Syd Baumann wrote:
> In P5 the who= attribute of <sp> is a URI. That means, among other
> things that it can point to an element in a different file. Thus,
> Martin can have the who= attribute of normal characters point to the
> <role> in the <teiHeader> of their play, but the who= of what Michael
> calls "meaps" point to a <role> in the <teiHeader> that is a child of
> the <teiCorpus> element, possibly in a different file.
> Whether processed as a single play or as a corpus, who= can still
> perform its function (as long as the corpus file is available and in
> the right place -- a big "if", I admit).
I'm not sure this is really a new possibility enabled by P5. After all, the
same could have been accomplished via ID-IDREFS (assuming the underlying
problem about apparent non-uniqueness of the original ID's had been sorted
out, which is a practical issue rather than a conceptual one). The IDREF
"who" value on <sp> for "one-off" characters would point as usual into the
dramatis personae list, whereas the IDREF on a "meap" <sp> would point into
a list of meap's in the corpus header. OK, the use of URIs in P5 makes the
different targetting very plain to a human reader, but then so could a
suitably devised scheme of ID/IDREF generation and assignment.
The reason why I'd be hesitant about doing things this way, whether by
IDREFS or URI, is that, from both an editorial and potential user
standpoint, I'm uneasy about a markup scheme where the "who" attribute on
adjacent <sp>s can point to two sorts of places, each involving a
conceptually distinct type of editorial judgement, where one type of
judgement is (usually) uncontentious but the other inevitably involves a
level of conjecture, or espousal of a theoretical view about the "identity"
of characters, which not everyone will be comfortable with.
Hence my suggestion of a layer of indirection that allows these two types of
editorial intervention to co-exist while remaining manifestly distinct. In
what I sketched out, all <sp> "who" values point initially into the same
list which is derived from the dramatis personae listing of the edition
(though even here of course editorial interventions may be needed in some
cases to bring the printed list into line with the actual repertoire of
characters). Someone (or a process acting on their behalf) who doesn't want
to go beyond that point can then simply ignore any additional pointers into
the list of meap's. Or they can follow them up and then, in a further level
of optional indirection, follow the IDREFS in the meap list back to the
other embodiments of the ap in other works.
As well as allowing clear delineation of different levels and aspects of
editorial judgement, this approach may lessen the potential negative effects
if for some reason the corpus-header-located "meap list" isn't accessible.
In such a case, the lookup failure will occur only if the user requires meap
information, and *after* resolution of the first target (in the dramatis
personae list) has succeeded and so yielded one significant part of the
information the "who" attribute is it meant to deliver (and which under this
approach it can still deliver equally for all <sp>s). Whereas if some <sp>
values point straight into a list in the corpus header and that header isn't
accessible, processing of those values will fail with no information yield
other than an error message, even though the information which the user was
seeking may well still be there to be had in the dramatis personae listing
of the individual work.