Hello tfennell ,

In SDL Trados Studio, you will typically select file based TM's at the 'All Language Pairs' Level (Global Settings),

These settings are then use for each language pair contained below (multilingual project)

If I now take a specific language pair- you will see the settings are taken from 'All Language Pairs'

If you wish to use alternative 'Translation providers' and different settings for a specific Language Pair- you are able to do this-

Now you can change the settings (search/penalties etc) for the individual Language Pair- which won't affect the Global settings 

Here for example the original All Language Pairs default setting and the changed Minimum Match Value for the individual language pair.

Hope this helps.

I understand all this, but this did not answer my question.  As I said to Paul above:

"I understand how you would want to set something like formatting penalties or the minimum match value at the All-Language level, and then adjust at the Language Pair level as necessary/desired.

What I don't see is why you would want to ever want to include a bi-lingual, file-based TM at the All Language Pair level.

It would seem like the option for using or creating a file-based TM should be unavailable at this level, and that it should only be possible to create or use a multi-lingual server-based TM at this level."

tfennell

tfennell said:

What I don't see is why you would want to ever want to include a bi-lingual, file-based TM at the All Language Pair level.

I thought I'd clarified that point... my bad then.  If you have a multilingual project with 40 language pairs and they all use the same settings (the ones available at All Language Pair level) then this avoids setting them up 40 times.  You have plenty of options at this level too.

Dear Paul,

I trust we will eventually get to mutual understanding.  I fear my question may be so basic it may be causing some cognitive dissonance.

Yes, as I said, I can absolutely understand you would have a 65 percent minimum match value ***setting*** for all 40 languages.

But why would you create or use an EN-fr-FR bilingual ****file-based TM*** at the "All Languages Level?"  It would seem necessary to have *bilingual* file-based TMs created or used for each of the 40 languages at the Language Pair level.

It would seem that only multi-lingual server-based TMs should be available for use or creation at the "All Languages Level."

Is my question more clear now?

Tom

tfennell

tfennell said:

But why would you create or use an EN-fr-FR bilingual ****file-based TM*** at the "All Languages Level?"  It would seem necessary to have *bilingual* file-based TMs created or used for each of the 40 languages at the Language Pair level.

If they were all using the same settings then you would just add all of these 40 TMs in one place, at the All Language Pair Level.

Perhaps you are just reading this phrase incorrectly... it's the place to add "All....... language pairs" when they have the same settings. 

If you like, because you are only working with one TM, you can put it at the specific language pair level.  It doesn't matter at all!  The only time you could see problems is if you mix and match.  So if you use an en-fr TM at all language pairs, and also add one at the en-fr language pair level, then it can cause odd results.  I've seen this when translators receive packages and ignore the message (not as obvious as it used to be in previous versions) that a specific language pair is being used, and the add their own TM at All Language pair level.  So best to use one or the other.

Does this help?

Dear Paul,

I think we may be getting there.  Oh, and I do have more than one TM - I have my big "Mother TM," my "Client TM" and perhaps an "End Client" TM and even an "End Client Matter" TM, and these in 1 or several language pairs. 

Let me put it this way.  I understand that usually I want to configure settings, like

All Language Pairs>Translation Memory and Automated Translation> Search>Minimum TM Match

at the "All Languages Pairs" level.

If I had a multi-lingual Server TM, I would add it at the "All Language Pairs" level

But I see no advantage in putting actual bilingual TMs at the "All Language Pairs" level - this only causes confusion.  Better to put the actual TMs at the "Language Pair" level to avoid confusion and conflicts.

If I had 5 language pairs, putting individual bi-lingual TMs for all these at the "All Language Pairs" level seems very confusing.  Especially if each of the 5 languages has 2-4 TMs ("Mother," "Client," "End Client")

I have been doing a lot of experimentation today with segmentation, and it seems like a TM only has an effect on segmentation if it is placed at the "Language Pair" level.  The only way to have an effect on segmentation at the "All Languages" level is to designate a Language Resource Template at this level.

(I won't go into the sad case of my lost LRT settings here).

Even more confusing, when I try to add a bi-lingual TM at the "All Language Pairs" level, I get the message:
----------------
"This translation memory '4460 PMEM' will not be used for the following language pairs:

Russian (Russia) -> English (United States)

To edit the translation memories used for these language pairs, go to the individual language pair level."
-------------
But then it goes ahead and adds the TM at the "All Languages" level, but apparently ignores the TM settings for segmentation.

