[tei-council] Choice and App/rdg again

[Daniel O'Donnell]

So if I understand things aright, what we're saying is that we have two
kinds of apparatus and two kinds of ways of indicating alternate
encodings: one for phrases, and one for units larger than the phrase.

In the case of textual apparatus, this seems bizarre. If Anne Frank
deletes a word from one draft to the next or deletes an entire diary
entry, I fail to see how these are different from a point of view of
collation. They are both textual variants. The question is whether or
not we can capture the extra information that we know--i.e. that a
missing material is actually an entire entry or just a word. Likewise
with Cædmon's Hymn: when Dobbie doesn't indicate that the extra "words"
he includes in his apparatus are actually a metrical line, he is hiding
information he knows. I don't see why we would willingly continue this.
The fact remains that for some texts--including real use cases--the unit
of collation is larger than the phrase. This is a fact of textual
criticism.

As for choice, Syd raises an interesting question: why have choice at
all if 

> One could argue that
> >     <choice>
> >       <sic>Dairy</sic>
> >       <corr>Diary</corr>
> >     </choice>
> > is essentially the same as
> >    <sic xml:id="e1" exclude="#c1">Dairy</sic>
> >    <corr xml:id="c1" exclude="#e1">Diary</corr>

If @exclude was renamed @alternate or @choice or something, and doing
this removed the problem of limiting choice only to the level of the
phrase, then I'd say get rid of it and make @alternate|@choice global.
My point is that the idea of simultaneity is not hierarchical--it is a
logical distinction. Even if we stick to the most restrictive
understanding of choice--that it involves encoding only and not content
(something I'd argue is a mistake)--it is simply not true that there is
anything inherently phrasal about the logical connection between the
options choice contains. Thus logically, given our definition of choice
as "groups a number of alternative encodings for the same point in a
text" both of the following encodings ought to be correct:

<choice>
<expan>Text Encoding Initiative</expan>
<abbr>TEI</abbr>
</choice>

<choice>
<div type="option1">
<p>The text encoding initiative is an organisation devoted to the
development and maintenance of standards for the encoding of texts. It
is built on a consortium model, with four hosts and an elected board and
council</p>
</div>
<div type="option2">
<p>The text encoding initiative is an organisation devoted to the
development and maintenance of standards for the encoding of texts.</p>
<p>It is built on a consortium model, with four hosts and an elected
board and council</p>
</div>
</choice>

In fact, only the first is.

Since I had to add the divs in the second example purely to distinguish
between the options, this may point to a deeper problem. Maybe what is
really going on here is that we have a couple of standard cases in which
we understand what the options are (abbr|expan, sic|corr, etc.) and a
more generic model where what is going on is something like
choice/option/otherStuff|option/otherStuff.

Me, I'm agnostic about whether the alterneity/simultaneity is
represented by element or attribute--attribute is cleaner, given the
issues of divs within segs Syd raises, perhaps--but I don't see any
evidence arguing that the actual idea of choice is hierarchical. A quick
and dirty element solution is to have phraseChoice, chunkChoice,
etc.--not my solution, but evidence that it ain't insurmountable.

I think choice is one of the most interesting additions to P5--I'd
really like to get the idea right.

-dan



On Sat, 2007-07-14 at 16:07 +0100, Lou's Laptop wrote:
> Just for the record, I am in 100% agreement with Syd on this! <choice> 
> is not meant to be used in the way Daniel suggests; <alt> is the 
> recommended method for dealing with such things.
> 
> Lou
> 
> Syd Bauman wrote:
> > Daniel --
> >
> >   
> >> ... let me re-raise the issue with choice and app/rdg: my view is
> >> that the current phrase-level model is [too] restrictive.
> >>     
> >
> > OK, but let me point out that even if you convince us that <choice>
> > should contain paragraph-level or div-level chunks, this is *way* too
> > big a change to even be *considered* for P5 1.0.
> >
> > But convincing me, at least, is going to be hard, not because of any
> > lack of truth in your argument that encodings of things larger than
> > phrases may occur in alternation -- I'm completely with you there.
> >
> > But rather it is difficult because changing the content of <choice>
> > to allow things larger than a place where it itself can appear is
> > very problematic: all of a sudden you can have <div>s inside of
> > <seg>s so long as there's a <choice> between the two. Quite bizarre.
> > Perhaps not impossible to get around with sincere warnings and
> > Schematron, but no small matter either way.
> >
> > But perhaps equally important, this may not be necessary. We already
> > have reasonable encoding methods for alternation. These include the
> > <alt> (and <altGrp>) element(s):
> >
> >     <div type="dairyEntry" xml:id="z23a">
> >       <p/>
> >     </div>
> >     <div type="dairyEntry" xml:id="z23b">
> >       <p/>
> >     </div>
> >     <alt mode="excl" targets="#z23a #z23b"/>
> >
> > And the exclude= attribute:
> >
> >     <sp who="#dg">
> >       <p xml:id="y12a" exclude="#y12b">Holy alternation, Batman!</p>
> >       <p xml:id="y12b" exclude="#y12a">Holy fledermaus, Altman!</p>
> >     </sp>
> >
> > One could argue that
> >     <choice>
> >       <sic>Dairy</sic>
> >       <corr>Diary</corr>
> >     </choice>
> > is essentially the same as
> >    <sic xml:id="e1" exclude="#c1">Dairy</sic>
> >    <corr xml:id="c1" exclude="#e1">Diary</corr>
> >
> > _______________________________________________
> > tei-council mailing list
> > tei-council at lists.village.Virginia.EDU
> > http://lists.village.Virginia.EDU/mailman/listinfo/tei-council
> >
> >   
> 
> _______________________________________________
> tei-council mailing list
> tei-council at lists.village.Virginia.EDU
> http://lists.village.Virginia.EDU/mailman/listinfo/tei-council
-- 
Daniel Paul O'Donnell, PhD
Department Chair and Associate Professor of English
Director, Digital Medievalist Project http://www.digitalmedievalist.org/
Chair, Text Encoding Initiative http://www.tei-c.org/

Department of English
University of Lethbridge
Lethbridge AB T1K 3M4
Vox +1 403 329-2377
Fax +1 403 382-7191
Email: daniel.odonnell at uleth.ca
WWW: http://people.uleth.ca/~daniel.odonnell/