Learn Japanese with JapanesePod101.com

View topic - Learn Japanese with Anki

Learn Japanese with Anki

Have a textbook or grammar book that you find particularly helpful? What about a learning tip to share with others?

RE: New Anki (0.2.6 + new Mac build)

Postby resolve » Mon 05.21.2007 10:28 am

Hi Yufina,

You have entered a username and password to keep a copy of your deck sychronized with the server. If you opened one of the sample decks and then closed Anki, it will have automatically synced those cards with your server deck. If you don't want that to happen, don't use 'sync on open' and 'sync on close' in the Anki preferences.

Ezrach: the stats are all HTML already. You can copy and paste them into a rich text aware editor and they should copy verbatim. I plan in the future to add a customizable 'full report' that can be saved as a HTML file.

Cheers,
resolve
 
Posts: 135
Joined: Thu 09.14.2006 12:41 pm

RE: New Anki (0.2.6 + new Mac build)

Postby Chrisou » Mon 05.21.2007 7:08 pm

resolve wrote:
Hi Yufina,

[snip]
How is Stackz partially better? I've had a brief look at in past, but it didn't look too impressive. This is what I wrote about it:

resolve wrote:
I had a quick look at the stacks website. They seem to use a varient of the leitner system which classifies cards into 5 "levels". The vast majority of flashcard programs out there use a leitner based system as it's simple to implement.

Stacks uses coloured graphs to indicate the age of cards, and like most flashcard programs out there, puts the responsibility of when to review on you - you can choose to drill a set of cards over and over, and when to review the cards. I think is a really bad idea. For one, people have a tendency to overly review cards they have difficulty with. It doesn't matter how many times you practice a card in a small amount of time - the important part is spaced repetition. You're much better off getting a card correct once and trying again the next day than practicing one more time.
[snip]

Damien




Dear Resolve,

I like Anki, it looks nice and is probably useful on the long run.

Concerning Stackz, maybe your look was a bit too brief...

Stackz has a mode called "RepeatAdvisor", where the cards that are recommended for repetition are visually highlighted in orange color. The key point there is that the difficult ones are highlighted earlier than the easy ones. This algorithm is customizable, the "difficulty estimation" is either based on the promotion level (i.e. the box of the card), or the success ratio of the previous tests (i.e. the percentage of correct tests).

You can say that the cards in the best box should be refreshed after 3 weeks, the card in the second best one after 2 weeks, etc, just as you like. Alternatively, you can say that the cards with a success ratio of 20% or better should be repeated after 3 weeks, the ones between 20 and 40% after 2 weeks etc...

The color makes the distinction, just keep the cards out of the orange zone. Very simple, very effective.

http://www.stackz.com/KooHelp/Stackz/DataOrganization/ColorModes.htm
Last edited by Chrisou on Mon 05.21.2007 8:14 pm, edited 1 time in total.
Chrisou
 
Posts: 5
Joined: Mon 05.21.2007 6:51 pm

RE: New Anki (0.2.6 + new Mac build)

Postby resolve » Tue 05.22.2007 12:00 am

Hi Chrisou,

Thanks for your comment.

When I looked into Stackz last time, I did have a look at that page. I think it demonstrates two fundamental problems that perhaps I didn't explain clearly enough last time.

You said that it's "very simple" - but I think it's too simple. For example, imagine one of the boxes is set to review cards after a month. If you add one card to that box on a day, and then another card to that box 25 days later, what happens? If the review time is set from the first addition of the card, you're repeating the second card much too early to be effective. If the review time is set from the last card, the other cards have to wait too long for review and you risk forgetting them. Box-based scheduling is not fine-grained enough to be really effective.

The colours are a gimmick. By turning orange, the program is telling you "you should review this box now". If the program knows I should review, why doesn't it just show me these cards? Also, by giving the user the ability to easily review cards they've practiced recently, you risk the user wasting time, and multiple repetitions in a short period are not the best way to learn (see
http://en.wikipedia.org/wiki/Spaced_repetition)

Anyway, if you're enjoying Stackz then that's great - any flashcard program is better than no flashcard program. But I think that box-based flashcard programs (the "Leitner" system) are fundamentally less effective than flashcard programs which schedule based on each card.

Cheers,

Damien
resolve
 
Posts: 135
Joined: Thu 09.14.2006 12:41 pm

RE: New Anki (0.2.6 + new Mac build)

Postby Chrisou » Tue 05.22.2007 4:03 am

Sorry, I was not successful in explaining it correctly...

If you look at the screenshot again, you see that every CARD has its own color, not the box.

