Translation Memories (TMS)
    1. Phrase
    2. Global
    3. Translation Memories
    4. Translation Memories (TMS)

    Modify Translation Memories

    Content is machine translated from English by Phrase Language AI.

    Translation memories can be edited. This allows for the addition, modification, or deletion of segments. This can be done in Phrase or by exporting the TM for modification in external applications. 

    Dependent on user rights

    • TM attributes can be changed.

    • TM can be deleted (moved to recycle bin)

    • Content can be searched for individual entries to be changed.

    • Import content from other TMs or other CAT tools.

    • Export TM content for editing in other CAT tools and import it back.

    • Align previously translated files with their originals and import them to a new TM.

    Edit Translation Memory Content

    Only Project managers and Linguists with special user rights have the access to edit or delete the content of a translation memory in a specified language.

    To edit translation memory content, follow these steps:

    1. From a translation memory page, enter text in the Search form. Wildcards are supported.

    2. Select the required language from the dropdown list and click Search.

      Results are presented in the search results pane.

    3. Sort results by rearranging the language columns, and clicking on Set up columns to select what columns are visible.

      Hover over TM_search_info.png to see details of the translation unit.

    4. Double click on translation unit text to edit it and TM_deleteunit.png to delete it.

      Caution

      Editing translation units within a TM is permanent. There is no undo function. If performing major maintenance, export the TM in .TMX format and save in .ZIP format as a backup.

    5. Click outside the editing field to save the edit.

    Edit Translation Memory Content Externally

    Translation memories can be exported into a file and edited in an external program without restrictions. Administrators and Project managers can then import the file back. Translation memories can only be exported one by one.

    Exported File Formats

    • .TMX 

      An exchange format for use in other CAT tools. Can also be used for editing content in external tools such as Okapi Olifant.

    • .XLSX 

      An XML based file format for spreadsheet applications.

    Export a Translation Memory

    Exported files contain translation unit metadata.

    To export a translation memory, follow these steps:

    1. From a translation memory page, click Export.

      The Export TMX/XLSX window opens.

    2. Select a format from the dropdown menu:

      • TMX

      • XLSX

        Select which parameters should be exported. If file is to be imported back, it is best to export it with all of the parameters selected to ensure no loss of data.

    3. Click Export.

      File is downloaded to your system.

    Edit Translation Memories in .XLSX Files

    TM replacement

    • .XLSX files can be edited and then imported back into new translation memories that replace the previous version. The old TM should be deleted as to not cause confusion.

    TM update

    • The file can be edited with |update or |delete added to the segment's ID. Marked translation units will then be updated or deleted when the .XLSX file is imported back into the TM.

      • Marking a TU with |delete will delete that entire TU upon being uploaded back to the TMS even if applied to only a subset of languages.

      • |delete cannot be used to delete an entire language from a TM. This can only be done manually through the UI.

      Duplicates created with the update function cannot be automatically removed and must be deleted manually; an error message is displayed if duplicates are found.

      If exporting an .XLSX file without context, edit the translation. If segments are marked with |update and imported back, these segments will be imported as new.

      Context information (Previous, Next and Segment Key) is updated only when the source is updated at the same time.

    Deleting multiple 100% matches

    A 100% match is a segment with the same source but different context. If segments don't require context (every source text has only one translation, regardless of the context in the file), the TM can be exported with no context and imported into a new TM.

    Every duplicate segment in the .XLSX file will be overwritten by the new version when imported back.

    To delete multiple 100% matches, follow these steps:

    1. From a translation memory page, click Export.

      The Export TMX/XLSX window opens.

    2. Select the XLSX format from the dropdown menu.

    3. Deselect these options:

      • Previous segment

      • Next segment

      • Context key

    4. Click Export.

      File is downloaded to your system.

    5. Create a new TM.

    6. Import the downloaded file from step 4 into the new TM.

    7. If no longer required, delete the old TM.

    Deleting segments from one user

    To delete all segments created (or modified) by one user, follow these steps:

    1. From a translation memory page, click Export.

      The Export TMX/XLSX window opens.

