Hey guys,

I'm still in what some call the low level blues.
I completed RTK1, I did a third of core2k and work Genki2 and immerse.

I live in Japan now. Sometimes I find myself recognizing quite a few Kanji and understanding the meaning.
For instance, recently I was at a public restroom and I could understand what was written. "forbidden, smoke. sensor will activate, police, drive here." I had hardly an idea on how to pronounce that sentence however (except for driving)

Right now I often get the feeling that I'll never reach a point where I just read and know how to pronounce sentences in japanese. Has anyone ever had that kind of feeling.

So how does one go from recognition to pronouncing as quickly as possible? just continue graded readers and core2k? what about RTK2, would it help a great deal with that?

When I open even easy novels in a bookstore, I get the same feeling btw. How should I ever be able to read this, while at the same time "recognizing" most of the kanji...

many thanks

By pronounce, do you mean know the readings of the kanji? Such as for example (禁煙-きんえん- no smoking). In that case, practicing reading is probably one of the best ways, although I found working through the readings in the Genki lessons for the kanji at the back of the back was a good way to give me a basic group of readings. At some point when you see it everywhere you start to remember it. Also, be sure to read stuff at your level, meaning you don't have to look up every third word/kanji (although there are some exceptions)

Or did you mean actual pronunciation, like with sounds? In that case my personal recommendation is a Japanese person to give you lessons/classes, that worked pretty well for me in getting a basically accurate level of pronunciation down.

Regardless, it's not something that happens overnight, or even after a week or two. I still find a lot of stuff I don't recognize, and I certainly can't just walk into a bookstore and pick up a novel aimed at my age group and start reading it. But stuff like 出口、禁煙,目黒、目白 gets to be recognizable as I see it everywhere and use it. Personally I don't notice that I'm improving until about 6-12 months pass, then I wake up one day and say "Hey, I understood the garbled loudspeaker voice of the train conductor, the train is late due to a power outage" or something like that. Not to say that I wasn't making progress until then, but I didn't really notice it at the time.

thanks, no I'm not talking about pronouncing as in "making the right sound". I don't find that part all that hard, especially in comparison to what I mean.

I'm talking about going from recognizing kanji to actually reading japanese.

The words you mentioned I can pronounce too when seeing them, as like you said you see them all the time. but there are tens of of thousands of words in japanese that you would have to be able to read (pronoubce) so you can finally read a newspaper, right?

How is that possible without learning 3 hours a day for 10 years? not at all I guess, huh...

thanks for your comments

maxwell777 Wrote:thanks, no I'm not talking about pronouncing as in "making the right sound". I don't find that part all that hard, especially in comparison to what I mean.

I'm talking about going from recognizing kanji to actually reading japanese.

The words you mentioned I can pronounce too when seeing them, as like you said you see them all the time. but there are tens of of thousands of words in japanese that you would have to be able to read (pronoubce) so you can finally read a newspaper, right?

How is that possible without learning 3 hours a day for 10 years? not at all I guess, huh...

thanks for your comments

Maxwell, can you make an example of a word which you can understand but not read?
For example, do you know how to read a word like 警官隊 or 警察 (as you mentioned "police")?

From recognizing Kanji, the next step is recognizing words. That's what reading is: the ability to recognize words just by briefly looking at them (without the need to look at each individual character).

The way to get there is simply through practice. I doubt bothering with shortcuts (stuff like RtK 2) is worth it.

But, of course, this assumes that you already understand the words you're trying to read (you already understand the spoken version). If you don't, then, unlike in other languages, things get tricky in Japanese. With other languages, learning the language through reading/writing is pretty much the same process as learning it through listening/speaking.

With Japanese, they're two very different methods. Learning through reading, while possible, might get pretty frustrating and convoluted. I don't understand the process well enough to give much meaningful advice. I prefer to only bother reading words I already know. It just seems more comfortable to me.

So, the question is, how well do you speak the language? Depending on the answer to that, you might want to focus on getting better at the actual language, and worry about reading later, or move right along to practicing reading.

maxwell777 Wrote:How is that possible without learning 3 hours a day for 10 years? not at all I guess, huh...

50 words a day takes about 1 hour in Anki and adds up to around 20,000 words per year.

kameden Wrote:

maxwell777 Wrote:How is that possible without learning 3 hours a day for 10 years? not at all I guess, huh...

50 words a day takes about 1 hour in Anki and adds up to around 20,000 words per year.

Without considering reviews, which accumulate... But if the only problem is with reading then there is no need to add every single word. After a while you will learn the common reading/readings for each kanji, and after this you don't need to add to Anki words which use the common reading. And even words which use less-common readings, it's not a given that you need to add it to Anki to remember how to read it.

