Why did you switch your Heisig reviews to Anki? Or why did you decide to keep using this site for reviewing Kanji instead of using Anki (or mnemosyne, etc)?
I love both RevTK and Anki. This seems like a useful discussion for Fabrice and Resolve as they continue to implement improvements to their fantastic programs. Plus, it may be helpful for those deciding the best way to review the Kanji they are learning.

If anki had the option to allow you just to click remembered instead of having to decide when the next time you saw the character was then I'd use it, unfortunately I'm not too good of a judge how long I'll remember a character I let the Leitner system do this for me which seems to work.

Plus I like seeing the box graph on the review page it gives me a clearer indication of how I'm doing, very motivating.

I originally was going to use Anki only for vocab and RevTK for Heisig kanji. I used the Leitner system on this site until I finished the book and had no intention of switching. I decided to switch to Anki after I missed "Spring". I drew the kanji for "springtime" and was mad because I knew "spring" but find "spring" to be a bad keyword. Just like that I decided to switch over to Anki where I could easily change the keywords (there are others that I similarily had trouble with).
Looking back I think my decision was a bit rash. Nevertheless I have been happy with Anki so I don't really regret doing it. Though I still don't think it was really necessary (eg i could do something similar with greasmonkey scripts).
If I was to recommend to someone, I would recommend to stick to one or the other but not go back and forth. Other than that I think it depends on the person. For example Anki doesn't require the internet use (which is another huge plus in my case).

meolox Wrote:If anki had the option to allow you just to click remembered instead of having to decide when the next time you saw the character was then I'd use it.

Couldn't 0 and 4 be the same as clicking yes or no?

meolox Wrote:Plus I like seeing the box graph on the review page it gives me a clearer indication of how I'm doing, very motivating.

I agree. Anki has many graphs but none are nearly as good as the simple box graph on the review page. Plus the new pie graph is a great addition, ne?

Currently I prefer RevTK.

I find RevTK with the "yes" and "no" button much simpler to handle than Ankit with the different numbers. The Leitnersystem seems to suite me, though I'd prefer to have one extra box which would ask to review one day after adding new cards, in addition ot the three days. Oh, and I really love to box graphs too.

Anki I'm trying to use for learning vocab, but currently I have trouble to open it on the computer (when logged in as admin there is no trouble, but logging on with a different account I have to use the "search"-option to find the .exe) and I have other little difficulties with it. So I'll stick to my cardboard-cards (which also give me more freedom where to learn) until I have more time to figure out how everything works.

I kept up with RevTK because of the shared stories, it had most of my stories on it and my data. Simple as that.

Anki is great, and when you're doing sentences the choice of 5 buttons. Yes, 0 and 4 are the "no" and "yes" options. In between you have the "same review time", "review earlier", "review more earlier". Once I got used to that idea, it worked better. I had thought choosing 3 was the "yes" equivalent, but was wrong. Anyway, with this wider choice, you can decide how well you did that sentence (a discussion thread of its own merits).

So, RevTK for Kanji (barring the Trinity update), Anki for sentences.

The other reason is unique to my situation (not using my own computer). Nothing really relevant here.

I love the elegant, simple design of RevtK. No fuss, very effective, visually appealing. Very little eye-punishing white; instead, pleasing creams and deep blue. The web-only aspect has bit me a couple of times-- like two weeks with no internet access, so that I drowned in orange cards on the way back. But in the end I made more progress by slowing getting to know other learners' stories & progress here. Alone would have been discouraging.

I like Anki too as the best downloadable SRS out there. Its design is three shades less elegant than RevtK, but a swiss army knife is uglier than a sushi knife. Anki has really helped me with vocab and phrases, and I'll use it for Spanish when I get around to that, I'm sure.

(I also find I like the look of it better after tweaking the BG & font color and hitting ctrl-option-cmd-8 on Mac.)

And I would have used neither if they hadn't had a keyboard infterface (as opposed to mouse).

Anki always crashes on my computer. On top of that I just don't like the design that much. I actually like using mnemosyne better because of its simplicity.

