Thank you for your great comments -- I've responded below.
Additionally, has anyone actually used the language corpora tagset for
letters, or considered using the tagset, and rejected it in favour of
> To be honest, I've never actually used a <textDesc> except
> for test- ing, so I don't know if I'd find I like it or not.
> But it seems reasonable. However, the content element
> children of <textDesc> must occur in a particular order,
> which was not the order in your message.
Sorry about that, I created the example hastily.
> One thing that really bugs me about <person> is that there's
> no key= attribute. I think it makes perfect sense to want to
> have a <partic-
> Desc> with <person>s and <personGrp>s to assign IDs and role=s to
> various people involved with a particular text, but to want
> to list their demographic information in an external database
> (am I nuts?).
We will indeed be storing information about people, and possibly places,
> Sadly, to do that as currently written, one
> would have to do something like put a <persName> with a key=
> inside <person>.
The key= attribute does seem to be placed one level too deep given that
everything under <person> is associated with a unique person.
> Of course, to use a <persName> one must also
> include the additional tagset for names and dates.
We will probably include the additional tagset for names and dates
anyhow, as we need to describe dates in some detail. I actually have
another question about indicating the uncertainty of days, months, or
years, rather than the date as a whole; I'll post that question
> > <personGrp id="g1" role="sender"></personGrp>
> > <personGrp id="g2" role="recipient"></personGrp>
> I'm not sure I understand the purpose of these groups; do you
> have letters for whom the sender or recipient are unknown?
Well yes, I imagine there are letters with unknown senders or recipients
(I'd not thought of using a <personGrp> to describe this situation).
However, I was thinking more of letters sent to or from a group of
people, whose individual members aren't known, and so can't be included
as <person>s. Using <personGrp> would also indicate that a letter was
sent to or from a recognized group (but not necessarily an
organization), rather than from an otherwise unrelated collection of
> Hope this has been helpful, even if it doesn't actually
> answer your question.
Very helpful. Thank you again.