Back

Anki efficiency - readings vs meaning

#1
Something that anotherjohn posted in the milestone thread called me attention:

(2016-08-01, 10:00 am)anotherjohn Wrote:
(2016-08-01, 9:39 am)yukamina Wrote: @anotherjohn, are those 53000 cards unique words?
Nooooooooooooooo!

Maybe 40k words/names -> reading, 3.5k kanji -> meaning, plus overlapping audio/sentences.

Also I review words for reading only, not meaning (though many words to require at least a vague impression of the meaning in order to get the right reading).

I find that once the reading is familiar, the meaning tends to follow from the kanji meanings + contextual cues.
Now, I know most people focus on both readings and meaning when doing reviews. But is it really that efficient to focus on both things and fail cards if you don't get the meaning quite right?

I'm not saying you should ignore the meaning of a word altogether. But being able to immediately process the meaning of a word when you hear it or see it oftentimes take a certain amount of meaningful input in my experience. Or better yet, once you already know how to read the word properly - which is a big deal -, sometimes just a little bit of meaningful exposure is all it takes.

For instance, I've been reviewing the card 日中 for a while and it's not too hard to remember how to read it. After all, readings are as objective as basic math. 1+1=2, end of story. There's no nuance to interpret, you either recall the way a word is pronunced or you don't.

But it was only a couple of days ago when I read it on NHK that my brain made a proper connection between the spelling/pronunciation 日中 and the actual meaning. Despite the fact that I'd recall how to pronounce, more often than not I'd think it was something along the lines of "for the entire day" and end failing the card. But now I ask myself, if focusing on internalizing the meaning of the word in context can be so efficient, should I really bother if I don't get it quite right in my Anki reviews? Wouldn't it be enough to pass or fail it based on readings alone?

My reviews aren't the fastest. Some people seem to be able to plow through words in a matter of seconds, but I usually take a few seconds to recall both the reading and meaning of a single word. Focusing exclusively on the meaning would speed up the process, which means I get to spend more meaningful time with the language internalizing words more efficiently.

Now, that doesn't mean this approach wouldn't help me recall the meaning of a word. The difference is that the meaning would be a reminder to slowly get the word in my head. Rather than drilling a somewhat imprecise definitions over and over, they would be used a "bonus input" to help me not forget these words altogether. It's like taking a glance at your notes to recall what you've studied reguarly, but accepting that what some of the words may need is a little more meaningful input rather than hours on end drilling them.

I'm curious to know if other people have tried anything like this. It seems like a valid approach to me. I don't think some balance would be bad though. Perhaps choosing hard if you feel you need more exposure to the word, or even failing a word if it looks so alien that you can't recall ever adding it?

So what do you guys think? Have you ever felt that failing cards when you managed to recall their readings was almost like a diversion? In that it was perhaps not a very efficient way to absorb the overall meaning of that word? Do you ever feel seeing/hearing the word a few times in context is enough to "get it" once you already know how to read it with ease? I'd love to heard your experiences with Anki and vocabulary acquisition.
Reply
#2
I actually have the opposite effect: I can recall the meaning but for some reason the readings I jumble up. Especially with words containing and . This is the Core 2k/6k deck.

上[のぼ]り、年上[としうえ]、頂上[ちょうじょう]and 下[お]りる、下[さ]がる、下手[へた]、年下[としした] are just some examples. And to answer your question: yes, I fail if I either don't know the reading or meaning. I will only press "Hard" if my reading was really close. If I read and recall the meaning correctly, it is good. If I do so instantaneously, it is "Easy".

With these "Core problems" as I like to call them, I might consider making a separate deck and use the MCBs method-One Character per Card setup. It has been successful for me in the past when I was learning vocab for my Japanese classes.
Reply
#3
HdI have separate decks for separate formats since certain words or sentences work better with certain formats. I have one for readings where I only focus on the reading so I put the kanji and meaning on the front. There's a deck where I do Mcd which focuses on the word and the meaning for example I have multiple cards for 大船に乗る where I blank out part of  the phrase and part of the meaning . Another example is 疑心暗鬼 where I made a few cards with different parts of ぎしんあんぎ blanked out so I can remeber the word itself because me doing a card with the kanji in the front and the reading in the back is pointless becaus all im doing is reading the readings of kanji im very familiar with
Edited: 2016-08-01, 11:20 pm
Reply
Thanksgiving Sale: 30% OFF Basic, Premium & Premium PLUS Subscriptions! (Nov 13 - 22)
JapanesePod101
#4
Well, the way I study most words is to add two cards for each fact (or note or whatever we're calling it now). I always have a context sentence on the front, and the first version of the card has the kana version of the word on the front, while the second version of the card has the kanji version (and requires a typed answer).

