I have been thinking about translion of the open source software didi, even just the readme.
At what stage do I start to ask for translators? Or do I look for free machine translation software?
I feel that a translation of the readme would be more inclusive and might assist with onboarding new collaborators at this early stage.
What do you think?
Top comments (10)
Create an html button ( read article in your language) . Link it with translate.google.com or anything similar.
do I look for free machine translation software?
In early stage of Didi , I believe it should work without any translation software
Sure that sounds like a good first step. Google translate does terify me somewhat though, it's something I personally cannot test the output and so it's not trusted.
True but all ML translation software are not trustworthy. You can ask devs to convert it into there mother language
Already working on that, I'm keeping my eye out for articles that need some English translation in exchange.
I guess the best would be to approach the authors of the software to make sure this is ok to them. Judging from the way you ask I guess you're not overly familiar with localization issues so I'll throw some generic advice regarding the topic as if you'd be the one doing translation and it would be your first time.
Translation is a challenging thing to do. While you can get some sense out of tools like Google Translate there is the fact that machine translation software isn't great with stuff like contexts, idioms, wordplays, etc.
When doing translation yourself if you're not familiar with the original language you can really go wrong unless you have some healthy paranoia, do exhaustive and careful search on every single word that seems to be a bit off, or ask from someone who speaks the language. Word-by-word translation is very bad. Finding the true meaning is important. You really want to "retell a story" in the way it would be told in the target language.
So ideally you want to find a person who know this is the way translating stuff works. In general if you want good quality you better hire a person to do it if you're unsure about doing it yourself. Translating stuff well and with a decent pace requires a lot of skill, knowledge and practise so asking to do it for free might be evil.
Not sure what didi is, but I would not integrate Google Translate to any application. The translation quality fluctuates a lot, from quite good to completely wrong. Integrating Google Translate says that I back up these translations. End users can use Google Translate or whatever translation mechanism they prefer themselves.
Translations are an architectural choice. That means it is a big decision and also hard to change later on. You want to support translations early on, with at least two languages. Preferably the translations would be in some standard format, so the translations can be loaded to a translation interface (e.g. Poedit), making life easier for translators.
Then you also have to consider localization, things that are not translations, but other things that are culturally different. That can be things like date formats, but also layouts.
I suppose to give some context, didi is a cli and configuration frontend for a compiler. That and the docs are all public facing. The project is open source (community driven project) and if didi facilitates the ability to easily add translations localization then that might be the way to go. There is a wonderful i18next library I found today for such a task.
Why don't you want to do localization? I believe that this is the best translation possible. It may be more expensive financially, but still much better. My experience is a little different, I did site localization, not software, and I read a lot of descriptions and information about services here translationreport.com/. I was very pleased with the result and there is a big difference in quality.
Oh but I do 🤷♂️
Check this page out as it’s a good starting point to understand translation vs localisation (and how to approach)
en.wikipedia.org/wiki/Language_loc...