Pages in topic:   < [1 2 3 4] >
How do you use MT?
Thread poster: ..... (X)
Bernhard Sulzer
Bernhard Sulzer  Identity Verified
United States
Local time: 17:16
English to German
+ ...
TM is not MT, AFAIC Sep 6, 2015

Jeff Allen wrote:

Bernhard Sulzer wrote:
Word of warning: "Translating" is something humans do. Using bilingual segments such as translation memories is not "translation" and it's not "machine translation."


Bernhard,

Translation memories are in fact one type of MT.


Translation memories (TMs) are source and target segments which are then analyzed by the CAT tool and when you work on a similar but completely new text, there will indeed be suggestions this CAT tool, or engine, makes that might be very helpful. That's what I use, and the CAT tool I use contains segments I entered in the first place (or other humans whom I can trust) - that is "target segments" I came up with in earlier projects or earlier in this project and that were entered into the TM. So who is the translator in the current project here? It's still me. I decide what to keep from the TM and what to discard, what to leave as if and what to change. This is not based at any point on a text engineered by a machine and then "edited" by me. Instead, I am the one creating the text whereby I might use some suggestions from my CAT tool that are based on previous input by me or other persons I can trust. I call it a translation when I am done because I indeed put together a text in the target language that is acceptable as a translation.

MT text, on the other hand, is something that someone serves up as completely engineered ("translated") by a machine, after the translation engine was programmed by humans; most importantly, the "engineered" text is presented as a complete text. Based on this text, outsourcers now simply ask for "post-editing" said content, postulating the rough draft is something that should be "edited" at editing rates.

Do you see the difference between these two examples? For one, I put trust into the first one because I am the translator throughout, re-using my own (or other trusted humans') TM input.
The second example would have me take a text arrived at by a machine based on one of various translation engine technologies, and asks me to trust the content to a degree that would have me agree to simply "edit" this "translation."

I don't have that trust.

Which engines do you want me to trust? Now, if you say build your own, then I would be trying to feed it with my knowledge for sure or again trust some technology that was developed by someone else . Which technology am I to trust and why? I haven't seen one text over 2 or more pages (short text) that could be called a translation that should simply be edited since it's really comparable with what a human would have presented as a rough draft. I am not talking about simple example sentences, I am talking marketing, technology (with many times little but significant changes in the text), and law, among other subjects.

I don't see my working with TM as a form of machine translation because it is not based on MT output but my own (the human translator's) input into the first and all the following updates of the TM.
That's not a form of MT the way MT is presented to us.

I know CAT tools compare source and target segments and that's a technological process but would you agree that that's not what a "good" translation engine is supposed to do? There are several types of translation engine models and types arriving at various different types/kinds of results.


For me there are really two big issues why I reject MT as "translation".

1) very pragmatic one, since outsourcers are really not interested in an MT output really being a "translation" but only to have a reason to pay less, and you can probably blame the clients too that go along with that thinking.
It's all about money. At the same time though, some are still demanding the "post-edited" content to be of excellent quality" and if I were to accept such a job, that's what you could indeed expect from me because that's the only thing I do. I don't provide sub-par results or simply "edit" something that's hardly a translation to a degree that will still make any reader's hair stand up on their head.

So, don't call something a translation when it surely isn't just because it suits the goal to get a real translation at editing rates.

To me, a translation is something that a human can indeed accept as a translation. So to prove your point, I would have to give you a source text, you run it through your machine and I then will evaluate it and say if I can accept it as a translation.

Problem is it isn't really a translation by a human and because of this one important reason, I must assume that a human has to indeed go through every single sentence of source and target text and examine it exactly to be sure that it is something that could be called a translation.

I am aware of the technological advances, and I might be willing to do an experiment and see how this would work for me.

2) The second problem is probably the bigger one and unfortunately, too many people have simply accepted the term MT (machine translation) as something that can ALWAYS be compared to the output of a human translator - why else call it a translation?

