{like the original message, this is posted both to BIBFRAME and to RDA-L]

On Tue, 8 Nov 2011 18:59:50 +0000, Akerman, Laura <[log in to unmask]> wrote
(in part):

>To me, the original "main entry" concept could more usefully be thought
about in a larger context of  "for any field that is repeatable in a set of
bibliographic description fields, is it useful to be able to designate one
such fields as "primary" for purposes of selection for display,
categorization (where a particular application requires one to "select one
box" to characterize a resource) or other functionalities?  If so, should
the designation be stored with the field, or separately from it?

There are, to my mind, certain structures which are properly inherent in the
bibliographical data, *because* they're inherent in the documents (and
document-related entities) we're describing. The fact is that we name the
creator of a text along with its title as, or as part of, a *citation*
(which may be regarded as a form of the name of the document [FRBR
"manifestation"] or of the text [FRBR "work"]. We do this because we have to
use that name in order to make consistent and coherent reference to it when
it occurs elsewehere; as Mac Elrod pointed out, when a document or a work
has to be cited as the subject of another document, or as a related work.

More than that, however, users work with citations, and the widely-used
style guides provide rules and patterns for creating them. If the records we
formulate don't correspond with those patterns, we place a barrier between
users and our data. Since the first principle of the catalogue is to serve
the convenience of the reader rather than the creator of the catalogue, we
have no right to build or perpetuate such a barrier. Therefore the data as
recorded should include the creator relationship. Whether it's recorded by
making the creator a special element, or by attaching the relationship as a
connection between the creator element and the title element, is secondary.
It occurs to me to point out that in MARC 21, the relationship is recorded
by using a special MARC tag (100/110/111) but in Unimarc it's recorded by an
attachment (a special indicator value) to the tag used for any
responsible/related name. In other data formats there are other mechanisms.
Some of them may even be able to accomodate shared multiple, responsibility
(e.g. W.S. Gilbert and Arthur Sullivan both responsible for _H.M.S.
Pinafore_ but conventionally cited as Gilbert first, then Sullivan (contra
AACR2 practice).

>For fields like subject, I believe there was a convention that the most
important subject (the one upon which the primary classification number was
based) had the first position in the record.   Since many modern systems
permit or even force re-ordering tags in numerical order, that positional
value can and often is easily lost.  Many of us stopped lamenting this a
long time ago, but was it valuable?

Well, I still lament it! Seriously, I think it worth having the principal
subject(s) specifically indicated, ancillary subjects being able to be
omitted from brief display of data but available for more comprehensive
display.
>
>What I don't think is valuable, is having to pick one author of a work with
multiple authors and designate that person as the "main" one, based on the
almost arbitrary factor of position of the name on the title page, (which is
often alphabetical), and ending up deeming this person "Creator" and
relegating the other author(s) to "Contributor" status.  (Nor do I think
that dichotomy is particularly useful.)

As you may gather from the above, I disagree.  The reason for my
disagreement is twofold: it needs to be plain that the record (or other set
of data if "record" is inappropriate) corresponds with the document in hand;
and it needs to correspond as well as possible with the citations users
bring to the catalogue -- the fact that citations may be imperfect does not
mean the data content doesn't matter. The implication is that joint
responsibility needs to be acknowledged (as, I believe, does editorial
compilation of existing material, like volumes of readings -- the citations
are written that way, why not cataloguing also?) otherwise we place a
barrier of ambiguity between user and data.  And one of the functions of
catalogue records is to be used to create citations -- Endnote, anyone?

So long as different elements are distinguished (names, relationships,
titles, other significant data elements) the format in which it's recorded
is very secondary.

Hal Cain
Melbourne, Australia
[log in to unmask]