Back

sentence mining from native media

#1
Is there anybody out there that puts in anki directly the native sentence which contains the word he wants to learn, and not an example sentence taken from a vocabulary with that same vocab? I'm not talking about who does single vocab cards with an example sentence to read just in case... But actual sentence mining. I understand sub2srs is all about this but what about novels? Who uses sub2srs tends to do the
same with novels and manga? Tell me about your experience Smile
Edited: 2015-05-03, 6:07 pm
Reply
#2
[edited]

I do this. I started with this style of deck for grammar. I imported grammar points out of Shin Kanzen Master and other books in order to get N1 and N2 grammar down.

These days, I use rikaisama to grab sentences with unknown words, or with sentence patterns that I find useful for speech. This is by far the easiest way to add example sentences, and it counts for the majority of my content. To help with scanning cards quickly when I'm busy, I usually post-process the results by balding the word or phrase that I'm studying. That way, if I'm not in the mood to read the full sentence, I can focus on just the part I'm learning.

I also add sentences from manga, novels, etc. My strategy is either to undermine unknown words as I read, or take a picture of the manga with the expression/unknown word that I want to study. At some later date, I go back to the book, scan for the underlines, and enter the sentences directly into Anki. If I don't get to this (if I'm too busy, e.g.), I don't consider it a big deal. If I do get to it, I end up with some great material for study.

I find using this style of deck + a subs2srs deck to be a powerful one-two combo. The mined sentences help greatly with vocab, reading speed, and sentence structure, while the subs2srs deck builds vocab, gives me useful expressions for conversation, and elevates my listening comprehension.
Reply
#3
gaiaslastlaugh Wrote:[edited]

I do this. I started with this style of deck for grammar. I imported grammar points out of Shin Kanzen Master and other books in order to get N1 and N2 grammar down.

These days, I use rikaisama to grab sentences with unknown words, or with sentence patterns that I find useful for speech. This is by far the easiest way to add example sentences, and it counts for the majority of my content. To help with scanning cards quickly when I'm busy, I usually post-process the results by balding the word or phrase that I'm studying. That way, if I'm not in the mood to read the full sentence, I can focus on just the part I'm learning.

I also add sentences from manga, novels, etc. My strategy is either to undermine unknown words as I read, or take a picture of the manga with the expression/unknown word that I want to study. At some later date, I go back to the book, scan for the underlines, and enter the sentences directly into Anki. If I don't get to this (if I'm too busy, e.g.), I don't consider it a big deal. If I do get to it, I end up with some great material for study.

I find using this style of deck + a subs2srs deck to be a powerful one-two combo. The mined sentences help greatly with vocab, reading speed, and sentence structure, while the subs2srs deck builds vocab, gives me useful expressions for conversation, and elevates my listening comprehension.
gaiaslastlaught thank you for your answer!

Often I read, for example on jalup, they suggest you to pick up simple sentences from a vocabulary but I've seen more often than not the original sentence is easiest than those I find on dictionaries. Moreover the more I read the more I see some words can have more than one meaning so I must find a different sentence for each meaning I want to review and this takes time. Also about patterns, I think exactly the same as you, on native media you find pattern, ways of saying, and grammar usages you do not find on dictionary example sentences.

Also as my main reading media is the kindle paperwhite, where you can save sentences and also retrieve all the dictionary checks you do with even the original sentence, I think this is by far the easiest way to make cards. And nowaday that I feel Anki is not more my main study tool, I want to lose the less time possible on it.

Could I ask you what do you put on the answer side?

Also, do you review kanji too?
Reply
May 16 - 30 : Pretty Big Deal: Save 31% on all Premium Subscriptions! - Sign up here
JapanesePod101
#4
Sounds like a good system. Smile I use the Kindle app for iPad, but I also don't want to be limited in what I read, so I don't restrict myself to it. The book I'm currently reading is of the old-fashioned, dead-tree variety. A lot of the manga I read is physical too, as I prefer to own the physical copies.

My answer is usually the word or grammar point I'm studying, pronunciation, and then a definition of the term in either English or (increasingly) Japanese. If I don't know the kanji used by the word well, then the card counts as kanji review as well. I don't do a separate kanji review; my Sentence and Subs2srs decks are my only two decks in Anki.

I agree about not using dictionary sentences. Those are often lifeless to me, and sometimes make little sense without additional context. I prefer to cull example sentences from native media. The only thing to watch for is that you don't pull a sentence that's overly long or complicated. This is particularly a problem when pulling sentences from Wikipedia, whose authors seem to believe that endless sentences make one appear erudite.
Reply
#5
gaiaslastlaugh Wrote:Sounds like a good system. Smile I use the Kindle app for iPad, but I also don't want to be limited in what I read, so I don't restrict myself to it. The book I'm currently reading is of the old-fashioned, dead-tree variety. A lot of the manga I read is physical too, as I prefer to own the physical copies.