Based on this, we have the current post-editing scheme.
But we need to first accept what is happening here is really "translation." Well it's not in a human kind of way even though scientists and technicians who build these engines try to model them after the human brain processes while translating. Now that is clearly a very difficult task because it will take quite a while to get to know and define all our brain activities and (often intuitive) comparison-work between two languages. While the machine depends on its programming, the human takes a look at a sentence in the source language and can immediately come up with a correct (not necessarily the best version yet) translation, stylistically, grammatically, with the right terminology. And if he/she's not clear on the right word, he/she will do the necessary research, based on all kinds of factors that the machine will almost surely be lacking. And I don't accept completely wrong terminology (especially if found very frequently) as something that should simply be fixed by the "editor."

So, I would accept it as "translation" if it were what we humans do. Then it is logical to do so.

Others are less picky and say they accept it as long as the output is indeed acceptable, because why should we care about how we arrived at it if it's correct, right? That seems like a good assumption, but I am afraid I can't just expect that at this point of the technological development. Having said that, I am even less inclined to accept something as an "editing" job that was engineered by a translation engine with no human having had any say in the actual composition of the target text (besides having programmed or fed the machine) - no human reviewed it yet either.

I don't know what I should call this text in the target language engineered by a machine. Engineered, yes. Translated? Nah.

But hey, if you can develop a machine that produces the same result as a professional human translator, even if it's just a decent first draft (emphasis on something I can accept as a decent text), I might indeed consider using it. What you never get that way though is a text translated by me (or any other specific translator). My own personal/typical output might also change because I am working with a text that wasn't based on my own language skills but on something programmed to come out a certain way.

As you can see, individualism and lack thereof might be another price to pay when relying on MT.
My clients get text that is actually translated by me or in cooperation with another trusted colleague, no MT involved. If I were to ever accept any MT output and then edit it to arrive at the final product, it had to be damn good and afford me the opportunity to retain my individualism, my style, my overall quality. And all of that will not lead to lower rates because a machine will always just be a machine.

If a customer simply wants a machine's output, then he/she should just take that, that's fine with me. And if machines take over, so be it. But it will take a hell of a lot to replace the excellent translators that are working today.


Jeff Allen wrote:
I've just mentioned this in my immediately previous post in this thread. Over the past 15-20 years, I've given probably 2 dozen conference talks on this subject (many of them even being peer reviewed articles), and often in the presence of colleagues from the various TM/CAT tool solution providers. Even gave a special invited talk to the Trados development team in 1998.
Not a single person from any of those companies, nor any review comments received for the submitted articles, have ever mentioned that my statements were off-track from a technical standpoint nor from a marketing/sales perspective.
On the contrary, a VP from SDL sitting next to me during the presentation at Localization World 2004 asked afterward for a copy of the presentation because it helped simplify the explanation of such technologies.


I understand why the people you mention here might be interested in new technology since that's what they sell. But I am sure that I must see this from my point of view as a translator who lacks proof of the quality of MT and sees what it has produced: the new sub category within our industry, namely that of "post-editing MT" with all its wrong assumptions and consequences.


Jeff Allen wrote:
The term "Translation Memory" likely continues to be used by all such solution providers because it is just a more user-friendly term than "Example-based Machine Translation".


It's much easier to ask to "edit a translation," and I hold that "example-based machine translation" is a misnomer for CAT tools. Using a TM in one's translation work is not the same as working with a text that was created in its entirety by a translation engine, "readied" for post-editing.

Jeff Allen wrote:
Andy Way is the only person who has told me that he thinks there are some differences. This was several years before he left the academic arena to spend 3 years in commercial translation software and service companies. I should follow up with him now and see what he thinks after having lived daily in such trenches for those 3 years.


It would be interesting what he has to say.

Jeff Allen wrote:
There have been dozens, even hundreds, of posts in these various discussion groups over the years in which freelance translators claim that TM is not MT. Yet, there is no substantial evidence, and that being based on technical descriptions, that TM is different from EBMT.


Well, as I explained above, I believe they are very different.

Thanks for your input! You could take this post and feed it through your English-to-German engine. Let's see what we get.


