| Pages in topic: < [1 2 3 4 5 6] > | How long will human translation last? Thread poster: LucyPatterso (X)
|
|---|
| Human translation will last as long as human race. | Oct 24, 2012 |
MT is not overrated -- it is totally inappropriate for professional translation. Some people who have nothing to do with translation get the wrong idea that MT is an alternative to human translation, and they also have another crazy idea about Wikipedia -- that its various articles were machine translated. This is according to some research I did. A few people told me: "Oh, Machine Translation is really great -- look at Wikipedia. You choose the translation on the left-hand side and here it com... See more MT is not overrated -- it is totally inappropriate for professional translation. Some people who have nothing to do with translation get the wrong idea that MT is an alternative to human translation, and they also have another crazy idea about Wikipedia -- that its various articles were machine translated. This is according to some research I did. A few people told me: "Oh, Machine Translation is really great -- look at Wikipedia. You choose the translation on the left-hand side and here it comes -- a perfect translation -- into so many languages, instantaneous. They don't realize that this is not MT, but hundreds of professional linguists have toiled for days, trying to translate the texts to the highest quality human translation standards. These are just multilingual versions of the articles -- 100% human (top of the profession) translation. It has nothing to do with MT, or translation programs. I just wanted to point it out, because I really got shocked with the ideas some people have about Wikipedia's multilingual articles. They had the impression that it was translated instantaneously through Google, or a different program. Some articles have even been written from scratch in another language -- not translated. Machine translation could work, if you put a human in the machine.
[Edited at 2012-10-24 13:00 GMT] ▲ Collapse
| | | | | Not worried at all | Oct 24, 2012 |
If you look closely at it, MT is currently so mediocre that no translator should feel threatened by it. The best example would be GT, which has been around for years now. Do you feel like it is getting more intelligent? Would you blindly trust it, even a for a 3 words sentence? Well, it might be somewhat competent when it comes to a simple sentence with a basic structure but definitely not "intelligent".
The day computers will be able to detect and use genuine forms of humor, sarcasm or s... See more If you look closely at it, MT is currently so mediocre that no translator should feel threatened by it. The best example would be GT, which has been around for years now. Do you feel like it is getting more intelligent? Would you blindly trust it, even a for a 3 words sentence? Well, it might be somewhat competent when it comes to a simple sentence with a basic structure but definitely not "intelligent".
The day computers will be able to detect and use genuine forms of humor, sarcasm or subtlety in a text will be the day your career might be threatened. With such brilliant technology, it might also be the day machines decide that they are going to rule the world and get rid of human beings. Killer lawnmowers, psychotic blenders and crazy electric toothbrushes... that should answer the question "How long will human (translation) last?".
On a more serious note, you need a real brain to translate, and today, we don't even really know how the brain works. Video games developers still struggle with elementary tasks like pathfinding (plotting by a computer of the shortest route between two points). MT has existed and has been available online for almost 10 years now, (I think it started with Babel Fish back in 2003) and it is still as clumsy and brainless as it was the first day.
If people are willing to use this system in a "professional" way, let them. They're digging their own grave and are showing the world how poorly they are running their business. Would you really like to deal with them, anyway? ▲ Collapse
| | | | | As long as I'm alive to scream | Oct 24, 2012 |
About the c..p content (supplied by a 3rd party) linked to from my in-house company's professionally, carefully and tenderly translated translated/transcreated site.
Not sure if it was the split compounds, incorrect word order or the odd bit of Norwegian/Danish that made me think the 3rd party might have outsourced to Mr GT...
| | | | inkweaver 
Germany
Local time: 09:49
French to German
+ ...
| My thoughts exactly | Oct 24, 2012 |
Nicolas Machado wrote:
If you look closely at it, MT is currently so mediocre that no translator should feel threatened by it. The best example would be GT, which has been around for years now. Do you feel like it is getting more intelligent? Would you blindly trust it, even a for a 3 words sentence? Well, it might be somewhat competent when it comes to a simple sentence with a basic structure but definitely not "intelligent".
As far as I'm concerned, MT is good for having a laugh now and then when you have nothing better to do.
Relying on it could be downright dangerous, since GT was not able to translate the Italian word for never a couple of years ago and simply omitted it in the translation, so a machine translated manual would have advised the user to immerse their electrical appliance in water.
An Italian horticulture business's website would have read like a bunch of assassins offering their services. Part of the company's services was felling unsound trees and disposing of them, according to GT they were offering to kill people and dispose of them.
