Translation Quality Assurance Tools - are they the next thing after MTPE? Thread poster: Tom in London
|
|---|
Tom in London
United Kingdom
Local time: 07:09
Member (2008)
Italian to English
I'd be interested to know if anyone here uses so-called Translation Quality Assurance Tools such as ApSIC Xbench, Verifika, ErrorSpy, etc.
On their websites they market themselves as heavy-duty, globally important, massively powerful, lightning fast AI tools that will take dozens of translations done by different translators who have been required to work to a house style or a required lexicon, find errors or inconsistencies in their work, highlight them, and correct them.
<... See more I'd be interested to know if anyone here uses so-called Translation Quality Assurance Tools such as ApSIC Xbench, Verifika, ErrorSpy, etc.
On their websites they market themselves as heavy-duty, globally important, massively powerful, lightning fast AI tools that will take dozens of translations done by different translators who have been required to work to a house style or a required lexicon, find errors or inconsistencies in their work, highlight them, and correct them.
Basically: they do what any translator can do by running a spellcheck or a find/replace to make sure all the language and terminology are consistent (on top of the other thing all translators always do: actually reading through a translation and checking it). Which is also what any competent PM would do every day of the week.
These TQATs (when I pronounce the acronym, I must say I find the sound very appropriate to an English-speaking ear) seem to have set themselves the goal of completely eliminating human editing; instead of someone actually reading through a load of translations and checking/correcting them, they are offering technology that will do this (for a price).
But they confess that even when you've finished running the software you *still* need a human being to do the final check.
To me these TQATs look like yet another emanation from a deranged world populated by people who don't like having to think, don't trust thinking, and just want IT to do everything for them.
Your thoughts?
[Edited at 2022-11-25 17:10 GMT] ▲ Collapse
| | | | John Fossey 
Canada
Local time: 02:09
Member (2008)
French to English
+ ...
They still have no idea of context.
| | | | | I use Verifika when working in Trados | Nov 25, 2022 |
Because Trados can't do a proper QA. When working in memoQ, I rely on their built-in QA tool for all things I need to verify. Verifika can check if you're following TB, but I only used this feature a couple of times after one of our tech guys spent a week to create a TB for Verifika where each target term was available in every form imaginable. Just a waste of time, hundreds more entries to check just to find out there are no actual mistakes. This tool is pretty stupid, but when customized, it c... See more Because Trados can't do a proper QA. When working in memoQ, I rely on their built-in QA tool for all things I need to verify. Verifika can check if you're following TB, but I only used this feature a couple of times after one of our tech guys spent a week to create a TB for Verifika where each target term was available in every form imaginable. Just a waste of time, hundreds more entries to check just to find out there are no actual mistakes. This tool is pretty stupid, but when customized, it can be used for numbers, spelling, translation consistency and the like ▲ Collapse
| | | | | Three keys... | Nov 25, 2022 |
...or key combinations in my CAT (Deja Vu) are sufficient enough to check my translation automatically on spelling errors (F7), numeral errors (Ctrl + Shift + F7), and for tags (Ctrl + Shift + F8). When the German target text is an MS Word file, I also can perform an automatic check for grammar errors too. The rest is pure brain work, which means all the features Tom has mentioned and for which my brain is being paid for. Why should I then pay for other external programs, that only cause trouble... See more ...or key combinations in my CAT (Deja Vu) are sufficient enough to check my translation automatically on spelling errors (F7), numeral errors (Ctrl + Shift + F7), and for tags (Ctrl + Shift + F8). When the German target text is an MS Word file, I also can perform an automatic check for grammar errors too. The rest is pure brain work, which means all the features Tom has mentioned and for which my brain is being paid for. Why should I then pay for other external programs, that only cause trouble where no trouble was and which cause only extra work when I receive a list with dozens of pseudo errors by the PM those programs generate, errors that aren't justified at all? But I lost at least two or three customers because I refused to apply (and buy!) Xbench and/or Verifica. I'm sure I would have lost them anyway today because of DeepL (or, if you don't believe in the negative effects of DeepL, because of the German economical crisis to come in the near future, because business climate indexes tell us so). Therefore, these tools are not the next AI-based step after AI-based translation. They just could aid a non-native speaker to control the native speaker to not cheat the non-native speaker by sloppy work. Oh yes, and I forgot to mention that 7 years ago a PM tried to tell me that their QA tool is intelligent and that I better should use it because it is very clever. Wow, that really gave me the chills: A QA tool that is based on life experience! ▲ Collapse
| | |
|
|
|
| It's a crutch for the lame | Nov 25, 2022 |
On a few occasions, a client sent me these... TQAT reports and asked me to review my translations accordingly. 98-99% of the flagged "errors" were false positives, so I think the client eventually realised it was not worth the trouble.
