Offshore Wrote:Really just a matter of if "you" think it's a waste of your time imo. Not everyone needs to be able to write Japanese, handwritten or just normally.

I write out all my vocab reviews not because I ever expect to make any use out of it, but mostly just because I enjoy writing the characters (and some of my friends think I'm crazy for being able to write them).

This! it's about your priorities, not someone elses.

Muscle memory certainly helps a lot with retention, but at the same time, writing everything out is very time consuming compared to doing normal reviews. So, theres always a trade off between the amount learned and the benefits you get from improved retention due to muscle memory.

Depending on the way you do your cards, it might be helpful as output practise though, i don't know, haven't tried it...

On one hand, i'm pretty sure i wouldn't have got as good at listening and reading if i'd spent all the time writing out stuff, i'd have had to put in a lot more time and effort. On the other hand, i found that in Japan i wanted to handwrite stuff a lot more than i thought i would... attending lectures, or just making notes in conversation, that kind of thing. Writing in kana is time consuming, and translating even more so.

Still, on balance, i'm not unhappy with the way i've gone about things. There's always a tomorrow to learn new things...

IceCream Wrote:Muscle memory certainly helps a lot with retention, but at the same time, writing everything out is very time consuming compared to doing normal reviews. So, theres always a trade off between the amount learned and the benefits you get from improved retention due to muscle memory.

It's better to do it once right than to do it a bunch of times and keep forgetting. No one saves time when constantly failing cards, am I right?

And nest0r, imagine what the world will be like when they cut out writing entirely from the school system and repalce it with just learning by keyboard. You know it's coming since school systems are more concerned with efficiency and cost rather than proper learning.

KMDES Wrote:You know it's coming since school systems are more concerned with efficiency and cost rather than proper learning.

What's the difference between proper and efficient learning?

KMDES didn't say it was efficient learning they were shooting for.

KMDES Wrote:

IceCream Wrote:Muscle memory certainly helps a lot with retention, but at the same time, writing everything out is very time consuming compared to doing normal reviews. So, theres always a trade off between the amount learned and the benefits you get from improved retention due to muscle memory.

It's better to do it once right than to do it a bunch of times and keep forgetting. No one saves time when constantly failing cards, am I right?

Well... i don't think it's about constantly failing, and i don't think anyone's going to assert that writing it once is going to do the job either.

We don't have the exact numbers / time spent, so it's hard to know which way is most efficient. Then you have to take into account context, and other things and methods that can help retention also and compare those with writing added. Writing is unlikely to add efficiency in a linear manner, i imagine.

Remember that efficiency in an SRS over the long term is likely to be around 85% or 90% even without writing, so you've got to consider the potential number of extra cards you could have added instead of, say, getting 96% over much fewer while writing them by hand.

Like i said though, it could help output also, which would tip the balance as well. So, it does depend on your priorities, i think.

The thing about writing is that it can help encode the memory into your brain better, allowing you to have a better retention quicker, resultings in the need for less SRS as all memories are not created equal. So by writing you culd cement the memory in better and save time by not needing to SRS nearly as much which anyone with a large review deck would attest to would be a life saver.

And what time frame would long term be considered?

Output is of course vital to the learning process as many people can relate to, so that should be part of their learning process regardless.

jettyke Wrote:

KMDES Wrote:You know it's coming since school systems are more concerned with efficiency and cost rather than proper learning.

What's the difference between proper and efficient learning?

Basically they want the proverbial 'Out the door and straight to McDonalds' type of efficiency. This why they don't go over material that has been learned in other years of school.

Again, maybe it leads to better retention, but it depends on how much. when i had a vocab deck, i could get through around 500 reviews an hour. Supposing you even just write in the air, i at least think it would take double the time, so now you're down to 250 reviews an hour. I deleted my anki decks ages ago now, so i don't have exact stats, but i remember clearly that the majority of my cards were in box 4 by the time they were classed as mature, which is at 21 days in anki.

Without the exact figures, it's really difficult to make a comparison. Certainly writing helps with retention, but i would estimate that at least in decks where you have high context to begin with, the effects aren't worth the time cost. In decks with low context, it might turn out that it's more efficient, but i'd think it was pretty close either way.

