https://heb.proz.com/forum/poll_discussion/335359-poll_have_you_noticed_improvements_in_machine_translation_in_your_language_pairs_in_the_last_two_years.html

דפים בנושא:   [1 2] >
Poll: Have you noticed improvements in machine translation in your language pair(s) in the last two years?
מפרסם התגובה: ProZ.com Staff

ProZ.com Staff
Local time: 16:37
צוות האתר
Jun 10, 2019

This forum topic is for the discussion of the poll question "Have you noticed improvements in machine translation in your language pair(s) in the last two years?".

View the poll results »



 

Teresa Borges
פורטוגל
Local time: 00:37
חבר (2007)
מאנגלית לפורטוגזית
+ ...
Other Jun 10, 2019

I don't use machine translation in my projects!

Mónica Algazi
Kay Denney
 

Alexandra Speirs  Identity Verified
Local time: 01:37
מאיטלקית לאנגלית
+ ...
Yes Jun 10, 2019

You may not use it, Teresa, but you can't deny it is out there and is definitely getting better.
It used to be good just for a laugh, but try running a few sentences on a good MT platform and you may be surprised by the results.


Edoardo Lorenzo Corda
neilmac
Jorge Payan
Alan Corbo, CT
TOMBE Franklin
Gibril Koroma
Eckhard Boehle
 

Kaspars Melkis  Identity Verified
הממלכה המאוחדת
Local time: 00:37
מאנגלית ללטבית
+ ...
Define better Jun 10, 2019

Alexandra Speirs wrote:

You may not use it, Teresa, but you can't deny it is out there and is definitely getting better.
It used to be good just for a laugh, but try running a few sentences on a good MT platform and you may be surprised by the results.


If you put the translation quality on the scale from 0 to 10 where 0 is the worst and 10 is the best, MT might have improved from 0.5 to 0.6. Whereas professional human translations usually range somewhere from 7-9. In other words, don't bother unless you are working in the market that needs the quality of 1 or 2.


writeaway
Mónica Algazi
 

Christine Andersen  Identity Verified
דנמרק
Local time: 01:37
חבר (2003)
מדנית לאנגלית
+ ...
Wavering between 'No' and 'Other' Jun 10, 2019

I would like to say no, but that could be because I have not seriously looked at much machine translation.

It is hard to say when, but there have been improvements in my language pairs. The big difference was probably more than two years ago. It is usable in some situations, but still not reliable, and comes up now and then with some odd mistakes.


vixen
Liena Vijupe
Alexandra Speirs
Elena Mordenti
Michele Fauble
Gibril Koroma
 

Teresa Borges
פורטוגל
Local time: 00:37
חבר (2007)
מאנגלית לפורטוגזית
+ ...
@Alexandra Jun 10, 2019

Alexandra Speirs wrote:

You may not use it, Teresa, but you can't deny it is out there and is definitely getting better.
It used to be good just for a laugh, but try running a few sentences on a good MT platform and you may be surprised by the results.


You might be right but as I don't use it I can't say that it has improved. The growing translation mechanization and automization worry me greatly.

N.B. I must add that I don't translate technical stuff...


Mónica Algazi
 

Elizabeth Tamblin  Identity Verified
הממלכה המאוחדת
Local time: 00:37
חבר (2012)
מצרפתית לאנגלית
N/A Jun 10, 2019

I never use it.

Teresa Borges
Mónica Algazi
 

neilmac  Identity Verified
ספרד
Local time: 01:37
מספרדית לאנגלית
+ ...
Yes Jun 10, 2019

GT4T has improved in the past two years. DeepL is quite good and can be quite useful as well. What people don't seem to realise is that these tools are only as good as the person using them. They can be a real help to a translator who knows what they're doing with them. However, there are not a panacea and you can't just feed in screeds of text and expect to get a decent result as output. For example, I had one project which is basically just a very very long list of chemicals. I tried feeding a... See more
GT4T has improved in the past two years. DeepL is quite good and can be quite useful as well. What people don't seem to realise is that these tools are only as good as the person using them. They can be a real help to a translator who knows what they're doing with them. However, there are not a panacea and you can't just feed in screeds of text and expect to get a decent result as output. For example, I had one project which is basically just a very very long list of chemicals. I tried feeding a whole page at a time into both of the two aforementioned MT programs to see what came out. Both did quite a good job, but neither of them were 100% accurate, and it took as long to go over the results as it would have to translate them manually from scratch.
Having said that, in my experience, they are useful for some types of texts and not so useful for others. And there is a difference between a freelancer exploring these tools individually to big businesses or agencies ramming them down your throat so they can pay less for "post editing"…
Collapse