PS: Because I watched your video: I don't consider 100% matches found (from example-based MT as you refer to it) as "translation" - it's simply matching something that has occurred before - most likely in the same document; it's not very likely this happens again in other documents and even if it does, it's not translation - someone (a human) translated it first, and all the machine does is alert you that it occurs again. That is very helpful, but it's not translation. And it's not what outsourcers mean when they say "post-edit MT." If the match is not 100%, the CAT tool will insert something and tell you it's say a 65% match (if you allow that percentage). That is still helpful, mostly, and it can (but doesn't necessarily have to) be easier, even faster, to use the information and then create the correct translation (human task) of that sentence. It is not likely that by using this method, we can simply have the entire text auto-translated (which means you would need an already populated TM before you start - so that's out if you start from scratch) and expect to receive a draft version that would resemble a draft by a professional human translator and can be edited easily. Never seen it. And if it's not, rearranging something that contains a lot of errors is no walk in the park. And I don't consider it "editing a translation."

[Edited at 2015-09-06 05:31 GMT]


 
Michael Beijer
Michael Beijer  Identity Verified
United Kingdom
Local time: 21:16
Member (2009)
Dutch to English
+ ...
@Bernhard: Sep 6, 2015

Oh no, here we go again. Sorry Bernhard, but can we pls avoid another marathon debate on whether MT can be called ‘translation’? Kevin started this thread to enquire about how we use MT, not to argue about semantic details.

If you think MT ≠ translation, that's fine, but start another thread to discuss it.

Michael

@Kevin: I haven't had time yet to properly answer your original question, but will
... See more
Oh no, here we go again. Sorry Bernhard, but can we pls avoid another marathon debate on whether MT can be called ‘translation’? Kevin started this thread to enquire about how we use MT, not to argue about semantic details.

If you think MT ≠ translation, that's fine, but start another thread to discuss it.

Michael

@Kevin: I haven't had time yet to properly answer your original question, but will do so ASAP! But first, we're off to the Sun. morning Elm Tree car boot sale in Icklesham!

[Edited at 2015-09-06 08:01 GMT]
Collapse


 
Jeff Allen
Jeff Allen  Identity Verified
France
Local time: 22:16
Multiplelanguages
+ ...
thanks Michael Sep 6, 2015

Michael Beijer wrote:

Oh no, here we go again. Sorry Bernhard, but can we pls avoid another marathon debate on whether MT can be called ‘translation’? Kevin started this thread to enquire about how we use MT, not to argue about semantic details.

If you think MT ≠ translation, that's fine, but start another thread to discuss it.


Thanks Michael. I was going to write a 1-2 sentence reply saying the same thing. No need to do that now.

Jeff


 
Jeff Allen
Jeff Allen  Identity Verified
France
Local time: 22:16
Multiplelanguages
+ ...
different levels of usage with MS Translator Hub Sep 6, 2015

Michael J.H. Davies wrote:
I use mainly MyMemory and SDL Language Cloud and, more occasionally, Microsoft's MT Enhanced using Microsoft Translator (particularly for texts within the IT domain) and sometimes Google Translate (which, in my view, has vastly improved since I first used it a few years ago).


Michael Beijer wrote:

Google Translate Microsoft Translator Google Translate
Interesting Michael, can you give us a bit more info on how this works? I keep meaning to check out the Microsoft Translator Hub. I use Google Translate and Microsoft Translator (via their respective APIs), both in little boxes in my CAT tool, and sometimes my CAT tools's various little tricks where it tries to combine them with my own resources, and auto-assembly, etc.


I inquired specifically about the Microsoft Translator Hub about a year ago because I was looking at different options for the MT projects we are doing at work.
There are different levels that you can have. There is free option for low-level volume usage. Beyond that, it is a paid service. Once you consistently go beyond a certain volume threshold on a regular basis, then you can choose the option to make your content non-shareable.
I didn't take any notes about it during the discussion with the new product manager of the MS team at that time because none of the scenarios applied to the needs of my team at the office. It is very possible that the MS Translator Hub offer has evolved since then.