As long as MT is not even able to translate relatively straightforward texts in fairly common language pairs I don't see any reason to be worried.
| | |
|
|
|
Shai Navé
Israel
Local time: 10:49
English to Hebrew
+ ...
| Don't confuse generic online MT with "trained" offline MT solutions | Oct 25, 2012 |
The various free online MT engines are just one type of MT, and a generic one.
For a couple of decades now there are other "offline" MT solutions that work well in certain controlled environments.
There are valid points for and against the use of MT, which I won't go into right now and that could be discussed in length, but the main thing that bothers me in the entire "MT good or bad?" discussion is the lack of context. MT, as CAT tools, is a just a tool and nothing more; but people,... See more The various free online MT engines are just one type of MT, and a generic one.
For a couple of decades now there are other "offline" MT solutions that work well in certain controlled environments.
There are valid points for and against the use of MT, which I won't go into right now and that could be discussed in length, but the main thing that bothers me in the entire "MT good or bad?" discussion is the lack of context. MT, as CAT tools, is a just a tool and nothing more; but people, even translation professionals, portray it directly or indirectly (by opposing it for the "wrong" reasons) as a professional equivalent substitute.
Mocking the quality of the readily available online generic solutions is easy, but some will stare with disbelief at what some trained MT engines can produce in certain controlled environments. However, even those MT solutions are only as good as the data put into them and the tweaks made to their algorithms with time (hence their better performance in a specific controlled environment). All that training is not cheap nor trivial, and most importantly, relays on constant human input.
Therefore, it is vital to understand the context of the various MT solutions, and that responsibility lies first and foremost with the language service providers. They should understand that MT is a tool and not a substitute. Even when one chooses to use it because it makes sense in his/hers specific setting, it doesn't mean that the rates (the "per hour rate") should (automatically) be lowered and/or other unfounded terms and conditions now apply.
MT is an emerging technology and we should all get more familiar with it, mainly in order to judge whether we want to use it, and if so if it can be useful to us, or not. We should get familiar with it to understand its advantages (if any) and disadvantages, just like professionals do with any any other tool of the trade. We should educate ourselves for our sake. The less educated professionals are about certain tools and technologies, the more easy it would be for others to spread false claims, skew the perception towards their own interests and "convince" (usually be spreading fear and disinformation, as could already be evident) professionals to work into their disadvantage.
[Edited at 2012-10-25 00:53 GMT] ▲ Collapse
| | | | inkweaver
Germany
Local time: 09:49
French to German
+ ...
| I can only judge what I have access to | Oct 25, 2012 |
And since I do not have access to the type of MT described in the previous post, I cannot judge it.
However, I am convinced that nothing compares to the human brain when it comes to dealing with language and all its subtleties. MT may work when it comes to certain repetitive phrases but fortunately the types of text I usually deal with are quite different.
So even at the risk of appearing "uneducated" about all the more or less wonderful possibilities of MT I don't see... See more And since I do not have access to the type of MT described in the previous post, I cannot judge it.
However, I am convinced that nothing compares to the human brain when it comes to dealing with language and all its subtleties. MT may work when it comes to certain repetitive phrases but fortunately the types of text I usually deal with are quite different.
So even at the risk of appearing "uneducated" about all the more or less wonderful possibilities of MT I don't see it as an option for my work. I have edited a couple of MT translated texts in the past (for a fee about 80 % of my usual translation rate) and I found it incredibly dull and uninspiring and I hope I won't be asked to do it again (it was for one of my best clients, otherwise I would have declined).
Anyway, MT is something I will happily leave to others to use. As far as I know they are quite a few colleagues around who don't use any CATs and they seem to be doing well so I guess there is room and demand for everything. ▲ Collapse
| | | | Shai Navé
Israel
Local time: 10:49
English to Hebrew
+ ...
| Generic post MT editing is | Oct 25, 2012 |
Post generic MT editing is indeed dull, and I personally oppose it because I don't see any professional value in it. The only "value" associated with it probably for those who push it only to reduce their costs.
For example, what if you would have your own trained MT engine? That is, a system that contain your previous translations, TMs, Glossaries, and even has some ability to "learn" your style and preferences in certain scenarios as result of you "training" it with use (think alo... See more Post generic MT editing is indeed dull, and I personally oppose it because I don't see any professional value in it. The only "value" associated with it probably for those who push it only to reduce their costs.
