| Pages in topic: < [1 2] | Ethical dilemma about the contests Thread poster: Katalin Horváth McClure
|
|---|
Samuel Murray 
Netherlands
Local time: 17:18
Member (2006)
English to Afrikaans
+ ...
| I see it now, thanks. | Sep 12, 2009 |
Katalin Horvath McClure wrote:
Right there, where you see "Enter up to 3 language pairs in which you work...", you should see your current language pairs with their rates...
I see it now, thanks. I guess I was stymied by the wording "in which you WORK", but the option to choose a non-work interest-language is only visible after you've entered a language combination using the "in which you WORK" option. Weird.
Well, there is only 1 Spanish-Afrikaans translator and 1 Russian-Afrikaans translator on ProZ.com -- I wonder if a couple of us English-Afrikaans folks should help their language combination enter Alternatively, there are no Portuguese-Afrikaans or Catalan-Afrikaans translators at ProZ.com -- perhaps we should look at these. Or do you think we should target language combinations with at least a few translators at ProZ.com -- being French-Afrikaans (6) and German-Afrikaans (10), of which one translator has already won an English-Afrikaans translation contest?
| | | | Samuel Murray
Netherlands
Local time: 17:18
Member (2006)
English to Afrikaans
+ ...
| | I do not see very much - | Sep 19, 2009 |
what this has to do with ethics.
The process is about choosing the best translation out of the pool of submissions.
If someone considers it wise to submit a text done by a machine - or by relying heavily on other people - and this text wins as the best solution among the other submissions, it will be the best solution, or the one voted as such, no matter how that text will have been obtained. And as such the text "deserves" the prize.
If then the poster foo... See more what this has to do with ethics.
The process is about choosing the best translation out of the pool of submissions.
If someone considers it wise to submit a text done by a machine - or by relying heavily on other people - and this text wins as the best solution among the other submissions, it will be the best solution, or the one voted as such, no matter how that text will have been obtained. And as such the text "deserves" the prize.
If then the poster fools himself by thinking he has won it, although he did not really translate it, that will be his problem - but I do not think it is an ethical one.
An ethical problem would only arise if the text would have been taken from other people without their consent. - A machine has not to be asked for consent, or has it?
If someone hangs an Olympic medal around his neck, that he has not won, he is in my eyes simply a fool. ▲ Collapse
| | | | | there you go... | Sep 19, 2009 |
as soon as "prizes" are introduced for the winners, the cheating (potential or not) starts...
| | |
|
|
|
| Cheating and possibility to vote out fakes completely | Sep 21, 2009 |
Giovanni Guarnieri MITI, MIL wrote:
as soon as "prizes" are introduced for the winners, the cheating (potential or not) starts...
But this is not an "Ethical dilemma about the contests" - it is the normal "ethical dilemma" with cheating as such.
But I have just seen a text Es-De and another one Ru-En, that were very obviously produced with MT and not even worked over, and the voting/tagging system should provide the possibility to earmark and discard such pieces right away.
[Edited at 2009-09-21 09:36 GMT]
| | | | Zea_Mays
Italy
Local time: 17:18
Member (2009)
English to German
+ ...
| Machine translated text / Translation quality | Oct 21, 2009 |
Hi all,
don't know where to post the question, so I do it here:
How is it possible that (for the it>de section in this case)
a) there is a machine translated one??
b) the quality of the translations is so poor??
ps: the question is in part quite rethorical
| | | | | Machine translated texts all over - staff opinion? | Oct 21, 2009 |
I have seen straight machine translation texts in the English-Hungarian and in another pair, too (I am not saying which one because I believe voting is still on in that pair). I am not talking about somebody using MT as "help" (which is apparently allowed), but putting the text into Google translate and copying the result from there, without any editing.
I was wondering who in their right mind would do that. If the person really works in the actual language pair, what is the advantage of d... See more I have seen straight machine translation texts in the English-Hungarian and in another pair, too (I am not saying which one because I believe voting is still on in that pair). I am not talking about somebody using MT as "help" (which is apparently allowed), but putting the text into Google translate and copying the result from there, without any editing.
I was wondering who in their right mind would do that. If the person really works in the actual language pair, what is the advantage of doing it? There is no way to win.
Or perhaps somebody did it just for kicks? To see how a machine translation fairs against human translations? Or prove some other point? As we know, you don't even have to work in a language pair to be able to submit an entry, so you could put a bunch of languages as "interest languages" in your profile, and have your way with Google - and the contest.
