Keld says:
The chinese GB standard has been tabled with chinese romanisation
and the japanese with japanese romanisation.
Also JIS X0212 has been tabled with chinese romanization.
Which westerner's romanisation for Chinese ? Which dialect did you
use for the romanisation ? How do you handle the many homonym glyphs
that all map to the identical character ? What about the BIG5
encoding for Chinese which is in wider current use than GB ? How are
you handling Conventional versus Simplified forms ? The differences
are not entirely in the font. How will a user be able to take the
romanisation and know which homonym is meant ?
The several dictionaries of Chinese that I use have different dictionary
numbers for the same character.
We have chosen two of those dictiotionaries for the numbering
and have each of the charactes coded with those two numbers.
This isn't very useful since different dictionaries use different
numbers. In essence, it makes the mnemonic draft specific to two
specific proprietary dictionaries. That isn't helpful.
What is meant by "pattern description" ?? Do you mean radicals ?
If so that is unlikely to work also. Research done in Asia and at
places like PARC have shown that the existing radical-encoding schemes
all have many-to-one mappings in parts of their encoding tables. That
isn't likely to be useful.
Pattern description is not radical/stroke. We do have radical stroke
too. Pattern description is based on the left downward corner of the
character, as far as I know.
If it is based on the left downward corner of the character then that
sounds like the WANG schema of some years back that had the same problems
with many characters looking identical in the corners. Again, not useful.
Radical encodings have the same problems with lacking 1-1 mapping.
It seems to me that there is still a gaping chasm (though Keld does
seem to be working hard to close it) between the mnemonic
representation and a useful representation. I remain skeptical that
the gap can be narrowed sufficiently to make mnemonic really useful in
the CJK area. The lack of a 1-1 mapping between mnemonic's
romanisation and the character is very serious and it is not clear how
that is ever going to be closed. (This ignores the fact that many
Chinese users, especially native speakers, have much difficulty
transforming romanisation into characters.)
Ran
atkinson(_at_)itd(_dot_)nrl(_dot_)navy(_dot_)mil