Input methods at the age of Unicode

Marcel Schneider charupdate at
Sun Jul 19 08:15:59 CDT 2015


thank you very much for this many information I didn't know and that is very useful to put into perspective the Windows Chinese IME new experience I referred to on the Mailing List.

Best regards,


> Message du 19/07/15 15:01
> De : "gfb hjjhjh" 
> A : unicode at
> Copie à : 
> Objet : Re: Input methods at the age of Unicode
forget to add Unicode maillist to reply address in my previous mail...add back and resend

---------- 轉寄的郵件 ----------
> 寄件者:"gfb hjjhjh" 
> 日期:2015年7月19日 上午9:38
> 主旨:Re: Input methods at the age of Unicode
> 收件者:"Marcel Schneider" 
> 副本:

the input method of type in the sound and pick corresponding characters have been developed for more than 20 years by many Chinese companies. Featues include prioritize characters to be selected according to usage frequency, if multiple sounds are input together without selection then it would provide selection of best fit vocabulary, with database constantly updating from network database, analyzing and personalizing its wordbank from social application, contact list, email, SMS and what you type, and if you input even more sounds together it can also give out candidates that fit natural sentence structure. And for those more commonly used characters or vocabulary, entering the first latin character of each letter's romanization is already enough for the input method to provide a list of best fit words, and thus saving typing time as each chinese character can romanize up to six or seven characters. It have also been developed that input methods have included some auto correction capability such that even if you have not master mandarin Chinese pronounciation and make some common mistake durung romanization, the program can still understand what you want to type. And on the other hand for  increasing speed, as typing each chinese character directly by their romamization often involve typing up to 6 characters, people map each vowel and each syllables into individual keys so that only 2 key strokes is needed to press before people start selecting which characters they want. However, as all the above mentioned methods involve body-eye coordination to select word they want, those who really emphasis speed would stock to some older input methods where they decompose characters base on glyph's shape, convert that into a series of string which if designed properly those string can be unique to most of the characters, or even if it really come down to repeated code or when you are using a scheme that uses shorter code which yield higher repeat rate, people would memorize the candidate # as part of the string so that they can type without looking at the screen. The typing speed using such method (with regular keyboard) have been recorded at more than 220 characters per minute which have already exceeded the Chinese national standard for stenographer that utilize specialized stenotype machine. On the other hand it appears that some Chinese stenotype machinese [in mainland China] used sound of characters to type just like those mentioned at the beginning, and some of them even used an application that compatible with the one used in desktop environment... So it's hard to say if it help or hinder the typing speed by letting typer rely on visual hint...

2015年7月19日 上午4:39於 "Marcel Schneider" 寫道:

> On 18 Jul 2015, at 17:30, Eli Zaretskii  wrote:
> > > Date: Sat, 18 Jul 2015 16:33:23 +0200 (CEST)
> > > From: Marcel Schneider 
> > > You might wish also to use the Windows on-screen keyboard which allows to see
> > > what's exactly on each key while typing on whatever physical keyboard, without
> > > any need to have the keycap labels match the layout. This on-screen keyboard is
> > > built-in, only it does not support Kana shift states.
> > 
> > That makes typing much more slow, unless you already know, at least
> > approximately, where the keys are. you are talking to someone who is
> > almost touch typist in English, but cannot remember for the life of me
> > the Russian keyboard. Transliteration is the way to go in such cases,
> > and it's strange that transliteration-based input methods are not
> > readily available on Windows out of the box.
> The Chinese IME new style is a very smart tool based on transliteration. You type just the syllables like they sound in English, and you get plenty of suggestions among which to choose. There is still the Chinese old style IME shipped with, too. I don't know Chinese so I can't tell more but visually I believe these tools are very performative. Perhaps for Russian no transliteration based input tool was built for Windows because we are meant to use the keyboard straightforward. Now, the osk.exe should probably include on each key picture the letter that is on the current physical keyboard. That is what I often missed on such UIs, that you cannot make the link with the base layout as the user knows it. I will say, too, that when the OS is in Russian, the OSK should display cyrillic letters following the Russian keyboard when the OSK displays a QWERTY keyboard layout. As you can have the OSK always above, you just look at it and see the keys you're striking.
> There is also the old solution with a keymap on a paper. You can open the Russian layout in the MSKLC, choose a nice font, font-size, window size (to get square keys; don't let the default rectangles), nice background colors. Then save it as a picture, in the File menu > Save as image. Open this in Paint or Gimp and add the Latin letters. 
> Marcel

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the Unicode mailing list