As Stanfield said, once the spoken word clicks, it's easy to associate the right pronounciation for its written form.
One on LingQ forum I wrote that sometimes I used to misread 老人 as "ろうにん" and a Japanese person replied: "it's easy, it cannot be ろうにん because ろうにん means a samurai".
And I was like "well, thank you! (<---sarcastic tone) You are Japanese, you already know this! ._. ".
But now I know what ろうにん means and what ろうじん means, so even if "人" has more than one reading, this poses not a problem. And I don't even have an Anki deck for vocabulary ._.

Another example: 水着. The first time I saw this word I thought:

"well, it's a compound, so it's likely that it uses onyomi. 水 only onyomi is スイ and 着 only COMMON onyomi is チャク, so it could be "すいちゃく".
But then I discovered it was みずぎ and I had not problem to remember it as I already know みず and き・る.

In other words, find a way to learn the most common readings for each kanji (mnemonics, kanji in context, or whatever) and then you will need Anki only for a minority of words.
If you think about it, it's the same for meanings, you don't put on anki each single word, because as you already said, most of the time you can easily predict, and then learn, words meanings, thank to rtk keywords.

cophnia61 Wrote:Without considering reviews, which accumulate...

No I was including those, unless you think it takes 1 hour to just go through 50 new words.

cophnia61 Wrote:there is no need to add every single word

Not every single word, but there is at least 20,000 words that are worth adding if you want to be efficient about learning to read.

kameden Wrote:

cophnia61 Wrote:Without considering reviews, which accumulate...

No I was including those, unless you think it takes 1 hour to just go through 50 new words.

cophnia61 Wrote:there is no need to add every single word

Not every single word, but there is at least 20,000 words that are worth adding if you want to be efficient about learning to read.

Sorry, I don't use Anki for words so I don't know very well how it works and how much time does it take.
What I meant to say is that to solve Maxwell issue you don't necessarely need to learn 20k words with Anki (not that I think this is not useful or infefficient, on the countrary!).

cophnia61 Wrote:Sorry, I don't use Anki for words so I don't know very well how it works and how much time does it take.
What I meant to say is that to solve Maxwell issue you don't necessarely need to learn 20k words with Anki (not that I think this is not useful or infefficient, on the countrary!).

To be fair I was responding to a post speculating how long it would take to be able to pronounce words with reading.

maxwell777 Wrote:How is that possible without learning 3 hours a day for 10 years? not at all I guess, huh...

kameden Wrote:

cophnia61 Wrote:Sorry, I don't use Anki for words so I don't know very well how it works and how much time does it take.
What I meant to say is that to solve Maxwell issue you don't necessarely need to learn 20k words with Anki (not that I think this is not useful or infefficient, on the countrary!).

To be fair I was responding to a post speculating how long it would take to be able to pronounce words with reading.

3 hours a day for 10 years

Yes you're right! I wanted only to add that even if you don't add 20k words to Anki you can still learn at least the most common readings and then you will learn and remember easily new words as you encounter them in the wild.
In other words you can learn how to read words even with a minimum use of Anki, (which maybe is inefficient). I said it only for Maxwell sake for him to know that you can learn how to read japanese even without relying heavily on Anki Just in case he is one of those people which hate Anki xD

This might be something that might help you. I read it on a blog and he calls it "Massive (or Moderate (or Minimal)) Context Bold cards (MCBs)" which is ajatt's MCD inspired but imo easier. Here is a post about dealing with readings of kanji. I actually used this technique while I was taking Japanese classes a few years back and it helped me do well on the reading exams which consisted of an entire page of text in Japanese and having to add ふりがな on top of the necessary words.

http://jeffben.com/mcbs-part-1b-one-character-per-card/

maxwell777 Wrote:So how does one go from recognition to pronouncing as quickly as possible? just continue graded readers and core2k? what about RTK2, would it help a great deal with that?

Without going through all of RTK2, you could memorize the perfect groups (kanji that always have the same reading and share the same primitive). There are 182 groups containing 429 perfect joyou kanji. You probably already know a lot of this anyway, so you could relatively quickly be certain of the reading of close to 1/4 of the kanji you encounter. Here's a thread with some linked anki decks and spreadsheets.

Parallel texts with audio is always the best way to learn to read, and it would be nice to see someone set up a whole system for people in your situation.

Aozora Bunko has books out of copyright and translations also out of copyright.

Stansfield,
that makes sense even though it's not what I hope to hear, as I went through RTK first, because I thought it would help me a lot in aquiring vocabulary, because I'd be able to read. If I have to learn the words through speaking and listening first though, that means it won't help for that.

kameden,
if you can learn 50 new words + reviews in an hour, good job. I can't do that. After a while it's 150 to 200 sentences everyday, as the reviews stack up. And even if you can do that, it's not nearly like you can read a newspaper with those 20,000 words at the end of the year, right? That will take many years, even with hours of practice a day. My guess was 10 years to get quite good at reading something like a newspaper or a novel really.