Jeff


 
..... (X)
..... (X)
Local time: 06:16
TOPIC STARTER
My understanding of Microsoft Translator Hub Sep 6, 2015

I have been playing around a little with Microsoft Translator Hub over the past few weeks and here is how I understand it to work.

1) You can train an engine for free. You sign up for an account and can load documents to train 1 or more engines.
You need to have at least 10,000 segments loaded and selected before you can train an engine. It took about a day for it to train a 12,000 segment engine. It might take up to 2 days if you are training a lot more segments.

... See more
I have been playing around a little with Microsoft Translator Hub over the past few weeks and here is how I understand it to work.

1) You can train an engine for free. You sign up for an account and can load documents to train 1 or more engines.
You need to have at least 10,000 segments loaded and selected before you can train an engine. It took about a day for it to train a 12,000 segment engine. It might take up to 2 days if you are training a lot more segments.

2) Once your new engine is trained you are given a Category key. With this you can query the normal Microsoft Translator API and if you include the Category key in your API request, instead of querying the main (normal) Microsoft engine, it will query your customized engine.

3) Data is charged based on the number of characters you query per month. The first tier includes up to 2,000,000 characters per month for free.

4) Microsoft offers a "Keep Data Private" option, a special terms of service where they will not use the data you submit to train your own engine to also train their MT; however, you need to be paying at least the 250,000,000 characters per month ($2,055 per month) to be able to qualify for this separate terms of service.

Kevin
Collapse


 
Jeff Allen
Jeff Allen  Identity Verified
France
Local time: 22:16
Multiplelanguages
+ ...
This corresponds with what Microsoft told me last year Sep 6, 2015

Kevin Dias wrote:

I have been playing around a little with Microsoft Translator Hub over the past few weeks and here is how I understand it to work.


Kevin, thanks for these details. These corresponds closely to what the product manager of Microsoft Translator Hub told me last year. There were too many points on which it would not work with a number of various constraints of my project, so I did not use it.

Jeff


 
..... (X)
..... (X)
Local time: 06:16
TOPIC STARTER
Thanks for all the responses so far Sep 6, 2015

Hi all,

Thanks for all the great responses so far. Out of the translators who do use MT as part of their workflow, it seems like there are many different ways it is used (or could/would be used if confidentiality concerns were met). To summarize some of the ways mentioned so far:
- It keys in texts for me
- To seek idioms
- Only as a last resort, and for the Auto-Complete feature
- I mostly use it as a reference
- I use MT as one of my autosuggest dicti
... See more
Hi all,

Thanks for all the great responses so far. Out of the translators who do use MT as part of their workflow, it seems like there are many different ways it is used (or could/would be used if confidentiality concerns were met). To summarize some of the ways mentioned so far:
- It keys in texts for me
- To seek idioms
- Only as a last resort, and for the Auto-Complete feature
- I mostly use it as a reference
- I use MT as one of my autosuggest dictionaries: plenty of useful suggestions
- If nothing else it would be an instant gisting function. The result might be awful or wonderful, but given that I can delete or populate text in a target segment with a keystroke, what's the downside?
- I turn my source text into a machine translated TM and then load it as a TM in my CAT tool with a 30% penalty
- Post-editing MT
- For me MT is a modern version of the dictionary, only that MT often finds terms that are not in any dictionary

For those who do not use MT, the reasons include:
- I don't trust it for reference - there are far more reliable dictionaries and resources in my languages
- It is often more trouble than it is worth in the languages I work with
- It just isn't good enough, it's not worth the trouble to post-edit the results, but every once in a while it can be useful
- In my language pair the result is not understandable
- With most of my clients I have signed conditions prohibiting me from using MT
- And even with their permission, I would be very wary of using anything that passed sensitive information - such as, for example, a financial report that has not yet been released - back to the MT provider
- I can't trust the results because we don't know / can't understand how the machine is forming those results (if it is my own TMs I can trust them because I wrote them myself)


Michael Beijer wrote:
Oh no, here we go again. Sorry Bernhard, but can we pls avoid another marathon debate on whether MT can be called ‘translation’? Kevin started this thread to enquire about how we use MT, not to argue about semantic details.

