| Pages in topic: < [1 2 3 4 5 6 7] > | The "Crime" of Using MT Thread poster: Michelangela
|
|---|
Samuel Murray 
Netherlands
Local time: 19:57
Member (2006)
English to Afrikaans
+ ...
| Miguel isn't telling | Jun 17, 2014 |
Mikhail Kropotov wrote:
If a translator posts a KudoZ question asking how to translate, say, 'file' or 'FTP server,' I wouldn't label them as a bad translator overall, but...
Yes, but now you're moving the goalposts. Neither of us knows what the exact query was that Miquel was referring to. All I know is that he connects the dots as follows: the translator doesn't know X, only bad translators don't know X, the translator supports MT, therefore MT is bad.
| | | | Samuel Murray
Netherlands
Local time: 19:57
Member (2006)
English to Afrikaans
+ ...
Mikhail Kropotov wrote:
I'm afraid you're just an inexperienced translator ... [if] you cannot think of a synonym that MT can [think of].
Michal Fabian wrote:
If MT helps you produce better (quicker, neater, more accurate...) translations than you would without it, chances are you're incompetent as a translator.
I'm afraid I do not share these sentiments.
I remember at a time that I had been using CAT for about two years, I was asked to do a number of translations without CAT, and I really struggled to do so, because by that time I had become used to the CAT method of translation.
A similar thing happens whenever I try to translate using the methods that we were taught at college (i.e. using strict application of linguistic theory). I had been translating using a different method for so long that I had forgotten how to do that which I was once quite capable of.
I now experience the same thing with subject fields in which I have been using MT for the past two years: if I try to translate in those fields without MT, I struggle to do so (somewhat)... not because I have become a less competent translator in those fields (as you two seem to believe) but simply because I've gotten used to using a specific, different method for it.
In certain fields, CAT helps me to produce translators that are "neater", and MT helps me to produce translations "quicker". That certainly does not mean that I'm incompetent as a translator in those fields.
| | | | | False comparisons | Jun 17, 2014 |
Samuel Murray wrote:
I remember at a time that I had been using CAT for about two years, I was asked to do a number of translations without CAT, and I really struggled to do so, because by that time I had become used to the CAT method of translation.
A similar thing happens whenever I try to translate using the methods that we were taught at college (i.e. using strict application of linguistic theory). I had been translating using a different method for so long that I had forgotten how to do that which I was once quite capable of.
I now experience the same thing with subject fields in which I have been using MT for the past two years: if I try to translate in those fields without MT, I struggle to do so (somewhat)... not because I have become a less competent translator in those fields (as you two seem to believe) but simply because I've gotten used to using a specific, different method for it.
In certain fields, CAT helps me to produce translators that are "neater", and MT helps me to produce translations "quicker". That certainly does not mean that I'm incompetent as a translator in those fields.
I believe you're muddying the waters by bringing CAT into the discussion as it has nothing whatsoever to do with MT.
MT produces an entire computer-created translation FOR YOU. CAT does not produce anything. It lets you enter YOUR OWN translations into the target segment by segment, allowing you to consult YOUR OWN previous translations (TM) for terminology (concordance) and sometimes reuse YOUR OWN previous translations in whole or in part (fuzzy matches). Plus of course it has spell-checking, QA, etc. Still, you are the author of each translated word. If you don't know how to translate some term, no CAT will do it for you. MT will. That's why MT can easily help people pass off as translators when they don't know what they're talking about. Of course, you could create a poor translation even without MT, but that would go much slower.
If you're an expert translator and you use MT for productivity, that's fine. Just don't tell me it lets you come up with better wording than you would on your own.
[Edited at 2014-06-17 18:49 GMT]
| | | | Michelangela
United States
Local time: 10:57
German to English
+ ...
TOPIC STARTER
Mikhail Kropotov wrote:
I believe you're muddying the waters by bringing CAT into the discussion as it has nothing whatsoever to do with MT.
The interesting part is that some CAT systems such as Trados have something to do with MT. Trados can be configured to use various sources of MT (e.g. google Translate, BeGlobal).
