I currently go at a rate of 25 a day, which I'm comfortable with. But I have a lot more time in which I can and want to practice so I though about increasing my rate.
The problem is creating the flashcards for anki is what takes most time. I can spend 1.5-2 hours creating cards for 25 new kanji, while only spending ~30 minutes actually reviewing that day.

My method right now is: read the kanji in the book, try to think of a story (only give myself ~10sec for that), Open the kanji in this site and look at the stories and notes, fill in the flashcard with all the info (keyword, mnemonic (hidden), kanji and frame number).
I find it rather ineffective. I need the anki because most of my reviews are not near my computer, and are therefore done through ankidroid.
Is using a premade deck a good idea? I really feel like creating the cards help me remember, if I'll just read it in the book and press a button in anki to add the next 25 kanji I don't think I'll remember them as well. It will also not have my own mnemonics (otherwise it would take the exact amount of time).

If you think creating your own cards helps you remember, by all means continue to do that.
In my opinion though, adding stories/mnemonics to your cards is pointless, so I think you should drop that part if it takes you much time.

When I was going through RTk, my process was pretty much the same as described here: http://rtkwiki.koohii.com/wiki/Study_gui...he_Kanji_1
So I would learn around 20 kanji, then review them quickly before taking a break. After the break I tested myself on those kanji again. I repeated this maybe 2-3 times, then at the end of the day before going to sleep, I'd add them to anki.
Worked really well for me, and didn't take too much time per kanji.

I use the shared Heisig deck on Anki. I usually don't have a set number of new kanji per day, I usually aim to about 20 - 30, stopping when a lesson ends or a new primitive appears that goes well beyond my limit.

First thing I do is get all my reviews done. The "New cards per day" option is always at 0. I always click "Learn More" after finishing my reviews and go from there. When seeing a new kanji, I show the answer and spend a few seconds thinking of a story, then I click the keyword to go to the RTK page and read some of the shared stories. Unless I have a really good story, I copy and paste something of interest from the shared stories onto the card in Anki, bolding the keyword and italicizing the primitives. I set my card layout to show the story in the question field. Then I write the kanji out by hand, sometimes more than once if it looks weird. I try to do it quickly because its funner and makes me feel great. After that I always press "good" on Anki, no matter how much I think I know the kanji, so I'll see the new card again tomorrow. Then I go on to my next kanji.

Each kanji takes maybe about 2 minutes max. Some people are opposed to having the story in the question field, but I find it more effective. I try not to read the story when I think I can remember the kanji without it.

Personally, I just use the 'add 25 kanji'-button on this website, and even if I don't enter the Kanji themselves, I tend to remember most of them.

To my understanding, it is not very productive to spend much time on one card; it may be much more effective to spend 7x 10 seconds repeating a particular kanji in the SRS than to spend 4 minutes on just creating a card for it.

My basic strategy is to look at the Kanji in Heisig's book, create a short story (since I've read 'Moonwalking with Einstein' I tend to put this story at a place I associate with the keyword, if possible), then move on to the next Kanji.

If for some reason I fail a card, I give it more attention. I may try make the story more emotional or impressive, or think of a new story. Next to that, I'll then also try let my body learn, making up a gesture for the keyword, following it with gestures for the primitives.

Every night, before going to bed, I scan Japanese printouts to circle Kanji I already know, mainly to keep my motivation and make my progress more visible.

In sum, you could consider learning the kanji like people like Tiger Woods learn golf; they don't spend minutes pondering every ball when training; at first they spend a little time on each ball, then hit, and only if they find that they have trouble with some particular hole they spend more time on it. Only try hard, time-consuming ways for those kanji for which short, easy ways obviously fail. You may not remember all cards in one go then, but well, that's what an SRS is for...

Try out if this works for you: learn 25 new cards within two hours of going to bed in the evening. Do not review those cards until the next day, but flip through them until you can recall everything from short-term memory.

Memory is only stabilized in deep sleep, but the memory cannot have been 'created' too long before the night's rest. There's a solid study on that but I forgot where I read it.

I didn't do this for RTK1 (pre-Anki), but later for Core6k to 10k, and it worked like a charm.

You could try using two separate decks for that, one purely for learning, the other for reviewing, and delete the frames learned from the learning deck, unsuspend them from the reviewing deck.

Also, really consider using a premade deck. If you can think of a story in ~10secs, it'll probably work for you, but if not, there are some really funny and memorable ones among those contributed here.

