| Pages in topic: < [1 2 3 4] > | Poll: How often do you use DeepL or other machine translation tools? Thread poster: ProZ.com Staff
|
|---|
Tom in London
United Kingdom
Local time: 12:36
Member (2008)
Italian to English
Ice Scream wrote:
I'm genuinely shocked. It's not that long since you said you'd use CAT over your dead body!!
I, however, refuse to get with the kidz.
Maybe you're like Stewie then.
https://www.youtube.com/watch?v=LHpdgHTINik
| | | |
.... if this is the result (see below), you can better avoid MT like the plague.
"When the product with high mode, accumulated a lot of heat, in order to avoid the shell temperature affect the use of comfort, this product joined the intelligent temperature control, when the built-in chip detected at high temperature, will gradually reduce the work power to avoid overheating flashlight rapidly, after being temperature decrease, automatically gradually restore the full power of curren... See more .... if this is the result (see below), you can better avoid MT like the plague.
"When the product with high mode, accumulated a lot of heat, in order to avoid the shell temperature affect the use of comfort, this product joined the intelligent temperature control, when the built-in chip detected at high temperature, will gradually reduce the work power to avoid overheating flashlight rapidly, after being temperature decrease, automatically gradually restore the full power of current gear work."
PS) This piece of text was sent to me by the end client and has MT (from German actually) written all over it. Happy translating!
[Edited at 2022-05-10 15:51 GMT]
[Edited at 2022-05-11 07:57 GMT] ▲ Collapse
| | | | Lieven Malaise
Belgium
Local time: 13:36
Member (2020)
French to Dutch
+ ...
| Probably not Deeple. | May 10, 2022 |
Robert Rietvelt wrote:
"When the product with high mode, accumulated a lot of heat, in order to avoid the shell temperature affect the use of comfort, this product joined the intelligent temperature control, when the built-in chip detected at high temperature, will gradually reduce the work power to avoid overheating flashlight rapidly, after being temperature decrease, automatically gradually restore the full power of current gear work."
PS) This piece of text was sent to me by the end client and has MT (from German actually) all written over it. Happy translating!
[Edited at 2022-05-10 15:51 GMT]
Probably not translated with Deeple. Other option: the source text was crap. Let's not forget that a lot of source texts are of bad quality. MT is not miraculously going to translate that (more or less) correctly.
Anyway, I have used quite some MT the last year and I can tell you I have never ever had to deal with such poorly translated MT content (not even close). But perhaps my language pairs are better suited, who knows.
| | | |
Lieven Malaise wrote: Robert Rietvelt wrote:
"When the product with high mode, accumulated a lot of heat, in order to avoid the shell temperature affect the use of comfort, this product joined the intelligent temperature control, when the built-in chip detected at high temperature, will gradually reduce the work power to avoid overheating flashlight rapidly, after being temperature decrease, automatically gradually restore the full power of current gear work."
PS) This piece of text was sent to me by the end client and has MT (from German actually) all written over it. Happy translating!
[Edited at 2022-05-10 15:51 GMT] Probably not translated with Deeple. Other option: the source text was crap. Let's not forget that a lot of source texts are of bad quality. MT is not miraculously going to translate that (more or less) correctly. Anyway, I have used quite some MT the last year and I can tell you I have never ever had to deal with such poorly translated MT content (not even close). But perhaps my language pairs are better suited, who knows.
***
Good point. MT varies DRAMATICALLY in quality, with Deepl being the best there is right now, in my experience.
GoogleTranslate, once considered cutting edge, pales in comparison. Not even close.
| | |
|
|
|
| Poor source texts | May 10, 2022 |
Lieven Malaise wrote:
Let's not forget that a lot of source texts are of bad quality. MT is not miraculously going to translate that (more or less) correctly.
A human translator will.
| | | | | My jaw didn't drop | May 10, 2022 |
Justin Peterson wrote:
If you put, for example, a newspaper article into it ... I guarantee you will not be laughing.
In fact, your jaw might drop at how good it is.