[Edited at 2014-06-17 19:05 GMT]
| | |
|
|
|
Alex Lago
Spain
Local time: 19:57
English to Spanish
+ ...
As always whenever MT is mentioned you will find loads of people telling you all the reasons why it is bad for you and how real translators don't use it.
All I can say is if it works for you then use it.
MT is just a tool, like all tools some people will find it useful, some people will never use it.
Don't let anyone tell you what your work flow should be, you decide what works best for you.
| | | | Michelangela
United States
Local time: 10:57
German to English
+ ...
TOPIC STARTER
When I started this thread my intention was to allow translators to express pro & con arguments regarding the use of MT. I was personally more interested in con opinions.
It has gone far beyond what I ever expected. I appreciate the many helpful comments, and deplore the few derogatory ones.
I'd like to express my appreciation for people who responded in a dignified manner, and brought logical arguments to the discussions. Many others felt the need to express their grea... See more When I started this thread my intention was to allow translators to express pro & con arguments regarding the use of MT. I was personally more interested in con opinions.
It has gone far beyond what I ever expected. I appreciate the many helpful comments, and deplore the few derogatory ones.
I'd like to express my appreciation for people who responded in a dignified manner, and brought logical arguments to the discussions. Many others felt the need to express their greatness in not using MT. I thank them as well, because they helped me understand their reasons.
In conclusion ... “Don’t judge, so that you won’t be judged. For the way you judge others is how you will be judged — the measure with which you measure out will be used to measure to you". ▲ Collapse
| | | | Alex Lago
Spain
Local time: 19:57
English to Spanish
+ ...
| Should have looked at other threads | Jun 17, 2014 |
Michelangela wrote:
It has gone far beyond what I ever expected.
You obviously hadn't looked at other MT threads in Proz, this is a typical response.
| | | | Michelangela
United States
Local time: 10:57
German to English
+ ...
TOPIC STARTER | Incorrect Assumption | Jun 17, 2014 |
Alex Lago wrote:
You obviously hadn't looked at other MT threads in Proz, this is a typical response.
I did. The point of my comment was that "I appreciate the many helpful comments, and deplore the few derogatory ones."
| | |
|
|
|
Samuel Murray
Netherlands
Local time: 19:57
Member (2006)
English to Afrikaans
+ ...
| Muddying the waters | Jun 17, 2014 |
Mikhail Kropotov wrote:
If you're an expert translator and you use MT for productivity, that's fine. Just don't tell me it lets you come up with better wording than you would on your own.
The MT system contains snippets of translations from thousands of other translators, some of whom are better than you or who have come up with a better way of saying something, and it is not unlikely that some of those translations will end up in the results offered by the MT system, particularly if the MT system has a reliable method of evaluating the quality of its sources.
By this reasoning, allow me to tell you: MT can let a translator come up with better wording that he would have been able to do on his own. Not because MT is intelligent (it is not) but because MT is powered by a database of translations by thousands of other professional translators.
Samuel Murray wrote:
A similar thing happens whenever I try to translate using the methods that we were taught at college (i.e. using strict application of linguistic theory). I had been translating using a different method for so long that I had forgotten how to do that which I was once quite capable of. I believe you're muddying the waters by bringing CAT into the discussion as it has nothing whatsoever to do with MT.
We did not learn how to use CAT at college.
I mention CAT in my previous post for the same reason that I mention college, i.e. as an example, to show that this particular aspect of dealing with MT (i.e. that you unlearn unrelated tasks) is not limited to dealing with MT.
MT produces an entire computer-created translation FOR YOU.
MT puts a best guess of all the right words in roughly the right order for you. Your job is to fix the word order and to select new words if the proposed ones are not satisfactory, in order to produce a target text that means the same as the source text.
CAT lets you enter YOUR OWN translations into the target segment by segment...
CAT does not let you enter your own translations... unless it is set up to not insert the best fuzzy match or the source text into the target field. And CAT does not always work with one's own translations -- it can involve evaluating matches done by other translators (who were paid less, slept less, argued with the client less, etc), which you have to evaluate for quality anyway.
