Masoud Moghaddam

Hi,

Thank you for your very good question as it touches so many core concepts and best practices depending on your own specific use case.

• Default behaviour
• Settings (possible solution)
• Integrated QA
• Project TM’s

Default TM behaviour

Firstly let me confirm that TM’s don’t intentionally store “duplicate” segments.

What constitutes duplicate in the eyes of SDL Trados Studio is primary based on 2 source segments being identical, regardless of different target translations. Keeping both versions of the translation is possible for legitimate reasons, but it does require a change of settings or user behaviour.  

The ideal is as you described, smart to recognize and save the new version of the segment over the old ones in TM 

Settings (possible solution)

When making changes to a segment, how are you updating the TM?

What I would check is the following:

1. File – Options – Automation – After Confirming Segments Manually = “Update Translation Memory” should be enabled

2. During translation, changes are saved (segment confirmed) on the fly by Ctrl + Enter and not via the default short cut of Ctrl + Shift + U (Add as new TU).
   *If need be we could try rest your short cuts

        3. If as part of a batch task, it is possible that your Update TM settings are enabled to keep both versions?

May I also take the time to explain best practices or additional steps one can take to help safeguard both the quality of what goes into your TM 

Integrated QA

As you say unintentional mistakes are near enough guaranteed, but indeed the technology should be intelligent enough to aid the end user. This is where the integrated quality assurance settings really showcase consistency of quality and most certainly boost end user efficiently.

QA settings cover a wide range of problem area’s to be cautious of. Ranging from caps, through to extra spaces, end punctuation differences and so much more.  Depending on the success criteria such mistakes can be categorised as simply a note, warning or a flat out error which you should not leave as is.

Once your QA settings have been setup, they run on the fly as your translate (in parallel when confirming a segment) making it hard to ignore those warnings/errors. Alternatively they can also be run at the end of your translation project by selecting F7. Either way it’s a means to safe guard what goes into your TM, as I understand the importance of maintaining the quality of your TM for future reuse.

More information about QA settings can be found here:  https://docs.sdl.com/813470/662346/sdl-trados-studio-2021-sr1/verifying-translations

Project TM’s

Because you mentioned revising segments and version control, what you may also find comforting is the concept of Project TM’s.

These can be created as part of your project and work alongside your main TM. All your legacy translations still originate from your trusted main TM. But any alternations, or new translations get temporarily stored in this second TM known as a project TM.

Only once the translations have officially been signed off, do you then update your main TM based on the translations within your sdlxliff files.

More information on project TM’s can be found here: https://docs.sdl.com/813470/548261/sdl-trados-studio-2021-sr1/running-batch-tasks--populate-project-translation-memories

I hope this was helpful and insightful. I do appreciate the effort it takes to keep TM databases clean so if there is anything more we can do to help please do reach out.

Regards

Lydia

Masoud Moghaddam

Hi,

Thank you for your very good question as it touches so many core concepts and best practices depending on your own specific use case.

• Default behaviour
• Settings (possible solution)
• Integrated QA
• Project TM’s

Default TM behaviour

Firstly let me confirm that TM’s don’t intentionally store “duplicate” segments.

What constitutes duplicate in the eyes of SDL Trados Studio is primary based on 2 source segments being identical, regardless of different target translations. Keeping both versions of the translation is possible for legitimate reasons, but it does require a change of settings or user behaviour.  

The ideal is as you described, smart to recognize and save the new version of the segment over the old ones in TM 

Settings (possible solution)

When making changes to a segment, how are you updating the TM?

What I would check is the following:

1. File – Options – Automation – After Confirming Segments Manually = “Update Translation Memory” should be enabled

2. During translation, changes are saved (segment confirmed) on the fly by Ctrl + Enter and not via the default short cut of Ctrl + Shift + U (Add as new TU).
   *If need be we could try rest your short cuts

        3. If as part of a batch task, it is possible that your Update TM settings are enabled to keep both versions?

May I also take the time to explain best practices or additional steps one can take to help safeguard both the quality of what goes into your TM 

Integrated QA

As you say unintentional mistakes are near enough guaranteed, but indeed the technology should be intelligent enough to aid the end user. This is where the integrated quality assurance settings really showcase consistency of quality and most certainly boost end user efficiently.

QA settings cover a wide range of problem area’s to be cautious of. Ranging from caps, through to extra spaces, end punctuation differences and so much more.  Depending on the success criteria such mistakes can be categorised as simply a note, warning or a flat out error which you should not leave as is.

Once your QA settings have been setup, they run on the fly as your translate (in parallel when confirming a segment) making it hard to ignore those warnings/errors. Alternatively they can also be run at the end of your translation project by selecting F7. Either way it’s a means to safe guard what goes into your TM, as I understand the importance of maintaining the quality of your TM for future reuse.

More information about QA settings can be found here:  https://docs.sdl.com/813470/662346/sdl-trados-studio-2021-sr1/verifying-translations

Project TM’s

Because you mentioned revising segments and version control, what you may also find comforting is the concept of Project TM’s.

These can be created as part of your project and work alongside your main TM. All your legacy translations still originate from your trusted main TM. But any alternations, or new translations get temporarily stored in this second TM known as a project TM.

Only once the translations have officially been signed off, do you then update your main TM based on the translations within your sdlxliff files.

More information on project TM’s can be found here: https://docs.sdl.com/813470/548261/sdl-trados-studio-2021-sr1/running-batch-tasks--populate-project-translation-memories

I hope this was helpful and insightful. I do appreciate the effort it takes to keep TM databases clean so if there is anything more we can do to help please do reach out.

Regards

Lydia

Lydia Simplicio said:

I hope this was helpful and insightful. I do appreciate the effort it takes to keep TM databases clean so if there is anything more we can do to help please do reach out.

Hi Lydia

Thank you for your perfect description and helpful answer. The "Update Translation Memory” has already enabled. If I have more questions about this subject, I'll ask them here.

Best

Masoud

What a wonderful way to start the day! Thank you for letting me know what the true source of the issue was.
Hopefully the the additional pieces of information such as QA Verifications will be something to review as you see fit.