Lost opportunity for learning in Duolingo
Hello. I use Duolingo daily to practice several languages.
I believe the review system misses an opportunity for a much more effective learning and this could be achieved with a simple feature.
When we review a unit we have already studied we are presented with the same sentences we have already solved previously, namely translation from target language to English. It is very seldom that we get to translate from English to Target Language.
I believe it would be much more challenging to have translation from English to Target Language as a default for reviewing lessons that one has already studied.
In that way the reviews would differ from the normal study and a proper learning would be guaranteed.
I totally agree with you on this one. A balance between the two would allready be great. But from english to the target language would be much more helpfull as it will help you to more easily use it in a real life situation. A Balance between the two would be perfect because then when hearing we will understand and when wanting to speak we will know how to say it
I have found the more often you strengthen a skill, the more often you get English to target language translations.
The converse is that if too many/complex English to target language translations are introduced too soon, it can be a demotivating experience. The trick (for Duolingo or any other foreign language course) is to get the balance right and make things sufficiently challenging at the right time, but not too soon.
Getting this balance right while maintaining user engagement is a big priority for Duolingo. Unfortunately, translation into target language sees people giving up quicker.
However, the balance does tilt as you show Duo that you know the vocab, I think in particular if you ace the questions where a sentence is read to you in the target language and you write it down. In the trees where I'm further along in actually learning the material there is quite a good amount of translation into target language (easily enough that I have almost never used Strengthen Skills in my reverse trees since I'm just translating into a language I know well), in others almost none bar the transcription ones.
Some people (I am tentatively one of them) also think the balance might set more intelligently once you've finished the tree, an argument for going through with somewhat less review than you might otherwise, I suppose, and then diving deeper on a second time through.
It is interesting that you say that people give up quicker. In fact what I'm arguing for is to have the sentences swap (FL -> ENG to ENG -> FL) when you are doing the 2nd or 3rd reviewing so that it is material you know very well already. I am not sure what you mean with reverse tree, can you explain?
Completely agree with needing less review once you have finished the tree, but actually even before that. I mean, if I speak good Norwegian is not like I am suddendly going to forget how to say "Hello" and "Thank you" so why should I ever review that?
"Reverse tree" is the term commonly used for e.g. using the English tree for French speakers to help oneself (as an English speaker) learn French, like Thomas.Heiss and yibemajam mentioned.
I don't think things are as simple as "doing the 2nd or 3rd reviewing." For one, I have seen there be up to about 4 times as many sentences in a skill as one comes across just in the lessons. Unless one does quite a number of skill specific strengthenings, one won't even have seen all the sentence available a first time, never mind a second. And the frequency with which each individual sentence is presented comes out of the alchemy of Duolingo's algorithms. In the Russian course a sentences can be presented to users as much as 100 times more often than a very similar sentence. So, intuitively, one should base decisions on translation direction on a sentence level, so all or nothing rules aren't going to be optimal. "2nd or 3rd reviewing" doesn't have any well-defined meaning in general. One could of course base such decisions on overall progress through tree, overall XP level, overall "success" level with translations. This is what Duolingo is trying to do. Do I fully agree there should be a user option to heavily prioritize translation into target language? Yes, but do I think it should just switch by default, no. It should switch contextually as it does now.
If you already have experience in a language, I'd recommend going pretty quickly through the tree and then going back to strengthen skill by skill, focusing on timed practice. I think you'll quickly see an increased percentage of translation into target language.
I agree with you piguy3, especially concerning strengthening skill by skill and focusing on timed practice. I think you'll have an increased percentage of translation into the target language the further down the tree you go, and the harder the challenge becomes especially if you had no previous knowledge of or very basic knowledge of the target language.
For me the only language I had studied was French and some basic Spanish, and virtually no German. I therefore find it especially challenging the further down the tree I go for German right now. I have learnt virtually everything I know about German and most of the other languages on duo - cases, sentence structure etc, especially from others in the commentary section. I therefore encourage everyone not to give up if you think it's too easy - French was relatively easy for me; try to finish that course and try another language you had never studied and see how challenging it becomes. Good luck everyone.
piguy3 Very interesting points:
Just some remarks: "one won't even have seen all the sentence available a first time" I believe language learning should not be completely mnemonic, so the fact that you haven't seen a sentence should not hinder you from being able to make the translation. Learning a language is not about being able to answer to Duolingo's quiz, it's about using the language in practice.
You talk about the alchemy of Duolingo.
What I am kind of ventilating in this thread is that I perceive the Duolingo alghoritm as much less refinated than what I'd expected, since I get presented with the same sentences again and again to a higher degree than what I think I need as a learner.
My Duolingo experience can be frustrating because the system pushes me to review things in an ineffective way for the reasons I mentioned above. For the languages I am actively learning by just using Duolingo for 5 minutes a day, like Spanish (I'm well aware this is not an effective way to learn a language, but it is not a goal I want to prioritize more for the time being), I'd rather have the system challenging me to produce sentences in the target language rather than the opposite. I use time practice daily because it gives me the impression that I am reviewing the topics faster and I can let the time run out and get XP points for the questions I answered correctly, thus maintaining my streak, which is my main motivator.
Like I said, I, too, wish for a user option to facilitate translation into target language. I think it would be a very broadly popular feature among the kinds of people who care enough about language learning to write paragraphs about optimizing their process thereof :)
But I venture to say the vast majority of Duolingo users are not like you and I. They don't actually enjoy being challenged too much. So Duo stays the way it is on this point despite the hundreds if not thousands of times variants of the suggestion you've made are put forward. In part I suspect the people who care enough to notice they're not translating into the language they're learning very much figure out to just do a reverse tree.
Duolingo is no panacea. There are good arguments to be made for Memrise, which I think indisputably does a better job on strict spaced repetition. So if you're learning a language close to one you already know, the fact it's word based and not a good medium for grammar may not be much of a hindrance. lingvist.io is also an excellent choice for the languages available there. I generally prefer it to Memrise because it at least provides whole sentences to look at.
I believe you can already do this in a way. I am currently learning English from French and there I have to translate English phrases into French. I think if you want to try to translate from English to the target language, just start a new course in Duo, indicating that you speak the target language and then indicate you want to learn English from that language. I don't know whether it will work for all the courses, but that's what I did for French. Hope this helps Tomasso.