And systems that serve MT translations also let you enter your own translation (remember, this thread is not about MT post-editing but about using MT as a translation aid for freelancers).
Still, [with CAT] you are the author of each translated word.
This is only true if you never use TMs that contain segments that were translated by other translators. Think of MT as a sophisticated dictionary look-up system, so then you are also the author of each word translated, because you are the one who must decide in the end which suggestions from the dictionary are right and which ones are not right for this text.
If you don't know how to translate some term, no CAT will do it for you. MT will.
If you don't know how to translate some term, you can use CAT to search a database of old translations (yours or someone else's) or you can use your CAT tool's built-in dictionary look-up feature (if available). Or, you can use MT to search a much bigger database of translations. It stands to reason that the bigger the database the higher the chance of finding a match.
MT can easily help people pass off as translators when they don't know what they're talking about.
Yes, but the fault does not lie with MT -- rather with those translators. I have seen exceptionally bad translations perpetrated by people who did not use MT, and quite acceptable translations by people who did use MT (I can think of a project in which Google Translate was used specifically, and the translations from experienced translators were of acceptable quality, and the translations from beginner translators were of unacceptable quality).
| | | | Phil Hand
China
Local time: 01:57
Chinese to English
| Samuel - could you give us an example? | Jun 17, 2014 |
This conversation is all a bit fact-free, and we just end up hating each other. Apparently I'm vehement, and Samuel is a bad translator!
I asked this last time the topic came up, and I'll ask it again: Samuel, seeing as you're here and you've taken the time to explain clearly how MT works for you, could you give us an example? Find a paragraph of representative text, run it through your system, and put the results up? (Could you do it AF-EN, so that we can all read the results?) The... See more This conversation is all a bit fact-free, and we just end up hating each other. Apparently I'm vehement, and Samuel is a bad translator!
I asked this last time the topic came up, and I'll ask it again: Samuel, seeing as you're here and you've taken the time to explain clearly how MT works for you, could you give us an example? Find a paragraph of representative text, run it through your system, and put the results up? (Could you do it AF-EN, so that we can all read the results?) Then either you can put up your edit, or we can all say how many edits we think it requires. You could also point out any particularly good suggestions that came from the system.
NB. If you do this, some idiot will undoubtedly turn up in the thread and abuse you because they disagree with your word choices. So feel free to say no, or to invite a moderator to come and watch the thread first. But for me, some real practical examples are the only way to advance this conversation. ▲ Collapse
| | | | Dan Lucas
United Kingdom
Local time: 18:57
Member (2014)
Japanese to English
| Advancement not required | Jun 17, 2014 |
Phil Hand wrote:
Apparently I'm vehement, and Samuel is a bad translator!
...
But for me, some real practical examples are the only way to advance this conversation.
Don't worry Phil, nobody has a monopoly on vehemency round here.
But I'm skeptical that the conversation needs advancing, partly because I don't see how we can arrive at a general proof that MT in its current form is good or bad.
I'd argue that like so many things in life, if it works for you as an individual that's good enough. If MT contributes in some way to your job, then it's useful. If it doesn't it can safely be ignored.
| | | | Phil Hand
China
Local time: 01:57
Chinese to English
| Not proof but persuasion | Jun 17, 2014 |
Thanks, Dan.
Yeah, we can never get proof of much at all, but I would like to be persuaded. I'll admit that I do kind of use the MT thing as a prejudicial rule of thumb when I'm looking at other translators. And that's backed up by lots of experience, particularly in my pair (and I presume yours as well). If I'm wrong, I'd like to know about it! I'd also like to stay on top of the technical trends, so if MT can be a useful tool, I'd like to know how.
Also, from a broade... See more Thanks, Dan.
Yeah, we can never get proof of much at all, but I would like to be persuaded. I'll admit that I do kind of use the MT thing as a prejudicial rule of thumb when I'm looking at other translators. And that's backed up by lots of experience, particularly in my pair (and I presume yours as well). If I'm wrong, I'd like to know about it! I'd also like to stay on top of the technical trends, so if MT can be a useful tool, I'd like to know how.