My answer is usually the word or grammar point I'm studying, pronunciation, and then a definition of the term in either English or (increasingly) Japanese. If I don't know the kanji used by the word well, then the card counts as kanji review as well. I don't do a separate kanji review; my Sentence and Subs2srs decks are my only two decks in Anki.

I agree about not using dictionary sentences. Those are often lifeless to me, and sometimes make little sense without additional context. I prefer to cull example sentences from native media. The only thing to watch for is that you don't pull a sentence that's overly long or complicated. This is particularly a problem when pulling sentences from Wikipedia, whose authors seem to believe that endless sentences make one appear erudite.
Understood! So no kanji writing at all? Sorry I switched recently to a recognition deck and I am still in a phase of uncertainty Tongue

Ahaha you're perfectly right about the wikipedia thing xD But as I'm trying to read only things at my level I ended up finding that the actual sentences are perfect to put them on anki Tongue

Thank you very much for all your suggestions!
Reply
#6
I do some kanji writing occasionally outside of Anki. Before this I mainly used Skritter to study vocab, so I got lots of kanji writing practice. Most of my input these days is on a computer, so I'm not worried too much about writing. My basic kanji writing skills are decent enough that I feel like if I have a need to pick it up sometime in the future, I can.

Good luck!
Reply
#7
cophnia61 Wrote:Is there anybody out there that puts in anki directly the native sentence which contains the word he wants to learn, and not an example sentence taken from a vocabulary with that same vocab?
Well there was this one guy, who wrote a blog called All Japanese All The Time about it, but he started selling the decks he made, and spent all the money on coke and hookers.

Sure, he learned Japanese, but he's dead now. A hooker slit his throat for his stash. So best avoid native stuff altogether if you don't want to end up like him.
gaiaslastlaugh Wrote:I do this.
It's been nice knowing you.
Edited: 2015-05-05, 4:30 pm
Reply
#8
What I do for manga is basically the same as what I do for games that I put in this thread:
http://forum.koohii.com/showthread.php?tid=12678

The only difference is I use a tablet to read them on and just screenshot the page and upload it to box.com then do all the other stuff after I have enough. If I was using physical manga I'd probably put stickers/tabs on the pages with new words and then scan them later. I've found images really help a lot with both recall and really cementing the concept of the word in my mind.
Reply
#9
Stansfield123 Wrote:
cophnia61 Wrote:Is there anybody out there that puts in anki directly the native sentence which contains the word he wants to learn, and not an example sentence taken from a vocabulary with that same vocab?
Well there was this one guy, who wrote a blog called All Japanese All The Time about it, but he started selling the decks he made, and spent all the money on coke and hookers.

Sure, he learned Japanese, but he's dead now. A hooker slit his throat for his stash. So best avoid native stuff altogether if you don't want to end up like him.
gaiaslastlaugh Wrote:I do this.
It's been nice knowing you.
Stansfield123, there's nothing substantive about this message, and it's vaguely judgmental and insulting. It does nothing to further discussion on this topic. Consider this a moderator's warning to put more thought behind your posts.
Reply
#10
So, I used to do take sentences from native texts, but then I stopped for several reasons.

First, let me explain that I put sentences on the front of the card. For that reason, my ideal sentence is concise, because I want to read it quickly. It demonstrates the word's usage grammatically without giving away the meaning with the context. It is the most common usage, or it is a common usage that is very concrete - many words have both a literal and a metaphorical usage; the metaphor is usually easy to understand if you first learn the literal meaning, the reverse not so much. It is a commonplace sentence that is similar to many other sentences, so that I don't learn the sentence instead of the word.

On the other hand, with native sentences, I had these problems,

1. Sentences in books are often quite long. Longer than you want to read when you're trying to review a hundred cards or more a day. You may start glossing over the sentence and only reading the word, but then why have the sentence at all?
2. The sentence may provide a ton of context that completely gives away the meaning of the word even if you have never seen it before, or the sentence may be so unique that you can easily learn the sentence and answer the question before you even read the word because you've learned to associate the answer with the sentence.
3. If I'm learning a new word from the sentence - or perhaps multiple new words in the worst case - then I may not understand the sentence correctly. I may think I do, but actually not, and be mis-learning. Dictionary sentences from J-E dictionaries come with an English translation.
4. The use of the word in the native work may be atypical, used in an 'interesting' way to create a certain flavor, used metaphorically, or worst case if it's dialogue, misused in a way that would be funny to a native speaker to set a tone for a character.

Of course, you -can- still use native sentences. 1 and 2 and 4 can be solved by being selective in your sentence choices, if you can find suitable sentences that contain the word. Authors often use the same vocabulary repeatedly, so within a book this is often possible with enough attention and patience. It does, however, mean you can't just take the first sentence with a new and interesting word. Which means you have not added the word to your learning list as you continue through the work, and you may find yourself looking it up repeatedly until you do reach a suitable sentence.
3 can be solved for works that have an English translation, or asking native speakers; also, with sufficient understanding of Japanese, you become more firm in your understanding of even sentences with unknown words in them (though you may always be in an overconfidence trap).

