Michael Sperberg-McQueen writes :
>The TEI Applications Page, at
>    http://www-tei.uic.edu/orgs/tei/app
>currently includes information on over fifty projects which report that
>they use TEI for all or some of their text encoding.
...
>If you are using TEI and your project is not listed, please let us
>know,
 
Are the projects which FAILED to use TEI also expected to be listed ?
 
This is intended not at all as a sarcasm, but to serve as a source for
future improvements.
 
Here is an example :
 
A JSA (Japanese Standards Association) committee funded by MITI
(Ministry of Technology and Industry, Japan) has been working on a draft
for the standard of ``format for the information interchange for the
machine-readable dictionaries of Japanese input methods'', of which the
draft public review is scheduled during Oct 1 -- Dec 7, 1996.
``Japanese input methods'' are softwares for transforming Japanese
KANA-scripts into KANJI-scripts, or so-called `FEP', which is more
popular.
 
As a member of the committee, I submitted a draft based on the TEI
dictionary tags, but the draft could not be completed, and thus was not
accepted.
 
The reasons were clear :
 
  1. E-text dictionaries for `Jp input methods' need to be merged.
 
     TEI-based documents cannot be merged, because each doc can
     have its own-encoded DTD.
 
  2. In the to-be standard, the `Parts-of-speech' tags have a unique
     hierarchy, but the levels of minuteness (`granule') need not be the
     same between conforming dictionaries.
       eg. One dict. may have `proper-noun-for-humans', `proper-noun-for-
       locations', etc, and another may have just `proper-noun'. But, all
       understand `proper-noun-for-humans' comes under `proper-noun', and
       `verb-of-irregular-conjugation-SU-type' under `verb', not `noun'.
 
     TEI-based documents cannot assume hierarchy in ATTRIBUTEs.
     The hierarchy between attributes, ie. inheritance, should be spelled-out
     simply by copying by hand [TEI P3, 3.7.1, (p.59)].
 
  3. TEI does not guarantee that Japanese character strings
     can be used in TEI documents (or does it?)
 
     Where is the Writing-System-Declaration for the Japanese language ?
 
  4. TEI does not specify conformance, as is called for by Industrial
     Standards.
 
     eg. documents which contain mismatched `ids' (reference of
     non-existing anchors) are still `conforming' to the TEI.
 
  5. No validator was available. (still is true)
 
The (non-TEI) draft of this to-be standard is scheduled to be available by
Oct, 1st, and will be announced by then.
 
Readers on this list who understand Japanese are encouraged to contribute
to the public review process, and submit a TEI-based draft as an alternative
when necessary, which I was not able to supply.
 
Masayuki TOYOSHIMA / Dept. of Japanese linguistics, Faculty of Letters,
Hokkaido university, Sapporo 060 Japan, FAX +81-11-726-0919 (24h)
[log in to unmask], [log in to unmask]