| Pages in topic: [1 2 3] > | The Future of Translation Thread poster: Isely Mills
|
|---|
Isely Mills 
United States
Local time: 09:22
Japanese to English
Recently, I was offered a project wherein I would provide translations of random bits of text from around the internet, with the goal being to provide a data set for a translator neural net to train off of.
If you do not know what a neural net is, I can explain, but basically it is a programmed structure that learns in the same way our brains learn, and as such is able to learn very complicated things on its own, without the need for a programmer to specifically teach it the task in... See more Recently, I was offered a project wherein I would provide translations of random bits of text from around the internet, with the goal being to provide a data set for a translator neural net to train off of.
If you do not know what a neural net is, I can explain, but basically it is a programmed structure that learns in the same way our brains learn, and as such is able to learn very complicated things on its own, without the need for a programmer to specifically teach it the task in question.
In general, a neural net needs 3 things to learn a task:
1) A robust neural net/learning algorithm
2) A large data set
3) Time
So, it seems like it is only a matter of time (and money) before automated translation becomes a central player in the translation business (whether it usurps translators, or becomes another translation tool is another matter).
Basically, I would love to hear from anyone who has insight in this matter. What do you think the future of this industry looks like? Do you think that freelancers will be able to compete with agencies for business? Will there still be a place for manual translation? Will translators need to become post-editors to stay relevant in the next 5-10 years? ▲ Collapse
| | | | Mario Chavez (X)
Local time: 11:22
English to Spanish
+ ...
| Some thoughts | Nov 26, 2017 |
What do you think the future of this industry looks like?
The future of the translation profession (not to be confused with the so-called “industry” —more on that later) is the same as it has always been. As long as there are humans having something to say, inform or teach, there will be translators to meet the challenge.
Do you think that freelancers will be able to compete with agencies for business?
Individual translators and translation agencies may do some of the same things, but they have different goals. Translators have professional goals: good writing quality and good research, proper style and register, caring for the right reader and audience, proper language usage, etc. Agencies's goals are more business-oriented: customer service and satisfaction, profitability, cost management, etc. Of course translators have to attend to some business goals of their own.
Will there still be a place for manual translation?
Please refer to my answer above under “The future of the translation profession.”
Will translators need to become post-editors to stay relevant in the next 5-10 years?
Not necessarily. Only those who are engaged in PEMT projects and translators whose paycheck depends on working as employees of PEMT companies, MT companies and the like. Historically, they're in the minority. MT, whether conventional or neural-network based, is a niche and will remain so because annual licenses and maintenance are prohibitively expensive (only large companies can afford them) and because the customization of an MT engine is time consuming (is dependent on translators, not just software tweaks) and never reaches 100% accuracy under any meaning of the word.
What's my insight? I worked with PEMT for about two years for two different clients. First client used Lingotek, a web-based MT engine that provided us translators with a web-based TM interface and glossary, a basic tool compared to established TEnT tools we know and love (Déjà Vu X3, SDL Trados, Wordfast, memoQ). With Lingotek, we were allowed to rewrite the suggested MT translations, and rewrite we did. Without having hard numbers to go by, I'm assuming my language pair (English to US Spanish) yielded a pre-editing accuracy of 50-75%.
In the second case, the client, an oil conglomerate, provided Word files that were already processed and needed PEMT while preserving the document's preferred formatting. In confidence, one of the staff members handling the preparation of such files told me with a hint of resignation that the MT engine only achieved about 75% accuracy.
That software companies selling and licensing MT, AI-based MT and PEMT products are computationalists doesn't surprise me. On the other hand, there are many fervid computationalists among well-respected translators and TS (Translation Studies) academics (one of them was part of the examination board for my PhD thesis proposal recently), and that's a very sad state of affairs. For basic information on computationalism, see https://mitpress.mit.edu/books/computationalism
A word regarding “industry”: software companies, affiliates and many translation agencies that adopted ISO standards are task-oriented entities. To them rationalizing translation, proofreading and editing processes as if they were predictable and measurable units makes economic sense. Not to us translation practitioners. The differentiation between industry and profession is more than a tussle over semantics.
| | | | Tom in London
United Kingdom
Local time: 16:22
Member (2008)
Italian to English
| Long way to go | Nov 26, 2017 |
As far as CAT tools are concerned, we're still in the stone age. I can't understand why people seem concerned. At their current stage of development, they're clunky - all of them, and they keep breaking down. It will be a long time before CAT tools (or whatever replaces them) are accurate and reliable and require no human input. That won't be happening any time soon - not before there's a complete rethink of what translation is, and what kind of IT has the ability not only to translate a text bu... See more As far as CAT tools are concerned, we're still in the stone age. I can't understand why people seem concerned. At their current stage of development, they're clunky - all of them, and they keep breaking down. It will be a long time before CAT tools (or whatever replaces them) are accurate and reliable and require no human input. That won't be happening any time soon - not before there's a complete rethink of what translation is, and what kind of IT has the ability not only to translate a text but to understand what it means and what its intention is, as well as irony, wit, the development of an idea, etc. CAT tools are kind of OK (though still not reliable) for mechanically written texts, but that's all. They're no good where some sort of style and elegance are called for, be it a business letter or an academic thesis.
