Perhaps a more verbose but consistent way could be to use the @ana attribute, linked to a set of <interp> tags that enumerates all the interactions in the text <sp ana="#character1-to-character2">what char1 says</sp> ... <interp xml:id="character1-to-character2">Interaction form Char1 to Char2</interp> 2013/11/5 Lou Burnard <[log in to unmask]>: > On 05/11/13 20:43, Sebastian Rahtz wrote: >> >> On 5 Nov 2013, at 20:28, Lou Burnard <[log in to unmask]> wrote: >>> >>> I would do this with >>> >>> <sp who="#character1" whom="#character2">what char1 says</sp> where >>> @whom is an attribute I have defined in my ODD. >>> >> but then its not a TEI solution, if you can't define an archival >> interchange version >> based on pure TEI (your notation is fine as an editing format). >> >> > > But the TEI allows me to add extensions so long as I document them, I > believe. And new attributes dont even have to be in a special namespace, > since all TEI attributes are in the null namespace.... > > Oh very well. If you insist, I can't stop you using @corresp in the sense > of "vaguely related": > > <sp who="#character1" corresp="#character2">what char1 says</sp> > > > This at least has sort of the right semantics and is much more expressive > than your original proposal!