All that is assuming that you're going to be marking on the same criteria as i would, which for many people who are writing, isn't going to be the case, since they may fail the card if they used the wrong kanji / wrote the kanji incorrectly, etc. If you mark that way, it's debateable whether you will see any impact on your srs reviews, given that you're marking on a much harder criteria than those people who don't write.

I don't think there's necessarily one correct answer, writing is definitely beneficial to memory, and perhaps the types of benefits it brings might mean that some people want to trade amount-done efficiency for that. I think doing RTK is pretty much the same trade off. People can choose based on what they're happy with doing, and how much they think it benefits them personally... personally, i'm happy with a more sloppy style.

The problem with the math is you're figuring it out as if someone is doing the same amount of reps AND writing it all out anyway. Assuming those 500 cards turned into 50 cards and then you double the time it'd take a about 12 minutes. This may be on the genrous side, but it's quite possible. Let's do a little math wth an encoding method I've been doing.

Assuming it takes me double the time to review as it does your time (Which is about an average of 7.2 seconds per card by the way.) and I've done 240 reviews at 14.4 seconds a piece at 4 reps over the last 24 days (Should of been 3, but as it was an experiment I did it anyway though it probably was never needed) equaling 960 reps at 14.4 seconds, or 3.84 hours. This would be 8 hours of reviewing at 500 cards with 2000 reps. Now to compare this I'd need to know how many reps on average it'd take for you to mature a card. In my system, the 4th rep would be considered mature.

This is with a 90% retention rate and failed cards do not get renewed reps in my system. Why I do this is beyond the scope of this thread right now, so I'll leave it at 'take my word for it' right now.

Since you're reviewing in context, your memory gets a recall bonus of about 15%. (stat via the study with people recalling data in water and out of water and other such studies.) With writing, you can avoid needing context for recall (which drops SRS's stats that you have to around 70%-75% recall) Now, with a little more data we can formulate a recall vs time efficiency average between the two of us using each of our methods (which will be an estimate of course without certain variables.)

And being able to write will be considered an added perk.

Also we can assume that it takes people more than 21 days to reach an interval of 21 days. So I'm not sure how long you took to delete your deck.

wait, does anki treat mature cards as those with intervals over 21 days, or is it cards that are over 21 days old? (i was assuming the latter).

the 500 reps was on vocab only cards, those cards with context would be much lower rate per hour. (But i don't think writing for context is a straight swap either, though... i don't think writing can replace context at all, as it has more benefits than writing alone.)

You can't drop my efficiency down to 70-75%, since i quoted the efficiency of my vocab only deck to begin with. For vocab, it was around 85%, my high context deck was something ridiculous like 99%.

i also don't understand how you can compare if you don't review failed cards...?

But anyway, i would say the average number of fails per card to get it to maturity was between 8-12. Maybe someone with an existing deck could give you better statistics though? (that's just a guess)

I didn't read everything, but in regards to the OPs post: "Do whatever you need for your goals" is a language learners number 1 rule. If you want to be able to write in Kanji and don't mind the extra time do it. If you don't really mind or don't need it then don't bother. I've stopped writing myself as I want to use that time elsewhere. There's no reason why you can't learn to speak, type, read and understand Japanese without writing it - it all depends on what you want.

To give you an example of what I am saying. I want to understand (like 100% of almost everything heard) and sort of read Japanese (just what I need to get by). I do this in three ways:

1) Anki: Video and song cards as this is what I am going to use the Japanese to do. (I'm using the AJATT sentence method here with a few tweaks.)
1.1) Repetition of the audio and video I learn in my cards above, so I hear the whole thing often.
2) Writing producing from grammar points and getting correction via lang-8 (basically conversation school for free).