    2. Select the XLSX format from the dropdown menu.

    3. Select these options:

      • Previous segment

      • Next segment

      • Context key

      • Created by

        and/or

      • Last Modified by

    4. Click Export.

      File is downloaded to your system.

    5. Open the XLSX file in a spreadsheet editor.

    6. Filter the created_by (or modified_by) columns and select user for whose segments are to be deleted.

    7. In the ID column, append all IDs with |delete.

    8. Turn off the filtering and check that only the intended segment IDs have been appended.

    9. Save the file and import it back into the translation memory.

    Editing a segment's source or target

    Segment IDs appended with |update in an .XLSX file will update while non appended IDs are skipped if the source, target and context is unchanged.

    1. From a translation memory page, click Export.

      The Export TMX/XLSX window opens.

    2. Select the XLSX format from the dropdown menu.

    3. Select these options:

      • Previous segment

      • Next segment

      • Context key

    4. Click Export.

      File is downloaded to your system.

    5. Open the XLSX file in a spreadsheet editor.

    6. In the ID column, append updated items with |update.

    7. Save the file and import it back into the translation memory.

      Appended items are updated in the translation memory.

    Updating new translations to old jobs

    If the translation of a specific word is changed after part of a job (or even after all jobs in a project) are already translated, the translation memory can be updated and reapplied to the earlier work.

    To apply a modified TM to earlier work, follow these steps:

    1. Ensure all old translation is saved in old TM.

    2. From a translation memory page, click Export.

      The Export TMX/XLSX window opens.

    3. Select the XLSX format from the dropdown menu.

    4. Select these options:

      • (optional) ID

      • Text

    5. Click Export.

      File is downloaded to your system.

    6. Open the XLSX file in a spreadsheet editor.

    7. Remove the duplicates.

    8. Edit required segments.

    9. Save the file and import it back into a new translation memory.

    10. Remove the old TM from the project (keep as a backup).

    11. From the job table, select all earlier jobs that require updating.

    12. Click Pre -translate and select Delete all translations.

      Verify the deletion.

    13. Select the same jobs again, click Pre -translate and select Pre -translate where empty.

    14. Set the pre-translation threshold to 100% and set all 100% non-translatables be confirmed.

    15. Click Pre -translate.

      New translations are applied to the earlier work.

    Import Content to a Translation Memory

    External data can be imported into translation memories.

    Segment Handling

    When importing content into a TM that already contains segments, this content is added as new segments, overwrites existing segments, or is skipped.

    • If a segment has the same source, target, and context as a pre-existing segment, the new segment will be skipped and not added to the TM.

    • If the new segment has the same source and context, but with a different target than another segment, the new segment will overwrite the pre-existing segment. This is also applied if no context is available.

    • Segments without context information and identical source but different target will be deduplicated on import.

    • All other segments are added to the translation memory.

    Supported file types

    • .TMX

      An exchange format for use in other CAT tools. Can also be used for editing content in external tools such as Okapi Olifant.

      File must match the source language of the translation memory and at least one of the target languages. If multiple target languages are included, only those present in the translation memory will be imported.

      TMX versions 1.4 and 1.4b are supported.

      .TMX files exported from SDL Trados will not contain all attributes and the following will be changed:

      • creationdate will be converted to created_at

      • creationid will be converted to created_by

      • changedate will be converted to modified_at

      • changeid will be converted to modified_by

      .TMX files exported from WorldServer will not contain all attributes and the following will be changed:

      • creationdate will be converted to created_at

      • changedate will be converted to modified_at

      • x-idiom-source-ipath will be converted to file

      Segment context is coded into WorldServer TMX files and can only be deduced at best.

    • .MXLIFF

      Phrase bilingual XLIFF.

      .MXLIFF files can be imported into a translation memory but this import feature has been designed as a backup solution.

    • .XLSX

      An XML based file format for spreadsheet applications.

      .XLS and .XLSX can be directly imported into a translation memory. Files should be formatted as lists of plain segments without tags and with the language code at the top of each column. Ensure the languages listed in the file correspond to the languages in the translation memory.

      Only use supported language codes.

      Duplicates within the file will be removed.

      Segments translated in Phrase are saved to the TM with context.

      Segments with the same source and target will not be duplicated unless they also have the same context.

