Sebastian, Conal, and TEI-L,

I'm afraid I don't have time to write properly (teaching today and 
tomorrow), certainly not to say what I feel has to be said in a form 
brief enough to be useful. So I won't.

But I will say, just to throw more wood on the fire:

No, the question of what would go into a more minimalistic tag set with 
constraints is not at all trivial.

I agree in general that much of the problem is that the content models 
of P5 are too tight as well as too loose. Sometimes both at the same time.

I also agree that it has not yet been satisfactorily demonstrated that a 
workable interchange tag set across this domain is possible. (After all 
there isn't a single domain.) Probably that can be demonstrated only in 
the demonstration. But we can learn by trying.

In particular:

On 8/25/2011 5:03 AM, Sebastian Rahtz wrote:
> I am reminded that quite a few elements in the TEI are known to
> be syntactic sugar or specialization. So<persName>   ==<name type="persName">
> for example. Although my heart sinks at the work involved, we could/should formally
> record these links, and then generating the programme which resolves them
> is trivial. similarly<div1>  ==<div type="div1">
> more generally, it's a truism that "with<div>,<ab>  and<seg>  I can mark up anything",
> but reducing the whole TEI set to those three seems to go too far...

I agree. However it might not be too far to go to say there's a properly 
descriptive set including p, lg, l, and a couple dozen more, and work 
from there, falling back to microformat-like markup on ab and seg where 
necessary. While the scoping problem is key, it is not impossible to 
allow for a certain grey area where description is less fully controlled 
than it was supposed to be in the old days (when control had to be much 
more all-or-none than it now does), and nevertheless possible and not at 
all useless, if done right. (This is essentially the thesis of my 
Balisage paper.)

One might also start assessing the 80% by performing analytics over 
representative data sets to see how elements have actually been used in 
the past.

But I'm teaching again tomorrow so I won't be able to address what 
follows with anything like the level of attention it deserves. :-(


Wendell Piez                            mailto:[log in to unmask]
Mulberry Technologies, Inc.      
17 West Jefferson Street                    Direct Phone: 301/315-9635
Suite 207                                          Phone: 301/315-9631
Rockville, MD  20850                                 Fax: 301/315-8285
   Mulberry Technologies: A Consultancy Specializing in SGML and XML