Hacker News new | past | comments | ask | show | jobs | submit login

Dealing with this many languages is difficult. I recently worked on a project where many of the languages were not supported by Windows, so the locale information wasn't easily selectable as resource files in ASP.NET.

If translating content to 1000+ languages interests you, the JWs have released, at times, some articles explaining how they achieve simultaneous publishing in 1034 languages using a piece of software from Oxford called MEPS [ https://www.jw.org/en/choose-language ].




Since you didn't link the article: https://wol.jw.org/en/wol/d/r1/lp-e/101984287

This is honestly amazing, never did I think that the most translated website would be for something like JWs.


This is fascinating, but I was intrigued by why in the world they were using IBM mainframes to do their translation into multiple languages.

And then I realized that the article was published in 1984 (!) so it appears that MEPS / IPS dates back to that time and is almost 40 years old now.

Is MEPS still in use? I'm assuming it has been superseded by newer technologies, Unicode, etc?


JW and Swe here with xp volunteering on projects internally!

While not being directly involved with meps development ( there is so many different depts and service to volunteer and help out), I know that the MEPS has always been an umbrella for the multi lingual publishing tools. At its origin it was a revolutionary system that helped layout, compositing and preprint workflows and scaled to a Hundred something langs an provided features we take for granted today (typesetting and automatic layouting in 120+ Lang’s well before it was cool)

Nowadays it’s highly probable that the meps also encompasses web publishing workflows (now in 1000+ Langs), pdf, Braille and print, using whatever industry standard appropriate and filling the gaps where commercial products fall short ( its a quite unique set of challenge that would stress test any CMS and publishing software)

Finally I have lots of friends working as volunteers in the content translation itself and I’m even more amazed by the quality of training, quality assurance process and attention to detail.

For instance small teams of translators are geographically located where the target language are spoken idiomatically (even for small communities) to ensure smooth and practical translations. Just imagine having to coordinate hundreds of distributed teams in every single country and region of the world, all volunteers, and maintaining high quality translation while most of translators are locals with only goodwill and no formal training in translation ( sometimes even dictionaries does not exist )

And of course, no corporate backing, and strictly every single member is a volunteer !


Shame no pictures.


Pretty impressive, thanks! It is quite a challenge that I've been tinkering with for over ten years now, really giving Unicode a workout. Plus, often the pre-modified characters (say, for the sake of argument, "a" with an acute accent) often don't turn out well in some fonts and I would have to fix that by using the regular character plus the combining character — two Unicode entities — then manipulating the latter with inline CSS to position it pixel by pixel.

I also faced the challenge of combining Arabic and/or Hebrew (which read right-to-left) and Roman in the same line. It drove both my browser and text editor nuts. In at least one such case in the book, Mozarabic, where I needed all three, separated by slashes, I just went for the nuclear option of simply bitmapping them.

As the title implies, I was able to exploit Google's Noto series to the hilt for my book. The Cardo I'm using for the body copy (sorry that some don't care for the ligatures, which I deliberately sought) was originally a web display-only font, but that's been fixed. Also, its italics wouldn't respond to CSS; fixed, also.




Consider applying for YC's Spring batch! Applications are open till Feb 11.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: