A translation blunder by Facebook shows just how difficult localisation can be, and how easily it can end in a good joke.

In its page metrics in Lithuanian, Facebook translated engagement as susižadėjimas, which means all those folks out there have gotten engaged and are about to get married.


The betrothed.

Context is the #1 Localisation Challenge

Facebook is unlikely to feel too embarrassed, as it experiments with its translation algos. But this example highlights two common challenges of multilingual localisation. Both are associated with an inability to process context adequately:

1. Over-reliance on translation automation. It is standard practice to use automated translation in localisation. But machines struggle with culture and multiple meanings, so they still need humans to pick up after them.
2. Humans work with just text, without seeing the context and layout. This is common because the layouts and context may not even be available (e.g. they may still be in the design phase).

In either case, a final review or user tests would prevent such mistranslations from going public.