| Pages in topic: [1 2 3 4 5] > | DeepL Thread poster: Gerard Barry
|
|---|
I work as an in-house translator. My company has recently introduced DeepL. This means that everyone at the company will now be able to have their texts translated by DeepL if they so wish (as opposed to giving the work to us in the translation department). I've seen some of the DeepL translations and they're pretty damn good. I'm just wondering what other translators' thoughts are on the subject. Will DeepL replace a lot of translators? It's clearly faster than a human translator, and I presume... See more I work as an in-house translator. My company has recently introduced DeepL. This means that everyone at the company will now be able to have their texts translated by DeepL if they so wish (as opposed to giving the work to us in the translation department). I've seen some of the DeepL translations and they're pretty damn good. I'm just wondering what other translators' thoughts are on the subject. Will DeepL replace a lot of translators? It's clearly faster than a human translator, and I presume cheaper too. And as I said, the quality seems pretty good in the translated texts I've seen. ▲ Collapse
| | | |
Gerard Barry wrote:
Will DeepL replace a lot of translators? It's clearly faster than a human translator, and I presume cheaper too. And as I said, the quality seems pretty good in the translated texts I've seen.
Hi Gerard,
It's hard to say. While DeepL is pretty good on general texts, it is also quite awful on specialized technical terminology, as well as marketing and legal texts, apart from stereotypes, and I would not rely on it for important documents.
It is quite likely that there will be a lot more editing work and less translation, possibly for a smaller number of translators.
In any case, this is the general trend.
Sandra
| | | | Samuel Murray 
Netherlands
Local time: 00:52
Member (2006)
English to Afrikaans
+ ...
Gerard Barry wrote:
I'm just wondering what other translators' thoughts are on the subject. Will DeepL replace a lot of translators?
Yes, I think machine translation in general will result in a reduction of the number of translators. This is particularly true in situations where the translator is not liable for the quality of his translation. Machines still make mistakes, and it will become the job of translators operating these machines to spot those mistakes, but as clients keep pushing rates lower and lower, such translators start to work faster and faster, and eventually they start making mistakes themselves (or fail to spot mistakes made by the machine).
Feel free to start using DeepL, but know that your role then changes from the person deciding what to write to the person responsible for finding mistakes, and you must ware against becoming complacent when doing so.
| | | | Tom in London
United Kingdom
Local time: 23:52
Member (2008)
Italian to English
| Can be disastrous | Jun 20, 2021 |
I occasionally use Deepl for a quick and dirty rough translation of a particular phrase or even a paragraph.
It can be surprisingly good. But in the source language I translate (Italian) it can't handle complex sentence construction. Italians often put qualifying phrases in places where they wouldn't work in English and Deepl makes a dog's breakfast of trying to reconstruct them.
And it always makes the same mistakes over and over again (such as insisting that the col... See more I occasionally use Deepl for a quick and dirty rough translation of a particular phrase or even a paragraph.
It can be surprisingly good. But in the source language I translate (Italian) it can't handle complex sentence construction. Italians often put qualifying phrases in places where they wouldn't work in English and Deepl makes a dog's breakfast of trying to reconstruct them.
And it always makes the same mistakes over and over again (such as insisting that the columns in a building are "pillars", and so on).
So like all tools and as in every walk of life, I find that Deepl is an instrument that only works when it is in the hands of a skilled artisan. It is not a substitute for translation.
[Edited at 2021-06-20 09:00 GMT] ▲ Collapse
| | |
|
|
|
Kevin Fulton
United States
Local time: 18:52
German to English
| German to English improving | Jun 20, 2021 |
From time to time I run a few sentences through Deepl and Google Translate. I've noticed a steady improvement over the years, especially if long German sentences are broken down into more digestible components. Compared to GT, Deepl appears to make fewer terminology errors.
Machine translation can be useful, for example "gist translation", if the user needs to determine whether the content of a document warrants human translation. A few decades ago I was part of a team translating t... See more From time to time I run a few sentences through Deepl and Google Translate. I've noticed a steady improvement over the years, especially if long German sentences are broken down into more digestible components. Compared to GT, Deepl appears to make fewer terminology errors.
Machine translation can be useful, for example "gist translation", if the user needs to determine whether the content of a document warrants human translation. A few decades ago I was part of a team translating thousands of documents – at a bulk rate – relating to a legal discovery process having to do with patent infringement. Of the dozens of documents I translated, most involved maintenance schedules, expense reports and office visits; few related to the intellectual property issue at the heart of the lawsuit. Machine translation could have been used to screen these documents, saving the parties involved a lot of money. Although translators were kept busy for an extended period, they could have made more money doing something else. ▲ Collapse
| | | | Adieu
Ukrainian to English
+ ...
| I'm more scared of people going paperless | Jun 20, 2021 |
At the moment, bad PDF scans, clumsy OCR, and a reluctance to clean such documents is the biggest obstacle to MT implementation.
And another pleasant side effect is it royally screws up fuzzy match discount calculations. MT systems just don't recognize the 101% matches behind all those errors and often instead pick up the bad habit of producing very different results for the same string with a single typo or OCR error in different places.
I daresay paperless is the real... See more At the moment, bad PDF scans, clumsy OCR, and a reluctance to clean such documents is the biggest obstacle to MT implementation.
And another pleasant side effect is it royally screws up fuzzy match discount calculations. MT systems just don't recognize the 101% matches behind all those errors and often instead pick up the bad habit of producing very different results for the same string with a single typo or OCR error in different places.
I daresay paperless is the real nightmare.