What I'm doing is practicing only for what I need and what I am suggesting is that you do too. If you want to just understand TV, radio, news, anime and dramas then what I am doing is probably enough at my level to get there slowly. If you want to be full on N1, all four skills, and near native as you can get then you will need to do everything (literally everything a Japanese person does - I mean general life - in Japanese) for as long as it takes to get there. Hence, practice for what you want to do and do whatever you need to get to your goals.

nohika Wrote:

howtwosavealif3 Wrote:don't worry you don't have to follow his post.
i used to like believe/follow him but now i don't since I passed hiim in Japanese (as far as i can tell)> Like because of him I was never going to do RTK. He didn't say DON'T DO IT but he was like oh it didn't work out for me but you can try it but still what I got was oh he's not approving of it, I won't do it. well at this point i finished rtk but I would've finished earlier if i didn't listen to him since i would've started earlier.

and like a couple weeks ago he did a post on yoroshiku and i guess he doesn't know this b.c. he doesn't read japanese that much or like diddn't RTK but he thought the kanji for よろしく　Ｗａｓ　　夜露死苦... like that was the actual kanji when it's just 宜しく
and i've actually seen both b.c. before reaidng that post I recently watched some akbingo episode and they used the yoroshiku with the yoru/tsuyu/shinu/kurushii kanji and i looked it up b.c. i was curious what the significance was. I think he's stuck at some level at japanese and honestly it just becomes more obvious the more i get better at japanese

Both of those kanji can be read as Yoroshiku, so...

Anyways. I can't write the kanji right now, but I'm fine with it since it's not my goal at the moment. My best friend is a native speaker - she was born in Hiroshima, moved over here with her mom when she was 5. Went to Japanese school every weekend + extra during the summer from elementary school to high school (it was a special program that basically kept the students up with what the students in Japan were doing. They did a week's worth of material in a day).

However, when we were chatting and she was writing stuff out, she'd not know how to write some of the more complex kanji. If you don't see the skill or use it very often (she reads a lot of Japanese, talks with her Mom in the language, but she's been out of Japanese school for over a year), you're going to lose it, native speaker or no.

it's not just Both of those kanji can be read as Yoroshiku, so...
I'm saying 宜しく　　IS LIKE the official way and it's SEEN WAYWAYYWAY more often than the other version (I've only encountered it in that AKBINGO episode like I said before and then i read about it online) which I'm like how can you make that mistake.

IceCream Wrote:wait, does anki treat mature cards as those with intervals over 21 days, or is it cards that are over 21 days old? (i was assuming the latter).)

'Mature cards
Cards that have an interval over 21 days. By the time cards become mature, they are usually quite easy to remember.' At least that's what the anki site gives me as a definition.

IceCream Wrote:the 500 reps was on vocab only cards, those cards with context would be much lower rate per hour. (But i don't think writing for context is a straight swap either, though... i don't think writing can replace context at all, as it has more benefits than writing alone.)

The more encodings the better of course to the point that a single exposure can cause a memory to last an entire lifetime. I'm not sure of anyone that actually writes out the entire context out as even I don't do that.

IceCream Wrote:You can't drop my efficiency down to 70-75%, since i quoted the efficiency of my vocab only deck to begin with. For vocab, it was around 85%, my high context deck was something ridiculous like 99%.

I can because you activated my trap card! Seriously though, I can because anki retention doesn't equal real world retention as it is a static enviroment that you experience that doesn't involve writing or any othe encodings that can negate the effects of that bonus being taken away when context is removed. So the 99% deck would probably be more like 84% real world.

IceCream Wrote:i also don't understand how you can compare if you don't review failed cards...?

Let's consider it a wild variable as we'd be here all day if I asked you to come up with how often you failed cards and then getting them back up to being mature.

IceCream Wrote:But anyway, i would say the average number of fails per card to get it to maturity was between 8-12. Maybe someone with an existing deck could give you better statistics though? (that's just a guess)

So for your method it would be between 8-12 hours of review to get to maturity and mine would be between 6-8 hours of reviews. On average via percentage I would retain
about 64 per hour of reps. Yours would be about 42 retained per hour of reps (Given the average of 70-99%).

While this data is approximate at the very best considering we didnt even factor in genetics, health, emotions, time of day of reps, drugs or anything of the sort. But can you consider that writing can possibly produce a better learning system than without even on a time efficiency basis?

KMDES Wrote:

IceCream Wrote:You can't drop my efficiency down to 70-75%, since i quoted the efficiency of my vocab only deck to begin with. For vocab, it was around 85%, my high context deck was something ridiculous like 99%.

