How easy is it to run both Japanese- and English-language applications on a Mac, or on a Windows 95 machine? Which environment offers the better help system in both languages? How intuitive is the operation if I have only a rudimentary grasp of one of the languages, but must nonetheless use it occasionally in my work?
These are just some of the questions that readers in Japan and overseas have been asking Computing Japan. This month, we address those questions, and more, by taking a comparative look at the procedures for processing Japanese and English on both Windows 95 and MacOS machines, with an emphasis on the Japanese input method editors used by each.
by Steven Myers
Apple Computer's recent ad campaign has had considerable fun at Microsoft's expense by pointing out "new" Windows 95 features that have long been standard fare for Mac users. ("It lets you drag objects anywhere on the desktop.... Imagine that.") Looking past the gibes and hype, though -- in terms of basic Japanese-language handling and general bilingual capabilities -- how do the most recent solutions offered by Microsoft and Apple compare? And, more importantly, which is best for your own needs?
When comparing the Japanese and English text-handling features of different computing systems, such as the MacOS and Windows 95, it is important to look both at the specific method each system uses for basic Japanese input (a process not nearly as straightforward as typing in English), and at the "big picture," or overall bilinguality of the system.
"How can I use both English- and Japanese-language applications on my computer?" is a question being asked by an ever-increasing number of native English-speaking computer users who have acquired at least a fair command of the Japanese language. From the perspective of such a user, the ideal bilingual computer should be able to run Japanese applications without difficulty, yet at the same time allow the user to operate in a predominantly English environment (with English menus, dialog boxes, and online help).
In this respect, MS-Windows' support for bilinguality has traditionally been somewhat lacking. Japanese Windows 3.1 allowed users to run Japanese applications reliably, but it was not always so reliable for running English-language applications. English Windows 3.1, meanwhile, provided no support for Japanese applications whatsoever. Third-party add-on products, such as Win/V and TwinBridge (both covered in previous issues of Computing Japan), gave English Windows 3.1 users some Japanese capability, but Microsoft itself has produced practically nothing in the way of software for bilingual Windows 3.1 users.
Nor has this situation improved much with the arrival of Windows 95. While the Japanese edition of Windows 95 is now able to reliably run "all" English-language applications, there is no provision for non-kanji menus or help files. And there is no product like Win/V on the market that will allow users to run Japanese applications from within an English environment. Windows 95 users who want to use English-language applications in an English environment, yet also want to run Japanese-language applications occasionally, can buy the Japanese and English editions of Windows 95 and install both on the same machine, but the procedure is rather convoluted and requires rebooting to switch between English and Japanese. (See "Ensuring Peaceful Coexistence," page 25, in December's Computing Japan.) This type of fix, however, forces the user into an entirely Japanese environment for running Japanese applications.
Microsoft's stance is that there simply is not sufficient demand at the current time to warrant the effort of making Japanese applications more accessible to users of English Windows 95. It appears that the company has divided the world market into three main linguistic regions -- Western/European, Middle East, and Far East -- and intends to focus on each region separately and individually. Providing bilingual solutions between language regions is not a priority.
For many years, Apple Computer took the same approach to bilingual computing as Windows 95 has done: you choose one language for your OS, and do all of your work in that language. Eventually, however, Apple released the Macintosh Japanese Language Kit (JLK), which performs functions similar to what Win/V and TwinBridge did for Windows 3.1, only much more smoothly and without the need to reboot every time you change environments.
The Mac JLK not only lets the user run Japanese programs in an English environment, but it also provides the capability for entering Japanese and English in the same document even when using a non-Japanese program (provided that the program is compatible with Apple's WorldScript). Additionally, Apple provides English-language manuals for both the JLK and for Kotoeri, the Apple Input Method Editor (IME). This contrasts sharply with the Japanese version of Windows 95, which comes with no English-language documentation or online help for the latest Microsoft input editor, MS-IME 95.
In general, Apple appears (at this point, at least) to be far more progressive than Microsoft in promoting the capability to work in multiple languages on one system. In addition to the JLK, Apple produces several other language kits, all of which work in the same modular fashion. When a user installs a new Language Kit, choices for keyboard and text controls corresponding to that language are immediately made available, and the user can switch languages simply by choosing the appropriate keyboard. Apple seems to espouse the forward-looking view that, rather than having several different language versions of an operating system (OS), there should only be one OS -- with language-selection treated as just another object.
Several excellent third-party Japanese IMEs have been produced for both Macs and Windows PCs. The scope of this article, however, is limited to the IMEs that ship as "standard": Apple's Kotoeri, which comes with the Japanese Language Kit and with KanjiTalk (the Japanese version of the MacOS), and Microsoft's MS-IME 95, which comes with the Japanese edition of Win 95.
An Input Method Editor functions as the front-end for Japanese character conversion by intercepting all keyboard input. When used with a US-standard keyboard, the IME is responsible for determining the katakana, hiragana, and kanji characters that correspond phonetically to the alphabetic characters that the user types.
The quality and ultimate utility of an IME is determined primarily by the ease and speed of its character conversion. Nearly all IMEs today employ a variety of artificial intelligence (AI) techniques for finding the desired kanji character from a list of possibilities. The system "learns" to use specific characters under certain circumstances, based on recorded patterns of behavior. For example, if a user repeatedly selects the same character for a certain phonetic reading, that character will appear higher up on the list of choices the next time that reading is to be converted.
Both Kotoeri and MS-IME 95 use very similar basic methods for character input and conversion. With both, the user can choose to have his or her alphabetic input displayed on the screen in hiragana, katakana, or romaji. In addition, these characters can be either full- or half-width (compared to the standard width of a kanji character). The control of character input and other properties for the IME are handled using the Operation Palette on Kotoeri or the IME Toolbar for Windows 95.
