john555 Wrote:I agree with part of what you said. I learned all 2,042 kanji in RTK1 WITHOUT USING SRS AND ANKI and without doing thousands of reviews.

The secret was using effective stories/mnemonics for each kanji. If you carefully craft your stories you shouldn't have to do too many reviews.

Very true. It's just that coming up with 2,200 (6th edition of RTK1) really effective mnemonics (or stories, or whatever we call them) has proved to be beyond me. Sure, about 1,600-1,700 I'm quite satisfied with and I rarely forget the associated Kanji, but the rest I tried with less-than-satisfactory mnemonics and lots of intense reviewing using SRS. But basically to no avail, as I've mentioned elsewhere in this thread.

john555 Wrote:Personally I find staring at same flashcards over and over again without effective mnemonics doesn't do anything for me. I tried that once it just doesn't work.

"Effective mnemonics" quite nicely describes the dividing line between remembering and not remembering, with SRS making little or no difference. The division that some other posters have made between the "primary study method" (i.e. the mnemonics/stories for memorizing Kanji) on the one hand and SRS as something quite apart from that "primary study method" with no other purpose than to combat "memory degradation" on the other hand, still seems artificial to me. To put it bluntly: if the "primary study method" is good, then there should be little or no "memory degradation" and hence very little need for SRS. And if it isn't good (enough), then no amount of SRS can help overcome that. Which brings me right back to... Einstein's (or whoever ascribed it to him) definition of insanity.

Eminem2 Wrote:To put it bluntly: if the "primary study method" is good, then there should be little or no "memory degradation" and hence very little need for SRS. And if it isn't good (enough), then no amount of SRS can help overcome that. Which brings me right back to... Einstein's (or whoever ascribed it to him) definition of insanity.

Although I don't use SRS, I find I do need to periodically review the kanji I've learned, even if the mnemonic stories were good.

I just think SRS/anki is too scientific. From what I understand, you end up with a bunch of different piles like "this pile I have to review every 8.777 days; this other pile I don't have to review again until 31.334 days have passed" etc.

I just take the list of 2,000 kanji, sort them in random order, print out the pages and on my coffee breaks etc. fill in the kanji with a pen. As I write in the kanji I think of the story in my head. Then I check my work. Some I get wrong, some that I got right last time I get wrong the next time, others I didn't get right the last time I get right this time. So far mistakes are well under 100. I do this once a month.

At the same time I'm working my way through a graded reader in Japanese which uses 700 kanji. When I'm finished this book, I'm going to work through the Miller reader which some learners consider "too difficult."

It occurred to me this morning, that in all my foreign language studies (elementary school, high school and university) none of my teachers ever mentioned the Leitner system or SRS or even using flashcards. I guess they assumed you would just "learn the vocabulary somehow".

Eminem2 Wrote:To put it bluntly: if the "primary study method" is good, then there should be little or no "memory degradation" and hence very little need for SRS. And if it isn't good (enough), then no amount of SRS can help overcome that. Which brings me right back to... Einstein's (or whoever ascribed it to him) definition of insanity.

It doesn't matter, you still need to review. It doesn't matter how memorable your mnemonic is, if you don't review it, you'll still forget it. Of course, it definitely helps, especially for kanji, it's never going to stick if you don't remind yourself of it.
If it helps, think of this: the most prioritized memories are those of trauma and those of embarrassment (unless there is repression involved, but that's not really forgetting, even though the person may be unable to access the memories); these are deemed the most important because they directly affect your ability to survive in the future. Now, tell me you remember every injury or embarrassment you suffered in your life (or even in just highschool or something, when these events are prevalent); chances are, you don't. Despite being some of the most likely memories to stick, most were deemed unimportant at some point and forgotten. Of course, the worst ones will still be there (I can still remember where and when I was shoved down onto a gravel walkway, getting a deep gash in my knee that had to have several pieces of gravel removed from it and required several stitches), but save for a select few, I've forgotten almost all of them. Even the one that I just mentioned was only put back together fully a year or so ago when I sat down and thought about it when I saw the scar.

Basically, my point is that, no matter how good your mnemonic or whatever primary study method you use is, you still need to review it. Memory simply doesn't work on a one time basis; even the strongest memories you have are only there because you couldn't get them out of your head and kept thinking about it.

The most common words can easily be reviewed with reading, but the more obscure ones will take a while to get to that strength and will probably stick best with artificial review, whether with flashcards or SRS.

SRS is just a more efficient way to review than flashcards, it's nothing more or less.

Eminem2 Wrote:Einstein once defined insanity as "doing the same thing over and over again and expecting different results."

1. No, he didn't. The Internet is full of quotes people randomly attribute to various recognizable names for no apparent reason.
2. Even if he did say it (and he didn't), it's just a witticism. Doesn't mean much, it's just meant to express a general sentiment in a clever way.

sholum Wrote:

Eminem2 Wrote:To put it bluntly: if the "primary study method" is good, then there should be little or no "memory degradation" and hence very little need for SRS. And if it isn't good (enough), then no amount of SRS can help overcome that. Which brings me right back to... Einstein's (or whoever ascribed it to him) definition of insanity.