Randomly selected Swedish newspaper headline today:
Så mycket kostar ett svenskt Natomedlemskap
Google:
That is how much a Swedish NATO membership costs (wrong tense and weird English)
DeepL:
How much does Swedish membership of NATO cost? (wrong tense and it's not a question)
Me:
What Swedish membership of NATO would cost
So how much time has MT saved me there?
| | | | | Never, And Never Will... | May 10, 2022 |
...since I am, almost exclusively, a literary translator. I have some inkling that certain MT developers are now experimenting to see how MT could eventually be used to produce literary translations. I hope I won't be around at the time that might occur.
| | | | Lieven Malaise
Belgium
Local time: 13:36
Member (2020)
French to Dutch
+ ...
Ice Scream wrote:
A human translator will.
Of course, but the point is that crappy source texts aren't suitable for MT and shouldn't be used for it (nor to 'prove' MT is no good).
| | |
|
|
|
Lieven Malaise
Belgium
Local time: 13:36
Member (2020)
French to Dutch
+ ...
Ice Scream wrote: Justin Peterson wrote:
If you put, for example, a newspaper article into it ... I guarantee you will not be laughing.
In fact, your jaw might drop at how good it is.
Randomly selected Swedish newspaper headline today: Så mycket kostar ett svenskt Natomedlemskap Google: That is how much a Swedish NATO membership costs (wrong tense and weird English) DeepL: How much does Swedish membership of NATO cost? (wrong tense and it's not a question) Me: What Swedish membership of NATO would cost So how much time has MT saved me there?
MT is no good for headlines (no context, concise language). At least I saved you some time by avoiding that you will use it again for that purpose in the future.
It's funny how you are trying to prove that a food processor is no good because you can't paint your house with it.
| | | | expressisverbis
Portugal
Local time: 12:36
Member (2015)
English to Portuguese
+ ...
| We are free to choose and the client has the final say | May 10, 2022 |
Translators are free to choose the translation tools according to their needs and their own preferences, regardless of whether it is pen and paper, MT, termbases, glossaries, online and printed dictionaries, AI, translation memory, etc.
A teacher isn't expected to teach their students by using the same methods; a lawyer doesn't resolve your legal matters in the same way as others; a civil construction worker doesn't knock down an internal wall by complying with the same code of practice i... See more Translators are free to choose the translation tools according to their needs and their own preferences, regardless of whether it is pen and paper, MT, termbases, glossaries, online and printed dictionaries, AI, translation memory, etc.
A teacher isn't expected to teach their students by using the same methods; a lawyer doesn't resolve your legal matters in the same way as others; a civil construction worker doesn't knock down an internal wall by complying with the same code of practice in construction work. And so forth...
The client has the final say: if he/she is happy with the quality of the job, he/she retains the translator, otherwise will look for another.
Every tool that is part of our work is useful and I respect both colleagues who use more rudimentary practices and colleagues who are avid users of modern technologies, like MT.
Regarding the poll, I don't use DeepL in my projects, but I have one or two agencies that allow me to work with Language Weaver.
I can agree that for some languages and translation areas, MT may not work (so) well, and the same goes when we use it incorrectly.
[Edited at 2022-05-10 22:46 GMT]
[Edited at 2022-05-10 22:48 GMT] ▲ Collapse
| | | | | So what is a food processor good for? | May 10, 2022 |
Lieven Malaise wrote:
MT is no good for headlines (no context, concise language).
The list of things MT is not suitable for just keeps growing… It seems all that is left is incredibly basic texts that a human could type/dictate straight out without any need for post-editing…
It's funny how you are trying to prove that a food processor is no good because you can't paint your house with it.
😂
I did actually put the whole article through, with similar results, but thought I’d keep it brief here.
MT is very good at getting it almost right, but that is no use to me. I know my first draft won’t contain weird errors or misunderstandings or incorrect terms. I think that is important.
| | | |
After resisting the trend for a long time, I eventually dipped my toe into CafeTran and found to my surprise that its MT engine can save me a lot of typing time with most of the source texts I deal with. These are various kinds of medical texts and, as such, contain a lot of long words - drug names and medical terminology - that the MT engine generally gets right (try typing "hepaticojejunostomy" or "abciximab" repeatedly, without error).
