Regarding multiple titles and not multiple names - the MODS record would record how the manifestation represents the information - you'd only include multiple titles found on the manifestation (not all possible titles given over the various expressions and manifestations over time - that would go in an authority record).  Similarly for names - in a MODS record you'd only be using the names found on the manifestation (which indeed could be full or shortened forms, but hopefully the same name and not all the person's alternative names).  Providing the alternate and variant forms of names in MODS records requires doing it redundantly over and over each time a person is identified in a MODS record (I'd say not recommended!), whereas, having the information once in an authority record, removes that redundancy.  Hopefully the MODS and MADS records' information would be searched together and not require separate searching.   - Barbara Tillett

Dr. Barbara B. Tillett, Ph.D.
Chief, Cataloging Policy and Support Office
Library of Congress
101 Independence Ave., S.E.
Washington, D.C. 20540-4305

tel.: +1 (202) 707-4714
fax: +1 (202) 707-6629
email: [log in to unmask]

>>> [log in to unmask] 11/22/05 11:11 AM >>>
On 11/22/05, Marc Truitt <[log in to unmask]> wrote:

> Actually, I think 'sensible' is a better descriptor than 'arcane'.  I
> have no illusions that I can explain it more clearly than has Barbara
> Tillett in her posting, but perhaps this may help...

I'd call the FRBR/FRAR way for modelling this "sensible"; but not what
we see in MODS and MADS.

> Think of it from a database design perspective.  If you view books and
> persons who have some responsibility for them ('authors', for brevity's
> sake) as database objects, of which object (the book or the author) is
> the name 'Mark Twain' more properly an attribute?  The book _Tom Sawyer_
> or the person who also has the name 'Samuel Clemens'?  The 'arcane
> cataloguing practice' of authority control is designed so that the
> reader needn't know under what name Clemens wrote _Tom Sawyer_ (or any
> other of his numerous works).  If you regard the name as an attribute of
> the work, then you would either have to append every known variant name
> to each of his work objects or else you would have to settle for knowing
> in advance which variant name went with which work object, would you not?

No, this argument has always confused a number of things. For those
arguing this perspective, you are in part seeming to only view MODS
and MADS as long-term storage formats. But that's rather limiting.

James and I are both interested in end-user oriented citation
metadata, where those files might be generated on the fly from some
database. There the XML becomes a transport format basically.

But even there, if we were to do this in a relational way (say RDF), we'd say:

[a book] is authored by [a person]
[a person] has name [x] and alternate name [y]

Admittedly, it gets a little tricky to then indicate that the book is
authored by a person under their pseudonym, but I stand by my claim
that notwithstanding AACR2, it's completeely arbitrary to say that one
can include multiple titles in a MODS records, but not multiple names.

I certainly wouldn't use it to include different names like "Samuel
Clemens", but I would use it to include different language forms.