I've noticed a scattering of posts lately relating to inputting sentences into Anki (or your SRS of choice). I'd like to create one place to discuss different methodologies, theories, and ideas regarding exactly how we should be putting all of this information we're finding into our virtual brain.

Of course there is no one correct way, so that's why I've created this thread.

The problem I've been faced with lately is shortening my sentences, but the shorter your sentences are, the less context you have. How exactly do you compromise this situation?

Here's an idea that just came to me: input the whole sentence, but make the part you aren't concentrating on either a light grey or white (so you have to roll over it in order to see the text). When the sentence appears, you can quickly scan the black text to see if you understand the word, and if isn't enough context, you can let your eyes focus on the grey portion of the text.

Example:

神殿が築かれるずっと以前から現在まで脈々と続けられてきた人類の創造の営みによって

My problem is that the whole sentence just keeps going and going. The word I'm concentrating on is 脈々. So I could possibly do this:

神殿が築かれるずっと以前から現在まで脈々と続けられてきた人類の創造の営みによって

What do you think? If this were your sentence, would you find the surrounding information superfluous in the first place? How would you deal with this beast? Do you think turning the surrounding text a different color indicate it isn't important will work?

Let's discuss.

I've taken the stance now to let one item be the focus of the sentence, and everything else is in support of that one thing. I've also gotten away from writing out the entire sentence (during production) except for the first time. Production reviews just have me focusing on that one thing

*this can be reversed for production*
Q: 「死ぬ」　「僕の犬は病気で死んでました」
A:　「しぬ」　「ぼくのいぬはびょうきでしんでました」

Show the word I need to know, show it in a sentence. I get the card wrong if I cannot pronounce the word or know what it means. The question side can show a photo relating to the sentence. The answer side can play the audio.

*here's a production sample*
Q:　「しぬ」　「ぼくのいぬはびょうきでしんでました」
A: 「死ぬ」　「僕の犬は病気で死んでました」

So here, I will just write out the one word and have to know what it means to get it right. The audio and photo can be played on the question side.

The benefit with this set-up is it's a stripping from the iKnow website. So it's a simple matter of getting it into Anki with minimal effort (sorry, I'm lazy so I like it streamlined). If I get something wrong in the rest of the sentence, no biggee. I chastise myself, but it doesn't effect if the card is wrong.

Grammar Sentences:

These are sentences that come from Grammar sources (All About Particles, Tae Kim's site, UBJG). For these, I make the following recommendation:

DO NOT DO PRODUCTIONS (well, maybe write them out the first time). Just do recognition. However, do not mark the card wrong because you got nouns and verbs wrong if they're just supporting the grammar concept.

*example from Tae Kim about compound sentences, causation*
Q: 時間がなかったからパーティーに行きませんでした。
A: Compound Sentence - Expressing reason or causation using 「から」 and 「ので」

I'll have the kana sentence, english translation showing in the answer also. But the point was I look at these and think about grammar. "There wasn't time in the past, which was the reason he did not go to the party, saying it politely" (for politeness or gender, I just picture a social situation in my head to justify the level).

Anyway, those are my two types of sentences: Vocabulary and Grammar.

I think it depends on what kind of context you're worried about cutting out. I think you should cut your sentences into whatever meaningful chunks you need. Pretty much as long as you have at least two words, you have some grammatical context at work, and by inputting those two words in your SRS you'll be acquiring that grammar. One idea behind SRS is to have repetitive input of language that you can understand that is in a [grammatical] context, albeit limited one. Even the longer sentences, unless uttered/written outside of any conversation/paragraph or larger situation/context are necessarily being cut away from their [informational/situational] context. So deciding how short to make a sentence to put in is pretty much up to you, because you will always being cutting out context, and cutting at the sentence mark is almost as arbitrary as any other cut. (Of course, you won't want to cut at parts that change meaning or affect grammatical accuracy.)

The technique you showed is a pretty neat way to avoid cutting away the whole sentence. It's nice to highlight what you're focusing on. But I don't think it's necessary. If you're focusing on that section but leave the lighter section which has clues, you might end up always relying on the lighter section to help remember. Doing it in white would help avoid this, like you mentioned.

Another idea behind the sentences/SRS is that you're reading/hearing 50-100 (or more) sentences for every sentence that goes in. Included with this idea is that putting sentences in instead of single words allows your brain to acquire vocabulary and grammar at the same time. But if you have really long sentences, they're bound to include a lot of new/shaky/semi-acquired stuff. I think shortening them will let your brain focus better while still have a grammatical context. You may be losing the informational context of the whole sentence, but you're also losing the context of the paragraph/chapter/conversation/episode/book/series/etc. That informational context isn't as important in the SRS; it's most important in the living in Japanese for 10,000 hours.

samesong Wrote:My problem is that the whole sentence just keeps going and going. The word I'm concentrating on is 脈々. So I could possibly do this:

I cut it wherever I like and have a "context" field in Anki. I put as much context as I want there along with any explanations for what happened or where I got it from. Example:

Sentence: 頭を噛み砕いて

Reading: あたまをかみくだいて

Context: 大和 is stalking 朝比奈 and is at her house in wherever. She's not there, but the dog is.

