2013/11/8 Gabriel Bodard <[log in to unmask]>:
> Yes, and you may want to treat and point to #s2 and #s3 separately in other
> contexts, so just joining them into a single segment/sentence is entirely
> sidestepping the problem.

Ok but you can assign @xml:id also to sub-components so you can point
them, can't you?

> On 2013-11-08 14:33, Lou Burnard wrote:
>> sorry, that;'s not legal: if you tesselate with <s> you cannot introduce
>> an <ab> in between them.

I didn't mean to make a real example of continous <s> elements (they
could be in different places). I was just suggesting that you can
enclose <s> (or whatever) elements inside an abstract container (<ab>
or <seg> depending on the cases) and treat them as a block if and when
you need. I agree that <join> is a better and more general solution,
but in some situation enclosing could be more "natural" and

>> but this is really just evading the inherent ambiguity of @corresp!

Yes, that is true, just ad hoc tricks for a very specific encoding
situation. But would you really resolve that ambiguity in the
Guidelines, and how? That is why I think there should be a way to
state locally the intended semantic of markup constructs where TEI do
not want to or can not enforce it, in order to keep safe its