Cophnia,
The Ronin example seems obvious, yes, and I agree that there are those. But isn't that kind of thing the exception. Most of the times you don't have these clues when learning new words right. Like Stansfield, you're also saying reading is easier when you know the word. Which makes me wonder sometimes why I learned all the Kanji, if it's not so much help with getting better at the language, which is what you guys say if I understand you right?

Rawrpk,
thanks for the link - isn't that more about getting better at the meaning for certain characters and their readings though, as opposed to reading entire words and sentences?

yogert909,
that's a great concept, thanks a lot. I already downloaded the Anki file for these perfect groups and am going through that.

kapalama,
I agree it's helpful. Even though at this point I have to stop the audio after every sentence and repeat an entire part after 30 seconds at most. Certainly useful, I wish it was more fun though lol

maxwell777 Wrote:it's not nearly like you can read a newspaper with those 20,000 words at the end of the year, right? That will take many years, even with hours of practice a day. My guess was 10 years to get quite good at reading something like a newspaper or a novel really.

It depends on how much you read of course. If you have a big vocabulary base it shouldn't take too much reading to get used to grammar. Probably around 1000 quality hours spent reading is enough to more or less understand any grammar you encounter without having to reread the sentence. Of course the more you read the better you get at reading, similar to how a native who reads a lot will be able to read better than a native who does not. Years it's such an arbitrary metric. You could read the same amount in 10 years as you do in 1, it just depends on how much overall reading you do, not how long of a period you do it over (although that may play a small factor).

The more words you know, the easier it gets to learn new vocabulary. One thing you can do is learn all the vocabulary in Genki (and how to write it) and then work your way through an easy manga that isn't too light on kanji, adding all new vocabulary to Anki. It'll take ages at first but after a couple of volumes it should get much faster.

Songs are also an effective way of learning vocabulary - find a song you like that's easy to translate and learn it by heart. The vocabulary should stick.

What I intend to do once I'm done with RTK is to have two decks: one for all unknown vocabulary I encounter while reading, and another one where I assign a single word to each kanji, as simple and useful as possible, to phase the Heisig keywords out and hopefully learn some vocabulary. I'm sure most kanji are used in more than one useful word, but it's a start, and this way I can consolidate the kanji in my mind past the Heisig stage. Plus, as I said, the more words you know, the easier it is to learn new ones.

kameden Wrote:

cophnia61 Wrote:Without considering reviews, which accumulate...

No I was including those, unless you think it takes 1 hour to just go through 50 new words.

cophnia61 Wrote:there is no need to add every single word

Not every single word, but there is at least 20,000 words that are worth adding if you want to be efficient about learning to read.

You're probably right about the 20,000 words, but your estimate is way off on the one hour/day, if you add 50 cards to your deck every day.

Your daily reviews would quickly rise, well over a thousand/day. And that's not even the biggest problem: adding 50 new words, every day, would make the deck frustratingly difficult.

Anki is best used with easier material, that allows for quick reviews. That's why so many people recommend i+1 type sentences (i+1 is terminology from Krashen's input theory), instead of vocab reviews.

maxwell777 Wrote:Stansfield,
that makes sense even though it's not what I hope to hear, as I went through RTK first, because I thought it would help me a lot in aquiring vocabulary, because I'd be able to read. If I have to learn the words through speaking and listening first though, that means it won't help for that.

Didn't say you can't learn Japanese through reading. Just that it's different from how you would do it in other languages (and I haven't tried it, so I can't really be of much help on how you'd go about it, and on how difficult it really is).

What I have done is this: I did sentence reviews, with both the audio and the Kanji version of the sentence in the question (and I passed the cards when I understood the sentence - irrespective of whether it was the audio or the Kanji that helped me understand). I used a bit of Core6k, but later I switched to Nayr's deck, because it's better.

I have no idea what made it work: could've been mostly my RtK knowledge, could've been mostly the auditory memory. Most likely it was closer to a 50/50 split. What matters is that it really worked, and allowed me to get through new material much faster than if I had only used reading or only used audio. And that's what it comes down to: how much (comprehensible) material of increasing difficulty can you get through. Once you get through enough, you will have learned Japanese.

In other languages, the quickest way to get through large amounts of materials is reading (at least if you love reading). With Japanese, reading is tough, but you can use audio to help you. (in the interest of full disclosure, I can't actually read Japanese...or rather, I have no idea how close I am to being able to; so far, I've been focusing on listening comprehension, my reading is limited to the occasional blog - for anything bigger, I only do it if it has furigana).

Stansfield123 Wrote:Your daily reviews would quickly rise, well over a thousand/day.

