Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!


Forgot your password?

Japanese Input Support For Western OSes? 19

RobM asks: "I'm currently studying Japanese, both for fun (Anime) and for profit (possible employement in Japan). To be able to better learn the language, I was planning to use my PC to read Japanese pages and then to manually translate them using one or more of the open Japanese dictionaries available. To be able to input the Kanjis I have to add to the system a Kanji input method and the related software , but none of the western distro has support for this. Moreover, there are a lot of tools for similar tasks, many of which are poorly documented and/or very old." How do the more common Operating Systems (Windows, Linux, BSD, etc.) compare when it comes to Japanese support?
This discussion has been archived. No new comments can be posted.

Japanese Input Support For Western OSes?

Comments Filter:
  • I don't know for japanese, but for the Win32 Platforms with IE (5.0 or 5.5) there are additional input methods (such as pinyin, etc) to type chinese characters. Just install the correct IME (it shoud go with IE), and select it from your keyboard prefs.

    (Works fine with win2k German and chinese keyboard ;)

    Samba Information HQ
  • Apple MacOS

    In MacOS 9, extra language packs can be installed using the cusom install option in the installer on the MacOS CDROM. With the Japanese language pack, you can display and enter Japanese text in various applications. SimpleText localized in Japanese is included in the extras folder on the CD, also. Previous version of the MacOS (to 8.0, I think) included 'Multilingual Internet Access' - basically, Japanese fonts, but no Japanese input method.

    Microsoft Windows

    In Windows 2000 (only, I believe, correct me if I'm wrong), the Japanese language pack can be installed in the initial operating system install. The Microsoft IME has various useful features, including the ability to draw a Kanji character on the screen and have it converted the appropriate actual character. Japanese language support can be installed in older versions of Windows, but I'm not familiar with them.


    Using a combination of tools like kterm, kinput2, and various dictionary servers, you can happily input Japanse in UNIX too. A good resource to look at is the Debian GNU/Linux task-japanese [debian.org] package. In that link is a list of sub-packages, which you download for various other UNIXes too, other than Debian GNU/Linux (from seperate websites, of course).


  • by fosh ( 106184 ) on Monday January 08, 2001 @06:04PM (#521666) Journal
    I am not sure how far along you are in your study of japanese, so this may be new, or not.

    In Japanese, every "sound" can be created by stringing together charectars in the core alphabet (also known as hiragana) which consists of 40 something charectars. Kanji's are pictures which are equivelent to one or more hiragana, depending on thier context. For example:

    The Word JAPAN is: (individual hiragana charectars are seperated by spaces)
    ni ho n

    It is comprised of two kanji's. Sun, and book.
    Sun == ni
    book == hon

    I.E. Japan == Sun (Kanji) + Book (Kanji) == ni + hon.

    However, there is one little wrinkle here, the kanji's can change thier pronunciation (and sometimes meaning) depending on thier context.

    I.E. Monday
    Monday == Sun (Kanji) + Day Of Week (kanji)

    But, in this case,
    Monday == ni chi
    (in case you are curious, Day of Week == yobii)

    To remedy this situation, most westerners input japanese by inputting the sounds, and then choosing kanji's for them once they are input. (The computer has a database of hiragana->kanji).

    This is do-able through many many packages for linux and Windows. Check out the MULE package.

    As for looking up Kanji's you see, this is done by stroke order and radicals. The Japanese language is extremely modular, and mosr complex Kanji are made up of sevral radicals. When looking them up, you give the dictionary a few radicals and it gives you a list. There is an excellent palm program for this (Called Hanabi,) and sevral linux programs, such as gjiten and jdict.

    Finally, there is an extremely cool Kanji handwriting recognition project for linux called kanjipad. It is a stand alone program as well as a widget, However it is very very difficult to draw Kanji with a mouse.

    Good Luck in you Quest! (ganbatte)
    --Alex Fishman
  • Sorry. Formatted the links wrong. 8 P

    Urls are, respectively,
    Jim Breen's Japanese Page-
    and the Monash Nihongo ftp archive-
    http://ftp.cc.monash.edu.au/pub/nihongo/00INDEX. ht ml

    Hope this helps,
  • First, the online reference of choice for Japanese computing...Jim Breen's web site [monash.edu.au] and ftp archive [monash.edu.au] (or a mirror [monash.edu.au]).

    For western Win32 versions before win2k, i think you are limited to things like this hack [microsoft.com] from MS for IE5, which allows CJK (Chinese/Japanese/Korean) input in IE5, Word2000, and Outlook, or to using apps which handle the language on their own.

    For X (should apply to most versions/most *nix variants), you need fonts (included with XFree86, possibly in one of the optional font packages), an Input Method (IM) like kinput2 and a kana-kanji conversion server like Canna or WNN. You probably also will need localized variants of many apps, kterm for xterm, jvim for vi/vim, etc. Recent Emacs should work, look for 'International Character Set Support' in the emacs info pages for details. Mozilla handles display fairly well, though i haven't messed with setting it up for input yet. XJDIC (or a variant) rounds out the list of things to get, providing a nice Japanese-English and English-Japanese dictionary.