Also, from a broader perspective: some of us like to think that we're a bunch of professionals, but we're not behaving like them in this thread. One of the hallmarks of professionalism is the ability to regulate practice and spread beneficial new practices. The MT issue is clearly quite divisive, so I would like to see the debate advanced. I guess on my side, I should simply choose not to air my prejudices; I hope on the pro-MT side, MT users will be willing to illustrate how their processes work. And then we really can advance to a state of saying, MT works for this, doesn't work for that; MT works for me, it doesn't work for you. The model would be Peter Simon's post on page 2 of this thread.
My problem is that as of today, I have literally never seen MT output that I thought was editable. So I'm still finding it hard to believe. I wouldn't trust the hype from advertisers, so I ask my colleagues. ▲ Collapse
| | |
|
|
|
| It's never advisable to generalize, because the next guy may be a very happy user of MT systems | Jun 17, 2014 |
Why do some people believe that their experience is "the only pebble on the beach"?
Many of us find that the MT+revision process has two problems.
1) By providing you with a (bad) version, it interrupts your own natural thought processes. MT works on sentences; people write texts. If you're constantly looking at MT output, you never develop a flow, you never get into the feel of the text, so you don't produce a cohesive target.
... See more Why do some people believe that their experience is "the only pebble on the beach"?
Many of us find that the MT+revision process has two problems.
1) By providing you with a (bad) version, it interrupts your own natural thought processes. MT works on sentences; people write texts. If you're constantly looking at MT output, you never develop a flow, you never get into the feel of the text, so you don't produce a cohesive target.
1. The MT+revision process does not interrupt my own natural thought processes, but rather complements them... and it can suggest an absolutely new approach to handling a particularly difficult sentence or paragraph.
2. I am constantly looking at MT output, yet I can easily develop a flow, and I have no problem getting into the feel of the text... and producing a cohesive target, which is evidenced by a very high repeat business rate in my case.
2) Revising MT just takes longer. Typing a new sentence doesn't take very long. Editing a sentence takes longer - for me, if there are more than two errors in a sentence, it would literally be quicker to type it from scratch than to use mouse/cursor keys to go to error 1, delete, correct, then use mouse/cursors to go to error 2, delete, correct. And MT translations never have less than 2 errors per sentence. So physically, MT revision is always slower.
1. I have been following the developments in MT for two decades. I always tried each new major version of several MT systems (e.g. PROMT). I never used them because MT revision took just as much time as doing the translation from scratch... Now, the situation is a bit different. The recent developments in MT are really impressive, and I see that I progress faster whenever I activate an API to feed MT inputs into my SDL Trados Studio.
2. It is very often that I just accept an MT-suggested segment when I translate UN-related material.
The reason many of us (including me) are suspicious of MT is that we think MT users don't correct enough. I've asked on this forum before for examples of "good MT", and an MT enthusiast showed me a "good" translation from Dutch to English. In less than 30 words, there were 5 errors, two of them major. Like I say, the correction process physically takes longer - and it would still be a bad translation, because that only fixed the gross errors. I'm not yet talking about style or flow. To fix those, you'd have to rewrite completely.
1. The reason many of us (including me) are happy with MT is that we think that MT provides a readily available (and useful) 'helping hand' in many cases.
2. Have you ever tried to use a generic MT with a couple of specialist dictionaries attached (e.g. Oil & Gas, or Information Technology, or Legal)?
3. The post-edited translation may be poor, medium or just brilliant, depending on the editor
So if you're going to rewrite completely, what's the benefit?
My process: 1) Read source 2) Write target
MT process: 1) Read source 2) Read target 3) Realise target is unusable 4) Delete target 5) Rewrite target
One of those processes is obviously more efficient.
I dare say that it's not that obvious in my case
So when I hear someone say "I use MT", I'm afraid that I assume they don't do enough editing, and they are satisfied with poor quality texts.
