| Páginas sobre el tema: [1 2] > | Poll: How do you feel about post-editing machine translation (PEMT)? Autor de la hebra: ProZ.com Staff
|
|---|
This forum topic is for the discussion of the poll question "How do you feel about post-editing machine translation (PEMT)?".
This poll was originally submitted by Ana Vozone. View the poll results »
| | | | | If you can't beat 'em, join 'em | Aug 12, 2019 |
I've been postediting machine translation since 1980 - coming up on 40 years. It's not my favorite thing to do, but I will say that the system I've been using sometimes turns out a decent job. My output ranges from 6,000 words a day when it's at its best to 1,800 words when the system isn't working well for various reasons.
When the system is performing decently and I can work fast, my postedited product is fully understandable and as error-free as I know how to make it. However, w... See more I've been postediting machine translation since 1980 - coming up on 40 years. It's not my favorite thing to do, but I will say that the system I've been using sometimes turns out a decent job. My output ranges from 6,000 words a day when it's at its best to 1,800 words when the system isn't working well for various reasons.
When the system is performing decently and I can work fast, my postedited product is fully understandable and as error-free as I know how to make it. However, when I re-read my work, I find that the language doesn't flow smoothly. It's not "organic" English. I end up revising it for flow and improved improved comprehension. On the other hand, when I revise my "from scratch" translations, I rarely change the language when I re-read it (just typos and stuff like that).
Trust me, I wouldn't do postediting if the compensation weren't decent. I don't believe translators should be paid any less for postediting than for real "human" translation.
[Edited at 2019-08-12 08:33 GMT] ▲ Collapse
| | | | neilmac
España
Local time: 05:32
español al inglés
+ ...
The poll, as usual when discussion of PEMT rears its ugly head, appears to be from the point of view of agencies, or perhaps direct clients, foisting PEMT jobs on to long-suffering translators in order to save a few measly cents. I prefer not to work in that kind of scenario, and I'm currently lucky enough to be able to reject any offers of that kind that might come my way.
However, most of my revision/correction/post-editing work consists of checking academic/scientific articles written ... See more The poll, as usual when discussion of PEMT rears its ugly head, appears to be from the point of view of agencies, or perhaps direct clients, foisting PEMT jobs on to long-suffering translators in order to save a few measly cents. I prefer not to work in that kind of scenario, and I'm currently lucky enough to be able to reject any offers of that kind that might come my way.
However, most of my revision/correction/post-editing work consists of checking academic/scientific articles written for publication in English by non-native authors, and nowadays they are obviously leveraging MT programs and online dictionaries and similar to produce their drafts. In that situation, I have no qualms about it, and it can even be fun spotting the typical MT howlers that crop up.
In general, I know from experience that even early attempts at MT (I'm thinking Systran, around the turn-of-the-century) could be useful for translating certain types of text, in the hands of a competent human translator. However, in general I wouldn't recommend PEMT to clients or businesses unless pennypinching is their prime concern. Caveat emptor. ▲ Collapse
| | | | | Depends on the payment | Aug 12, 2019 |
I voted "indifferent" here because I have no general objection against PEMT, as well as I have no general objection against any other tool or method. The main issue is the productivity and consequently the hourly compensation I can achieve with it. If the payment is so good and the MT quality so high that the resulting compensation is attractive, why not? However, I have tried it here and there, and MT quality in my pair mostly was pretty poor. Also, it seems that most clients/agencies offering ... See more I voted "indifferent" here because I have no general objection against PEMT, as well as I have no general objection against any other tool or method. The main issue is the productivity and consequently the hourly compensation I can achieve with it. If the payment is so good and the MT quality so high that the resulting compensation is attractive, why not? However, I have tried it here and there, and MT quality in my pair mostly was pretty poor. Also, it seems that most clients/agencies offering PEMT jobs seem to be in the bottom rate range, so that it does not work out. ▲ Collapse
| | |
|
|
|
Let’s start by saying that I don’t enjoy editing or proofreading other’s work as much as enjoy translating and so what I feel about PEMT is tainted from the very beginning. That being said, if I haven’t done PEMT so far has everything to do with the compensation. If and when I happen to find a client willing to pay a decent rate I might do like Muriel…
| | | | | I have tried it and I don´t like it | Aug 12, 2019 |
I am in the lucky position that I can just turn down PEMT jobs. I have enough translation work to do, and I can just retire completely if and when I get tired of them - or am no longer able to maintain a high standard.
