[tei-council] stemma modeling

[David J Birnbaum]

Dear Council,

To adapt the markup used in my stemma report to current TEI practice:

1) Use TEI <eTree> where I use <node>.

2) Change the attributes for identification and pointing as James proposed.

3) Change my <contaminates> to TEI <relation>, as Arianna proposed.

Problem: <relation> is not currently permitted inside <eTree>, and it 
would need to be for this to work. I think the question for Council, 
then, is whether we want to permit <relation> inside <eTree>. If so, the 
adaptation is easy, and we can fold my (modified) stemma example into 
the graphing chapter as an example of using <relation> to model a 
contaminated tree. If not, we should proceed without the proposed stemma 
model.

Argument for: A stemma is not a tree, but it can be understood as a tree 
with something else (and not simply as a non-tree). Accordingly, 
modeling it as a tree (with <eTree>) but adding something to reflect the 
contamination (<relation>) conforms to what's really going on.

Argument against: A contaminated tree is by definition a non-tree, and 
tree-oriented markup (such as <eTree>) is therefore inappropriate.

For what it's worth, I find the argument in favor more useful for those 
who might want to model stemmata.

Best,

David