However, I find it easier to check a J-E dictionary in most cases and grab a nice short bland sentence. Note that I refer here to professionally published dictionaries with typically bland dictionary sentences - shogakukan, kenkyuusha, etc. The Tanaka corpus is often linked to the EDICT database for examples (that is the set on wwwjdic, jisho, rikai, many free cell phone dictionary apps, etc.), which isn't really ideal. There -are- sentences that meet my criteria in the Tanaka corpus, and on alc.co.jp, and weblio.jp, and tatoeba.org ; you just often have to look through a lot of sentences to find a suitably concise one (and btw, don't trust the furigana on tatoeba.org, it's automated not curated and in cases that are actually ambiguous, it is likely to be wrong). I still use native sentences if I can't find a good example in the dictionaries.
Reply
#11
mh... both of you have good points xD SomeCallMeChris, what about doing vocabulary cards with an hidden sentence to read only after you've identified the word in isolation? For example, considering Stanfield123 has mentioned him, when I started learning japanese I was about to start the sentence method in AJATT and I was worried about the issue of "too much context". I even wrote to Katsumoto about it and the example was from a jpop song:

深い沈黙どれくらい続いたろう

I was worried about the fact that I know the song so well I would have ended knowing the answer even without reading the word in question.

So what about doing a vocabulary card with an hidden sentence to read only after I have identified the word in isolation? This way I think I have both advantages of lack of context first and then a sentence to see the word in context and also to practice with reading and grammar.

About the atypical use I must say I always read the dictionary definition when I check a word I don't know so I would see if the sentence uses the word in an atypical way. For example the 抜ける in the beginning of the very first sentence of zero no tsukaima...

Well I will think about it and see, because I've still not decided xD

to cracky: I see, but you do production cards right? In other words you must write (or visualize) the kanji which compound the word as part of the answer? I don't think I'm ready for this, too much work for me xD
Reply
#12
cophnia61 Wrote:to cracky: I see, but you do production cards right? In other words you must write (or visualize) the kanji which compound the word as part of the answer? I don't think I'm ready for this, too much work for me xD
I do usually visualize the kanji but I don't mark it wrong if I don't. I focus on the kanji in the words when I'm first learning the card and while the card is still young. Usually by the time it's mature, the kanji just kind of comes with it. I really don't stress too much about it.

I used to hate production cards a lot until I really figured out how to present the information for myself. Now I like them a lot more than recognition cards. It's hard to say if it's too early for them, it's probably a personal choice and preference.
Reply
#13
cophnia61 Wrote:For example the 抜ける in the beginning of the very first sentence of zero no tsukaima...
抜けるような青空 A blue sky so clear as to seem like it's missing/falling away.

Seems to be something of a pat phrase.

Handy definition here:

どこまでも果てがないような透き通った青色の空を形容する表現。
Reply
#14
cophnia61 Wrote:mh... both of you have good points xD SomeCallMeChris, what about doing vocabulary cards with an hidden sentence to read only after you've identified the word in isolation?
Sure, if you want to do single-kanij-word on front, that's an option many people take. Of course, that leaves you needing a special case for kana-only words. Anyway, I don't prefer it personally.
For all my cards, I have a kana-in-context-sentence card that I learn first, followed by learning the kanij-in-context-sentence card. (Of course these are siblings from the same set of information.) Given the number of homonyms in Japanese, the value of a front-side sentence for the kana version of the card should be obvious.

Sometimes one word can have two or more highly different uses, in which case I may create different cards for the same kanji word for different uses - which may either be different words or just strongly different meanings of the same word. Especially for learning to disambiguate kanji with multiple readings (便、入る、値、and all of the numbers, to name a few), there is in some cases simply no substitute for a context sentence - or at least a context phrase.

Front-side context sentences are something that I personally feel is very important for my properly learning a word - I don't feel like context free words sink in properly as recognizing them in isolation seems to work differently from -reading-, an act which can only be done with a string of words. I don't know enough about it to say if that's my personal issue or if it's a natural part of language learning, but in any case, I find a lot of value in sentence cards, where on the other hand I find very little value in isolated vocabulary cards.

Of course you can say that you'll read the sentence on the back, but it takes a lot of discipline to actually do so, and it has no apparent reward. It becomes extremely difficult to keep it up when it's so easy to just click through to the next card. Someone did once post a bit of script that let you click once to see the context sentence and then again to actually go to the back side of the card; that would be one way to give yourself time with both the isolated word and the context sentence before you were 'done' with the card.
Edited: 2015-05-07, 1:22 am
Reply
#15
gaiaslastlaugh Wrote:Stansfield123, there's nothing substantive about this message, and it's vaguely judgmental and insulting. It does nothing to further discussion on this topic.
You just summed up what a joke is perfectly, by describing a joke, but without realizing that that's what you're describing. In a way, that's almost genius.
gaiaslastlaugh Wrote:Consider this a moderator's warning to put more thought behind your posts.
I think I'll just consider it what it is: you missing a joke.
Reply