Notes: くだ・く【砕く・▼摧く】
(動カ五[四])① 固まっているものを，打撃や圧力を加えて細かい破片にする。かたまりを細かくする。「氷を－・く」「土のかたまりを鍬(くわ)で－・く」

In this case I cut it off after the いて, but I didn't feel the need to include the rest in my context field. Often I will, or give the previous/next sentence in the context field.

Isn't using english context against the idea of using japanese deffinitions?

Nukemarine Wrote:*here's a production sample*
Q:　　「ぼくのいぬはびょうきでしんでました」
A: 「死ぬ」　

So here, I will just write out the one word and have to know what it means to get it right. [...] The benefit with this set-up is it's a stripping from the iKnow website. So it's a simple matter of getting it into Anki with minimal effort [...]

Consider leaving the production Q in kanji (other than the target vocab). This way, you practice reading the normal sentence. (I find all kana sentences hard to read.) It would require a bit more fiddling with your deck, but the active process (of copying the kanji sentence into a new production sentence field and changing the target vocab into hiragana) will probably help you memorize stuff from a pre-made deck.

Q: 「しぬ」
「僕の犬は病気でしんでました」
A: 「死ぬ」　「ぼくのいぬはびょうきでしんでました」

I put the hiragana prompt in large font and first test if I understand the word. Then I read the small font sentence to get the context and hopefully the meaning (if I wasn't sure). Basically, it tests vocab meaning, kanji writing and sentence reading all in one card.

edit: added A

Thora Wrote:Consider leaving the production Q in kanji (other than the target vocab). This way, you practice reading the normal sentence. (I find all kana sentences hard to read.)

Q: 「しぬ」
「僕の犬は病気でしんでました」
A: 「死ぬ」　「ぼくのいぬはびょうきでしんでました」

This is exactly how I do it. Everything is in it's normal (Kanji) form except for the vocab word in question. I don't even bother bolding the hiragana word.

I found that long sentences slow me down, but chopped-up sentences sound unnatural. When I take a long sentence, I take it as a complete thought, so it makes sense as a whole linguistic unit. In order to keep from flunking it too often, I'll add extra, short cards to reinforce the vocab that shows up in the bigger sentence. So the idea is that if I have:

Q:市役所で介護の講習会があったので、出席した。とても役に立った。
A:しやくしょでかいごのこうしゅうかいがあったので、しゅっせきした。とてもやくにたった。

(Yes, I know it's 2 sentences.)

I would add extra cards for vocab like 市役所, 介護, 講習会, 出席, and 役に立つ, if I was feeling uneasy about any/all of them. The more cards I add, the more easily the bigger sentence goes into my memory.

Fortunately, I don't run into sentences that are this unwieldy very often, because when I do, I can add as many as 10 cards... *but*, those cards oftentimes will lead to other, related vocab, so I don't worry about it too much. It's all Japanese.

That said, I use short sentences for the most part. The exception is Kanji Odyssey, which has some long, oddball sentences like the one above, but I love the vocab order the book uses, so I tough it out. And when I have to add extra cards, I use short sentences from online dictionaries.

The sentences come from "safe" sources for the most part. If I don't know the source, I don't use it. I don't write my own, and I don't change the ones I find. I don't use EDICT, I don't use Tanaka.

Rather than torture a sentence to fit it into my deck, I'll just find another one. It's not that hard to do.

Production is key for me. Production improves my retention by an order of magnitude. Production is also a Royal Pain in the Butt, but it's that important, so I keep it. I've streamlined it down to the vocab I'm targeting, and I use the finger-in-the-air method. No paper. It looks weird, but it's fast, I don't get cramps, and the environment is saved.

I *do* use all kana sentences for production. You get used to it over time. It's not particularly natural, but it's good kana reading practice.

When it comes to grammar sentences, I have production cards, but since I only produce kanji when they're odd/uncommon, I tend to breeze through these, since most grammar sentences use really easy vocab. The main reason I keep them in the deck is so that I see the grammar examples twice as many times as I'd see them otherwise.

It's not really ideal, though.

I'm probably going to try a new approach for grammar, and that is going to be fill-in-the blank. So there will be a recognition card, and a FITB card. Should be handy for remembering just what the heck I'm trying to learn. I've got a few books on particles that I've been itching to attack in such a manner.

I'm debating between doing it like this:
Q:二年前の戦争_大勢の人々が死んだ。
A:二年前の戦争で大勢の人々が死んだ。

Or like this:
Q:二年前の戦争(で/から)大勢の人々が死んだ。
A:二年前の戦争で大勢の人々が死んだ。

Maybe I'll use both. Dunno.

I really like that idea for longer sentences. I think it might be better than what I've been doing, which is this (left):



The snipped sentence on the front and then the full one on the back. Also something I've been doing for a while with manga is putting little stickers on the words I don't know, and then later going through a downloaded scan with SnagIt to grab those frames. I like the extra context it gives but mostly I just like the way it saves me from my own typos.



I've been experimenting with a small number of gap fills. These are mostly for only kana words, because I've realized that I often don't recognize them when I see them outside my deck. I must have added しっかり like 6 times before I realized they were all the same word (with slightly different meanings but still). I'm keeping the number of these low though. Even though I have a bunch of different kinds for different purposes (I also have some dictation ones with iKnow sentences), I'd say like 90% of the 3497 cards in my deck are still simple sentence ones like this: