I have a datafield set up for a concept: #commercial-relationship^purpose-of-cooperation. When I translate the clause to Lithuanian, the datafield shows as untranslated and unrecognised in red - #komerciniai-santykiai^purpose-of-cooperation. Datafields for other concepts are untranslated but recognised (#recipient^gender vs #gavėjas^gender). Any thoughts on where I’m going wrong with the purpose-of-cooperation datafield?
You are not doing anything wrong — this is an issue somewhere in the black box of our third party translation service (DeepL), over which we unfortunately do not have any control. ClauseBase always sends source text in the same way, but — depending on the target language — DeepL, somewhat unpredictably, does a great or not-so-good great job.
I personally also noticed that DeepL’s French module behaves more erratically than, say, Dutch or German. Lithuanian was one of the last languages added by them, so I am actually not surprised that you encounter these issues.
In general, we always emphasize that you should consider the machine translation as a “good start”, and nothing more. Problem is, many machine translation results of DeepL tend to be brilliant, so that it is only natural for users to then get disappointed when the results are suboptimal.
Thanks, I figured it was a DeepL issue. What’s best practice on this - how can I make sure purpose-of-cooperation is recognised and not highlighted in red when switching to Lithuanian or any other language?
Could it be the case that you translated the label of the datafield to LT, but not its alias?
The label is only used within the user interface, towards end-users — it can for example contain spaces, parentheses, or whatever else you want that best communicates the purpose of the datafield towards your end-users.
The alias, on the other hand, is used for internal purposes within the clause (i.e., for clause authors). Its typical purpose is to either create an abbreviation (e.g., “poc” instead of “purpose-of-cooperation”, which is much longer to type each time), or to use in other languages. If you have an alias, then it should get translated by the software before everything gets sent to DeepL.
1 Like
Thanks for pointing me in the right direction! I didn’t have an English label for that datafield, I don’t know why. Once I had both the English and Lithuanian labels, all is recognised.
For future feature readers: see also the further explanation and examples on Datafield aliases – Help
1 Like