I can because you activated my trap card! Seriously though, I can because anki retention doesn't equal real world retention as it is a static enviroment that you experience that doesn't involve writing or any othe encodings that can negate the effects of that bonus being taken away when context is removed. So the 99% deck would probably be more like 84% real world.

hahah
but... hmmm... certainly that's true for context decks, but if you're talking about a vocab only deck? (i.e. just a plain kanji compound written). There's definately still some outside interference encoding problems occasionally, but it's definately not as high as with a context deck.

KMDES Wrote:

IceCream Wrote:i also don't understand how you can compare if you don't review failed cards...?

Let's consider it a wild variable as we'd be here all day if I asked you to come up with how often you failed cards and then getting them back up to being mature.

ok, cool, then let's assume neither of us fail any cards... but isn't that one of the main advantages of writing, that overall you fail less often?

KMDES Wrote:

IceCream Wrote:But anyway, i would say the average number of fails per card to get it to maturity was between 8-12. Maybe someone with an existing deck could give you better statistics though? (that's just a guess)

So for your method it would be between 8-12 hours of review to get to maturity and mine would be between 6-8 hours of reviews. On average via percentage I would retain
about 64 per hour of reps. Yours would be about 42 retained per hour of reps (Given the average of 70-99%).

ummm... i'm lost...
it takes me 8-12 reps to get to maturity, not 8-12 hours!!! each rep is approximately 7.2 seconds, giving a total of 86.4 seconds per card. Those 8-12 reps are already including failed cards, but in case i'm totally wrong (i remember my figure to be at 37,000 reps on the core 6K at the point i finished it, so it's an probably not more than triple that to get it all to maturity (i guess), which equals roughly 15 reps per card, so we're taking 12 as the estimate.)

So, if yours take an average of 7 reps to get to maturity, at an average of 14 seconds per card, that gives 98 seconds per card to reach maturity.
It seems like there's not so very much to choose between them really... as long as you mark your cards on the same basis as mine. Mine is a little lower, but probably not enough to really be outside of deviation between different people.

KMDES Wrote:But can you consider that writing can possibly produce a better learning system than without even on a time efficiency basis?

I can agree that writing has very well researched benefits, and is useful in many other ways. It could also be useful for output, which would tip the balance. But, it really is about your preferences, i guess. If those figures are all correct, there's not really anything to choose between them one way or the other...

Writing can reduce your failed cards and reduce the reps you need, but how would we calculate that without getting a bunch of data?

I wnet 500 * 7.2 * (8-12) /60 /60 to get the data. Coincidentally, 500 * 7.2 /60 /60 = 1. So the hour equal the reps through math fluke basically.

There is no 7 average for my cards. My cards reach maturity at 3 to 4 reps. Not 3 + 4 reps. o_O; (Maturity being next interval of over 21 days or more) And I calculated it at 14.4 seconds. Where did you get 7?

but your number of reps to get to maturity should also include failed cards too right?

ah, i thought since you'd given mine as 8-12 hours and yours as 6-8 hours, it meant your average to maturity was 6-8 reps (7 average). That actually sounds reasonable, because with only 3 or 4 reps wouldn't that mean that everything you did was always marked as easy? (first rep is 7 days interval, then 14 days, then 21 days, next is over 21?)

Saying that, i did think that cards were mature at 21 days old, so my estimate could also be completely wrong... maybe it's better if people with real numbers step in, we could work it out a lot better...

or someone could maths isn't my strong point hahah

Failed cards aren't taken to maturty, ailed cards are removed from the deck completely if failed once. I'm a harsh harsh man. I don't have easy hard or anything. It's either I know it completely or I remove it from the 'deck'.

The average maturity was 6-8 hours, not reps as the math gives me 14.4 seconds which doesn't give a perfect 1. If my reps were 7.2 seconds, that'd be a average of 7 reps, yes, but I'm not at 7.2 seconds. And I only do 3-4 reps before maturity.

I think some of the confusion here is that you think I'm using anki to time my reps, but I'm not. My formula is custom made but is still giving me the same results as if I were using something like anki with less reps by using more encoding methods. Which is kind of my point for the whole 'less reps = saved time' thing. The formula and system are still in the works too, but it's showing great promise.

