Sunday, January 20, 2008


A reader writes that he thinks a self-explanatory interface is a goal (perhaps unattainable). I'm not sure that it actually is a goal, at least for most of the things I want to use computers for (the discussion started a long time ago in connection with Chinese, Japanese and Korean input on the Mac).

Natural languages are almost entirely arbitrary systems of signs. The number of spoken words that sound like the things they represent is generally a very small fraction of the number of words in the language. The number of written words that look like the things they represent is very small even in languages that use ideographs; in languages with phonetic writing systems ALL written words are arbitrary. And as we all know, mastery of even the simplest writing systems takes years of practice. English and French have atrocious spelling 'systems', Italian and German fairly sensible ones - but even an Italian schoolchild does not read and write fluently after an hour's exposure to the alphabet. Ancient Greek is tricky; written Chinese and Japanese are fiendish.

If we take the case of ancient Greek, there are no native speakers of the language. The thing that would be really useful would be a print-out that not only explained the keystrokes needed to input letters and diacritical marks, but also actually explained the use of accents and breathings to the novice. To type the language correctly, the novice needs to know, for instance, that a final acute accent normally changes to a grave before another word; that a circumflex may be take the form of a rounded French circumflex in some fonts, a tilde in others. Virtually all published Greek is input by novices - that is, by typesetters who are at home in Quark or Indesign but have never studied the language. They would be better served by a crib sheet which gave the information needed. It generally takes students a week or so to use smooth and rough breathings correctly without thinking about it; accents take much longer to use correctly. It would be surprising if the casual typist or typesetter could get this sort of thing right without any explanation at all.

It would be easy to multiply examples, but I spent yesterday translating English into English and I have a headache. The singer/songwriter Alyosha Blinov asked me to go over his band's entry on Myspace, which had been written by a fellow Russian and bandmember. Poor head. The revised entry for Classic a la Punk can be seen here.

1 comment:

Anonymous said...

Is it that
(a) a self-explanatory computer interface should not be a goal?
(b) it is an unattainable goal given that much of computing is based on written language, which does not have such an interface?

I think it's still a nice idea, and demonstrably possible in areas not predicated on unintuitive interfaces like written language. Even if we know we can't get there because Greek gets in the way, why shouldn't we approach the interface problem as an opportunity to maximize intuitive use?

And of course we can side-step the issue entirely by saying "well, it ought to be intuitive insofar as it relies on skills the user already has". The purpose of an input system for Greek is not to teach you Greek. If we assume you know the written language, is there anything preventing the computer's input system from being self-explanatory? Certainly not: a native-language keyboard gets you most of the way there already.

Sorry to harp on this; apparently it is a subject that angries up the blood.