| Pages in topic: < [1 2 3 4 5 6] > | How long will human translation last? Thread poster: LucyPatterso (X)
|
|---|
Shai Navé 
Israel
Local time: 04:45
English to Hebrew
+ ...
1. The argument about typing comes up every time because it is valid. As I've said, by using the word typing people, mostly, refer to the mundane, dull and repetitive tasks that are part of the translation process, and not just the typing itself. Speech recognition is a good and efficient tool (though in current state not for every language), but:
A. It is not for everyone (due to the technical aspects or just preference).
B. It still doesn't help you to construct the sentences that ... See more 1. The argument about typing comes up every time because it is valid. As I've said, by using the word typing people, mostly, refer to the mundane, dull and repetitive tasks that are part of the translation process, and not just the typing itself. Speech recognition is a good and efficient tool (though in current state not for every language), but:
A. It is not for everyone (due to the technical aspects or just preference).
B. It still doesn't help you to construct the sentences that you speak to it.
Also, you may consider typing (the physical task) negligible, others may think differently due to their lack of speed, health concerns, or just because they don't like to type.
I do agree, however, that claiming that the major benefit of MT is a reduced typing effort is wrong, because this is not its main function, and those who just look for that can find other solutions, mainly speech recognition.
2. I oppose any MT post editing that is not done on a text that was created by one's owned MT environment (i.e. text that was created by the MT environments of others). This is one of the bad practices that try to creep in.
3. Forget about the generic online MT engines (which also pose a confidentiality issue), and even generic offline ones, they are almost irrelevant in the greater professional MT scheme. The mere fact that they are being used as an MT benchmark of sorts is misleading, because there are other MT systems there, and they can be useful, at least in theory. When I speak about potentially useful MT solution I mean an offline customized solution that each professional owns and maintains. This means that you put all your TMs, glossaries and past work into it, and while editing the first generic results you "teach" the solution your style. Think about the great progress that text prediction softwares on mobile devices have made in the last couple of years, from generic not very useful ones, to a solution that analyses your input and identify patterns that you can actually use. Today, some of these solutions are a real time saver. So the offline MT environment is an hybrid between a CAT tool and a "text prediction" component. When you use it you won't just get a generic wall of MT generated text, but rather (with time and work) a text that feels much more like something that you would write, that would be a lot easier to edit, and yes, that has the potential to increase your productivity by saving typing and structuring of basic, dull sentences that in one form or another you have already translated thousands of times before.
Granted, this won't work for anyone or anything. It won't work for literature or highly creative or otherwise complex texts; it will work better for specialized professionals that work in specific fields and less for the Jack-and-Jills of all trades that seem to work on whatever is thrown their way; it will take some technical skill to setup and maintain so it's not for everyone, and so forth. So yes, it won't be useful to all, but for other it could be. Dismissing the technology altogether based on the online generic engines is a bit avoiding the issue.
This is how an MT solution can be potentially useful. This doesn't mean that it should be adopted by everyone or used for any thing, and I strongly recommend against working for any cloud based or otherwise shared system of this type, but the potential benefits are there, and dismissing them, thus limiting the discussion (the eventually suppose to help each and every professional to come to an educated decision) is a shame.
Just to be clear, at this current time point I personally don't intend to use MT anytime soon. I don't see it as mature enough to be really useful, and I think that it is used solely as another mechanism to exploit naive translators and to reduce rates (and quality). I also like to think that I won't use MT ever because I take pride in my work, even in structuring "dull" sentences. However, I acknowledge that if the technology would mature to something along the line that I have described above - to a customized MT environment - it can be useful; not to everyone and not for any thing, but in general it has some merits.
So I'm really not advocating to start using it (you can search the forum to read previous posts that I wrote on the subject and I have voiced my opinion int Proz's last virtual event on this subject).
What I'm certain about though, is that avoiding the issue and dismissing MT altogether just because the generic MT solutions try to mimic the human mind is a mistake. We should discuss the technology and learn about it because it is here. Otherwise, others will exploit that lack of education and knowledge (we can refer to it the "collective naiveness") and spin that to their favor. ▲ Collapse
| | | | | Voice recognition -- I agree. It can be perfected one day and make | Oct 26, 2012 |
typing just an option to use, or not to use. However, in the modern culture of internet fora, blogs and texting is it really that important? Some people can type faster than they can speak. Some may have to take diction classes to be able to use the program. It is possible to develop it to quite high standards, I think, because there is really nothing intelligent about it. What I mean is the computer does not have to think or make that many decisions -- it basically sound recording in a written ... See more typing just an option to use, or not to use. However, in the modern culture of internet fora, blogs and texting is it really that important? Some people can type faster than they can speak. Some may have to take diction classes to be able to use the program. It is possible to develop it to quite high standards, I think, because there is really nothing intelligent about it. What I mean is the computer does not have to think or make that many decisions -- it basically sound recording in a written form. If you mean programs that would interpret speech into another language -- I am sorry, but this is just another pipe dream. It is hard enough for a human to interpret sometimes, not to mention poor machines. ▲ Collapse
| | | | Shai Navé
Israel
Local time: 04:45
English to Hebrew
+ ...
