I'm about 1 week in, and I'm up to frame 77.

I like this. IMHO James Heisig is a genius and should get the nobel prize in japanese pedagogy. I'm using anki and it seems like the quantity of repetitions there will be the limiting factor in how quickly I get thru the book.

Do other people use anki for RkT? If so how do you deal with the large number of daily reviews that inevitably arise? The RtK deck that I got has 2200 cards, which is gonna wind up being a huge daily time sink with anki at some point no matter how good your memory is. My concern is that if the number of daily repetitions gets to be too high then I might not make it thru the deck.

The reviews will tend to add up, but in a few months, the cards you are studying right now will only come around every few weeks or months. So you'll be spending the bulk of your time studying recently added cards and only a few minutes on mature cards. For instance the 400 kanji I added last year have an average interval of 1.4 years, so I only spend about 2.2 minutes per day on reviews.

It might seem more fun studying new kanji, but the reviews are the important part. You are unlikely to remember a kanji after the first or second times seeing it, and the ones that do will quickly gain high intervals, so you won't be spending time on on them anyway. This is the power of SRS and anki. Trust me, you'll want to keep up with your reviews.

Studied 966 cards in 33 minutes today. (over half of which was adding ~110 new cards)

At your current pace you'll wind up with 100-200 reviews a day. It's not really that much if you do them fast.

Vempele Wrote:Studied 966 cards in 33 minutes today. (over half of which was adding ~110 new cards)

At your current pace you'll wind up with 100-200 reviews a day. It's not really that much if you do them fast.

How in the world do you study 966 cards in 33 minutes? That's 2 seconds per card. I trust you are doing what you say, but I don't see how that is possible to sustain. 110 new cards no less.

Don't worry, the review count for 2200 is not going to be 30 times what you have today :-).

A lot of the reviews come from newly studied cards. If you stopped studying more kanji now, after 1 month you will only have to review 2 or 3 kanji per day. In 2 months less than 1 per day, and so on.

The review load for older kanji decreases with time, so you then have more time to keep studying new kanji.

yogert909 Wrote:How in the world do you study 966 cards in 33 minutes? That's 2 seconds per card. I trust you are doing what you say, but I don't see how that is possible to sustain. 110 new cards no less.

I started at 5 seconds/review in March 2013 (first month after switching RTK recognition from Memrise to Anki; I was probably a bit slower on Memrise) and got faster from there. I gave up RTK nearly a year ago, and it's consequently my slowest deck by far.

I test one fact per card: Kanjified word (occasionally expression) -> reading (not meaning, though I do have definitions on the back side), randomized kanji (containing a specific phonetic primitive) -> on'yomi (abandoned), Swedish word <-> English word. Checking correctness is fast, and when I get a card wrong, I don't need to keep looking at the correct answer, I just move on.

My learning steps are 1 10 60 (learn ahead limit 1). With premade decks, I usually add 5 cards at a time, adding another 5 when the first 5 are all at the 10-minute step, and so on.

Vocab recognition is quick. I'm averaging 3.2s on that, testing reading & meaning.

For RTK I usually go through the steps of writing it in my palm & counting the strokes, so it takes me on average 9.8s, but I only have 15~20 daily reviews on that right now.

My sentence deck is taking 12.5s average per-card, so sentences take me the longest to review. At some point I need to probably prune sentences that are less useful.

Vempele Wrote:I started at 5 seconds/review in March 2013 (first month after switching RTK recognition from Memrise to Anki; I was probably a bit slower on Memrise) and got faster from there. I gave up RTK nearly a year ago, and it's consequently my slowest deck by far.

I test one fact per card: Kanjified word (occasionally expression) -> reading (not meaning, though I do have definitions on the back side), randomized kanji (containing a specific phonetic primitive) -> on'yomi (abandoned), Swedish word <-> English word. Checking correctness is fast, and when I get a card wrong, I don't need to keep looking at the correct answer, I just move on.

My learning steps are 1 10 60 (learn ahead limit 1). With premade decks, I usually add 5 cards at a time, adding another 5 when the first 5 are all at the 10-minute step, and so on.

Here's my problem with this: I often come across words where I don't immediately identify one or more Kanji in them. At that point, I don't want to just look at the reading and fail the card, because I want to be paying attention to the individual Kanji in the words I learn. I think that's important (because Kanji represent morphemes, which are meaningful elements in the language - you have to pay attention to them, you can't just keep memorizing compounds as if they don't share the same morphemes).

So I have no choice but to stop my review and look up the Kanji on this site, and read the story (I could rack my brain to figure out the Kanji without help, but it wouldn't take significantly less time and it's a needless drain of energy). That drives my review times up (I'm a little under 10 seconds/ review).