The kana cards I have to get the meaning exactly right or it's a fail -- that's the whole point. And with front-side context, there's really no excuse. (It's still best to choose context sentences that don't give away the meaning of the word of course, so shorter simpler sentences are better.) For the kanji version, I have to get the reading exactly right or it's a fail ; if I -totally- flub the meaning on the kanji card, I'll still fail it even with a correct reading but I'm pretty liberal... as long as I have -some- idea what it means I'll let 'near misses' slide.

This way I'm somewhat separating memorizing 'meaning' and memorizing 'reading' into separate tasks. Plus, of course, memorizing the phonetic spelling of words makes it easier to recognize them when you hear them or encounter them in kana or (heavens forfend) romaji.

You could split this up all kinds of ways, giving the translation on the front when strictly testing for reading, creating another kind of card where you give the reading and the meaning and require producing the kanji, etc, etc., but the way I'm doing it now works fine for me.
Reply
#5
My cards are Jalup style: Sentence on the front, dictionary definition of the new word and tts-audio on the back.
I read the sentence and if I understand it and get the readings right it's a pass. I don't translate anything.

Even if I didn't grasp the whole meaning of the word with just this sentence, it's not really a problem. I will see the word again and again when reading or watching TV, so the concept I have of that word is always evolving in my head.

I wouldn't go so far as to only study readings and no meaning at all at the moment. But when your vocab is already quite big and new words mostly consist of known Kanji anyway, then I believe this approach might work.
Reply
#6
I test for reading and meaning, failing if I can't recall one or the other. However, I hardly ever fail for meaning because probably >80% of my cards I am using the meaning to recall the reading. In other words the kanji usually gives away the meaning or gives a strong hint that helps me remember the meaning. So if I can't remember the meaning, I can't remember the reading either. I guess that is only for words with kanji in them, but you don't need to test readings for words normally written in kanji so.

So I don't think it's a bad idea to experiment with testing for reading only and see how it goes. Or you could move the cards without a strong hint from the kanji (and of course the kana only words) into another deck where you test for meaning.
Reply
#7
I test for reading and meaning, and my failures tend to be an even mix of the two, but for any particular card it's generally only one of the two I have trouble with, so I don't feel in practice like it's trying to learn two things with one card. Reading failures are usually things like forgetting whether 火山 is かざん or かさん, or wrong vowel-length. Meaning failures are on gitaigo or where the word isn't easily guessable from its constituent kanji (I don't put sentences on the card front or I'd fail meaning much less often). And I really wouldn't want to have to double the number of cards I'm dealing with :-)
Reply
#8
Same as pm215, with one exception, I'll put the example sentence on the front, a bit below the word, so I can't see the sentence too easily, and if I can only figure out the **meaning** from the example sentence (and get the reading right), I'll mark it as 2 instead of 3. To get a 3, I have to get meaning/reading right just from the word cue within a couple of seconds. To get a 4, the recognition has to be immediate. If the reading or the meaning is wrong, it's a fail.

I'm pretty much using that Core 10k v23 deck that was posted here a while back, but I've been modifying about half of the images in it to suit me, and I didn't change the layout all that much.

One thing I'm doing now is that I'm using the notes field to add stuff like "this 上る is for Tokyo-bound trains" or "this is the 'formal' version of 去年" so I don't waste time flunking cards where the kanji and/or the okurigana are the same as another card's, but either the reading and/or the meaning are different.
Edited: 2016-08-03, 12:27 pm
Reply
#9
(2016-08-03, 12:26 pm)rich_f Wrote: One thing I'm doing now is that I'm using the notes field to add stuff like "this 上る is for Tokyo-bound trains" or "this is the 'formal' version of 去年" so I don't waste time flunking cards  where the kanji and/or the okurigana are the same as another card's, but either the reading and/or the meaning are different.

That's a good idea. Do you put that info on the front? It seems like it would be too much of a hint having it on the front, but wouldn't be of much help having it on the back. For words written identically with multiple readings and/or meanings I've been recalling both readings/meanings.
Reply
#10
I put that kind of disambiguation on the front (there's a hint field in the card template I'm using). I did a pass through using Anki's find-duplicates feature to fish out the cards with duplicate fronts so I could add hints (or just suspend the cards in some cases). I often end up making the hint be "not <some other thing>" because usually there's two options and I naturally remember the other one better.
Reply
#11
(2016-08-03, 5:46 pm)pm215 Wrote: I put that kind of disambiguation on the front (there's a hint field in the card template I'm using). I did a pass through using Anki's find-duplicates feature to fish out the cards with duplicate fronts so I could add hints (or just suspend the cards in some cases). I often end up making the hint be "not <some other thing>" because usually there's two options and I naturally remember the other one better.

