| Pages in topic: < [1 2 3] | What do we consider a sustainable MTPE rate? Thread poster: Adieu
|
|---|
I must admit, this is something I expected. I believe that if the machine translation offered a really significant advantage, like translation memories and terminology management tools, all or most translators would use it and would do it consistently.
I am not saying I don’t appreciate the existence of this technology. It can be definitely used as a hint (and only as a hint) when I’m completely out of ideas. I sometimes use it for language learning when there is no teacher arou... See more I must admit, this is something I expected. I believe that if the machine translation offered a really significant advantage, like translation memories and terminology management tools, all or most translators would use it and would do it consistently.
I am not saying I don’t appreciate the existence of this technology. It can be definitely used as a hint (and only as a hint) when I’m completely out of ideas. I sometimes use it for language learning when there is no teacher around and I don’t have time to check up every single word (Google Translate can work as a cheap pocket dictionary). But as a realiable tool for professional translation, we’re simply not there yet.
In practical terms, I suspect machine translation could save a little bit of typing time and that’s about it.
[Edited at 2022-04-13 11:46 GMT] ▲ Collapse
| | | | | I no longer accept MTPE assignments | Apr 13, 2022 |
Anton Konashenok wrote:
In general, a fair price means you earn the same amount per hour for all linguistic services you provide. Suppose you can translate X words per hour or post-edit Y words per hour for the same kind of text, and your translation price is Z per word.
Then your fair MTPE price = Z × X/Y.
The ratio X/Y varies widely depending on the subject field, source text complexity, MT output quality, your professional experience and your criteria of acceptable quality. Don't be surprised if it takes you longer to post-edit than to translate the same text from scratch - to me, this is the case all the time, and since nobody wants to pay more for MTPE than for human translation, I don't do MTPE.
For the same reasons - but, what's more, for some ethical reasons as well - I no longer accept MTPE assignments either. Period!
| | | | | Accepting MTPE assignments comes for translators to digging their own grave... | Apr 13, 2022 |
Tom in London wrote:
There is no acceptable rate for MTPE.
MTPE is what used to be called "working on the first very rough draft of a translation" before machine translation was invented.
It is usually the first step in a lengthy and demanding process of writing and rewriting, always referring back to the source text, in order to arrive at a faithful and accurate translation that is usable for publication or other important purposes such as uploading to a website.
Machine translation, at best, will help you with the first 10% of the task; but the remaining 90% cannot be done by a machine. It requires brainwork and a level of literacy and linguistic creativity that even the best MTs, like Deepl, cannot do (no matter how close they seem to come, sometimes).
About an hour ago a client sent me a 200-word press release that I ran through CafeTran, also using Deepl to try out alternative ways of saying the same thing; but it then took me the best part of that hour to rephrase the translated text and polish it until it was ready to see the light of day.
MTPE is a scam. In my opinion no self-respecting, experienced translator would ever do it because they know the amount of work that's involved.
I couldn't have put it better, dear Colleague!
I simply would stress that accepting MTPE assignments comes for translators to digging their own grave...
Time will tell, but it'll be much too late for some of us. You've just be warned...
| | | | | The solution is quite simple : refusing to accept humiliating MTPE assignments | Apr 13, 2022 |
Jean Dimitriadis wrote: Claudio Porcellana wrote:
...
All that said, and to reply to the OP, 70% of the normal revision rate for heavy MTPE is considered fair nowadays, in the EU at least, and I think it can be applied to the per word rate as well
Claudio's analysis seems interesting and knowledgeable, but with all due respect, I think the rate suggestion ( 70% of the human revision rate, at most) is doing a disservice to fellow professional translators who will look for MTPE pricing inspiration in these forums. 70% of the human translation (per word) rate (on average, can be more, can be less, depending on the specifics) would be more like it. Standard per hour rate would be even better. I had objected to a similar suggestion by Claudio in the past, and I feel compelled to do so again. https://www.proz.com/forum/machine_translation_mt/318856-rates_per_hour_for_mtpe.html In page 2 of that same forum thread, I also make a case for eschewing MTPE altogether. [Edited at 2021-12-09 07:40 GMT]
Glad to have you joining the silent minority of translators out there who categorically refuse to accept humiliating MTPE assignments.
[Modifié le 2022-04-14 04:58 GMT]
| | |
|
|
|
Metin Demirel 
Türkiye
Local time: 21:41
Member (2018)
Italian to Turkish
+ ...
There are translators out there accepting regular translation projects with rates so low as MT rates. And they are happier to take MTPE jobs instead of tiresome human translations because they are getting paid the same. The threat is not MT but the fellow colleagues justifying their decisions or feeling compelled to accept the dictated rates. I've long lost hope that this would change. MT is just another facet of the philistine market ethic. MT is innocent. It's just another tool. If it wasn't ... See more There are translators out there accepting regular translation projects with rates so low as MT rates. And they are happier to take MTPE jobs instead of tiresome human translations because they are getting paid the same. The threat is not MT but the fellow colleagues justifying their decisions or feeling compelled to accept the dictated rates. I've long lost hope that this would change. MT is just another facet of the philistine market ethic. MT is innocent. It's just another tool. If it wasn't MT, we would find another way to shoot ourselves in the foot anyway. ▲ Collapse
| | | | Samuel Murray
Netherlands
Local time: 20:41
Member (2006)
English to Afrikaans
+ ...
