Back to top image

Using MT to Overwrite Existing Machine Translations

 

If you would like to use an MT workflow to overwrite published Machine Translations, please follow the steps outlined below. 

This may be necessary in any of the following scenarios:

  • If Glossary Term Insertion was enabled after the original Machine Translations had reached the Published step, or the project's glossaries were updated. 
  • If TM Match Insertion was enabled after the original Machine Translations had reached the Published step, or after the associated TM was updated.
  • If a change was made to the threshold for TM Match Insertion after the original Machine Translations had reached the Published step.
  • After a change was made to the translation prompt (if GPT is used as a translation provider).
  • If the desired level of formality was updated (where supported) after the original Machine Translations had reached the Published step.

Tip: If you would like to overwrite existing Machine Translations for a different reason, please reach out to your Customer Success Manager for further instructions.

How to Overwrite Published Machine Translations

The previous Machine Translation needs to be discarded before a new one can be provided.
To do so, please follow these steps:

  1. If a post-translation step was used and the original translations were saved to the Translation Memory, we would recommend correcting the relevant TM entries before proceeding.
    • This step is not necessary if a human linguist will edit the translations again and save them to the TM once the content is authorized for translation in step 8. 
  2. In the Leverage Configuration used for your project, temporarily disable the SmartMatch setting “Restore translation: Exact match for previously removed unpublished translations”.
    This step is necessary to avoid re-using the previous translation automatically. 
  3. [Optional] As an optional, but recommended step, apply a tag to the relevant strings, so they can be found more easily: 
    • Filter for the strings you would like to re-translate and select all strings matching that filter. 
    • From the “Actions” menu, select “Add / Remove Tags”. 
    • Apply a tag that you will be able to search for in step 5.
  4. Exclude the strings that you would like to re-translate:
    • Filter for the strings you would like to re-translate and select all strings matching that filter. 
    • Select the option “Exclude” from the “Actions” menu.
    • These strings are now excluded and not part of the original Job anymore.
  5. Navigate to the Strings View of your Smartling project and search for the strings to re-translate:
    • Use the Languages filter to narrow down your search to the relevant target languages. 
    • [Optional] Use the Tags filter to search for the tag you added earlier, in step 3. 
    • If you did not apply any tags, use any other necessary filters to locate the strings that you would like to re-translate.
  6. The excluded strings can now be restored:
    • Select all strings to re-translate. 
    • From the “Actions” menu, select the option “Restore Excluded”.
    • The strings are now restored and “Awaiting Authorization”. 
  7. The strings then need to be added to a translation Job:
    • From the “Actions” menu, select “Change Job”. 
    • Add the strings to a new or existing Job. You can add the strings to the original Job or create a new one.
  8. Authorize the strings for translation again.
    • You can choose the same Machine Translation workflow that was used for the original translations, or a different workflow.

The existing translations can now be overwritten with new (Machine) translations. 

Note: The steps outlined above only describe how to re-translate strings in Smartling. If you are using a custom API integration or a direct integration with a third-party platform, for example one of our connectors or plugins, additional steps may be required to reflect the change also in the connecting platform.

 

Table of Contents