| Technology is a tool | Oct 26, 2012 |
LilianBoland wrote:
typing just an option to use, or not to use. However, in the modern culture of internet fora, blogs and texting is it really that important?
Yes it is, because some find it useful, otherwise the technology would not have been in demand. Mind you, there are other fields/uses for a technology outside the translation, or any other specific, industry.
It is possible to develop it to quite high standards, I think, because there is really nothing intelligent about it...the computer does not have to think or make that many decisions -- it basically sound recording in a written form
I disagree; it might seem that way (just take an audio input and convert it to text), but it is not that simple from a technical standpoint. If it was so simple, the technology would have advanced more or matured a long time ago. Thinking that the voice recognition technology is a simple process that merely turns A into B in unintelligent way is no different than what people outside of the translation field think about translation, although we all know that in reality there is nothing furthest from the truth. Personally, my first rule of business is to practice what I "preach". I never disrespect or dismiss something just because I don't know about/understand it or don't find it useful for my needs. Every profession and every technology require some degree of skill, knowledge and effort, and that should be respected.
If you mean programs that would interpret speech into another language -- I am sorry, but this is just another pipe dream. It is hard enough for a human to interpret sometimes, not to mention poor machines.
You keep repeating the point of "machines will never replace humans" and therefore we should not care about any of this. While you are right about the fact that technology cannot replace humans, I think that dismissing the importance and impact of technology is approaching the subject from the wrong angle. Technology is a tool, it cannot replace humans in skilled professions, but it can help make the work more accurate and efficient, sometimes it even enables things that otherwise were impossible.
For example, an airplane is nothing without a human pilot and crew, there are technological advances there all the time and no one thinks that it will be used to replace the flight crew. The technology is just another tool to make the life of the crew easier and the work more efficient, accurate (and safe, and so on).
So stirring the conversation to the never ending discussion about how machine cannot replace humans is counter productive if you ask me, is the wrong approach. It doesn't offer any real value, and If anything, it only helps strengthening that idea in the mind of others (if it is discussed so fiercely, there must be something to it, thinks the average layman) from outside and within the translation industry. We should talk about the technology in terms of efficiency, workflow, fit for purpose, usefulness and shortcomings, and not dismiss it just because it seems unimpressive in a narrow, localized context; given the right technology-human tandem, great things can happen.
| | | | | I fully agree with Shai | Oct 26, 2012 |
While we turn around the machine-will-never-replace-humans thing ad infinitum, companies and translation agencies are experimenting with and implementing MT workflows to their own and sole benefit, with more and more relative success cases being reported. Cost reduction and/or profit increase are the big drivers for R&D here as in any other industry. I fear that, as long as those companies and the MT developers keep the initiative, we might be more and more faced to accomplished facts as regards... See more While we turn around the machine-will-never-replace-humans thing ad infinitum, companies and translation agencies are experimenting with and implementing MT workflows to their own and sole benefit, with more and more relative success cases being reported. Cost reduction and/or profit increase are the big drivers for R&D here as in any other industry. I fear that, as long as those companies and the MT developers keep the initiative, we might be more and more faced to accomplished facts as regards our position in this industry. There is a world beyond Google Translate and similar generalist and publicly available MT platforms, and the least we can do is keep ourselves well informed and see what MT can do for us, for our profession, and how it can help us to improve the service we offer to our customers. After all, we professional translators are in the best position to get the most from good customizable MT packages and any other computer assisted translation tools without compromising effective communication and language quality. MT developers would be wise to keep this in mind.
[Edited at 2012-10-26 16:25 GMT] ▲ Collapse
| | |
|
|
|
Zhihua Liu
China
Local time: 09:45
English to Chinese
+ ...
| World evolves | Oct 28, 2012 |
The world evolves and so do the translators. As long as there's a need for translation, there's need for human translator, just a little diffrence in the way they work.
| | | | | MT helps translators | Nov 26, 2012 |
I think it's quite clear that MT is not going to replace human translators. But MT definitely can be used as a tool that helps a translator to work faster. And that obviously should be good and very interesting for most translators.