      Source segments cannot be empty: importing a spreadsheet with empty source segments and content in corresponding target segments will fail with an error.

    • SDL WorldServer XLIFF

      Bilingual files created in WorldServer.

    Caution

    The created_by and modified_by attributes are associated with the Phrase Users database so only Phrase Usernames are supported. If a username does not correspond with a Phrase username, the field will be imported to the TM as empty.

    Import Content

    To import content into a TM, follow these steps:

    1. From a translation memory page, click Import.

      The Import TMX/XLSX/MXLIFF/SDLXLIFF (WorldServer) window opens.

    2. Click Choose file to add a file, or drag and drop the file to the window.

    3. Apply options:

      • Remove tags

        Removes all of the tags imported from TMX files excepting ({1} or <2}) tags in XLSX and MXLIFF.

      • Strict locale matching

        Prevents the import of a language if it has a different locale than the project.

        Example:

        A file with an EN_US designation will not be imported into a TM designated with just EN and not EN_US.

      • Ignore not confirmed segments (MXLIFF)

        Prevents unconfirmed segments in an MXLIFF file from being imported into the TM.

    4. Click Import.

      The window closes and the file is imported into the TM.

      If there is an error, Warning.png is shown in the Last import field of the translation memory page. Hover over the warning for error details.

      Hover over TM_search_info.png in the Last import field to see the number of processed translation units prior to deduplication.

      The number displayed next to Segments is the number of translation units stored in the TM. Hover over it to see the number of each target translation unit variant with its language code.

    Align Documents to Create a Translation Memory

    Previously translated texts can be converted into a translation memory even if not translated in any CAT tool. Both the original and the translated files are required and must have the same file format. The format must also be supported.

    Note

    Aligning .XLSX files that contain many columns with different content and metadata can cause segmentation issues. In this case, it is recommended to copy and paste content from the original Excel files into a single Excel file, and import the new file into a job as a multilingual Excel.

    Single file pairs can be aligned or batches of files (in ZIP folders).

    Files are imported and segmented. The content will then be aligned and inserted into an .XLSX file for download. The .XLSX file can then be opened in a spreadsheet application for review and if correct, uploaded as an addition to a TM or a new TM.

    Recommendations for use:

    • Create a separate translation memory for the aligned segments (Aligned TM is a suggested name). This helps distinguish between the segments originating from the alignment and the actual translation work.

    • Ensure a review of the .XLSX is done in a spreadsheet application. Alignment results may not be perfect due to differences in content, incorrectly set segmentation, inconsistent formatting in the source and target files, etc..

    • Help translators by adding the aligned .XLSX file as a reference to the project. If incorrect alignments are identified, they can be checked in the reference file.

    Single Pair Alignment

    To align a single pair, follow these steps:

    1. From a Translation memory page, click Align and select 1 source + 1 target file.

      The Align window opens.

    2. Click Browse to upload the original file.

    3. Apply Segmentation rules if required.

    4. Select the Language of the translated file from the dropdown list.

    5. Click Browse to upload the translated file.

    6. Apply Segmentation rules if required.

    7. Click Align.

      The segments are aligned in an .XLSX file and offered for download.

    8. Download the file to your system.

    Batch File Alignment

    To align a batch of files in ZIP format, follow these steps:

    1. Create a .ZIP file with two folders named source and target. These folders contain source and target files. Matching files in the source and target folders must have identical names in order to be paired and aligned.

    2. From a Translation Memory Page, click Align and select Multiple files.

      The Align window opens.

    3. Click Browse to upload the .ZIP file.

    4. Apply Segmentation rules if required.

    5. Select the Language of the translated file from the dropdown list.

    6. Apply Segmentation rules if required.

    7. Click Align.

      The segments are aligned in an .XLSX file and offered for download.

    8. Download the file to your system.

    Was this article helpful?

    Sorry about that! In what way was it not helpful?

    The article didn’t address my problem.
    I couldn’t understand the article.
    The feature doesn’t do what I need.
    Other reason.

    Note that feedback is provided anonymously so we aren't able to reply to questions.
    If you'd like to ask a question, submit a request to our Support team.
    Thank you for your feedback.

    Articles related to Translation Memory