The error message is not generated when I use the TM at the Language Pair Level.

It seems like it would be much less confusing to only allow bilingual TMs at the "Language Pair" level and only multi-lingual server TMs at the "All Language Pairs" level.

I see no benefit to allowing 1 or even worse several **purely bi-lingual*** file-based TMs at the "All Languages" level - only additional confusion.

I realize my conclusion may rankle.

I would be very open to any advantages you can state for putting one or several purely bi-lingual TMs in the "All Languages" level, but alas, I have not seen any yet.

Sorry to be such a nag with this, but it would seem very important to get this straight before forging ahead with developing and optimizing workflows that are very much based on the proper use of TMs, especially in the configuration of templates.

I am seriously considering upgrading to Trados PRO, but need substantial clarity in issues like this to move ahead.

If this is an architectural shortcoming of Trados, that is fine - maybe it can be changed in Studio 2021, but if I am missing some advantage from using bilingual TMs at the "All Language" level, I want to understand these advantages.

Thanks once again for your time and attention to this issue.

I know I may be pushing things, but I spend a lot of time on these posts - not to speak of figuring out the Langauge Resource Template issues - if you would be open to a Skype call (which may be shorter) I would be very open to that.  My profile name is Tom.Fennell on Skype.

Best regards,

Tom

tfennell

tfennell said:

But I see no advantage in putting actual bilingual TMs at the "All Language Pairs" level - this only causes confusion.  Better to put the actual TMs at the "Language Pair" level to avoid confusion and conflicts.

You clearly never had to manage a project with 40 language pairs!

But if you see no advantage then don't do it.  Nobody is forcing you.

tfennell said:

I have been doing a lot of experimentation today with segmentation, and it seems like a TM only has an effect on segmentation if it is placed at the "Language Pair" level.  The only way to have an effect on segmentation at the "All Languages" level is to designate a Language Resource Template at this level.

Incorrect.  The vast majority of users don't use the specific language pairs at all so I think we'd have noticed ;-)  Segmentation takes place based on the first TM in your list at the All Language Pairs level.

tfennell said:

Even more confusing, when I try to add a bi-lingual TM at the "All Language Pairs" level, I get the message:
----------------
"This translation memory '4460 PMEM' will not be used for the following language pairs:

Russian (Russia) -> English (United States)

To edit the translation memories used for these language pairs, go to the individual language pair level."

Just informational because you have done what I said is a bad idea... mixed and matched.

tfennell said:

I realize my conclusion may rankle.

Not really.  You can do whatever you want.  I'm just trying to explain to you how it works.

tfennell said:

I would be very open to any advantages you can state for putting one or several purely bi-lingual TMs in the "All Languages" level, but alas, I have not seen any yet.

Well... I don't really need to convince you.  But the obvious is that you don't work with enough language pairs and you probably don't use many of the available features.  If you did then not having to change the Search settings, TM penalties, filter penalties, field updates, match repair settings or even verification options on every TM would be a blessing.

tfennell said:

I am seriously considering upgrading to Trados PRO, but need substantial clarity in issues like this to move ahead.

Why?

Dear Paul,

Thank you for retaining your equanimity with me.

I really am trying to wrap my head around this.  I really am trying my best, and I am not just trying to be obstinant.  I have spent more time testing, and seem to have validated my perspective, and have attached the results. I have spent most of my day trying to get this right.  Let me try rephrasing again.

1. We have no argument about the value of using the "All Languages" level to configure settings for 40 languages.  The thing is it seems to me you can do this without adding a single TM.  Even though the following settings:
Search
Penalties
Filters
Update
Auto-Substitution
Are located in the hierarchy under:
"Translation Memory and Automated Translation"

***You do not actually need to add a single translation memory in order to utilize these settings.****

Further, even if you do put a TM here, it has no effect on these settings.  You still have to set the settings at the level of:
All Language Pairs>Translation Memory and Automated Translation>Search
**whether you have a TM there or not**
These settings are just not stored in a TM.

And when you do use or create a TM at the "All Languages" level, its "settings" (Gear Button) have to do only with things that apply to that specific TM, like its fields. And then there is also the key issue of the Language Resources (Segmentation) in such a TM, which I will analyze in item number 2 below.

You can see what I have described, setting minimum match value at the Language Pair level, without a TM at the Language Pair level, applied in the attached project:
"LTR Test 14.0"

I put the setting of a 100% minimum match value in:

All Language Pairs>Translation Memory and Automated Translation>Search

without ***any TMs at the "All Languages"*** level, and it applied beautifully to my Language Pair (Russian->English) I did not change the standard setting from 70%>100% at the Ru>En Language Pair level - this was done only at the "All Languages" level.