It doesn't matter, you still need to review. It doesn't matter how memorable your mnemonic is, if you don't review it, you'll still forget it.

Perhaps after a (very) long period of time, but not nearly as quickly as the most common SRS algorithms suggest. If you try to memorize something and you found it hasn't stuck after about a month, then something is wrong with your primary learning method (or specific application thereof for the particular word or symbol). At its very best, SRS can alert you to this fact. But quickly running through the material you hope you've memorized will yield the same results without the need to regurgitate everything that you *have* memorized.

sholum Wrote:Now, tell me you remember every injury or embarrassment you suffered in your life (or even in just highschool or something, when these events are prevalent); chances are, you don't. Despite being some of the most likely memories to stick, most were deemed unimportant at some point and forgotten.

Since I don't have a list handy with all the injuries or embarrassments I suffered in just high school, it is pretty much impossible for me to determine if I can remember them all. As for them being some of the most likely experiences to stick? Perhaps, perhaps not. At the very least, memories of useful techniques that may come in handy at some time in everyday life are strong candidates for long term retention. Things like tricks for getting certain rare stains out of clothing, potentially dangerous situations in traffic that you may never have actually personally experienced but were warned against during driving lessons in the distant past, or certain scams that you may have read about but that you have not actually become a victim of. No need to SRS any of these, since all of them carry the tag "beware or be sorry" without any concrete embarrassment of traumatic experiences to be based upon. At least, that's how it works for me.

Another category that should never need SRS consists of important (positive) life events of family and spouses. Where your parents grew up, what line of work your grandparents were in, what line of work your brothers or sisters are in (even if you never really talk about their experiences at work), what your spouse's favourite romantic movie is (even if you never watched it more than once or twice together and it almost never crops up in conversation either). All of these are things that are probably marked "important" in your memory and that *do* stick in it for the (very) long term, even without the benefit of repeated reminders over the years.

My point here is that memorization is not just a matter of a certain learning technique or learning frequency. There must also be a certain amount of tagging involved, as I outlined above.

If something carries the tag "potentially dangerous and may be encountered some day", then it will stick (much) better than something without that tag.

If something carries the tag "is important to relatives, spouses or good friends", then it will stick (much better than something without that tag.

And if something carries the tag "may come in very useful some day in the context of speaking a certain language", then it will stick better than something without that tag.

So that's why I believe that seeing words, phrases or pictograms in action in real-life situations such as texts, cartoons, bits of speech or bits of film will stick much, much better in memory because they automatically carry the tag that says "may come in very useful some day in the context of this language" in ways that lists of words or symbols on a page never can. Unless they are given mnemonics that are so good that they become hard to distinguish from such real life applications.

And that there will always be some category of more obscure or confusing words or Kanji that do not stick easily, does not validate the concept of SRS for me. That it may work for some is not nearly as relevant to me as that is hasn't worked for me, even after giving it more than a fair chance.

Stansfield123 Wrote:

Eminem2 Wrote:Einstein once defined insanity as "doing the same thing over and over again and expecting different results."

1. No, he didn't. The Internet is full of quotes people randomly attribute to various recognizable names for no apparent reason.

I acknowledged as much in later posts in this thread.

Stansfield123 Wrote:2. Even if he did say it (and he didn't), it's just a witticism. Doesn't mean much, it's just meant to express a general sentiment in a clever way.

You cannot possibly know that Einstein never said this. Even if the quote only popped up as late as the 1980's then it may still be something that he once said to someone who only reported this later. Or if he never said it, it may still be a fair summing up of his way of thinking. That it's just a "witticism" is nothing more than your opinion. Given how many times solutions that haven't worked in the past get proposed again and again just because no one has managed to come up with a better idea, shows that there is more than a grain of truth to this quote.

Eminem2 Wrote:You cannot possibly know that Einstein never said this.

So your new position is that Einstein may have said it, therefor it is a universally applicable logical truth?

In that case, my reply is that insanity is the inability to see the immediately obvious. Another way to phrase the same thing is that insanity is an error in judgment that can't be fixed by appealing to logic and reality.

[EDIT]: I'm talking about using SRS, not in the context of RTK, but with Core 10k and adding your own words. Also, the "quotes" at the end are meant to be a jab at misattributions.

Whether Einstein said it or not is irrelevant. Whether it's a "fair summing up of his way of thinking" is irrelevant. Yes, he contributed greatly to the field of physics, but that doesn't mean he's right about everything or never made mistakes. Einstein's a popular figure to stand in for the "genius archetype", but no one's right about everything.

Stansfield123 Wrote:Given how many times solutions that haven't worked in the past get proposed again and again just because no one has managed to come up with a better idea, shows that there is more than a grain of truth to this quote.

Please provide some examples, otherwise it'll seem you're pulling things from the air. Hopefully you're not confusing "solutions that get proposed again and again" with prototypes that get modified. Just look at the Wright brothers and the path to the airplane.

