A short comparison between the contents of TMX data coming from different translation tools (Trados 2007, Studio, memoQ).
TMX can contain tool-specific information (additional fields, segment status, segment context, alignment penalty, etc.) that’s not easily imported into other tools.
If a TMX import does not work, look at the language identifiers first.
A memoQ TMX file contains TM-level (project. client, etc.) and segment-level (project etc., but also changeID, client, corrected, aligned, context)metadata.
The information entered into the User and meta-information fields is case-sensitive, so this can lead to data duplication.
For the moment, the meta-information fields for each new project is limited to username, project ID, domain, client, subject.
Since the user ID is overwritten when a new user changes a field, a workaround is to use the “subject” (or another) field to specify the name of the original translator.
Trados has both creation ID and change ID fields. memoQ only has the change ID field, so when you import from Trados, the change ID from Trados will be imported and the creation ID will be lost.
Trados Studio includes context information in its TMX exports, but these are in form of hashes (a long string of digits), and do not contain the actual context strings like memoQ does. As a consequence, context information cannot be imported into memoQ from Trados.
The memoQ TM import settings
The field Process TRADOS TMX for best results in memoQ should be used if both the TM and the translatable files are in Trados format.
Import <ut> as memoQ tag: this allows support for legacy TMs. “UT” means “Unknown tag” here.
For Trados versions up to 2007, it’s very important to apply the penalties. Otherwise the statistics will treat segments that are almost identical except for punctuation, tags, and formatting, as identical.
If Use context is selected, you should not use Allow multiple translations. Two identical segments with different context will both be saved to the TM.
Angelika showed us some quick methods for doing maintenance on TMX files using Olifant. Delete duplicate and inconsistent segments.
It’s probably best to re-create a TM from scratch after the edits, rather to import back into the original memoQ TM.
Import of a Trados TM into memoQ
In order to map the Trados fields to the corresponding memoQ fields, a search & replace is done in the TMX file, using a text editor. The values contained between quotes in the <prop type> fields are replaced by the corresponding hardcoded field names for memoQ.
If the search&replace is too complex, it’s probably better to export from CSV in Olifant and import the CSV into memoQ, where fields mappings can be set up during the pre-import procedure.
Based on the participants’ inputs, here are some of the issues that users face when migrating their resources to memoQ from other tools.
- Loss of metadata
- Loss of tags
- TM files too big for import
- Moving TMs from one server to another
Moving term bases