I've given a lot of thought to transliteration and transcription at the glyph level:
Which comes back to the issue of ciphers. It would seem to me that glyph-level transliteration is the accepted behavior for ciphers (else we would actually have to address whether such things as Theban should be encoded, and Braille would have been a non-issue from the get-go). What determines whether a script is a cipher of another?
I offer this:
Any sign can be made a cipher by changing the signified. Writing systems are collections of conventional signs, which means that there is conventional agreement as to the signified. For example, the signifier 'A' is conventionally agreed by users of Latin script languages to signify the signified 'Latin uppercase letter A'. Users of Greek, Cyrillic, etc. writing systems conventionally agree on different although historically related sign relationships for the same signifier: so the Greek Alpha is not a cipher of the Latin A, because its principle conventional association is with a Greek letter. A cipher occurs when either 1) a signifier is associated with something other than its conventional signified, or 2) the signifier is associated with a signified that is conventionally associated with a different signifier. The Theban cipher is an example of the latter: these are a collection of signifiers that are associated with signified that are conventionally associated with other signifiers, e.g. 'Latin uppercase letter A'.
In Unicode terms, one could say that Unicode encodes what is signified as characters, and these are signified, conventionally or otherwise, by glyph signifiers. So there is no point in encoding the Theban cipher because its signified are already encoded as Latin characters.
John Hudson
Tiro Typeworks www.tiro.com Vancouver, BC [EMAIL PROTECTED]
I sometimes think that good readers are as singular,
and as awesome, as great authors themselves.
- JL Borges
