Hello Dear SDL Team :)

At our company we are about to handle as a daily task huge lists of product names and values and we need to properly approach the issue of homonyms.

PERFECT SCENARIO:

We want Trados to be able to automatically (when confirming a segment) save different translations of the same word with a multiple translations penalty so that they are never recognized as 100% matches and used for autopropagation so that the translator always needs to select and confirm one of the terms stored in TM for the specific source word/segment.

The example we tested is for pair EN > DE, for the word "Tube". We wanted to first save two translations for this entry: "Schlauch" and "Tube" and then we tested it for newly added files where we had couple of lines populated with source word Tube. First we set the Multiple translations penalty for the connected TM at 1 and ticked the Translation Memory Update setting: "When target differs.... Add new Translation Units". We imported the file, got the first instance of Tube, translated as Schlauch and Confirmed with Ctrl+Enter , and when the following instance got pretranslated with Schlauch and confirmed as 100% match we changed it to Tube and also pressed Ctrl+Enter so save this entry. This didn't work, in the TM only the last TU ("Tube") was saved as if overwriting the previous one, even  though we ticked the option to add new TUs when target differs. Why is that happening? Is there some other setting that is blocking it?

 
Then we tested the "Add as New Translation" option before confirming the segment (which I would like to avoid as I want to keep the process as simple as possible for our translators, and would like to be able to save different translations for the same segment only by Confirming and Saving the TU (with Ctrl + Enter) and at the same time applying the penalty to those TUs. Is this possible?

Anyways, when I used the Add as New Translation option that actually worked and the translation memory now showed both TUs "Schlauch" and "Tube" with the set penalty, so they were not saved as 100% matches. Still when I added new file to the project with couple of lines populated with Tube source word (Excel files) after the pretranslation was applied (with a setting applied to pretranslate only with 100% matches) the second segment gets pretranslated with TU from this TM and this TU used for pretranslation is marked as Context Match, even though it is saved in TM as a 99% match. How can we avoid this situation that pretranslation applies penalized TUs from TM as context matches?

Thank you very much for any advice

Best!

Kasia