Obviously, it takes all types. Maybe for you, in your field, MT does work well. But there are good reasons for not bringing MT into a translation flow. You can do what you want, but you're going to have to live with the fact that many of us will treat it with suspicion for a good few years yet. [/quote]
1. I am here to share experiences and new ideas... You can do what you want, and you can treat MT with suspicion for your lifetime. You can even make do with an old good mechanical typewriter, for all I care:) It's "everyone to his trade", I have no problem with that.
Personally, I believe that people who are still debating whether using an MT (or CAT) tool is a good or bad thing, are just missing the gravy train here and now!
2. Moreover, I am afraid you may be either (1) squeezed out of the market in 2-3 years, because ALL potential clients would expect their language service provider to use an MT (or CAT) tool compatible with their business process; or (2) forced to obtain and learn one of those... but by that time you may find yourself dragging behind the happy crowd of early adopters who would be able to do 5,000-7,000 words per day, thus earning twice as much ▲ Collapse
| | | | | @Little Woods: Don't count me in whenever you say "we", please | Jun 18, 2014 |
[quote]Little Woods wrote:
The most we need is the glossary or terms.
[quote]
I need VERY MUCH MORE than the glossary. It's quite possible that you can't even dream about what I might need and use right this minute.
e.g. I was already reading e-books back in Sep 2006 on my iRex iLiad which started it all. I don't think you even know the name, because they are long gone...
QUOTE
The distributor of the iLiad is iRex Technologies, a Philips spin-off company. It was initially advertised in December 2005, to be launched in April 2006, but was delayed until July, when it started to be sold as a beta product. It was released to the general public near the end of July, and since then has undergone considerable software revisions.
Its list price in Europe is €649, and in US $699, however it is no longer available in North America due to FCC regulation non-compliance.
...
In early June 2010, iRex Technologies of the Netherlands filed for Bankruptcy protection. According to CEO Hans Brons (2010), the decline in sales were a direct result of a delayed response from the U.S. Federal Communications Commission (FCC) to approve the device.
UNQUOTE
By way of example (a disclaimer to avoid any misunderstanding: the scenario below is for demonstration purposes only... but you may try and reproduce it or a similar setup... if you do that, get back to this thread, please, and share your experience:))
You have two windows open:
- SDL Trados Studio in the 'Editor' view, with Dragon Naturally Speaking 12 plugged-in;
- Notepad, or WordPad, or Word, or ... with an MT plugged-in;
The MT provides a suggestion in its window, e.g. "Come to think of it, an MT tool might well be the future of the translation industry, right here on your doorstep."
You read it out, DNS 12 style, i.e. "Come to think of it COMMA an MT tool might well be the future of the translation industry COMMA right here on your doorstep PERIOD"
While reading out loud, you see DNS 12 in action, "writing down" this sentence for you at the rate of 160 wpm with an 99.9% accuracy.
For those who tend to be suspicious of new technology, please find below a couple of links to illustrate the functionality of the latest release:
1) http://www.youtube.com/watch?v=8LCBsC9wuFc
2) http://www.nuance.com/for-business/by-product/dragon/dragon-for-the-pc/legal-edition/index.htm#dr_dragon_inaction
I can vouch for the product because I have tried it myself, and it delivered as promised. I can imagine that you might have another suspicion now, namely that I am on the developer's payroll... Well, I'm not... but I may consider joining their reseller program one of these days
[Edited at 2014-06-18 00:35 GMT]
| | | | | Do you really mean to say I am satisfied with poor quality texts just because I do use MT? | Jun 18, 2014 |
[quote]Maksym Petrov wrote:
I couldn't have put it better myself:
Phil Hand wrote:
So when I hear someone say "I use MT", I'm afraid that I assume they don't do enough editing, and they are satisfied with poor quality texts.
I might retaliate this accusation by assuming that you are a less-skilled, less-respected and, resultantly, less-paid 'freelancer', as opposed to an 'independent language professional'... No offense meant, just in case.
[Edited at 2014-06-18 00:36 GMT]
| | | | | Pages in topic: < [1 2 3 4 5 6 7] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » The "Crime" of Using MT | 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! » |
| | TM-Town | Manage your TMs and Terms ... and boost your translation business
Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
Translation news
