[openbeos] Re: Messages Translation ?
- From: pascal@xxxxxxxxxx
- To: openbeos@xxxxxxxxxxxxx
- Date: Wed, 26 Jun 2002 20:40:52 -0700 (PDT)
On Thu, 27 June 2002, "Matt Verran" wrote
>
> Looking to expand my knowledge here and play devils
>advocate! Two good friends of mine, one from Kobe and
>another who just lived there for a while say Kanji is
>one of many written forms of the Japanese language,
>which raises the point, what about the others?
Well, this sounds quite off-topic, but...
I don't know exactly what he wanted to tell you, but
as for writing, there are not many, there are three.
One writing form is the kanji, as you mentioned.
There are also two syllabials of about 50 characters
each (46 to be accurate, and a few extra that are not
used).If you count the accentuations, you end up with
2 sets of syllabs of about 100 characters.
As for the kanjis, 1900 characters in the education
ministry's official list (+45 extra characters).
With that, you can basically read newspapers except
place names and person names that may be off-list.
Maybe your friend wanted to tell you that a single kanji
can have many readings?
[other author from here ]
>> A rudimentary knowledge of japanese is really all
that
>>is necessary,
Errr, you were describing an input method, right?
A rudimentary knowledge of Japanese can only make a
rudimentary input method. But this has nothing to
do with the localization itself. Localization of
a program can be made on a machine that has an
input engine and exported to a machine that doesn't.
Localization of menus and messages, and input method
are separate issues.
>>since you won't have to compile a dictionary of
>>japanese words, or anything. Actually, the dictionary
>>that ships with BeOS is amazing; it includes
place-names
>>for Okinawa, which really has its own language and
>>has sounds that aren't "japanese".
I would be curious for these non-japanese sounds in
Okinawa.
Can you give some examples?
As for the daily use, the input method in BeOS cannot be
used. It is basically a word-by-word method, which is
inefficient.
Example: try to type with Be's input method:
atsui fuku // warm (thick) clothes
atsui ocha // hot tea
atsui tenki // hot weather
The kanjis for atsui should be all different. You can
type
that with BeOS, but a good input method gives the right
"atsui" at once because there is some kind of semantics,
or a huge database, I don't know, that somewhat
"understands"
the context. I just made the experiment on BeOS with
EGBridge and the 3 different "atsui" were right at once.
Basically you type a whole sentence, press the space
at the end of it, and it should be right at once. But
in the real world it still can be stuck in some cases.
Note that I am not saying Be's input method is bad.
I think Hiroshi did a good job when he built it.
So there is an input method, and it works, which is
better than no method.
But when you use it, you have to deal word by word, and
it's really a pain. No surprise that a god input method
costs a couple of hundreds of dollars by itself.
Pascal
Other related posts:
