Back

Finished RTK at the end of 2014

#1
Hello there

Some years ago, I decided to learn japanese. In december the twenty-sixth, twenty fourteen, eleven in the evening, I wrote the message http://forum.koohii.com/thread-95-post-2...#pid216269:


Quote:I began in august 2013 and finished today, 26. december 2014. I made 2200 flashcards.

*happy*


Now, more than two years later, I am still learning japanese. It has never been my main activity. But nonetheless, I tried to steadily improve my knowledge. Currently, I experiment with a spaced repetition system made out of paper. I used Anki before, but stopped after some data loss. With my 2cm x 4 cm flashcards I learned the complete vocabulary for JLPT N5. Now I am learning the Kanji for those words. At the end of this year, I will take the JLPT N5 in switzerland.

With paper cards, I have more freedom over the intervals used for the spaced repetition system. In my opinion, it does not make sense to continue to learn the same flash card with an interval of more than 7 days. I extended my cards with Kanji after I knew they would remain easily recallable from memory for at least 7 days. The same intervals I used for the flashcards only with words did not work as well for the flashcards with kanji added. I did not know the kanji when I tried to recall them after the assigned interval. So I started to make a learn session twice a day instead of only once while using the same interval growth rate I had by having just one session a day.

Interval growth rates for vocabulary:

I recalled the card the first time -> wait 1 day (wait till the next learn session)
second time -> wait 2 days
3. -> 3 days
4. -> 5 days
5. -> 7 days
6. -> remove from system

I added the flash card again into the system after I had extended it with Kanji. Now I am trying the following system:

2 sessions a day, morning and evening:
1. -> wait 0.5 days
2. -> 1 day
3. -> 1.5 days
4. -> 2.5 days
5. -> 3.5 days
6. -> remove from System

I just forward the cards the same number of fields (one field -> one day) on my grid, but twice a day.

Why paper?
1. I have to write the flash cards myself -> I am already learning while creating the flash cards.
2. I don't look like a mobile phone zombie when I am riding the train. ( I have the cards in a small plastic bag in my pocket. I can easily take them out and learn. It does not need any electricity or devices that may break).
3. It is much clearer what I have already learnt and what I am about to learn.

What is the main purpose of a SRS? Use more time for difficult flash cards and fewer time for easy flash cards. Each time I don't know a card, I remove all tokens and shift the card only one field further. When this happens several times, I know I have to do some research about the Kanji to make it easier to memorize (e.g. looking up the components in RTK).

Here is an image of my setup:


http://imgur.com/a/7b3Sr

Thanks for noticing.
Edited: 2017-02-19, 4:01 pm
Reply
#2
Hey, whatever works for you! I spent years (starting in the 90s) learning vocabulary (for multiple languages) with paper flashcards, and the day in 2008 that I discovered anki was one of the happiest days of my life. But everyone has to find their own system.

(Your image link seems to be broken btw.)

Good luck with your next steps. Having been working on Japanese ever since 2003, I am resigned to the fact that the day I stop studying will be the day I die. Luckily I enjoy the studying.
Reply
#3
Nice answer. (added the image link)
Reply
August Sale (14th - 25th): 30% OFF Premium PLUS - 25% OFF Premium
JapanesePod101
#4
Nothing wrong with using paper, though it can take lots of physical space...

(...but we can start a nice argument about killing trees to make paper vs burning oil and expelling CO₂ to generate energy -- lol just kidding :-P).

OTOH, in my own experience seven days isn't even remotely close to long term memory. I have strong doubts about your confidence on being able to remember something 6, 12, 24 or 240 months later just by proving to yourself you can retain it in your mind for mere days. That's actually the point of SRS: by a ridiculously small amount of additional effort, you can retain 85% ~ 90% accuracy several years later.

Obviously it's you who should make that call: it's not impossible your memory is that gifted... But I'm just trying to warn you about a possible mistake you're making and that you'll regret later.

Anyway, as the self learners we are, and as said by tanaquil, whatever works for you ;-).
Reply
#5
i know anki has settings you can tweak like interval modifier. not sure if those settings are enough for you but for me recently I found out you can change the STEP  thing. that stuff was driving me crazy because the card I failed would show up 1 minute later or 10 minutes later. I changed that sucker to 1440 minutes!  I WISH I found out about it sooner.
Reply
#6
I use Anki strictly as an electronic flashcard system, without the spaced repetition aspect.  To do this, in the options I set the "Steps (in minutes)" to 4800 and 9600.  This allows me to just flip through the decks one card at a time, with no spaced repetition.

I like how with Anki I can easily "flip" the cards and test myself back to front instead of front to back.

I can't use paper flashcards because 5,000+ cards would be too many to carry around in my briefcase.