To me it looks like good GUI design if the user can control the action, not the program. You are not every day in the same mood, and you don't have every day the same amount of time. There will be learning breaks once the course is over. First shorter ones, then longer ones. If you want to pick up your training again at any moment, you need to know where you left off.

Somehow the system is just too good to explain, one has to experience it. I think the website needs better explanations!
Chrisou
 
Posts: 5
Joined: Mon 05.21.2007 6:51 pm

RE: New Anki (0.2.6 + new Mac build)

Postby resolve » Tue 05.22.2007 5:26 am

Okay, I spent some time looking at the docs more carefully. But I don't think my original observations were wrong :-)

Each card is represented by a line in the box graph, and the lines are colour coded, so that older cards can be distinguished from newer cards, right? So the idea is to decide to review a box when a large enough percentage of the cards become orange.

But this is exactly what I was saying before - if you wait too long you risk forgetting some of the cards, if you don't wait long enough you waste your time by reviewing too much. Yes, it's possible to review half the cards in the box and then stop and review again later, but that requires you to micromanage your deck, and if you aim to do a card exactly 4 weeks from when it was scheduled, you'll be forever clicking around different boxes.

I don't understand your arguments about mood or time, either. Are you suggesting the program should let users practice things they don't need to practice, because they're not in the "same mood"? I'm all for programs being flexible, but I think people review because they feel like they need to - even if we're not the best judge of when we will forget something again. People tend to be more conservative than they need to be, and it leads to us practicing more than is necessary. One of the design goals of Anki is to help you learn as much as possible with the minimum amount of effort. If we let users review stuff that doesn't need to be reviewed (because the algorithm knows the chances of forgetting are quite low), then we are effectively letting the user waste their time, when the time could be better spent learning new material, or just having fun. Also, some studies indicate that waiting longer actually strengthens your memory more than waiting for less time, so in some cases studying more will make your memory less strong than if you'd just waited!

Your comments about taking a break indicate that you haven't actually used Anki before. Scheduling is fine-grained so that a card is scheduled at an exact time in the future, but the program doesn't force you to do it at that time. Delays and vacations are a natural part of life and Anki doesn't punish you for that. In stackz if you take a break, you'll come back to a deck with lots of orange everywhere, and you need to click on each box in turn, and study the orange cards (while keeping an eye on the graphs to make sure you're not studying the green cards you just added!). With Anki, you just open the program and it will ask you each card which is overdue in turn. You don't need to scan a bunch of graphs to see which cards are a particular colour. And as a bonus, it notices that you've answered a card later than you should have, and if you still remember, it will factor that into the next scheduling, putting the card further in the future next time.

Oh, and Anki is free, and always will be. :-)
Last edited by resolve on Tue 05.22.2007 5:28 am, edited 1 time in total.
resolve
 
Posts: 135
Joined: Thu 09.14.2006 12:41 pm

RE: New Anki (0.2.6 + new Mac build)

Postby IanLewis » Tue 05.22.2007 6:18 am

An edited version of our IRC conversation.

Me:
I think you misunderstood how stackz functions. Granule functions the same way. Each card has a next review time so you don't review cards too early or too soon. So if you review box 5, say, you only review the cards in box five who are up for review. It's still different than Anki in that you are inserting cards into boxes so there are only 5 time intervals you can attach to cards rather than using a formula based on previous reviews of the card.

Basically if you review a box, and the cards you get right go into box 4, then every card you put in box 4 during that session will be reviewed x days from now, where x is the setting attached to box 4.

The only difference between it an granule from the looks of it is that it attaches a color to the card. (And it looks ugly).

resolve:
right, so it ensures a minimum amount of waiting?

Me:
well, I think you can technically review the same cards over and over. But it keeps track of when each card should be reviewed.

resolve:
even then, you have to click around to choose the box to review

Me:
yes, you have to choose the box to review.

It would be easy to implement something that allowed you to review all boxes at once but you're basically heading towards what Anki already is anyway. I tend to agree but as I said earlier it's easy to get caught up micromanaging your study rather than studying. Even anki has some distractions like graphs and stats. ;) Though they tend to be more useful than the micromanaging features of granule and stackz.

resolve:
The graphs and stats aren't part of the process, though. I think they serve to motivate. The graphs in stackz are part of the process

Me:
But if it's 85% motivation it's 10-15% distraction. It's a balance.
IanLewis
 
Posts: 1
Joined: Tue 05.22.2007 6:10 am

RE: New Anki (0.2.6 + new Mac build)

Postby Chrisou » Tue 05.22.2007 8:09 am

I think your current learning mood is really different from time to time... at least my brain does not work like a computer :-)

1) If I work on an entire lesson, as for the tests in my intensive spanish course, I select all boxes of that lesson and Stackz works as if no boxes were there, it presents the hard ones first regardless of their box (unless the box is used for difficulty estimation of the word of course).