This is annoying on several levels:
1. People reviewing the texts have to waste time tagging the text or voting it down to be able to get it out of the list. This is not only disrespectful of the whole contest, but may even be categorized as abuse of resources.
2. In language pairs with only a few submissions, this practice can alter the outcome quite a bit. For example, if there are two real translations and one machine translation submitted by the first (original) submission deadline, the pair goes directly into final voting.
This means there are actually only two contestants. On the other hand, if the machine translation was not submitted, the pair would have had two submissions, it would have gone into extended submission, waiting for more contestants.
So, allowing the machine translation, and not filtering it out during the qualification phase changes the contest scene. This actually happened in at least one of the contest pairs this time.
What complicates the issue is that if the pair goes into the final voting round directly, even if you are a native speaker of the target language and it is obvious to you that the text is an unedited machine translation, if you don't work in the exact pair, you can do nothing, you cannot help the process by tagging the fake entry. On the other hand, if there was a qualification phase, you could do it.
I am wondering what ProZ staff thinks about this, whether they are planning on investigating how these unedited machine translations got into the contest, and want to figure out a way to deal with them in the future. I think it would be good to have some way of getting rid of these as early in the process as possible, as having these may demotivate contestants and voters for the reasons mentioned above.
Thanks
Katalin ▲ Collapse
| | | | | practicability - tagging from 0 to 5 | Oct 24, 2009 |
Katalin Horvath McClure wrote:
I am wondering what ProZ staff thinks about this, whether they are planning on investigating how these unedited machine translations got into the contest, and want to figure out a way to deal with them in the future. I think it would be good to have some way of getting rid of these as early in the process as possible, as having these may demotivate contestants and voters for the reasons mentioned above.
All your observations about this are perfectly right.
But I think it would be asking too much of the contest organizers to do anything like a screening as you are proposing it here. After all, this would mean no less than to check each and every entry in each and every pairing - clearly an impossibility. All the more as the case that you are describing, that someone will, out of sheer mischief, put in a completely unedited MT piece will after all be rather rare.
Anybody who participated a few times in these contests, will have been aware that there are "cheating" or even just "distorting possibilities" in them. And maybe what you are describing here is even a rather less important possibility of cheat or distortion. After all, the only harm produced by such "mock or bogus entries" is that they can deform the competition on quantitative grounds - in blowing up the number of submissions and changing the quorum situation, as you have well described. - For it is practically just unthinkable that people working in the concerned pairing would mark a mere machine translation as winner. - As you have also justly said, submitting an edited MT might be a more or less legitimate procedure. But I have doubts that the mere linguistic polishing will work for an MT of non technical content. All that I have seen from MTs so far is that there are hardly any two consecutive lines in them, that, independently from their linguistic value or non-value, do not contain complete mis"understandings" of meaning, mostly of a thoroughly laughable nature.
Much more "pernicious" than these more or less mischievous "mock or bogus submissions" at the bottom line of any winning chances, is in my opinion the possibility of "buddy" or "teletubbies love each other voting" at the top of the list. The organizers had, in order to reduce this possibility, early on arranged for a different sequence of submissions visible for every person logging in to the contest. But that can of course be easily circumvented by giving others some specific clues about the wording of a particular piece and make it so easily identifiable.
I think, Katalin, one just has always to be well aware, that these contests are a nice distraction. That they have a certain value, are a nice exercise and, for me particularly interesting, give an idea what colleagues consider as "good translation". But, in many ways and for many reasons, they are not really conclusive or representative.
The only way to rule out "mock or bogus submissions", abuses and vote rigging would be to conduct the contests indoors and have the voting done by a jury - of whatsoever qualification.
But that would steer the event into quite a different dimension - and is just not a practicability.
Nonetheless, the problem that you describe, is relevant for a shortcoming that I observed in the present tagging system of the qualification round. As the tagging options go from 1 to 5, for "accurateness" as well as for "good writing" [a category inadmissible for me when independent from accurateness], giving the lowest note would mean giving 20% of value - which is 20% too much for submissions without any value at all.
It could help the process of early elimination of "bogus submissions" if the tagging range would not go from 1 to 5, but from 0 to 5 [and if, btw, the category "good writing" would be scratched altogether]. - I for instance am reluctant to give a 1 to a nonsense piece. But this means that a piece with no tags at all could stay in longer than pieces that got several 1 point tags. - -Therefore it might be useful to have the option to give 0 points.