For example, what if you would have your own trained MT engine? That is, a system that contain your previous translations, TMs, Glossaries, and even has some ability to "learn" your style and preferences in certain scenarios as result of you "training" it with use (think along the lines of predictive text systems). Some kind of an hybrid between a CAT tool and a language engine. This has the potential to increase one's productivity by allowing one to get most of the "typing" out of the way, while still getting a relatively quality text, and focus more on improving the content and the language rather on typing or structuring "basic" sentences. Of course, different scenarios have different considerations and different professionals have different approaches and methods, and what I just described won't work for literature for instance and not everyone would find it to their liking, but for some it could prove valuable in their setting.
But all of that is true when training our own systems and does not apply to cloud based systems or any other type of "shared" system. What is the point in investing time and effort in "training" such a system, expending its database and help improving its performance, and then just voluntarily hand over all that work to a third-party to capitalize on it?
I personally don't find any value in MT in its current state, and strongly against post MT editing of any text that is not a product of one's local MT owned solution. I also know that building an "offline" MT engine and training it is not a matter of couple clicks of a button and it takes time, money and knowledge. The technology is still not mature enough to be run locally by just any individual.
I want to believe that I will never use MT because I prefer to keep using my voice and style and not replace them with something synthetic (even when using my past work), but I'm quite aware that it is quite possible that in due time the technology would mature and actually become useful as a tool (not as a human replacement), maybe even in ways that we can't imagine at this point in time. I also know, well, its already happening, that some stakeholders will try to use MT, even in its current relatively immature state, as yet another mechanism to reduce rates and otherwise skew the market in their favor. This is why I think we should educate ourselves in order to protect ourselves and our profession from such conduct. By the time MT will become more mature and possibly useful (again, as a tool), it might be too late, and too many misconceptions, de-facto standards and other questionable practices would already be a reality; and to me this is the main issue and risk. From a professional standpoint I'm not concerned about a machine replacing us, I am much more concerned about the lowering of the standards (when poor to mediocre quality become acceptable) and by professionals falling into whatever ploy some stakeholder might throw at them. ▲ Collapse
| | | | Heinrich Pesch
Finland
Local time: 10:49
Member (2003)
Finnish to German
+ ...
| Wikipedia is not translated at all | Oct 25, 2012 |
LilianBoland wrote:
MT is not overrated -- it is totally inappropriate for professional translation. Some people who have nothing to do with translation get the wrong idea that MT is an alternative to human translation, and they also have another crazy idea about Wikipedia -- that its various articles were machine translated. This is according to some research I did. A few people told me: "Oh, Machine Translation is really great -- look at Wikipedia. You choose the translation on the left-hand side and here it comes -- a perfect translation -- into so many languages, instantaneous. They don't realize that this is not MT, but hundreds of professional linguists have toiled for days, trying to translate the texts to the highest quality human translation standards. These are just multilingual versions of the articles -- 100% human (top of the profession) translation. It has nothing to do with MT, or translation programs. I just wanted to point it out, because I really got shocked with the ideas some people have about Wikipedia's multilingual articles. They had the impression that it was translated instantaneously through Google, or a different program. Some articles have even been written from scratch in another language -- not translated. Machine translation could work, if you put a human in the machine.
When you look at different language versions, you'll notice that almost each has been written from scratch and reflects the interest level of the subject for the specific language and the devotion of the authors to the subject.
But really interesting news, that people might think Wikipedia is MT-generated. Haha.
| | |
|
|
|
Phil Hand
China
Local time: 15:49
Chinese to English
| What is this obsession with (not) typing? | Oct 25, 2012 |
Shai Nave wrote:
...to increase one's productivity by allowing one to get most of the "typing" out of the way...
We translate maybe 3000 words per day. At a normal, untrained typing speed (i.e. mine), that's one hour's work. A little more than 10% of the day, if that. And of course, it's not the case that my typing time is dead time. I think and type at the same time.
If you don't want to type, get Dragon.
This weird argument that MT reduces typing just shows what a crazy idea it is. If the best that can be said about an MT system is that it does the job of speech recognition software, then we're really scraping the barrel.
| | | | | Typing is a pleasure, and MT has just a purely entertaining use. | Oct 25, 2012 |
You can actually just waste a lot of time using MT for any professional purposes because you have to retranslate everything from scratch. I think MT programers should concentrate more on
great dictionaries, phrase books, glossaries and things like that, rather than on the impossible -- trying to make machines make intelligent decisions and translate. In my opinion, it is really like trying to create gold out of sand.
| | | | Shai Navé
Israel
Local time: 10:49
English to Hebrew
+ ...
| Some clarifications | Oct 25, 2012 |
1. There is no obsession about typing.
2. I used the term "Typing" to describe all the mundane, repetitive tasks involved with the translation creation process, hence the double quotes.