Davide Fezzardi wrote: A few months ago, Charles University (Czech Republic) published a report on post-editing titled "Neural Machine Translation Quality and Post-Editing Performance" available at http://ufal.mff.cuni.cz/biblio/attachments/2021-zouhar-m7239318061676785041.pdf. As regards the possible increase of work efficiency by using machine translation, it includes a note, in page no. 9, reading: Contrary to current results, translating from scratch was not significantly slower than post-editing in either of the two phases.
This line and this study shows the importance of context. The sentence doesn't mean what it seems to mean. Well, to be honest, there is a lot that is unclear about the study. For example, "phases" is never defined. There are two "rounds", but I'm not sure if the rounds and phases are supposed to be the same thing. But let's assume that.
Note that the study uses the term "post-editing" to refer to reviewing (regardless of whether the original translation was human-translated or machine-translated). This means that editing a machine translation and editing a human translation are both called "post-editing" in this study. From what I can tell, the expression "from scratch" is used in this particular study for a human translation that does not undergo any revision.
So, the sentence means that translation without any additional revision rounds ("translating from scratch") is not significantly slower than two rounds of revision ("post-editing in either of the two phases") of a reference translation that already exists, e.g. a machine-translation or a previously translated human-translation.
What this study doesn't take into account (and that is no criticism), is that although translation without any revision is considered "safe" for academic purposes (e.g. to act as reference points for a BLEU score), no proper translation agency is going to pass along a human translation without at least one round of revision.
| | | | jyuan_us
United States
Local time: 14:41
Member (2005)
English to Chinese
+ ...
| You need to ask this question | Sep 2, 2022 |
What will the agency do with the copy you will deliver to them for a MTPE assignment? Will they publish/print it without having anybody review it, or will they have another translator further edit and finalize it?
I think the time you need to spend would vary greatly for each of these cases.
| | | | Ivana Bjelac
Croatia
Local time: 20:41
English to Croatian
+ ...
| The most sensible reply! | Dec 1, 2022 |
Anton Konashenok wrote:
.......................... if it takes you longer to post-edit than to translate the same text from scratch - to me, this is the case all the time, and since nobody wants to pay more for MTPE than for human translation, I don't do MTPE.
This is the most sensible reply!
Actually, MTPE suites the translation agencies only.
It is just the way to get the translation for less money.
| | |
|
|
|
Tony Keily
Local time: 20:41
Italian to English
+ ...
| Interesting implicit commentary by the Italian Revenue Agency | Dec 2, 2022 |
I've just been jumping through the hoops of my annual ISA ('synthetic reliability indexes' - you've just got to love them!) submission for Italian Revenue, like a trained circus dog. This involves telling the agency, in detail, how much work I've done in my sector in terms of hours, lines, words, pages, etc. and from which languages, in a desperate attempt to appear a 'reliable' taxpayer and not suffer the fate of Josef K.
This year the Revenue Agency has decided to introduce a new... See more I've just been jumping through the hoops of my annual ISA ('synthetic reliability indexes' - you've just got to love them!) submission for Italian Revenue, like a trained circus dog. This involves telling the agency, in detail, how much work I've done in my sector in terms of hours, lines, words, pages, etc. and from which languages, in a desperate attempt to appear a 'reliable' taxpayer and not suffer the fate of Josef K.
This year the Revenue Agency has decided to introduce a new category of work: 'light MTPE' (the agency actually uses the English word 'light').
I find this interesting because there's no mention of heavy MPTE or even plain old MPTE. This makes sense because it seems to suggest that if MPTE isn't light, it's basically the same as translation, which makes it pointless.
Which is actually what one of my agencies suggested to me about 10 years ago before MPTE was all the rage: 'We have some customers who sometimes just want to get the gist of documents, to see if they're of interest. They don't a need a standard translation-level product (at least initially). Could you tidy up MT to help them out for a lower fee than your translation fee?' ▲ Collapse
| | | | Sadek_A
Local time: 22:41
English to Arabic
+ ...
Let's get a few major things straight for any and all of the obstinates (not to say the thick skulls) plotting for (or, against, actually) Translation:
1. Human Translation must not be cheap. It's a tailored service, and must remain so for as long as it's around.
2. Human Translation must not be for/by amateurs, wannabes and/or passers-by.
3. Post Editing Machine Translation (PEMT), or Machine Translation Post Editing (MTPE), is a valuable, costly knowledge... See more Let's get a few major things straight for any and all of the obstinates (not to say the thick skulls) plotting for (or, against, actually) Translation:
1. Human Translation must not be cheap. It's a tailored service, and must remain so for as long as it's around.
2. Human Translation must not be for/by amateurs, wannabes and/or passers-by.
3. Post Editing Machine Translation (PEMT), or Machine Translation Post Editing (MTPE), is a valuable, costly knowledge-transfer service and must be paid as such, or otherwise eliminated and Machine is instead fed with pure translation inputs from thoroughly-screened, profession-dedicated and well-paid human translators.
