Machine translations don't consistently obey the glossary

Machine translations don’t appear to obey the glossary, in most cases they translate placeholders, markup, and other items in the glossary.

Hello,
Thank you for telling, that it’s important to you. This is a known issue and we already have a plan to improve this behaviour soon.

1 Like

Also worth mentioning when you click glossary words in the translation flow, even in situations where the the word has been translated to the target language in the glossary, the source language is always copied over into the text field.

Any updates on this? As a new user I’m finding the usefulness of MT to be limited because they translate interpolation expressions, resulting in a broken string. e.g., MT from English: over {{count, number}} results to German produces über {{Zahl, Anzahl}} Ergebnisse, which is invalid.

1 Like

Hi @Bryan_Smith,
This hasn’t been released yet, but delivering more reliable MT translations is high in our priority list.
I will update this topic on the progress.

1 Like

Hi @janbilek !
I’m also curious if you have any update on this. We are currently running a trial period with Localazy and this would be a much appreciated function - mostly for when we want to bulk-translate our 10k+ keys to new languages…

I think DeepL currently supports Glossary’s as part of their own app as well. Or do we need to enable the OpenAI (GPT) translations to benefit from this?