... and check out the user study methods, you might find great inspiration there: http://forum.koohii.com/showthread.php?tid=4943

@Irixmark

As someone who worked with and without support of memory consolidation through deep sleep, could you reflect on how your repetitions were different in those two modes?

Inny Jan Wrote:@Irixmark

As someone who worked with and without support of memory consolidation through deep sleep, could you reflect on how your repetitions were different in those two modes?

Not sure I quite understand your question...?

Here's what I've been doing before going to sleep (ideally before I'm too tired to focus).

Setup: word shown in hiragana with audio of the word played simultaneously, then only the audio of the sentence.

With the learning deck, I look at/listened to the card front side and the answer once, press 'show me the answer', repeat the audio aloud, copy the keyword in kanji, then press `again.' This is the learning phase.

By the time I've worked through my 50 cards once each, Anki starts displaying the first card again. Now I keep testing myself like this: see the word in hiragana, hear the audio. I answer by repeating the audio and write the kanji from (short-term) memory, and grade myself on this and on correctly understood meaning. If I get both correct, I press any of good or easy or very easy, it doesn't matter as the only point is to remove the card and consider it "learned" for the night. Repeat until I can correctly answer all 50 new words and Anki says I'm done. Then I delete the newly-learned words from the learning deck in Anki, and notice that satisfyingly the remaining card count goes down. That's the short-term memory recall phase. It's really like flipping through vocab cards, nothing more, with Anki for convenience and audio. I suppose I could even use the cramming mode for that, but I want to keep the reviews and scheduling completely separate.

Then I try to get a good night's sleep.

At some point during the next day, I open my reviewing deck, unsuspend the same 50 cards I've studied the night before, and start reviewing. Getting everything correct usually means pressing 'good' or 'easy'. 5/50 cards I'll likely have forgotten overnight, although usually not completely, so those get 'again' and another review that day. From then on, it's just the usual reviews.

I could easily learn more than 50/day, but the reviews would pile up too quickly and I don't have more than 30 minutes/day for that at the best of times. Then there are days when I was too tired. Or sometimes you go to the pub for a pint or two and there's no point studying after that.

This is only for cramming vocabulary. These days, when I take a sentence and keyword from everyday input there's no learning phase because I've already seen it in context, looked it up, clipped the audio and example sentence if needed.

For RTK1, it should be: form/choose a story, construct the image as described by Heisig, test yourself on short-term memory while also practicing the writing. Sleep, start reviewing the next day.

My current rate is ~20 cards a day (19.5577 to be exact ) and that's mainly because of the time that I can spend on reviews. I didn't care about when during the day I learn new cards but your numbers are promising (30mins review for 50 new cards/day) - or my memory is not performing as well as yours

Anyway, thanks for sharing your experiences.

@InnyJan

Something that might make my experience slightly unusual was that I found the Core6k and 10k words really useful and interesting because I watch a lot of Japanese news and read financial newspapers for work. Here are the links to the papers I mentioned earlier in the thread:

http://learnmem.cshlp.org/content/13/3/259.full

http://phys.org/news171875597.html

Seems especially plausible to me because my required reviews during the day went down when I switched to evening learning sessions even though the vocabulary didn't change, if anything it became harder over time.

Interesting studies, though I think its been known for many years that sleep aids in memory. I'd caution statements like:

Irixmark Wrote:[Long-term] Memory is only stabilized in deep sleep

Your hypothalamus works even when you are awake, so long-term potentiation happens even while awake. The effects of sleep defiantly make the process more efficient though I believe.

vix86 Wrote:

Irixmark Wrote:[Long-term] Memory is only stabilized in deep sleep

Your hypothalamus works even when you are awake, so long-term potentiation happens even while awake.

Right, of course. What I meant to emphasize was that studying kanji/vocab before bedtime is more effective for me than doing so in the morning, and the studies confirm that. It obviously doesn't mean that you learn nothing unless you go to sleep right away.

Actually when you read the studies I linked to above, the difference between studying at 8am and at 8pm is statistically significant, but substantively not overwhelming---a 20-25% gain in retention. That again confirms that there is no magic bullet for learning kanji and vocabulary. Though when you consider that you'll probably spend 1000-2000 hours learning and reviewing kanji and vocab until fluency, shaving off 25% is quite significant, and that's probably to some extent cumulative with the efficiency gain of an SRS.