[Edited at 2009-10-24 16:25 GMT]
| | |
|
|
|
| Machine translations in the contest | Oct 24, 2009 |
Why so much fuss about machine translations in the contest?
I wouldn't waste my time with them... Just vote for the best! As long as "the best" is a human translation, there is hope in the world for our profession
| | | | | I did not mean screening all entries by staff | Oct 24, 2009 |
But I think it would be asking too much of the contest organizers to do anything like a screening as you are proposing it here. After all, this would mean no less than to check each and every entry in each and every pairing - clearly an impossibility.
I agree it would be too much to ask. That's why I did not ask or propose that.
However, what I think is doable, is when there are 3 submissions in a certain pair, to NOT move that pair into final voting right away automatically, but do one of these two things:
a) Check if any of the submissions is a straight machine translation and if yes, either remove it, or leave it in but keep the pair open for more submissions
b) Move the pair into Qualification phase, and let the audience disqualify the entry before moving the pair into the final round.
Your proposal for the "0" mark is good, too.
All the more as the case that you are describing, that someone will, out of sheer mischief, put in a completely unedited MT piece will after all be rather rare.
I am not sure. Take a look at the German-Hungarian pair. Since voting is closed now, I can say that the 3rd place entry is one of those I talked about. It is a straight copy of what Google Translate gives you, if you put the text in. If you look who submitted it, you will see that this person signed up in September, admits that "he" is a software, and from the list of the languages "he" apparently works in, I am sure I am not alone to think that all the machine translations in this contest were the works of this master of all languages.
I am not sure whether there was any other pair where the MT entry got into the finals, so I cannot check if the same contestant was the culprit. However, staff can, and they would know the identity of the person as well, as he/she paid for full membership.
I know it is against the rules to speculate who submitted what entry, but this person proudly displays its user name of the entry in question, which makes me think that there must be a point he/she is trying to prove.
In addition, staff should be able to see when a single user submits entries in multiple language pairs (and I am not talking about 2 or 3, but many more), and that should be a signal that something may not be entirely right.
I know many people say "what is the big deal, just ignore it", and they do have a point, but I think it is really demotivational for participants (both for those that submit legitimate entries, and also those that evaluate them), even if the whole contest is not to be taken too seriously.
I really would like to hear from staff, what their view on this is.
Katalin
| | | | | That can be a genius | Oct 25, 2009 |
Basically, I got the same view by Heinrich. If one knew little about one or more the languages pairs in the contest and finally manage to have the perfect expression, or we say 'translation', no more words have to say. That's a genius who can make it like that.
But the fact is most of the translation text are not so easy to be cracked by the outsiders not in that language pairs. For example, German, rather the French or Spain, or Italian, etc., can hardly be used to provide a satis... See more Basically, I got the same view by Heinrich. If one knew little about one or more the languages pairs in the contest and finally manage to have the perfect expression, or we say 'translation', no more words have to say. That's a genius who can make it like that.
But the fact is most of the translation text are not so easy to be cracked by the outsiders not in that language pairs. For example, German, rather the French or Spain, or Italian, etc., can hardly be used to provide a satisfied translation text (target), unless the translators do have the enough understanding of German and the expression skills.
Machines, or that sort of stuff, probably used by translators at different levels. Yet, the really perfect target text still needs strenuous efforts before it can be provided.
LIN Shu, a legendary Chinese translator in early 20th century, who did a lot famous translation works including the Chinese version of Hamlet by Shakespeare. but amazingly, he had little English knowledge. Apart from with the help of others who knew more about English and read for him, more importantly, He made it by virtue of his powerful language basis that did count in his 'translation' process (regardless of some funny context errors in his versions:).
To some extent, he's a genius. Nonetheless, this kind of people is not always found here and there in the world. To the end, perfect translation only comes from those who really have got the hang of it.
Thanks~
[Edited at 2009-10-25 11:01 GMT] ▲ Collapse
| | | | Luca Tutino
Italy
Member (2002)
English to Italian
+ ...
| The ethical dilemma comes later | Oct 25, 2009 |
The problem is not wether you do it or not, but rather wether saying it (if you win) or not.
It might be unethical to hide the fact that you used MT, but if you declare it openly, even if after the end of the contest, you are just proving you are smart and MT is better than we thought.