If you think MT ≠ translation, that's fine, but start another thread to discuss it.


Yes, I agree. Please try to stay on topic. Bernhard, I realize that you are not happy that there is a trend in the industry for rates offered for PEMT jobs to be less than regular proofreading or translating rates when you feel that the work required is the same or more to produce a high quality final product. I think that is a very valid point but let's leave that for another thread. In this thread I am more interested if you use MT in your personal workflow or not, and if so, how do you use it.

Kevin
Collapse


 
Jeff Allen
Jeff Allen  Identity Verified
France
Local time: 22:16
Multiplelanguages
+ ...
yes, for many different needs and workflows Sep 6, 2015

Kevin,
I have used and continue to use MT for many different scenarios and workflows.
- many language pairs
- many industry sectors
- rule-based, example-based, stat-based, knowledge-based, etc
- locally-installed installations to enterprise-level deployments
- various types of user profiles ranging from professional translators, to software developers, to marketing content creators, to techdoc writers, etc
- with and without dictionaries
- with a
... See more
Kevin,
I have used and continue to use MT for many different scenarios and workflows.
- many language pairs
- many industry sectors
- rule-based, example-based, stat-based, knowledge-based, etc
- locally-installed installations to enterprise-level deployments
- various types of user profiles ranging from professional translators, to software developers, to marketing content creators, to techdoc writers, etc
- with and without dictionaries
- with and without postediting

Just over the past several months, I have been working specifically on projects involving new approaches to MT postediting, and some comparison of various MT system types based upon modified source language rewriting, etc

Now also embarking again on speech translation projects.

For me, the answer is usually "All of the above (& even more)".

Jeff
Collapse


 
Dan Lucas
Dan Lucas  Identity Verified
United Kingdom
Local time: 21:16
Member (2014)
Japanese to English
This is the key issue Sep 6, 2015

Kevin Dias wrote:
4) Microsoft offers a "Keep Data Private" option, a special terms of service where they will not use the data you submit to train your own engine to also train their MT; however, you need to be paying at least the 250,000,000 characters per month ($2,055 per month) to be able to qualify for this separate terms of service.

Earlier in the thread I noted that "My first condition for use, then, would be absolute, guaranteed privacy at a reasonable price."

Apparently the price for preserving confidentiality that is considered "reasonable" by Microsoft is more than $24,000 a year. At that level it's clearly not a sensible option for the vast majority of freelancers, but it might make sense for a team.

Dan


 
Jeff Allen
Jeff Allen  Identity Verified
France
Local time: 22:16
Multiplelanguages
+ ...
high volume to make the data private Sep 6, 2015

Kevin Dias wrote:
4) Microsoft offers a "Keep Data Private" option, a special terms of service where they will not use the data you submit to train your own engine to also train their MT; however, you need to be paying at least the 250,000,000 characters per month ($2,055 per month) to be able to qualify for this separate terms of service.


Dan Lucas wrote:
Earlier in the thread I noted that "My first condition for use, then, would be absolute, guaranteed privacy at a reasonable price."
Apparently the price for preserving confidentiality that is considered "reasonable" by Microsoft is more than $24,000 a year. At that level it's clearly not a sensible option for the vast majority of freelancers, but it might make sense for a team.


Yes, it also seemed to me, in talking with Microsoft, that it is aimed at helping teams and projects (like mine) which are mandated to do this type of project yet don't want to create and manage their own system from scratch, and can bill it to their IT dept as an ongoing, regular, expense.


 
Michael Beijer
Michael Beijer  Identity Verified
United Kingdom
Local time: 21:16
Member (2009)
Dutch to English
+ ...
@Kevin: Sep 6, 2015

Kevin Dias wrote:

I have been playing around a little with Microsoft Translator Hub over the past few weeks and here is how I understand it to work.

