mr_hans_moleman Wrote:To tell you the truth, Anki is only good for the long term. What you put in today, don't expect to to learn it until 1 year later.
Anki is basically a more user-friendly version of SuperMemo for MS-DOS. SM-2 algorithm + facts / model paradigm, tagging, plugins; the scheduling algorithm is basically the same. Thus, the
advice given by Dr. Wozniak applies. In particular, from my experience:
1) If you don't know it, Anki won't help any more than route repetition--in fact it's worse, since SRS, by design, jumps around from topic to topic like a hyperactive gradeschooler. (This is what makes it addictive. Fine for review, not so for initial learning.) Learn, then use Anki for review. (Rules 1, 2)
2) Simple and easy beats hard and complex. Subdivide complex sentences. (Rule 3, 4, 9, 10)
3) Do not beat yourself up about not "finishing my reviews today." Start every day, stick to whatever baseline you've decided, and let finishing fend for itself.
4) But, don't add new cards to an un-reviewed deck. Anki is simply not tuned for that: new stuff can easily displace old and Anki will mindlessly try to pack it all in. I find that if more than 24 hours elapses between adding and first review, I start having to learn stuff in Anki.
So, if I take on a bulk project (currently, the Core 2000, previously
an Esperanto/English glossary.), it has to go in a separate deck. Otherwise, new cards go to the end and it might be weeks before I first review them.
Real-life example of simplification from my reading:
I think this is an interesting sentence: (From
鳥をとるやなぎ)
「エレキの楊の木？」と私が尋ね返そうとしましたとき、慶次郎はあんまり短くて書けなくなった鉛筆を、一番前の原吉になけつけました。
I tried to ask "An electric willow tree?", but then Keijirou chucked a too-short, can't-write-with-it-anymore pencil at Genkichi, the student immediately in front of him.
Can you imagine trying to do dictation of that? Ouch.
kanji-card:
purple-willow / 楊 Normally, I can resist the cries of a weeping willow, but this one, she was a
purple willow! I just couldn't let the poor
tree sit there, so I gave her my
piggy bank. (Credit: Megaqwerty)
reading cards: (new vocabulary)
尋ねる
返す
投げ付ける
reading/dictation (or kana-only-recognition) cards:
尋ね返そうとしました (illustrates grammar:
-(y)ou to suru / "tried to do X, but something interrupted)
書けなくなった鉛筆 (illustrates complex grammar: "®enaku natta" / "has become unable to") -- note that learning this requires being comfortable with the underlying simple stuff: potential, negative, naru, etc. If it's not comfy, I'll delete this.
あんまり短くて書けない (illustrates grammar "ammari" / "so (much) that")
鉛筆を原吉(げんきち)に投げつけました。(illustrates usage of 投げ付ける)
One sentence -> eight facts, twelve cards (1 kanji, 3 vocabulary, 4 reading, 4 dictation).
Fragmenting sentences inflates the number of cards, but I think that's a good thing. Simpler cards train faster (because the pattern of neural activation is more consistent), and can be scheduled more precisely (the hardest part slows everything else down). The only downside is that people might accuse you of "cheating"--but honestly, comparing deck size kinda puerile.
As for deleting or suspending cards, I say "go for it!" Delete hard cards. Delete awkward cards. Delete too-easy, recently-added cards. But,
don't delete easy, mature cards. Those cards require very little upkeep--reviewing them for the next decade will take less time and effort than initially learning an equal number of new cards. Just grade them "easy" and move on. Likewise, don't delete a deck because it's hard right now. By design, Anki only shows you the hard cards--the easy, cheap-to-retain cards are the silent majority.
I think deleting your RTK deck was a huge mistake. Anyone who's continued to review afterwards, can open to a random page of the keyword index, cover the characters, and write well over 90% of them. In your case, having both Japanese and English keywords would be a more applicable test (if you can write them from Japanese prompts, there's no point retaining the English), but I still encourage you to try--I hope you can still write them all, but you might be surprised by how many you've lost.
(If it bothers you, my suggestion is to get a premade deck that is keyword (maybe Japanese keywords) -> character only. Write stories only for characters you fail and grade easy ones "easy" even if you can't remember a story. Choose how much time to dedicate to reviewing them each day (5 minutes is fine), and don't worry about leaving the task unfinished.)
But, other than that, I think you're basically in the right place. Four thousand sentences a year sounds slow to me--but you've obviously learned a lot. I can't argue with success. Do add easy stuff--stuff that's easy now is still subject to forgetting.
As far as going forward, while I can't speak from experience, if I were in the same place, I'd try sub2srs based on
Nukemarine's experience. Actually, even being well behind you, I think I'll try it anyway. (must order DVDs with exact subs, must order DVDs with exact subs--and pick up a Region 2 drive).
And, if I wanted to grow my vocabulary, I'd at least consider
going back to school. Anki can be used more traditionally (monolingual flashcards, cloze deletion, mind-maps, etc--sentences are kind of an awkward fit: how
do you grade understanding?), and there's no one telling me to do boring homework or show up for boring class or read boring literature, so I can skip all that for the yummy, fun, interesting learning. (Admittedly, that's based on my preferences--I always was a nerdy bookworm--and may not fit with your goals.)
Above all, congratulations, and here's hoping for breaking through the intermediate phase to epic-awesome kung-fu proficiency.
