Support Chinese translations

[marco]

Add Chinese as a supported language for translations in Firefox.

[Jon]

(Note: similar ideas have been merged into this thread)

[mukdenia]

@Winston_Sung 

>Not at all. I've seen cases with misconversions like:

> For the word "information", which was misconverted into "message"
> zh-Hant "資訊" (information)
> => (correct conversion) => zh-Hans "信息" (information)
> => (incorrect conversion) => zh-Hant "訊息" (message)
> => (incorrect conversion) => zh-Hans "讯息" (message, Traditional Han term)

Maybe "not at all" is an overstatement?

The difference between "信息", "資訊" and "訊息" is subtle. Also please note that in most cases people read articles or paragraphs, or at least sentences, rarely a lonely single word like "信息" or "資訊". That is, even if some words are misconverted, one can still infer the main idea from the whole context.

Thus, a simple mutual conversion might still be an acceptable option for translation between zh-Hant and zh-Hans.

[Winston_Sung]

> Even if some words are misconverted, one can still infer the main idea from the whole context.

This is not a valid reason to give out incorrect translations / conversion results.

> A simple mutual conversion might still be an acceptable option for translation between zh-Hant and zh-Hans.

It is discussed that it should at least contain word-to-word (regional term) conversion (instead of only character-to-character conversion).

[c933103]

It make text more difficult to understand. In fact, this is a main reason that when I use Google Translate, I would choose to view the result in English instead of Chinese, because its Traditional Chinese translation output being converted directly from its Simplified Chinese translation output make the terms less understandable than they are in English despite Chinese language is my primary written language.

Also note that, a bigger problem with using conversion to deal with difference between Traditional and Simplified Chinese is that, the characters do not correspond with each other. Tools like Google Translate most likely made use of technologies like neural network to find out when should which characters be converted to which, and thus offer better result than a simple mutual conversion tools that are currently available as browser addon, also show why it is not a good idea to achieve the language support through a simple conversion table.

In addition, in recent decades, Traditional Chinese users use a lot of Cantonese, Taiwanese, and Zhuyin in their online writing, while Simplified Chinese users use a lot of Pinyin Acronyms and Regional China Dialect. These words and expressions are unlikely to be recognized by users of Chinese languages of another script, hence it will be desirable for mechanical translation software to work on it, although a big question is whether there are enough bilingual material on these text for the purpose.

[laoshubaby]

> > Even if some words are misconverted, one can still infer the main idea from the whole context.

> This is not a valid reason to give out incorrect translations / conversion results.

Still, if we adopt a rules-based transformation similar to mediawiki/opencc, errors are inevitable at the beginning. We should not reject it because it is not perfect, but instead expect Mozilla to allow us to fix it in an agile way.

Unfortunately, Mozilla has not yet pointed out in what form the community can work hard.

[Winston_Sung]

> It is discussed that it should at least contain word-to-word (regional term) conversion (instead of only character-to-character conversion).

(what I mean was, [more acceptable] translating as different (written) language > word conversion >> character converson [not acceptable])

[wutongtaiwan]

@c933103 Chinese are also eligible to use this feature, you don't have to worry so much, it's an unnecessary worry

[wi11dey]

@dawger007I think @tempwall was being sarcastic with their statement "nothing happens in China" since it seems that they were dismayed with the lack of a Chinese translation option in Firefox 118, and also pointing out that Mozilla Connect might not be the best way to gauge interest on this particular idea since this is a primarily-English forum and Chinese users won't be as likely to see this idea.

I also support adding a Chinese translation option, maybe the work done in https://github.com/mozilla/firefox-translations-training/pull/192 could be used to train student models on https://huggingface.co/Helsinki-NLP/opus-mt-zh-en and https://huggingface.co/Helsinki-NLP/opus-mt-en-zh, then we could all easily get Chinese translation. Also, Simplified Chinese could be used as a Pivot language internally, which is already supported https://firefox-source-docs.mozilla.org/toolkit/components/translations/resources/01_overview.html#p..., then a lightweight model could do the translation between Traditional and Simplified since even though it isn't a direct character-to-character mapping like @Winston_Sung said, it's much simpler than most language translations.

[tempwall]

@wi11dey  It is indeed as you say. Thank you for your understanding. 😃

Even in our official Chinese forum, when people asks about the translation features，someone with the mark of "mozilla staff" could only suggests 3rd-party add-ons. It seems that nobody cares about their requirements.

I'm not sure if mozilla have convinced that Chinese market is no longer important or not. Hope I'm wrong.

[xxnuo]

Has there been any progress thus far? May I offer my assistance in any way, such as contributing code or whatever help I can provide? The native translation speed of Firefox is truly astounding, and it would be splendid if it could support Chinese.

[marco]

@xxnuohelp is always appreciated, you could chat with us on Matrix in the #firefoxtranslations room (https://wiki.mozilla.org/Matrix).

[marco]

We are planning to start working on Chinese in the second half of 2024.