How can the proofreader see the fuzzy matches?
Thread poster: Samuel Murray

Samuel Murray  Identity Verified
Netherlands
Local time: 02:58
Member (2006)
English to Afrikaans
+ ...
Oct 24, 2014

Hello

If I translate a file in Trados 2009, using a client's TM, and the proofreader then edits that file in Trados (presumably 2009 or 2014), is there any way that the proofreader can see the TM matches that I saw when I translated the file, so that he can see how much of the text (and which text) I had to change?

If not, then the proofreader will be forced to carefully reread every single word, even if the segment was a high fuzzy match, because he would not know which words in the matched text I had changed.

Samuel


Direct link Reply with quote
 

Triston Goodwin  Identity Verified
United States
Local time: 18:58
Spanish to English
+ ...
Only one experience Oct 24, 2014

I have only had one project like what you have described and I was able to see the fuzzy and 100% matches. The client just sent me the same project files that the translator had sent to them.

Direct link Reply with quote
 

SDL Community  Identity Verified
United Kingdom
Local time: 02:58
English
Of course... Oct 25, 2014

Samuel Murray wrote:

If I translate a file in Trados 2009, using a client's TM, and the proofreader then edits that file in Trados (presumably 2009 or 2014), is there any way that the proofreader can see the TM matches that I saw when I translated the file, so that he can see how much of the text (and which text) I had to change?



... the match values are held in the sdlxliff so they will see exactly what you do. The same applies to external review files (in Word) where the match value is written in the second column, and you can also colour the rows by value to make it more obvious.

Regards


Paul


Direct link Reply with quote
 

Anton Konashenok  Identity Verified
Czech Republic
Local time: 02:58
English to Russian
+ ...
I'd rather NOT have the proofreader see the fuzzy matches Oct 25, 2014

In my humble opinion, relying on fuzzy matches in the course of proofreading defies the very purpose of proofreading - that is, assuring the quality of the resulting translation: firstly, no one guarantees the quality of existing TM, and it's the duty of the proofreader to check it; secondly, there are situations when changing just one word in a source segment would require a complete overhaul of the target one. In the bottom line, 'carefully reread every single word' is exactly what a proofreader ideally has to do.

Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 02:58
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Paul Oct 25, 2014

SDL Support wrote:
The match values are held in the sdlxliff so they will see exactly what you do.


When you say "match value", do you mean the match percentage or the actual value (content) of the match?

The same applies to external review files (in Word) where the match value is written in the second column...


Aah, so you do mean "match percentage" when you say "match value". Well, match percentage is not useful by itself. A segment of 50 words that came from a 90% match will have had very few changes applied to it, but the number "90" does not tell the proofreader which changes. So the proofreader is forced to re-proofread all 50 words. But if the 10% difference was made up of just two or three specific words, it would be great if the proofreader could see which three words they were.


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 02:58
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Anton Oct 25, 2014

Anton Konashenok wrote:
In my humble opinion, relying on fuzzy matches in the course of proofreading defies the very purpose of proofreading - that is, assuring the quality of the resulting translation...


I hear you, and I agree with you, in cases where the client's TM is rubbish and the proofreader is required to fix those problems. But in my case the content of the TM originated with me and my proofreader, and we're very happy with the quality of it. This means that if I (the translator) get a 90% match on a 50-word segment, I can safely fix just those two or three words, but the poor proofreader is forced to reread the entire 50 words, very, very carefully.

Firstly, no one guarantees the quality of existing TM, and it's the duty of the proofreader to check it


It is not always the proofreader's task to do that. In fact, the proofreader is sometimes asked not to do that, if the client is happy with his TM (went through his QA processes successfully) and there is a risk that the proofreader may make too many unnecessary, stylistic changes that might even undo all the hard work that the client's previous translators and proofreaders did. Sometimes the proofreader's given task is to spot translation errors (i.e. to see if the translator had not made a mistake when he fixed the fuzzy matches) and nothing else.

Secondly, there are situations when changing just one word in a source segment would require a complete overhaul of the target one.


That is true, but that would only make a difference if the bulk of the segments are of that type (and in my language combination, it seldomly is).


