Pages in topic:   [1 2] >
Is translation a field I can work in despite the rise of MT?
Thread poster: Andrew Dirr

Andrew Dirr
Germany
Local time: 01:23
German to English
+ ...
Mar 1

Hi everyone,

I'm hoping for some insights from a few people who have been able to gather some experience in the professional world of translation. I'm an English native speaker (U.S.) who moved to Germany at the age of 18. I have worked a number of different jobs in Germany and obtained German citizenship the year before last. However, I still hold no degree other than a German Abitur (something like a highschool diploma).

As I am a bit older (28) now, I have been worried about not having any other kind of education. This can be especially difficult in Germany. So I started studying a social science last year. While doing so, I was also given a job as a translator in a research network on campus. I soon found that I actually had a lot more fun translating than studying in my field. I began noticing that I had to painfully force myself to study for exams and write papers, but when it came time to translate, I was on top of it right away, enjoying it and forgetting all about the time. The feedback I receive from the research network's professors and postdocs (English and German speakers alike) is also very positive.

As if the universe wanted to give me a further sign, a good German friend of mine has also recently approached me with a very large work of literature. I am now translating it from English to German, being paid for it and having a wonderful time doing it. My German is usually mistaken for that of a German native. So far, it hasn't been a problem to translate in two directions. I used to spend hours upon hours reading German dictionaries (Duden and Wahrig) and absorbing the words. I simply love the language.

After studying a few semesters here and there in different fields and trying things out, I must admit that translation is really what does it for me. I feel like I have finally found what fulfills me in life! So it was much to my dismay that I stumbled upon DeepL. I ran a few texts through it and found that it did quite a good job on them. I was planning on starting a degree in translation this winter, but now I fear that human translation will become quite obsolete due to the exponential progress being made by machine translation and A.I. in general.

I was hoping for some opinions on how some of you see the future of translation. I'm trying to gather data and information before committing to a degree. I don't want a degree that is completely worthless, but am also not looking to earn tons of money. If I can live off of translation decently, afford healthy food, rent a small place, maybe have a cat and also have time to continue learning new things and reading lots of books, I will be more than happy. I just want to be sure I am not entering a "dying field" where there will be literally nothing left to do in the near future. I don't know what I would do otherwise. I feel that this is really it, what I have been searching for! It would be very sad to have finally found it, only to lose it to technology.

How do you see this? Any thoughts or opinions?

Thanks so much in advance. Every word is highly appreciated!

Best,

Andrew


Direct link Reply with quote
 

Michael J.H. Davies  Identity Verified
Denmark
Local time: 01:23
Member (2009)
English to Danish
+ ...
the future of translation ... Mar 3

I wonder if anyone is able to state without a shadow of doubt that translation by humans will never be only of historical interest. I certainly wouldn't like to make any kind of categorical statement either one way or the other.

Fortunately, I started with translation work as a source of income late in my career (I am also a professionally qualified engineer and independent consultant) - mainly (to begin with) as a 'gap filler' between consultancy assignments. In recent years, however. translation has become my main source of income with occasional consultancy work now and again.

I have noticed in recent years that MT has become increasingly better and I now often use MT (with reliable sources of translation and guaranteed confidentiality - as a paid service) as a first stage of translation before 2 separate stages of 'post editing' - the first to ensure that all ideas from the source text are correctly represented in the target text and the second to ensure that the target text reads correctly and fulfils the required quality criteria. This process enables me to achieve the required quality of translation with reduced effort (and especially time) so that the price remains competitive

I believe that the combination of MT and post editing is likely to become more and more a requirement rather than an option and have already seen this development in several of the translation bureaux with whom I cooperate.

Apart from this, I would advise you to consider the possibility of combining translation with another field of activity so that you have an alternative source of income if the demand for (human) translation does indeed become reduced during your active career lifetime. Disruption has already today had a significant effect in a several fields and who knows what new fields will be affected in the (unforeseeable) future.

I hope that my comments might give you food for thought but, in the final analysis, only you yourself can make your own career decisions.

Best,

Michael


Direct link Reply with quote
 

Christine Andersen  Identity Verified
Denmark
Local time: 01:23
Member (2003)
Danish to English
+ ...
You have to go for the top end of the market Mar 3

Whatever you do, keep up your studies besides translation.

