The major headaches for Western software vendors in adapting their software to Japanese needs is Japanese text entry integration, Japanese menus, help, manuals, and improved quality.


“Well-behaved” graphical or Java applications that rely on the operating system’s text-handling routines can generally accept Japanese double-byte characters. EUC and Unicode character sets properly display and handle mixtures of single-byte and double-byte characters. However, sometimes software companies develop custom text handling routines to optimize performance, and these can garble text, for example, by allowing users to position the cursor in the middle of a double-byte character and separate the two bytes thus rendering them unintelligible.

Japanese support

Japanese has two parallel alphabets (kana) of 50 letters each, in addition to some 7,000 kanji or Chinese characters. Rounded-looking “hiragana” evolved as the calligraphic abbreviation of a Chinese character used phonetically. Hiragana are used for Japanese words. Squarish-looking “katakana” or “partial” kana, evolved from the portion of a Chinese phonetic character used to write the character in shorthand. Katakana are used for foreign words and phrases (“table” – te-buru, “computer” – conpyu-ta, etc.) and for word emphasis in a manner similar to italics.

Kanji are like the Latin roots of words (hereby represented in bold letters). Kanji are used in pairs to form a compound word (inter national), or as the semantic portion of a verb containing the hiragana conjunction (we arewalk ing around). Individual kanji may have many pronunciations. Because of the large number of Kanji, two bytes are needed for a computer to address them. In addition, there are some 40,000 infrequently occurring characters, of which the user may define up to 80 or 90 user-defined “gaiji.”

Collating sequence drivers

Japanese kanji sort by the number of brushstrokes, so a database needs to preserve the pronunciation of, for example, proper names and place names, in order to allow sorting. Sort and search functions must take into account that there are up to five possible ways to write the exact same thing with the two single or double-byte alphabets and, of course, kanji.

Front-end processors (FEP’s)

The conversion from the keyboard’s nearly 100 keys into over 7,000 Japanese letters and kanji on-screen is made through a kana-kanji conversion program or front-end processor (FEP). Entry is made phonetically in kana or roman letters, and then the user hits a “convert” key which through grammatical parsing and other software techniques, selects the appropriate kanji to express the intended meaning of the phonetic entry.

computer keyboard

The FEP acts like a terminate-stay resident program, popping into the foreground whenever the user hits the keyboard, for data entry or for menu selection. However, in word processing programs and products such as Lotus 1-2-3 with well-integrated kana-kanji support, the FEP only intrudes into the user session when kana-kanji conversion is necessary, not every time a key is punched.

The software must be compatible with, through Windows drivers, the most common FEP’s available, including ATOK (from the Ichitaro word processor), Matsutake (from the Matsu word processor), VJE (a stand-alone FEP), and optionally, IBM’s kana-kanji conversion program, and Fujitsu’s OAK FEP.


The product documentation, error and other screen messages, and help must be translated into Japanese to gain wide adoption by unsophisticated end users. Product documentation examples must be rewritten to reflect Japanese customs and business practices. For example, Japanese do not often take out home or car loans.

A rule of thumb is that it will cost $50-100 per page to translate the manuals.


Japanese users are sticklers for quality. Bugs and even purely cosmetic problems are not tolerated.