Slightly orthogonal to this discussion, but only slightly, it occurs to
me to ask where the topic-map itself should reside in a TEI universe. It
could, as Patrick suggests here, be a free standing document, associated
with the documents described by it by means of xpointers or  similar.
But why not consider locating it within the TEI Header? The <textClass>
and <classDecl> elements seem to have purposes very similar to those
served by topic maps: to identify the  subject matter and classification
of parts of a document or a suite of documents.
Now that we have namespaces, simply adding an xtm element into the
header is a very real option -- we just need to decide on the right
place, and on some usage rules...

One point that occurs to me: you might well want to have your topic map
describe a whole set of TEI instances, rather than a single one. But
that's why God gave you the <teiCorpus> element,  more specifically the
concept of a global teiHeader (which contains data about your whole
collection) as opposed to the specific teiHeader (containing data about
a specific document)

Lou

Patrick Durusau wrote:

...

>
>>
>>PS: Given the lead your project has taken in the use of topic maps with
>>TEI, I wonder if there is enough interest to support discussions/drafts
>>of applying XTM to TEI document instances and/or developing TEI markup
>>for topic maps? While not a universally shared opinion,  ;-) , I think
>>understanding of the paradigm has progressed since the formulation of
>>XTM. Perhaps TEI markup for topic maps could take advantage of that more
>>mature understanding.
>>
>>
>
>
>
>