Also, it's pretty well known that memories with strong emotional/physical stimuli tend to stick around longer. Conversely, memories that are "neutral" are harder to keep around. Let's say you're reading a book on Monday and come across a word you don't know. If you want to know what it means, you can look it up. Sure, you can make a mnemonic or use imagery or whatever you like to remember it for the future. Now, let's say you come across 50 new words during that session. Maybe if you spent a good amount of time and effort, you might be able to still remember them after a few hours. But if you don't see them again, can you still do so after Wednesday? Now, let's say Tuesday, you came across 20 new words. You try to remember those. On Wednesday, you encounter 10 words. After this you go back and reread the passages you've read during the three days. I seriously doubt you'll be able to remember all 70 new words having seen them just once. I know you've acknowledged this before, but the point I'm trying to make is that there's just way way too many words to have everything paired with a stimulus. That's why SRS is good.

So basically, what I'm saying is this: You might pick up a few (hundred?) words based on how memorable they are for you. But it's nowhere near the thousands and thousands and thousands needed for a solid vocabulary. That's why you need some way to review, be it flashcards/SRS or reading. You can't see every word every day. And people who claim to to not use SRS for reviewing usually use a method that's similar, but less automated (based off the posts I've read). Sure, I could print out lists of 10,000+ words and review sections throughout the day but...


And if you want to argue "witticisms" or whatever, here's two that you're probably more familiar with.
"At first if you don't succeed, try, try again." - Barack Obama
"Practice makes perfect" - Marie Curie


If you're not averse to a little reading, I suggest checking out Ebbinghaus's Forgetting Curve. This is basically the cornerstone of SRS (at least in principle, not sure if it was intentional).

There's no need for 2200 perfect mnemonics and perfect recall. The book is made to enable you to learn the Kanji first, initially. You won't stop there. You'll then learn words using the Kanji, which provide more connections. And then other words using the same Kanji, and even more connections. And then you'll learn ways to express, aka. noun+verb pairs, those are pretty common in Japanese (i.e. you use 引く with 風邪 or 干す with 洗濯物 even those other verbs would do the job just fine - they are just unnatural). Step by step, the Kanji will become parts of other things, and those will then become part of even more complex things. See the trend?

That's why learning and maybe reviewing them a few times is enough. No need to ANKI 'em up till they're all on a several-years-timer. One day 人 will come up and you will connect it with 人間、友人、人類、保証人 and so on. The least you want to connect it with is "person". There will be no more Mr. T, no more Spiderman and no more Gandalf. They're training wheels to teach you how to (dis)assemble the Kanji, and that's it.

Actually, no ANKI needed at all. It does help a little tho, till you assigned words to all of the Kanji you did with RTK.

Eminem2 Wrote:But quickly running through the material you hope you've memorized will yield the same results without the need to regurgitate everything that you *have* memorized.

Reviewing everything over again takes a lot longer than using an SRS

EDIT: I do agree that anki isn't really needed for rtk, but that's because you eventually don't need to remember the connection and story by itself.

andikaze Wrote:There's no need for 2200 perfect mnemonics and perfect recall. The book is made to enable you to learn the Kanji first, initially. You won't stop there. You'll then learn words using the Kanji, which provide more connections. And then other words using the same Kanji, and even more connections. And then you'll learn ways to express, aka. noun+verb pairs, those are pretty common in Japanese (i.e. you use 引く with 風邪 or 干す with 洗濯物 even those other verbs would do the job just fine - they are just unnatural). Step by step, the Kanji will become parts of other things, and those will then become part of even more complex things. See the trend?

Yes, I see the trend. The trend is that, unless you make sure to keep things simple by learning in increments, your learning material will grow increasingly complex and it will be exponentially more difficult to manage.

If, when you encounter a word for the first time, you have to look up the stories for several Kanji, work to remember them by visualizing those stories, and then go about the separate task of trying to actually learn the word, that means you failed at reducing the task of learning that word to its simplest stand-alone components.

Heisig's motivation for creating RtK was to give western students of Japanese the luxury Chinese students of Japanese have: they already know the basic meaning of the Kanji.

I assure you, SRS-ing the Kanji separately is an effort worth making. You more than recoup the time spent on that simple task, later on, when the otherwise complex task of learning Japanese compounds becomes an order of magnitude simpler. Especially since SRS-ing 2000 simple cards (maybe less, if you eliminate Kanji you don't really need for the initial vocab) doesn't take up that much time.

andikaze Wrote:That's why learning and maybe reviewing them a few times is enough. No need to ANKI 'em up till they're all on a several-years-timer. One day 人 will come up and you will connect it with 人間、友人、人類、保証人 and so on. The least you want to connect it with is "person". There will be no more Mr. T, no more Spiderman and no more Gandalf. They're training wheels to teach you how to (dis)assemble the Kanji, and that's it.

Actually, no ANKI needed at all. It does help a little tho, till you assigned words to all of the Kanji you did with RTK.