*On PEMT/MTPE: you unintentionally, coercively and effectively lose your own style, vocabulary and train of thought to the Machine's input. The alternative is that you change everything in the input, which would consequently render the task T not PE.
- Someday, there might be studies on that first point.
4. Mass Production Machine Translation (MPMT) must be released as an independent, for-profit service only after it has been 100% perfected and 0% PEMT/MTPE-dependent.
Any other arrangement than the above is just a disruption, a serious one too. ▲ Collapse
| | | |
How has your view of machine translation changed, assuming it has, with the onset of ChatGPT?
| | | | | ChatGPT's not there yet, but... | Sep 25, 2023 |
ChatGPT is useful as a tool for translators, I use it by asking questions such as "which verbs correspond to the adjective ?" or "help me find a more concise way to say ".
As a Machine Translation tool, it is no better (yet) than DeepL as far as I can tell, though it's a little more flexible as you can give it specific instructions in natural language (on the other hand, you can't click on a word and look for the alternative suggestions like you can do in DeepL).
ChatGP... See more ChatGPT is useful as a tool for translators, I use it by asking questions such as "which verbs correspond to the adjective ?" or "help me find a more concise way to say ".
As a Machine Translation tool, it is no better (yet) than DeepL as far as I can tell, though it's a little more flexible as you can give it specific instructions in natural language (on the other hand, you can't click on a word and look for the alternative suggestions like you can do in DeepL).
ChatGPT, like DeepL, still have the flaws of Machine Translation that we know all too well. The problem is... many clients/prospects don't seem to know or understand the issue with Machine Translation, or how much work remains to be done by the translator to have good results.
At best, people are like "well, the MT is like 80% there, so you just have 10% of the work left to be done and we can just pay you 20%", not realizing that those "20%" actually represent 80% of the time we spend working on a given project.
At worse, people are like "wow, this MT translation is top-notch", when it really isn't. I've seen it quite a few times. In the English-French language pair, I can see how "translated" the text is, I can see all the calques, all the loan translations, all the wrong tenses being used. I can read the English thinly veiled behind the French MT.
But it seems many of my fellow countrymen can't. They seem to be just happy with the "English logic with French words slapped in place" text they read. They don't seem to even be able to spot the difference. And if they aren't able to tell... then what more can we do to convince them that no, this really sucks, this really isn't our true, natural language and they ought to hire a professional instead?
When people are perfectly happy with eating nearly-free synthetic junk, how do you convince them to pay a huge premium for actual food, when their tastebuds are so fucked they can't even tell the difference? I think this is the state of the translation industry right now.
And I have very little hope we can recover from this, when that difference in quality is only going to get narrower.
Those of us who only have a few years left before retirement can hope to get through this. The youngsters among us that are barely getting started? This is hopeless. ▲ Collapse
| | |
|
|
|
Marta Kumar
United Kingdom
Local time: 19:41
English to Polish
+ ...
| Pessimistic view | Nov 28, 2023 |
Gauthier Casimiro wrote:
When people are perfectly happy with eating nearly-free synthetic junk, how do you convince them to pay a huge premium for actual food, when their tastebuds are so fucked they can't even tell the difference? I think this is the state of the translation industry right now.
And I have very little hope we can recover from this, when that difference in quality is only going to get narrower.
Those of us who only have a few years left before retirement can hope to get through this. The youngsters among us that are barely getting started? This is hopeless.
100% agree with this as someone who has worked in the industry for over 15 years.
We live now in a world of mass-translations and functional illiteracy. High quality doesn't matter for the majority of people, they see no difference to MT. You can compare it to fast-food. Instant result, low quality. This world has no demand for high quality translations anymore exept for very specific, narrow fields. The big translation agencies know it and exploit it and profit from it. Just recently I was offered medical-related work, but the client failed to mention that the majority of their projects are MTPE. Big company, maybe the biggest on the market. If even the medical content is now machine translated, then I see no hope for the future...
| | | | Lingua 5B
Bosnia and Herzegovina
Local time: 20:41
Member (2009)
English to Croatian
+ ...
Marta Kumar wrote:
If even the medical content is now machine translated, then I see no hope for the future...
Now? In my language pair, I’ve been receiving MTPE offers in the medical field for at least 5 past years. That of course I ignore, but they’ve been around for quite some time, they aren’t new.
| | | | | Pages in topic: < [1 2 3] | There is no moderator assigned specifically to this forum. To report site rules violations or get help, please contact site staff » What do we consider a sustainable MTPE rate? | 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 » |
| | 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 
Translation news
