Back

dismissing sentence mining for a better japanese mining?

#1
Hello i am somebody who has just finished rtk, i've been reading this forum and it gave me the motivation of doing 100 kanji a day, the last two days of rtk i did 165, and this forum have given me many goods ideas of learning a language.

Now from what i've been reading here, the best and faster way to go after rtk is the ajaat method or antimoon method of sentence mining, using the classification of ko2001, which i think is a good idea and i was willing to follow, but is not what i wanted to do after rtk, i just want to start having fun with real japanese, and for me the idea of doing another "crusade" is not very attractive.

Now i'm going to share the way i learn english and the way i intend to learn japanese, i learn english without even try, all i did was having an electric dictionary and just the wish of finish the video games i was playing at the time, i was like twelve. Without even notice i was capable of reading and somehow speaking english, a natural acquisition of the language. I only have been doing english input and nothing of output so i'll probably make mistakes writing this post though.

But this natural acquisition it may be too hard for japanese, with the huge barrier of the kanjis. So i did rtk but still the problem with kannas and pronunciation, thats why sentence mining is a good systematic way of studying japanese but not so fun like playing video games all the time, so then i need another systematically method that allow me to have fun in japanese while i learn, this is what i'm planning to do

1. find an article of a topic i like with 100 new vocabulary words, it could be a wikipedia article.
2. learn those 100 words using iverson method, iknow method, or any mnemonic that works.
3. read the article, understanding every word, first i will not understand the article completely because i don't have the natural understanding of grammar but later i would acquire doing it this way.
4. put the name of the article in a new anki deck.

Each time i search for a new article it would have to be longer, i could begin with the extraction of a little part of an article in wikipedia, and in the end a whole article from wikipedia.

The benefits of studying this way i believe they are:

- more fun to read than those sentences.
- the words are learned in their natural form in real context, which i believe is the best way to know a word at the long term.
- you don't have to do the task of adding and reviewing 10000 cards on anki, probably just 100 will do.
- each time you review an article you'll try to read it faster, giving you better abilities of reading.
- a faster way of understanding grammar.

I strongly recommend using lingq with this method, in lingq you could easily keep tracks of the words you know, and it allow you to import any material for private use.

I don't know if this would be more effective than sentence mining, maybe someone have already try it and can comment about it, i personally believe that is a better way of study, what do you guys think?
Reply
#2
I thought about that for news articles. Cos sometimes one little sentence on it's own is boring. I've found that after doing KO2001, something like this is far more access-able. I think if you were reviewing short articles or paragraphs, eventually you're reading speed would increase quite quickly I would think.

It's easy to spot that you're not a native English speaker btw xD
Reply
#3
I tried native sentences immediately after RtK, but it was too hard. Basically you're trying to learn vocab + kanji readings + grammar all at once, which is really tough and goes against the whole i+1 principle of learning.

Sources like 2001KO are popular simply because they have pre-made English translations, and are fairly well structured. This is the biggest issue I have with the AJATT method; if you're a total newb and you start mining native sentences, how do you know if you're accurately capturing every nuance of a new sentence?

I'd be interested to hear from anyone on this board who went straight to native sentences from RtK and was really successful; I'm sure there are a few brave souls!
Reply
May 16 - 30 : Pretty Big Deal: Save 31% on all Premium Subscriptions! - Sign up here
JapanesePod101
#4
Good English translations actually really help me get a feel for what a particular piece of grammar actually conveys! It's a very good point that a total newbie would have no clue without aid if they are getting the correct meaning from something. I too tried going straight from RTK to native sentences but there was just too much and KO was far faster, it definitely paid off.
Reply
#5
mezbup Wrote:It's easy to spot that you're not a native English speaker btw xD
lol, yeah even i reading the post found it unnaturally, but is because i don't practice output, i'm too lazy, i never have study english nor i will, but i know if i do some output practice like writing i won't have any trouble.