I agree that you don't need to review them separately for years (unless you go back to practice your writing or something), I'm just arguing that SRS itself is a very useful tool that should be employed by most students.
Personally, I didn't spend much time on RTK and I'm doing fine (about four months, including review). I just used it to be able to 'see' the kanji.

I also agree that the kanji become abstract ideographs after you study a lot of vocabulary. When learning new words, I usually only have difficulty with words that use less frequent kanji.

Eminem2 Wrote:Since I don't have a list handy with all the injuries or embarrassments I suffered in just high school, it is pretty much impossible for me to determine if I can remember them all.

That's not required to know that you don't remember something. For instance, I know I've been stung by wasps more than once in my life, but I only remember one particular event because it was useful (I learned not to swat at unidentified bugs that land on me).
I also know that I've been cut by metal numerous times, but only really remember a couple of times (such as one time where I was careless when using an old handrail and caught a nail sticking out of it).

That's just how memory works; in most cases, only excessively positive or negative events are given such priority that they rarely require review (though you usually review them in your head anyway). All current understanding of memory suggests that, for the average human, review is required to retain memories for more than a few days, it doesn't matter if the review is in your head or managed through some physical interaction. If you think otherwise, by all means, design a valid experiment that can test your claims and publish it, 'cause if you've found a way to use mnemonics better than everybody else, a lot of people will want to know about it. I'm being serious. I've not been given any valid information that suggests that the average human can intentionally remember a data set for multiple weeks after only seeing it once.

So, are your mnemonics just so memorable that you think about them a couple of times that same day? What about later in the week? If yes, that's reviewing. I don't remember if you said how many new kanji you study per day or if/how many you review, but if that's how your memories are being reinforced, then I'm fairly sure you'd be unable to scale that up to a level where you could have thirty or more new entries a day.

I'm not trying to say you're lying or that your method doesn't work (if it works it works), I'm just trying to express my doubts at its sustainability with larger data sets or at higher rates of entry. SRS is a review management system, not a learning system (though it's just as easy to learn some words in it, if you look up some example sentences).
It's also useful for obscure vocabulary as well, but some might decide it's just as well to only bother with it when they see it.

Anyway, I really should quit messing with a pointless argument (does it benefit anyone? Not that I can think of; thus it's pointless) and just go back to studying. This argument isn't even applicable to it, since I'm working on listening.

Kuzunoha13 Wrote:[EDIT]: I'm talking about using SRS, not in the context of RTK, but with Core 10k and adding your own words. Also, the "quotes" at the end are meant to be a jab at misattributions.

Stansfield123 Wrote:Given how many times solutions that haven't worked in the past get proposed again and again just because no one has managed to come up with a better idea, shows that there is more than a grain of truth to this quote.

Please provide some examples, otherwise it'll seem you're pulling things from the air.

Since I'm the one who really said that, I'll provide some examples (mostly from the field of economics):
- whenever there's a recession, calls for government stimulus packages crop up again, even though their efficiency is quite dubious, to say the least. (And speaking of misattributions, Keynes is often cited in defense of such measures, even though he never called for such extreme ones).
- calling for a heavier crack down by law enforcement agencies on drug abuse, even though the problem hasn't really been alleviated over the past few decades by doing just that (just look at what the power of drug cartels has done to Mexico). Yet, since the alternative of legalization is deemed too horrid to contemplate (and not without cause), the "War on Drugs" continues unabated. And uncelebrated.
- the insistence on simplifying the actual economy to the reassuring maxims (perfect information and perfectly rational decision makers, to name a few) once posed by the neo-classic economists (not be confused with the new-classic economists). This led to the conclusion that every economy always tended towards full employment, so that any unemployment that existed was by definition of the "friction" variety. Even in the face of the massive unemployment of the 1930's, neoclassic economists maintained their defence that this was all just frictional unemployment. And in case it wasn't, then this could only mean that wages were simply too high because all markets (including the labour market) always tend towards equilibrium. Keynes brilliantly critiqued all the false assumptions underlying these obviously false conclusions in his "General Theory", only to become the unintended champion of massive deficit spending as a result of it.

Perhaps the well-known concept of "cognitive dissonance" has a lot to do with the persistence of irrationality. If the mind is so strongly entrenched in the idea that reality works in a certain way, then any fact that contradicts that idea may very well be rejected. The fairy tale of the "Emperor's New Clothes" might be seen as an old illustration of this idea. Why couldn't the same be true for people who defend SRS after having invested a lot of time and energy in it?

Kuzunoha13 Wrote:I know you've acknowledged this before, but the point I'm trying to make is that there's just way way too many words to have everything paired with a stimulus. That's why SRS is good.

SRS is good, because the alternative is... not workable (at least you believe it to be so)?

Kuzunoha13 Wrote:So basically, what I'm saying is this: You might pick up a few (hundred?) words based on how memorable they are for you. But it's nowhere near the thousands and thousands and thousands needed for a solid vocabulary. That's why you need some way to review, be it flashcards/SRS or reading.

