Patrick Durusau <[log in to unmask]> writes:
 
[what about fragility wrt corrections which perturb the ID set when
using stand-off markup?]
 
You're right to worry, and to be sure to design an ID numbering rubric
which will allow interpolation: I'd suggest alphabetic augmentation
for interpolated or overriding corrections:
 
<c type="consonant" id="00001a">w</c>
<c type="consonant" id="00001b">h</c>
<c type="vowel" id="00002">a</c>
<c type="consonant" id="00003">t</c>
 
Note it's actually a GOOD thing that links to 00001 are now broken:
the change happened at a word boundary, so you need to go back and
look at everything that pointed at the changed area.
 
But if e.g. 00002 had changed from 'a' to 'o', and therefore to
00002a, probably nothing would break, and that's correct.
 
For a discussion of how to do this without editing, using an addition
to the repertoire of stand-off markup semantics, see [sorry to be
referencing the same thing twice in succession :-]
 
    http://www.ltg.ed.ac.uk/~ht/sgmleu97.html
 
ht
--
  Henry S. Thompson, HCRC Language Technology Group, University of Edinburgh
     2 Buccleuch Place, Edinburgh EH8 9LW, SCOTLAND -- (44) 131 650-4440
            Fax: (44) 131 650-4587, e-mail: [log in to unmask]
                     URL: http://www.ltg.ed.ac.uk/~ht/