aphasiac Wrote:if you're a total newb and you start mining native sentences, how do you know if you're accurately capturing every nuance of a new sentence?!
yeah is tough but having to read content that you enjoy will make it easier, and each time you read the same article you'll understand a little bit more. you'll capture the language in a natural way, just like you did with your native language.
Reply
#6
aphasiac Wrote:This is the biggest issue I have with the AJATT method; if you're a total newb and you start mining native sentences, how do you know if you're accurately capturing every nuance of a new sentence?
Well actually I think that capturing the nuances of sentences is precisely what AJATT is good for. Think about how you learned, or rather acquired, all the nuances in your native language. For instance, I'm sure you were never explicitly taught the difference between "large" and "big" but you would have no problem using them both correctly. Things like this are never taught explicitly (in fact it would probably be quite difficult to articulate rules like this), only through huge amounts of input and hearing the words used in real sentences do you acquire the subtleties of the language.
Reply
#7
Indeed but I believe that having a grounding in something like KO will make it easier to mine native material, if you do follow AJATT I think this grounding can make the process quicker and easier. Everyones different though.
Reply
#8
There's a fine line here. Using real Japanese to learn works wonders. But you said yourself Kanji is a big obstacle. On top of that, common grammar and vocabulary will be a huge obstacle. I use pre-made sentence packs for vocabulary and grammar for the same reason I used RTK: To use a structured means to get the point across very, very quickly. However, if that was all I did, then I would be horrible at Japanese.

To really know Japanese comes not only from immersion (which should be occuring even during the "pre-packaged" phase), but learning from native material.

It doesn't have to be an either/or here. You realize that RTK is equivalent of KO2001 (iKnow, JLPT 4) and Tae Kim (UBJG, Genki) in that you're using explanation in your native language (or stronger second language) to understand Japanese a little bit better. Once you have that leverage, then use that understanding to more effectively learn from native material.

If it helps, AJATT's author Khatzumoto used the book "All about Particles" as his first group of sentences. That gave him the leverage for when he started mining real Japanese. The "pre-packaged" material was not a base, but a jumping off point.
Reply
#9
As tempting as the whole "fun with Japanese" thing is, its not really optimal in terms of time management. Like everyone before me pointed out it's just so much easier when using a structured source in the beginning. Interestingly spending lots of time mining 2-3 sentences from manga doesn't really spell fun, so it defeats the whole purpose of doing it.

Best way is to balance fun & learning performance and I find such balance in SRS and KO2001. I'm at the very beginning and I see progress with each sentence I add/review, thats really good for my motivation and its fun.

Learning 1000+ most common kanji, 4000+ words, 3100+ sentences with audio = yes please. Whats even better is that it can be done within 3 months or so. Add RTK to that and you have about 6 months of time spent and in return you have a solid groundwork for reading & decent vocab. That leaves you 12 months to make it fluent and you're on par with an idol for many: Khatzumoto Wink
Reply
#10
nukemarine, i do believe that kanji is a big obstacle but just for learning naturally, so i needed the help of rtk, my point is that grammar and vocabulary indeed can be learned naturally. I don't discard the help of structured material with explanation to grammar, i think it just helps you to notice the rules of grammar maybe make it a little bit fast to really learn unconsciously which in my opinion is the only way to understanding grammar, so something like teach yourself basic grammar is probably a good idea but in a language there are so many rules that to learn then explicitly it would be a waste of time and energy. i should have pointed that i have a very basic understanding of japanese grammar, and i think it is needed when you want to learn fast, but nothing to big to say i studied grammar for learning japanese, i think the time spending in reading grammar rules should be better spend in reading natural content.

now for something structured with no real context like ko2001, is something that everyone here agrees that it works and i'm no denying it, but it is the fastest way? i think in the short term ko2001 is attractive and not so hard as reading a whole article that's why nobody has tried using articles, and i accept that i'm going to feel lost, but it is only at the beginning, the benefits in "article mining" instead of structured sentence mining are going to show in the long term. easier to review vocabulary only reading the article, less cards in anki, reviewing topics you like, more exposure to grammar.

I don't know thurd said that he is having fun with ko2001, maybe like i just did 100 kanjis a day i'm thinking that sentence mining is going to be another test of will i don't want to have that pressure again, so i haven't decided doing ko2001, thurd also say that ko2001 is fast, yes but theoretically article mining is not slower, for instance instead of adding 100 sentences with 100 new words a day i'll be adding one article with 100 words, i still haven't decided what to do, still i think is better idea reviewing articles in the long term, but if i found out that doing structured sentence mining is better i'll do it then.
Edited: 2009-11-08, 8:16 am
Reply
#11
thurd Wrote:Like everyone before me pointed out it's just so much easier when using a structured source in the beginning.
And so much less fun. Wink
Well, it's a personal thing I guess. "Structured sources" don't seem to work for me personally.
Reply
#12
@the OP, I can see where your coming from dude and I agree. However, you say that you don't want to face the grind of srs again, but if you are going to follow the sentence mining method of learning...it's inevitable. Native sources will always be more fun, and if you have a basic grounding in the language, you say you have basic grammar, then you are probably ready to jump into native material.