PEMT requires a special approach, and I have not managed to develop it. I have always felt it would be quicker and easier to translate from scratch instead, and PEMT takes me ages, so it is not economically viable for the client to pay me a satisfactory hourly rate.... See more I am in the lucky position that I can just turn down PEMT jobs. I have enough translation work to do, and I can just retire completely if and when I get tired of them - or am no longer able to maintain a high standard.
PEMT requires a special approach, and I have not managed to develop it. I have always felt it would be quicker and easier to translate from scratch instead, and PEMT takes me ages, so it is not economically viable for the client to pay me a satisfactory hourly rate.
And it is not viable for me if the client pays a low rate. 
I think it is going to be necessary for younger translators to accept PEMT, and some already find it quite useful.
I hope the market will settle into an end where edited MT is OK, and an end where clients understand that MT is never going to be adequate, and they will pay reasonable rates for human translators.
[Edited at 2019-08-12 09:28 GMT] ▲ Collapse
| | | |
I am about conscious job taking.
| | | | Thayenga 
Alemania
Local time: 05:32
inglés al alemán
+ ...
I've tried it twice, so there will definitely be no third attempt. Too much work for too little pay.
The only good thing about MT is that it sometimes really makes you laugh at the awkward results it produces.
[Edited at 2019-08-12 13:22 GMT]
| | |
|
|
|
Adam Warren
Francia
Local time: 05:32
Miembro 2005
francés al inglés
| From the wrong end | Aug 12, 2019 |
Elsewhere, a colleague said that MT should be used as a resource for translators to draw on upstream in the translation process. The clients have introduced this unmatured technology from the wrong end, instead of involving the translation community from the outset.
| | | | Ana Vozone
Local time: 04:32
Miembro 2010
inglés al portugués
+ ...
| If you can't beat them, join them | Aug 12, 2019 |
I have done quite a lot of post-editing jobs, but never accept this type of work before having a look at the actual file. I think you need to balance quality of the machine translation and budget offered by client. And sometimes the combination is good. I have found nearly perfect translations of long and tedious segments where I only have to change a small detail. So, post-editing is just a new area of work for me which I quite enjoy, after some initial rejection/aversion to the concept. PEMT i... See more I have done quite a lot of post-editing jobs, but never accept this type of work before having a look at the actual file. I think you need to balance quality of the machine translation and budget offered by client. And sometimes the combination is good. I have found nearly perfect translations of long and tedious segments where I only have to change a small detail. So, post-editing is just a new area of work for me which I quite enjoy, after some initial rejection/aversion to the concept. PEMT is here, and it is getting bigger and, in many cases, better and easier to work with. The quality of the client's platform features is also key. (I have never edited this type of work using MS Office files. My experinece is only with jobs done on client's platforms.) ▲ Collapse
| | | | | It is inevitable | Aug 12, 2019 |
MT is getting better and better, and it is inevitable that clients ask for revisions of machine translations.
When technologies make a viable product cheaper, clients want to take advantage of that price reduction, and suppliers want to offer those discounts to be more competitive. it is basic economics, though many translators seem to think it is some kind of abusive or insulting practice.
In fact, I am shocked at how well "traditional" translation has resisted this ... See more MT is getting better and better, and it is inevitable that clients ask for revisions of machine translations.
When technologies make a viable product cheaper, clients want to take advantage of that price reduction, and suppliers want to offer those discounts to be more competitive. it is basic economics, though many translators seem to think it is some kind of abusive or insulting practice.
In fact, I am shocked at how well "traditional" translation has resisted this trend to date.
That said, they can't expect miracle rates, unless the results of the MT are very good, and some MT translations are, of course, simply awful, to the point that a revision is just not even viable.
In any case, this will be a rising and unstoppable trend, and translators need to recognize it, unless among the select few who translate things like novels, immune to the MT trend.
[Edited at 2019-08-12 14:32 GMT] ▲ Collapse
| | | | | This is pretty much the future | Aug 12, 2019 |
I took on a French-English MT job the other day and was impressed with the overall quality -remember a lot of machine translation is generated from corpora which has been produced by professional human translators.
The quality is only going to get better - the industry is already changing and it is getting harder and harder for "generalist" translators to make a living.
Translators are going to have to adapt to survive, I'm afraid.
| | |
|
|
|
| MTPE or PEMT is mainly a euphemism | Aug 12, 2019 |
MT can't be trusted for accuracy. Point 1 Couple of other points:
You need a human translator to check the original text and the translation. We're not talking about just fixing grammatical errors or stylistic shortcomings.