I totally agree with Michael Davies' comments. MT is improving, but I am very sceptical about the idea that it will replace humans entirely. Artificial Intelligence is about as close to the real thing as a photograph is to a person. MT is based on probabilities and algorithms, and in a lot of standard situations human language is repetitive and predictable. Here it is possible to find what would be said in one language and replace it with what would be said in another. Machine translation is a game of probabilities, and it is possible to load the odds in by various means. However, machines can never actually understand the meaning of language,

There are still many situations where there are far too many factors for MT to take into consideration. Human life and language is just not predictable beyond a certain point. Patterns and probabilities are not precise enough for individual cases. Humans can see when they dealing with a typical case and when they are faced with a 2% minority, or whatever. Machines have trouble with gender in translations...

Cultural differences cannot be boiled down to formulae and algorithms.
What linguists call phatic communication - small-talk - is a challenge for MT. In some countries people talk about the weather, and talking about health, beyond 'I'm fine, thanks' is considered bad manners... Others happily discuss their digestive problems... Some British people hate 'I'm good', while for others it is perfectly normal... 'Have a nice day' or in Danish 'Continued nice day' are similarly normal for some and infuriating to others.

Between greeting and parting remarks, the chitchat goes on at an apparently meaningless level, but if you run it through Google Translate, the cultural elements go wrong. The whole function of chitchat is far too complicated for machines! A human who has lived in both countries can find suitable equivalents, depending on the purpose of the text and the target readers. It may be a challenge, but an experienced translator knows how to tackle it. Then there are all the other cultural issues...

Technical translation calls for real, human intelligence too. I wept with frustration over a terminology exercise which was part of my language diploma. It was about car parts, which you would expect to be quite prosaic. I know very little about car engines, and it taught me that I should definitely hand over anything like that to the experts. You would expect each part to have a name, with an equivalent in each language... Forget it! It depends on the make of the car: Japanese, German, Swedish and no doubt Italians, French, British and all the others... I never got that far... are all different. Cars are not manufactured in Denmark, but of course Danish mechanics and engineers maintain them and talk about them.

I would NEVER trust MT with a text like that.

I would not trust it in my own specialist fields - Law and Medicine - either.

The day machines take over marketing is far into the future and will probably never come. Texts have to be 'localised' or rewritten from scratch rather than simply translated. Nothing Sucks like Electrolux may be a myth, but it illustrates what happens when you translate the words, but not the meaning. There are plenty of true stories around about how machines bungle translation in marketing.

Even with MT, Post Editors and SEO experts will be needed for a long time. Both fields call for highly qualified humans with a solid knowledge of specialist subjects.

MT may be taking over the routine work, but there will still be a market for humans for the foreseeable future.


Direct link Reply with quote
 

Daniel Frisano
Monaco
Local time: 01:23
Member (2008)
English to Italian
+ ...
Why get a degree? Mar 3

First of all, you don't need a degree in translation. It won't find you clients, it won't teach you how to write properly in your target language, it won't teach you the subtleties of your source language(s), it won't make you an expert in any field, it won't help in all the tasks involved in self-employment.

That said, the future is murky. Translation won't be killed by machines, but it will certainly change. More and more post-editing will probably be required, and that's a pain in the patootie. With a growing mass of already translated material, there will be less request for translation from scratch and more adaptations and micro-jobs to update existing material. More pain in the aforementioned.

The good news is that passion, talent, ethics, willingness to work long hours, flexibility to change and other qualities can pave your way to success. Nothing is assured or granted though: it's a bet for which you are given no odds.


Direct link Reply with quote
 

Michael Beijer  Identity Verified
United Kingdom
Local time: 00:23
Member (2009)
Dutch to English
+ ...
Completely agree: a degree in translation is a waste of time and money Mar 4

Daniel Frisano wrote:

First of all, you don't need a degree in translation. It won't find you clients, it won't teach you how to write properly in your target language, it won't teach you the subtleties of your source language(s), it won't make you an expert in any field, it won't help in all the tasks involved in self-employment.

[snip]


Find a mentor (who is an actual translator), start with low-paid agency work, and work your way up. You can also do this while holding down another job, to keep paying the bills.

Regarding the main topic: I'm not really worried that machine translation is going to ever completely take over our work. However, it is definitely going to change a lot of things, so I do think it is advisable to try to start thinking about a Plan B, just in case. In my case, for example, I have been toying with the idea of learning how to code (in my spare time), since coding seems to be something that will definitely be around for another 20-40 years, whereas translation as we currently know it might've changed beyond recognition by then. However, I also think that if you are technically inclined, there will always be translation related work available, even if artificial intelligence, etc. takes over the bulk of repetitive technical/legal translation. For example, there will probably always be demand for language engineers, CAT tool specialists, etc. — basically all the people that handle the technical side of translation. I'm also pretty sure there will always be a need for translators who can translate extremely flowery/beautiful/etc. stuff, as I suspect machines will never fully master that type of thing.