2. I understand your skepticism if the vast majority of users seem to set segmentation at the "All Languages" level by using a bilingual file-based TM at the "All Languages" level.  I am just looking at the results when I did so in my current version of Studio 19, which I attach.

In the attached project LTR Test 15.0, when I used the TM "4460 PMEM" with advanced segmentation rules at the "All Languages" level, the resulting file has only basic, not advanced segmentation.  I did **not**  include the Russian language resource template as a stand-alone setting (for when there is no TM) at the level of:  
All Language Pairs>Translation Memory and Automated Translation>Language Resources.

I understand if I had specified the Russian Language Resource Template in:
All Language Pairs>Translation Memory and Automated Translation>Language Resources.
The segmentation would have been fine, but again this does ***not require specifying a TM.*** 

In addition, I get the error message "No Translation memories found to look up translations for language pair "Russian (Russia)->English (United States). Origin Pre-Translate Files."

This stands in contrast to Project 14.0, where I use TM "4460 PMEM" at the Language Pair level, and the resulting file is processed with beautiful advanced segmentation (abbreviation and segmentation rules applied). 

3. As for me not "mixing and matching," as they say in the great "Show Me" State of Missouri - I can show you!

In the attached archive LTR Test 13.0 I created a project with NO TMs, but when I try to add 4460 PMEM to the "All Languages" level, I get the error message:
"This translation memory '4460 PMEM' will not be used for the following language pairs:
Russian (Russia) -> English (United States)
To edit the translation memories used for these language pairs, go to the individual language pair level."

I am not trying to "mix and match."  Try attaching the TM and show yourself! No need to take my word for it!

----------------

Thanks once again for your patience with this.  I understand that if I am right about these things, it may create some real headaches for you.  But if what I am saying is indeed true and not being dealt with by the vast majority of "nicer," less insistent clients, then I am sure you will agree it will be better to address the matter, for many reasons.

FYI - some good news, I have manually recreated my Language Resource Template with all my abbreviations (over 300) and segmentation rules, and I am storing that file in numerous places to avoid whatever it was that happened to the old one.  I agree we may never know what happened there so don't worry, I don't plan to bug you more on that one!

Best regards,

Tom Fennell

Three archives with three Studio projects.zip

To summarize - if you can apply segmentation rules using a TM a the "All Languages" level, then there is some sense in using a TM (rather than a stand-alone Language Resource Template) at the "All Languages" level.

If you can only use a TM to apply segmentation rules at the "Language Pairs" level, then there would seem to be little reason for using a bilingual file-based TM at the "Language Pairs" level, because all the other settings which would apply to a variety of languages are set with other settings, not with the TM.

Hi tfennell

The segmentation rules are applied (i.e. created, changed, added, removed, reset) by the user to the TM, by accessing the TM settings via the Translation Memories View, as here:

These rules are then applied by the TM itself on creation of the project.You can't apply segmentation rules to an existing project by changing the TM segmentation rules via the 'All Languages' or individual 'Language Pairs' levels as they have already been applied to the sdlxliff by creation of the project (or by using File>Translate Single File).

Any changes you make at either level will only alter the settings in the TM itself and any subsequently-created sdlxliffs/projects.

All the best,

Ali

tfennell

Wow... this is really difficult and time consuming to work through.  It's also wrong on many levels.  'll try to explain but it's quite difficult because I feel as though I'm just repeating myself and then you do something else and think it's right, when it's not.

So some notes as I work through this, maybe not in the right order... just how my brain is working:

tfennell said:

In the attached archive LTR Test 13.0 I created a project with NO TMs, but when I try to add 4460 PMEM to the "All Languages" level, I get the error message:
"This translation memory '4460 PMEM' will not be used for the following language pairs:
Russian (Russia) -> English (United States)
To edit the translation memories used for these language pairs, go to the individual language pair level."

This is because you have checked this box:

You are telling Studio to use a TM at the "Language Pair" level and then you tried to add a TM at the "All Language Pair" level.  Result: works as expected and you are advised to add the TM where you already told Studio you wanted to.  So the application is making sure you do the right thing.

tfennell said:

In the attached project LTR Test 15.0, when I used the TM "4460 PMEM" with advanced segmentation rules at the "All Languages" level, the resulting file has only basic, not advanced segmentation.  I did **not**  include the Russian language resource template as a stand-alone setting (for when there is no TM) at the level of:  
All Language Pairs>Translation Memory and Automated Translation>Language Resources.