Nukemarine is just basically saying some people prefer to use KO or similar pre made decks as a stepping stone to begin with. Going in at the deep end is also a proven method so I say goodluck Smile
Reply
#13
WeTsTICK thanks though i'm not planning to stop srs at all, what scares me about ajaat or ko2001 are those big numbers i should add and review, 10000 senteces, 3000 senteces. Right now i have to review 400 cards of the rtk deck, i imagine doing 10000 sentence and having to review 1000 facts, is too much. So i'm planning instead of adding sentences to srs just add an entire article. All day i'll be memorizing words in the short term memory later i'll read the article with those words, and then i'll put the name of the article in anki, so in the end instead of having a deck of 10000 cards i will have a deck of 100 cards.
Reply
#14
aphasiac Wrote:if you're a total newb and you start mining native sentences, how do you know if you're accurately capturing every nuance of a new sentence?
You don't. You absolutely don't. Don't even worry about it. Nuance is for intermediate and advanced learners. 'Newbs' should stick to just understanding the literal meaning of the sentence.

As you read/listen/etc more and more, you will pick up the nuances.

The more I get into Japanese, the more I notice people telling beginners 'X' means 'Y'. When I read it, I realize it literally means 'Z', but due to a TON of nuances, it would be -translated- as 'Y' for an English-only audience. They are -not- doing the beginners any favors by doing this.

In short, worry less, have more fun, and just enjoy learning Japanese. You'll get there eventually, and if you have more fun, you'll get there faster.
Reply
#15
The problem isn't missing some subtle nuances. It's very likely that, as a newbie or even as a more advanced student, you're going to completely misinterpret many sentences even if you know the vocabulary and grammar used.

In Japanese, some ideas are just expressed in a way that is very different than what we would expect.

Using sentences that come with a correct translation will teach you those differences. Repeating a misinterpreted sentence in your SRS isn't so good. It won't completely rot your brain forever and eventually you'll learn the correct meaning but it's best to avoid too many of these at the beginning.
Reply
#16
scissor61 Wrote:WeTsTICK thanks though i'm not planning to stop srs at all, what scares me about ajaat or ko2001 are those big numbers i should add and review, 10000 senteces, 3000 senteces. Right now i have to review 400 cards of the rtk deck, i imagine doing 10000 sentence and having to review 1000 facts, is too much. So i'm planning instead of adding sentences to srs just add an entire article. All day i'll be memorizing words in the short term memory later i'll read the article with those words, and then i'll put the name of the article in anki, so in the end instead of having a deck of 10000 cards i will have a deck of 100 cards.
That's where you have to stop looking at some far off destination and just do one step at a time. You don't do 1000 or 3000 or 10,000 sentences. You do so many sentences a day. At some point, you look back and see all the sentences you left in your wake. The beauty of an SRS is not that you're reviewing 5000 cards, it's that you're reviewing 100 cards a day and adding 20 or 30 more every day.

It's not about 2000 study hours. It's that you "studied" 2 to 3 hours a day, then after 2 years you look back and notice there's 1500 to 2000 hours that you utilized. On top of that are the thousands of hours you spent reading and listening to Japanese for fun.

As for your plan, sounds like you're using Anki to plan a reading schedule for you. I think others have done similar so hopefully they chime in with results. However, when I read articles (as opposed to sentences), I tend to pass-over words I don't know or gleam a bit too much from the kanji. There's nothing to "test" that I know the word.

Now, with grammar, or kanji, or vocabulary sentences I'm confronted with a card that's testing my grasp of it. It's either I know it or I don't. The benefit of this is all the kanji and grammar and vocabulary I've added are pretty well at my beck and call. So, when I'm reading or listening, these words are not the ones I need to worry about.

However, you can also try this: Use vocabulary packs (like 2001ko), but only activate sentences that have words that you come across in your articles. That just leaves figuring out more advanced grammar concepts. It's similar with what I do with J-Dramas using subs2srs, which gives a great reading resource as they also have a script.
Reply
#17
I don't see how going straight from RTK1 to native materials is even possible. How do you figure out the meaning of the sentences in the native materials with no grammatical background? How do you know how to parse the sentences or look up conjugated forms? I've seen people recommend this but they've only given the vaguest hints of how to actually do it.