evelyn_hero

Michael


Direct link Reply with quote
 

Mirko Mainardi  Identity Verified
Italy
Local time: 01:23
Member
English to Italian
+ ...
Specifically "trained" MT engines Mar 4

Christine Andersen wrote:

Technical translation calls for real, human intelligence too. I wept with frustration over a terminology exercise which was part of my language diploma. It was about car parts, which you would expect to be quite prosaic. I know very little about car engines, and it taught me that I should definitely hand over anything like that to the experts. You would expect each part to have a name, with an equivalent in each language... Forget it! It depends on the make of the car: Japanese, German, Swedish and no doubt Italians, French, British and all the others... I never got that far... are all different. Cars are not manufactured in Denmark, but of course Danish mechanics and engineers maintain them and talk about them.

I would NEVER trust MT with a text like that.

I would not trust it in my own specialist fields - Law and Medicine - either.


If MT were used in such cases, it would be with a MT engine "trained" with content for that specific client, including existing TUs and glossaries. Legal texts usually tend to be quite formulaic, so IMO they are also "at risk", although post-editing would still be an integral part of the process.

So, while I agree that MT should not be (en)trusted with highly sensitive content (and based on past cases I witnessed) I highly doubt this isn't already happening... (without the end-client necessarily knowing)

To the OP: the main risk I see in the near future is that the market share of post-editing will keep growing, eroding that of "human translation", so, if what 'fulfills you in life' and enjoy is translating, by putting your mind and heart into it in order to carefully craft something new, original and "beautiful", then you might be not feel as thrilled and fulfilled with PEMT...

So IMO you should focus on market segments and areas that will remain relatively MT-free (at least in the near future), particularly content that requires a creative flair or that is too sensitive to be routinely fed into a MT engine.


Direct link Reply with quote
 

Susan Welsh  Identity Verified
United States
Local time: 19:23
Member (2008)
Russian to English
+ ...
The Shallowness of Google Translate and other stories Mar 4

See this interesting article:
https://www.theatlantic.com/technology/archive/2018/01/the-shallowness-of-google-translate/551570/

It does not answer your career questions, but raises some of the key issues about MT.
For the reasons others have written above, I don't think MT is ever going to be able to cope with actual literature, rich with ambiguities, puns, metaphors.

Just try putting this sentence through online Google Translate to English (this is an academic article about Roman archeology, specifically digs where they found military equipment):

Bei den frühkaiserzeitlichen Ortbandscheiden sind die metallenen Scheidenbeschläge auf ein Mundblech am oberen Scheidenende, mit dem auch eine oder beide Tragezwingen verbunden sind, und ein sogenanntes Ortband im Spitzenbereich der Scheide beschränkt.

Have fun!


Direct link Reply with quote
 

Christine Andersen  Identity Verified
Denmark
Local time: 01:23
Member (2003)
Danish to English
+ ...
People will go to quite incredible lengths... Mar 4

... to convince themselves that one day, somehow, MT is going to be worth all the time and effort it takes to get acceptable results out of it.
I worked on pre-editing texts for machine processing in another life. Admittedly, computers have gone through quantum developments since then, but the principles and human effort are roughly comparable.

What makes PEMT so soul-destroying is the knowledge that a competent human can often produce better and/or faster results in the same time. Instead of simply reading the source, translating and checking, a post editor has to read the source and the target, check and assess... and then translate when the MT is not good enough. Many translators will tell you that the extra step takes longer, and I personally get a block after a short time - I miss errors, or when I see them, I can't think how to correct them.

I compare the process with a sculpture, where you start with a rough block of wood or stone, and chisel or chip away gradually, until the final details emerge. Mistakes cannot always be corrected if you chip too much away...

Translation is more like a water-colour painting. You have to get it right first time, and then you can perhaps highlight a few details with fine brushstrokes at the end.
It is not helpful to begin with a rough draft in MT and pick out the mistakes, just as it is difficult to clean water-colour away from the paper if you are not satisfied with your painting. In oils you can paint over mistakes, but it still takes time and effort...

OK, MT may be good enough for the 'here today and mulched tomorrow' kinds of texts. FAQs, chatbots and so on.

I mentioned earlier that MT has trouble with gender from one language to another. Translating from a language like English into a language with inflections and grammatical gender can challenge the algorithms to their limits.
Pronouns and possessives - he, she, it, his/her/its/their ...

Syntax comes over as unidiomatic, and MT can be very difficult to read. Negatives are sometimes left out, with disastrous results. If the translation reads well, there is still no guarantee that the meaning is really accurately rendered. Wherever this is important - and when is meaning NOT important??? - the translation has to be carefully checked against the source, and it all takes time and human effort.