Also, the 240 I qouted was learned in one day. I plan to do an experiement this summer to see how many cards I can learn and retain before I hit a 'brick wall' so to speak. That's gonna be fun. Also wil give me a idea of what the upper limits of the system would be as I have not attemptd to go beyond 240 in one day and push the limits the followng days. Hypothesis: burn out and insanity.

*edit* Almost forgot, another experiment I want to try is to see if the system can be built up like a muscle, so you're able to learn more per day/faster as you 'work out'.

i'm totally totally confused!!!... if i took out all failed cards, then mine would also be the minimum reps possible! It wouldn't be 8-12 any more, it'd also be 3-4! lol!

Yeah, i've also often wondered what the optimal amount of cards you can learn in a day, and whether theres a point after which the number of reviews start to climb massively due to failures. (or in your case the number of deleted cards). But, yeah, after around 2 weeks i think more than 1000 reviews per day = burnout + insanity, definately!!!

i think, yeah, you can train your brain to take in more stuff, definately. Certainly reviewing was far easier when i stopped than right back at the beginning, i think!

IceCream Wrote:i'm totally totally confused!!!... if i took out all failed cards, then mine would also be the minimum reps possible! It wouldn't be 8-12 any more, it'd also be 3-4! lol!

Yeah, i've also often wondered what the optimal amount of cards you can learn in a day, and whether theres a point after which the number of reviews start to climb massively due to failures. (or in your case the number of deleted cards). But, yeah, after around 2 weeks i think more than 1000 reviews per day = burnout + insanity, definately!!!

i think, yeah, you can train your brain to take in more stuff, definately. Certainly reviewing was far easier when i stopped than right back at the beginning, i think!

Is 3-4 the minimum for getting to maturity in Anki? Also, the thing is with my method, 3-4 is the minimum and maximum, I don't do no more or less. I've been tweaking the exact times, but the thing is the amount of time it takes is fairly static.

From when I did the 240, the ones that failed (24 in total in 24 days) tended to be either random, cue dependant forgetting (which is not really forgetting, but I fail them anyway) or chunks where I can see I just wasn't paying enough attention to the task and doing bad encoding. Also, there were a few involving confused memory (similar words/kanji with various meanings being misattributed to the other words with similar features) which is a beef I have with RTK, too many of the same indentifying item with variations at one time.

It is if you don't use hard or easy as well, otherwise it goes up to around 5 or 6 i think...

But, really we can scrub the above calculations anyway, because we are working off entirely different premises to begin with.

i made absolutely no effort to encode any information correctly whatsoever, and never tried to remember anything... my initial pass rate was probably only 20% or so. There are too many different variables here to calculate... you're talking about 240 new cards over 24 days while im talking about 6000 new cards over 40 days! If we were studying the same things with the same encoding methods with only the writing or not writing different, we could compare, but...

But please post your results from your experiments in the summer, i'm definately interested to see the results!!!

Cranks Wrote:To give you an example of what I am saying. I want to understand (like 100% of almost everything heard) and sort of read Japanese (just what I need to get by). I do this in three ways:

1) Anki: Video and song cards as this is what I am going to use the Japanese to do. (I'm using the AJATT sentence method here with a few tweaks.)
1.1) Repetition of the audio and video I learn in my cards above, so I hear the whole thing often.
2) Writing producing from grammar points and getting correction via lang-8 (basically conversation school for free).

What I'm doing is practicing only for what I need and what I am suggesting is that you do too. If you want to just understand TV, radio, news, anime and dramas then what I am doing is probably enough at my level to get there slowly. If you want to be full on N1, all four skills, and near native as you can get then you will need to do everything (literally everything a Japanese person does - I mean general life - in Japanese) for as long as it takes to get there. Hence, practice for what you want to do and do whatever you need to get to your goals.

Could you go into a little more detail on how you do your anki please? I'm just about to start my first sentence pack ala AJATT and would like any pointers you could give. I'm not a computer guy so I don't know much about making video or sound cards.