I can imagine such a system being used by a freelancer who is aware of being somewhat scatterbrained.
In a corporate situation, it may be useful for terminology control in a large team of translators or technical writers working on the same topic, espe... See more On a few occasions, a client sent me these... TQAT reports and asked me to review my translations accordingly. 98-99% of the flagged "errors" were false positives, so I think the client eventually realised it was not worth the trouble.
I can imagine such a system being used by a freelancer who is aware of being somewhat scatterbrained.
In a corporate situation, it may be useful for terminology control in a large team of translators or technical writers working on the same topic, especially when new members are added every day.
However, personally I find these systems to be of more harm than help.
When working in memoQ, I rely on their built-in QA tool for all things I need to verify.
In my experience, MemoQ QA tool also yields mostly false positives, but your mileage may vary.
[Edited at 2022-11-25 22:40 GMT] ▲ Collapse
| | | | Kaspars Melkis
United Kingdom
Local time: 07:09
English to Latvian
+ ...
| these tools are awful | Nov 26, 2022 |
Anton Konashenok wrote:
On a few occasions, a client sent me these... TQAT reports and asked me to review my translations accordingly. 98-99% of the flagged "errors" were false positives, so I think the client eventually realised it was not worth the trouble.
I can imagine such a system being used by a freelancer who is aware of being somewhat scatterbrained.
In a corporate situation, it may be useful for terminology control in a large team of translators or technical writers working on the same topic, especially when new members are added every day.
However, personally I find these systems to be of more harm than help.
These tools are very primitive, they don't even take into account case endings, nor special cases with numbers, punctuation etc.
I personally think that all CAT tools are awful, very slow, poor workflow and UI. I believe that there is a lot of unused potential in this area. From one side we are talking about latest generation neural net machine translations that almost achieve the same level as human translation. And on the other side we are here with tools that cannot even ensure good and proper terminology management. Does anybody else see the disconnect here?
| | | | Stepan Konev
Russian Federation
Local time: 09:09
English to Russian
| Any tool is awful and primitive when you don't know how to use it | Nov 26, 2022 |
Kaspars Melkis wrote:
These tools are very primitive, they don't even take into account case endings, nor special cases with numbers, punctuation etc. It's up to you to set up your tool to recognize case endings, numbers and other things like that. Wildcards and regular expressions could help you if you were trained in them. Probably you simply can't use the tool? Hence your misbelief about CAT tools. I can't use a snowboard. Am I right in my belief that snowboard is always awful and very slow for moving? 
[Edited at 2022-11-26 02:10 GMT]
| | | | Kaspars Melkis
United Kingdom
Local time: 07:09
English to Latvian
+ ...
| no, it's not up to me | Nov 26, 2022 |
Stepan Konev wrote: Kaspars Melkis wrote:
These tools are very primitive, they don't even take into account case endings, nor special cases with numbers, punctuation etc. It's up to you to set up your tool to recognize case endings, numbers and other things like that. Wildcards and regular expressions could help you if you were trained in them. Probably you simply can't use the tool? Hence your misbelief about CAT tools. I can't use a snowboard. Am I right in my belief that snowboard is always awful and very slow for moving? [Edited at 2022-11-26 02:10 GMT]
If I had to program a spellchecker to recognise Latvian case-endings (not a trivial task) I would be justified to say that this tool is primitive and useless in my work.
