Too much translation from learning language, not enough translation from native to learning language
Is there any way to adjust settings so that I get more practice translating from my native language (English) into my learning language (German)? I find that almost all of the questions ask me to translate from German to English. Consequently, my comprehension is growing pretty well, but my ability to put a sentence together in German is still very weak. I think there is probably not a way to adjust this setting, though I might try changing my native language from English to German and "learning" English as a workaround. This is more of a suggestion for the developers to consider balancing the ratio of the direction of translation.
Thank you for starting this discussion again. I am now doing the refresh French, and in twenty sentences maybe two are English to French. Of course it is very easy, I do it very fast, beat the clock every time . It is impossible to learn a language properly by translating from target language to your own (English) language. This has been commented on several times, but it seems DL doesn't want to change it. However, they could make a course 2, where everything is from English (or Spanish, Italian etc) to your target language. That would be a challenge, and much more interesting.
If you do the reverse tree, turn off the sound. And don't refresh. Then most of the time you'll write German.
I've found that repeating the same lesson helps. If you progress really quickly, the algorithm always thinks you are not ready for more advanced stuff. Slow down, repeat lessons and strengthem them and you'll see, you'll start to get translations into your target language soon enough. Then if you make a mistake, it goes back to easier excersices for a while (might just be one sentence). I did the whole French tree like this, and I think towards the end I started to feel like I don't always get enough translations from French to English, since the algorithm had me on a higher skill level, or something. Then there were some individual skills. like the subjunctive, which I think are difficult for most learners, those are usually set to give you FR-->EN exercises for a good long while. I think it was only the fourth or fifth strengthening that I finally started to get a balanced amount of those.
Ah, so the recognition vs expression ratio shifts as you progress further? That actually makes a lot of sense from a learning-curve perspective, even though it might feel a little unbalanced at first. Appreciate the information.
I can confirm chirel's observation: The first time I revise a Russian lesson, I hardly have to do any translation into Russian at all, but the nth time I do it (on the website -- the apps are different), I sigh because there's so much translation into Russian.
i think this may have been what caught me out when i got to perfectives the second time throught the russian tree. i had been doing well up to then so it gave me difficult questions which i failed to master so now it has become easier again.
i have also noticed i'm in an a/b trial giving more translations from from language. since i am doing the french from german course in order to practice german but am now getting more to translate into french this is not as welcome as if i were doing another from english course. however my french is benefitting! the ratio has definitely improved, they are working on it:) but not everyone will see the change yet.
I only noticed this as I started doing the French tree. I'd decided to do every lesson at least twice and strengthen the skill the next day and the day after that (and more often, if I felt I needed it).
The difference to the Swedish tree was clear. With the Swedish tree, I had rushed through, and only got EN-SW translations every so often and thought it was annoying. I even made flashcards of the sentences so that I could practise EN-SW direction enough.
Interesting. Have you repeated the lessons several times before moving on?
II started last year in January. The first time I did the French tree I tried to keep the tree golden, so I refreshed when I had to. I knew French before I started so I finished the tree rather fast. Then I did the reverse which took more time. In November I took a break for two months. When I started agsin I had to refresh even rhe first lessons. Now I refresh with the time limit, it's fun but very easy. I write mostly English. English is not my native language, so I have to consentrate but it's not a big challenge. When my French tree is golden I will refresh the reverse with the time limit, it's more fun.
This is how I did my Swedish tree which didn't give me enough EN-SW exercises.
Try this: Repeat every lesson at least twice (or even three times, I did have prior knowledge in French, so if you make more mistakes than I did, I bet you need to repeat more). Then strengthen the skill two days in a row even if it is golden and if it was a more difficult one, strengthen more than just one even if it is golden. If you proceed slowly enough for Duo to believe you know what you are doing, you'll start to get more translations into your target language. If you breeze through the tree because you believe, you know enough, then you clearly won't.
At least I tend to forget how to spell яблоко if I don't repeat it every now and then. So I'm not exactly repeating exercises to please Duolingo.
Do note that this seems to be an A/B-test, so it could just be that you and I are in different test groups, since I've noticed this difference in the ratio of questions depending on my own learning strategy and you haven't. Let's hope that the test shows that giving more translations into the target language as one practices more leads to better learning, in which case it will be rolled out to everyone.
How exactly would I "please" an algorithm? :D
I do it because I've noticed I learn better that way. I pay attention to every detail so later I don't need to juggle a bunch of things I almost can use or can use most of the time, but not always. As a bonus, I noticed I get more translations into my target language.
Here's something you are missing. When you breeze through the skills (even if you make no mistakes) you don't encounter all the words. Those words drag your skills down so you need to strengthen the skill soon and often. If you see all the words in a skill AND make very few or no mistakes, the skill stays golden longer. And later on, when those words come in more difficult sentences or skills, they are not new to you any more (if they are, and you make a mistake, they drag the original skill down, not the one where you are). I've had to repeat a lesson more than ten times and I still didn't get all the words in it. You just will miss a lot of words if you progress quickly. If you don't mind your skills not staying golden, then it's not a big deal. I decided I want to get most words in the skills where I'm supposed to get them and to progress slowly to make sure I don't just think I know something because I knew it twenty years ago. As a result my tree stayed golden up until I was at the third check point and I got a nice amount of EN-FR translations, which I really liked.
This doesn't happen for me, even on lessons I've done more than ten times already.