That's a good idea.  I might try something similar.  Thanks!
Reply
#12
I wonder how long you guys take per vocabulary card. I was having some issues with deck corruption a couple of weeks ago, so I decided to make a new account last month (still able to review my old cards on ankiweb just fine though) and my average with this new vocabulary deck is 6.6 seconds per card. That's a lot of time imo, but I can't seem to review stuff any faster.

I don't think having a deck just for meaning would be a good idea at all. Of course, some words aren't written in kanji, and in that case it's probably a better idea to take meaning into consideration as well. But these are exceptions.

The point of this approach would be to save time by focusing more on the more objective aspect of a word (readings). I don't mind picking hard for a card that I don't know very well or failing a card that seems completely alien to me, even if I get the meaning right.

I'm just not sure reviewing the same cards a whole bunch of times outweighs the benefits of having more natural exposure of the language + reviewing even more cards. Anki is a great approach when something needs repetition and you want to drill it. Natural exposure is a great way to internalize the meaning of words. In other words:

(1) Every time I read, I have to add new cards. To read a lot, I need to add a LOT of cards.
(2) More leniency lowers the total amount of cards to review. This allows me to read more without being overwhelmed by the amount of new cards.
(3) Reading more helps me nail the meaning some of the vocabulary that otherwise I would end up just reviewing in Anki.
(4) Some words will take a little longer to find. But I'll still see the meaning of these words in Anki sooner or later. It gives me a little more exposure to these words (akin to flipping through the pages of a vocabulary notebook, but more efficient since you can pick "hard"), which is better than no exposure. I just won't be failing a whole bunch of cards, just like I don't open Anki to fail cards when I don't recall them in the wild, or fail a card in Anki just because I saw it in the wild in the previous day (which is like cheating because it's as if you'd failed the card and then went back on your decision to fail it).

Does this train of thought make any sense to you guys? Or do you think there's a better way to go about doing this and achieving better or similar results?
Edited: 2016-08-03, 8:46 pm
Reply
#13
Some others might have more experience, but I personally think it's best to try to pronounce the item immediately, and then think of its meaning. I do them in that order and don't have any problem doing both with the same card.

Ideally the pronounced Japanese word will "mean" the right thing to me, in which case I don't worry about trying to recall the exact English definition. If the English definition more or less matches the meaning that came up in my head, I count it as a correctly answered card.
Reply
#14
It's worth giving a try. Since you are doing a lot of reading, you will get some practice decoding the meaning of words so I don't think it's a bad idea. Making more time for reading should make that even more true. I think it's really dependent on the individual whether the extra reading makes up for the lack of practice in anki or not, so you have to try it to see if it works for you. It's certainly not a bad idea.
Edited: 2016-08-03, 10:55 pm
Reply
#15
Try it out and report back how it goes for you? It can be a challenge to balance reading/immersion vs study time. Personally I don't think 6 seconds a card is terrible (I do sentence cards, so my average is even longer). You can go for quantity over quality, but your comprehension could suffer for it. For example, I find the words I studied with Memrise (isolated vocabulary, multiple choice instead of active recall) don't stick in my head as well as words I learned with sentences in Anki. The sentence cards take longer, but I remember them much better.
Reply
#16
6.6 seconds isn't terrible. I think I average around 5 seconds per card, more on bad days, less on good days. Ideally, I want to get it down to around 3 seconds, but for some cards, it takes a little longer, or I have that "WTF is this?" moment that bogs me down a bit. Or sometimes I have to read the sentence for the meaning when my brain freezes.

I usually try to sound it out in Japanese first, too, so if I can't even do that, I know I'm going to flunk it, but II try to remember the meaning too, because sometimes I can figure out the meaning even if I can't say it.

I really try to keep it all in Japanese in my head, but that's hard to do with English definitions on the cards and with translations for the sentences. I don't really get too worked up over that, now that I think about it, but generally, I try to avoid the English as much as possible. It's all relative depending on where you are in your Japanese study.

Did anybody ever come out with a decent speed reading training app for Japanese (for Android?) That would be awesome.
Reply
#17
Ignoring meaning just seems kind of weird to me because it feels like that's the major aspect of a word. I wouldn't say I knew a word if I happened to be able to pronounce it but didn't know the meaning. Also for a lot of words: (1) you'll encounter them in reading and not in conversation (2) the reading will be fairly easily guessable from the kanji (3) if you don't actually know the meaning you're stuck (whereas not knowing the reading doesn't prevent you from understanding the sentence and moving on). I'd sooner be lenient on not knowing the reading than drop testing for meaning.

