RandomQuotes Wrote:If you know 16,000 words you should be advanced as far as reading goes. If your're not, I think you need to consider what you mean by the word "know."

For sure. I think I'm just under 16k (though its kind of hard to measure, obviously), but I consider myself "advanced" for reading. I'm by no means perfect but I can read and enjoy Japanese novels in Japanese quite easily. On average I come across one new word every page or so.

My question for kameden then is, where do you *find* these words so quickly? Its all well and good saying you can whack 70 words into Anki and review perfectly without an hour (and hey, maybe you can) but that's only if you have some good reference source in front of you to get them from. I'd love to spend more time drilling vocabulary and learning hundreds of words a day, but I literally can't because I don't have the time to actually read enough to find that many new words.

100 reviews every 10 minutes is about right I think, I try to aim for around that.

Roketzu Wrote:

RandomQuotes Wrote:If you know 16,000 words you should be advanced as far as reading goes. If your're not, I think you need to consider what you mean by the word "know."

Even if you do know 16000 words inside out you'll still constantly come across words you've never seen before. I'm not sure what you mean by advanced, but I'd consider native level advanced, as in maybe seeing a few new words per book. At 16000 you'd be seeing new words every few pages.

At 16,000 words (lexemes) you're going to have roughly 96-97% coverage of the language, Considering 97-98% is where you hit the ability to read without a dictionary and learn the meaning from context(another 3-4000 words), I would safely say that is advanced. Native level is beyond advanced, in almost every scale I've seen for learners of a language. Native level is a noble goal, but to think you're not advanced until you hit native level is a fallacy.

The genral scale goes something like this:
Low Beginner, High Beginner, Low Intermediate, High Intermediate, Low Advanced, High Advanced, Native

RandomQuotes Wrote:If you know 16,000 words you should be advanced as far as reading goes. If your're not, I think you need to consider what you mean by the word "know."

I think you need to consider what you mean by the word "advanced".

NightSky Wrote:My question for kameden then is, where do you *find* these words so quickly?

It's not difficult because 16,000 words really isn't that much. I constantly run into new words. I would definitely not consider myself advanced at all. I think you are either reading really easy content or are exaggerating your reading ability.

NightSky Wrote:100 reviews every 10 minutes is about right I think, I try to aim for around that.

I go slightly slower than that. 600 reviews takes me about 70 minutes. I usually get 500-600 a day.

RandomQuotes Wrote:The genral scale goes something like this:
Low Beginner, High Beginner, Low Intermediate, High Intermediate, Low Advanced, High Advanced, Native

That scale is kind of meaningless without any information of what those words mean. In my opinion advanced is around native level. If we're talking vocabulary, average high school graduate native knows 40-50k, so if I had to define it:

Beginner: 1-10k
Intermediate: 10-30k
Advanced: 30-50k

Anything beyond that would be expertise in a specific field, but probably beyond general knowledge of a language, aka words that (most) natives know.

Your scale puts anyone below JLPT N1 at beginner. Almost everybody else would consider even the N2 (~6000 words and everyday grammar) at least intermediate.

RandomQuotes Wrote:

Roketzu Wrote:

RandomQuotes Wrote:If you know 16,000 words you should be advanced as far as reading goes. If your're not, I think you need to consider what you mean by the word "know."

Even if you do know 16000 words inside out you'll still constantly come across words you've never seen before. I'm not sure what you mean by advanced, but I'd consider native level advanced, as in maybe seeing a few new words per book. At 16000 you'd be seeing new words every few pages.

At 16,000 words (lexemes) you're going to have roughly 96-97% coverage of the language, Considering 97-98% is where you hit the ability to read without a dictionary and learn the meaning from context(another 3-4000 words), I would safely say that is advanced. Native level is beyond advanced, in almost every scale I've seen for learners of a language. Native level is a noble goal, but to think you're not advanced until you hit native level is a fallacy.

The genral scale goes something like this:
Low Beginner, High Beginner, Low Intermediate, High Intermediate, Low Advanced, High Advanced, Native

Fair enough, I know it may just be semantics but I feel that someone at an advanced level would feel like they had little left to learn, and I think anyone around that level (16k vocab) would be fooling themselves if they thought they didn't have a whole lot more to learn.

kameden Wrote:I think you need to consider what you mean by the word "advanced".

I would consider JLPT N1 to be advanced Japanese, and that covers roughly 10,000 words or so.

kameden Wrote:That scale is kind of meaningless without any information of what those words mean. In my opinion advanced is around native level. If we're talking vocabulary, average high school graduate native knows 40-50k, so if I had to define it:

Beginner: 1-10k
Intermediate: 10-30k
Advanced: 30-50k

Anything beyond that would be expertise in a specific field, but probably beyond general knowledge of a language, aka words that (most) natives know.

When you say 40-50k, are you talking about individual words, lexemes, or word families? 50k individual words? Sure, 50k lexemes highly unlikely.

Vempele Wrote:Your scale puts anyone below JLPT N1 at beginner. Almost everybody else would consider even the N2 (~6000 words and everyday grammar) at least intermediate.

