Convert TMX to SDLXLIFF or use another technique to achieve this re-reading task?

Simply: 

A client has sent me a 20,000 word document: the source and the target, both Word documents. The source is in French target is in English and the job is a "re-read".

By converting all tables to text in both files, then converting all text to tables, then making these two vertical columns in the same Word file, I then made an Excel file.

I tried converting this to tmx using something like Olifant... or was it using something in SDL itself? Can't recall. Half the units failed. Not unexpectedly (since this is SDL) the "error results" failed to be visible.

Then I found an online spot to convert from Excel to tmx. This file is larger and on examination it appears to contain all TUs. Everything (i.e. source and text) appears to be perfectly "aligned".

Now I want to do an EXTREMELY simple thing: produce an sdlxliff (the memory can be a blank one) so I can go through the file segment by segment in SDL. 

I have tried to do this several times ... but to no surprise whatsoever, nothing appears to be able to achieve this. 

Any suggestions?

emoji
Parents Reply Children
  • OK thanks. I thought it might be something like that. But by God it's annoying.

    Earlier today I said this (replying to Jesús):

    Thanks.

    I ran the plugin for v 2019 (apparently successfully), then closed and re-opened SDL, before doing this. 

    There is no "Add" button. There is a "New" button ... but I couldn't add TMX. But there was a new link on the "File types" page. This let me do it. Link then disappeared.

    Then I tried creating a new project... added the TMX file, .... and got a dreaded Error message: "The source language of this file (French) is different from the project source language".The thing wouldn't open (i.e. wouldn't process the .tmx file).

    This is incorrect. Under language pairs there is "All Language Pairs" and under that "French (France) --> English (United Kingdom)". Furthermore, when I open that .tmx in Olifant (with all the TUs, perfectly aligned) it says very clearly Source: FR, Target: EN.

    Any suggestions?

    In addition to finding out what's wrong there, I'm now trying to find out about the "bilingual Excel file" solution. I currently have no idea how that works... Found a vid though.

    NB a "bilingual Excel" solution is not ideal, because I would then need to reconstruct the re-read Word file. 

    emoji
  •  

    This is incorrect. Under language pairs there is "All Language Pairs" and under that "French (France) --> English (United Kingdom)". Furthermore, when I open that .tmx in Olifant (with all the TUs, perfectly aligned) it says very clearly Source: FR, Target: EN.

    Does it say fr-FR and en-GB?  Or just very clearly the wrong thing?

    Paul Filkin | RWS Group

    ________________________
    Design your own training!

    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

    emoji
  • You mean in Olifant? It says "Source: FR" and "Target: EN". That's all it says.

    Could there be some way of editing some bytes of the .tmx to make it says PRECISELY what SDL needs it to say, if "fr-FR" / "en-GB", for example? I have Notedpad++ and have some experienced of this byte-level intervention due to some of the most infuratingly intractable SDL errors.

    emoji
  •   

    You mean in Olifant? It says "Source: FR" and "Target: EN". That's all it says.

    Yes... and probably in a text editor too.  You don't need Olifant just to look at a TMX.  Does the file look like this:

    <?xml version="1.0" encoding="UTF-8"?>
    <!DOCTYPE tmx SYSTEM "tmx14.dtd">
    <tmx version="1.4">
      <header
        creationtool="ExampleTool"
        creationtoolversion="1.0"
        segtype="sentence"
        adminlang="EN"
        srclang="FR"
        datatype="PlainText"
        o-tmf="TMX"/>
      <body>
        <tu>
          <tuv xml:lang="FR"><seg>Bonjour</seg></tuv>
          <tuv xml:lang="EN"><seg>Hello</seg></tuv>
        </tu>
      </body>
    </tmx>
    

    Or this:

    <?xml version="1.0" encoding="UTF-8"?>
    <!DOCTYPE tmx SYSTEM "tmx14.dtd">
    <tmx version="1.4">
      <header
        creationtool="ExampleTool"
        creationtoolversion="1.0"
        segtype="sentence"
        adminlang="en-GB"
        srclang="fr-FR"
        datatype="PlainText"
        o-tmf="TMX"/>
      <body>
        <tu>
          <tuv xml:lang="fr-FR"><seg>Bonjour</seg></tuv>
          <tuv xml:lang="en-GB"><seg>Hello</seg></tuv>
        </tu>
      </body>
    </tmx>

    If it's not the latter then that might be your problem.  Just edit with a text editor - search and replace the language codes.

    Excel is probably easier for you.

    Paul Filkin | RWS Group

    ________________________
    Design your own training!

    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

    emoji
  • That did it ... thanks a lot!

    ("Verify answer" appears to mean "Thanks Paul, this is the accepted answer"). Unusual forum this one. 

    emoji
  •  

    ("Verify answer" appears to mean "Thanks Paul, this is the accepted answer"). Unusual forum this one. 

    It means, please verify it as the correct answer, if its helped.  Not all users do that but it's very helpful for several reasons:

    1. Users finding this thread will see this at the top:
      Screenshot of a Trados Studio forum thread titled 'Convert TMX to SDLXLIFF or use another technique to achieve this re-reading task?' with a verified answer marked by a green checkmark.
      Which can be helpful in really long threads as you probably just want the answer to a similar question.

    2. Al verified answers go into a moderation queue that we use to continually train the AI.  In that queue moderators can make sure the correct queation and answer are being mapped, validate them, and add the Q&A into the training for the AI.
      Users sometimes mark their own questions as verified, or when there are multiple questions in the same thread verify an answer that is not related to the original question.  So we have to be sure that the Q&A makes sense.
      This is also why we prefer to have one thread per question.

    Paul Filkin | RWS Group

    ________________________
    Design your own training!

    You've done the courses and still need to go a little further, or still not clear? 
    Tell us what you need in our Community Solutions Hub

    emoji


    Generated Image Alt-Text
    [edited by: RWS Community AI at 10:25 AM (GMT 1) on 23 Oct 2025]