In linguistics, examples, glosses, or parts thereof are
often prefixed with symbols like "*", "?", "?*", "(?)", "#"
etc., specifying the type or degree of (un)acceptability.
(Unfortunately, there is no consensus as to the
inventory and exact interpretation of these symbols.)
These specifications could be easily represented in TEI
by an optional "accept" attribute on <mentioned>,
<gloss>, and <seg>. The value of the attribute should
directly give the symbol instead of some fixed
meta-value like "yes" or "no".
Logged In: YES
user_id=1021146
Intriguing suggestion! Would you not also want it for <eg>,
<q>, <cit> etc? though, e.g. to mark spurious sentences and
quotations?
The certainty module provides attributes which might be of
relevance to this application. Again, if the only purpose of the
proposal is to affect the rendition of the output, I think the
REND attribute is a better example.
Logged In: YES
user_id=950793
The purpose of the proposed "accept" attribute is to encode
a meta-linguistic acceptability statement on some linguistic
entity. Of course, an acceptability statement is more than a
rendition specification of the entity it applies to (but see
below).
Unfortunately, linguists often record acceptability statements
by vague and partly undefined marks, though. While "*" is
usually interpreted as 'unacceptable', the interpretation of
other notations like "?", "?*", "(?)", "#", "%", "**" etc. are
rarely specified explicitly.
Thus, someone encoding or quoting a pre-existing linguistic
text with undefined acceptability marks has to encode their
original form--i.e. their original rendition.
The proposed "accept" attribute could of course be applied
to additional elements besides <mentioned>, <gloss>, and
<seg>--namely to all elements representing some linguistic
entity, including rule-like entities (recall the Generative
practice of attaching acceptability marks--or, rather,
grammaticality marks--to rules or parts thereof).
By the way, the <seg> element may be useful for marking up
acceptability statements with a restricted scope, for
instance:
<mentioned>Bill loves/<seg accept="'*">love</seg>
Mary.</mentioned>
Logged In: YES
user_id=686243
While the whole idea is new to me, my first reaction is that
this is a fine idea, and Lou's right, should probably
include quite a few elements in the class of elements that
gain this attribute.
On the other hand, it is worth spending at least a little
time and effort considering other mechanisms for acheiving
this goal before settling on one method. Possible examples
follow, not at all thoroughly thought through, rather just
thrown out to demonstrate that there are a lot of possibilities.
* A new <acceptability> empty element which would operate
much like the current <certainty> (and which might fit well
into the certainty module) which would bear target=, given=,
degree=, and desc=. (And rend= for the actual symbol; or
perhaps the symbol should be the value of degree= if a more
precise degree is not known.)
* A new <accept> element which operates somewhat like <sic>,
but has degree= to indicate the level of (un)acceptability.
* Using ana= to point to an element that describes the
elements (un)acceptability.
It also occurs to me that this problem isn't so dissimilar
to the problem of wanting to demonstrate how *not* to do
something by putting incorrect grammer, erroneous computer
code, or (perhaps the worst for us) invalid XML into an <eg>.
Logged In: YES
user_id=1021146
We agree that this should be investigated further, and
probably involving definition of a new attribute class in the
certainty module.
Logged In: YES
user_id=686243
Originator: NO
While we all seem agreed that this is a good idea, it is clear that more thought & effort are needed than we will be able to muster for the 1.0 release of P5. I have therefore opened a ticket (#299) on this subject over at trac (the TEI's new place to keep track of P5 progress), and scheduled it for version 1.1. The Sourceforge version of this ticket can now be closed.