Probability score: 0.8  

Question improvement suggestion: Provide specific details about the settings you used when importing the TMX file and clarify if you have adjusted any of the default settings in the Project Settings.  

In the meantime, based on the information you have provided, here is a suggested answer to your question:  

Hello Rosanna,  

Thank you for your detailed question! It sounds like you're encountering a common challenge when working with Machine Translation (MT) in Trados Studio. Here’s a step-by-step guide to help you prioritize your Translation Memory (TM) matches over the MT results and ensure that fragment matches are visible:  

**Step 1:** **Adjust Project Settings**  

- Go to **Home > Project Settings**.  

- Under **Language Pairs > All Language Pairs > Translation Memory and Automated Translation > Search**, ensure that the option **Look up segments in MT even if a TM match has been found** is **unchecked**. This will prevent MT from overriding TM matches.  

**Step 2:** **Set Minimum Match Value**  

- In the same settings area, check the **Minimum match value**. You may want to set this to **100%** for pre-translation tasks. This way, only segments with exact matches will be pre-populated, and MT will not be used unless there are no TM matches.  

**Step 3:** **Enable Fragment Matching**  

- Still in the Project Settings, under **Segment Fragment Matching Options**, ensure that **TU fragment** is checked. This will allow fragment matches to be displayed.  

**Step 4:** **Check AutoSuggest Settings**  

- Go to **File > Options > AutoSuggest** and ensure that the **Fragment Matches** option is selected. This will help you see fragment matches as you type.  

**Step 5:** **Review the Fragment Matches Window**  

- If you want to see fragment matches even when there are TM results, go to **File > Options > Editor > Fragment Matches Window** and ensure that the option **Display results from the Fragment Matches window if the TM lookup returns no results** is selected.  

**Step 6:** **Pre-translate Your Files**  

- After adjusting these settings, run a **Pre-translate** batch task to ensure that your TM matches are prioritized and that the MT does not overwrite them.  

By following these steps, you should be able to see your TM matches more prominently and still access fragment matches without them being overridden by MT results. If you continue to experience issues, please double-check the settings and consider creating a new Project TM specifically for this task.  

I hope this answer addressed your question. If not, please reject the answer, add the question improvement suggestion information, and click the RWS Community AI checkbox on the right to try again.  

RWS Community AI

I have followed these steps, but I still get "exact match found (no fragment search performed)" in the fragment matches window. 

Hi Rosanna Lithgow 
I guess this intended behaviour as Translation Units (TUs) of a TM are always seen as higher quality translation source compared to concordance or fragement matches.
Studio displays only fragment and concordance matches, if no match can be found in the TM(s) of your project:

File > Options > Editor

I think, the best way to achieve what you want, namely having MT from your customer displayed as well as using your TMs for exact, fuzzy, concordance or fragment matches, is to ask your customer not to send you a tmx but rather a pre-translated sdlxliff or even better a project package in which the sdlxliff is pre-translated with the customer's MT engine (means the target text column in the editor is already filled with the MT outputs of your customer).

If you continue working with the tmx export, the only way to dinstinguish between your own TM matches and the ones from your customer is to work with field values and filter penalties. However, this only shows you where the matches come from in the Translation Result Window, you won't receive concordance/fragment matches this way.

To use the filter options:

Add field values in the tmx import when creating your customer's MT TM:

Then, set filter penalties in the projects settings according to the field values you added before:

Test if you need to set the condition to "NOT" as filters work reverse.

A concordance search can always be performed manually by highlighting the source text or parts of it and pressing F3.

Regarding the order in which matches are displayed in the Translation Results Window: You can choose TM order in the project settings by moving them up/down using the blue arrow:

By moving the TMs up/down you can decide which matches of which TM should be displayed first/second, etc. in your Translation Results Window.

If you'd like to see only your TMs and therefore see concordance/fragment matches of your TMs but still use your customer's MT outputs, I'd recommend asking for a pre-translated document.

I hope this helps.

Best

Seraina

Thank you, Seraina. I will try the field values and penalties suggestion on my next project. 