Not to get to far off topic, but do you really think reviews become well over 20x what you're adding? You'd have to get like 40% correct for something like that to happen. I've never gotten more than like 8x what I was adding.

kameden Wrote:Not to get to far off topic, but do you really think reviews become well over 20x what you're adding? You'd have to get like 40% correct for something like that to happen. I've never gotten more than like 8x what I was adding.

With 40% correct, you'd be adding 50 cards, plus 60% of your daily reviews, every day. Things would escalate pretty quickly. I don't know exactly how high that would go in a year, but I gotta think it would have to be above 5,000 daily reviews.

With a more reasonable 80% or so correct rate, I think you'd be hitting 1,000 after 4-5 months. At least that's what I remember from my experiences, and from threads of people posting their volumes in the past. And trust me, as a beginner, with a vocab deck of the 20,000 most popular Japanese words, you won't be getting an 80% correct rate. Japanese vocab, in isolation, is difficult to remember.

Stansfield123 Wrote:

kameden Wrote:Not to get to far off topic, but do you really think reviews become well over 20x what you're adding? You'd have to get like 40% correct for something like that to happen. I've never gotten more than like 8x what I was adding.

With 40% correct, you'd be adding 50 cards, plus 60% of your daily reviews, every day. Things would escalate pretty quickly. I don't know exactly how high that would go in a year, but I gotta think it would have to be above 5,000 daily reviews.

With a more reasonable 80% or so correct rate, I think you'd be hitting 1,000 after 4-5 months. At least that's what I remember from my experiences, and from threads of people posting their volumes in the past. And trust me, as a beginner, with a vocab deck of the 20,000 most popular Japanese words, you won't be getting an 80% correct rate. Japanese vocab, in isolation, is difficult to remember.

Adding 50/day at 80% accuracy you'd be at about 500 reviews after 5 months(or 10x). Accuracy makes a big difference though as you get large compounding effects on higher numbers of relearning cards. Just to be clear, these numbers aren't counting relearning. Rather, the numbers are simply the workload you'd see first thing in the morning before starting reviews.

I made a chart a while back and adapted it to 50 new cards/day.

yogert909 Wrote:https://docs.google.com/spreadsheets/d/1...rmat=image

That is great work. Very enlightening. Shows that you're basically doubling your work load (after counting the re-learning numbers) every time you lose 10% in accuracy.

It's something everyone should see before starting SRS-ing, because the intuitive thing to do is to condense a large volume of information into fewer, more difficult cards. In reality, the way to save time is to split it up into more cards, until you have at least 90% accuracy.

yogert909 Wrote:I made a chart a while back and adapted it to 50 new cards/day.

This seems to fit with what I experienced, I usually got around 90%.

maxwell777 Wrote:Rawrpk,
thanks for the link - isn't that more about getting better at the meaning for certain characters and their readings though, as opposed to reading entire words and sentences?

It actually will give you the readings of the entire word but you "divide and conquer" by making multiple vocab cards in this format. Here is the same blogger's older example of how he uses MCBs in a sentence format: http://jeffben.com/mcbs/
Was that something more of the lines you were looking for?


I probably should had given examples of how I personally do it. Let's use a word posted here 「警察」police.

Card 1
けい察

-----
警察
police

警察は彼に目をつけている。



Card 2

警さつ
-----
警察
police

警察は彼に目をつけている。



Other words can of course be derived using other forms of the kanji. Let's say we want other words formed by the first kanji, 警:
警笛 けいてき (n) horn; alarm; whistle; foghorn
警部 けいぶ (n) police inspector
警句 けいく (n,adj-no) aphorism

You can do the same with the 2nd kanji, 察 :
洞察 どうさつ (n,vs) discernment; insight; (P)
診察 しんさつ (n,vs) medical examination; (P)
明察 めいさつ (n,vs) discernment; penetration; insight; intellectual acumen

[Advanced] A cool way to learn new words : this topic sorta takes this concept into an extreme mindmap form which I never really thought of doing. Which one of the replies lead to a site I used while I was still taking Japanese classes: Rikai Kanji Map

tl;dr: divide and conquer readings. Use kanji in other forms/words for reinforcement.

Stansfield123 Wrote:It's something everyone should see before starting SRS-ing, because the intuitive thing to do is to condense a large volume of information into fewer, more difficult cards. In reality, the way to save time is to split it up into more cards, until you have at least 90% accuracy.

Well, according to the supermemo guys, peak efficiency is around 70-80%(chart) because, as they put it:

Quote:The greatest overall knowledge acquisition rate is obtained for the forgetting index of about 20-30% (see Figure 3). This results from the trade-off between reducing the repetition workload and increasing the relearning workload as the forgetting index progresses upward. In other words, high values of the forgetting index result in longer intervals, but the gain is offset by an additional workload coming from a greater number of forgotten items that have to be relearned.

Personally, I go for 90% because it's more satisfying and I'm not in a hurry, but I don't think it's nearly as efficient as if I was closer to 80%.