Obviously everyone is going to have different experiences, but for me, learning from a textbook was much more fun than spending an hour on three sentences and only understanding 50%.

Quote:Now i'm going to share the way i learn english and the way i intend to learn japanese, i learn english without even try, all i did was having an electric dictionary and just the wish of finish the video games i was playing at the time, i was like twelve.
Had you taken any classes before that?

The antimoon site does not give advice for people with no grammatical basis. The authors of the site took years of English classes and so when they started their learning method, they already had a good grammatical basis in English.

My usual recommendation is that you are ready to use native materials as a primary learning source when you can read them at a pace and level of understanding that doesn't bore you. If you try to force it earlier than that it's going to become more tedious than using a textbook.
Edited: 2009-11-08, 12:17 pm
Reply
#18
wccrawford Wrote:The more I get into Japanese, the more I notice people telling beginners 'X' means 'Y'. When I read it, I realize it literally means 'Z', but due to a TON of nuances, it would be -translated- as 'Y' for an English-only audience. They are -not- doing the beginners any favors by doing this.
Personally when I do this what I'm trying to get across is "'X' is the natural and idiomatic way of conveying the meaning 'Y'". I assume that the person I'm talking to is intelligent enough that they can look at the 'literal' meaning of the words being used and come up with 'Z' for themselves. The point is that the beginner shouldn't be thinking of 'X' as some weird obscure case that would only be used when you'd say 'Z' in English, because it isn't so.
Reply
#19
yudantaiteki Wrote:Obviously everyone is going to have different experiences, but for me, learning from a textbook was much more fun than spending an hour on three sentences and only understanding 50%.
I've had the same problem. Its NOT FUN when you're reading one manga page an hour, trying to guess what thats supposed to mean and probably get it wrong anyway.

If sitting in a dictionary all the time after getting a glimpse of a sentence is the fun way of learning Japanese then thank you I'll pass. It would probably be a little easier if it was some digitalized text (webpage) or something, since you don't have to look up kanji in dictionary but still its a chore.

Sure with time it would get easier but by the time you got there, the guy that has been doing "not fun" SRS thing, would have had a lot more kanji, vocabulary and reading experience (!) than you. My personal opinion is that he also would have more fun then you...
Reply
#20
Codexus Wrote:The problem isn't missing some subtle nuances. It's very likely that, as a newbie or even as a more advanced student, you're going to completely misinterpret many sentences even if you know the vocabulary and grammar used.

In Japanese, some ideas are just expressed in a way that is very different than what we would expect.

Using sentences that come with a correct translation will teach you those differences. Repeating a misinterpreted sentence in your SRS isn't so good. It won't completely rot your brain forever and eventually you'll learn the correct meaning but it's best to avoid too many of these at the beginning.
How do you know that it will rot my brain forever? khatz and stephen kaufmann learned with natural content and without correct translation, i might not be the best example but i learn english only with real content, but i'll be a test if you want, if i fail and ruin my japanese forever i'll let you know.

Nukemarine Wrote:That's where you have to stop looking at some far off destination and just do one step at a time. You don't do 1000 or 3000 or 10,000 sentences. You do so many sentences a day. At some point, you look back and see all the sentences you left in your wake. The beauty of an SRS is not that you're reviewing 5000 cards, it's that you're reviewing 100 cards a day and adding 20 or 30 more every day.

It's not about 2000 study hours. It's that you "studied" 2 to 3 hours a day, then after 2 years you look back and notice there's 1500 to 2000 hours that you utilized. On top of that are the thousands of hours you spent reading and listening to Japanese for fun.
Yeah, i do have the tendency of looking things too far ahead, is in my nature, don't know why but i cannot avoid it, i don't know if is positive or negative for me, but when i finally make a decision even if it is a radical one i do it, so i was planning to do 100 sentences a day in that way in three months finish with the 10000 sentences of ajaat. really a crazy idea, right? that's why i thought in a way to do this crazy task, which is instead of doing 100 sentence per day, doing one article per day

Nukemarine Wrote:As for your plan, sounds like you're using Anki to plan a reading schedule for you. I think others have done similar so hopefully they chime in with results. However, when I read articles (as opposed to sentences), I tend to pass-over words I don't know or gleam a bit too much from the kanji. There's nothing to "test" that I know the word.