Same problem as before:

You added your customised TM at the "All Language Pairs" level, but you have told Studio to ignore that and use the resources at the "Language Pair" level.  Result: Studio did what you told it!

Frankly I think all of your problems are becoming clear now.  You are adding resources to the "All Language Pairs" and then telling Studio not to use them.  I can't do anything with your other project as there seems to be something wrong with it and the translatable file isn't there... but I reckon the same problem is the one you need to investigate.

tfennell said:

As for me not "mixing and matching," as they say in the great "Show Me" State of Missouri - I can show you!

Hopefully this shows you why I said it's not a good idea to mix and match.

tfennell

Wow... this is really difficult and time consuming to work through.  It's also wrong on many levels.  'll try to explain but it's quite difficult because I feel as though I'm just repeating myself and then you do something else and think it's right, when it's not.

So some notes as I work through this, maybe not in the right order... just how my brain is working:

tfennell said:

In the attached archive LTR Test 13.0 I created a project with NO TMs, but when I try to add 4460 PMEM to the "All Languages" level, I get the error message:
"This translation memory '4460 PMEM' will not be used for the following language pairs:
Russian (Russia) -> English (United States)
To edit the translation memories used for these language pairs, go to the individual language pair level."

This is because you have checked this box:

You are telling Studio to use a TM at the "Language Pair" level and then you tried to add a TM at the "All Language Pair" level.  Result: works as expected and you are advised to add the TM where you already told Studio you wanted to.  So the application is making sure you do the right thing.

tfennell said:

In the attached project LTR Test 15.0, when I used the TM "4460 PMEM" with advanced segmentation rules at the "All Languages" level, the resulting file has only basic, not advanced segmentation.  I did **not**  include the Russian language resource template as a stand-alone setting (for when there is no TM) at the level of:  
All Language Pairs>Translation Memory and Automated Translation>Language Resources.

Same problem as before:

You added your customised TM at the "All Language Pairs" level, but you have told Studio to ignore that and use the resources at the "Language Pair" level.  Result: Studio did what you told it!

Frankly I think all of your problems are becoming clear now.  You are adding resources to the "All Language Pairs" and then telling Studio not to use them.  I can't do anything with your other project as there seems to be something wrong with it and the translatable file isn't there... but I reckon the same problem is the one you need to investigate.

tfennell said:

As for me not "mixing and matching," as they say in the great "Show Me" State of Missouri - I can show you!

Hopefully this shows you why I said it's not a good idea to mix and match.

Dear Paul,

OMG!!!!!  Hurrah!!! 

Yes, I just tested it with a new Project 17, and everything works as you described it was supposed to.

Everything came down to that little box "use different translation providers for this language pair," at the Language Pair which I did not previously see or pay attention to!  I was focusing on actually adding bilingual TMs at the All Languages or the Language Pair level, not on this little box which has such a tremendous effect on how such TMs are used.

And that little box gets checked automatically somehow (I won't take your time or mine to figure out that one now, but I trust I'll eventually get it sorted out now).

Steven had pointed this box out his extensive post, but I focused on the issue of bi-lingual TMs, and I skipped over it.

How embarrassing for me, how frustrating for both of us!

The one consolation I can give you is that you have saved me much, much grief if I had continued to ignore this checkbox as I move forward with Studio.

Now I have lots of questions about best practices with using All Languages and Language resource templates, but I will save those for another thread, and another day, so as to give us all a break and so as not to get that issue all mucked up in the midst of this voluminous mess all caused by my blindness in regard to one little checkbox!

I hope you share my feeling that things are no longer "so wrong on many levels", and that things have now been set right.

I definitely owe you one big time for getting me to the "promised land" with this one.  Thank you so much!

Very best regards,

Tom Fennell

Yes, I just looked again, and that checkmark snuck its way into my Default template as checked, which is why the TM I put in at the All Languages level had no effect after I removed the TMs at the Language Pair level.

I pledge to pay attention to every checkbox, radio button and dropdown list box with the most Zen-like attention I can muster going forward.

Thanks once again!

Einen shoenen Feierabend und ein ganz shoenes Wochenende wuensche ich Dir!!! :) :) :)

<sigh_of_relief>Phew... </sigh_of_relief>

Have a great weekend!