My goals are simply to be able to read and converse in the language. I live in Japan 1/2 the time so I'd like to be able to communicate and I don't want to be illiterate. Now, I suppose if you define literate as being able to write(by hand) 2000+kanji then so be it. At this point, I would be pleased as punch to be able to text an e-mail message to my friends (and in fact when I do text something like 今は勉強しています。　　They're pretty impressed even though I wouldn't have been able to produce that from my little grey cells (sorry if that sentence isn't 100% correct btw,I am just a beginner after all). Sure, eventually I'd like to be able to read, write, and speak like a native but If I could take some shortcuts in the meantime and then go back to doing a little skritter or something it won't hurt my feelings.

IceCream Wrote:It is if you don't use hard or easy as well, otherwise it goes up to around 5 or 6 i think...

But, really we can scrub the above calculations anyway, because we are working off entirely different premises to begin with.

i made absolutely no effort to encode any information correctly whatsoever, and never tried to remember anything... my initial pass rate was probably only 20% or so. There are too many different variables here to calculate... you're talking about 240 new cards over 24 days while im talking about 6000 new cards over 40 days! If we were studying the same things with the same encoding methods with only the writing or not writing different, we could compare, but...

But please post your results from your experiments in the summer, i'm definately interested to see the results!!!

6000 in 40 days?! That's, like 150 in one day each day! 150 in a day? That's unplausible! (hehe.)

I have been keeping a mostly detailed log of my attempts and wrote up a tutorial of how my system works and how to use it. It's also 5.5 pages long, which makes it too much effort to read for 99% of the people I've mentioned it to. I don't think anyone has actually made it past the first few paragraphs yet. I don't think I could even pay people to read the darn thing. :/

KMDES Wrote:The thing about writing is that it can help encode the memory into your brain better, allowing you to have a better retention quicker, resultings in the need for less SRS as all memories are not created equal. So by writing you could cement the memory in better and save time by not needing to SRS nearly as much which anyone with a large review deck would attest to would be a life saver.

And what time frame would long term be considered?

Output is of course vital to the learning process as many people can relate to, so that should be part of their learning process regardless.

Agreed, integrating a couple seconds of writing into reviews for encoding and then where one feels it useful during subsequent reinforcement results in stronger and more stable memorization by integrating the motor memory, so that not writing and spending less time per review, or even spending those extra seconds of not writing trying harder per review, will result in less effective learning. You'll see that exact hypothesis born out in Longcamp's research, where exposure to information is of the same length, but those who used handwriting methods learned the material better. Sensorimotor encoding is awesome.

Of course, blending this with other senses such as audio is even more effective, when done in a complementary fashion according to the target information on the card and the purpose of the deck, etc. I don't write for my video subs2srs decks that are designed for listening. But for RTK especially, writing and active recall is so perfect it's essential. Likewise for writing in vocabulary-oriented decks, be they single words or sentences. A few seconds of writing when they're new, a few when they're being restudied/touched up.

Edit: Although I might start incorporating gestural techniques during listening-specific cards also, actually: http://www.sciencedaily.com/releases/200...210039.htm - Here it involves complementing working memory where the phonological loop and articulatory rehearsal are involved by providing a mild distractor to the sensorimotor/visuospatial component that actually enhances the other areas, I imagine from some bottom-up process, it's been a while since I read the paper that article's based on. Similar I imagine to the way in which multitasking practice trains executive attentional modulation to get better at multitasking by improving management of memory bottlenecks: http://www.physorg.com/news170015185.html

High five for 29% retention boost!

KMDES Wrote:High five for 29% retention boost!

I think it will take more than that study, however, to convince teachers their students are paying attention even better than the students making eye contact, when doodling in class. ^_^

To be honest, I don't think retaining kanji writing ability even requires constant practice such as writing a word out every-time you review it. Once you have learned them once, and particularly recognise all the radicals, if you just very occasionally write words (such as when noting down unknown vocabulary) , that is enough to maintain that radical writing ability, it really does get embedded in your muscle memory extremely robustly. As long as you then keep reading, you will be reminded of a japanese word's 'spelling' (which radicals and how they are placed, any patterns and so on) and therefore be able to write all common words extremely easily without having to deliberately practice that skill. This is the same as it is for English, you don't have to practice English spelling, you just know it from reading a lot.