Luca
[Edited at 2009-10-25 10:59 GMT]
| | |
|
|
|
| more about practicabilities - | Oct 27, 2009 |
Katalin Horvath McClure wrote: All the more as the case that you are describing, that someone will, out of sheer mischief, put in a completely unedited MT piece will after all be rather rare. I am not sure. Take a look at the German-Hungarian pair. Since voting is closed now, I can say that the 3rd place entry is one of those I talked about. It is a straight copy of what Google Translate gives you, if you put the text in. If you look who submitted it, you will see that this person signed up in September, admits that "he" is a software, and from the list of the languages "he" apparently works in, I am sure I am not alone to think that all the machine translations in this contest were the works of this master of all languages. I am not sure whether there was any other pair where the MT entry got into the finals, so I cannot check if the same contestant was the culprit. However, staff can, and they would know the identity of the person as well, as he/she paid for full membership.
But essentially, Katalin, you confirm yourself that this case of a probable jester or mischief-maker is rare. If it is what you describe, that one person alone has supplied a lot of such bogus pieces, then it is one person alone, and that is not much. -
I still do not see what is the "ethical" aspect in this kind of bogus intervention, because such a piece can never "steal" the win from anybody, unless it would be voted by "colleagues" into first place. But that, for all practical reasoning, can only happen if the "bogus submission" would be flanked by "bogus voting". And that, "bogus voting", would indeed constitute an ethical problem.
When I see, as I just did, a winning piece with at least 5 grave imprecisions, with a completely arbitrary fancy invention that has nothing to do with the source - and, btw, is of a kind that even an MT would not have "dreamed" of - I can hardly think of anything else as that there has to have been some "bogus or cheat voting" involved in order to get such a piece on top.
I know many people say "what is the big deal, just ignore it", and they do have a point, but I think it is really demotivational for participants (both for those that submit legitimate entries, and also those that evaluate them), even if the whole contest is not to be taken too seriously.
There you say what it is. - BTW, I would not say that it is demotivational to compete against these pieces. On the contrary, as long as there is "normal" voting they will never win anyhow.;-) - As Maria Diaconu has said correctly - "Just vote for the best!"
It remains that you are right in saying that such pieces can lead to voting procedures in pairings with low submission that would not have a voting without them.
I really would like to hear from staff, what their view on this is.
Maybe it could be practical for the organizers to introduce into the tagging system, in addition to providing for the possibility of a "0 points" tagging, a checkbox called something like "Suspicion of MT". - This could help them to get early feedback by competent people for the relevant pairings, have then a look into the data of the poster, and, if there would be a case as you described here, with someone having put in a series of such nonsense entries, really remove them early on, as you propose.
[Edited at 2009-10-27 01:05 GMT]
| | | | | Clarification of the original question | Oct 27, 2009 |
Roland Nienerza wrote:
I still do not see what is the "ethical" aspect in this kind of bogus intervention,
Roland,
What is being discussed in the latest postings is very different from the original posting.
Recently the discussion shifted towards the submission of UNEDITED machine translation pieces, and that is different from what the thread was originally about.
My original posting described a theoretical scenario involving MT and 3 languages, where the person does not know Language A, uses MT to translate the text from Language A to Language B that he/she understands. Then, produce a "translation" in Language C, that is his/her mother tongue, and submits that piece in the A-C language pair. For the sake of argument, I asked to assume that the submitted "translation" is actually very good, in terms of the text written in C is of high quality and also accurate in terms of the content. My questions were about this scenario. (I will not repeat them here.)
I do think there is an ethical dilemma when one submits such a "translation" for the contest, and there is a point where it would definitely come up: there is a checkbox to state that the translation is his/her own work. Would in this case the "translation" from A to C be the translator's own work (when he/she does not even work in that pair)? Could the translator say that? Would it be honest to say that?
In my mind, the originally described scenario is not really "translation", it is more like when people hire professional writers to write novels based on their ideas. The resulting book may turn out a bestseller, so who should get the credit? The writer would have never written it without the ideas or scenarios created by the other person, so if the writer gets credit, it is like he/she stole the idea from the other person. On the other hand, the person with the idea would have never written a bestseller by him/herself, so if the writer does not get credit, that is strange, too.
(I think there is a parallel between that hired writer and the translator that works in the B to C pair, and the "idea person", who presented the ideas to the writer is the MT tool.)
