New French: Exhausting but Educational
I did a review lesson today for French. It took me 45 answers (in 45 minutes) to finish it. I'm not really complaining because I did learn quite a bit--I just wasn't expecting to. On the bright side, it gave me the full 10 XP. :-)
The recent changes to the French tree definitely had an impact. I have proceeded slowly, only doing two of the new lessons per day, and then only when nothing "weakened" during the night. Today, the Prepositions 3 skill weakened, and when I went to strengthen it, I wasn't terribly surprised to see entirely new sentences. That's usually kind of fun, after all, and I usually learn a bit in the process.
Today I learned a bit more than I bargained for. In fact, in order to understand some of the new sentences, I spent a lot of time reading in Wordreference forums and even reading from a French reference grammar ("Advanced French Grammar," L'Huillier, 1999). I do think I learned a lot, but, my gosh, it was quite a struggle. Especially for a review lesson!
Usually, when I end up getting lots of questions wrong, a good number of them are really errors or omissions by Duolingo. Not today, though. I managed to make these mistakes all on my own.
It is a very good thing that Duolingo got rid of the hearts system a few months ago. Otherwise this would have been too discouraging to imagine doing. This is also the first time I've appreciated the fact that in review lessons Duo often asks the exact same question three or four times. Usually that's annoying, but today, it was what I needed!
Again, I don't mean to be complaining--just expressing a little surprise and wondering if others are having the same experience.
Yes, a side-effect of the new tree is that we also get more to practice with different contexts, and that courses can improve at an exponential rate that was not possible before.
The "no-hearts" system effectively makes people more likely to try out different phrases, more likely to reach the end of the lessons, and report errors. The main problem remaining is that newer sections and the bottom of the tree will always be reviewed by fewer users and thus contain more errors/omissions.
Overall, the team's work was admirable.
Anyone who tries to keep his/her tree gold will soon discover that the review lessons focus heavily on the lessons below the last checkpoint, though. It becomes really important to get errors fixed because otherwise you keep seeing the same ones over and over, since Duolingo thinks these are sentences you're have trouble with.
Duo should give higher priority to error reports from people who have all (or mostly) gold trees. Not just because those people have a greater need for it but also because those are far more likely to be correct.
I totally agree that the revisions of the French tree are admirable. Lots of good thought and work went into what we have now. The repetition of sentences, both written and aural, is much appreciated. It helps to "set" the patterns in both modes. I am grateful.
I'm glad too that they got rid of the hearts and it is great to hear they have added some new words to the French tree. Your blog looks very interesting also!
My schedule is to do three lessons each day, but having had so much trouble with the first one, I dragged my feet for a few hours. I just did the two general reviews and didn't miss a single question on either one. Of course it was all different material, but it was still refreshing to be reminded that I really have learned some French over the past 15 months. :-)
Hi Greg, I checked out the link to your blog and tried to reply there to the entry about guilding your tree, but could not post there, so here is my comment:
Great advice, thanks! Very useful! I am currently doing Spanish rapidly because I have prior knowledge, and German very slowly because of no prior knowledge. Your advice will help me keep my Spanish progress more "durable gold." As for German, I am in exactly the boat you described where I got to level 5 and got overwhelmed and confused, in spite of having an all gold tree. In German I have now gone back to the beginning, and I am re-working through all the lessons in order, repeating them until I really understand them and get everything right. This feels good to me. Thanks for your great blog!
I appreciate (and agree with) the points you have brought up, Greg. But the revision came at a difficult time for me, as I had decided to review by doing the reverse tree. And then, feeling confident, I finally added German which had been tempting me for some time. The new French tree requires three lessons a day to keep gilded, and then I am trying to do two reviews to one lesson on both the Reverse French tree and the German tree. I can do it- but at the cost of neglecting immersion, which I love and which I feel has a great deal to offer after one has lived with the language for a while. I am currently wondering whether to stay with it or not.
I would suggest you just let the forward French tree decay and use that time for Immersion instead. You can always regild it later, if you decide you feel like it.
Thanks. If the tree doesn't "settle down" and require just the occasional upgrade within a week or so, I am going to do just that.
Stick with it! You're an inspiration to me and I'm probably the age of your grandkids! Keep it up!
My tree has totally deteriorated too. While so many people are glad the hearts are gone, they were what worked for me: they kept the lessons short. I still have them on the iPad (although if I slip and install the most recent update, I will lose them there too).
The idea of a 45 minute review session for ONE review, as the first author described, would be the end of me! I try to squeeze DL into spare corners of a busy day...
Good luck with the German and the regilding.
As crush points out, I do take longer sometimes because I want to understand everything. I'm a linguist, so that's not a big surprise. Further down the tree, I think that makes me a lot faster. Even when I make a mistake, I usually look at it and go "oh yeah, I knew that." In Spanish and Italian, I do the review lessons in 21 questions or fewer, and they only take 5 minutes.
My goal with these languages is to read novels, so that means I eventually have to learn the whole grammar anyway. I prefer to do it up front, and, yes, that can lead to the occasional hour-long lesson. I'm retired, so that's not a big deal for me. The big surprise was needing 45 minutes for a review lesson.