Do you think N1 is fluent or something? It's the high end of extremely basic comprehension, which is where it falls in my scale.

By logical extension, anyone who knows hold less than a kanken 1級 is a beginner at kanji.

RandomQuotes Wrote:By logical extension, anyone who knows hold less than a kanken 1級 is a beginner at kanji.

I don't know what that is, but if you were to divide it up it would be something like 2000/3000/3500 at least in terms of recognition.

kameden Wrote:It's not difficult because 16,000 words really isn't that much. I constantly run into new words. I would definitely not consider myself advanced at all. I think you are either reading really easy content or are exaggerating your reading ability.

Ah perfect, this kind of answers my questions perfectly.

You don't know 16,000 words.

Its really simple, because I know myself I'm around almost exactly the same level. Yes there are lots of words I don't know and plenty of things I don't understand, and I don't deny that. But I know for sure, the act of *finding* those 70 unknown words takes much much longer than the time it takes to throw them into Anki and review. I've always done tons of anki reviews, but now my number of daily reviews are really manageable because I'm not capable of finding words fast enough otherwise.

I think it normally takes me about 2-3 hours of reading native material to find 100 unknown words. Apparently you can find 70, put them into Anki and do all of your reviews, all within 1 hour?

I believe its totally possible for a person to do that, but not if they already know 16,000 words and don't have another resource that throws new vocab at them quickly.

(FWIW I'm still reading 1Q84. I don't claim its the 'hardest' content, but its a Japanese novel by a nobel prize nominated author, its good enough for me)

The kanken 準1級 is less 3000 characters, the pass rate for that is less that 20%. The 3000th character on the frequency list is 癇．If you were to cover from 1-3000 that would give you coverage of 99.896% out of a 32,819,412 word corpus. I'm all for increasing vocab and what not, but your standards would make a high school student and intermediate user of their native language.

RandomQuotes Wrote:The kanken 準1級 is less 3000 characters, the pass rate for that is less that 20%. The 3000th character on the frequency list is 癇．If you were to cover from 1-3000 that would give you coverage of 99.896% out of a 32,819,412 word corpus. I'm all for increasing vocab and what not, but your standards would make a high school student and intermediate user of their native language.

Firstly, I said recognition, not production. I also have no idea where you are getting your statistics from. Are you not including names? I don't consider names words, but natives can still recognize the kanji that make them up.

I think we're just misunderstanding each other then. 100 words in 2-3 hours of reading is about what I get too, maybe less if I'm reading manga or something. All I was saying is if you add 60-70 words a day, it takes around an hour for reviews.

kameden Wrote:

RandomQuotes Wrote:The kanken 準1級 is less 3000 characters, the pass rate for that is less that 20%. The 3000th character on the frequency list is 癇．If you were to cover from 1-3000 that would give you coverage of 99.896% out of a 32,819,412 word corpus. I'm all for increasing vocab and what not, but your standards would make a high school student and intermediate user of their native language.

Firstly, I said recognition, not production. I also have no idea where you are getting your statistics from. Are you not including names? I don't consider names words, but natives can still recognize the kanji that make them up.

Database of Kanji used in Japan based on a 32 million word corpus.
http://www17408ui.sakura.ne.jp/tatsum/CD...Ver2_0.xls

kameden Wrote:

Vempele Wrote:Your scale puts anyone below JLPT N1 at beginner. Almost everybody else would consider even the N2 (~6000 words and everyday grammar) at least intermediate.

Do you think N1 is fluent or something? It's the high end of extremely basic comprehension, which is where it falls in my scale.

No, I'm just saying that if that's what you call intermediate, you're using the word in a different way than everyone else. Pick a different word or you'll be forever stuck explaining what you mean by "intermediate" every time you use it in a new conversation.

People have low expectations about language learning. "Native level" is seen as nearly impossible, not as the default objective. Hence, for there to be an "advanced" for regular people to strive towards, you have to set the bar pretty low. You might get away with your definition of the word on a forum where everyone is N1+, but this isn't that forum. This is a forum for people who do RTK. Most people who do RTK don't even finish it.

Not disagreeing with your point, mind you.

Ah okay, probably misunderstanding each other then. Yes I otherwise agree with you!

At this level to learn another 100 words a day probably requires around 3 hours reading, followed by another 2 hours to enter that all into Anki and do daily reviews (of 1000 or so). I did that for a while earlier this year, its hard to keep up but if you don't have a full time job its achievable.

RandomQuotes Wrote:When you say 40-50k, are you talking about individual words, lexemes, or word families? 50k individual words? Sure, 50k lexemes highly unlikely.

I already asked this question but still I feel I don't quite understand the difference between lexeme and individual words.

I downloaed the word frequency lists that MaxHayden linked here and, while I was reading some comments on an idol's blog, I searched some of those words in that frequency list, and one of those was the 15000th entry c.ca. It was "絶賛", a pretty common word I think. And it appears after 5 mins of random reading of teen-ager's comments. And here we are talking about active usage of the word. So I think a native has a passive knowledge of way above 15000 of individual words. Or are those morphemes?