But I always use RvTK. My favorite thing about RvTK, besides its appealing design, is the bar-charts, which really go well with Heisig's principle of "divide and conquer". With RvTK progress is very clearly charted right from the review page. With the other programs you start off in the middle of a stack which can be, for me, very depressing.

I also like that RvTK uses TTFs for the characters as opposed to images (gif, jpg) so that I can copy and paste any character onto a "漢字の正しい書き順" page or anywhere I wish.

Can't wait for the new site updates*

I'm sticking with this site for Kanji, the shared stories are an invaluable feature. I also agree with billyclyde that the simple and clear design is superb. Leitner works fine for me and seeing the green stack in the last box grow and grow is a great motivator (I use a spreadsheet to track the sizes of my green stacks). I'm currently working through RTK3, but even when I'm finished, I expect to still be using this site for years to come...

On the other hand, I do use Anki for vocab and phrases. It is the best general SRS I've come across and the inclusion of a reading field has proved very useful. However, I'm not yet convinced that the five different levels of answer are that beneficial, but then again I never use the Easy option on this site either.

For drilling Kanji into my thick skull its RefTK. I like the system and it works for me.
For sentences. and vocab I use Anki cause I can attach Sound files to the cards. This is important to me because , unless its very easy vocab, I need to hear the words being spoken to make sure I get them correct.

I switched to anki at about the half way point. The main reasons for the switch were subjective - for one, I did not think that I was getting enough reviews in the early stages. At this point it seems I need to see the cards several times before waiting a day or more before the next review. I really like anki`s option to rate your performance and determine the next scheduled review time.

Another reason - the graphical representations of the failed/expired decks were very very distracting for me. I would often get discouraged and dedicate several days to review, only to find that I had lost my momentum when I wanted to start learning again. This might sound like a dumb reason, but I really do begin reviews with a much better attitude now that I use anki. I start it up, do my reviews, and move on to learning more kanji - no intimidating red and orange towers to stare me down.

I still visit this website every day for the community and mneumonics - there`s no subsitute for that.

I`m currently at frame 1320 - so I certainly can`t say that my approach has worked yet. But here`s hoping, right?

I started out reviewing with RevTK and it was good, but after about 1/3rd through the book I stopped RTK. A year later I started again and this time used Anki. I'm working on chapter 50 now, and its been great.

I like that Anki has made me keep up with all of my reviews. I'm past frame 1850 and have no big stacks of failed kanji to come back to. Every frame has been addressed.

I like the fact that I can influence the intervals myself. In the beginning I may want to review some kanji several times before letting them go for a whole week. After cards have made it into long term memory, I like that I can choose if I remembered easily, or after thinking for a while, or barely remembered. It helps optimize the review schedule to suit the strength of the memory.

Other than that, I think Anki is a little more speedy than RevTK. It shaves a little off my review times. I also like the stats.

[ファブリス]

Thanks for the very constructive replies so far, it's very helpful to see why you use one feature or the other, and how the programs make a difference.

I've noticed some users find the orange stacks to be intimidating, but it wasn't meant so, maybe "expired" is not the best choice of words. I always thought they are cards "ready to be reviewed". The urgency of review varies from stack to stack, but no program in the world can read your mind, it's a best guess from an algorithm which is more or less skilled at deducting the optimal remembering interval for you.

I've been pretty stubborn with the red stack. I reckon there should be an easier way to review from the red pile, but as I thought about it again lately I started wondering if the SuperMemo approach may not be the best approach for material such as RtK.

Lately I reflected on the fact that if a SRS tries to be too smart, it wants to find the best intervals for you. But are those intervals really based on your memory's "capacity", or are they based on what best use you could make of your memory at the time of learning with the information at hand and the skills you built so far? To give a simple example, if the interval is set in stone to 6 days and you didn't pay attention while learning, you fail. But a SuperMemo-like SRS will say, ok let's review this in two days instead based on your rating of the answer. The result is that the program doesn't tell you the ugly truth : you didn't pay attention while learning.

