Thanks for this extensive reply, Syd (deleted below because it was so
extensive, let alone my previous mail), and again I've learnt so much
(although my nose hurts a bit, but that can't be helped).
Some clarifications first:
span and fragment: these may of course coincide. I was thinking of the
following where they don't:
<p><q>There may indeed be problems with using the part attribute for
<p><q>Especially,</q> the sages added, <q>if you keep the standard
This is a speech, consisting of two spans (There ...
Especially; if ... values), the first of which is split into two fragments
(There ... for Q; Especially).
I hold no interest whatsoever in linking all the pieces of the whole
speech. But I want to link the fragments of a span. And this elegantly and
in a manner that's not time consuming. Not all that hassle with
thinking up IDS (and often two for one fragment), making sure that
everything fits and matches. I did this for one text and I'm still dizzy.
Linking took me almost as long as all the rest taken together.
(It was another Bible and speeches easily cover half a dozen verses.)
I didn't mention user-friendliness last time, did I?
Just linking the fragments, and then leave the spans unconnected, seems
wrong to me. But I may have wrong notions about linking.
I wouldn't have those qualms with linking via a PART-attribute.
As for linking fragments (into spans) and spans (into speeches)
separately: say, for some reason you want to keep the speech together
(get all the dialogue out of a book and keep all the chunks together arranged
by speaker) and for some other reason you want to keep the span together as
well, but separately (to later squeeze in the quotation marks, say).
Yes, imagine you're really forced to link all the bits and pieces of a
speech, like this:
[The speech covers three divs and gets interrupted in the middle of div
<q id="qtag1" next="qtag2a">
There may indeed be problems with using the part attribute for Q.
<q id="qtag2a" next="qtag2b" prev="qtag1">
the sages added,
<q id="qtag2b" next="qtag3" "prev="qtag2a">
if you keep the standard values.
<q id="qtag3" prev="qtag2b">
So better leave the DTD alone.
How can you make sure that you get the correct quotation marks?
next alone = beginning
prev+next = middle
prev alone = end
the two fragments in div No. 2 (qtag2a, qtag2b) will be analyzed as middle
parts. In English they would both get quotation marks at the beginning, but
not at the end, which will be wrong for the first fragment (qtag2a), which
should get both.
I would like to point out that other languages restrict quotation marks to
the very beginning and the very end of a span of quotation, no matter
over how many paragraphs etc they extend.
Take German for instance (three paragraphs of direct speech):
So in German, qtag2a and qtag2b wouldn't get any quotation marks, being
analysed as middle parts, which is completely wrong - qtag2a should get
closing quotation marks (only), qtag2b opening quotation marks (only).
Is there anything here that I don't see?
BTW: if there's a better term than 'span' I'm happy to use it. I couldn't
come up with anything better.
Now, to modify TEI Lite, or not to modify TEI Lite.
If it's not possible, why does the XML version of the TEI Lite DTD say
"Please do not edit this file!
To modify TEI Lite, please change the modification files
teilitex.dtd and teilitex.ent" etcetc.
Modifying some DTD by adding the part attribute to Q.
You, Syd, said you did this at the WWP. Given the problem with the
values and DIRECT, what did you eventually do? Chuck out or change direct?
(Or did I miss that piece of information?)
Basically, it would be the better idea to change DIRECT as it's unique to
Q, whereas PART is, if not exactly global, quite common.
Anyway, it's irrelevant what you call the values as long as they are
unique for an element, right?
With Q, that means Y and N are nonos.
So, I, M, F, C, U would do.
C for complete (would be the default), U for uncomplete (although I wonder
whether anybody would ever need uncomplete here, unless mice have
feasted on the manuscript [<gap reason="eaten">]).
Everything else in that reply still needs absorption.