Quote:(1) Every time I read, I have to add new cards. To read a lot, I need to add a LOT of cards.
You don't *have* to add cards for words you don't recognise in your reading. I certainly don't and I feel like it would eat an awful lot of time creating cards if I did.
Reply
#18
(2016-08-04, 11:20 am)rich_f Wrote: 6.6 seconds isn't terrible. I think I average around 5 seconds per card, more on bad days, less on good days. Ideally, I want to get it down to around 3 seconds, but for some cards, it takes a little longer, or I have that "WTF is this?" moment that bogs me down a bit. Or sometimes I have to read the sentence for the meaning when my brain freezes.

I usually try to sound it out in Japanese first, too, so if I can't even do that, I know I'm going to flunk it, but II try to remember the meaning too, because sometimes I can figure out the meaning even if I can't say it.

I really try to keep it all in Japanese in my head, but that's hard to do with English definitions on the cards and with translations for the sentences. I don't really get too worked up over that, now that I think about it, but generally, I try to avoid the English as much as possible. It's all relative depending on where you are in your Japanese study.

Did anybody ever come out with a decent speed reading training app for Japanese (for Android?) That would be awesome.

I just happened to have for a pretty decent app actually Smile It's called MONDO. It's for iOS too! Posted in the Essential Resources forum along with another reading speed resource.
Reply
#19
(2016-08-04, 3:21 pm)pm215 Wrote: Ignoring meaning just seems kind of weird to me because it feels like that's the major aspect of a word. I wouldn't say I knew a word if I happened to be able to pronounce it but didn't know the meaning. Also for a lot of words: (1) you'll encounter them in reading and not in conversation (2) the reading will be fairly easily guessable from the kanji (3) if you don't actually know the meaning you're stuck (whereas not knowing the reading doesn't prevent you from understanding the sentence and moving on). I'd sooner be lenient on not knowing the reading than drop testing for meaning.

Quote:(1) Every time I read, I have to add new cards. To read a lot, I need to add a LOT of cards.
You don't *have* to add cards for words you don't recognise in your reading. I certainly don't and I feel like it would eat an awful lot of time creating cards if I did.
I see your point. It's not like being lenient and passively reviewing the meaning of a word is the same as from ignoring it though. Whenever you read anything, you're passively reviewing vocabulary without testing it. Unless you literally open up Anki to fail flashcards, you can't really "fail" words you see in the wild - you just look them up if you don't recall the reading or meaning.

And yeah, I know that not knowing the reading doesn't prevent readers from moving on. But I'm the kind of person who won't be willing to move on without knowing the reading until become relatively fluent, unless I can't find it anywhere. Names in Japanese really annoy me because I suddenly can't read the whole sentence. In fact, I made a deck with the name of the 173 most important cities/prefectures + tokyo wards just to be able to read the news a little more comfortably. Sooner or later, I'll be adding the name of all remaining cities. I know there'll be a lot of places and names I won't know, but a significant part of the issue will be dealt with.

As for not having to add words I don't recognize, do you mean words I don't recall seeing more than once? That seems like a smart approach if you want to read more. I'm afraid that would be a little difficult for a guy like me though. I like to add almost everything that I learn and manage to understand in context. But maybe that'll be fine if I review the uncommon words more passively as I mentioned above. That doesn't mean I can't fail words that I consider more important. Does that make more sense to you?
Edited: 2016-08-06, 2:17 pm
Reply
#20
I just meant that if you want you can read a book, look up words where you need to, or not if you can guess from context, and not bother to put anything into anki at all. So you can read a lot without worrying about it meaning your anki workload is going to go up. The only anki reviews I do is working through core10k (which then shows up as words I'd probably have skipped over in reading turning into ones I recognise). Personally I'd find reading a lot less enjoyable if it was tied to the chore which is anki reviews. But what works for me may be different from what works for you.
Reply
#21
I'd probably be like that if we were talking about languages more similar to Portuguese and English. No wonder I hardly ever review words in katakana in Anki. But everything else in Japanese is entirely different from the languages I know.

I think I've studied about 6600 words so far, but I still need to look up quite a bit of stuff depending on what I read, which is a pain. My hope is that if ramp up my vocabulary acquisition now, reading will become more manageable in the near future. But that's something you can't do with Anki alone, hence why I'm trying to find the middle ground between both approaches.
Edited: 2016-08-06, 3:06 pm
Reply