Trados Studio API - mark string or terms as untranslatable for translation provider

Hi,

While programming our own TermInjector plugin we came across the issue that some strings do not only need to be worked on in target after translation but to get some strings within a segment back without translating that part or even to define our own translation for it so that MT does not touch these untranslatable strings or injected translations (quite often MT returns groups of words translated differently across segments) but just implements them within their own translation.

Is it possible to mark terms, or groups of words as untranslatable or even define the translation to be applied for specific terms before sending them to a translation provider API (e.g. MT)?

So that we get back results translated around that term or group of words.

e.g.

- names of British legal acts that must remain in English as there is no translation for them when referring to them.

Italian to German: 'Scalinata di Trinità dei Monti' must be translated as 'Spanische Treppe'

Auntie Edit says:  marking them as untranslatables could be seen as define a translation for that string that must be applied where source is the same as target in that case.

Best regards,

Pascal



some clarification added
[edited by: Pascal Zotto at 6:43 PM (GMT 0) on 15 Feb 2024]
Parents Reply
  • Hi,

    because the rules can be changed on the fly, during translation to optimize results for all translation provider queries (and for all used translation providers at once) after the segment where you discovered you need a new rule to get better results. In that case you add a new rule or edit an existing one.

    Sometimes the rules are NOT known before files get prepared and we also don't use the inbuilt Trados kategories (Date, currency, variables, aso. from TM settings). They are quite often a pain in the a$$ as they offer wrong results (or need to be set for each TM individually) so we override them with our settings to get more correct results. Therefore most inbuilt tagging options might not work.

    With our tool you create the rule globally and it works for all translation providers (TM, MT, aso.) that have that specific rule collection implemented.

    e.g. for one of my last translations I had a name of a machine that was quite straightforward (2 words with a hyphen in between), nevertheless MT managed to offer me around 10 different (often nonsense) translations for it... oO

    So I had to adapt the rules to get it correct all the time and to adapt the term to grammar rules. As the files were also prepared by the client you guess I could not even have implemented interal locked tags or anything else when importing the project, even if I would have known the rules to be used before.

Children
No Data