I definitely don't think that everyone else has to do it the same way I do, though!
I think that's a bit of an exaggerated case, though. Most people won't be looking every single thing up online, taking out grammar books, and writing/reading posts online. But the author apparently did learn quite a bit in the process, though.
Hi Greg, Guru of Goldness! I have a question: My Spanish tree is all gold and I want to practice the more recent lessons to strengthen them per your advice. Does it still help if I re-do individual "lessons" as opposed to practicing a "skill," which is the module with several lessons in it?
I'm sure it helps, although I haven't tried it. The advantage of the whole-skill exercise is that it should focus on the things you have trouble with. If you redo lessons, it will show you more things that you already know well.
Good point, Thanks! and by tree being all gold, I meant the part completed so far, lol!
It has become much more difficult to keep my tree gold. Three to four modules weaken each day. I am having to do Objs 1,2,3 Abstracts 1,2,3, Infinitives 1,2,3, etc. everyday. It shows me just how poorly I've memorized these but it is a little frustrating that many of these that I bring to full strength simply weaken the very next day. Obviously this is part of the new algorithms and that will only make us better.
Three or four a day is way too many. Are you using hints or timed practice or something like that?
I don't use timed practice. I'll use the hints if it's a word I haven't recognized, but even then I generally refrain from using them. I believe that it may have something to do with my success rate, though I have no proof for that. Oddly enough, the modules that weaken the most are all clumped together as the middle part of the fourth leg of the tree.
I have 5 weakened modules at least per day, I'm about 15 modules to finishing now, sometimes it got up to 8 or 10 modules I don't even want to count, just insane, even I only tried timed practice yesterday out of desperation. And I don't skip anything in timed practice, so timed practice actually is even slower than normal, I felt so.
My tree is usually golden but the app still tells me I have weak skills every day. I do at least 50 XP in French every day, plus some Dutch and German. At the start I would only have to review a couple of times before the app said "No weak skills!" and I could move on to a new lesson. Now I have 26 completed topics but haven't finished a new one for weeks. I want to make sure I have a good grasp of the completed lessons before starting a new topic. I don't want to rush through the whole tree! Can anyone explain why Duolingo thinks my skills are weakening faster than I can improve them? I don't think I make that many mistakes and I often often complete a review lesson without any errors!
Do you use hints? And when you say "weak" do you mean the skills have dropped to 4 bars? Or are you trying to review until it says "no weak skills?"
Sometimes the skills drop a bar but usually not. What do you mean by hints? Do you mean clicking an underlined word to see its meaning? If so I almost never do that! I'm trying to get it to say "No weak skills", which it currently only does if the internet connection temporarily drops out (i.e. trying it again immediately results in a review lesson). Also, in the past it has said "No weak skills. Learn something new" (I've forgotten the exact message) even when some topics have dropped bars (i.e. some topics were not full). I used to do "Practice Weak Skills" first until there weren't any left, then any non-golden topics that remained, then new topics if I hadn't reached 50 XP yet.
Mine weaken like mad and I rarely use hints and often get all correct. It is not rare for me to find 10 freshly ungilded skills on a tree. I do keep them gold for the most part and the weakening often seems completely arbitrary.
I would say that it's a mistake to try to get it to say "no weak skills." That's too much work for too little return. Just try to keep the skills gold.
A strategy that works is to plan to do three strengthens per day if your tree is not all gold, and do just two if it is all gold. Strengthen any weak (that is, non-gold) skill first. Do general strengthens when there are no weak skills. Only do a new lesson if there were no weak skills. (So you always do three lessons a day until the tree is complete.)
I have just discovered that "Practice Weak Skills" on the main page is useless! If you only do this it "strengthens" the same skills every day! It doesn't keep track of which words you are getting better at! However, the method you suggest is equally bad! I don't want to rush through and have many words I've forgotten as soon as I was taught them (which I already do). I've seen on the website it actually tells you individually which words are weak but I only learn on the app and it doesn't have that feature. So unless I spend extra time on the computer as well as the app there's no way for me to actually find out and strengthen my weak words. I wrote an algorithm in the late '80s for learning French vocabulary (not grammar though) that was far superior to Duo's system!
What's my purpose here? To have a golden tree? No. To get through the tree as quickly as possible? No. To learn all of Duo's words perfectly? No. I'm here because I want to learn to speak French well! I was giving Duo more credit for having a good system than it deserves at the moment. Don't get me wrong, I'm not saying that Duo is bad and I certainly intend to keep using it. It's just that Duo gives the impression it does more than it actually does.
Only do a new lesson if there were no weak skills? I guess if you want your tree to take years that's a strategy. If I followed that I'd never get to another new skill as I'd completely lose interest and give up by then.
My strategy on an unfinished tree is to regild every skill that needs it and do a few new ones daily, even if it means leaving a few ungilded until the next day. I look forward to finishing my tree, as I feel that the review afterwards of keeping it gold is where my learning really takes off.