1) You can train an engine for free. You sign up for an account and can load documents to train 1 or more engines.
You need to have at least 10,000 segments loaded and selected before you can train an engine. It took about a day for it to train a 12,000 segment engine. It might take up to 2 days if you are training a lot more segments.

2) Once your new engine is trained you are given a Category key. With this you can query the normal Microsoft Translator API and if you include the Category key in your API request, instead of querying the main (normal) Microsoft engine, it will query your customized engine.

3) Data is charged based on the number of characters you query per month. The first tier includes up to 2,000,000 characters per month for free.

4) Microsoft offers a "Keep Data Private" option, a special terms of service where they will not use the data you submit to train your own engine to also train their MT; however, you need to be paying at least the 250,000,000 characters per month ($2,055 per month) to be able to qualify for this separate terms of service.

Kevin


In an ideal world, someone would offer what Microsoft does (the ability to train your own engine and use it in your CAT tool via an API/plugin), but with a "Keep Data Private" option, and without the crazy high characters-per-month limit. I'd be willing to pay a modest free for such a thing. Sounds like a job for TM-Town to me .

Michael


 
Bernhard Sulzer
Bernhard Sulzer  Identity Verified
United States
Local time: 17:16
English to German
+ ...
Probably no marathon debate Sep 6, 2015

Michael Beijer wrote:

Oh no, here we go again. Sorry Bernhard, but can we pls avoid another marathon debate on whether MT can be called ‘translation’? Kevin started this thread to enquire about how we use MT, not to argue about semantic details.

If you think MT ≠ translation, that's fine, but start another thread to discuss it.

Michael

@Kevin: I haven't had time yet to properly answer your original question, but will do so ASAP! But first, we're off to the Sun. morning Elm Tree car boot sale in Icklesham!

[Edited at 2015-09-06 08:01 GMT]


You didn't read Kevin's last request then :

Kevin Dias wrote:
- any other thoughts on the topic?


Even without that request, translators are not all in your camp as you know, and since most participants here clearly favor MT, I took the liberty and commented on the topic from my (a different) point of view.

I don't intend to make this into "another marathon debate, and even though I find your colorfully highlighted comment less than fair, I am not going to get riled up about it.

Fact is, there are many people not in your camp (the MT camp), and if I believe that what people here usually refer to as MT is not a valid term, I will say so. You can agree or disagree.

As long as the majority of contributors here are all for MT, there won't be a marathon debate anyway. Only if other people thinking like me chime in here will it make sense to spend a lot of time on this debate. Otherwise, you guys are just going to drown me out - and that's never much fun.

People reading this thread can make up their own mind. But I am sure you don't encourage anyone to accept a concept making people believe they should carry out "post-editing MT" jobs at editing rates, i.e. for less than they would charge for a translation. Or do you? If so, then you should back up your claims of how good MT is. And that means having put a text through a translation engine and then give it to a translator for evaluation


 
Bernhard Sulzer
Bernhard Sulzer  Identity Verified
United States
Local time: 17:16
English to German
+ ...
Debate? What debate? Sep 6, 2015

Jeff Allen wrote:

Michael Beijer wrote:

Oh no, here we go again. Sorry Bernhard, but can we pls avoid another marathon debate on whether MT can be called ‘translation’? Kevin started this thread to enquire about how we use MT, not to argue about semantic details.

If you think MT ≠ translation, that's fine, but start another thread to discuss it.


Thanks Michael. I was going to write a 1-2 sentence reply saying the same thing. No need to do that now.

Jeff


This way of reacting seems to be easier for you than addressing the points I raised. Can't say I am surprised.
But the points are there.


 
Michael Beijer
Michael Beijer  Identity Verified
United Kingdom
Local time: 21:16
Member (2009)
Dutch to English
+ ...
We are traying to discuss referents, not terms/concepts Sep 6, 2015

Bernhard Sulzer wrote:

Michael Beijer wrote:

Oh no, here we go again. Sorry Bernhard, but can we pls avoid another marathon debate on whether MT can be called ‘translation’? Kevin started this thread to enquire about how we use MT, not to argue about semantic details.

