to Home Page

From the Multilingual Battlefield (2)

by Urs APP


Abstract

This article (which is also published in the Electronic Bodhidharma No. 4) takes up some issues related to multilingual computing on various personal computer hardware and software platforms. The focus is on the ability to handle one or more East Asian languages and other script systems in the same document. An earlier article of this series was published in No. 3 of the same journal (1993).

(For additional evaluations of Windows and Windows software see Christian Wittern's report Windows and the World in the same issue.)


In the two years since my last article in this series, various changes took place above and underneath my working table; but I am sorry to note that the term "battlefield" still is appropriate, and that no "killer application" or "killer OS" has made my half-baked solutions and pitiful workarounds unnecessary. But my set of warriers has undergone a slight change:

1. The NEC-9801 Compatible

My trusted Epson 286 is still functional, but I use it so little that I disconnected screen and keyboard and may soon reduce the machine to backup status. There simply is nothing left that my two other workhorses cannot do. 9801-compatibles survive largely because of Windows, and they are doomed for the very same reason. IBM PC compatibles are simply better, cheaper, more standard, and less closed-in.

2. The IBM PC Compatible

The Pentium-90 machine took over all hard-core DOS duties with panache. The Windows-interface was puzzling at first -- until I understood that it really consists of two discrete parts: the hard-core DOS cake (file manager) and a fluffy topping (the program manager) made mostly of air and some Apple-flavored cream (ingredients: nothing but aliases). Using mainly the Japanese and Chinese Windows flavors, I have in the meantime installed and de-installed enough Windows operating systems and pr ograms to be a weird kind of expert. The whole still puzzles me, though. But I spare you the lurid details of the "sys" purgatory and the "ini" hell.

What do I use the Pentium for? Mostly for word processing, data tagging, and text-related utilities. The Chinese and Japanese versions of Word 6 have great table capabilities, and they handle large files -- though the patented Windows inability to set the amount of memory available to an application lets Word run out of memory all the time. But at least it tells you so and does not simply leave you literally in the dark, like Nisus Writer on the Mac. Since Christian Wittern has described some of the problems and advantages of Word 6 in this number (Windows and the World), I will refrain from going into detail here.

The Microsoft approach ("divide et impera," in the words of Christian) is totally hygienic in that it makes sure that no Japanese pollutes his screen and files with Taiwanese characters, no Taiwanese with mainland characters, no mainland Chinese with Tibetan, and so on. The basic rule is: stick to your language, plus English if you must. Thus a non-English European who dares to use his mother tongue on a Japanese, Chinese, or Korean Windows system literally has to cross his fingers to even get an accented character on screen. Alternatively, he can write rather complicated macros that do things like clean up mixed-language documents (see the Word macros on the ZenBase CD). This is light-years away from Apple's "language kit" approach which lets you implement a German system with German menus/keyboard layout and Chinese, Japanese, and Korean capabilities, if you want so. In Windows, everything works in lock-step: for a Japanese system you need Japanese DOS/V an d you get Japanese menus and Japanese help and Japanese manuals and so on. Even on the application level this holds true: Word 6J does not even recognize your ordinary English Word template (normal.dot), and having (as one does in Nisus Writer on the Mac) the option of getting English or Japanese menus is out of question: Japanese is for the Japanese, mind you, and French for the French.

There may be a little glimmer of hope for multilinguals at the end of the tunnel: Unicode. When the first Chinese character of the much-touted Unicode will finally flicker on my screen next to a Tibetan glyph and a diacritic for Sanskrit, I'll perform a number of prostrations in direction of Portland, Cupertino, New York, or wherever. I don't expect this to happen too soon, though. Remember: almost a decade ago, the Japanese government published a supplemental code standard of over 6000 ch aracters. To my knowledge there still is no computer or printer in Japan that has actually implemented this character set.

In the meantime, I keep switching between Chinese Windows and Japanese Windows and DOS-US mode and DOS-Japanese mode, use fgrep and (Hidemaru)etc., and hope for better days while using the Mac for producing letters, papers, books, HTML-coded documents, and camera-ready copy. Possibly against Microsoft's hopes, one of the boons of the fixed-width Windows TrueType kanji fonts is that working on linguistically cleansed Chinese or Japanese files w ith essentially alphabet-only programs such as Author/Editor (Softquad's SGML editor) is possible. It is not very clean, though; line breaks etc. pose problems, and it is possible to set the cursor into the middle of a character. We're still in the stone age. Therefore, tagging texts is safer when one uses Chinese or Japanese Word 6 or Nisus Writer with appropriate macros. On the Mac, I have not been able to use Author/Editor (which is not Worldscript-compatible) with Japanese or Chinese, n ot even with the fixed-width kanji fonts supplied with Apple's System 7.5J.

The other day, a friend from the Windows planet watched me move the Word folder on a Mac to a different hard disk volume. I explained: "Just pull the Word folder onto the hard disk or a partition, and drop it. That's it." "You mean to say," he asked in amazement, "that this the whole installation?" Anyone who has installed Word on a Windows system and later tried to change the directory or hard disk knows: better keep a whole evening free for the task. Because Windows is an essentially cons ervative system, relying as it were on a Kafkaesque disk bureaucracy. Lots of files all over the place who do not reveal their identity, or even deliberately conceal it. One soon feels like a Japanese politician trying to change the ruling bureaucrats' minds: there's always one more bureaucrat who will not change his ways and blocks the whole process. Only: who is he, and where? Windows is very fond of the past. That's the essence of conservatism. Once it runs, though, it keeps running on a nd on; but try installing, de-installing, or moving anything essential, and you're in for one of those late nights that all of us and our spouses love so dearly. More flexibility would probably kill it, though.

