Hacker News new | ask | show | jobs
by codingdave 2513 days ago
I like the idea, but you are putting more weight behind the tooling than I would. I don't find translation tooling to be cumbersome, so especially if machine translations are free, I don't find the price point for human translations to be compelling.

What would be compelling is if you could pro-actively call out the bigger gotchas in translation - grammatical differences that make you change word orders, different mechanism for handling plurals, etc. If you could preemptively warn us, even before a "push" that we may hit a problem, I'd take a closer look. For example, flagging a line saying, "Hey, it looks like you are using phrasing that will be problematic in <Italian/Hindi/Russian/etc.> Here's why..."

2 comments

Great point, we would like to add support for common gotchas. Right now we abstract all plural logic away from the user by automatically requesting all plural forms of a phrase for each language. We also have noticed that translations often break layouts so we are pushing a tool that notifies you when translations are x% longer or shorter than the original string.

We would love to explore more ways we could help pro-actively solve these problems. Ping me at abhi@langapi.co if you want to talk more!

Mozilla’s Fluent project [1] seems like a really thoughtful and comprehensive approach to many of these problems.

[1] https://projectfluent.org/

We analyzed fluent when we built the product and found that it didn't offer much more than the international standard ICU. Our platforms have full support for ICU syntax including plurals.