If you think MT ≠ translation, that's fine, but start another thread to discuss it.

Michael

@Kevin: I haven't had time yet to properly answer your original question, but will do so ASAP! But first, we're off to the Sun. morning Elm Tree car boot sale in Icklesham!

[Edited at 2015-09-06 08:01 GMT]


You didn't read Kevin's last request then :

Kevin Dias wrote:
- any other thoughts on the topic?


Even without that request, translators are not all in your camp as you know, and since most participants here clearly favor MT, I took the liberty and commented on the topic from my (a different) point of view.

I don't intend to make this into "another marathon debate, and even though I find your colorfully highlighted comment less than fair, I am not going to get riled up about it.

Fact is, there are many people not in your camp (the MT camp), and if I believe that what people here usually refer to as MT is not a valid term, I will say so. You can agree or disagree.

As long as the majority of contributors here are all for MT, there won't be a marathon debate anyway. Only if other people thinking like me chime in here will it make sense to spend a lot of time on this debate. Otherwise, you guys are just going to drown me out - and that's never much fun.

People reading this thread can make up their own mind. But I am sure you don't encourage anyone to accept a concept making people believe they should carry out "post-editing MT" jobs at editing rates, i.e. for less than they would charge for a translation. Or do you? If so, then you should back up your claims of how good MT is. And that means having put a text through a translation engine and then give it to a translator for evaluation


Bernhard,

I am not in any camp. I don't give two hoots whether people use MT or not. That's their business. I just don't want the topic to get all confusing because someone starts arguing endlessly about the words we use, rather than their referents. The terminology of translation is very interesting, but completely irrelevant (in this thread).

Think about it:

Kevin asks: How do you use MT (when translating)?

And you answer: "Translation is not machine translation!", "TMs are not MT!", and finally, "I am against MT!".

How does that make any sense?

These might be interesting topics (for a separate thread), but how do they relate to the current thread? I don't mind arguing about stuff, but not if it is all so confusingly put that no one knows what is going on. Then it's just a waste of our time. That's why I used my special blue font for Oh no, here we go again.

You said:


"Fact is, there are many people not in your camp (the MT camp), and if I believe that what people here usually refer to as MT is not a valid term, I will say so. You can agree or disagree."


But what do you actually mean by "what people here usually refer to as MT is not a valid term"? I can't agree or disagree with it because I don't understand it.

You also said:


"People reading this thread can make up their own mind. But I am sure you don't encourage anyone to accept a concept making people believe they should carry out "post-editing MT" jobs at editing rates, i.e. for less than they would charge for a translation. Or do you? If so, then you should back up your claims of how good MT is. And that means having put a text through a translation engine and then give it to a translator for evaluation."


Again, this is very confusing.

If anyone is dumb enough to accept a post-editing job for extremely low rates, it won't be because of a concept or a term! It will be because they misjudged the quality of the MT output. Do you see what I mean? You are linking certain lamentable, current practices regarding poorly paid post-editing jobs with the concepts/terms we use to talk about matters such as TMs (translation memories), translation, MT (machine translation), etc.

I too get offered those idiot post-editing jobs. I just say "No thanks, !*$$! off". Arguing about whether MT is a form of translation isn’t going to stop anyone from offering them.


 
Samuel Murray
Samuel Murray  Identity Verified
Netherlands
Local time: 22:16
Member (2006)
English to Afrikaans
+ ...
What Bernhard means Sep 6, 2015

Bernhard Sulzer wrote:
I think it would be a good idea to learn your definition of MT. We have been discussing this topic for a while now here on Proz.com. Plenty of threads.


Many of those "plenty of" threads were started by Bernhard (-:

In those threads Bernhard explained that he believes that the term "translation" has a particular definition which he believes everyone else should have as well. He explained that he believes that the term "translation" can only refer to some kind of cognitive process in the brain, and that if a human and a machine produces an identical target text, the human produced a "translation" (because the human is sentient) and the machine did not.


 
Pages in topic:   < [1 2 3 4] >


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

How do you use MT?






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 »
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! »