imperfect translations corrected by another

Do I get partial points for translating something imperfectly (because I am not a native speaker of the target language), when another translator, who does not speak the source language but is a native speaker of the target language, takes my imperfect translation and produces a perfect translation from it?

This happened with a translation I did from Turkish to Greek, which was rejected, but it was later corrected by a member who doesn't list Turkish as one of her languages, and this translation was then accepted.

Laura 3

I'm not sure to get everything but you must not translate when you are not fuent. Because in a perfect world we would not have to reject any translations, and we shouldn't have to edit any translation too. As experts are volunteers, their time is not unlimited. That's why we ask that only fluent speakers do the transations.
Does it answer to your questions ?

Not really. That means that a translation from Turkish to Greek would have to be done by someone who is fluent in both languages, and I don't know if we have that here. Between Irini and me, we got the job done.

My reading ability is fluent in both languages, although I make a few small mistakes when writing, but a native speaker of the target language can easily figure out what I meant.

Laura 3

Only the target language has to be a fluent/native one. In your case you have one rejected greek translation, maybe I guess your greek level is not good enough to translate into greek but good enough to translate from greek.

Keep in mind that the goal is not to provide only the meanings but also perfect/fluent sentence constructions. Imagine that you request a translation for a letter you want to send to a foreign friend, would you like it to be perfect or only understandable with some mistakes inside ?

I agree that in some cases, for example a translation from turk to greek, there are very few (none?) translators that can do it here, and an association of 2 translators like you and irini is the best solution. That's also why it's asked to require an english translation as a gateway from the source language to the target language.

OK; I get it. Merci pour l'explication.

Laura 3

I got a follow up question:
I always see the option: "Add another target language". I never did that, but does that mean, if I see e.g. a request from Arabic into Albanian, is it a good thing to add an English translation even if it is not requested because chances are much higher that there is someone on here who speaks English & Albanian and therefore the text has a higher chance of getting translated?
Thanks!

yes, but you will spend some points
That's why we encourage (in the submition form) people to ask for english.

I have a suggestion. I realize the importance of having accurate translations done, based on your example, but it seems to me the site would attract more would-be translators if it was more open to any translation. Please allow me to explain.

I have acquired a fair level of efficiency in Japanese, but if someone asked me if I am fluent I would say no. Despite that, there are a number of translation request that would be easy for me to complete, even at my current level, but when I first signed up I ended up not doing any translations because of the strict disclaimers about fluency.

My suggestion is that translators be able to give their translation a confidence score. The higher the score the more confident they are. As a reviewer scores the translations, the system can make its own interpretation of the users confidence score and let people who are using the translation know. In this way I believe it would encourage more users to translate and avoid putting off users who may be perfectly capable of translating certain things.

Hope this helps.
I, for one, was fairly confident in some translations that I wanted to do when I first signed up for the site, but ended up not doing them based on the disclaimers and agreements posted when doing the translation.

This is a good idea. The other thing is that sometimes the requester only wants to know the meaning of the text, whereas other times the translation will be used in a publication or something like that, so different degrees of fluency in the target language are required for these two different cases. Sometimes the requester specifies this, by saying something like "I just want to understand this text" or "I want to put this on my website" or something like that. This kind of note is very helpful to the evaluator (especially if the translator has translated the note).

Interesting thoughts
Actually, the warnings about fluency are mainly adressed to new translators. Without this warnings we would not have enough time to moderate the wrong translations.
Experimented users know that they can do translations depending on the difficulty, even if they are not totally fluent.

The problem of the confidence score is that we only allow only one translation by request. If somebody submit a translation with a confidence score of 1/10. The native speaker who come just after this submition will not see this request, until the translation is evaluated (several days). This would increase the waiting period for translations done. This is interesting remark but I don't think I will do that.

Nevertheless, kafetzou, I think like you that there are 2 kinds of translation request : those for meaning and those for publication. The requests only for meaning, would not display the usual warnings. I will add this in the todo list.

Rather than "meaning" or "publication", perhaps it would be more helpful to grade translations as either "approximate" or "exact". General idea or precise meaning. Rough or smooth.

Yes - that would work.

In my opinion, 'Rough'/'Smooth' is nearly the same than 'meaning'/'publication'.

'General idea'/'Precise meaning' and 'approximate'/'exact' would be ver interesting too but would not be easy to implement. Each time a translation is accepted the numbers of characters of the source text and the translation are taken into account in the these statistics. I guess "general idea" would result in shorter texts and that would be very hard to keep accurate statistics about numbers of charcters which are used to calculate the cost (in points) of each translations. I guess this kind of request would better fit in the forum. I will keep this in mind anyway.

A rough translation should be worth fewer points per character than a smooth translation.

I don't quite agree that "Only the target language has to be a fluent/native one."

I have seen translations done in flawless grammar that totally miss the whole point of the original text.

Those translators seem to have learnt the "textbook version" of the original language and didn't understand the colloquialism contained within the original text...

(Example, "We are all one.". Does it mean "Each of us is an unique individual." or "There are no differences among us"? The "textbook" guys wouldn't be able to decide which is the right choice.)

I would rather the translator understands the original text perfectly and give a grammatically bad translation. The experts will be able to catch such "bad" translations and make necessary changes...that's where we can use the "Accept rating 1" option. ("Accepted" because the meaning is correct. "Rating 1" because it's totally not fluent.)

Yes "Accept 1" is OK but "Accept 10" is better. I agree meaning is as important as grammar. If you would ahve to edit all translations, I'm not sure you would ahve time, that's why I prefer to limit the translators to their native languages.

In my opinion flawless grammar is the first condition to be able to evaluate the meaning. Sometimes the grammar is so bad that I can't even guess the precise meaning (on some arabic-french translations done by marhaban for example).

As you maybe know, now, we have a tool to to be sure the meaning is perfect too. When you're not sure about this, just click on the button "I don't understand the meaning of the source text" . Then after a couple of days you can decide to accept it or not, depending on the votes of the cucumis users.

There is also a feature that will appear sooner or later on cucumis about "meaning only" requests. This request will not require that you speak fluently the target language.

Sorry about jumping in here, but jp I think maybe you missed the essence of what samanthalee was saying. She was saying that you need to be a very high-level speaker of the SOURCE language as well as the TARGET language in order to do a good translation.

I agree with her, but not always. For example, I was able to translate that recipe that was requested recently from Portuguese, which I don't even speak. On the other hand, there have been translations from French, which I do speak, albeit quite imperfectly, to English, which is my native language, but which I felt unqualified to translate because my French wasn't good enough, never having lived in a francophone area.

I think it would be better to have two different ratings: for the source language and for the target language.

I agree, but how would it work? Maybe also based on points earned for translation FROM various languages.

yes, maybe.