If using MT would help a translator to work e.g. 10% faster (without compromising quality) why would you not use it? Or if MT helped to work 20% faster? Etc.
Niko
Multilizer
... See more I think it's quite clear that MT is not going to replace human translators. But MT definitely can be used as a tool that helps a translator to work faster. And that obviously should be good and very interesting for most translators.
If using MT would help a translator to work e.g. 10% faster (without compromising quality) why would you not use it? Or if MT helped to work 20% faster? Etc.
Niko
Multilizer
http://mt-quality.multilizer.com ▲ Collapse
| | | | Ali Alsaqqa
United States
Local time: 21:45
English to Arabic
| Interesting. | Nov 26, 2012 |
Sergei Tumanov wrote:
qte
someone was saying "machine translations are getting quite good now".
uqte
I'd rather say
that technical writing is getting quite poor now....
And this is the main reason for the machine translation success in this field.
[Edited at 2012-09-17 08:36 GMT]
This is an interesting response, indeed!
| | | | |
|
|
|
Tom in London
United Kingdom
Local time: 02:45
Member (2008)
Italian to English
| I agree - especially---- | Nov 26, 2012 |
Multilizer wrote:
I think it's quite clear that MT is not going to replace human translators. But MT definitely can be used as a tool that helps a translator to work faster.
But does it? 90% of these forums are cries for help from people who are having problems with CAT tools and losing time because of them.
| | | | | More comments about translating faster with MT | Nov 27, 2012 |
Tom in London wrote: Multilizer wrote:
I think it's quite clear that MT is not going to replace human translators. But MT definitely can be used as a tool that helps a translator to work faster. But does it? 90% of these forums are cries for help from people who are having problems with CAT tools and losing time because of them.
I think this excellent comment from yesterday answers Tom's question well:
Cristóbal del Río Faura wrote:
While we turn around the machine-will-never-replace-humans thing ad infinitum, companies and translation agencies are experimenting with and implementing MT workflows to their own and sole benefit, with more and more relative success cases being reported.[Edited at 2012-10-26 16:25 GMT]
I think using MT to speed translation is getting more and more common, and it benefits translators. Probably the best ways to utilize MT are still in development but MT already helps. I know of several methods currently in use:
* customizing MT: this is getting easier and easier. I know of one translator who did this himself and gained >10% better translation speed
* using general-purpose MT for pre-translating easy texts: it certainly helps, the problem is to know which texts are easy enough (actually the share is quite small currently)
* our own clients are experimenting with our technology that detects good MT that should be post-edited (and omitting bad MT that is better to translate from scratch)
All of the above methods are known to speed up translation, at least in some cases. I think the challenge is to learn which methods should be used, when and how. I am also certain that MT does currently not help in every situation (e.g. translating literature).
In general, a technology that helps translators to work faster certainly should be very beneficial to translators.
Niko
MULTILIZER
http://mt-quality.multilizer.com
| | | | apk12
Germany
Local time: 03:45
English to German
+ ...
"machine translations are getting quite good now"
yeah...
MT is absolutely needed for a good laugh in a coffee break, I would miss it.
If you want to know how long human translation will last: as long as the people will be curious what the newspapers in that other country are -really- saying. There are, and will always be, phrases, idioms, connotations, reference jokes, hints, that no MT will be ever able to get a grasp of. Language develops fast. As news are.
And of course, as long as companies prefer to address clients concerned about quality, they also will always prefer to work with a human, experienced translator. (If they want to avoid a place on the list of the Most Hilarious MT PR Cases Last Year.) (Somebody should start a project with that, a kind of an award, given to the best, most hilarious MT "PR", "Advertisement" and "Web Copy" cases).
Something following the example of the "Goldenes Brett" ("Blockhead of the Year"), an award given to the biggest BS of the year in science (the most hilarious theories, stupidities, in science but also in media). The continuously updated list of the nominees is here (it includes also Thilo Sarrazin), and the winner last year was somebody who, according to the paper of one of his 'students', has absolutely no idea how a student paper should look like (but well, in Germany, we have a couple of such cases, just remember the Guttenberg case). Anyway, this paper was also hilarious, simply hilarious. Not a plagiarism case this time, but ... in German, we say in such a moment: "ohne Worte". There are simply no words to discribe that. It was about a tube that according to the "original inventor" makes a lot of funny stuff possible (contact to ghosts, UFOs, time travel and simply the absolute blast), which the student took and made a "model" of, he "minimized it". Then he made "tests" (you do not need any methodology in such areas, so why should you actually bother about that, must have been obviously what the student was thinking) and the model "worked". "Somehow". The text of the "paper" was... it was just... - I think here, just a quotation will help:
"Am ehesten umschreiben kann man es vielleicht mit: Aufladen, wie an einem Akku - schwerelos - zeitlos - wooohlfüüühlen." (page 13 here) (and don't miss page 12 when you take a look :-] ) - this was called a "Master" paper. It got the best grade for it (in German grading system) from the person who now received the golden slow uptake award.
