Input methods

In order to input text in any particular language and writing system, we need a Unicode font to display it in, a rendering engine that knows how to display it, and a keyboard layout or Input Method Editor (IME) that provides a way to get all of the needed characters. Most alphabetic and syllabic languages can be typed on fairly simple keyboards that produce one Unicode character per key combination, using the ordinary typing keys together with Meta (usually Alt) and Compose keys. Any of several keys, including Menu and Windows keys, can be set to act as Compose. Then on Latin keyboards Compose-a-' produces á, Compose-c-, produces ç, and so on. Any accented letter that is included in Unicode in precomposed form falls within this capability. This covers letters that occur in any widely-used pre-Unicode character set, such as Latin-1 (ISO-8859-1), which supports French, German, Spanish, Italian, Scandinavian languages, and some other languages that use only the accented letters in Latin-1.

Multiple diacritics can be entered sequentially on simple keyboards of this type, while more elaborate input methods can enter more than one Unicode character code into the input buffer for each key combination. Yoruba is an example of a language that poses this choice, because it has vowel letters with an acute accent above and a dot below that are not available precomposed in Unicode.

The most elaborate IMEs are for input of CJKV characters for Chinese, Japanese, Korean, and the historical Vietnamese Chu Nomh writing. Each of these languages requires several thousand characters at a minimum, and there is a desire to have much more extensive CJKV sets available, including a number of Hong Kong characters and other recent additions, or the tens of thousands of historical characters important for scholarship.

Several hundred methods for entering CJKV characters have been invented over several decades. Among the most important (due to efficiency of use or ease of learning, or in a few cases both) are language-specific phonetic conversion systems for Chinese, Japanese, or Korean, and shape-based systems that are in principle independent of language, but in practice specific to particular countries up to now.

See also countries, languages, writing systems, fonts, locales, and keyboard layouts.

Tools

Tools for keyboard layouts, to come. loadkeys utility to load keyboard layouts.

Tools for IMEs, to come.

Input Methods

Phonetic conversion

The concept of phonetic conversion is that any CJKV language typed in any alphabet or other sound-based writing system can be converted using a combination of dictionary lookup together with grammatical and semantic analysis. The first successful phonetic conversion word processor was the Xerox 8010 J-Star, an outgrowth of the Xerox Alto computer and Smalltalk programming language in 1981. Thanks go to Alan Kay for the Alto and Smalltalk ideas, and to Joseph Becker for the language handling software. Phonetic conversion to CJKV characters exists for the following combinations, in many variations.

• Romazi (Latin alphabet) or Zhuyin to either Traditional or Simplified Chinese hanzi 漢字

• Romaja 로마자 (Latin alphabet) or Hangeul 한글 Korean alphabet to hanja 漢字

• Romaji ローマ字 (Latin alphabet) or hiragana ひらがな syllabary to Japanese kanji 漢字

Phonetic conversion systems depend on a native alphabetic or syllabic representation, or on one or more Romanizations of the target language.

• Chinese: Pinyin 拼音, Gwoyeu Romatzyh 國語羅馬字, Wade-Giles, and Yale are a few of hundreds

• Japanese: Hepburn, Kunrei-shiki, Nippon-shiki, Yale

• Korean: McCune-Reischauer (MR), Revised Romanization of Korean (RR), Yale

(Yes, the Yale Department of Linguistics was busy on the issue for decades.)

Shape-based

It has been evident to everyone who has studied Chinese characters, including ancient Chinese scholars, that one can analyze most Chinese characters into a variety of smaller parts, many of which are characters themselves, and further, into a number of specific brush strokes (or the corresponding elements for characters cast in bronze, carved into seals, scratched into oracle bones, and so on). Many attempts were made over thousands of years to analyze the character shapes for various reasons, among them dictionary making. The best known is the system of 214 radicals set forth in the 1615 CE Zihui (字彙 "Character Glossary"), edited by Mei Yingzuo (梅膺祚) during the Ming Dynasty, and made a de facto standard by the 1716 CE Kangxi Zidian (康熙字典 "Kangxi Dictionary"), compiled under the Kangxi Emperor of the Qing Dynasty.

In the computer age, numerous analyses of character shapes were made as the basis for methods for typing Chinese, including those described below, which remain in widespread use, and others. As with the radicals, shapes are generally somewhat variable. In shape-based IMEs, several shapes may be grouped together and assigned a single key. Different shape-based IMEs are usually used for Traditional and Simplified characters.

Methods by language

Traditional Chinese

Zhuyin conversion

Zhuyin 注音, or Bopomofo ㄅㄆㄇㄈ, is a Chinese alphabet used for teaching children as well as for typing Chinese input. It has multiple keyboard layouts. Zhuyin is one of the standard conversion methods for Chinese, appearing on almost all computers and some cell phones.

