John Narins

John Narins said:

How do I switch those specific TUs in FILE 1 out for the better translations in the early TUs of FILE TWO?

If you have this in your TM, presumably replacing the one you did in File #1, then just pre-translate File #1 with the updated TM and use this option:

As long as you have this set to 100% you should achieve what you're after.  You could also just review File #1 using this option:

This will run through your file and stop at every segment where you have something different in your TM.  So this way you can actively validate the work as you go.

The comment from Steve is also valid.  Working with all files in the project is valid, although it wouldn't help in your specific case since you'd still end up with the same problem of how to deal with correcting translations in the first file.

That sounds promising! Let me answer questions and check that it's what I mean.

Yes, I translated File 1 and confirmed the results.

Then I created and translated File 2 and did the same. There is some overlap - that is, there are some TUs that were translated twice.

(It is not always convenient to translate one gargantuan file.)

Now, for each of the overlap TUs, I have two confirmed translations in memory, right?

So I think I want two things here:

A) for the overlap TUs in File 1 to convert, en masse, to the versions created in File 2. Paul, will your method know that the File 2 translation is the better match? Since both the File 1 and File 2 translations have been confirmed, wouldn't they both be 100% matches?

and

B) I really only want the File 2 translations for the overlap TUs to remain in memory.

I suppose I could just ERASE the contents of all the overlap TUs in File 1, confirm them with nothing in them... would that erase the File 1 versions of those translations from the TM? And then auto-translate those sections.

What do you guys think?

John Narins

John Narins said:

What do you guys think?

I think you should make this easy on your self and do this:

1. create a new TM and add it to your project for the second file.  Remove or disable all other TMs
2. populate this new TM with your translations (Batch Tasks -> Update Main Translation Memories)
3. add this updated TM to your first project and remove/disable all other TMs
4. now run the "Confirm and translate until next fuzzy match"

It may make sense to do this from the segment you know this requirement starts... so from the point you actually have a match in your updated TM.  Otherwise you'll have to confirm every segment in the first part of the file that isn't in your new TM at all yet.

Paul, thanks! That sounds right, almost perfect.

The only point I want to clarify is this: when you say "add this updated TM to your first project"...

...there's only one project. Two files within a single project. The first file was an entire novel. The second file, which I added, was the 2nd (middle) 1/3 of that same novel. There will be a third file, but no conflicts, as only about 49% of the first file was translated.

Does that change the process at all?

John Narins

John Narins said:

...there's only one project

Perfect!

John Narins said:

Does that change the process at all?

Nope... overall idea is the same.

1. create a new TM and add it to your project. Remove or disable all other TMs
2. populate this new TM with your translations from the second file (Batch Tasks -> Update Main Translation Memories)
3. now run the "Confirm and translate until next fuzzy match" against the first file only, while the file is open in the editor (forgot that last part...)

Paul, thanks! Before Step 3, though, I need to re-enable the old TM, right? Otherwise none of the NONdoubled TUs will have ANY matches, not just fuzzy matches, I think.

Also, for the Step 3 stage, might the "keep most recent translation units" option in [batch tasks/Update Main Translation Memories] have been useful here?

John Narins

I think the key point here is use the knowledge you have here to apply it in the most sensible way.  Your question was this:

John Narins said:

How do I switch those specific TUs in FILE 1 out for the better translations in the early TUs of FILE TWO?

To do this here are the complete instructions:

1. create a new TM and add it to your project. Remove or disable all other TMs
2. populate this new TM with your translations from the second file (Batch Tasks -> Update Main Translation Memories)
3. position the active segment in the first file so that it's the first segment before the improved translations
4. now run the "Confirm and translate until next fuzzy match" while the file is open in the editor
5. this will allow you to correct all the translations based on the second file
6. once this is complete you now have a file that is fully translated as the first part of file #1 is already complete and presumably contains translations you are happy with
7. if you want all of the file in this new TM then run the batch task to update your Main TM again and now you have the contents for both files in the same TM

If you want to tidy up your original TM use the TM Maintenance View to find all the duplicates and delete the ones you don't want.  I don't know enough about the history of your TM to know how best to do this efficiently.  Maybe you used a field for this project that you can use to remove all translations against this field and then you could import the new TM into your old one?  Maybe you could do this on the basis of dates?  Maybe you just need to go through a manual process?  Or maybe you remove/disable the new TM, add the original back, run the "Confirm and translate until next fuzzy match" to find the ones that you don't want and delete them in the TM Results window as you work?  Many ways to tackle this... depends on your preference.

