[tei-council] [Fwd: Re: Proposal <idno> coverage -SF 2493417]

[Peter Boot]

Another message from Syd

-------- Originele bericht --------
Onderwerp: Re: [tei-council] Proposal <idno> coverage -SF 2493417
Datum: Fri, 23 Jan 2009 09:30:54 -0500
Van: Syd Bauman <Syd_Bauman at Brown.edu>

Some almost stream-of-consciousness thoughts about this follow.

* I like the <authorStruct> idea.

* Probably should be modeled on the now defunct <biblItem> idea,
   rather than <biblStruct> itself. That is, no PCDATA, and only a
   certain subset of elements, but that subset could be any number in
   any order, rather then prescribing a particular structure.

* Need to make sure the new element is flexible enough to describe a
   person, an organization, or one or more of each, probably in any
   order.

* We already have excellent elements for describing people and
   organizations: <person> and <org>.

* So content of <authorStruct> could be
      ( model.nameLike.agent | model.personLike | listOrg )*

* That doesn't leave a method for indicating an author that is *not*
   something you would indicate by a name. We would need mechanisms
   for indicating the concepts indicated by:
   - unknown
   - anonymous
   - various
   - et al.

* Don't need to be able to transcribe the actual words used to
   indicate these concepts, though, as one could use <author> for
   that!

* Perhaps there should be a formal way of indicating which author is
   the "primary" author, and which one to contact for reprints.


LB> As to the bizarre things that crop up in the TEI's current
LB> content model (which is model.phraseSeq) -- that's because we
LB> never really finished the job of defining
LB> model.headerRestrictedText -- i.e. a subset of phrase level
LB> elements which make sense as header content.

Actually, we finished defining model.limitedPhrase, which contains 67
elements, I think. But because <author> is used not only to record
authorship attribution in the header or in a structured bibliography,
but also to record authorship attribution "in the wild" of text being
transcribed, it has a content model of model.phraseSeq, which permits
134 or so elements.

It may be worth looking at the lists of elements permitted in
phraseSeq but not limitedPhrase and asking if any of them are
actually useful in <author>. If the answer is "no", we should
seriously consider changing <author> to model.limitedPhrase.

And heck, even model.limitedPhrase has things it's hard to imagine
needing in <author>. It would be fun to come up with a bibliographic
citation that needed <geogFeat> or <watermark>!