I ask that the Council please consider changing the datatype of @rend from 1–∞ occurrences of data.word to simply "text". There was a lengthy discussion of this issue on TEI-L recently. Arguments in favor of this change include:
—It is backwards compatible with current usage. 1–∞ occurrences of data.word is also valid as Datatype text.
—It supports the use of CSS in @rend. The current Datatype, if one follows both "the spirit and the letter" of the Datatype, does not support CSS.
It is useful for a markup language to support both references to styles (as with the <rendition>/@rendition model) and inline styles. There is little point in creating a "class" for a style using <rendition> if that style only appears once in a document. There are many other scenarios in which the incline style could be useful.
The current Datatype discourages users from using a widely adopted standard (CSS) in @rend to describe the rendition of an element. And I believe it would be wise for TEI to at least support the use of this standard. Users prefer to describe renditional using a non-standard language, may continue to do so.
In the discussion on TEI-L it became clear that a number of users (including some current and former members of Council) are already using CSS in @rend and advocated official support for this usage. Others acknowledged the benefits of supporting CSS in @rend. There was of course, some dissension, but while some may believe the change is unnecessary, I don't believe anyone argued that it was harmful or dangerous.
There was also discussion of establishing a mechanism for declaring how one was using @rend. I believe such a mechanism would be helpful, but is certainly not necessary. I think it's much more important to change the Datatype. The other issue can be sorted out later, and the current <encodingDesc> provides mechanisms for making such declarations.