There are still translators who make quite a good living without even using CAT tools. They can actually compete with those who do. I use my CAT, not always in the ways the designers expected, but all the translation that comes out is human, either mine or the work of trusted colleagues.

Machines are fine for mass-produced industrial products like drinking glasses, plates, chairs... perfectly flat window glass... but language is not that simple and it is not identical every time.

One day I hope we will learn that just as machines are better at some things, humans are better at others, and quality translation is something humans do best. Getting it right first time is really more efficient than all the effort that goes into editing and programming machines...

Translating is not a dirty job that damages your health like working with chemicals and lifting heavy loads. Build powerful, non-corrosive machines to do that kind of thing, and let humans do the jobs that are fun!

I have a nasty suspicion that IT nerds and programmers regard translation as a kind of grail that they must go after, and programming can be just as clean and fascinating as translation. So the IT guys are not likely to give in, just because they have not put us out of work yet.
That does not mean they are going to find their grail - they have been telling us for decades that within five years... and now things are really happening...

Translators must divide the market. Let the nerds and machines have what machines can be allowed to do, but distinguish and defend the areas where humans really are best. We are not machines, and nor are our clients. Make it a selling point!


Direct link Reply with quote
 

Sheila Wilson  Identity Verified
Spain
Local time: 00:23
Member (2007)
English
+ ...
Position yourself carefully for the future Mar 5

Christine Andersen wrote:
Translators must divide the market. Let the nerds and machines have what machines can be allowed to do, but distinguish and defend the areas where humans really are best. We are not machines, and nor are our clients. Make it a selling point!

It's been happening for some time now, but more translators must decide where they're aiming for:
- The translation industry, where MT (free - as in Google Translate - or trained) does the lion's share of the work. Sometimes humans will perform basic post-editing for a fraction of the cost of a market-rate human translation. Or they may even be assigned high volumes of words to translate with a deadline that forces them to do 1k+ per hour, leaving no time for checking or proofreading, for a stupidly low rate that leaves them no time to do any marketing so they're unable to lift themselves up and out of the downward spiral they're in.
- The translation profession, where experienced and/or qualified humans translate sensitive texts, texts in highly specialised areas, or texts which require a 'human touch', such as marketing, advertising, literature... These human translators know their worth and are able to convince enough clients of that worth.

Now, those of us who arrive in the world of translation as a second or third career can mostly manage without a high qualification in translation; indeed without any recognised qualification in some cases (e.g. if, like me, you started off at a time when fewer people got degrees, and anyway that was 40-odd years ago and you've done a lot since then). But the easiest way for inexperienced translators to differentiate themselves from translators working in the 'industry' is to get qualified. Either in translation or in a subject area, preferably both. And if the language pair is a common one (as is the case for EN DE), then specialisation is fundamental, IMHO. You absolutely have to reduce the number of peers (good, bad and indifferent) who are able to offer what you can offer, otherwise you have to compete solely on the basis of price. And anyone living in Germany is bound to lose that contest, I'm afraid .


Direct link Reply with quote
 

Michael Wetzel  Identity Verified
Germany
Local time: 01:23
German to English
To repeat what Christine said ... Mar 6

(Edited to display quotation correctly)

Andrew Dirr wrote:

The feedback I receive from the research network's professors and postdocs (English and German speakers alike) is also very positive. (...)

As if the universe wanted to give me a further sign, a good German friend of mine has also recently approached me with a very large work of literature. (...)

So it was much to my dismay that I stumbled upon DeepL. I ran a few texts through it and found that it did quite a good job on them.


Social science is a gigantic field that ranges from statistical analysis to Sloterdijk. I suppose DeepL or neuro-Google-uber-plex might produce halfway plausible results in a few areas ...

But are you really saying that you fed it with paragraphs (not sentences) of academic papers or literature and it spit out results that were consistently reasonably accurate and convincing and - what seems most unlikely to me - well-written enough that someone could effectively read and retain (and, at least in the case of literature, enjoy) hundreds and hundreds of pages of it week in and week out?

Take another look at the results: Aren't you being too generous?

PS: And I would certainly suggest studying something, even if you don't get a degree: It's free and enjoyable and it's an amazing way to find clients (as you've already noticed).