[Edited at 2017-11-26 17:12 GMT] ▲ Collapse
| | | | Daniel Frisano
Italy
Local time: 17:22
Member (2008)
English to Italian
+ ...
Good translators will still produce texts that are far superior to those produced by MT, and for a long time - I'm with Tom here. Top clients will still appreciate, seek and reward this kind of service.
Mediocre-to-bad translators will increasingly need to become post-editors. In fact, I am sure that some of them ALREADY post-edit MTed material and peddle it as a translation, if the stuff I proofread is any indication. Or perhaps it's just that their style is indistinguishable from ... See more Good translators will still produce texts that are far superior to those produced by MT, and for a long time - I'm with Tom here. Top clients will still appreciate, seek and reward this kind of service.
Mediocre-to-bad translators will increasingly need to become post-editors. In fact, I am sure that some of them ALREADY post-edit MTed material and peddle it as a translation, if the stuff I proofread is any indication. Or perhaps it's just that their style is indistinguishable from MT. Either case, this segment alone will have to worry about MT taking their place.
There will always be BMWs and Hyundais, and they won't compete with each other. To each its own market segment.
[Edited at 2017-11-26 17:39 GMT] ▲ Collapse
| | |
|
|
|
Andrew78 (X)
Local time: 17:22
| We will still need humans | Nov 26, 2017 |
As much as machines could become intelligent, even if just for a tiny detail, they'll never be able to stand in for human. Personally I would never trust a machine translated test unless proofreaded by a translator. Then, if we are speaking about a future of Androids (not the phone, but the Robot) then we won't live enough to see it. It takes time to reach that level. Even nowdays, who does not rely on pc and CDs,but when you have something really important to keep, you choose paper.Maybe if you... See more As much as machines could become intelligent, even if just for a tiny detail, they'll never be able to stand in for human. Personally I would never trust a machine translated test unless proofreaded by a translator. Then, if we are speaking about a future of Androids (not the phone, but the Robot) then we won't live enough to see it. It takes time to reach that level. Even nowdays, who does not rely on pc and CDs,but when you have something really important to keep, you choose paper.Maybe if you just need to fill with contents a website even now Google translatore might do, as long as you don't need visitors to actually need the text, but think about a promotional brouchere os a dentist service, for instance, no, no machine could take now of in the folliwing 20 years such responsibility. ▲ Collapse
| | | | | Not yet, nor any time soon. | Nov 26, 2017 |
Two thoughts in support of “no”:
1. If neural net-based translation output provides a better or on par result as opposed to a human-produced output, we will be talking about a major paradigm shift in the functioning of the world as we know it today. This means that there will be no human soldiers, accountants or even musicians. That will probably come, but not in out lifetime. I recall I used my first mobile phone in 1997. A few years later, a service called “WAP” appeared (... See more Two thoughts in support of “no”:
1. If neural net-based translation output provides a better or on par result as opposed to a human-produced output, we will be talking about a major paradigm shift in the functioning of the world as we know it today. This means that there will be no human soldiers, accountants or even musicians. That will probably come, but not in out lifetime. I recall I used my first mobile phone in 1997. A few years later, a service called “WAP” appeared (you could access internet and move some data around). Well, my smartphone in 2017 looks pretty much the same thing to me, just with improved interface, more data capacity, etc. There is also a downside; the battery drains faster…
2. So long translation output addresses a human being (contracts, marketing-related texts, poetry, etc.), it will have to be produced by a human being, preferably from the same cultural and social background as the recipient of the message.
[Edited at 2017-11-26 22:15 GMT] ▲ Collapse
| | | | Gerard de Noord
France
Local time: 17:22
Member (2003)
English to Dutch
+ ...
| The future of translation won't resemble the past of translation | Nov 26, 2017 |
For translators the next ten years will be like the last ten years olive oil producers have gone through.
The artisans were still making their extra virgin olive oil in huge medieval presses, while technicians found a way to obtain a product of the same quality by centrifugal force.
I’ve been translating about extra virgin olive oil for at least 20 years, and I won’t be the one to divulge that a product is no longer as special as it used to be.
I became... See more For translators the next ten years will be like the last ten years olive oil producers have gone through.