Trained and customized MT solution is far different from the online generic ones. Speaking about the technology, including potential benefits and disadvantages, is important. Burying our heads in the sand will only hurt us in the long run, and not because we should all adopt it at some point, b... See more 1. There is no obsession about typing.
2. I used the term "Typing" to describe all the mundane, repetitive tasks involved with the translation creation process, hence the double quotes.
Trained and customized MT solution is far different from the online generic ones. Speaking about the technology, including potential benefits and disadvantages, is important. Burying our heads in the sand will only hurt us in the long run, and not because we should all adopt it at some point, but rather because being uneducated about a technology or tool that exists in one's market puts one in disadvantage.
Oppose it if you will (I know I do), but do it out of an educated position and for the right reasons. Those who dismiss it because "it will never be able to replace a human" or out of fear of "being replaced" might be right, but they ignore some issues and points that make up the bigger picture, and even worse, by doing so out of fear without understanding what this technology is about, they just help to perpetuate the message that MT is indeed a human replacement and now translators are fighting to keep their jobs (to the delight of those who work hard to implement that idea; and sadly, some translators actually think that).
[Edited at 2012-10-25 11:21 GMT] ▲ Collapse
| | | | | It is not the fear at all -- there are thousands of other interesting things to do, | Oct 25, 2012 |
if translation were to be taken over by the machines. It is just not going to happen -- perhaps in science fiction movies like "The Trucks." I am just worried that many disasters may happen because of MT, such as accidents caused by improperly functioning machines assembled based on MT translated manuals, people may die because of wrongly translated medical research materials, drug leaflets, etc. When machines will publish their first books, then I will believe that MT is really something that c... See more if translation were to be taken over by the machines. It is just not going to happen -- perhaps in science fiction movies like "The Trucks." I am just worried that many disasters may happen because of MT, such as accidents caused by improperly functioning machines assembled based on MT translated manuals, people may die because of wrongly translated medical research materials, drug leaflets, etc. When machines will publish their first books, then I will believe that MT is really something that can be treated seriously. ▲ Collapse
| | |
|
|
|
Translating is becoming more like banking and finance, nobody really knows how it works. Once upon a time you knew the bank manager and the translator, personally or by reputation, now everything's done in a flash by mirrors and the latest quick fix.
Our local Red Cross shop accepts books, other than dictionaries. Does that tell us anything?
| | | | Phil Hand
China
Local time: 15:49
Chinese to English
| I agree with this part... | Oct 25, 2012 |
Shai Nave wrote:
Speaking about the technology, including potential benefits and disadvantages, is important. Burying our heads in the sand will only hurt us in the long run, and not because we should all adopt it at some point, but rather because being uneducated about a technology or tool that exists in one's market puts one in disadvantage.
I used the word "obsession", Shai, because it seems like every time we talk about MT this is the argument that comes up. And it's a really stupid one.
I completely agree that we should talk about the technology and what its uses, advantages and drawbacks might be. I think that that conversation is seriously impeded at this moment in history by two things:
1) The fact that whenever we start talking about MT, several voices will chime in with how great it is to cut out all the typing.
2) The fact that the current dominant MT working model is "post-editing", which is the worst idea in the history of translation.
There could be uses for MT, though I struggle to see them at the moment. I would prefer, as I said in a recent post (http://www.proz.com/forum/machine_translation_mt/232996-software_misnomers_and_how_to_use_mt.html) to have a tool that gave me multiple dictionary look-up.
If you have more ideas, please share them. What can MT do that is actually useful? I.e. not saving on typing (the job of speech recognition programs); or consistency (the job of CAT programs); or translating for post-editing (the job of monkeys).
It can give approximate meanings in non-critical texts. That's a useful thing. Past that, I'm drawing a blank.
| | | | | Create great, multiple dictionaries, phrase books with real speech samples | Oct 25, 2012 |
Create specialized dictionaries, easy to access -- legal, medical, technical. Suggest declensional forms, and other grammatical forms, which may be very useful for many students of foreign languages, and some translators as well, especially if they translate into a second language. (conjugators, and things like that.) There are many things technology can help with.
| | | | | Pages in topic: < [1 2 3 4 5 6] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » How long will human translation last? | 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 » |
| | 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
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
