Use Translation Memory from dialect for Automatic Pre-translation

Hi!
We are using Crowdin with many dialects of languages, because we deal with different stakeholders from different countries with different requirements on wording. In 95% of the cases, translations for de-DE should be the same as for de-AT.

This is why I would like to use pre-translation via translation memory to automatically translate keys which have been translated in any dialect to all other dialects (and base language). For example, if some translator from Austria decides to translate a newly added translation key to de-AT first, then I would like to add the same translation by default to de-CH and de-DE as well without having to go through the translation keys one by one in all dialects again. If I click PRE-TRANSLATE → via Translation Memory in the dashboard and try to pre translate de-AT based on existing translations in de-CH or de-DE, then I always see “0 translations added. 348 skipped” even if some translations are present in de-CH, which I want to apply for de-AT automatically. I even decreased the precision from 100% match to 98, but still no translations are applied.

Is my desired behavior not supported or can I configure something like this? I am aware of the export setting “Automatically fill in regional dialects”, but this will not get translations from de-CH to de-DE or de-AT. Do you have any suggestions on how to automate my desired behaviour using Crowdin?

Thank you!

Hi @FilipHesse this is expected behaviour. Pre-translate works only if there are suggestions for this language in the tm since it can’t grab translations with different language codes and apply it to another language (this behaviour might create a mess with translations since system won’t know for sure what dialect language you want to use)

The option you mentioned is to make sure you get the translations from the main language populated to all the dialects on export.

As a workaround, you may down’oad translations from the needed language and upload them to a different one (pay attention to the language code in the file)