Pretranslating segments including + or - signs

Hi all,

We have lately found an issue with segments containing the "+" sign that were not correctly pretranslated.

It seemed at first that the problem came from the TM but I kept on investigating and realised that the issue just came from pretranslation.

I checked the TMs and they seem correct so I did different tests with and without TMs and with different file formats and came to the conclusion that when the segment in the files is exactly the same as the TU in the TM, then the + sign is maintained. When the segment changes because of a different number, for example, then the + sign disappears.

Case A - Exactly the same content in the segment and the TM:

Screenshot of Trados Studio showing segments with '+' sign not correctly pretranslated, highlighted in blue and yellow.

However, in the below image, the TU appears as 100% but it's not exactly the same, as the number varies from 2,1% to 3% in the source segment, and in that case it does not include the + sign in the target (see in blue in the TM box and in yellow in the rest of the segments), while when the source segment is exactly the same as the TU in the TM, it applies the number and the + sign correctly (in green in the below image):

Case B - Number differences between the segment and the TM:

Screenshot of Trados Studio with segments where '+' sign is correctly included in green highlight, and incorrectly excluded in yellow.

This does not happen in IT, where the + sign is not used (it's replaced by text) and the numbers are therefore well localised:

Screenshot of Trados Studio displaying segments in Italian where '+' sign is replaced by text, highlighted in green.

And for some reason the ES segments are also well localised:

Screenshot of Trados Studio showing segments in Spanish correctly localized with '+' sign, highlighted in green.
I checked the measurement settings in all TMs but all languages are using the same options, so there shouldn't be a difference there.
On the other hand, we have a different behavior when we use the "-" signIn that case, Studio seems to maintain it as part of the number, although it does not process it well either. Please see the below examples.

Case A: Segment 2 lacks the +, while segment 4 includes it when it shouldn't, just because I added it manually in segment 3:

Close-up of Trados Studio segment where '+' sign is manually added, highlighted in yellow.

Case B: Segment 3 has been updated adding the + manually too, so now segment 4 recognises it but adds it when it shouldn't, maintaining the - as well:

Close-up of Trados Studio segment where '+' sign is recognized after manual addition, highlighted in yellow.

And the same happens with IT where we saw that the files behave differently because + is not included in IT, but now adding also the -:

Close-up of Trados Studio segment where '+' sign is added incorrectly after manual input, highlighted in red.

Another case is when I do not use a TM, so TUs don't get updated:

Screenshot of Trados Studio with segments not using TM, '+' sign stays but number is not recognized as localized.

In those cases the + sign stays but I have also added a space between the + and the number and between the number and the % and then Studio does not really recognise the number as correctly localised.

To sum up, it would be great if the SDL development team could check this behavior as it's clear we cannot trust how Studio manages the segments containing the + sign, especially when using a TM, and this can change the whole meaning of a translation.

Sorry for the long post and please let me know in case of queries. 

Thanks very much and kind regards,

Elisa



Generated Image Alt-Text
[edited by: Trados AI at 1:25 AM (GMT 0) on 29 Feb 2024]
emoji
Parents Reply Children