Bob Smart writes:
This suggests that Ran and Mark believe that Mnemonic does not intend to
handle the glyphs of all the Japanese alphabets. I have the clear impression
that Keld intends Mnemonic to handle _all_ known glyphs. Perhaps Keld
can enlighten us on this.
Please allow me to try to enlighten some of you.
Japanese can be converted to mnemonics in such a way that they are
quite readable (if you can read the 26 letters of the English
alphabet).
An example of such a conversion would be:
K1 K2 K3 -> Naka Hara Yasushi
K1, K2, and K3 are three Kanji (Japanese ideographic) characters.
The problem, however, is that you cannot convert back from "Naka Hara
Yasushi" to the Kanjis, because several Kanjis (say, K4, K57, K3001)
also map to "Naka".
One solution to this, is to put a number after the mnemonic:
Naka1
Now, (as if that wasn't bad enough), we have the additional problem
that many Kanjis map to multiple pronunciations (mnemonics). For
example, the "Naka" above (K1), can also map to "Chuu", which varies
case by case, i.e. depending on the context.
Keld, being a pretty optimistic sort of guy, keeps saying that he will
be able to come up with usable mnemonics for the ideographic
characters. I doubt it. Strongly. (And I should know. I speak Japanese
fluently, and read and write Japanese email everyday. I even live in
Japan.)
The way I read this it suggests that the Japanese don't just want their
glyphs to be transported, they want them transported in suitably
Japanesey bit patterns. Since that doesn't make sense I'm obviously
missing something again.
Why doesn't it make sense? If I sent this message in, say, EBCDIC, you
wouldn't be able to read it, would you? The choice of bit pattern is
pretty important. (Would that be an understatement? :-)
Now, having said that Japanese text must be in a particular form of
ISO 2022 for it to be readable, I should also say that ISO 2022 cannot
be used in email headers, because it can contain unquoted special
characters like '<', '(', '"'. (And quoting would distort the
ideographic glyphs.)
That's why the Japanese have been using ordinary romanizations of
their names in email headers (e.g. Yasushi Nakahara).
The extension that I proposed allows people to continue to put their
romanized name in the headers for the benefit of both
non-Japanese-speaking people and Japanese-speaking people with old
email software. The Smart/Freed proposal doesn't satisfy this
requirement.
Regards,
EvdP