Kotoeri is slightly more sophisticated than MS-IME 95 at performing the basic romaji-to-kana conversion. Consider, as an example, the case of converting a word where a な (na), に (ni), ぬ (nu), ね (ne), or の (no) character follows the ん (n) character. The Microsoft IME does not convert this correctly unless the user adds an extra "n" in the input string. That is, the hiragana expression おんな (onna) can be input normally as "onna" on Kotoeri, but it must be entered as "onnna" on MS-IME 95. Inputting "onna" on MS-IME would result in conversion to おんあ. Also, on hiragana combinations such as きょ, きゅ, みょ, or ひょ, hitting the backspace key on MS-IME 95 has the effect of wiping out both characters, rather than just the small ょ or ゅ.
In my own testing, also, the MS-IME 95 toolbar occasionally would lock up for no apparent reason, preventing me from accessing the dictionary or changing input modes. When I minimized the Japanese application, and then restored it, however, the toolbar would then work fine. Go figure.
In addition to phonetic input, a good IME must be able to take a numeric JIS, Shift-JIS, or kuten code and convert it to the proper character. MS-IME 95 has a irritation in this area as well: it will only recognize five-digit kuten values. Unless the user adds a leading zero to a four-digit kuten value, the system will warn that the reading is invalid, without explaining why.
(For those not familiar with the kuten system, it is a method of assigning a value to a character based on its position in a 119 row (ku) by 94 column (ten) grid. The first two or three digits of the kuten value are the row number, followed by the two digits of the column number. Characters that are in rows 100 through 119 obviously have a five-digit kuten, but the rest of the characters will have four digit kuten values. Most IMEs allow the user to simply enter this value as is, without a leading zero.)
The heart of any IME is its dictionary utility. This is the part of the IME responsible for storing and arranging possible character combinations for the input sequences (not to be confused with an electronic word look-up dictionary). Most IMEs have one main system dictionary, which cannot be modified by the user, such as the msime95.dic file on Windows 95. In addition to this system dictionary, most IMEs allow each user to create his or her own personal dictionary of frequently used character combinations that may have unusual or infrequently used readings, such as 朧月夜 for おぼろづきよ (oborozukiyo, a "hazy moon night"). These personal dictionaries may also contain personal or geographic names, or abbreviations for frequently used addresses.
Besides the main system dictionary and personal dictionary, some IMEs also allow for special dictionaries that contain the technical terms for a particular field. While Apple's Kotoeri does not come with a specialized dictionary, it does allow users to develop their own. Oddly, though, once the user has created such a dictionary, Kotoeri does not allow words to be added or deleted. MS-IME 95, on the other hand, simply defines a one system dictionary and one user dictionary. Several different user dictionaries can be created, but only one of them may be open at any given time.
MS-IME 95 has one weakness that potential users should be aware of: it is incompatible with dictionaries created using previous versions of Microsoft's input utility, MSIME and MSIME2. If you have built up large dictionaries using these earlier IMEs, you'll find that you must recreate your dictionaries anew using MS-IME 95.
MS-IME does have some interesting and useful features not found on Kotoeri. Many users of Japanese applications for Windows, for example, have long opted for third-party IMEs, such as JustSystem's ATOK, instead of the Microsoft IME. Each IME has its own set of conventions for using the function keys to speed up the conversion process. (MS-IME 95 and ATOK, for example, both use the F7 key to quickly convert keyboard input to full-width katakana characters, while the WX2 IME uses Ctrl+F7 to perform the same function.)
Rather than force the users of other IMEs to give up the keyboard shortcuts they have grown accustomed to, Microsoft has smoothed the transition to MS-IME 95 by allowing the user to choose from "templates" for other IMEs, including WX2, VJE, and ATOK. When a particular template is chosen, all of the hot keys and keyboard shortcuts for that third-party IME become active. Users can also display a keyboard guide that lists all the relevant key sequences for a given IME.
MS-IME 95 also interacts very well with the gaiji editor utility (one of the accessories included with Win 95J). This program allows users to quickly create their own custom characters on the fly, assign them code points and fonts, and use them in documents. In the middle of a document, for example, the user can start the gaiji editor and create the new character, and then use that character in the document.
With Windows 95, Microsoft has greatly improved both its input utility and its general user-friendliness for Japanese users. Native Japanese speakers (and persons with nearly native Japanese reading ability) will have few problems working within the Windows 95J environment and running English applications from within that environment.
Those whose Japanese reading skills are not so highly advanced, however, might be advised to do their Japanese work on a Mac with aid of the Japanese Language Kit. That will allow them to take advantage of English-language online help, English manuals for the IME, and a more sophisticated front-end for Japanese input.
Essential vocabulary for reading Japanese IME online help
半角文字 (hankaku moji): half-width character
全角文字 (zenkaku moji): full-width character
文節 (bunsetsu): phrase
総画 (soukaku): stroke (for writing characters)
部首(bushu): radical (used for indexing characters)
記号 (kigou): symbol
句読点 (kutouten): punctuation
辞書 (jisho): dictionary
学習 (gakush4): learning
Products reviewed
Japanese Language Kit ver. 1.2
1.4MB disks and CD-ROM
Requirements:
Macintosh or Power Macintosh
5MB RAM, System 7.1 or later,
30MB HD space
Microsoft Windows 95J
1.4MB disks and/or CD-ROM
Requirements:
486SX processor or higher
8MB RAM, DOS/V (for upgrade version)
75MB HD space
Copyright 1996 Computing Japan Magazine