So when you stop, what happens? Did you really make good stories? That's one reason I kept with the red stack and probably will keep doing so. The red stack is there not to tell you you've failed, because nothing stops you from clicking "Learned" and drilling again after all; it's there to tell you that you should probably look at the story/mnemonic and see if that's not the culprit, because this RtK method heavily relies on mnemonics/stories. This is possible because the site started as a review site purely for RtK, but I wonder if this should not apply to many other materials.

A learning program is not primarily entertainment, if it can make learning fun that's great, but that's also a regular reflection of mine, the subject of .. can you really make a learning game? (but I disgress and this would make another very interesting topic )
And so I think pleasing the user is not the number one priority, it's better to be blunt and say "could you improve this?" rather than give the learner the feeling everything is alright and let them continue on shaky foundations. So that's one area where RevTK takes a stance, so to speak.

PS: my reflections above concern RtK material. Once you move onto vocab and material that does not rely heavily on mnemonics I think smart SRS'es like like Mnemosyne, Anki, SuperMemo etc are very well adapted. For the vocab area on this site I've already changed the functioning of the red stack, it's still red, but you can review it straight away.

i always click a review and change the number to 1 to review the failed stack just like any other expired stack.
perhaps its an idea to make a buton somewhere for the that same function, im sure not everyone has noticed this "trick"
and for anki, i started to use it recently for adding words from the jlpt lists, but for the heisig kanji this site is by far beter.

I use both. *shrug*

Thanks for that tip, ivoSF. I also like to review my failed stack and until now I had to wait until I was ready to see the answer to click the learned button of the previous kanji.

ファブリス Wrote:And so I think pleasing the user is not the number one priority, it's better to be blunt and say "could you improve this?" rather than give the learner the feeling everything is alright and let them continue on shaky foundations. So that's one area where RevTK takes a stance, so to speak.

It's a fair point, however it also seems to sometimes have the opposite effect where some people leave hundreds of failed cards unattended while rushing for the finish line thinking "everything is alright, I will look at these later", all the while not correcting their
methods.

Whereas Anki is less forgiving because it will keep asking failed cards. One would not get very far without studying properly because one would soon get swamped in failed cards.

Both have their plusses and minuses.

Hmm, here's my interpretation:

RevTK's stricter adherance to Leitner seems to work better for me. I like that it's either Yes or No, and it's next stack or first stack. Keeps it simple with less chance you'll try to play with wiggle room. I'd prefer a more spread out scheduling though. I posted that elsewhere but the idea is that you're less likely to get 100 cards one day then 60 the next.

For the missed stack, I also like that you must review those prior to getting them thrown into the mix again. Prevents a "push through" mentality where you're hammering the kanji into memory instead of letting the stories do their job.

So, I could say "You know, I shouldn't add more information since there's so much in my missed stacked" or I can say "I'll let my missed stacked build up, it's better to get all the kanji into the system first". That I have that option works great for me (I'm all about options).

I'll use both Anki and RevTK. But for Kanji, I'll stick to RevTK for now.

I started out using both the site and iFlash. I would study my iFlash deck in the morning and then do my RevTK pile before bed. After about 800 or so kanji I started to question the logic (and time requirement) of running two decks together and decided to focus on the iFlash system.

The sole reason I chose to stop RevTK and focus on iFlash was out of fear that if the site was for any reason to go down or disappear then all of my review data, stories and progress would be lost... and for me that was too much of a risk.

Well, I use RevTK for kanji and Anki for sentences.

Anyway, talking about kanji learning, both of these systems have their own pros and cons. What I love about Anki that is not supported by RevTK is mainly the "Cumulative view of due cards" graph, which tells me when my speed is ok and when I must hurry up. On the other hand, I really prefer (when studying kanji) the "red box" system. When I fail a kanji, it is because of my story being weak, so I need to give that kanji a rest and take some time to make a better one, and with Anki's system I would be brute-force-memorizing it rather than having time to make a better story. Besides, I really like actually being able to see all my stacks in the review section.