Alexandra Speirs
Elena Mordenti
Ivan Nieves
Bernhard Sulzer
Florian Stauber
Elaine Ruby
Gibril Koroma
 

Kevin Fulton  Identity Verified
ארצות הברית
Local time: 19:37
מגרמנית לאנגלית
Uneven results German > English Jun 10, 2019

Over the past few years I've occasionally run sentences through Google Translate, and – more recently – Deepl.
Initially the results ranged from gibberish to inadequately intelligible. The results of the recent test performed in March of this year ranged from barely intelligible to understandable (GT), and intelligible to almost acceptable (Deeple). I can tell when GT changes its algorithm, as the quality between tests either improves or declines substantially. I haven't tested Deepl l
... See more
Over the past few years I've occasionally run sentences through Google Translate, and – more recently – Deepl.
Initially the results ranged from gibberish to inadequately intelligible. The results of the recent test performed in March of this year ranged from barely intelligible to understandable (GT), and intelligible to almost acceptable (Deeple). I can tell when GT changes its algorithm, as the quality between tests either improves or declines substantially. I haven't tested Deepl long enough to recognize qualitative differences.

Oddly enough, when running a poorly-written source text through Deepl, the result is often more understandable than the original.

Not surprisingly, single-word terminology searches often yield better results than whole phrase translations, since no syntax parsing is required.
Collapse


Eckhard Boehle
Julio Madrid
Diana Obermeyer
 

Alan Corbo, CT  Identity Verified
אורוגוואי
Local time: 20:37
מאנגלית לספרדית
+ ...
They have certainly improved Jun 10, 2019

The worrying thing is that the so-called "multibillion dollar language/translation industry", which we are told is growing at whopping two-digit rates, is not ours for the taking. Yes, there is more content to translate. Yes, companies want to jump onto the localization bandwagon. Yes, we are all now potential authors who want to project our voice to the worldwide community (via Twitter and Instagram feeds which need to be translated). That should be good news for us translators. But it's not, i... See more
The worrying thing is that the so-called "multibillion dollar language/translation industry", which we are told is growing at whopping two-digit rates, is not ours for the taking. Yes, there is more content to translate. Yes, companies want to jump onto the localization bandwagon. Yes, we are all now potential authors who want to project our voice to the worldwide community (via Twitter and Instagram feeds which need to be translated). That should be good news for us translators. But it's not, in my opinion, for two main reasons:

1. There aren't enough translators to translate the massive amounts of content that is produced every day
2. Most of the content produced needs not to be translated to professional levels. A "good enough" translation suffices. And for those, MT works wonderfully. And let's face it: not all translators translate doctoral theses or philosophical treaties. So when they say, "The language industry growth is exponential", beware: we won't share in most of that growth, if at all.

MT is the technological answer to the scarcity of human translators, and to the general fall in people's expectations of what good language use is. If "good enough" is becoming the new standard for good communication (and, IMHO, that's what happening in several fields of human interaction), then MT is here to stay, and will keep improving (or maybe people in general will keep settling for the "good enough" creed).
Collapse


Jorge Payan
neilmac
Elaine Ruby
R-i-c-h-a-r-d
Julio Madrid
Diana Obermeyer
 

Gloria Teixeira  Identity Verified
ברזיל
Local time: 20:37
חבר (Mar 2020)
מאנגלית לפורטוגזית
+ ...
other ways Jun 10, 2019

It's not all jobs that I use machine translation

neilmac
 

Justin Peterson  Identity Verified
ספרד
Local time: 01:37
חבר (2007)
מספרדית לאנגלית
OMG, yes Jun 10, 2019

Oh yes, they have improved... dramatically.

OF COURSE they are not substitutes for a human translator, but, at this point I think they are INDISPENSABLE tools for most projects.

I think they have increased my speed by 2-3 times (!) while increasing, not sacrificing, quality. I can spend more time now revising, fine-tuning and thinking rather than just rattling away on the keyboard. Then there are the overlaps with previously translated material. And then sections tha
... See more
Oh yes, they have improved... dramatically.

OF COURSE they are not substitutes for a human translator, but, at this point I think they are INDISPENSABLE tools for most projects.

I think they have increased my speed by 2-3 times (!) while increasing, not sacrificing, quality. I can spend more time now revising, fine-tuning and thinking rather than just rattling away on the keyboard. Then there are the overlaps with previously translated material. And then sections that require no thinking whatsoever (lists of countries, for example).

