I lurk anonymously behind James's original post on MEI, since it was I
who mentioned this to him --- MEI was talked about at a planning
symposium organised by the Digital Archive of Medieval Music
(<>) in Oxford a week ago, and I was rather struck by
what I saw.  I am also responsible for the Cursus project, referred to
by James, in which he and I worked on xml encoding of medieval
liturgical manuscripts.  So I am very pleased to see Perry's
contribution to the thread.

In the case of our Cursus work only one Ms of the corpus we were
dealing with had music-notation, so it was not a burning issue.  But
as James remarks, liturgical manuscripts are very frequently sources
of music, and I remember thinking that we had made editions in which
it was possible to quiz up to a dozen-and-a-half Mss about their
precise readings of something like 7000 verbal texts of chants, thanks
to the employment of TEI principles, but that there would have been no
readily accessible way of doing a comparable job for the melodies too.
That is a rather down-to-earth reason for wanting a music-encoding
system that can slot into a text-encoding one, though it is not
unimportant: if one has (say) N sources for antiphon X, it would be
nice to check if they all use the same melody, and how they deal with
fitting their textual variants to the music-notation.  But beyond
that, the parallel incidence of verbal and musical text is not
unusual, whatever the period or source-type, and the provision of
techniques for encoding both elements in digital corpora might go some
way towards correcting the situation --- rather commom, I'm afraid ---
in which textual scholars look at the words, musicologists look at the
music, and the two do not much meet.

My own primary interest is medieval, where as Perry says there are
particular issues.  I would perhaps take issue with his remark that
"at some point ... this early notation is not really music notation at
all but an annotated text".  The implication here seems to me to be
that there is a definable 'thing' which we can call music-notation and
from which it is possible to diverge.  On a global scale as well as an
historical one, music-notation seems to be a bewilderingly variable
phenomenon, and one which has a relationship to possible sound-events
that is fundamentally unpredicatable.  The usual explanation involves
the reliance of notation-systems on unwritten traditions of
performance-practice.  That is undeniably (and in a few cases
demonstrably) true, though I remain sceptical that notations
consistently have a prescriptive relationship with the realisation of
sound-events as their primary (or at least sole) purpose.  Perhaps it
would be more useful to say simply that we have graphical signs, much
as words are, and not to bother too much in the encoding with what
they might 'mean'.  Even with adiastematic neumes there is _some_
information we can encode.  We should perhaps not forget that
neume-forms have since the Middle Ages had names, even though these
are sometimes problematic; and there is a modern vocabulary for
classifying types of neumatic notation, moderately well-developed in
French and German though pretty inadequate in English.  All this could
be a starting-point.

In a rather cursory look at the MEI files during the last week I have
been struck by its potential, not least because it allows one to cut
straight to essentials (inherent over-complexity seems to me something
that has bedevilled music-encoding attempts).  I can even envisage
ways in which it could accommodate quite 'unconventional'
music-notations.  I would be very interested to see some experiment in
in using it in parallel with TEI-encoded text.

Professor David Chadd

University of East Anglia, Norwich NR4 7TJ, UK

All Souls College, Oxford OX1 4AL, UK