription of names in field 245 and then repeat them in 1XX, 6XX, and
7XX fields.  The same questions would be raised about titles.  I think we
need to look closely at the granularity and perhaps too at areas where
there might be "useful redundancy" in MODS and rationalize our choices.

I'll end this message with a question:

With the ability to use initial and final wildcards in searching, in
addition to the power of string and Boolean search techniques, do we
really need to do ANYTHING special with the encoding of articles
and the parts of names/subjects anymore?

The answer(s) might guide us on whether any additional granularity is
needed in MODS name and title elements.

Randall K. Barry
Library of Congress
Network Development and MARC Standards Office
101 Independence Avenue, S.E.
Washington, DC 20540-4402  U.S.A.
TEL: +1-202-707-5118
FAX: +1-202-707-0115
NET: [log in to unmask]
Ideas and opinions expressed in this communication
are personal and do not necessarily reflect the
position of the Library of Congress.