If I had created the list of rules to work with those “quality checkers” then it would be my own tool I had created, not the one I had bought for immediate use. They don't even make it easy to create and maintain your own rules!
| | |
|
|
|
Recep Kurt
Türkiye
Local time: 09:09
Member (2011)
English to Turkish
+ ...
| Not for everybody | Nov 26, 2022 |
If you don't work with large/repetitive texts with a specific glossary and/or fuzzy matches and/or tags etc. you might not need them. The proofreading tools of the software you use to check your texts would be just fine.
I like using Verifika when delivering game translations (up to 150K words) as I find it to be more user-friendly compared to other similar tools. It highlights the issues that need to be checked in the original bilingual file and allows me to skip or change whatever needs ... See more If you don't work with large/repetitive texts with a specific glossary and/or fuzzy matches and/or tags etc. you might not need them. The proofreading tools of the software you use to check your texts would be just fine.
I like using Verifika when delivering game translations (up to 150K words) as I find it to be more user-friendly compared to other similar tools. It highlights the issues that need to be checked in the original bilingual file and allows me to skip or change whatever needs changing.
I agree with Stepan - you need to configure your QA tool to do what you need it to do. This would reduce the number of false positives and save you time. ▲ Collapse
| | | | Tom in London
United Kingdom
Local time: 07:09
Member (2008)
Italian to English
TOPIC STARTER
Recep Kurt wrote:
- you need to configure your QA tool to do what you need it to do. This would reduce the number of false positives and save you time.
I just use my brain for that. My brain is remarkably powerful, and fast.
| | | | Stepan Konev
Russian Federation
Local time: 09:09
English to Russian
| Error-free brain | Nov 26, 2022 |
If everybody could have a remarkably powerful brain, it would be an ideal world. However not everybody is lucky enough to have such a brain. Some (including myself) still have to use tools.
| | | | Kay Denney
France
Local time: 08:09
French to English
I do run the QA check in MemoQ and while yes it does throw up a lot of false positives, mostly flagging a missing number 2 because I've translated it as "twice", it does also point out little slip-ups with spacing etc. mostly caused by ... the bally tags that they themselves introduced.
And sometimes it points out that there is a space missing before the tag, and an extra space after the tag, and it doesn't understand that since the tag is destined to die during the export it doesn... See more I do run the QA check in MemoQ and while yes it does throw up a lot of false positives, mostly flagging a missing number 2 because I've translated it as "twice", it does also point out little slip-ups with spacing etc. mostly caused by ... the bally tags that they themselves introduced.
And sometimes it points out that there is a space missing before the tag, and an extra space after the tag, and it doesn't understand that since the tag is destined to die during the export it doesn't actually matter!
[Edited at 2022-11-26 12:08 GMT] ▲ Collapse
| | |
|
|
|
Samuel Murray
Netherlands
Local time: 08:09
Member (2006)
English to Afrikaans
+ ...
Tom in London wrote:
These TQATs seem to have set themselves the goal of completely eliminating human editing; instead of someone actually reading through a load of translations and checking/correcting them, they are offering technology that will do this...
These types of tools have existed almost for as long as CAT tools have existed. Many CAT tools have built-in functionality that is similar to these tools. They are meant to help a human translator check for things that are easier for a machine to find than for a human. But ultimately they are a tools that are meant to help a human do their job. QA tools are most useful if they catch most errors, and that means that they catch also things that are not errors, and it needs a human to tell the difference. I'm sure these tools are getting cleverer and cleverer each year, but any client who relies on them alone to "ensure quality" doesn't quite understand what they're for.
They are called "quality assurance" tools, but they don't assure quality. They help the human assure quality. You'll get the most out of these tools if you get to know them.
[Edited at 2022-11-26 19:43 GMT]
| | | |
Samuel Murray wrote:
These types of tools have existed almost for as long as CAT tools have existed. Many CAT tools have built-in functionality that is similar to these tools.
I compared Xbench with the QA checks that my tool offers:
https://www.proz.com/forum/cafetran_support/359362-do_we_need_xbench.html
I find terminological and consistency checks very useful if I have to translate a few new segments in projects that were translated by several other translators in the past, with different preferences and craftsmanship.
| | | | There is no moderator assigned specifically to this forum. To report site rules violations or get help, please contact site staff » Translation Quality Assurance Tools - are they the next thing after MTPE? | Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
| | Anycount & Translation Office 3000 | Translation Office 3000
Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
Translation news
