OK so I looked up some info on English vocabulary. (Which would be my native language so it might differ for some of you.)


Gulliver's Travels is 103,803 words long but there are only 8,183 different words. Apparently, 3,489 of those words only come up once.

Everywhere I look people are saying:

Highschool Grad: 10,000
Average College Grad: 17,000-20,000
Law/Med school Grad: 30,000

So, I'm thinking 20,000 might be enough for my goals. I'm no doctor.

yea. 20,000 seems to the right number.

jettyke Wrote:

nest0r Wrote:I'm at 42,000. Still a long way to go. You need 100,000 words memorized and then you'll be fluent. Language is just strings of words, so if you memorize a big list you can easily speak and write strings of words. You just need a big enough number and then you'll be as good as, say, Jarvik7.

HUHH???

I hope that this is some bad joke

it probably is, or isn't it? lol

I read a really interesting on this on Chinese vocabulary and apparently because of the nature of Chinese people rarely use dictionaries. All my friends are surprised when I tell them I still use a dictionary for English sometimes.

I think maybe these "lists" of words are different depending on the language.

5,000 words in Chinese could be the equivalent to, say, 10,000 words in English. (I'm not saying it is but think about the idea.)

@jettyke - Huh?

@ta12121 - lol

@gyuujuice - Does Chinese have words? Where are the spaces at?

Nest0r, Yes they have words but there are no spaces.


我是美國人。

我 I
是 am
美國人 American

3 words, 5 characters.



I found this interesting as well. (It was a huge study for Spanish but just take a look.)

Learning the first 1000 most frequently used words in the entire language will allow you to understand 76.0% of all non-fiction writing, 79.6% of all fiction writing, and an astounding 87.8% of all oral speech.

Learning the top 2000 most frequently used words will get you to 84% for non-fiction, 86.1% for fiction, and 92.7% for oral speech.

And learning the top 3000 most frequently used words will get you to 88.2% for non-fiction, 89.6% for fiction, and 94.0% for oral speech.

-----------------------------------------------------------------------------------------------
250 words constitute the essential core of a language, those without which you cannot construct any sentence.

750 words constitute those that are used every single day by every person who speaks the language.

2500 words constitute those that should enable you to express everything you could possibly want to say, albeit often by awkward circumlocutions.

5000 words constitute the active vocabulary of native speakers without higher education.

10,000 words constitute the active vocabulary of native speakers with higher education.

20,000 words constitute what you need to recognize passively in order to read, understand, and enjoy a work of literature such as a novel by a notable author.



...interesting

that's interesting. It definitely does make sense when you think about it.

gyuujuice Wrote:I found this interesting as well. (It was a huge study for Spanish but just take a look.)

Learning the first 1000 most frequently used words in the entire language will allow you to understand 76.0% of all non-fiction writing, 79.6% of all fiction writing, and an astounding 87.8% of all oral speech.

Well, but that's Spanish. It took me all of three months to reach a level of fluency I'm not approaching in Japanese after many years of study. I don't buy that word count because to test it, you'd have to teach 1000 words to learners of Spanish who don't recognize any Latin word root... perhaps to 1000 Chinese native speakers... and see how they cope with those 1000 words.

Does anyone have a suggested Rikaichan to Anki deck method? You end up with just single words and no context that way, right? I mean, I'm assuming you're exploiting the quick save feature of Rikaichan and then importing, so you're bypassing the typing in of the contextual sentence... Anyway, I've tried this, but I've noticed the file I save to has different tab delimitations depending on whether it was a kanji compound, a kana-only word, etc so the import into Anki gets a bit botched. Any suggestions? Thanks!

jettyke Wrote:

nest0r Wrote:I'm at 42,000. Still a long way to go. You need 100,000 words memorized and then you'll be fluent. Language is just strings of words, so if you memorize a big list you can easily speak and write strings of words. You just need a big enough number and then you'll be as good as, say, Jarvik7.

HUHH???

I hope that this is some bad joke

no.

it's actually a good joke.

---

i actually don't like thinking about things in percentage of understanding. let's give it a shot, i'll take away a few words from a sentence.

________ to what was previously thought, Mr. Johnson likes to _______ his apples before cooking them.

If you think about it, coverage percentages are less impressive than they seem.
76% could mean not knowing 2-3 words per sentence.

kodorakun Wrote:Does anyone have a suggested Rikaichan to Anki deck method? You end up with just single words and no context that way, right? I mean, I'm assuming you're exploiting the quick save feature of Rikaichan and then importing, so you're bypassing the typing in of the contextual sentence... Anyway, I've tried this, but I've noticed the file I save to has different tab delimitations depending on whether it was a kanji compound, a kana-only word, etc so the import into Anki gets a bit botched. Any suggestions? Thanks!

Search for wrightak's Rikaichan mod, that should do what you want.

@kainzero - Thanks. ;p

Jarvik7 Wrote:If you think about it, coverage percentages are less impressive than they seem.
76% could mean not knowing 2-3 words per sentence.

Yeah, well with the comments regarding dictionaries I find it's not much of a hassle to use my ipod touch to look words up... as long as I know the reading. If I don't, using the finger-writing input is too slow and screwy to function efficiently (maybe I've got fat fingers or something, but some of the simplest-looking characters often get mis-identified).

