I'm happy with everything except point 3, my reservation being due to my
understanding of the Leiden conventions used by papyrologists. Of
course, the TEI should not be constrained by what one specialist group
does. Nevertheless, here is an attempt to explain my unease.
The relevant parts of the Leiden conventions may be summarised thus
(from "Note on the Method of Publication and Abbreviations" in recent
vols of the _Oxyrhynchus Papyri_); my suggestions for TEI conversions
are given as well:
a\.b\.c\. (\. means dot under the preceding letter.)
The letters are doubtful.
Approx. three letters remain unread by the editor.
The letters are lost, but restored from a parallel or by conjecture.
<gap reason="lost" extent="3"/>
Approx. three letters are lost.
<choice><abbr type="nomen sacrum">ΘΣ</abbr><expan>ΘΕΟΣ</expan></choice>
Round brackets indicate the resolution of an abbreviation or a symbol.
The problem arises as follows. As a papyrology novice I was commanded to
only use square brackets for something that is really lost. I remember
being told, "Even if the slightest trace of a letter remains, it belongs
outside square brackets." An unreadable trace is what the bare dots
"..." are for--letters unread by the editor. (Note: these bare dots
appear outside square brackets. E.g. time fl... li\.ke an arro[w].)
It seems to me that "doubtful" text and "letters unread" should both be
converted to TEI in the same way: using the <unclear> element with a
type attribute indicating what variety of unclear. This is because in my
mind the difference between the two categories is one of degree:
"doubtful" for any letter of which a trace remains that the editor
thinks is a good bet, and "unread" for such a letter that due to the
extent of damage could be anything. The trace may be insubstantial.
Nevertheless, if there is a trace, the transcription should indicate the
By analogy with the papyrological injunction against using square
brackets for anything but lost text, I reserve <supplied> exclusively
for letters that are lost. This goes against the TEI definition, which
allows <supplied> to be used for letters that are damaged as well as for
those that are lost. I wish that the TEI definition of supplied would
drop the "due to damage" part and only allow supplied to be used for
text that is well and truly lost, with not the slightest bit remaining.
So the cause of my unease is a dogged determination to reserve
<supplied> for text that really is lost. This is also the reason for my
distaste of <supplied> or <supp> to expand abbreviations.
On Sun, 2007-03-11 at 13:45 +0000, Lou Burnard wrote:
> I promised I would summarize the consensus on this issue, and to some
> extent my "strawman" proposal of 21 feb was meant to do that. For
> avoidance of doubt, however, here's what I am now proposing to implement
> in P5 with regard to this issue:
> 1. Make explicit that <expan> should contain the full expanded form of
> an abbreviation, not just a part of it.
> 2. Explicitly licence the Bodard Practice (this is the use of <abbr>
> within <expan> to mark the part/s abbreviated)
> 3. Make explicit that <supplied> can be used to supply three kinds of
> editorial addition
> (a) letters that are legitimately missing from the original e.g. within
> an expansion
> (b) letters that are accidentally missing from the original e.g. by
> scribal error
> (c) letters that are missing from the original because of damage or
> other non scribal reason
> I have invented the term "legitimately" to describe the first case: I
> hope it is clear what I mean -- if you could point the omission out to
> the scribe he would look at you a bit funny like and say "quid?".
> Whereas in the second case, he would probably say "Doh!" and slap his
> pate. Certainly there does seem to be a general agreement that the above
> three cases should be distinguished. There is less consensus on whether
> the distinction should be made by a type attribute or a distinct
> element, and if the latter what it should be called (the names <e> and
> <supp> were both proposed, but neither got much support, even from their
> I toyed with, and have not yet definitively abandoned, the notion of a
> generic "editorial intervention" element <ed> which could be used to
> delimit *anything* not present in the original source but supplied by
> an editor, without further precision as to whether it is the result of a
> correction, an expansion, a regularization or whatever. This would
> parallel the use of <hi> as a generic marker for "stuff that looks
> different but I haven't time or energy to explain why". And you could
> use it instead of <supplied type="a"> .
> I quite like this idea myself, probably on account of being incorrigibly
> lazy. I am not sure though whether it's a good idea to encourage this in