The artisans were still making their extra virgin olive oil in huge medieval presses, while technicians found a way to obtain a product of the same quality by centrifugal force.
I’ve been translating about extra virgin olive oil for at least 20 years, and I won’t be the one to divulge that a product is no longer as special as it used to be.
I became a translator because of the new opportunities the Internet afforded me, and I must have been among the first Wordfast users. Twenty years ago, we were awed by the idea that we would always be able to compare a new translation with former translations we had made.
Today, I witness that Deepl presses the same oil as medieval presses. Again, I won’t be the one to divulge anything.
Cheers,
Gerard ▲ Collapse
| | | | Hamish Young
New Zealand
Local time: 03:22
Chinese to English
| Hard to predict the future! | Nov 27, 2017 |
I was offered the same project, although it doesn't seem to have materialized as yet. But I did work on a very similar project nearly 10 years ago, producing output that was designed to improve machine translation.
It's hard to predict the future but we can look to the past and realize that the proliferation of translation jobs is quite a recent phenomenon. There have always been translators but surely not this many. There is no reason to expect that so many translators will be requ... See more I was offered the same project, although it doesn't seem to have materialized as yet. But I did work on a very similar project nearly 10 years ago, producing output that was designed to improve machine translation.
It's hard to predict the future but we can look to the past and realize that the proliferation of translation jobs is quite a recent phenomenon. There have always been translators but surely not this many. There is no reason to expect that so many translators will be required in the future.
I have a hard time getting my head around the concept of neural translation but I've definitely noticed a massive increase in the quality of google translate in the past couple of years in my area of expertise. I would like to think of myself as a very experienced and capable translator but it is not uncommon now for google to translate a sentence better than I could have done myself. Occasionally I will even see whole paragraphs translated quite expertly, requiring only a few adjustments. This was not possible even two years ago. It would be no understatement to say that at a paragraph level in my work, google is capable of outperforming many human translators and can actually be faster to proofread, but at a document level it is still deficient. It must be said that I work on technical texts rather than creative texts. ▲ Collapse
| | |
|
|
|
Tom in London
United Kingdom
Local time: 16:22
Member (2008)
Italian to English
| That's because | Nov 27, 2017 |
Hamish Young wrote:
.....Occasionally I will even see whole paragraphs translated quite expertly, requiring only a few adjustments. This was not possible even two years ago....
- and do you know why that happened? Because real, human translators using Google Translate were foolish enough to help Google, for free, by suggesting better terms and phrases.
If anyone reading this has been doing that: please stop.
| | | | Isely Mills
United States
Local time: 09:22
Japanese to English
TOPIC STARTER
Thank you all for the perspective.
It seems like we are still a ways off from highly reliable neural net-based translation that can deliver the same quality of a human at an equal or lesser price, especially in domain-specific and creative translation. I suppose we will just have to wait and see what happens. I hope that if machine translation does become a major competitor, that we can use it as a tool, rather than be replaced by it.
I really do enjoy translating, and... See more Thank you all for the perspective.
It seems like we are still a ways off from highly reliable neural net-based translation that can deliver the same quality of a human at an equal or lesser price, especially in domain-specific and creative translation. I suppose we will just have to wait and see what happens. I hope that if machine translation does become a major competitor, that we can use it as a tool, rather than be replaced by it.
I really do enjoy translating, and I do not enjoy editing machine translations, so I hope there is still a place for us in the future. Anyway, your responses give me hope for that future. ▲ Collapse
| | | | Mario Chavez (X)
Local time: 11:22
English to Spanish
+ ...
| Let's not fret | Nov 27, 2017 |
Isely Mills wrote:
Thank you all for the perspective.
It seems like we are still a ways off from highly reliable neural net-based translation that can deliver the same quality of a human at an equal or lesser price, especially in domain-specific and creative translation. I suppose we will just have to wait and see what happens. I hope that if machine translation does become a major competitor, that we can use it as a tool, rather than be replaced by it.
I really do enjoy translating, and I do not enjoy editing machine translations, so I hope there is still a place for us in the future. Anyway, your responses give me hope for that future.
Technologies, whatever they are, are always dependent on their creators and on human mind and labor. It's easy to feel uneasy, even scared, on the face of these so-called AI and neural network advances, but careful reading of human history should dispel such nonsense.
| | | | Roy OConnor (X)
Local time: 17:22
German to English
| A dying profession and industry | Nov 28, 2017 |
Let’s face it. The last 25 years have seen HUGE changes for translators. The Internet has made global communications easier and has supplied instant infinite knowledge to everyone. My view is that neural systems and the harvesting of mega-data will have a similar disruptive effect, particularly where languages are concerned. We shouldn’t keep our heads in the sand any longer.