It's almost like I'm hearing the echoes of a lecture of a neoclassic economist from the 1930's: "because all markets always tend towards equilibrium, there can be no such thing as structural mass unemployment. Don't let all those disshevelled people standing in line for some free soup convince you otherwise: they are simply unwilling to work for the lower wage at which a full employment equilibrium could be reached!"

Kuzunoha13 Wrote:You can't see every word every day. And people who claim to to not use SRS for reviewing usually use a method that's similar, but less automated (based off the posts I've read). Sure, I could print out lists of 10,000+ words and review sections throughout the day but...

But...? But only SRS will somehow help you accomplish that? Even the inventor of SRS (to whose research links were provided earlier in this thread) admitted that it doesn't work that well for databases of over 1,000 items.

Kuzunoha13 Wrote:If you're not averse to a little reading, I suggest checking out Ebbinghaus's Forgetting Curve. This is basically the cornerstone of SRS (at least in principle, not sure if it was intentional).

I'm not adverse to a little reading, but after reading through the links given earlier in this thread and finding how surprisingly little support they contained for SRS for larger databases, I'm not looking for a repeat of that. And what's more: my own experience of language learning indicates that SRS isn't necessary (neither are pseudo-SRS methods that you refer to above). Now, I don't expect anyone else to accept my own experiences as proof, but they do count rather heavily for me.

cracky Wrote:

Eminem2 Wrote:But quickly running through the material you hope you've memorized will yield the same results without the need to regurgitate everything that you *have* memorized.

Reviewing everything over again takes a lot longer than using an SRS

Perhaps an example with numbers is better suited for the point I'm trying to make regarding SRS and Kanji. Let's say that on some random day 100 cards are up for review in Anki (or another SRS-based system). If the reviewer has hit on a decent primary method of learning, then (s)he may get 85 of them right, while 15 of them go into the "failed" stack. If it takes 10-15 seconds to completely draw each Kanji (whether correct or incorrect), then the entire review takes between 1,000-1,500 seconds. Or between 16 and 25 minutes.

If, however, the reviewer randomly selects a number of pages from the book (perhaps using a random number generator) and is honest with him/herself which Kanji they could have faithfully reproduced, then it should take much, much less time than between 16-25 minutes to identify which need to be relearned. Since none of the tedious reproducing of Kanji that the reviewer already knew is necessary. (And even then, I strongly doubt if applying the same method of learning again and again to the failed cards is all that useful. But I've expounded on that point of criticism in many of my other posts in this thread.)

The whole Heisig system is just an aid to learn the Japanese "alphabet". The same way you'd want to learn the Cyrillic alphabet if you were learning Russian. This English keyword system is a good method for those who have the ambition to power through the books in the beginning - the same way some people would want to learn any foreign alphabet when learning a new language.

Now, you could do the Heisig system while studying with materials that use furigana (which most beginner and even many intermediate materials do), and in that way spread out the learning over a longer period of time. And that would be fine too. If you do that though, you're inevitably going to lose some connections between keyword and kanji. I don't think it matters. A lot of keywords are similar.

But the longer the spread of time you use, the less effective the Heisig method will be, it seems to me. You'd be better off learning the radicals and then learning kanji as you encounter them in your studies, perhaps doing part 1 of Heisig just to get used to the idea of using visual memory. Where is that tipping point with regards to time? I don't know. It seems reasonable to me that you could do Book 1 of R the K over maybe 1 year.

But you WILL lose the keywords, because they'll no longer be important, so your brain discards them. And you WILL make the readings visible at some point, because Kanji don't only have one, and some are quite arbitrary. The single Kanji will also only serve a purpose until you got them connected to at least ONE word. so 人 will become ひと and 安 will become あん, and later you'll learn more vocab and other readings for them - but this will happen automatically. And the sooner you start to actually learn vocab through normal reading, the sooner this will happen. Sure, you may facepalm because you just looked up 表情 for the 20th time, but then you will encounter 時給 after you already knew 給料and you'll never have to review this word again.

What people also seem to believe is that knowledge is like living beings. They're "dead when they die". This is NOT the case. You may come across 調停 because your girlfriend is in the middle of a divorce, and 10 years later you will look it up because you forgot it, and a light bulb will go an. Oh yeah, 調停 - and when you need it again 10 years later, you might not even have to look it up again. Knowledge does not vanish. The connections might go, but once you re-establish them, you'll have that puzzle piece back.

And why would you have to remember, of all things, 調停 100% of the time when you only need it like 3 times in your life?

Forgetting stuff seems to scare learners. In my own experience, stuff you need to know will become solid eventually. I know, I did it with English, and it worked for Japanese. There's nothing scary about forgetting some words or Kanji. Some will come back when you need them, and some you'll look up quickly.

Besides, you don't need 10000 words in your active vocabulary. Most native English speakers never use more than 2000 in their whole life. It's enough to know them passively, and for that, ANKI isn't needed.

You can use it to initially drill the stuff and then to hold on it until you're confident you will get it when you come across it, or can use it in case you find it useful, but there's no need to SRS it.

I've read expressions like よりによって or どこからともなく ONCE, then used them 2 or 3 times, and while I might not always come up with them when I could use them, at other times I do, and when I stumble across them I know what they mean.