Direct link Reply with quote
 

Eileen Cartoon  Identity Verified
Local time: 02:58
Italian to English
In my Studio Wish List Oct 25, 2014

This would also be possible in review. I would sometimes like to be able to double check what I did with the fuzzies while I am reviewing what I translated but in Review mode the original fuzzy no longer shows so I can't double check.
I suppose this is the same problem you are asking about for the proofreader.
Eileen


Direct link Reply with quote
 

SDL Community  Identity Verified
United Kingdom
Local time: 02:58
English
Seeing the changes... Oct 25, 2014

Hi,

Perhaps the answer is to work with tracked changes then and only accept them when you review? If you do your initial project preparation and then turn on track changes prior to starting your initial translation then when you do your review you'll be able to see exactly what you changed. You can even work in "Final Mode" as you edit so the changes don't get in the way of editing, and just toggle "Final Mode" on and off when you want to see what changes were actually made.

So for example, in this image below I first have a 92% match from my TM as a result of the Project preparation. I switch on tracked changes and I work in final mode. So in effect it's just like normal except that I'm recording every change in the sdlxliff.



Then if I need to see the detail behind the change when reviewing I just toggle the "Final Mode" on or off. This is a ribbon icon or a shortcut (Ctrl+Alt+Shift+F9... that's the default but it could be whatever you want it to be!)

Maybe this is what you'd like? The reviewer can do this too as the changes are held in the sdlxliff if you work in this way and remember to turn the tracked changes on when you start.

Another thing you could do is use Post-Edit Compare. With this app you can create snapshots of your projects whenever you like and it will keep a record of them in a new view in Studio. You can then compare any two views you like to see what changes had been made between them. You can find an article showing the Post-Edit Compare interface here.

Regards

Paul


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 02:58
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Paul, thanks for thinking of options Oct 25, 2014

SDL Support wrote:
Perhaps the answer is to work with tracked changes then and only accept them when you review?


This will show what changes the translator had made, but not what changes there were between the proposed match and the source text. Here's why it matters: what if the translator changed the wrong thing? That is precisely what the proofreader should catch. The translator might also decide to change something else in addition to the "difference", or decide that the change is so minor that it requires no editing.

If you do your initial project preparation and then turn on track changes prior to starting your initial translation then when you do your review you'll be able to see exactly what you changed.


Unfortunately (unfortunately in this case) Trados' track changes feature works in real time, and not by comparison after the segment is committed (which is how Wordfast Pro's track changes feature works). If I delete a word and then decide to put that word back in again by typing it in, then both my deletion and my insertion will be visible as edits (when in fact there is no edit). I'm afraid the proofreader is going to see track soup, because I don't sit back and carefully consider the whole segment for a while, before I start fixing things in it.

But thanks for thinking of the idea.

The reviewer can do this too as the changes are held in the sdlxliff if you work in this way and remember to turn the tracked changes on when you start.


Yes, I suppose if there is a keyboard shortcut for "accept all changes in current segment" (is there?), the proofreader can hit that shortcut before making his own edits, to ensure that only his edits are visible in the file that he delivers.

This would be necessary because Trados doesn't have the capacity to keep the edits of more than one person separate (not if the one person edits something inside someone else's edit) (I mean, if the proofreader makes a small change to an insertion that the translator had made, then that entire insertion is marked as the proofreader's edit, and not just the small change (is this a bug or a feature?)).

Another thing you could do is use Post-Edit Compare.


This looks interesting, thanks.

I suppose I could also use a program like Change Tracker to compare a pre-translated version of the SDLXLIFF files with the translated version, and export it to Excel and copy it to MS Word. Change Tracker sometimes fails to see all segments, but it shows very nicely where segments have changed. Again, this will only show the edits that the translator had made in the target text, and will not show the actual differences between the TM match's source text and the document's source text.

By the way, one of my l10n clients sends me updated text in an Excel file that has several columns, including a column for the current source text, the previous source text, the previous source text's translation, and a column that shows the source text differences in "tracked change" format. I realise that "updated source text" is not the same as fuzzy matching, but it comes close.

==
Thinking out loud...

Is there a keyboard shortcut for inserting not the current source text into the target field, but the source text of the TM match (that would be silly, but I'm asking anyway)?

I suppose if I knew about this beforehand (i.e. if it was a big project), I could generate a dummy TM from the client's TM that has the source text of all segments copied to the target text, and then do the pre-translation using it, on a version 1 of the files. Then I could create a version 2 of the files in which I copy the current source text into the target fields, for all segments, and then I could use e.g. Change Tracker to compare version 1 and version 2. This will yield a report that shows the source text changes of the fuzzy matches.