[Edited at 2021-06-20 14:14 GMT] ▲ Collapse
| | | | Gerard Barry
Germany
Local time: 00:52
German to English
TOPIC STARTER | I suppose we'll have to wait and see | Jun 20, 2021 |
Sandra & Kenneth Grossman wrote: Hi Gerard, It's hard to say. While DeepL is pretty good on general texts, it is also quite awful on specialized technical terminology, as well as marketing and legal texts, apart from stereotypes, and I would not rely on it for important documents. It is quite likely that there will be a lot more editing work and less translation, possibly for a smaller number of translators. In any case, this is the general trend. Sandra
Hi Sandra,
Thanks for your response! I suppose we'll have to wait and see how it will all play out. The texts I've had translated through DeepL so far have been well-written. DeepL probably wouldn't make a good job of poorly written German texts, the translation of which makes up a fair share of the work in my company!
| | | | Joakim Braun
Sweden
Local time: 00:52
German to Swedish
+ ...
| Very impressed | Jun 20, 2021 |
I've tried it with a few short newspaper articles in languages I know. Very good indeed! Certainly a lot better than some human translators I've edited. Not only was the terminology accurate, but the text was actually pleasing to read. A huge improvement on Google translate or MT-translated shopping sites. Food for thought...
| | |
|
|
|
Edward Potter
Spain
Local time: 00:52
Member (2003)
Spanish to English
+ ...
I only use DeepL when I am sent its result by a client for post editing.
Does DeepL have to be used through a CAT or are there other ways of using it?
| | | | Adam Dickinson
Canada
Local time: 18:52
Member (2016)
Spanish to English
+ ...
| Short texts + documents | Jun 21, 2021 |
Edward Potter wrote:
I only use DeepL when I am sent its result by a client for post editing.
Does DeepL have to be used through a CAT or are there other ways of using it?
DeepL's website allows anyone to use it in the way you would, say, Google Translate - drop in some text, see what DeepL puts out.
If you have a Pro subscription (which you also need to get the API to use in CAT tools), DeepL also has the option to translate entire documents - upload a .ppt or .doc and it'll create a matching document in the target language. It works about as well as you might expect - workable but not great since additional formatting makes a robot's life hard.
@Adieu - yes, it's kind of funny how much job security terrible photocopies provide. It's like a giant captcha that we get paid for resolving.
@Gerard - I'm curious, do the new rules around DeepL include specifications that certain things -must- go through the translation department? I'd be surprised if anything customer-facing were entrusted entirely to DeepL at this point, even something as innocuous as a quick email...
| | | | Stepan Konev
Russian Federation
Local time: 01:52
English to Russian
Edward Potter wrote:
Does DeepL have to be used through a CAT or are there other ways of using it? Same as Google, you can use it either through their website, or through API in CAT tools, or through third-party apps like Qtranslate or GT4T that replace any selected text with translation.
| | | | Gerard Barry
Germany
Local time: 00:52
German to English
TOPIC STARTER | Like proofreading, I suppose | Jun 21, 2021 |
Samuel Murray wrote:
Feel free to start using DeepL, but know that your role then changes from the person deciding what to write to the person responsible for finding mistakes, and you must ware against becoming complacent when doing so.
Hi Samuel, yes post-editing a DeepL translation is a lot like proofreading, I guess: one of my least favourite tasks as an in-house translator by the way:)
| | |
|
|
|
Tom in London
United Kingdom
Local time: 23:52
Member (2008)
Italian to English
I was personally contacted a few weeks ago by a company that I believe is associated with Deepl, asking me if I would be interested in doing translations for them (Italian to English). They specified a number of interesting conditions, e.g. that I must not use a CAT tool or MT-
They must have identified me using Proz. There was no other way they could have done it.
I suspected at the time that Deepl is hiring professional translators (using Proz) to improve the most pro... See more I was personally contacted a few weeks ago by a company that I believe is associated with Deepl, asking me if I would be interested in doing translations for them (Italian to English). They specified a number of interesting conditions, e.g. that I must not use a CAT tool or MT-
They must have identified me using Proz. There was no other way they could have done it.
I suspected at the time that Deepl is hiring professional translators (using Proz) to improve the most problematic things that their algorhythm can't handle.
Since as a translator I am not in the business of hastening my own demise, and because they were offering a miserable rate, I didn't take them up (and anyway I was busy).
I would be interested to know if others have also been contacted by the same people.
[Edited at 2021-06-21 10:45 GMT] ▲ Collapse
| | | | Adieu
Ukrainian to English
+ ...
| How miserable | Jun 21, 2021 |
Was their miserable rate?
| | | | Gerard Barry
Germany
Local time: 00:52
German to English
TOPIC STARTER
Tom in London wrote:
I occasionally use Deepl for a quick and dirty rough translation of a particular phrase or even a paragraph.
It can be surprisingly good. But in the source language I translate (Italian) it can't handle complex sentence construction. Italians often put qualifying phrases in places where they wouldn't work in English and Deepl makes a dog's breakfast of trying to reconstruct them.
And it always makes the same mistakes over and over again (such as insisting that the columns in a building are "pillars", and so on).
So like all tools and as in every walk of life, I find that Deepl is an instrument that only works when it is in the hands of a skilled artisan. It is not a substitute for translation.
[Edited at 2021-06-20 09:00 GMT]
I'm not so sure. Over the years, I've become convinced that many translators overestimate the importance of their work. Very few clients want a perfect translation (even if they did, most translators can't deliver this either - we're only human). They just want one that's good enough. I think for some types of text at least, DeepL can meet - and exceed - this standard.
| | | | | Pages in topic: [1 2 3 4 5] > | There is no moderator assigned specifically to this forum. To report site rules violations or get help, please contact site staff » DeepL | TM-Town | Manage your TMs and Terms ... and boost your translation business
Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.
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 
Translation news