I admit: I occasionally am tempted by the more appealing and intuitive interface of OS/2, and "Warp" sounds pretty flexible. I also like the option of using a true command-line and OS/2's advanced DOS capabilities (like running the Taiwanese ETEN in a DOS session) are a welcome tribute to the past. So are batch procedures: sometimes the graphical interface just does not cut it and makes ordinary tasks needlessly complicated and cute. But as the present OS/2 universe is just as fragmented as that of Windows and there are few applications fit for Far-Eastern languages (not to speak of front end processors etc), I would simply use it as another layer of whipped cream underneath the Windows cake. I confess: I de-installed OS/2 exactly as many times as I installed it. Not the best way to save time. But I read that OS/2's internal coding is being readied for Unicode: right on, implement it, IBM! I'll buy my fourth upgrade of OS/2 and may even leave it on my machine for more than a week.

3. The Macintosh

Macs are still the main workhorses under my desk and at university, and no other environment approaches Apple's when it comes to mixed scripts, strange diacritics, and other wonderful fauna. Though not unifying it, the Mac-OS system software and its Worldscript technology are one big window to the world of different languages and writing systems. This stands in stark contrast to Windows which is, as you would expect, a whole bunch of windows -- equipped with bars, if you li ke, to prevent people from seeing things in a broader perspective. When it comes to application programs, though, Apple users get a different kind of constricted feeling. To my knowledge, Hypercard is still the only program that in its database incarnation lets you mix all kinds of languages (even in the same field) and happily chugs along. Chapeau to Bill Atkinson for this wonderful and versatile tool. The Japanese version of Filemaker 2.1, for example, gets hiccups when you use an umlaut or an accent. Unusable with combinations of languages. Does that sound like an echo from the Windows world?

Equally, with word processing programs that handle mixed languages, the choice is practically non-existent: at present, it is Nisus Writer, period. MS Word 6J for the Mac is supposed to be out soon, but I expect it to be another Japanese-and-English-or-else-you're-dead kind of program. I'd be happy to be proven wrong because Nisus' annoying weaknesses, its perverse lust for memory and lack of speed with big files, its bad footnote handling, and that terrible terrible table function definitely need competition. I'd happily dispense with all of Nisus Writer's icons and toolbars in exchange for a freely sizable footnote window that lets me see the main text simultaneously with the footnote, scrolls with it if I would like it to, and allows me to index terms within notes. I've been pushing this ridiculously modest agenda (we're living at the end of the century!) for a number of years now... but the latest version (4.0.7) of Nisus Writer still shows no trace of any of this.

On the other hand, all other programs, including WordPerfect 3.0J, lack some of the basic text handling functions (for example forcing the font to any font you like and back) that make Nisus such a unique tool. Furthermore, Nisus Writer's extensive search-and-replace capabilities are simply wonderful (as long as they don't run out of memory after keeping me waiting for twenty minutes on large operations). The Nisus macro language, though, is to Word Basic what the assemble r language is to Visual Basic. It can do a lot, but who knows how. I also ought to mention that Nisus is the ultimate reason for me writing this piece on a Mac LC630. I once pulled my keyboard into a new position. This dislodged the Nisus copy protection device attached to the keyboard. If you need to handle more than one Asian language in one document in Nisus Writer, you need this scary device. The injudicious move blew the ADB fuse of my Quadra 700, which is shorthand for "motherboard re placement." A hideous way to punish multilingual pedestrians, the hard-core customers of Nisus. If the program's real strength is its implementation of Apple's Worldscript technology, why lock this feature?

As to desktop publishing, one of the undisputed strengths of the Mac approach: most word processing programs on Windows are still incapable of printing cropping marks. I think that Pagemaker 5.0J for the Mac may be the only page layout program that accepts Japanese, mainland Chinese, Taiwanese Big-5, and Korean without fainting. If only its footnoting and indexing capabilities would approach those of decent word processing applications! The Japanese version of Quark Express, on the other ha nd, reflects an uncannily compartmentalized Windows-mindset.

However, the future belongs to electronic text and its multiple incarnations. Electronic text has inherently a much longer future than printed matter because each copy is identical with the original and because it is much more malleable than printed text. And if it bridges ephemeral hardware and software platforms (that's what the Standardized Generalized Markup Language [SGML] is all about) as well as national code boundaries (Unicode is just a first attempt), future generations may surf t he past with some more profit. The future begins now: who takes the initiative? How long will take for ordinary word processing programs to implement not just print-age commands (justification, character format, font, size, etc.) but also structure and content-related menus that insert visible or invisible SGML tags into the text? Though it is a rather primitive implementation of HTML, the booming HyperText Markup Language (HTML) --which structures and links information on the World-Wide W eb-- demonstrates the potential of such an approach. Word processing will unfold much of its inherent potential as it discovers this sphere beyond print.

When visiting software stores or Internet sites, Mac users rightly feel a bit marginal. By their very nature, computers are supremely flexible machines that tend to do whatever software tells them to. And now Mac users finally have AppleScript, an ultra-batch utility that can accomplish complex system tasks and direct many application programs. If only there were more of these programs for AppleScript to call! Where are all the blindingly fast utilities, (say, for handling text files) that flood the DOS and UNIX worlds? AppleScript-aware implementations of tools (such as Mathias Neracher's MacPerl) are still the exception rather than the rule. But I do hope that some company soon takes a giant step ahead by producing the first operating system that fully implements Unicode, complete with fonts and whistles and bells: an operating system that supports multilanguage and multiscript work from the bottom of its heart.


Author:Urs APP
Last updated: 95/05/03