People, especially on this forum, seem to be quite obsessive when it comes to certain tools or certain methods. But there's no need to limit your horizon by that and stubbornly (or anxiously?) stick to one thing when others work well, too, and when forgetting is not REALLY a big deal.

Yeah, I meant to quote you, but it got mixed up.
Anyway, I'm not too familiar with those issues. However, I have heard about the pitfalls of stimulus packages, but then again, this issue and the other ones you outlined are possibly meant to be to be stopgap measures. And while they don't directly solve the problems at hand, it's better than doing nothing. On an aside note, this is the same reason I'm against affirmative action in academia (and other fields), but that's a different topic. I wouldn't call these things "insane", but rather, a consequence of having to weigh several different variables at once, with the "superior" outcome coming with its own set of problems.

Cognitive dissonance is not the appropriate term to use here. The reason SRS is recommended enthusiastically is because the results are tangible...er..."readable" for people who've invested time into it. Personally speaking, I went from being able to read nothing (about <50 words) to 60-70% of a text in around 8 months. Of course, I was reading grammar books at the same time, but most of those honestly use pretty common vocab. If you have any other method that'll give the same results --IN THE SAME TIME FRAME-- feel free to share. According to you, it took you 3-3.5 years to do something similar? If that's the case, why is it better to spend 3+ years doing something when you can do in under 1?

Oh, BTW, I'm not trying to defend SRS specifically. SRS is just a method of review (but for me, the most optimum). And honestly, I don't think anyone who is learning Japanese is just SRSing all day - they're reading and listening and watching stuff. And that in itself is review as well. But if I want to learn a word, I can't know for sure when it'll come up again. But, when it does, I want to know it. Once again, you can't remember thousands and thousands of words without reviewing them (whether purposefully or coming across them), no matter how good your so-called "primary method" is.

Also, I don't understand your quote about the neoclassical economist - how is that related to my point? What is your suggestion for an equivalent experience that doesn't take years to develop?

You can learn a lot of stuff in context, but how to learn from context when you're looking up 9 words out of 10? Especially as a beginner, when pretty much every word is unfamiliar. Here's one answer - you look all of it up (or guess). Again and again, until it sticks. And hopefully, you won't forget. And if you forget...you look it up again. Or, you could make a list and study them purposefully. Guess which one is less tedious?


About the 1000+ card database:
If you're referring to this:
http://www.supermemo.com/english/ol/beginning.htm

I suspect you didn't actually read the whole article. The first database he created was paper only, and reviewed at irregular intervals. Once he experimented and fine tuned his algorithm, he achieved 80% retention, with decks well over 1000. Once again, SRS is just a tool you can use to review, but one that is highly customisable and controllable.

Eminem2 Wrote:review takes between 1,000-1,500 seconds. Or between 16 and 25 minutes.

If, however, the reviewer randomly selects a number of pages from the book (perhaps using a random number generator)

So it's possible they could just not review the material they actually need to review then, that doesn't sound like a better system but alright. Also this point is valid for early on when you do spend some time reviewing cards you already know very well, but cards mature. The book method would be spending much more time reviewing things you don't need to. Your example system seems like it will take more time overall and also be possibly totally useless depending on the random numbers that come up.

Eminem2 Wrote:(And even then, I strongly doubt if applying the same method of learning again and again to the failed cards is all that useful. But I've expounded on that point of criticism in many of my other posts in this thread.)

I think most people agree with that, that's why anki has the leech system to show you what cards you should either try something else with or delete. I think anki is already pretty well built to deal with most of your criticisms.

Edit: oh and one final thing

Eminem2 Wrote:If the mind is so strongly entrenched in the idea that reality works in a certain way, then any fact that contradicts that idea may very well be rejected.

Eminem2 Wrote:my own experience of language learning indicates that SRS isn't necessary (neither are pseudo-SRS methods that you refer to above). Now, I don't expect anyone else to accept my own experiences as proof, but they do count rather heavily for me.

Kuzunoha13 Wrote:Cognitive dissonance is not the appropriate term to use here. The reason SRS is recommended enthusiastically is because the results are tangible...er..."readable" for people who've invested time into it. Personally speaking, I went from being able to read nothing (about <50 words) to 60-70% of a text in around 8 months. Of course, I was reading grammar books at the same time, but most of those honestly use pretty common vocab. If you have any other method that'll give the same results --IN THE SAME TIME FRAME-- feel free to share.

No, I can't say I've found *any* method that does that within that time frame. Including SRS.

Kuzunoha13 Wrote:According to you, it took you 3-3.5 years to do something similar? If that's the case, why is it better to spend 3+ years doing something when you can do in under 1?

Since that method is the only one that's ever worked for me, there is no speed comparison. Just "works" versus "does not work". But assuming that SRS had worked for me, then the comparison wouldn't just have centered around the number of years something took, but also the amount of time it took on a weekly or monthly basis.