Every time I read a Japanese story, I type up all the vocabulary I have to look up in the dictionary and add them to my main deck (tagged with the name of the story).
Edited: 2017-02-19, 11:31 pm
Reply
#7
(2017-02-19, 11:30 pm)phil321 Wrote: I like how with Anki I can easily "flip" the cards and test myself back to front instead of front to back.

If you flip a deck is that change permanent until you change it back or does it go back to normal next time you open the deck.

Also if you flip does it flip the entire front and back or just the key words.
If you have a front=key word and back with say Kanji, Story, and something other items will it flip for everything or just the main [Kanji] one.

I have a deck for RTH that is done with Hanzi on the front and everything else [key word, pronunciation, etc] in the back. I'd like to switch key word and Hanzi but keep everything else on the back card.

I like Anki but those unhelp files are pretty brutal.
Reply
#8
I used to have flashcards almost exactly like the ones you have. I even used paper clops to hold the decks together. A also believe anki is a vast improvement to that old system for a few reasons that people have already mentioned. Carrying around all of those flash cards is a huge pain. I have +40k cards on my phone and I can study any of them any time I want. Try that with paper flashcards.

Also, as faneca mentioned, 7 days is hardly sufficient. Do you really remember words you learned in 2014 after just 6 reviews? That would be very surprising if you did unless there are other aspects of your study that I'm not getting. Try taking answering a few of your old flashcards from last summer and see how many you remember.

If you plan on taking JLPT this year, the test is more than just vocabulary. You'll need to practice grammar, reading, and listening if you want to pass the test.
Reply
#9
(2017-02-19, 4:00 pm)dlgltlzed Wrote: ...

Paper flashcards are fine, nothing wrong with them. But learning individual words isn't going to teach you Japanese, it's just going to teach you a bunch of Japanese words. In the future, you should study collocations/sentences, not just words.


Words are fine for learning the writing system, though (as long as it's words you already know).

(2017-02-19, 5:57 pm)faneca Wrote: OTOH, in my own experience seven days isn't even remotely close to long term memory.

Well it is. Everything beyond 30 seconds is long term memory.

Quote:I have strong doubts about your confidence on being able to remember something 6, 12, 24 or 240 months later just by proving to yourself you can retain it in your mind for mere days. That's actually the point of SRS: by a ridiculously small amount of additional effort, you can retain 85% ~ 90% accuracy several years later.

I've heard this claim before...but, as far as I know, there isn't any science behind it.

The way long term memory works (according to the leading scientific theory on this), is this: when a person places a piece of information into their short term memory (aka "working memory"), they strengthen it in long term memory. When the person does not "use" that piece of information (meaning it does not enter their "working memory") for a while, it fades. The longer that period of non-use, the more it fades.

If this theory holds true, the "spacing" doesn't really affect the strength of the memory. What affects it is the number of times it was "rehearsed" (meaning brought into the short term memory), as well as the length of each "rehearsal". So, for instance, rehearsing a piece of information once a day for 10 seconds, for a week, will create a memory that is just as strong as rehearsing it at ever increasing intervals, for a total of seven times, over the course of a year. Five years later, if all else is equal, both memories will have about the same strength.

Doesn't make SRS useless, of course. By spacing out the rehearsal times, you are trading shorter term strength for longer term strength, which can be a good idea in some contexts. But there's no evidence to say that it creates stronger memories, overall.

P.S. There are many other factors that affect the strength of a memory, beyond just what I mentioned here. But, all else being equal, SRS and more frequent rehearsal have the same effect towards strengthening memory.

One more thing: the big benefit of Anki isn't the review spacing it provides. It's the other functionality: it gives you the ability to measure your knowledge, and allocate your time accordingly. And it does so without mental effort on your part: it manages your time for you.

For instance, if you already know a card, Anki allows you to spend very little time on it (you just have to press "easy" 3-4 times...which takes a pretty much insignificant amount of time to do). On the other hand, if you don't know a card well, Anki allows you to increase the frequency with which you see it (by pressing hard), or even reset the counter altogether, by failing it.

And, finally, when you're really struggling with a card, Anki allows you to cut your losses, by automatically suspending it after 4-5 fails. 

All this, without any extra attention from the user. I think this functionality makes programs like Anki a huge time and energy saver, compared to other study methods, and is far more useful than the supposed memory benefits of spacing out reviews.

And sure, you can replicate this functionality with paper cards, but it's not automatic. You have to spend your effort on managing the cards, and making decisions on what to get rid of, what to set aside because you know it, etc., instead of on the contents of the cards. With Anki, the managing part is done for you.
Edited: 2017-02-21, 4:29 pm
Reply
#10
>>P.S. There are many other factors that affect the strength of a memory, beyond just what I mentioned here. But, all else being equal, SRS and more frequent rehearsal have the same effect towards strengthening memory.

