The Immersion experience can be quite frustrating
One can get quite attached to articles in immersion, particularly when one has been translating it for a while.
Unfortunately, with the 'new' system of translation where you edit the latest translation of a sentence, several issues can arise:
- 'Edit wars', which have been mentioned a few times, are the most obvious.
- Google Translate. People actually copy and paste the Google translation in. This isn't helpful as it hides DuoBot's translation (which can be a helpful source of inspiration).
- People overwriting correct translations with wrong ones. I'm not sure what motivation these people have. It could be abuse, lack of language skills, or just general ignorance.
- People keep editing the first few sentences. They don't even bother getting to the later paragraphs of a very long article.
- Reverting is painfully slow. One has to read the new suggestion, decide whether the last one was better and scroll down the page (excruciating if there are many revisions). If you want to put an explanation for your revision, you have to copy the one you want to revert to, edit the translation and paste it in.
I have been slowly working my way through this article for a few weeks: http://www.duolingo.com/translation/48639cd282fb77bfbb789b2a95297d50. On occasion all of these problems have arisen. I'm not sure what to do when one of them does. 'Reporting a translator' doesn't seem to do much, and I'm not even sure when it should be used.
I feel like moderators for individual articles would be helpful. An unbiased arbiter might be able to resolve some of the translation disputes. Discussions would also be very useful. They've been mentioned before.
I'm sure this is not an extensive list of the problems people face in Immersion, and overall I enjoy the experience (finishing an article is very satisfying), but I hope that voicing the frustration I feel might provide valuable feedback.
I agree with all your points. I gave up on immersion after I spent about an hour diligently translating what was quite a complicated article from idiomatic French to idiomatic English only to find a few days later someone had crossed out my version and replaced it with one which was a word for word literal translation that made no sense whatsoever (I think they must have done it by hovering over every French word and typing the first English word given). This translation was then marked as correct and checked by the community, even though the first line they had done was something like "time are of animals".
They're doing a kind of point farming I guess. I see how it would work but I have no idea what they gain from it!
I just worked through several paragraphs of something showing as untranslated and when I went back to it to look at something, found earlier translations from ten minutes before displaying! So sometimes one can translate without knowing that someone else has had a go at it. Perhaps it shouldn't always be taken as criticism if another version appears.
Yes for edition in the first hour after the article has been uploaded. But when really wrong editions (google translate type for example) are made more than some an hour after the previous, it's clearly not due to a simultaneous edition (simultaneous edition are rare but indeed shouldn't be considered as a problem).
I usually try to get round this by not looking now - it is too irritating when someone 'corrects' perfectly good English to something that is a literal translation and grammatically incorrect! I'm here to fix my lamentable language skills, not quarrel with someone trying to do the same.
"grammatically incorrect" - yes, that is a big problem with DL's business model. There is NO vetting for competency in the translation language whatsoever! If someone is translating, say, from Spanish to English, how does DL know that that person is an excellent English speaker and writer, versus someone with a sub-par grasp of English grammar? Part of the problem is that points are given for translation efforts and some folks will do ANYTHING to get points, even paste in Google translations. How SHOULD this work? Maybe points should only be awarded for translations AFTER a complete language skill tree is completed? DL might end up with fewer but better translations. I've been trying to puzzle out a solution and don't really see one yet.
You can't learn a language by translating individual sentences with no context, no audio comprehension and no translation or reading comprehension passages, though. The translation is part of the course. Without other types of more complex exercise the tree means very little. You can get through bits of it without having really assimilated the material (not least because the grammar is nowhere near sufficiently explicit). My relative 'levels' on the trees in no way reflects the relative levels of my comprehension in the different languages - the one I know least about is the one that's highest up the tree because that's the one I'm trying to learn.
The only real solution I see (but which is far from perfect) is to give points according to votes for the translation. Maybe one point for translating (except may be for the first who did the translation, but still not perfect) but the rest with votes. And one points when voting
- no incitation to translate at any price (at the price of google translate for example)
- really good translation would give more and more skill points to the translator
- We should be able to vote for all versions, so it needs changes.
- it would be so long to vote for all so users probably wouldn't do it and at the end nobody would get points for translating
- you could earn points but then lost them if downvoted after: quite disturbing seeing your skill points moving up and down, and also evil-minded users could use it to put down points of another user
- some users, in order to earn free points, could, when seeing a very good translation, edit by erasing one character then immediately changing back their edition to add again the character: so the translation would be theirs (also)... Well a way to avoid it would be that if you enter a translation that have been put before (exactly the same, character for character), then it replace your name by the one of the original translator (maybe adding: "revert by [your name]") as if you had reverted to this translation.
Well it has more "cons", so as I said: "far from perfect", but I don't see other solution (hope Duo team will) to make translating not attractive for users that want skill points at any price.
The after completion is an idea, but not sure it'll change anything: user that would do anything to earn skill points will still do anything to earn them after completing the tree.. and also it's not motivating for learners to earn nothing when they try to practice by translating.
I agree that it's frustrating to see that someone has incorrectly corrected your translation, and I am torn about what to do when it happens. Part of me never wants to look to see what has been corrected, but at the same time, that can be one of the most useful aspects of the site because you might get help from someone who has a better grasp of the language or of the subject of the article. It might help if the comment function were made mandatory for any edit after the initial translation. That way, people would be forced to explain why they think their translation is more accurate. It isn't a foolproof system, but it might make people stop and think.
I think everyone reacts in different ways, though. I have been far more exasperated by an ill-informed comment than by someone discreetly tidying up a typing error that I've made without comment or just re-translating because they want to translate for themselves. I think there will be as many perspectives as individuals. Short of figuring out how individuals prefer to interact and treating them accordingly (and I shall try not to change your text without comment, as I know we read the same articles!), I don't know what any of us can do for the best.
I have a concrete suggestion. Can there be a text box at the top of articles for immersion translation in which frequent or especially skilled translators can leave practical notes for all? For example, one article I've been working on constantly has less fluent translators wrongly "correcting" Brittany (province in France) to Britain (nation across the English channel from Brittany). It might help to be able to just write a simple note at the top of the page explaining how this word should be translated. Similar cases abound and might be easily solved by means of this expedient.
Not sure it would be read anyway: comments aren't (even if I know there not so visible, if you don't want to see them).
IMO, it would become quickly a chat where everyone would let there her/his opinion etc... Or you have to select who can write there: who will decide? on what criteria? etc... at the end would be quicker and simpler (IMO) to designate moderators for Immersion (but I'm not a pro of that).
Perhaps when you select something to translate, Duo can make a copy of it that you work on. That way while you can see others' translations, and suggestions, if someone goes stupid your copy is unaffected. Then when it is completed you can submit it, where i guess it would be integrated into the final copy sent to the client? Something like that. Like an instanced dungeon in an MMO.