It's really not my experience that MT is getting better all the time. It can be atrocious in the areas of marketing or literature.
In technical, medical and legal fields, you better check every single word, phrasing, and sentence structure of that MT ... See more MT can't be trusted for accuracy. Point 1 Couple of other points:
You need a human translator to check the original text and the translation. We're not talking about just fixing grammatical errors or stylistic shortcomings.
It's really not my experience that MT is getting better all the time. It can be atrocious in the areas of marketing or literature.
In technical, medical and legal fields, you better check every single word, phrasing, and sentence structure of that MT text before you deliver the translation.
I don't say you can't use MT in the translation process, but the translation becomes a translation only after it has been checked and double-checked by a human translator. And that process isn't a simple editing job. It's translating by a human translator, with or without the help of MT.
Even if it turns out that MT was helpful - and it can be - it's only as good as the translator who reviews it.
MTPE is a misnomer. It is supposed to convey that it is easier than translating from. scratch and should therefore be cheaper. It also implies that it's something that's done to an already pretty decent translation.
What it really is is a euphemism for "give me a cheaper rate than for translating." Most people asking for MTPE IMO have exactly that in mind.
To counter misconceptions others have because they honestly think MT warrants a discount because they heard somewhere that MT is making it so much easier and faster to provide translations, you need to point out the limits and shortcomings of MT, the skills required to judge the accuracy and quality of MT, the overall work involved - esp. the part about checking the OT against the TT, and the fact that you as a translator deliver the translation based on your knowledge and skill, and not the machine.
I don't think it's in our or the client's interest to taut MTPE as something that replaces the human translator in any shape or form or justifies lower rates. On the contrary, any tool that can assist the translator in improving the quality of his or her work should make his or her work more valuable. But it's for the human translator to decide what these tools are.
PS: It's also a good idea to point out to your clients that amateurs who lack the required knowledge, skills or experience might rely heavily on MT and that what they deliver might look like a perfectly fine translation but that it could possibly be of very poor quality or even worse, MT with a few changes to make it look "edited,". That's why many clients actually ask not to use MT.
But as always, only those who apply best practices will render top-notch translations. And they know how to do that and what that work is worth. ▲ Collapse
| | | | Mario Freitas
Brasil
Local time: 01:32
Miembro 2014
inglés al portugués
+ ...
| I used to reject it | Aug 12, 2019 |
But now I see it's a tend we'll hardly run away from. So, as long as the client pays an intermediate price between the translation and the revision rate, and as long as it's NOT Google Translate or other similar ones, it's acceptable, and we'll have to swallow it to get work in the future.
| | | | Saara Kauppinen (X)
Finlandia
Local time: 06:32
inglés al finlandés
+ ...
| No negative experiences | Aug 12, 2019 |
Reading the other comments here, I believe I've been extremely lucky with PEMT so far. I've done several jobs like this, and all of them have been of pretty good quality. I've also negotiated my regular translation rate for the parts I've had to translate from scratch, instead of having the whole project marked as PEMT and paid accordingly. You should never agree to do a project solely as "PEMT" if it includes translating anything from scratch, you're being exploited. It's important to negotiate... See more Reading the other comments here, I believe I've been extremely lucky with PEMT so far. I've done several jobs like this, and all of them have been of pretty good quality. I've also negotiated my regular translation rate for the parts I've had to translate from scratch, instead of having the whole project marked as PEMT and paid accordingly. You should never agree to do a project solely as "PEMT" if it includes translating anything from scratch, you're being exploited. It's important to negotiate and find out what the project actually entails! If the client is not willing to negotiate, you're probably better off not taking on the job.
Also, machine translation may be of worse quality than what you would produce from scratch, but that's not on you! If the client wants a perfect translation from scratch, they can pay for it. If not, let the machine do its thing and make sure it's legible. Your time is valuable!
Bottom line, I don't mind PEMT. I've actually enjoyed it so far. But I do understand MT can be terrible, and the translator may have to pick up the slack. I just think this is where you have to stand up for yourself, tell the client how atrocious the MT quality is and demand to be paid for translating, since that's basically what you'll be doing. ▲ Collapse
| | | | | Páginas sobre el tema: [1 2] > | 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 do you feel about post-editing machine translation (PEMT)? | Trados Business Manager Lite | Create customer quotes and invoices from within Trados Studio
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 » |
| | Trados Studio 2022 Freelance | The leading translation software used by over 270,000 translators.
Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop
and cloud solution, empowering you to work in the most efficient and cost-effective way.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
