Quarter Life Crisis

Translating¶

1549 words

Translating things is easy. Translating things is hard. Once you know the other language sufficiently, getting the message across somehow isn't too hard. However, getting it across with the text feeling right and belonging naturally to the other language, seems very hard to me. I certainly have great respect for translators of novels that make you not realise the novel wasn't written in the language you're reading it in. While I don't know Polish and thus cannot compare, the German translations of Stanisław Lem come to mind. They really feel like naturally German texts (and are widely considered very good). On the other hand I have great disrespect for people translating technical instructions in a way that they hardly get the grammar right.

A whole new twist is given to the translation game in the field of computer software. To make things harder – or easier – you are limited to a certain vocabulary that is set by the Human Interface Guidelines. You'll have to stick to that terminology and the recommended style to make your program blend in well with the others.

As most computer related terms are coined in English they seem more or less natural there: The have an original meaning, and a second, technical, meaning that is associated to them after being used in a computer environment or desktop metaphor. An example for this is the word file. It has an original meaning related to the physical files we have in offices and it has gained another meaning referring to data stored in a computer. This double-meaning of the word may may cause difficulties when being translated to other languages. In German, for example, many computer programs sound much more technical than they do in English. This may be because the German language lends itself more easily to long words, making it look clumsy, but it is also a question of the attitude of the translators.

The most prominent example for this is the translation of the File menu. Apple chose, in my opinion wisely, to translate it as Ablage. While Ablage is a rather unusual word referring to classical filing, thus reflecting what the File menu is for. It is not mainly for computer files but rather for activities related to to file. The other possible translation, used by Microsoft and Adobe, even on the Mac, is Datei. It refers to the file as stored on disk.

Another example is the translation of the Save menu item. Apple's translation is Sichern, which means, erm, Save, whereas the other common translation is Speichern which is more like Store, in line with the idea that the File menu has to do with files rather than with filing activities. However, Speichern as well as Datei are purely computer-related terms in German, making the programs using that terminology sound much more technical and much less alive. This is not just about grammar, and getting your message across. As in real life, the choice of vocabulary is a design statement on whether you're trying to be technical or not.

Many more examples can be found and perhaps this gives an idea about why people not using English as their system language are having a harder time. Wordings that are perfectly innocuous in English can become a source of ambiguity and requiring extra decisions when translated. And that's on top of the Find vs. Search debate or the possibilty of unwisely chosen words.

In part these problems can be avoided by a rigorous analysis of language. An example of particularly careless translation (and lack of quality control) is found in Apple's System Profiler application: There are at least two possible translations for the word extension to German: In the sense of file name extension Apple translates it as Suffix, and in the case of kernel extensions it's translated as Systemerweiterung. Somewhere in the program they got their strings mixed up and use both translations for the same thing as can be seen in the screenshot.

Thus, the same spelling doesn't mean you're dealing with the same word. And you'll have to consider the influence of grammar as well. Once you're aware of the grammar used you will know that you need separate translations for you read:

singular, present tense

Du liest

plural, present tense

Sie lesen

singular, past tense

Du lasest

plural, present tense

Sie lasen

with the additional twists that you'll want to use the polite plural form when translating the singular from English and that most people aren't as snobbish as me and might prefer to use the Perfekt tense Du hast gelesen, Sie haben gelesen instead of the correct past tense.

Only considering read, without the you you'll get even more ways to use it in English: For all persons (except he/she/it), as infinitive, as past participle, as in a good read and possibly more that didn't come to my mind now. All of these will require distinct translations in many languages, even European languages, as I gather that there may be a whole set of extra problems when going to Asian language that have a completely unrelated grammar and structure.

As most programmers seem to work in English, the convenience of the English grammar may be a disadvantage here as it causes many problems when no proper analysis if the language is done. While people like Erik may be aware of language, I still disagree with his assessment that most programmers are. I've seen too much badly translated software for that. (His claim may still be true for non-natural languages, but that's a completely different and in my opinion far less interesting issue).

Coming back to the 'extra twists' I mentioned above, suggests that simply knowing the words' usage and grammar doesn't do either. You'll need to know extra habits and the culture as well.

Adding all these up means that for a good translation you'll need someone who:

1. knows the language, preferably as a native speaker,
2. is aware of the language's subtleties,
3. knows the original language well,
4. is quite familiar with the operating system and it's language guidelines,
5. knows enough other software to make the program fit in and, of course,
6. despite all these skills, thinks translating programs is a worthwhile way of spending his or her time and investing a lot of effort and patience in.

And even with a translator as described above, things will only work out if the programmers are prepared to co-operate and think of later localisations right from the beginning. Frameworks like Cocoa certainly improve the willingness in that area. Considering all this probably explains why there is so much poorly localised software and makes you meet the localisations of large pieces of software as MacOS 9 with respect.

I have done a few localisations myself. As I grew aware of the different issues it became apparent that doing a reasonably good localisation is very times consuming and requires a lot of arguing the pros and cons of different possibilites as well as research on how similar things are translated elsewhere. Usually translating my own programs was easiest, as I knew exactly what things are supposed to mean and do and thus could easily settle the ambiguities. The translation of the ancient SETI Checker was OK, particularly as the terminology it uses is related to the English SETI@Home website. I am more unhappy with my recent translation of Rechnungs Checker. It still feels a bit German, which may be because most of the surrounding terminology is and I haven't really though about the subject in English that much.

Another localisation effort was that for GeburtstagsChecker. I bugged my Japanese and Mexican flatmates 知帆 and Job to do Japanese and (Mexican) Spanish localisations with me. That turned out to be very tricky. They didn't know much about software, Mac software in particular, and I had to explain everything in great detail to them, so they could find the appropriate wordings. Those explanations took a lot of time and it's best to see where they come into action within the program to see whether they're good or not. While I appreciate 知帆 and Job's efforts, I suspect that those localisations aren't 100% for the lack of background. As I don't know the languages myself, I can't assess that, though.

Another problem with this is that we can't easily add new strings to the program now. While this may lead to more conservative development (which may be good), it also led to a few English strings being present in the Japanese and Spanish versions.

My last attempt at translating things was contributing a German localisation to Nicholas Riley's latest release of IceCoffEE. It's very little text to translate, but even that raised many questions. Just the process of pinning down the problems and discussing them with Nicholas, sometimes led to the clarity needed for a satisfactory solution. Just don't read the translation of the read me. It's not particularly smooth. The problem I had with it may be that Nicholas' style of writing it is different than the style I would have used. I suppose that's another thing proper translators learn in their education: Translating any text well without distorting its style.

May 31, 2003, 14:47