Subject: Re: how should we localize locale names?
From: Tomas Frydrych (tomas@frydrych.uklinux.net)
Date: Mon Mar 12 2001 - 05:51:27 CST
> Tim:
> summarise: I believed (and still do) that if you plan to use a particular
> language, then you should definitely be able to recognise the name of that
> language as used by the speakers of that language. I don't think that is
> an unreasonable assumption. I gather Tomas is not so keen on this notion,
> though this doesn't seem to be the part that really upsets him.
The assumption itself is reasonable; what I think is a bad idea is to
make or not make features understandable by the user on the
assumption she is or is not going to be able to use them; the
interface itself should always be understandable. Also, I can
imagine users in a certain country rather confused by having to
look for German under D instead of G; but you are right, this is not
my biggest concern.
> Now, if you are going to try to display the names of all the other
> languages in their own localised names, you have a character set problem,
Yes, indeed, and one that is bigger than it might appear; we have
to think in 8-bit char sets, for our Unix interface was designed as 8-
bit (which is pretty standard in the Unix world). Also, we should not
expect the AW users to have to provide any other fonts than are
required by the locale of their choice to be able to have a proper
interface, so even if someone eventually implements font sets for
AW, using them in the interface is in my view not a viable option.
>for all the other languages. My proposal was to use romanised
names in
> that case, so that, for example, if I use AbiWord in en_US (or even
> en_AU), I'd see options like Bahasa Indonesia, Deutche, Francais, Nihon
> go, Sinhalawa... The reason for this proposal is that the examples I could
> think of (Japanese, Thai, various Indian and related languages) all have
> well-defined romanisation schemes that are well-known to most native
> speakers.
> If I understand Tomas correctly, he is telling us that this is not the
> case for Eastern European languages and, additionally, there is the
> problem that the character set in use may not actually have Roman
> characters available, so we may not have the option of romanisation, even
> if the users were able to cope with it.
I have been browsing through the different charsets over the
weekend, and pretty much all contain the chars of the English
alphabet, so technically the romanisation is possible. My problem
is not simply that the user is not going to be able cope, but that it
feels like when you want to buy a gadget in the UK that was made
abroad and comes with a pidgeon English manual -- it really puts
you off, suggesting that the product itself is going to be of a
comparable quality. The fact that the Japanese postman can cope
with an address written in English characters does not mean that
the Japanese would consider this to be correct way to write
Japanese. Our interface should be considered by the users correct,
not just usable. (In non-roman locales, bulk of the langauge list will
look like the pigeon-English manual). And at the end of the day,
yes, we should not assume that everyone can read the English
alphabet.
> easy answer to it. I'm definitely not keen on the idea that AbiWord should
> provide the translated names of every language in the world, translated
> into every supported language... I wonder whether any of our OS's already
> have this knowledge built in? It is a large body of knowledge for a mere
> word processor to carry around...
This is a misconception that Paul introduced when he started
talking in the terms of n x n matrix. The wordprocessor does not
need to carry names of all languages translated into all languages;
it carries names of all languages (it supports for spellchecking) in
English only, and it loads the translations for one particular
language to which it is localised when it starts up. It is not a large
body of knowledge. Word 2000 language list contains cca 170
lanugages, this amounts to around 340 words in each language;
compared to the localised helpfile this is really nothing (< 10
minutes of typing?).
My view is that the basic rule for localisation should be 'do for every
language what you would do if you were designing single language
interface.' If AW had an English-only interface, I doubt that we
would ever consider anything else but offering the list of languages
in English. I believe very strongly that as much as possible, the
application should feel as if it was designed by natives in every
language it is localised into.
All the best
Tomas
This archive was generated by hypermail 2b25 : Mon Mar 12 2001 - 05:51:28 CST