    Kon2 handles using CJK from the linux console (also possibly BSD and others), haven't used it in a while though, so don't really remember how well it worked.

  • by Anonymous Coward
    Its true that the HON kanji can mean book and it is also pronounced the same. However, in the case of NIHON, the meaning of HON is origin or root. So NIHON = the origin of the sun Chris
  • To expand on my problems (and on my level of knowledge), I know a little japanese (some grammar, hiragana and katakana and very few kanjis).

    I'd like to be able to use a computer dictionary to speed up my 'reading time' of a Japanese web page, and with that improve my understanding of the language using real world examples of it. Using the dictionary I can copy-paste the kanji in the right place in the dictionary program, and avoid the long search time on a paper dictionary (I do have a very good memory for symbols, but I need to be able to do something "interesting" while studying, or I get bored :)

    I've already tried kinput2 on a SuSE 6.4 with KDE 1.1, but with little success: when the kinput2 is active and configured (I've used cannaserver), nothing happens even with the kterm (self compiled) or with the Kanji dictionaries.

    Maybe SuSE's X isn't compiled for extended input methods? Or maybe kinput2 doesn't work with XFree 4.0? Guess I'll have to check specfiles and docs for X.

    I'll also check the various link you provided me for answers to my dilemmas ;), many thanks again!

  • I too have been striving to get japanese support working on my computers. The best luck I have had is with BeOS. I bought the "Pro Edition" and at install time I selected the japanese input system which works brilliantly. I have also gotten it to work with FreeBSD. One of the CDs in FreeBSD has *tons* of japanese programs, inluding a half dozen different input systems. I had the best luck using WNN as the server and kinput2 as the client in kterms. I don't have the URL handy, but Craig Oda has a great page about how to get japanese to work with linux. I have had no luck so far (on SuSE 6.4). X, or something seems not to be compiled with japanese support, so when I start GNOME in japanese, I get ascii garbage instead. And, I 've never gotten any server to work (and generally, not even compile). I've heard that TurboLinux, and Kondra have the best Japanese support.
  • On Linux, using KDE2 will give you the ability to view Japanese in Konqueror and other apps. To get typing ability, you need kinput2 and whole bunch of other stuff. Kondara MNU/Linux is an good distribution if you want English and Japanese support out of the box. TurboLinux and RedHat both have Japanese versions but I don't recommend these unless you can read Japanese pretty well already.

    Win2K is the best platform right now for multilingual support (particularly) Japanese. I hate to say it - especially since Win2K blue screened on me today at work (all I did was stick a CD in the drive!) But if you can deal with a blue-screen here or there it works.

    BeOS also works extremely well.

    On Win9x and NT 4, get the Microsoft Input Method . This will let you enter Japanese into just about anyplace where you're using HTML. Works for IE/Netscape, Outlook Express, FrontPage, etc...

    I've never tried using Japanese on BSD, but for what it's worth, there is a pretty lively BSD community in Japan. I have yet to meet a Linux person in Japan.

    Suggestion for you if you happen to live in the USA - find a Korean grocery store in you area and see if they rent tapes of Japanese tv. Immerse yourself in the wonders of Japanese television for a few days.

    Good luck!
  • Why NJStar is cool:

    1) Allows me, a Sino-geek with a strong background in Hanzi [or Kanji to those on the islands] but little knowledge of Japanese to communicate ideas in characters to people on Japanese machines. Ditto for Korean. Ditto for

    2) Broad support for Chinese/Japanese/Korean entry systems. IMEs work across character sets, as the set permits. (see point 1)

    3) Works for browsing under almost all Windows apps (Microsoft Asian Language packs seem to be confined to OLE-enabled apps).

    Why NJStar sucks:

    1) No TrueType font support.

    2) CJK rendering is sometimes a bit crufty.

    For "thinking of cool stuff" I'll give it four-and-a-half out of five. For implementation, I'll give it three out of five.

    _FLAME ON_

    Since this is Slashdot, and since I really hate MS on this point, let's kick them around a bit: I have had a hard-on for MS's implementation of CJK ever since I was working on East Asian Win95 testing in the Hive back in 1994. While implementation of the Language packs in Win 98 have fixed some of these issues, MS is (IMHO) taking an easy problem (rendering CJK characters, which share a lot more commonality between the various languages than differences, once you see past the details of which numeric value is assigned to which character), and making it hard. There are still FOUR seperate language packs, which do not work together with each other at all. When I read Guo2 Biao4 (PRC character encoding standard) content from Mainland China, Microsoft forces me to read it in simplified characters. Grrrr. If I am writing in Big-5 (Taiwan character encoding standard), I can't write in Pinyin. GRRRRRR!


    Gawsh, doesn't MS's implementation of CJK look purty?
  • In all my use with Chinese materials, I've always found NJStar to be the best company to use (for the word processor _as_well_as_the "internet viewer" or whatever they call it now.) If you are using windows, I'd strongly recommend it.

