[tei-council] span span span span span span span span glorious span

James Cummings James.Cummings at oucs.ox.ac.uk
Sat Aug 20 15:57:43 EDT 2011

On 20/08/11 10:20, Lou Burnard wrote:
> Some might say that these examples all indicate the confusion that
> ensues when you mix the purely documentary view with the textual,
> interpretive one. But just for fun, let's pretend there was some visible
> way of representing the boundaries of personal names in the written
> source in question. It's surely feasible that a scribe might wish to
> show that they were actually referring to a single person called Mr
> Holmes-Burnard? And hence, there needs to be *some* way of indicating
> that the markup is being deleted as well as the content.

That usually with <del> any enclosed markup is also deleted is, I admit, 
a good argument that the same should be true of delSpan. So on 
reflection I think I was wrong and that the default assumption by normal 
people is probably that all markup is deleted...so instead of needing a 
way to indicate that the markup is being deleted, maybe it is the other 
case, that the markup is preserved that needs a way to indicate it?

> My preference therefore for (b) in my original opposition is based on
> the purely pragmatic ground that if you really want to leave the markup
> behind you can do so explicitly by supplying two deletions, as Martin
> observed.

I think providing two deletions when there is a single line (without 
indicating that these are fragmentary elements or something like that) 
is encoding an untruth.

> What worries me more about this, and which no-one has yet observed, is
> the high probability of producing (or implying) an ill-formed document.

But that is true of all other spanning elements as well, no?  <addSpan/> 
or even things like <join/> if you chose to implement it without 
respecting the current XML hierarchy?  With none of these are we 
creating a well-formed XML document that represents both structures, we 
are documenting a textual or graphical phenomenon using one hierarchy or 
the other.  All of this makes me long for easy to use, generic, 
stand-off annotation/encoding tools.

> But using @spanTo is so fraught with the possibilities of
> own-foot-destruction (as James has noted) that one more or less seems
> less of a concern.

I recognise my examples were limited and artificial, I think I know how 
I might encode some of them but it is more that those seem to be 
completely obvious use cases, thus need documenting.


Dr James Cummings, Manager of InfoDev
Oxford University Computing Services
University of Oxford

More information about the tei-council mailing list