By the way, I really think that "Heisig method WITH SRS" has to be considered as something almost absolutely different from "Heisig method WITHOUT SRS".

????? Wrote:Lately I reflected on the fact that if a SRS tries to be too smart, it wants to find the best intervals for you. But are those intervals really based on your memory's "capacity", or are they based on what best use you could make of your memory at the time of learning with the information at hand and the skills you built so far? To give a simple example, if the interval is set in stone to 6 days and you didn't pay attention while learning, you fail. But a SuperMemo-like SRS will say, ok let's review this in two days instead based on your rating of the answer. The result is that the program doesn't tell you the ugly truth : you didn't pay attention while learning.

So when you stop, what happens? Did you really make good stories? That's one reason I kept with the red stack and probably will keep doing so.

Fabrice, I provisionally agree about the red stack-- I think Anki's approach of treating the failed cards first (I have my reviews there set for 45 min and 3 hrs for vocab/sentences) is meaningless for RevTK if you just dont' have a good story.

But I think a smart implementation may work for later stacks. I find myself forgetting cards I never forgot before at the 5th and 6th reviews. In this case, my stories are clealry fine-- it's the length of the review that trips me up. Leitner's guidelines are guideliness, of course, and were I doing this with my own system, I would have shortened those review times already to improve retention.

billyclyde Wrote:

Quote:So when you stop, what happens? Did you really make good stories? That's one reason I kept with the red stack and probably will keep doing so.

Fabrice, I provisionally agree about the red stack - I think Anki's approach of treating the failed cards first (I have my reviews there set for 45 min and 3 hrs for vocab/sentences) is meaningless for RevTK if you just dont' have a good story.

This is why when I fail a review I go back to the story. I don't just try to drill it instead. I doubt I could have gotten very far without sticking to Heisig's method.

However more often than not I don't have to revise the story; I just spend more time visualising them and just a small fraction tend ot need revising.

This is how my Anki Heisig deck is progressing. Cards are flowing up nicely. I think the "1.5-3 months" part is about equivalent to 4 or 5 correct reviews on RevTK. I usually put all my new cards through reviews at intervals 1, ~4, ~8, ~16, etc.

Cards over a week: 1635 (80.07%)
Cards under a week: 280 (13.71%)
Unseen cards: 127 (6.22%)

Correct over a week: 93.0% (1302 of 1400)
Correct under a week: 88.2% (5584 of 6330)
Correct unseen: 88.7% (1194 of 1346)

Average next time: 41 days
Average workload: 46.8 cards/day
Cards added: 14 a day, 425 a month

Period Cards Deck%
1 day 15 0.78%
3-5 days 93 4.86%
6-9 days 88 4.60%
9-21 days 233 12.17%
21-45 days 380 19.84%
1.5-3 months 1092 57.02%
3-6 months 14 0.73%


I owe a lot to this excellent site because its just gotten better & better since I first tried it over a year ago. There are a lot of good stories and the star system is great for finding them quicker than it was before.

I am thinking to import RevTK to anki, since I find myself reviewing so often from my keitai.

Does anyone know where the link is to export RevTK data to an excell sheet?

Also, does anyone know how to I add the data to mydeck in anki?

What exactly are you planning on importing to Anki? There is already a Heisig card set there...

That said, I wrote some Python scripts a while ago that pull data (HTML) from RevTK (reviewed cards, all stories for a particular card, recenty failed cards). I parsed some (not all) of the HTML to get some (again, not all) data out -- in particular, I used it so that I could get failed cards and look over them from time to time on my Zaurus.

If you like, I can give you the scripts, but it's likely that you will have to change them to do exactly that you want.

What I was hoping to do, (and I assumed was possible from the export to excell support of this site) was put the information into anki in a way that anki more or less understands when the cards would expire, so that I can continue my reviews on Anki.

Am I wrong about how the data can be used?

The link is here by the way:

http://forum.koohii.com/showthread.php?tid=553&page=40