Regarding encoding interpretation, you will need to beware about
overlapping tags. For example you may want to tag from half-way
through one line to half-way through the next as being an object of
interpretative value. I'm not sure what the current recommended way
to handle this is, but in a project I was involved in we used anchors
and then had joins at the end that identified that from anchor X to Y
there was interpretative feature Z. The problem with this is that it
is harder to write programs that can process this for analysis as
they need to first parse the joins and then pull the segments between
anchors. It wasn't a simple "search for all the action segments".
There is a deeper issue which is exploratory markup and handling
overlapping structures. A paper by Wendell Piez at http://
EML2001Piez01.html looks at this. Another paper that I noted builds
on Piez's LMNL to handle overlapping structures. See http://
www.hum.gu.se/allcach2004/AP/html/prop104.html . Others on this list
will be aware of more current research or other ways of combining
exploratory markup with descriptive markup.
COCOA (who remembers COCOA) was actually quite good at exploratory
markup, but terrible at everything else. Maybe stand-off markup has
come far enough to let you keep different structures independent so
you can have your interpretation and eat it too.
On 28-Sep-06, at 3:10 PM, Amélie Zöllner-Weber wrote:
> Dear Sirs,
> I have a small corpus of literary texts that I want to encode
> by the TEI. I am especially interessed in encoding text sections
> that are important for literary analysis. It would be very nice
> if I can mark sections according to descriptions or statements
> of literary characters, their actions and speeches. For analysing
> reasons it is also important to include information about the
> relations of the characters, like which character acts in relation to
> another. Here is an example of what I want to encode:
> <l><s function="statement" who="narrator"
> object="Mephistopheles">Er ist's!</s> am Baum hervor, aus Moos und
> <l><s function="description" who="narrator"
> object="Mephistopheles">Mit seiner Augen finsterem Geloder</s>,</l>
> <l><s function="action" who="Mephistopheles">Der
> Teufel blickt gewärtig und bereit</s>,</l>
> <l><s function="action" who="Mephistopheles"
> object="Faust">Und streckt sein Haupt in Faustens Einsamkeit</s>.</
> So, I want to ask if someone might have suggestions how to encode
> this. There might be completely different ways to encode the texts
> as well, by for instance by using feature structures. It would be
> very interesting to know how other projects have encoded texts
> for literary analysis.
> Yours sincerely,
> Amélie Zöllner-Weber