Paul, thanks! Now I'm having a slightly different problem.

This is what I did:

Within Project Settings, I created a new TM. Then, in Project Settings, I made sure the old TM was unchecked (disabled) and the new one was checked (enabled). I made sure the "update" box was ticked for the new TM. Then, with File 2 and only File 2 open in the editor, I and did Batch Tasks/Update Main Translation Memories. BUT when I put the cursor in a TU... no TM results. I tried Batch Tasks/Update Project Memories, I tried Batch Tasks/Populate Project TMs... didn't help.

HOWEVER... I tried adding a space to a TU and then deleting it, and then confirming that TU. THAT did help - when I put the cursor in the TU, I got the expected TM result. (And it isn't that the other TUs aren't confirmed, they are.)

I must be doing something wrong - something really obvious! But I can't quite figure out what it is!

Obviously, until I solve this, I will be unable to test and apply our fix!

John Narins

maybe something simple such as you didn’t confirm the segments yet? There are options to update with segments that have a different 

hard to help without seeing your settings.

Tell me what to show you!

Yes, I'm talking about confirmed segments, yes (I made sure to mention that, in fact, in the previous post). 

John Narins

Your project settings would be a good start.

Which settings? There are a ton of them, right?

Which settings? There are a ton of them, right?

John Narins

ok… the ones that show your TMs.

John Narins

You could also import your sdlxliff for file#2 into your new SDLTM directly.

This?

Paul said:

You could also import your sdlxliff for file#2 into your new SDLTM directly.

True, but then I wouldn't have figured out what's going wrong!

John Narins

I can't see that very well, it's a bit small!  But I can just about make out that you are using a Project TM and this isn't anything we mentioned here.  I'm not sure why you have done this because all I said was this:

Paul said:

• create a new TM and add it to your project. Remove or disable all other TMs
• populate this new TM with your translations from the second file (Batch Tasks -> Update Main Translation Memories)

To update a Project TM you need to use "Update Project Translation Memories".  However, that presumably contains everything you have done so far and would not really be the process we have discussed here.  You should disable all the TMs in that screenshot, create a brand new one and just add it, but not as a Project TM.

Interesting. I wouldn't have thought that would work much differently, but now I'll draw the distinction.

STILL mysterious - I *DID* do BatchTasks-->UpdateProject Translation Memories," and I *STILL* didn't get ANY TM results when I placed a cursor in a segment. Why would that be?

Let me try the other thing, then...

John Narins

This is taking far too much time.  If you have some time available after 1730hrs CET today I'll be happy to take a call with you so we can look at this together.  I think this will be a much more efficient way to deal with this for us both.  If that's ok then I'll drop you an email with a link to where we can do this later.

Okay, I tried it twice. The first time, I created a new TM. File 2 was open in the editor. In the project settings, Then I used batch tasks-->update main tms. It worked, the segment translations are now in the TM. BUT... I get something I've never seen before - the previous two TMS are there, disabled, as I'd left them. THEN there's a line for the specific xliff file for File 2. And only then the new Main TM. so it seems to be saying that it's applying that TM *ONLY* to that xliff, not the whole project. And I can't see how to tell it to apply that to ANOTHER xliff file...

SO... I removed all the TMs and created YET ANOTHER new main TM. This time I closed all the files first, nothing in the editor. THEN I opened File 2 in the editor, made sure the new TM was the only one there, made sure it was enabled, and ran the batch tasks--Update main TM memories. The TM populates. Then I close that file and open File 1. Just in case, for now, I un-tick the "update" option for the new TM. I go to the first segment that's in both files that I know is translated differently. I do get the option to substitute the new translation, so - yes, that works, thank you!

Remaining questions:

1.) The "Confirm and translate until next fuzzy match"  works if I want to choose each time... but if I want to, can I tell the program just to overwrite whatever's in the xliff with whatever's in the TM for all those segments, rather than reviewing each one?

2.) Why didn't the "update project TMs" batch task NOT work to update the TM when I had (incorrectly) created a project TM?

3.) So there's a function to set TMs to apply only to selected xliffs in a project? Where can I find instructions as to how that works?

Again, thanks for everything!

And happy to do that, although I think we're pretty much there. Yes, it did take longer than either of us expected!

Paul said:

This is taking far too much time.  If you have some time available after 1730hrs CET today I'll be happy to take a call with you so we can look at this together.

Paul, if this offer is still open, I'd love to do it!

I'm 7 hours ahead of (later than) EST. Let me know what works! And - my thanks in advance!