Maybe that's what it took you, but that doesn't mean it would be the same for everyone. Some people will retain the information more quickly than others, plus the algorithm changes from time to time and there is no telling when one will be put in a test group of "let's see what happens if we mess with this person this way or that."
Plus, ten months is quite a long time. You can create a month-old human in that span!
Two years is generally considered a reasonable amount of time to reach B1.
Okay, but that doesn't make me want to spend two years. I'd prefer to move ahead at a non-glacial pace so I'll have a bit more time on earth left to actually use the language.
I think of it as an error message that warns you that you've overloaded the review algorithm. :-)
Out of curiosity, how many lessons were added from the previous French tree? And what type of lessons were added. I started my tree after the update.
Overall I don't think the number changed much. They moved some lessons around and reorganized them, and added some new vocabulary (probably under 20 words).
I have found the updated tree somewhat annoying. For me about 70% of the new lessons was stuff I had already learnt previously. It isn't so much a new tree more of a reorganisation. I go through the lessons slowly and I am finally getting all of the added lessons done. But it has been somewhat frustrating to do these new lessons only to find this is stuff I have done before.
I like that Duo has got rid of the heart system too, you can take more of a risk and use a translation you are not 100% sure and it is good to know you the lesson won't end if you make a mistake. I do find it a bit of a chore when you are doing a difficult topic and it just feels like it will never end as you make mistake after mistake, when you get to around 40 it starts to frustrate. It is probably better than the previous system though and everything has its positives and negatives.
Do you think in general the new tree is better for people new to Duolingo/just starting their French tree now?
I continue to juggle my languages, and noticed the "Possessives" in French were finally made right. The lessons should have been called "Possessive Adjectives" before, because when I came upon "Possessive Pronouns" in other Internet study areas, I was confused until I got those two separated in my mind. The first two modules in Possessives are adjectives, and the added one deals with pronouns.
Lovely post thanks. I have found it more difficult since the tree was expanded. I often find that I need to review 5 lessons a day to keep my tree at full strength!
Why would you recommend avoiding timed practices? I think in theory at least it should have the same effect as non-timed practices. I'd think it would be more effective because even if you miss more than 3 answers you won't have to start the practice all over again and all those words will get strengthened anyway.
I haven't finished any of my trees yet but just spent the past few weeks "regilding" my Italian (1/2 done), German (1/2 done), and French (2/3 done) trees after about a year's absence. I finished regilding my Italian tree a few days ago and finished the German tree today, coincidentally following your rhythm of reviewing about 2-3 skills a day. But i always use the timed practice, because losing a practice session is really disheartening and just makes me want to quit for the day. I'm really glad they got rid of the heart system when learning new skills.
I can see the benefits of the non-timed practice, namely you focus more on the question and have time to read through the discussion (i really think even the timed practice should pause when you open up a discussion, or at least let you access the discussion topics after you finish the test). But for the purposes of maintaining a tree gold, i can't see how non-timed practices could be better unless you just tend to miss a lot more questions in the timed practices due to carelessness.
Anyway, it was a really interesting post, thanks for sharing it!
EDIT: Ah, i just noticed there are no more hearts in the non-timed practice, either! I think now i will spend more time doing the non-timed practices :)
I'm not sure what's wrong with the timed practice software. Perhaps it really is just as simple as "people make more errors when they rush." All I know is that when people say "I work and work and work, but my tree just keeps decaying," it always seems to come down to one of three things:
- They use timed practice.
- They use hints.
- They make lots and lots of errors (e.g. needing more than 25 or 30 responses to finish a lesson).
Folks who don't use timed practice, don't need hints, and need 25 or fewer answers to do review lessons find it pretty easy to maintain gold trees without spending a lot of time doing it.
Hmm... well i'll try using timed practice in my Italian tree and non-timed practices in my German tree and see if i really notice a difference, though it might not be a fair comparison because German is much harder for me than Italian. Though now that the heart system is gone, the only real incentive to do the timed practices is the extra points since otherwise you also miss out on the discussions.
My stats might also be a bit skewed because it's been so long since i've touched my tree, perhaps the delays between reviews are longer. But i've gone from the first skill to the about halfway through each course over the past 3-4 weeks and my trees are still completely golden.
I'll be very interested to see your results. If I may, let me suggest that you keep a spreadsheet recording what you did each day. I have done that myself for the past 16 months, and I find it makes it much easier for me to be confident I'm speaking about what really happened vs. what I'm "sure" must have happened.
This does not jibe with my experience.
First off, timed practice regilds a skill sometimes even if you've only managed to get one question correct before running out of time. This skill will indeed de-gild very quickly, but it's not because you used timed practice, but rather because you have barely strengthened it. When I use timed practice and actually manage to FINISH as opposed to running out of time, the skill seems to stay gold longer than when I do not use timed practice.
What seems to make my skills de-gild insanely quickly is using the app instead of the website.
You must also consider that we are not necessarily all using the same algorithm, as there are probably always tests going on, plus glitches.