The winner of that last "Goldenes Brett (vorm Kopf)" tried to explain that actually the "paper" of the "student" has proven the original inventor of the tube to be wrong and the tube to be not working. But the paper was online. And said something else. So.... finally... - well, finally, the University with this funny "Hogwarts an der Oder" (which is now the nickname of that university insitute) received a letter from the Hochschulstrukturkommission that advised in clear words to close it. Follow the links above and take a look at it - it is really simply hi-la-ri-ous.
We need something like that. The Best Most Hilarious MT Results Award. It should be a yearly award or so.
[Edited at 2012-11-28 13:40 GMT]
| | | | |
|
|
|
apk12
Germany
Local time: 03:45
English to German
+ ...
* Can MT help human, professional translators to translate faster? The answer is YES.
Disagree. Disagree deeply. My source: myself and my experience. Disagree deeply. I have seen/proofread stuff that I swear must have been from definitely experienced translators, complex material, 70% really fine and then boom. You see those places. In my experience, it just leads to an automatisation in your own brain/Sprachzentrum (speech center). It lets you get blind for what you're doing and forces questionable results. At least, this is how I read those places that you recognize in some material. You can see where the translator decided to embed a service and to just edit it. I personally like the work itself too much for that. Call me an old grey haired woman [which I am anyway turning to step by step, so it would not be wrong], that is stubbornly sticking to old-fashioned styles. But from my experience - no way. I do not find them helpful.
They create the illusion that you can indeed "normalise" language, which simply does not work. Take the brochures from that one bigger company - they may decide to use a certain term for a certain product or service description. Now when in this very same brochure the same word pops up in an other paragraph, being here however a different term since the word is a homonyme - in such a case the "helped" language center in your brain has automatically a higher risk to create the illusion you can freely and safely take what is presented to you. That it not how language is working.
[Edited at 2012-11-28 11:00 GMT]
| | | | | I am not sure if MT does more damage than good in the translation industry | Nov 28, 2012 |
I personally think it does more damage than anything else to the translation industry. (It might be good for entertainment and language learning to a certain extent, as long as the learner knows what to expect form it -- that it cannot really be 100% counted on, although it may actually suggest some phrases useful to a beginner learner). It diverts attention from the text and the creative process of translation, turning it into match and seek, or correct, procedure. It is really silly and time c... See more I personally think it does more damage than anything else to the translation industry. (It might be good for entertainment and language learning to a certain extent, as long as the learner knows what to expect form it -- that it cannot really be 100% counted on, although it may actually suggest some phrases useful to a beginner learner). It diverts attention from the text and the creative process of translation, turning it into match and seek, or correct, procedure. It is really silly and time consuming. A serious translator can always Google the terms they are not sure about, which takes considerably less time and is more reliable. I personally don't see the future of MT in professional translation. ▲ Collapse
| | | | Yaotl Altan
Mexico
Local time: 19:45
Member (2006)
English to Spanish
+ ...
apk12 wrote:
* Can MT help human, professional translators to translate faster? The answer is YES.
Disagree. Disagree deeply. My source: myself and my experience. Disagree deeply. I have seen/proofread stuff that I swear must have been from definitely experienced translators, complex material, 70% really fine and then boom. You see those places. In my experience, it just leads to an automatisation in your own brain/Sprachzentrum (speech center). It lets you get blind for what you're doing and forces questionable results. At least, this is how I read those places that you recognize in some material. You can see where the translator decided to embed a service and to just edit it. I personally like the work itself too much for that. Call me an old grey haired woman [which I am anyway turning to step by step, so it would not be wrong], that is stubbornly sticking to old-fashioned styles. But from my experience - no way. I do not find them helpful. They create the illusion that you can indeed "normalise" language, which simply does not work. Take the brochures from that one bigger company - they may decide to use a certain term for a certain product or service description. Now when in this very same brochure the same word pops up in an other paragraph, being here however a different term since the word is a homonyme - in such a case the "helped" language center in your brain has automatically a higher risk to create the illusion you can freely and safely take what is presented to you. That it not how language is working. [Edited at 2012-11-28 11:00 GMT]
Yes, it's a kind of unlearning, wherein our mind tends to omit certain processes. That's why many translators prefer to translate the whole text from the very beginning.
| | | | | 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 » |
| | 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 » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