Does this happen to you folks both with redoing lessons and timed practice? Because I mostly do the latter, and I don't really see any change in behavior. I often do timed practice on the same lesson five or six times in a row and it stays much the same ratio. Sometimes I've noticed a pattern where I get a few more translations into the target language every second time.
I haven't touched timed practice since I switched to the йцукен setup... :-)
LINHARS also mentioned doing timed practice -- I think it always used to be the case that timed practice, although a good way to regild skills, was actually a very bad way of keeping those skills gold. (Duolingo's algorithms keep changing, of course.) But the same might be true for determining what the mix of questions is like.
That sounds really frustrating :( and very different from my experience, so either it's an A/B test or all of this is just a coincidence.
It could indeed be part of the A/B test, since I do make plenty of mistakes in Russian, and still get an increasing amount of translation into Russian when practicing.
Let's hope this particular test succeeds/fails/does whatever makes the experience more useful and pleasant.
Today, two days after my last message (March 17 2016) I did the reverse, strengtening English from French. Here I had to write much more French. It should have been the opposite. I've never seen 133 likes after only one week before. Someone ought to do something about this.
I think my skills stay gold forever, but maybe I should try redoing lessons some more then.
At first I repeat the lessons normally, then later I stregthen with timed practise.
So you mean with your method you still get more translations into the target language with timed practice too? In that case I'm pretty sure we're in different A/B groups.
That's what I mean. And there's something else, too. I did start the tree in autumn already. I progressed quickly and it soom became apparent that I could not keep the tree golden and move on. Then just before the third check point I just got tired, since I'd assumed I knew stuff I had not really re-learned properly. Progressing became impossible. I took a break and decided to start over, this time slowly and patiently. That's when I noticed that I get more EN-FR translations than before (or EN-SW in theSwedish tree). I don't know if it's a test or what, but the difference is clear.
I'll try redoing lessons a bit more, but in my case I'm working on many trees that I've finished long ago and I don't want to restart anything. There are so many factors involved that it's hard to know what the real reason for different behavior is.
What's worked best for me so far is timed practice on the whole tree (when the whole tree is golden), when I do that I get to translate more into the target language. (Also of course doing trees like French for German or Italian for French, where I'm learning both languages).
Anyway I think we shouldn't have to second guess the system, it would be much better if there was an option so we could choose for ourselves.
I agree that it would be nice if the system was more open for users' adjustments. After all, we all have different needs and abilities concerning language learning.
I completed the swedish tree the first time in 4 months but reset it at level 16 as i couldnt use it offline . which happened around the update in that time im already 70% through and at level 12 .. I want a real hard Challenge and at the moment it's just look for the obvious thing. Its not even learning . siôn
It could be my imagination but it seems to me that when I joined Duolingo (a year or more ago) that there were many many more exercises for generating sentences in your learning language.
This is something the Duolingo team should see and address. I understand that they are only trying to make the language learning experience better but this is not okay in my honest opinion. I prefer translating to my target language as this helps me memorize it and learn better. Please change this :(
As you can see from chirelchirel's and bookrabbit's posts, it seems they are addressing it.
I totally agree! I wish there were more native-foreign language translation as well!
I really think there's no way for the algorithm to figure out what I want. There should be some way of customizing it. I don't want a gazillion buttons, but at least a 'harder' setting would be incredibly useful.
I would be good if it was set up like this
Serious 15 mins
And then this option
Casual multiple choice questions
Normal multiple choice and word match
Serious above and translate text with speech
Insane all above and translate both languages
You may also want to edit the title of your post as it does not make sense. There needs to be a to and from in the sentence somewhere!
I agree, there needs to be a better solution to this than the suggested 'learn NATIVE from NON-NATIVE' method. An important part of learning a new language is being able to construct and speak non-native sentences from your native thoughts. Another issue with the native from non-native method method is that it makes assumptions of your comprehension of the non-native language that will unlikely be accurate, not to mention all the useful missing popup information etc.
While this may be helpful it is not a real solution - it is only a partial workaround to a sub-par element of the learning process. I was only up to level 5 in Spanish when I noticed a gap between the understanding and expressing elements in my own learning.
My suggestion would not necessarily be to change the balance (maybe a little, but not a lot), because the 'recognition-weighted' system of lessons is not broken. Rather I might suggest a second Strengthen Skills style option that is 'expression-weighted' instead. This part of learning a new language is particularly important when the language structure, ordering, etc is different (such as English vs Spanish in my own example).
while that will help construct sentences, increasing the rate at which the courses have you translate into the learning language would be very helpful. Also, in my case im learning norwegian which doesnt have a reverse course yet.
There's also the consideration that each course is designed to teach that language, and so what is actually taught is not always optimal. For example, in the reverse Spanish course there's a lesson entirely devoted to teaching do/did as used in 'did you go to the cinema that night?' which is of no use to me, while at the same time complex grammatical structures may be used that are almost never used in the language you're learning through a reverse course, simply to make the translations closer. This is not only more difficult but also misleading for learners, who, for example, may learn to use estar+verb much more often than is really used in Spanish, simply because the equivalent is so overused in English. Various course devs such as the Korean ones have also cautioned against using the reverse course to learn the language because the sentences used are unnatural and complicated.
when i learn a language, I make it a priority to think everything in that language. Sounds weird, I guess, but it does help, especially if you want more translation from English to German.