I think of things in two categories: vocab coverage and character reading coverage. To realistically not need a dictionary at a really annoying frequency you need vocab coverage that will take _years_ to attain regardless of your methods (attaching a percentage is kind of goofy, given that we don't have a total number of words well defined -- what do we mean by something like 99.5% coverage?). However, there are a finite number of characters and a finite number of readings, so pursuing something like 99.9% coverage of all _readings_ is much more realistic and doable in a short time. Once you're in that camp you're in the no-hassle dictionary regime.

Anyway, that's just how I've thought about things for a while now. It's also my rational behind wanting to get example sentences for all joyo+some characters not covered by Core6k.

Wow, thanks for the replies! :p

No, I haven't gone through the Core 2K/6K per se; as I found that it was for the most part redundant. So, instead of deleting 90% of the deck, instead I started searching for kanji that I did not know one of the common readings of in it, yet this usually resulted in no hits or a really complicated sentence, so gave up. I had studied the Genki books and A Basic Understanding of Japanese Grammar prior to RTK1 and Core 2K/6K.

Up to now I have been following Khatz advice and doing sentences, though I must admit to be lazy, only 2000 sentences at this point. It contains 1366 unique kanji, and probably I know at least one reading of 1500+. Light novels generally cause me little trouble, but fantasy and SciFi does.

People seem to recommended different vocab centered approaches? The Rikaichan to Anki approach seems interesting - will try it out! Must admit though, I always assumed that mere exposure while using Rikaichan would yield a lot of vocabulary, like a manga with furigana does. Any thoughts?

Thanks again - you guys rock!

jarvik7 Wrote:Don't-look-it-up advice probably mostly comes from the wisdom of people who were studying without the aid of high-speed lookup electronic dictionaries (or copy & paste lookup, rikaichan, etc).

Thumbing through a paper dictionary is a definite hinderance, but I don't think looking something up in an electronic dictionary is a big deal.

Yeah, I see your point. But it kind of depends what level you're at too. If you need to look up 3 words words every sentence then whether it's electronic or not, if you're checking those words, you're no longer reading, you're decoding, which is fine but doesn't really train you like real reading does. You will learn vocabulary this way, but It's a lot slower than using parallel texts and so on because you don't have the translations right there staring you in the face from which to make simple 1 to 1 word associations. On the other hand, 2/3 words a page you can look up no problem, but when you can read that well checking those words is usually not critical for understanding, so again I often don't bother. The middle ground is probably a bit murky though.
I've recently started learning Mandarin just using parallel texts and phrase books and find I'm learning words and readings at a good pace without having to touch a dictionary at all or even use an SRS. I may start indiscriminately mass SRSing vocab later in the game though.

nadiatims Wrote:

jarvik7 Wrote:Don't-look-it-up advice probably mostly comes from the wisdom of people who were studying without the aid of high-speed lookup electronic dictionaries (or copy & paste lookup, rikaichan, etc).

Thumbing through a paper dictionary is a definite hinderance, but I don't think looking something up in an electronic dictionary is a big deal.

Yeah, I see your point. But it kind of depends what level you're at too. If you need to look up 3 words words every sentence then whether it's electronic or not, if you're checking those words, you're no longer reading, you're decoding, which is fine but doesn't really train you like real reading does.

Using rikaichan, looking up words isn't that big a deal, even back when there were a few words per sentence i didn't know.

My current policy is looking up every word i don't know and anki-ing if it crops up again on a different day (ie, not just a once off word). If it shows up a 3rd time, it gets prioritized (i have a bit of a backlog due to a holiday at present). One thing i've noticed by doing this is that there are a lot of words i was skipping as "rare" that were actually showing up sort of once a month. And once i started learning these "rare" words, my reading improved a lot.

Ideally books you read should be something you're really motivated to read, and not have too many new words per page, so if you're looking up a lot of words, you might want to try something easier. That said, when I'm reading novels, usually I use a reading technique that I found really makes new words stick (though not quite as well as a SRS), which I learned from a friend who had used it to help learn English & Japanese writing/reading (his English writing was indistinguishable from a native's 99.9 percent of the time, and from what I could tell, his Japanese was the same), the technique is:

Read Through #1: Read through the page normally, don't look up any words, just enjoy the story.

Read Through #2:Read through the page normally, looking up any words you don't know (I usually add a few sentences to a txt doc at this point, to add to Anki later)

Read Through #3: Read through the page as quickly as you can. If you usually mouth words, whisper, or mentally read aloud when you read, keep doing it, otherwise, say the new words you've looked up as you read them.

It doesn't take as long as it sounds, and even though it can be rough when I start a new book, I'm usually reading while hardly even having to use a dictionary by the time I get near the end.

Reading the book three times over sounds like a waste of time and a surefire way to get tired of the book before finishing it.

If you need to read something three times to understand it, stop and find something closer to your level.

Jarvik7 Wrote:Reading the book three times over sounds like a waste of time and a surefire way to get tired of the book before finishing it.

If you need to read something three times to understand it, stop and find something closer to your level.

I can already understand most of it the first time ^-^; , the other times are just for learning new vocabulary. Anyway, I think it's it's a useful compromise between the "don't look up anything" and the "look up every new word" methods of reading, and it doesn't take that long, plus my reading speed really increased because of it, but I realize that all methods don't work for everyone.