Multiple hierarchy raises their head down again

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view
|

Multiple hierarchy raises their head down again

Lou Burnard-7
[ A colleague raises the following questions about using SGML for
literary texts, which are of general interest. I pose them as they were
put to me, together with my responses. Any other views/comments
gratefully received.
                        LB      ]

"I have run into a couple of puzzles with SGML and am curious if
"you have ways out.

"1. quotes: <q> Quotes can be nested in quotes and for presentation can
"witch between single and double quotes. In text a quote can cross para-
"graph boundaries. The presentation convention is to precede each new para-
"graph included in the quote with an open quote mark. Structurally one
"could tag the beginning and the end of the quote, but any tag for the
"paragraphs would conflict with the first convention. What are you doing in
"that case?

Quotes can indeed be nested. If the use of single or double quotes in
the original is of importance, then the RENDITION attribute can be used
to signal it. Otherwise, the particular rendering for quotes-within-
quotes is an application-specific matter which should not, in my view,
be tagged at all. By the same token, I would not distinguish run-on from
block quotes.

The structural problem on the other hand is a killer. (And it applies
all over the place: see below) Either you have to use a concurrent
markupstream for the quotes -- arguing that this is quite a separate
structure from the paragraph one -- or you simply have to pretend that a
quoted passage which spans a para break is actually two adjacent quotes.
Or use a milestone tag for the para break. None of these solutions is
really wonderful, I agree, and it gets much worse in blank verse, where
you can have verse paragraphs, verse lines, and quoted passages or
speeches, the boundaries of all three of which resolutely fail to nest.


"I am experimenting with Donne's Religious Poetry to see what tags and
"dtd might be useful. For this I am using the Gardner edition. The higher
"structures are problematic. I have defined the following:
"  poem.set = the whole group of religious poetry
"     poem.subset - holy sonnets forms such a group, another is occasional
"            poems
"         poem.collect e.g. La Corona
"             poem e.g. a sonnet of La Corona
"         poem
"Is it a problem that poem can appear as a subsection of poem.subset and
"poem.collect? A litany poses another problem because it forms a poem.collect
"section for which there is no poem.subset unless I give it a dummy one. I
"actually would like to categorize it as poem.
"Any suggestions?

This issue, unless I've misunderstood you, is adressed in the TEI
Guidelines. You can use the neutral DIVn tags, with a name attribute
wherever a work can be hierarchically divided. You can use different
values for the `name' attribute in the same hierarchy if you like
Thus:
    <div1 name=section n='Religious Poetry'> for your poem.set
    <div2 name=subsect n='Holy Sonnets'> for your poem.subset
    <div3 name=group n='La Corona'> for your poem.collect
    <div4 name=sonnet n='lac.12'>  for the 12th sonnet in  La Corona
    <div2 name=Litany n='Whatever the title of the litany is'>
    <div2 name=poem n='This poem isnt subdivided at all'>

It's up to your application what happens to attributes of course. I
suppose the underlying problem is that the notion 'poem' is ill formed
as a structural component -- for some purposes 'La Corona' is a single
poem and for others it isn't, and the markup structure reflects that.

Lou