Katalin
| | | | | Yes, there has been a slight shifting of the topic - | Oct 27, 2009 |
Katalin Horvath McClure wrote: Roland Nienerza wrote:
I still do not see what is the "ethical" aspect in this kind of bogus intervention, What is being discussed in the latest postings is very different from the original posting. Recently the discussion shifted towards the submission of UNEDITED machine translation pieces, and that is different from what the thread was originally about.,
You are right to point out, and I actually had been aware, that the discussion was lately more about unedited - although I thought more of "un- or just slightly edited" - MTs, while the topic at the start of the thread concerned heavily edited material.
Although I had also replied at the beginning - I later did not read all the aspects so far discussed in the thread.
I have to say that what you are recapitulating here - as a sort of "détour","triangle" or "playing bank" approach is of course a rather weird procedure. But then I can only come back to what I had said before, i.e. that for me such a trick is more a foolish attitude than an ethical offense.
I think others have developed more in detail that with this procedure one might get a nice worded product for language C. But from all that I saw the misinterpretations in MT are often so grotesque, that putting them into correct wording still will lead to a grotesque translation.
But, in a curious way, we are coming here back to the topic that I had brought a year ago into the discussion, and in which you had intervened too, i.e., in my view, the inadmissibility of giving any merit to a translation on the ground of "nice wording" or "nice reading" - independently from source. -
"The sun is shining" is a simple phrase, but could, for whatever reason, be considered as "nice wording". - But it is complete nonsense, if, by exaggeration, it is supposed to be the translation of a source sentence that actually means "The weather is fine" - or even "They live in a new house". - And it would be preposterous to give "good reading" for the "The sun is shining" - when that bears only vague or rather no relationship to source.
On the other hand, and here, Katalin, I think we are really disagreeing, I would say that if the text produced by this "triangular procedure", implying, as you correctly said, the input of "borrowed or bought labour", would, however miraculously, be a palatable translation, it is in my view not "unethical" to submit it [indeed, in my view it is not "unethical" to submit it as a nonsense translation] - for the simple reason that there might be people "out there" who really "work" that way. - The proof of the pudding is in the eating. As long as there are clients who buy such stuff, it is a matter between those two parties. If the merchandise is bad, it is for the clients to seek remedy. - If it is good - however improbable - the client would have no reason to complain.
In my view, and this seems to be different from yours, the contest is about the most correct [and hardly the best reading, btw] translation, not necessarily the the most correct [and hardly the best reading, btw] human translation. If a piece concocted in a "triangular procedure" would outshine the human translations, it would be the best piece, and would deserve to named such. Voilà. -
There is a checkbox to state that the translation is his/her own work. Would in this case the "translation" from A to C be the translator's own work (when he/she does not even work in that pair)? Could the translator say that? Would it be honest to say that?
This is indeed an interesting "ethical" and even "legal" question. - "Ethically" it will rather not be quite "his/her work" - although, for sure, the man/woman will, after all, have supplied "the good wording" him/herself. And I note with some relief and amusement, Katalin, that within this puzzle or conundrum you at long last admit, at least implicitly, that "good wording" can be completely detached not only from "good translation", but from "translation as such". -
"Legally", I would say that this is clearly his/her work. - The contest is not about "checking vocabulary knowledge". - We all use dictionaries and databases, in work as well as in contests. Are the translations thus produced "unethical", let alone "illegal"? - Of course not.
Otherwise your question could be quite easily solved by declaring the events not just "Translation Contests" - but "Human Translation Contests". And, to preclude any "unethical", "illegal" or just "cheating" behaviour, they should be held indoors, with or without access to lexicography and reference. - Well?!
But as I said. - The more problematic, even "ethically" problematic aspect is the voting. But this too has been discussed, and there will be hardly any way to improve this. The best is still, what so many colleagues, contestants and not, have said -
"Just take it easy."
[Edited at 2009-10-28 00:43 GMT]
| | | | | Pages in topic: < [1 2] | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » Ethical dilemma about the contests | Protemos translation business management system | Create your account in minutes, and start working! 3-month trial for agencies, and free for freelancers!
The system lets you keep client/vendor database, with contacts and rates, manage projects and assign jobs to vendors, issue invoices, track payments, store and manage project files, generate business reports on turnover profit per client/manager etc.
More info » |
| | 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 » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
