RE: Transliteration

> I am open to suggestion regarding the mechanism/syntax.
> I prefer the _clean_ way where every bit of information is in
> his own little box and properly tagged.

Then define new attributes and/or headers for script, romanization
scheme, and other dimensions. _That's_ "cleaner" that redefining
and overloading "language").
 
> The reason for proposing the extension (over-loading)
> RF 1766 is because:
> 
>  - It feels like a natural extension (not too much overloading)
>  - Easier to implement
>  - Probably more acceptable to the standization community

It seems likely to do bad things if someone is actually trying
to select for particular languages or using the language/dialect
hierachy in some signfificant way. 

Consider what I'd have to do to select for Japanese language
regardless of how/if it was romanized.  Then consider the
problem of selecting a particular transcription for multiple
dialects.

If you re-use the same "variable" for different dimensions
things get messy.

--
    Albert Lunde                      Albert-Lunde@nwu.edu

Received on Thursday, 22 October 1998 10:04:58 UTC