2) If I want to improve my hard words, I select the colums on the left from various lessons and learn them somehow - not only with the computer, but in any other situation after printing them out. The Pocket PC version is a good compromise there because it's mobile... I don't want to learn in front of the PC.

3) And if I simply want to check my knowledge, I select the columns on the right and test the content. There I don't want to get frustrated with the many ones I don't know (yet), I just want to play with the content and update my memory... I think this is vital in Japanese more than in other languages, kanji just need to be refreshed if you don't live in a Japanese environment.

So yes, I think I am a bit flexible in my learning habits :-)
Chrisou
 
Posts: 5
Joined: Mon 05.21.2007 6:51 pm

RE: New Anki (0.2.6 + new Mac build)

Postby resolve » Tue 05.22.2007 11:16 am

I don't think it has anything to do with moods. For learning, there are all
sorts of things you can do, such as watching TV, reading a textbook, etc. The
time and order you do the things doesn't matter. You can aquire knowledge at
your own pace.

But flashcards are ultimately designed to help you remember things, not
learn them. How do you remember something? By practicing before you forget.
When's the best time to practice? Just before you forget. Our memories
follow a "forgetting curve" and there is an optimal time to review.
This
is a fundamental point of memory, and I believe that Stackz doesn't do a good
job of catering to it.

From the website:

The Stackz system does not take decisions for you, nor are actions
performed automatically. It rather makes the current situation transparent and
lets you to decide what you want to do about it.


Basically Stackz puts the responsibility of deciding when to review on you,
and gives you some pretty graphs so you can do it "transparently". I think
these graphs are not at all sufficient to achieve maximum recall in the
minimum amount of time.

You said "sometimes I want to check my knowledge". Why? Don't you trust the
time which Stackz has given you? You said "I don't get frustrated with words I
don't know". It's easy, right? That's because you're reviewing stuff you
already know, and don't need to review
. You might as well be doing
something else, like watching TV if your "current learning mood" doesn't suit.

That situation also illustrates a fundamental problem with Stackz. Stackz
is not designed for long term memory
. You said "Kanji need to be
refreshed". If you have to do that manually, then your flashcard program is
not doing what it should be doing! What it should be doing is showing you the
cards just enough to keep the memory strong, while not requiring you to feel
the need to "refresh" every month.

Part of the problem is that Stackz only has 5 boxes, which is completely
inadequate. We don't "master" a word after a month. We don't master many words
at all - almost anything we don't use or see enough we will forget eventually
- even in our native tongue. The whole concept of a "mastered" box is silly,
and to have it default to a time interval of something like a month is even
sillier. And having a visual indication that cards are a month old wastes a lot
of your time - cards certainly don't need to be refreshed every month once
you've practiced them enough.

There are some good things to be said for Stackz - the fact that there's a
pocket PC version is nice, although for that I'd strongly recommend
"supermemo" instead - its pocket interface is apparently a lot easier than the
clunky PC interface it has. Also Stackz is more mature than Anki. But
basically I think Stackz gives you a false sense of accomplishment. It
declares words "mastered" when they're in fact not, and then encourages
reviewing far more often than is necessary - and if you choose to ignore the
"transparent" interface you risk wasting your time. The division into boxes
and lessons requires you do all the grunt work of deciding what to study next.
And the scheduling system is not well thought out. If you insist on paying
money for a flashcard program, consider spending money on the much better
(though difficult to use) Supermemo, which has decades of research which
proves its effectiveness. Or use Anki, it's free. :-)
resolve
 
Posts: 135
Joined: Thu 09.14.2006 12:41 pm

RE: New Anki (0.2.6 + new Mac build)

Postby Chrisou » Tue 05.22.2007 11:39 am

Well... hmmm yes I can see that you are determined about your opinion. Good for you, I hope that you progress with this method.

My only intention here was to correct your erroneous informations you gave about the Stackz System, repeated times.