If you don’t believe me, then read the (often somewhat technical) literature. An easy book to read is ... See more Let’s face it. The last 25 years have seen HUGE changes for translators. The Internet has made global communications easier and has supplied instant infinite knowledge to everyone. My view is that neural systems and the harvesting of mega-data will have a similar disruptive effect, particularly where languages are concerned. We shouldn’t keep our heads in the sand any longer.
If you don’t believe me, then read the (often somewhat technical) literature. An easy book to read is Martin Ford’s “The Rise of the Robots” (Business Book of the Year 2015). If you don’t feel scared after reading this book, then you must be in the latter stages of your life when it doesn’t matter anyway! ▲ Collapse
| | |
|
|
|
Daniel Frisano
Italy
Local time: 17:22
Member (2008)
English to Italian
+ ...
| I sure hope Google’s self-driving car is better than Google Translator, yet... | Nov 28, 2017 |
Take the original posted message:
(...) is a programmed structure that learns in the same way our brains learn, and as such is able to learn very complicated things on its own, without the need for a programmer to specifically teach it the task in question.
(...) So, it seems like it is only a matter of time (and money) before automated translation becomes a central player (...)
Now substitute "driver" for "programmer" and "self-driving vehicles" for "automated translation" and you get:
(...) is a programmed structure that learns in the same way our brains learn, and as such is able to learn very complicated things on its own, without the need for a driver to specifically teach it the task in question.
(...) So, it seems like it is only a matter of time (and money) before self-driving vehicles become a central player (...)
I don't know about you, but the future is suddenly looking a bit scarier for pedestrians, cyclists, ...
| | | | Silvia Schulz
Netherlands
Local time: 17:22
English to German
+ ...
| Uncertain future | Nov 28, 2017 |
Hi there,
I also wanted to share my five cents. I do not share the optimism of some of the previous commentators. I am mainly working in the technical field with translation agencies and see that many of my clients are currently exploring the use of MT engines. Some have already achieved such a high quality output, that I was pretty shocked. Let's be honest, I will never be able to have the same amount of vocabulary as a machine.
I think the trend, at least in my field,... See more Hi there,
I also wanted to share my five cents. I do not share the optimism of some of the previous commentators. I am mainly working in the technical field with translation agencies and see that many of my clients are currently exploring the use of MT engines. Some have already achieved such a high quality output, that I was pretty shocked. Let's be honest, I will never be able to have the same amount of vocabulary as a machine.
I think the trend, at least in my field, for the next years will be a mixture of machine and human working together. Is my job still safe in 10 years? I am not sure. Can I work in this field until retirement in 30 years? Certainly not!
The industry is currently going through a shift. I can also see that in the increasing number of translation coaches out there, who have sprung up in the last year. Many translators seem to feel the impact and try to work on a better online visibility/branding etc. I am also currently working on rebranding my business, a marketing plan and building a new website.
Recently, I spoke to a manager from a translation agency at a networking event and she said that in Denmark the first translation department of a university is shutting down due to lack of new students. She predicts that in the near future there'll only be middle-aged translators left as young people will be hesitant to pick up this profession. She also said that I was still young enough to consider alternatives. Mmmh. I did not find this very encouraging.
Sorry for the long post. I don't have any conclusion either, but I think it is a good idea to keep track with the latest developments and adjust our businesses before we get a bad surprise. I for sure do not agree that MT will not have any impact on our profession any time soon, because it is already here.
Cheers,
Silvia
[Bearbeitet am 2017-11-28 18:16 GMT] ▲ Collapse
| | | | Mario Chavez (X)
Local time: 11:22
English to Spanish
+ ...
| May I suggest this other book? | Nov 28, 2017 |
Roy OConnor wrote:
Let’s face it. The last 25 years have seen HUGE changes for translators. The Internet has made global communications easier and has supplied instant infinite knowledge to everyone. My view is that neural systems and the harvesting of mega-data will have a similar disruptive effect, particularly where languages are concerned. We shouldn’t keep our heads in the sand any longer.
If you don’t believe me, then read the (often somewhat technical) literature. An easy book to read is Martin Ford’s “The Rise of the Robots” (Business Book of the Year 2015). If you don’t feel scared after reading this book, then you must be in the latter stages of your life when it doesn’t matter anyway!
I strongly suggest you read up on computationalism and on The Tides of Mind, written by (oh, surprise!) a computer scientist, David Gelernter.
http://amzn.to/2zNZj7D
I don't care for Mr. Gelernter's political biases but this book (which I'm reading these days) throws some skepticism into computationalism, or the theory that human thoughts and thought processes are “computational.”
| | | | | 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 » The Future of Translation | 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 » |
| | CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
Translation news