The interesting thing is, fragment matches show up as suggestions as I type, so they must be "hiding" somewhere.

I think the green and yellow squares indicate a fragment match, right? In the segment where this suggestion shows up as I type, no fragment match show up, just the MT in the translation results window, or I can find the term by doing a physical concordance match (F3). I find it very strange and frustrating that I can't get to see the fragment match that is obviously there somewhere. It should be obvious that these matches are more important than a machine translation and should be given priority by Trados.

Rosanna 

Hi Rosanna Lithgow 

It actually is always there in the Editor view:

But if a TM match is found, it automatically jumps back to the Translation Results Window (red arrow).

This is default behaviour.

Studio only jumps to the Fragment Match Window if no matching TU can be found the TM. In your case, there is always a matching TU form the TM as you created a TM which contains a match for every segment. Therefore, Studio will never jump into the fragment section automatically.

The only way to pin the Fragment Matches Window is to right click on the window and set it as "Floating":

You can now move it wherever you like (e.g. to a second screen)

or integrate it in the Editor View by moving to the position you'd like to have it displayed:

Result:

If you later on whish to change the Window Layout, click on Reset Window Layout under the View tab:

Maybe this helps.

Best 

Seraina

PS: Your MT is preceived as TM by Studio as you enter the MT outputs in a TM. The source of TUs in a TM doesn't matter for Studio to be detected as a TM.
Therefore, I'd still consider asking your client to provide a project with a pre-translation including the MT outputs.

No, the fragment matches are NOT there, that's the problem. When I click on the fragment match window I get the message "Exact match found [i.e. the MT] (No fragment search performed). This is the behaviour I want to change. I want it to perform a fragment search AS WELL as showing med the MT translation in the translation resaults window.

This is with "Search for fuzzy matches even if exact match is found" ticked under Project settings - translation memory - search:

Not sure if minimum match value is correct, but I get the same behaviour with it set to 70, so it doesn't seem to make a difference.

Rosanna

Hi Rosanna Lithgow 

As said, as long as you have exact or 100% matches from a TM, there won't be performed a Fragment Match Lookup.

Search for fuzzy matches even if exact match is found only applies to the Translation Results Window.

This means that not only 100% or CM match is displayed but also 99% or lower scored matches.

However, a Fragment Match Lookup still will not be performed.

Therefore, I'd suggest to either use the penalty feature or to ask your client for a pre-translate.

Best

Seraina

Ok, I think I understand. Thank you.

I couldn't quite let this go, so I asked ChatGPT. I haven't confirmed all the answers, but it makes sense. This is what it said:

In short: You can’t currently force Trados to display fragment matches when a 100% (exact) machine translation match is already present; the software is designed to skip fragment recall in that case, primarily for performance reasons.

So, just like you said, Seraina. Regarding the box "search for fuzzy matches even if exact match is found" it said:

• This applies to Translation Memories (TMs) only.

• Normally, if Trados finds a 100% TM match, it stops searching and won’t return fuzzies for the same segment.

• It does not apply to machine translation (MT) matches — only TM searches.

• It does not force Trados to also show fragment matches when an exact match is present.

And regarding fragment matches showing up as AutoSuggestion entries:

Those are indeed upLIFT fragment matches coming from your TM, but they’re only exposed via the AutoSuggest mechanism, not the main “Fragment Matches” or “Translation Results” windows, when a 100% match has already been found. 

Why you can’t see them in the Fragment Matches window

• Trados deliberately does not display fragment matches in the Fragment Matches window when there is a 100% match (from TM or MT).

• The background fragment recall still happens because AutoSuggest relies on it.

• That’s why you can “stumble upon them” when you type, but not browse them in a results pane.

This is a design choice by SDL/RWS for performance and to avoid “clutter” in the results window. There’s currently no setting to override this behavior — fragment matches are simply suppressed when an exact match exists.

I find this very strange, since a fragment match from a client provieded TM will/might be much more reliable than a MT. I found a Trados Studio Ideas about changing this behaviour, but it's from six years ago. Am I the only one who thinks this is a major problem?

Rosanna