Why not improve translations rather than write translations from scratch? Bots could provide the initial ones.
You would only obtain credit for improving a translation if others judge it to be an improvement.
Moreover, if your translation is improved by other(s), then this would provide excellent feedback to you.
All translation could be done in this way so that learners don't need to write sentences from scratch.
P.S. To see how this might work, consider a human guided genetic algorithm:
Would people still get to see the original sentence? My problem with this idea is that people would be more focussed on writing grammatically correct English sentences than actually understanding a language on their own. Your suggestion might help to improve sentences, but it defeats the purpose of language learning.
I agree with you Lenkvist.
Another argument against this is that professional translators don't use bots. For them it is faster to translate without bots than correct bot-ched translation. Forgive the pun. :-)
I agree to an extent with some suggestions but i.e improving translations but think that it should be more like the old system where people translate a passage then vote on a number of translations stating wether they are good bad or somewhere in-between.
Another suggestion when someone masters a language give them the option to translate the entire original or a few paragraphs from the original in one go as this would reduce the frequency of people translating an ambiguous sentence incorrectly because the context would be provided without having to look at the original