[tlug] Re: IM conversion quality

["Stephen J. Turnbull" <stephen@example.com>]

>>>>> "David" == David J Iannucci <David> writes:

    David> On Sat, 7 May 2005 18:40:37 +0900, "Evan Monroig" said:

    >> I use "anthy" with under Ubuntu (a debian-based distribution),
    >> and there are also big problems of kanji conversions

    David> Isn't there at least one open source IM that is really
    David> good?

Not in the sense you're thinking.  Why would you expect one?  This is
exactly the kind of task where open source is going to suck.

    David> Is everyone who writes Japanese in the Linux world really
    David> suffering with this?

No.  For example, doesn't bother me a bit; actually, I'm indifferent
between Canna and Kotoeri, and prefer either to MSIME, slightly.
Haven't tried ATOK in ages, nor Wnn (unless one of those is what the
Zaurus uses; that's no better than Canna).  In any case, I can still
type Japanese as fast as I can think it; you can't get better than
that.  :-)

    David> It's hard to imagine that the native speakers in the OSS
    David> geek community would be willing to put up with the cr*p
    David> we're getting from Canna, e.g.

They don't.  Canna (FreeWnn, Anthy, Quail, SKK, SJ3) is tuned to their
usage, not that of gaijin nor those Japanese who just want to type at
the terminal.  The problems that somebody's wife was having with the
personal dictionary in Canna or Wnn don't happen to them, because they
long ago worked out the problem (file permissions or location,
probably) and haven't done a fresh install since 1994 or so.
Unnatural conversions don't happen to them, because they long ago
converged their personal nyuryoku style and the conversion precedences
in their preferred conversion engine, not to mention that their
personal dictionary is by now bigger than the source for the 2.6 Linux
kernel.  (I see plenty of incorrect conversions in my Winblows-using
colleagues' email, but I don't dare point them out---too much
haji-inducing to have a gaijin catch them!)

There have been a couple of efforts at improved dictionaries, for
example, the Shion project[1] to create an improved Canna-compatible
dictionary.  I don't notice much difference, though.

FreeWnn is claimed to have a decent grammar-based sentence-level
engine, but I've never heard anybody get really excited about it
(except the developers).


Footnotes: 
[1]  Debian has a package for this; I would imagine RPMs are
available, too.

-- 
School of Systems and Information Engineering http://turnbull.sk.tsukuba.ac.jp
University of Tsukuba                    Tennodai 1-1-1 Tsukuba 305-8573 JAPAN
               Ask not how you can "do" free software business;
              ask what your business can "do for" free software.