Print

Print


The following is an account of the additional <SPAN> tag proposed
at the Myrdal TEI meeting by a working group of Elli Mylonas, David
Barnard and myself, and approved at the final plenary session.
Comments are invited.
 
The mark-up of literary texts of all kinds frequently requires the
identification and description of segments of text, often long ones, that
are non-structural, in the sense that they are not part of any system of
hierarchical divisions, may overlap with one another, and may
be discontinuous.  For instance an encoder may wish to mark
passages of text associated with particular characters, themes, images,
allusions, topoi, styles or modes of narration.  FSA offers one very
powerful means of encoding such features, but with the attendant
disadvantages of unreadability.  Another means is to use the <NOTE> tag:
<NOTE ed = DR> theme of love begins </NOTE> ... <NOTE ed = DR> theme of
love ends </NOTE>.  This tag can be appropriate for the encoding of
analyses that might be regarded as matters of interpretation, as personal
or controversial;  the fact that the tag indicates the originator of the
analysis is thus especially important, as a means of avoiding a spurious
air of objectivity in the mark-up in question. One disadvantage is that it
involves the interpolation in the original text of material which might
inadvertently be left in when a simple mark-up stripping facility is used
on the file.  (We therefore recommend that where the <note> tag is used in
this way, users' attention should be drawn to the fact in the document
declaration;  it would also be desirable to have tag-stripping software
that included the possibility of removing such material.)  A more important
disadvantage is that it is cumbersome and that, since there are no rules
for the syntax of its contents, these are less useful, potentially, than
they otherwise might be as paramaters for further processing.
 
2.  Where it is desirable to tag the sort of segments described above in a
rigorous and systematic way, the use of the <SPAN> tag is recommended:  eg
<SPAN END=XXX TYPE=THEME VALUE=LOVE RESP=DR> ... </SPAN>.  The "end"
attribute points to an anchor further on in the text;  the "type" and
"value" attributes can serve as parameters for further processing; the
"resp" attribute indicates the originator of the segmentation and makes the
point that it may be more of a matter of interpretation than, for instance,
a system of hierarchical divisions.  The contents of the tag are optional,
and should be used if required for further annotation of the feature in
question.
 
DAVID ROBEY