Now, with grammar, or kanji, or vocabulary sentences I'm confronted with a card that's testing my grasp of it. It's either I know it or I don't. The benefit of this is all the kanji and grammar and vocabulary I've added are pretty well at my beck and call. So, when I'm reading or listening, these words are not the ones I need to worry about.
I too tend to pass-over words I don't know, but i'll be reviewing an article that i should already know all the words so what i'll be testing is if i know or don't all the words in the article, if i cannot recall the pronunciation or meaning of one word i'll press fail, with this method i believe that reviewing so much text will give me a better understanding of grammar and improve my reading a lot, plus reviewing a lot of vocabulary faster.

Nukemarine Wrote:However, you can also try this: Use vocabulary packs (like 2001ko), but only activate sentences that have words that you come across in your articles. That just leaves figuring out more advanced grammar concepts. It's similar with what I do with J-Dramas using subs2srs, which gives a great reading resource as they also have a script.
Maybe i'll try if i fail with articles, and if i fail in that one too i'll use the established best method of learning japanese structured translated out of context sentence mining.

yudantaiteki Wrote:I don't see how going straight from RTK1 to native materials is even possible. How do you figure out the meaning of the sentences in the native materials with no grammatical background? How do you know how to parse the sentences or look up conjugated forms?
In a natural way, how you learn to speak is how you learn a second language, i don't see difference between learning a second language and learning to speak, the difficulty start because we realise that we don't understand the language and we search explanations, something that we don't pay attention as a child. i think the brain is already adapted to learn with no specifically explanation any language just by listen and reading.

yudantaiteki Wrote:Had you taken any classes before that?
No, not at all, worst yet i never studied any text that teach grammar or anything, and it surprise me, really, i didn't think it was possible to learn this way, but i did, and other people have learn this way, so is not something that i believe, but something i can say by experience

thurd Wrote:If sitting in a dictionary all the time after getting a glimpse of a sentence is the fun way of learning Japanese then thank you I'll pass. It would probably be a little easier if it was some digitalized text (webpage) or something, since you don't have to look up kanji in dictionary but still its a chore.
I should have already memorized all the words before actually reading, and at first i wouldn't try to understand all, just check if i know all the meanings and pronunciation. reviewing the text it would give me naturally the understanding of grammar.

Is going to be really hard at first but latter i'm sure that it will become "fun", is not something to enjoy in the short term is more about the long term and efficiency in reviewing.
Edited: 2009-11-08, 1:14 pm
Reply
#21
IceCream Wrote:BUT, i also think as a beginner, having translation is important too, so you can see whats going on.
Maybe but only in the beginning, having a translation would make you relied too much in that, not having one will force the brain to notice grammar.

IceCream Wrote:If you find it too difficult, you can always change your mind.

good luck! Smile
It will difficult, but i cannot change my mind before trying it for one month or two, giving the opportunity to the brain to understand in this way.
Reply
#22
I still think it could be interesting to create a structural template based on the Japanese the Manga Way book (or synthesize it from something else), but replace the series/media used with user-customized examples and share them. Might be easier to start that way than to fixate on a single, coherent and pre-existing narrative and trying to force that into a graduated structure.
Reply
#23
I am not sure why anyone would not be able to go directly from RTK/kana study into learning Japanese in context through watching Japanese video sources if Japanese transcripts are available and Rikaichan is incorporated. News, drama, anime, music... what ever a person's interest. Study the transcripts and enjoy the video. Sentence structure and grammar would be naturally learned. Though one could carry a pocket guide while waiting in lines or other lost time away from the computer.
Reply
#24
I would never be so bold as to say you can't learn Japanese by avoiding grammar explanations and SRS, et cetera. I would however say that it would be ***** ***** not to use them to enhance one's learning to the point that doing otherwise would in comparison have mediocre and torpid results. ;p

(***, etc. means 'cool and brilliant, of course.)
Edited: 2009-11-08, 4:40 pm
Reply
#25
I might be the only one with this view, but I am not so sure an artificial SRS like anki or smart.fm is better than taking on new sentences every day from media sources. Giving equal time to obscure words and phrases as one does the more common? ...Learning through natural exposure and building on that as more rare examples come up kind of makes more sense to me.
Reply