(One would have to use Change Tracker to create the visible "tracked changes", because Trados doesn't compare segments after the segment is committed (like Wordfast Pro does -- I'm not critisising Trados, just comparing methods). I mean, if a segment's source text says "All of your responses will be anonymous" and the source text of the fuzzy match says "All of our responses will be anonymous", and you put that source text into the target field, and then press Alt+Insert, it's not only the word "your/our" that will be marked as a change, but the entire segment will be marked as a change, even though the only change is the word "your/our".)





[Edited at 2014-10-25 22:26 GMT]


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 02:58
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Eileen Oct 25, 2014

Eileen Cartoon wrote:
I would sometimes like to be able to double check what I did with the fuzzies while I am reviewing what I translated but in Review mode the original fuzzy no longer shows so I can't double check.


Yes, exactly. In the case that prompted this question, I was thinking of the poor proofreader, but I do experience what you're talking about. When checking my own translation, it would be fantastic to see what the original fuzzy match was and how it differed from the source text.


Direct link Reply with quote
 

SDL Community  Identity Verified
United Kingdom
Local time: 02:58
English
Maybe you can use... Oct 25, 2014

Samuel Murray wrote:

SDL Support wrote:
Perhaps the answer is to work with tracked changes then and only accept them when you review?


This will show what changes the translator had made, but not what changes there were between the proposed match and the source text. Here's why it matters: what if the translator changed the wrong thing? That is precisely what the proofreader should catch. The translator might also decide to change something else in addition to the "difference", or decide that the change is so minor that it requires no editing.



... the option to "Search for fuzzy matches even if exact match found" and then you'd see it in the TM results window as you work? But I can see this wouldn't be helpful for the reviewer unless you gave them access to your TM too.

I must admit though... a part of me wonders if the proofreader would really pay attention to all of this detail. Are they there to QA how you work with your TM and what changes you made to a fuzzy that may or may not be useful? Or are they there to make sure that your translation makes sense and is well written?

Interesting discussion... would be good to see the opinions of others here too on whether a feature like this would be useful or not. I'd never considered it in this way before.

Regards

Paul


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 02:58
Member (2006)
English to Afrikaans
+ ...
TOPIC STARTER
@Paul Oct 26, 2014

SDL Support wrote:
Maybe [the proofreader] can use the option to "Search for fuzzy matches even if exact match found" and then [the proofreader would] see it in the TM results window...? But I can see this wouldn't be helpful for the reviewer unless you gave them access to your TM too.


In this particular case, the proofreader has access to the TM, so that would be a useful thing, yes.

I must admit though... a part of me wonders if the proofreader would really pay attention to all of this detail.


If I was a proofreader, and this feature would be available to me without too much hassle (i.e. if it didn't involve using separate programs to see the changes), then I would certainly use it (and it would save clients money, if the proofread was paid per hour instead of per word).

Are [proofreaders] there to QA how you work with your TM and what changes you made to a fuzzy that may or may not be useful? Or are they there to make sure that your translation makes sense and is well written?


Both, I think. A proofreader who does not compare the translation with the source text (but only reads the target text to see if it sounds nice) doesn't quite understand what bilingual proofreading is all about.

Yes, the translator is supposed to make sure that his translation is correct, but the proofreader is a second line of defence (particularly in cases where the translator is more likely to make small, hard-to-see mistakes, e.g. in rush jobs with rolling deliveries).

It also depends, I think, on how many proofreaders there are. My one client always uses two proofreaders... one to check that the translation is good and another to check that the target text sounds nice.


[Edited at 2014-10-26 12:18 GMT]


Direct link Reply with quote
 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

How can the proofreader see the fuzzy matches?

Advanced search







memoQ translator pro
Kilgray's memoQ is the world's fastest developing integrated localization & translation environment rendering you more productive and efficient.

With our advanced file filters, unlimited language and advanced file support, memoQ translator pro has been designed for translators and reviewers who work on their own, with other translators or in team-based translation projects.

More info »
SDL Trados Studio 2017 Freelance
The leading translation software used by over 250,000 translators.

SDL Trados Studio 2017 helps translators increase translation productivity whilst ensuring quality. Combining translation memory, terminology management and machine translation in one simple and easy-to-use environment.

More info »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search