How do you deal with that issue? Do you just ignore the Kanji you don't recognize?

I also don't think it's very useful to try and divorce meaning from reading. You gotta know the meaning. The reading and meaning are connected. Depending on the meaning of the word it's in, a Kanji can be read in different ways.

Vempele and Vosmiura, here's another question. I'll pick one simple example, but it's nowhere near the only one:

Quote:表 おもて 1: surface;
2: face (i.e. the visible side of an object);
3: front (of a building, etc.); obverse side (i.e. "head") of a coin;
4: outside; exterior;
5: appearance;
6: public;
7: first half (of an innings); top (of an inning);
8: cover (for tatami mats, etc.);
9: (Computer terminology) foreground

表 ひょう table (e.g. Tab 1); chart; list

How exactly do you decide on the reading, but more importantly, the meaning of 表 in 2 to 3.2 seconds?

The only way to know how it's read and what it means is by having an example sentence, and reading it.

And there are other examples of single Kanji, compound words, or verbs, with multiple readings. And there are many other examples of words with multiple meanings. Given that, I just don't understand what the point of studying words completely out of context is. You have to have an example sentence, and you often have to turn to it. Especially early on, as you're adding the word and learning it.

Stansfield123, several questions so I guess I can just describe how I've been doing things lately.

I've been keeping both a sentence deck and a vocab deck.
- My sentence deck is for learning word meaning and usage in context. The sentences are often not i+1. It's pretty boring to read i+1 sentences.
- My vocab deck is to make everything i+1, and to check my ability to skim words reliably. I have a sentence, but I only focus on one highlighted word.

In my sentence deck I'll have usually longer sentences, and I review reading and understanding all of the sentence. These take longer to review but they are also fairly information dense. Something like 表 appears many times in my sentences in different readings, and compounds like:
- 人には長所もあれば短所もあります。コインの表と裏のようなものです。
- 人の表情は、その人の気持ちを表しています。
- 自分の気持を上手に言葉で表現するのは、むずかしいです。
- 国連加盟国はそれぞれ、国連に代表を送っています。
- 調べたことをゼミで発表しました。 パワーポイントで図や表を見せながら説明しました。
- 多くの日本人は相手の表情を気にしながら、話をしています。
- 日本語の教科書の表紙のデザインは、良くないです。
...

In the vocab deck I'll have cards for each new vocab, and the vocab is the only focus of review. These are Core sentences, or Tangorin sentences. Example:
- この表を見てください。
- 表に人が来ています。
- 私たちは万歳をして喜びを表しました。
- 彼がクラスの代表だ。
...

When reviewing just vocab, if I see 表 I'll know I should look around for contextual hints. Also since I review the sentence deck, it helps learning what context to look for.

I don't tend to spend too much time on the kanji while reviewing. I certainly won't stop my reviewing to check a kanji. This is a given as I often review on my cell phone, and I'm not going to look up RTK. At the end of the day I look at the cards I failed and that day and if I don't know the kanji then I'll look it up in my RTK deck, and maybe add another related word.

What's helped me I think is I learn related words together, e.g. if you learn compounds like this together it enforces the readings: 長男, 長女, 長短, 長所, 短所, 所得税, 関税, 税関, 課税, 課長.

Also good to group are On groups like 求, 球, 救, so you can learn the signal primitives.

Having said all that, reviewing sentences is time consuming. I've thought about not doing sentences, but I don't know how it will affect my study so that's why I keep doing them.

Stansfield123 Wrote:So I have no choice but to stop my review and look up the Kanji on this site, and read the story (I could rack my brain to figure out the Kanji without help, but it wouldn't take significantly less time and it's a needless drain of energy). That drives my review times up (I'm a little under 10 seconds/ review).

How do you deal with that issue? Do you just ignore the Kanji you don't recognize?

After I'm done reviewing, I browse "rated:1:1 deck:current", look up the words in EBWin, any misrecognized kanji (plus whatever kanji I thought they were) in KanjiDic. If I notice I'm having trouble with a kanji, I try to add more words for it.

Stansfield123 Wrote:How exactly do you decide on the reading, but more importantly, the meaning of 表 in 2 to 3.2 seconds?

おもて is the default reading, so if I had to learn it now, the front would just be 表 (but since I already know it, the back would be 裏 ). For ひょう, I'd pick a short expression, maybe 表に載せる.

As I said, I only SRS readings for Japanese. I'd already learned ~2000 words just from reading native materials before I even started doing that.