I don't know about the scientific literature, but empirically, I have noticed that there is a striking divide between words that, once learned, I will NEVER get wrong again, and words that need to be continuously reinforced or they will be forgotten. The nasty thing is, you can't always tell which will be which.

I have twice had the experience of letting my very large anki database lapse for a year or more, and going through the labor of getting it back again. Both times, I have noticed the same thing: maybe 50% of words I get right despite being WAY overdue, and the other 50%, it's like I never saw them before (or I remember them, but only to remember that I can never remember if this means X or Y). Of course, there is some correlation with how common the words are and how early they are introduced, but a couple of my leeches are N5 vocabulary items (大体 and 大抵, I will hate you forever, please die).

One could argue that the 50% that I keep reviewing with a 100% success rate are a total waste of my time, and that wouldn't be wrong. But for me, it is so demoralizing to confront that 50% failure rate after a year or two, I would frankly rather spend the time doing the quick reviews of the easy stuff and get it out of my hair.

I agree that the OP might find it illuminating to review those cards that were set aside as "learned" to see how many of them stuck. Maybe I am too dependent on anki's algorithm, but in the long run, it seems to produce the greatest results for the least stress - for me.
Reply
#11
(2017-02-21, 3:35 pm)Stansfield123 Wrote: piece of information into their short term memory (aka "working memory"), they strengthen it in long term memory. When the person does not "use" that piece of information (meaning it does not enter their "working memory") for a while, it fades. The longer that period of non-use, the more it fades.

If this theory holds true, the "spacing" doesn't really affect the strength of the memory. What affects it is the number of times it was "rehearsed" (meaning brought into the short term memory), as well as the length of each "rehearsal". So, for instance, rehearsing a piece of information once a day for 10 seconds, for a week, will create a memory that is just as strong as rehearsing it at ever increasing intervals, for a total of seven times, over the course of a year. Five years later, if all else is equal, both memories will have about the same strength.

This certainly isn't true for me.  I've done some tests of extreme values in anki and 5 reps on the first day after failing a card had negligible benefit over one single review on the first day.

Other anecdotal evidence such as not remembering certain phone numbers despite dialing those phone numbers 1000's of times in the past suggests the same thing. The intervals between reviews is important. Too little or too much time between reviews is less efficient.

Research that I've read runs counter to your reading suggesting that there is a "spacing effect" where you are more likely to remember something after spaced out reviews rather than clustered reviews.  See Robert Bjork's research for instance.
Edited: 2017-02-21, 6:21 pm
Reply
#12
(2017-02-21, 3:35 pm)Stansfield123 Wrote:
(2017-02-19, 5:57 pm)faneca Wrote: OTOH, in my own experience seven days isn't even remotely close to long term memory.

Well it is. Everything beyond 30 seconds is long term memory.
Quote:I have strong doubts about your confidence on being able to remember something 6, 12, 24 or 240 months later just by proving to yourself you can retain it in your mind for mere days. That's actually the point of SRS: by a ridiculously small amount of additional effort, you can retain 85% ~ 90% accuracy several years later.

I've heard this claim before...but, as far as I know, there isn't any science behind it.

There is indeed lots of science on the subject, dating back to the '30s at the very least. The wikipedia article on SRS has a few pointers, but I'm really regretting right now (and for the n-th time, I must add) that I never cared to save the couple of studies that I read on this (or links to them). Anyway, yogert909 already posted a good link (wow, research done in 1885, didn't know about that).

[On an unrelated note, some studies from the '60s used (guess what) paper flashcards ^_^.]

In my case (and I know that just one experience is quite the opposite to science -- especially when it's your own, so it's impossible to have any objectivity at all) memory works in a really different way to what you described. I think memory is a multi-layered structure far from the plain landscape you painted, where every layer has varying characteristics (ease of recalling, persistence in time before fading away if not recalled, etc.) and, of course, without well defined boundaries in between them (for instance, with shocking events, several layers can be short-circuited across, vertically). So talking about the "throw-away working storage" you consider the short term memory and the "really long term memory" as if they were the only ways your brain deals with information sounds pretty far from the truth to me.

Another concern is that everyone here seems to talk about the "strength" of a memory as if it were a universally accepted and self-evident property, but a lot can be said about that too. There are several variables to consider (i.e. several degrees of freedom) when studying the different aspects of the way in which a memory took root, and just referring to them by using the informal term "strength" can't possibly convey them all (but yeah, most of us aren't memory scientists, so using informal terms is only logical).
Reply