I don't intend to turn this into some sort of religious discussion ;)
Chrisou
 
Posts: 5
Joined: Mon 05.21.2007 6:51 pm

RE: New Anki (0.2.6 + new Mac build)

Postby resolve » Tue 05.22.2007 12:10 pm

Perhaps I came on too strong. I'm very interested in learning effectively as I both learn Japanese and teach English, so every day I'm exposed to remembering and forgetting. My interest has prompted me to read a bit of research on the matter, and after spending some time on a supermemo like flashcard program, I was so impressed with the results (and the logic behind them) that I've invested hundreds of hours since, writing a free application for everyone to use so that they, too, might be able to benefit from effective memory training. If the other flashcard programs out there were good enough I would just be using them instead and wouldn't be talking about this now.

You say that you didn't want a religious discussion, but basically this conversation has consisted of you saying "I like stackz, and use it like this", and me talking about why I don't think that's the most effective way to remember. When I point out the reasons why I think Stackz is not as good, you label them "erroneous" and then retreat from the conversation because it's becoming "religious".
Last edited by resolve on Tue 05.22.2007 12:18 pm, edited 1 time in total.
resolve
 
Posts: 135
Joined: Thu 09.14.2006 12:41 pm

RE: New Anki (0.2.6 + new Mac build)

Postby Chrisou » Tue 05.22.2007 12:47 pm

...and I wanted to stop, hehehe.

I can see that you put a lot of effort in your program, that's great, and I wish you good luck! Creating software can be a lot of fun, I'm developing Software myself, I know what I'm talking about.

IMHO Stackz is based on the very same idea that you promote, it is just transparent in the sense that it lets you know with colors how things really are. This has many consequences, some may be good, some may be bad. If you had to find 10 positive and 10 negative consequences, you would find them with a neutral viewpoint, I'm sure.
Chrisou
 
Posts: 5
Joined: Mon 05.21.2007 6:51 pm

RE: New Anki (0.2.6 + new Mac build)

Postby debu » Tue 05.22.2007 4:13 pm

Resolve wrote:

If the other flashcard programs out there were good enough


Woah, pick on Stackz all you want, but leave my flashcard program out of this :D
Last edited by debu on Tue 05.22.2007 5:18 pm, edited 1 time in total.
Try Train Your Brain flashcard software, free for TJP members. http://www.digitaldominion.com pm or email me for registration key
debu
 
Posts: 58
Joined: Wed 10.25.2006 4:47 pm

RE: New Anki (0.2.6 + new Mac build)

Postby Ezrach » Tue 05.22.2007 9:15 pm

Stackz is pretty horrible because of the method it employs. If you don't use Anki, some other good programs are Mnemosyne or SuperMemo.

"Study Moods" is just a way of saying, "I want to cop out today." If you don't want to study using flash cards, then you should be doing something else, like watching a TV program in Japanese, or reading a page or two in a novel.

The point of the flash card system is simply answering a single question multiple times - "Do you know what this means?" Consider it a quiz. You can't pick what questions go on a quiz.
Last edited by Ezrach on Wed 05.23.2007 9:31 am, edited 1 time in total.
User avatar
Ezrach
 
Posts: 270
Joined: Tue 07.18.2006 12:05 am

RE: New Anki (0.2.6 + new Mac build)

Postby Ezrach » Thu 05.24.2007 11:39 pm

I don't know if this is a bug or just a lack of implementation, but on the mac version the hiragana isn't automatically detected and inserted when creating a new card, so I have to manually insert everything.

The older version did this automatically. Also, in the old version my system would automatically switch between Japanese and English input in the input boxes where I specified it, but the new version I have to keep switching back and forth manually.

Also, sometimes, I can't make input a number to score myself, and I have to exit the program and then restart it.

Other than that, there are no problems.
User avatar
Ezrach
 
Posts: 270
Joined: Tue 07.18.2006 12:05 am

RE: New Anki (0.2.6 + new Mac build)

Postby resolve » Fri 05.25.2007 1:51 pm

Hi Ezrach,

The hiragana should be automatically generated - if it's not, it's a bug. I'll look into it for you.

I don't understand what you mean by automatically switching between Japanese and English. Do you mean when you tabbed to the answer field, it would automatically revert to English input? That was never an intended feature (normally it's not possible to trigger the input method like that), and must have just been a bug which may have ended up being more convenient!

You are able to switch between Japanese and English using Ctrl-Space or similar, right?

Regarding the score input, I'm not sure what's going on there. I've had 2 similar reports on Windows, but one hasn't experienced it recently and the other said it turns out they had their IME set to Japanese mode. Could that be it?

Cheers,

Damien
resolve
 
Posts: 135
Joined: Thu 09.14.2006 12:41 pm

PreviousNext

Return to Learning Materials Reviews & Language Learning tips

Who is online

Users browsing this forum: No registered users and 4 guests