[Edited at 2018-03-06 10:56 GMT]


Direct link Reply with quote
 

neilmac  Identity Verified
Spain
Local time: 01:23
Spanish to English
+ ...
Go for it Mar 6

You obviously have such a a passion for translation that I think you should definitely concentrate on that route. There are so many lackadaisical number crunchers out there, or market-busting intruders who got into translation because they thought it was easy, that someone as inspired as you seem to be is a refreshing addition to our ranks. And you seem to move comfortably through the academic world, so if you took a translation course you would probably make some useful contacts and perhaps get work that way later.

And don't worry about MT. As a a language professional, you can use it as a tool to be leveraged rather than perceived as a threat.


Direct link Reply with quote
 

Alexandre Chetrite
France
Local time: 01:23
English to French
Human language only appeals to other humans (not machines) Mar 7

Hi,

I would like to say something really odd (maybe) and simple, yet maybe we miss the simple stuff here.

MT will never replace human translation for a simple reason: humans invented language to communicate with other humans (only).

"Obviously!" one may say. "Of course!" another will say.

But think about it. Who else is better suited to grasp the infinite patterns of language inflexions and nuances than...a human!

A machine doesn't "understand" emotions (at least not yet). A machine "interpretates" human emotions and tries to simulate human understanding.

But the difference between "virtual imitation" (machines) and real emotions (humans), is like having a real luxury brand bag in front of you and a Chinese imitation (sorry for the Chinese, I meant no disrespect to them).

Another abstraction might be too say that machines apply a filter to understand inflexions in languages , whereas human translators understand them right away, because it is built in our cognitive mind and functioning.

A filter is a additional step that makes Machine translation more difficult in a way.

Thereofre it is simple->

Language = Human-to-Human. Not Human-to-Machine or Machine-to-Human.

MT will barely be able to mimic the great and basic emotions of humans in poetry and language (hate, love, good, bad, etc).

But as for the inifinite patterns that I talked about in the beginning, there are as many variations per individual as there are stars in the Universe.

Therefore no MT will ever be able to translate "perfectly" the meaning of a complex text to another language, in my opinion.I may be wrong, but only future will tell (and AI).


Direct link Reply with quote
 

JaneD  Identity Verified
Sweden
Local time: 01:23
Member (2009)
Swedish to English
+ ...
Wonderful metaphor! And some advice Mar 7

Christine Andersen wrote:

I compare the process with a sculpture, where you start with a rough block of wood or stone, and chisel or chip away gradually, until the final details emerge. Mistakes cannot always be corrected if you chip too much away...


That's a brilliant way to put it, Christine, and I'm going to pinch that for next time I need an argument against PEMT!

As for the OP's question - you enjoy translation, so go ahead and do it. But I agree with other posters, don't bother getting a degree in it. Get a degree or other training in something else, instead. Refine your writing, go back to grammatical basics in case you've always used the semi-colon incorrectly without knowing it. Learn about other subject areas. Those are the kinds of training that are useful to translators, in my opinion (but then my degrees are in archaeology rather than translation, so perhaps I'm biased!)


Direct link Reply with quote
 

Merab Dekano  Identity Verified
Spain
Member (2014)
English to Spanish
+ ...
My opinion Mar 7

For some fields that are terminology-driven MT is already the rule (such as weather forecasts, etc.).

For fields where the actual meaning gone wrong can cost millions, if not billions, of dollars, humans will have the last word (such as law-related texts).

Whatever falls within “irrational” thinking will have to be produced and translated by humans (such as poetry, etc.). If machines can do it, they will take over and exterminate us (not just take away our jobs).

Since you cannot be a translator, but rather a translator in certain field(s), it will very much depend on your choice/capability/interests.


Direct link Reply with quote
 

Samuel Murray  Identity Verified
Netherlands
Local time: 01:23
Member (2006)
English to Afrikaans
+ ...
@Andrew Mar 7

Andrew Dirr wrote:
I just want to be sure I am not entering a "dying field" where there will be literally nothing left to do in the near future.


I don't think the rise of MT will have much of an effect on the demand for traditional translations. The trend in free MT seems to be to produce better-sounding translations, not more accurate translations. The only difficulty you'll face more often is convincing potential clients that just because the Google translation *sounds* very good doesn't mean it's actually very good.

In fact, you may find that future improvements in MT might even make you more productive, eventually.

There is a risk that more clients will want you to simply proofread machine translations, but if you're happy to do that and you quote a reasonable fee for it (reasonable for you), then that shouldn't be a problem, since the final product will still depend on your skill and creativity.


Direct link Reply with quote
 
Pages in topic:   [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 »

Is translation a field I can work in despite the rise of MT?

Advanced search






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

More info »
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 »



Forums
  • All of ProZ.com
  • Term search
  • Jobs
  • Forums
  • Multiple search