  • I took Japanese for four years in my high school.

    I often did my homework using njstar word processor. http://www.njstar.com.

    You type the romanji and it converts it to hiragana, katakana or kanji. Lets say you typed 'nihon.' The hiragana appears, and all the possible kanjis for that phonetic appear and you press '1' for the first kanji, 2 for the second, etc.
  • If you want to read _and_ write Japanese, there's one solution that tops them all - the atok conversion system. It's octane-enhanced, high-performance kanji/kana henkan makes conversion from romaji (alphabet) characters a breeze.

    Atok comes in two flavors -- bundled with Ichitaro, a windows-based (bulky) word processor, and as a separate program to be used in UNIX environments.

    If you're using Linux or FreeBSD, you'll use atok as part of a triumverate of three programs -- atok, kinput2x, and kterm. Throw in mule, the multilang version of emacs, and you're set.

    Getting the three programs to interoperate is not for the easily daunted. You're best bet is to buy a Japanese linux distro --Turbolinux or Redhat -- and do a full install. In most cases, the install will set up your rig so that the programs operate okay.

    Note that kinput2x is a version of kinput2 tweaked to operate with atok, and not Wnn (Wnn is an inferior conversion server with little of the ability of atok). Kinput2x has a bug that prevents backspacing in kterm in conversion mode -- the only way to beat it is to set the LOCALE variable to ja, for Japanese.

    This means that basic man pages, such as "ls" or "ping," will automatically spew forth in Japanese, and the menues of gnome and KDE will also appear in that language.

    Looking for an excuse to rev up on basic computer terms in Japanese? Install kinput2x, atok, kterm, and set the LANG variable to ja. Mind blowing linguistic fun awaits . . .
  • BeOS has a Japanese language option, which uses the type in romaji/pick your kanji method. It's not bundled in with the Personal Edition, but you can get it on their site.

    Jim Breen's Japanese Page annotates a lot of useful links, especially for software and language learning. He also maintains an ftp archive which mirrors stuff for most platforms (Amiga and Newton, frex).

    If you're on Windows, I recommend JWPce. It uses the Edict files so you can lookup meanings, readings, etc. as well as type bilingually. It also opens/saves as all the common Japanese encodings, plus Unicode.


  • when i was a PC boy i used s/w found here:


    of course now i am a Mac geek all the stuff is built in hehe

  • One possibly non-obvious thing about using kinput2/kterm is that you need to hit a special key sequence to start kana-kanji conversion, i think it defaults to shift-space...

    Another problem with using it for web pages is that (at least the last time i tried) mozilla doesn't seem to want to copy kanji to the clipboard to paste into the dictionary program.

    The 2 solutions i've tried are www.rikai.com [rikai.com] as someone else mentioned, which will grab a page and add nice little javascript popups to all the kanji, and using w3 mode to read the web page in emacs, and using edict.el for dictionary lookup(or cut and paste into xjdic). Both work fairly well, rikai is probably a bit easier to use, but working in emacs seems a bit easier if i want to chop out a difficult chunk and make notes as i try to read it....

  • > Microsoft Windows
    > In Windows 2000 (only, I believe, correct me if
    > I'm wrong)

    Well... Win 95/98 don't have the IME to begin with, but if you install IE 5 the option is there. Once installed it works fine with Office2k

    I did once have a problem where although the IME existed and allowed me to type in kana and kanji, nothing displayed them (I just got little black boxes). Then, suddenly it randomly fixed itself.

    Oh, and the IME only seems to come with certain versions of IE 5 (5.0 and 5.5 but not 5.01 or something like that!!).
  • As someone else pointed out, you can download the Japanese IME for Windows 98 from Microsoft. It only allows you to input japanese text on web pages though, so it probably is of little use to you. (Information on the Windows Global IME [microsoft.com]
    I've had the best luck using Windows 2000 - when you install the system, you can install a Japanese IME. If you then set your regional settings for Japan, it is also really easy to copy/cut/paste kanji, ICQ works right, etc. etc. It is pretty nice.
    Of course, if you don't want to bother with getting IME input to work right at the OS level, you can get multibyte support for ntemacs [washington.edu] up and running, which does have its own Japanese (and Chinese, Korean, Thai, etc.) IME.
    If you are interested in reading Japanese web pages, then you will probably love www.rikai.com [rikai.com]. The site uses a server side script to load in edict readings [monash.edu.au] for each kanji which pops up on mouseover.
    That's about all I have to say about that.
    BTW, if you are interested, I'm currently translating the Great Teacher Onidzuka manga [columbia.edu] using a Win2k system with Japanese input (along with a Java client / server thing)
  • The last company I was with used Japanese speaking/reading people fairly extensively and on the Win32 machines they were working on we used AsianSuite 2000 from http://www.unionway.com/ [unionway.com]. It provided an input method that was almost identical to the input method used in the Japanese version of Windows98. You could input hiragana, katakana, and kanji characters using the software. I believe they even have a demo available at their website.

UNIX enhancements aren't.