And my own method of gradual immersion cost very little time at all in that sense. I just kept comparing the spoken text I heard to the subtitles I saw, whenever I happened to watch TV. And kept trying to make out the words when listening to pop songs. And tried my hand at more and more advanced books of a type that I wanted to read anyway. And meanwhile, of course, I did the exercises in preparation for my high school classes (including vocab, but never repeating any material that I had already been tested on). But those were fairly limited in scope and depth, and virtually all of my classmates continued to struggle with the language while taking the same classes abd doing the same exercises I did.

So I guess for me the comparison is not so much between "1 year" and "3.5 years", rather than between "1. compressed language learning by doing lots of extra work in a relatively short time" and "2. effortlessly learning a language through gradual immersion, while none of the work really *feels* like work". (And method 1 so far hasn't worked at all for me.)

Kuzunoha13 Wrote:Oh, BTW, I'm not trying to defend SRS specifically. SRS is just a method of review (but for me, the most optimum). And honestly, I don't think anyone who is learning Japanese is just SRSing all day - they're reading and listening and watching stuff. And that in itself is review as well.

That in itself is probably so much more than just a review, seeing as it provides the reader/listener/viewer with the experience of seeing a word in action. And although I can't cite any scientific proof to back it up, I remain convinced that this makes for much better retention because the mind automatically assigns such words with a tag that says something like "seen it in action and it works". That kind of experience has to be worth more than dozens of reviews.

Kuzunoha13 Wrote:But if I want to learn a word, I can't know for sure when it'll come up again. But, when it does, I want to know it. Once again, you can't remember thousands and thousands of words without reviewing them (whether purposefully or coming across them), no matter how good your so-called "primary method" is.

And yet somehow, that is exactly how you learned the more difficult words of your native language. ;) Even if they only appeared rarely, far more rarely than an SRS algorithm would claim is necessary. Did you constantly have to review fairly rare words like "obsequious", "perspicacity" or "serendipity" before you remembered their meaning? (Their spelling might be a different matter, and I'm not sure I spelled them all correctly.)

Kuzunoha13 Wrote:Also, I don't understand your quote about the neoclassical economist - how is that related to my point? What is your suggestion for an equivalent experience that doesn't take years to develop?

Turns out it didn't apply to you. (I meant it as an example of repeating an analysis that has been proved to be incorrect, but since you claim to have made enormous strides in Japanese reading skills in less than a year, you at least have some concrete results to show for it. Which the neoclassical economists never did.)

Kuzunoha13 Wrote:You can learn a lot of stuff in context, but how to learn from context when you're looking up 9 words out of 10? Especially as a beginner, when pretty much every word is unfamiliar.

As I mentioned somewhere much earlier in this thread, that could be resolved by graded readers, so that the ratio of new:old words would be much more gentle. But those readers turn out to be rather rare and expensive for Japanese, I understand. But even without those, learning a language through gradual immersion is possible, in my experience. (Although not having access to simpler novels in the target language might prove troublesome.)

Kuzunoha13 Wrote:Here's one answer - you look all of it up (or guess). Again and again, until it sticks. And hopefully, you won't forget. And if you forget...you look it up again. Or, you could make a list and study them purposefully. Guess which one is less tedious?

Looking up stuff from material that you love (be they books, films, TV-shows or songs) is anything *but* tedious, in my experience. And because you love the material, the new words it provides you with will tend to stick around in your memory a lot easier. But since it's hard to quantify the effects of "love" or "fun" on learning, I doubt there will ever be scientific articles to support this notion...

Kuzunoha13 Wrote:About the 1000+ card database:
If you're referring to this:
http://www.supermemo.com/english/ol/beginning.htm

I suspect you didn't actually read the whole article.

I'll take your word for it, but my own experience with SRS still stands.

cracky Wrote:Edit: oh and one final thing

Eminem2 Wrote:If the mind is so strongly entrenched in the idea that reality works in a certain way, then any fact that contradicts that idea may very well be rejected.

Eminem2 Wrote:my own experience of language learning indicates that SRS isn't necessary (neither are pseudo-SRS methods that you refer to above). Now, I don't expect anyone else to accept my own experiences as proof, but they do count rather heavily for me.

Cognitive dissonance is generally more about denying actual experience in favour of theoretical convictions, than about convictions that are well rooted in actual experience. Nor did I try SRS with any negative expectations. (If anything, I was very hopeful that it would help me realize the advertised results. So I was biased in the opposite direction from the conclusion I finally reached.)

Eminem2 Wrote:Even if they only appeared rarely, far more rarely than an SRS algorithm would claim is necessary. Did you constantly have to review fairly rare words like "obsequious", "perspicacity" or "serendipity" before you remembered their meaning?

*Goes and looks up all three words*

At least I recognized the third, and may understand it if I saw it in context, even if I couldn't give a definition for it :p

Eminem2 Wrote:

cracky Wrote:Edit: oh and one final thing

Eminem2 Wrote:If the mind is so strongly entrenched in the idea that reality works in a certain way, then any fact that contradicts that idea may very well be rejected.