At this point I cannot even IMAGINE working without MTs. It would be like going to back to an axe after having a chainsaw to work with.

I do not understand how anyone today could forego using MT. I am not criticizing it, I just sincerely do not understand how they can afford to do that.

JRP

Then again, the TYPE of content one is translating is key. They are virtually worthless for a novel, for example (which I have translated too).


[Edited at 2019-06-10 16:32 GMT]
Collapse


Ivan Nieves
neilmac
Alexandra Speirs
Philip Lees
Eckhard Boehle
 

DZiW
אוקראינה
מאנגלית לרוסית
+ ...
something changes Jun 10, 2019

While I agree that AI and datamining algos in some language pairs may improve the MT quality, I doubt so many translators who can't do without a CAT or MT should be called "translators". CAT-operators and PEMTists, perhaps?

I know several such 'translators' who couldn't process a simple .DOC without TM or internet connection, let alone some of them were about to change their 5+year career because... of missing/damaged TM or moving to another CAT tool.


Julio Madrid
Diana Obermeyer
 

Jorge Payan  Identity Verified
קולומביה
Local time: 18:37
חבר (2002)
מגרמנית לספרדית
+ ...
Confusion? Jun 10, 2019

I think you are confusing MT (Machine Translation) engines, as Google Translate, DeepL, Bing Translator, etc., with TM (Translation Memory) software, such as Trados Studio, memoQ, DejaVu, etc.

Saludos

Justin Peterson wrote:

Oh yes, they have improved... dramatically.

OF COURSE they are not substitutes for a human translator, but, at this point I think they are INDISPENSABLE tools for most projects.

I think they have increased my speed by 2-3 times (!) while increasing, not sacrificing, quality. I can spend more time now revising, fine-tuning and thinking rather than just rattling away on the keyboard. Then there are the overlaps with previously translated material. And then sections that require no thinking whatsoever (lists of countries, for example).

At this point I cannot even IMAGINE working without MTs. It would be like going to back to an axe after having a chainsaw to work with.

I do not understand how anyone today could forego using MT. I am not criticizing it, I just sincerely do not understand how they can afford to do that.

JRP

Then again, the TYPE of content one is translating is key. They are virtually worthless for a novel, for example (which I have translated too).


[Edited at 2019-06-10 16:32 GMT]


 

Justin Peterson  Identity Verified
ספרד
Local time: 01:37
חבר (2007)
מספרדית לאנגלית
Nope ... MT. MateCat's, for example, is outstanding. Jun 10, 2019

Jorge Payan wrote:

I think you are confusing MT (Machine Translation) engines, as Google Translate, DeepL, Bing Translator, etc., with TM (Translation Memory) software, such as Trados Studio, memoQ, DejaVu, etc.

Saludos

Justin Peterson wrote:

Oh yes, they have improved... dramatically.

OF COURSE they are not substitutes for a human translator, but, at this point I think they are INDISPENSABLE tools for most projects.

I think they have increased my speed by 2-3 times (!) while increasing, not sacrificing, quality. I can spend more time now revising, fine-tuning and thinking rather than just rattling away on the keyboard. Then there are the overlaps with previously translated material. And then sections that require no thinking whatsoever (lists of countries, for example).

At this point I cannot even IMAGINE working without MTs. It would be like going to back to an axe after having a chainsaw to work with.

I do not understand how anyone today could forego using MT. I am not criticizing it, I just sincerely do not understand how they can afford to do that.

JRP

Then again, the TYPE of content one is translating is key. They are virtually worthless for a novel, for example (which I have translated too).


[Edited at 2019-06-10 16:32 GMT]


Jorge Payan
finnword1
 
דפים בנושא:   [1 2] >


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

מנהלי פורום זה
Jared Tabor[Call to this topic]

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

Poll: Have you noticed improvements in machine translation in your language pair(s) in the last two years?

Advanced search






SDL Trados Business Manager Lite
Create customer quotes and invoices from within SDL Trados Studio

SDL Trados Business Manager Lite helps to simplify and speed up some of the daily tasks, such as invoicing and reporting, associated with running your freelance translation business.

More info »
PerfectIt consistency checker
Faster Checking, Greater Accuracy

PerfectIt helps deliver error-free documents. It improves consistency, ensures quality and helps to enforce style guides. It’s a powerful tool for pro users, and comes with the assurance of a 30-day money back guarantee.

More info »