Of course, there are some howlers, which I c... See more After resisting the trend for a long time, I eventually dipped my toe into CafeTran and found to my surprise that its MT engine can save me a lot of typing time with most of the source texts I deal with. These are various kinds of medical texts and, as such, contain a lot of long words - drug names and medical terminology - that the MT engine generally gets right (try typing "hepaticojejunostomy" or "abciximab" repeatedly, without error).
Of course, there are some howlers, which I correct as I go, but the overall level in my language pair is much higher than it was a few years ago. By using this method to produce a rough first draft, then applying the same editing and polishing phases as before, I find I can save considerable time on a translation without sacrificing quality.
I don't accept PEMT jobs from other sources, so as often seems to happen here, none of the poll's answers apply to my case. ▲ Collapse
| | |
|
|
|
Lieven Malaise
Belgium
Local time: 13:36
Member (2020)
French to Dutch
+ ...
Philip Lees wrote:
After resisting the trend for a long time, I eventually dipped my toe into CafeTran and found to my surprise that its MT engine can save me a lot of typing time with most of the source texts I deal with. These are various kinds of medical texts and, as such, contain a lot of long words - drug names and medical terminology - that the MT engine generally gets right (try typing "hepaticojejunostomy" or "abciximab" repeatedly, without error).
The same goes for Deepl. It translates a lot of medical (and also technical) terminology correctly. Even to the point that it sometimes helps you in the right direction if you're not sure how to translate a certain term immediately.
The greatest 'BUT' is that it takes time to learn to deal with MT properly. You're not becoming a good (read: fast, delivering quality) MT post-editor overnight. You have to really dig into it. From a quality point of view it's also very important that you use the Auto Suggest function (with no more than 2 clicks Deeple changes the complete word order of even very long sentences without any grammar mistakes. So if you want to start your sentence with the last words of the Deeple generated translation, 2 clicks or typing the first word or two words you want to use is all it takes). A lot of people don't see it like that, but it's a skill that you have to develop.
I may seem a little bit too enthousiastic about MT, but it was a real eye-opener to me once I was able to get rid of my resistance. And I know how fierce that resistance can be. I've been there.
| | | | | Occasionally | May 11, 2022 |
I recently had the opportunity to use DeepL. As one who has been regularly exposed to MT output since 1980, I have to say that I was stunned by how good it was compared with any other system I've seen -- and I've seen many! Yes, my jaw dropped. The document I was asked to postedit was a long technical report from Portuguese to English. Other language combinations and domains may not be as good. Going forward, I definitely plan to use it from time to time. (BTW, the pay for that job was very good... See more I recently had the opportunity to use DeepL. As one who has been regularly exposed to MT output since 1980, I have to say that I was stunned by how good it was compared with any other system I've seen -- and I've seen many! Yes, my jaw dropped. The document I was asked to postedit was a long technical report from Portuguese to English. Other language combinations and domains may not be as good. Going forward, I definitely plan to use it from time to time. (BTW, the pay for that job was very good and the subject matter was extremely interesting, or I wouldn't have accepted it.)
[Edited at 2022-05-11 06:32 GMT] ▲ Collapse
| | | | | They even publish research on that | May 11, 2022 |
Barbara Cochran, MFA wrote:
...certain MT developers are now experimenting to see how MT could eventually be used to produce literary translations. I hope I won't be around at the time that might occur.
https://slator.com/post-editing-machine-translation-limits-creativity-in-literary-translation/
Excerpts:
"The reviewers were unanimous in ranking HT as an “extremely good translation,” MT as an “extremely bad translation,” and PE translation as “neither good nor bad.” The results also demonstrated that HT had the highest creativity score, followed by PE and, lastly, MT.
The translators thought that during post-editing, they were “primed or conditioned” by the text they were given, with one participant stating it was like having “a corset,” and that it was “difficult to step out” of the translation that was already there."
How surprising.
Philippe
| | | | | Pages in topic: < [1 2 3 4] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Poll: How often do you use DeepL or other machine translation tools? | Wordfast Pro | Translation Memory Software for Any Platform
Exclusive discount for ProZ.com users!
Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value
Buy now! » |
| | 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 » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