Eminem2 Wrote:my own experience of language learning indicates that SRS isn't necessary (neither are pseudo-SRS methods that you refer to above). Now, I don't expect anyone else to accept my own experiences as proof, but they do count rather heavily for me.

Cognitive dissonance is generally more about denying actual experience in favour of theoretical convictions, than about convictions that are well rooted in actual experience. Nor did I try SRS with any negative expectations. (If anything, I was very hopeful that it would help me realize the advertised results. So I was biased in the opposite direction from the conclusion I finally reached.)

Sure, but your main argument boils down to the fact that it didn't work for you so it doesn't work. You see how that's logically flawed right? It not working for you doesn't say anything about if it works for someone else.

I don't know much about cognitive dissonance so I can't speak on that but I am implying that you have a confirmation bias because of your personal experiences.

EDIT: The other part of your argument is that cognitive dissonance exists so it could be at work here but no real evidence to support that it is.

Eminem2 Wrote:Perhaps an example with numbers is better suited for the point I'm trying to make regarding SRS and Kanji. Let's say that on some random day 100 cards are up for review in Anki (or another SRS-based system). If the reviewer has hit on a decent primary method of learning, then (s)he may get 85 of them right, while 15 of them go into the "failed" stack. If it takes 10-15 seconds to completely draw each Kanji (whether correct or incorrect), then the entire review takes between 1,000-1,500 seconds. Or between 16 and 25 minutes.

If, however, the reviewer randomly selects a number of pages from the book (perhaps using a random number generator) and is honest with him/herself which Kanji they could have faithfully reproduced, then it should take much, much less time than between 16-25 minutes to identify which need to be relearned. Since none of the tedious reproducing of Kanji that the reviewer already knew is necessary. (And even then, I strongly doubt if applying the same method of learning again and again to the failed cards is all that useful. But I've expounded on that point of criticism in many of my other posts in this thread.)

Your argument is invalid because you are comparing two different procedures. The second procedure would take less time with a SRS. SRS doesn't mean you have to write out the kanji.

Eminem2 Wrote:Perhaps an example with numbers is better suited for the point I'm trying to make regarding SRS and Kanji. Let's say that on some random day 100 cards are up for review in Anki (or another SRS-based system). If the reviewer has hit on a decent primary method of learning, then (s)he may get 85 of them right, while 15 of them go into the "failed" stack. If it takes 10-15 seconds to completely draw each Kanji (whether correct or incorrect), then the entire review takes between 1,000-1,500 seconds. Or between 16 and 25 minutes.

If, however, the reviewer randomly selects a number of pages from the book (perhaps using a random number generator) and is honest with him/herself which Kanji they could have faithfully reproduced, then it should take much, much less time than between 16-25 minutes to identify which need to be relearned. Since none of the tedious reproducing of Kanji that the reviewer already knew is necessary. (And even then, I strongly doubt if applying the same method of learning again and again to the failed cards is all that useful. But I've expounded on that point of criticism in many of my other posts in this thread.)

Some of the false statements/assumptions you've made in this post, about using Anki with RtK, and Japanese writing in general.

1. There's no requirement to "tediously draw each Kanji", in Anki. There's no requirement to write them at all, unless you want to learn how to write by hand. Most people don't.

2. If you are learning to write, it's not true that writing (not drawing, writing) a Kanji you know takes 10-15 seconds. You're way off. Writing Japanese takes about the same time it takes to write English. A lot at first, but very little once you know it.

3. The average number of Kanji Anki would give you a day, during the review phase, assuming the failure rate you assumed, would be far less than 100.

4. The frequency with which Anki would give you cards you don't know, would be far more frequent than once in 20 days (it would be once every 10 minutes, and then once a day, by default).

5. Your alternate method is terrible. Once in 20 days (which is the frequency you would review unknown Kanji with using your alternate method) is not enough to efficiently learn a Kanji you don't know. No one in their right mind would try to learn something by looking at it once every 20 days. It's such an obviously absurd thing to do.

6. This is by far the most important one: the notion that Anki forces you to spend a lot of time interacting with Kanji you already know is blatantly false. The vast majority of time is spent on Kanji you don't know.

If you know 1000 Kanji perfectly well, and 1000 none at all, then the time you need to spend on the 1000 Kanji you know is less than 1% of the total. The amount of time you need to spend on a card you know in Anki is ONE SECOND. That's the time it takes to press Shift-1 and suspend that card.

If you know 1000 Kanji fairly well, then the time spent on them is still under 5% of the total, and on the other 1000, 95%. Even without suspending a Kanji, the total time you have to spend on it is no more than 10 seconds (that's how long it would take to review the card 4 times in its lifetime; and 4 is the total number of times Anki would show you an easy card, during the first year, if set up correctly). Meanwhile, if you used your alternate method for a year, you would have to take a look at that Kanji you know once every 20 days: that's 18 times a year. So it's really your alternate method that wastes time, not using Anki. Anki saves time, by hiding the cards you know well automatically, and only showing you the ones you have trouble with.

I'm surprised this thread went as far as it did considering how inane the initial post was, with the fake Einstein quote and the silly conclusions drawn from it.