Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watmath!clyde!rutgers!ames!sdcsvax!ucsdhub!hp-sdd!hplabs!hpcea!hpfcdc!hpfcrj!bayes
From: bayes@hpfcrj.UUCP
Newsgroups: comp.std.internat
Subject: Re: Computers and human languages (was Re: What is a byte)
Message-ID: <6990001@hpfcrj.HP.COM>
Date: Tue, 25-Aug-87 18:56:37 EDT
Article-I.D.: hpfcrj.6990001
Posted: Tue Aug 25 18:56:37 1987
Date-Received: Sat, 29-Aug-87 08:24:16 EDT
References: <717@maccs.UUCP>
Organization: HP/SSO Ft. Collins, CO.
Lines: 37

Back to Kanji:

As one who lived in Japan a number of years, and who has implemented
a Kanji utility on a computer, I have a couple of points to make with
some confidence:

o Japanese will not just "drop Kanji". When reading (and in an odd way, when
  speaking) Japanese, one thinks in Kanji, in some sense. Hiragana and
  Katakana can be used to represent the language, but they obscure, rather
  than illuminate the meaning. The fact that kana have been accepted in the
  past does not mean that kana are adequate or acceptable in a modern
  computing age.

  You might say:  "Well let them change from Kanji to
  Romaji/kana/whatever".  To which I might reply:  "The USA hasn't even
  gone to metric yet...".  Think what it means to change your whole way
  of representing your language and world.  Learning that 21C degrees is
  room temperature, or 2km is a 20 minute walk, is EASY by comparison.
  yet we're reluctant to make that concession to "modernity".

o There exists a standard for Japanese character representation digitally:
  JIS C 6226 (may have been updated). Any implementations had best superset
  that, as there's a lot of S/W and H/W around that now understands it.
  Shades of EBCIDC/360 compatibility.

o ALL the language must ultimately be representable, in some sense
  orthogonally. Saving a bit or 2 here and there, or forcing context-dependent
  decodings at the cost of performance and portability are false
  economies. So we need 32 bits. Big deal. Going to 21-1/2 bits or whatever
  has been suggested, or encoding only 90% of the language will just leave
  us with inadequate standards somewhere down the road (is this deja vu,
  or does my machine just not know how to accent the 'e' in deja :-)?

I cannot speak competently on Chinese, etc.

Scott Bayes
hpfcla!bayes