I scan that list and looked at words on 20k, 30k, 40k ranks and some of them seem rather common.

Levels aren't about raw numbers. You can't just do a statistical analysis of the words or kanji in a passage vs. what a person has studied and come up with their percentage of comprehension. That's not the way language, or reading, works.

cophnia61 Wrote:I already asked this question but still I feel I don't quite understand the difference between lexeme and individual words.

Using English examples [do, undo, redo, doable] to explain it: And individual word is any given word. That is do, undo, redo, doable are four different words.

Lexemes are the base for of any given word, think of the dictionary. do, undo, redo, doable, would all be 1 lexeme of the lemma "do"

A Morpheme is the smallest unit of a language that contains meaning so do, undo, redo, doable, would contain the morphemes do-, un-, re-, -able.

RandomQuotes Wrote:

cophnia61 Wrote:I already asked this question but still I feel I don't quite understand the difference between lexeme and individual words.

Using English examples [do, undo, redo, doable] to explain it: And individual word is any given word. That is do, undo, redo, doable are four different words.

Lexemes are the base for of any given word, think of the dictionary. do, undo, redo, doable, would all be 1 lexeme of the lemma "do"

A Morpheme is the smallest unit of a language that contains meaning so do, undo, redo, doable, would contain the morphemes do-, un-, re-, -able.

Thank you for the explanation! Sincerely I already understand the difference in english, but I don't know how to translate this in japanese. Could you give me an example?

The words in the list of Dr. Tatsuhiko Matsushita are lexemes?

cophnia61 Wrote:Thank you for the explanation! Sincerely I already understand the difference in english, but I don't know how to translate this in japanese. Could you give me an example?

So, lets looks at the words 食べる、食べさせる、食べられる、食べさせられる、食う、食品

食べる、食べさせる、食べられる、食べさせられる、食う、食品 are 6 individual words

食べる、食う、食品 are lexemes

食、-させる、-られる、品 are the morphemes

cophnia61 Wrote:The words in the list of Dr. Tatsuhiko Matsushita are lexemes?

Yes, that frequency list is a list of lexemes.

For clarification, at least based on RandomQuotes examples, I was talking about lexemes, not conjugations and stuff.

RandomQuotes Wrote:

cophnia61 Wrote:Thank you for the explanation! Sincerely I already understand the difference in english, but I don't know how to translate this in japanese. Could you give me an example?

So, lets looks at the words 食べる、食べさせる、食べられる、食べさせられる、食う、食品

食べる、食べさせる、食べられる、食べさせられる、食う、食品 are 6 individual words

食べる、食う、食品 are lexemes

食、-させる、-られる、品 are the morphemes

cophnia61 Wrote:The words in the list of Dr. Tatsuhiko Matsushita are lexemes?

Yes, that frequency list is a list of lexemes.

Wow, now it's clear! Thank you very much!

So, in the light of this, what do you think it is the number of lexeme one need to know to cover, let's say, 95% of reading comprehension? Or, what do you think is the number of lexeme one need to know to have a good understanding of things like comedy shows or teen-ager dramas? Where for "good" i mean a level where the process is enjoyable

cophnia61 Wrote:Wow, now it's clear! Thank you very much!

So, in the light of this, what do you think it is the number of lexeme one need to know to cover, let's say, 95% of reading comprehension? Or, what do you think is the number of lexeme one need to know to have a good understanding of things like comedy shows or teen-ager dramas? Where for "good" i mean a level where the process is enjoyable

According to the frequency list you linked to 95% comprehension would be at word number 9676 lexemes. However, based on research on polyglots and second language acquisition, being able to learn new words without needing a dictionary based solely upon reading, begins at like 98% or 99% comprehension. which would be 21073 and 32177 word respectively. Generally speaking, the amount of words need to be literate is much greater than the amount of words needed to be fluent. So you would, in theory, need more words to understand Yukiguni, than you would to understand GTO.

As far as my personal recommendation goes, the more words the better. I think 20k is a reasonable goal. At 30 new words a day it would take a little bit less than 2 years. As far as how to get there, I would suggest using a premade list for 1-10,000, if your interested in academics, go through the academic word list(2500 words), if your interested in literature, go through the literary word list(1600 words) and fill in the gaps. From there I would suggest actually using the language and building up your base from reading. Make cards for words where the you can't figure out the word from context.

Vempele Wrote:I have ~90% accuracy (might be as low as 86%: Anki conflates learn and relearn here, and my relearn accuracy is probably very close to 100%), 2.6 seconds per review (for new cards; my average time across all reviews is 2.8 seconds). 364 cards added per hour. Doesn't include the time it takes to look up the word in a dictionary and make the card, though.

Total time spent studying 2185 cards: 21 hours. 2046 mature cards.

Something is wrong with your math here (or we are talking about something completely different). You say that you studied 2185 cards in 21 hours. That's 2185/21=104 cards per hour but you say you add 364 cards per hour. At that rate you wouldn't even be able to study the added cards let alone the reviews. And 104 cards per hour is 34 seconds per review, not 2.6 or 2.8.