[Wikipedia: Zhuyin]

[Unicode: Bopomofo] code table (PDF)

Phonetic conversion method

Pinyin conversion

Phonetic conversion method

Cangjie

` 1 2 3 4 5 6 7 8 9 0 [ ]
   手 田 水 口 廿 卜 山 戈 人 心 ［ ］ 、
    日 尸 木 火 土 竹 十 大 中 ； ‘
     z  難 金 女 月 弓 一 ， 。 ／

Shape-based. The Cangjie keyboard layout has 24 simple Chinese characters on it, plus a key for "difficult" 難 characters. All of the common brush strokes and many common combinations are mapped to the 24 base characters. Characters are analyzed into 1 to 5 elements that can be coded in the base characters, according to a moderately complex set of rules.

Four Corners

Shape-based. The corners of a character are encoded, and a code for the whole character created from them.

etc.

Simplified Chinese

Wubi

etc.

Japanese

Kana conversion

etc.

Korean

Hangeul conversion

etc.

Descriptions of IMEs, to come.

SCIM Smart Common Input Method platform

The SCIM platform supports the following IMEs on Linux and other forms of Unix.

• English/European
• Ethiopic, including Amharic, Tigrigna-Eritrean, Tigrigna-Ethiopian

Amharic and other languages of Ethiopia are written in a syllabary with too many symbols to fit on a simple keyboard. However, it can be typed in an IME. One types a base syllable, then a vowel, to get the syllable with the consonant from the base syllable plus the specified vowel.

This is the basic keyboard layout.

` 1 2 3 4 5 6 7 8 9 0 [ ]
   ቅ ው እ ር ት ይ ኡ ኢ ኦ ፕ [ ] \
    አ ስ ድ ፍ ግ ህ ጅ ክ ል ፤   
     ዝ ሽ ች ቭ ብ ን ም ፣ . /

The consonant and vowel letters combine as in the following example.

For   Type
ታ     ት አ
ቶ     ት ኦ
ተ     ት እ
ቱ     ት ኡ
ቲ     ት ኢ

• Cyrillic

This is called a phonetic layout, where the letter on most keys is the closest Cyrillic letter in sound to the Latin letter on the same key in QWERTY. But it is not simply a keyboard layout. Letters not on the keyboard can be typed as combinations. For example, the sequence 'ы о' results in 'ё', and similarly 'ы у' results in 'ю'.

` 1 2 3 4 5 6 7 8 9 0 [ ]
   я в е р т ы у и о п [ ] \
    а с д ф г х й к л ; 
     з х ц в б н м , . 

• Inuktitut

Again, a base keyboard layout with combining rules.

` 1 2 3 4 5 6 7 8 9 0 - =
   ᖅ  w e ᕐ ᑦ ᔾ ᐅ ᐃ ᐅ ᑉ [ ] \
    ᐊ ᔅ d ᕝ ᒡ ᔅ ᔾ ᒃ ᓪ ; 
     z x ᒡ ᕝ b ᓐ ᒻ , . /

Example: ᔾ ᐅ results in ᔪ.

• IPA

Keyboard layout with combining rules.

` 1 2 ʒ 4 5 6 7 8 9 0 - =
   q w e r t y u i o p [ ] \
    a s d f g h j k l ; '
     z x c v b n m , . /

Note that the character on the 3 key is not a 3, but U+0292 LATIN SMALL LETTER EZH.

Example: 's h' results in 'ʃ'.

• Vietnamese

` 1 2 3 4 5 6 7 8 9 0 - =
   q w e r t y u i o p [ ] 
    a s d f g h j k l ; 
     z x c v b n m , . /

Example: 'd d' results in 'đ'

Japanese

• Hiragana (romaji-hiragane conversion)
• Katakana (romaji-katakana conversion)
• Nippon (romaji-kanji conversion)

Korean

• Hangul
• Hangul Romaja
• Hanja

Simplified Chinese

• Erbi
• Erbi-QS
• Wubi
• Ziranma

Traditional Chinese

• Wu
• Array30
• CNS11643
• Cangjie
• Cangjie 3
• Cangjie 5
• Canton HK
• Cantonese Pinyin
• Dayi3
• EasyBig
• Jyutping
• Quick
• Simplex
• Stroke 5
• ZhuYin
• ZhuYin Big

On Debian Linux, use apt-get, aptitude, or Synaptic to install the scim package along with im-switch (required for Chinese, Japanese, and Korean) and scim-gtk2-immodule. You can also install skim, the KDE graphical frontend, which puts an icon on the toolbar. On systems using Red Hat packages, install the corresponding software. Follow the configuration instructions to set the necessary environment variables. Then run

scim -d

from the command line. Open a supported application such as gedit. Right-click in the window to get the main IME menu. Select "SCIM Input Method" to enable Chinese, Korean, and Japanese,

xcin

Used with bitmap fonts and the rxvt Chinese terminal program.

IIIMF

Internet/Intranet Input Method Framework of OpenI18n.org is a new Unicode-based input method framework.

Chinput

Chinese XIM Input Server

Help

Tests of IMEs, need help from expert users.

External links

• Wikipedia article on IMEs with links to articles on IMEs for specific languages
• Wikipedia article on Chinese dictionaries
• Official Cangjie Home Page
• Cangjie Method book in English
• SCIM
• IIIMF
• How to create an IME in scim