Actually, we're working on exactly the same problem for the Internet
Shakespeare Editions project: cast lists which are the work of modern
editors rather than forming part of the source text.
I like the idea of making the content model of <role> similar to that of
<person>. That's a rich selection of elements, but those from namesdates
especially would be very useful in providing richer information about
<role>s. <role> itself already has a strange assortment of items from
namesdates, but lacks some of the most useful such as <age>, <birth>,
<death>, or <faith>.
On 2015-05-06 10:48 AM, Martin Mueller wrote:
> I have been thinking about ways of classifying characters in early
> modern plays in terms of sex, age, social status, and relations to other
> characters so that a corpus-wide view of the literature would give you
> some sense of change over time by just looking at cast lists or their
> equivalents. There doesn’t seem to be any obvious way of putting that
> information into the <castList> elements. The Folger Digital Editions of
> Shakespeare’s text have used <particDesc> in the header to convey some
> of that data. The <person> element that is the key to <particDesc> would
> do a lot of the work.
> Would it make sense to allow the child elements of <person> as child
> elements of <role>? Have others wrestled with similar problems and found
> good solutions? Graph databases (about which I know next to nothing)
> seem a promising avenue for storing and analyzing that kind of data. But
> they are a very different animal from TEI.