I was thinking about a system to help learn vocab without Kanji. In other words, to help speak the Japanese language (and read furigana) without learning the Kanji first.

When we memorize new vocab, many of us rely heavily on the Kanji. Not only that, we rely on the Heisig names the Kanji has. Some even use mnemonics. However, there are up to 2000 Kanji even in a beginner/intermediate student's vocab, so any mnemonic system uses up to 2000 primitives.

It occurred to me that, without the Kanji, a mnemonic system can have as many primitives as we wish and deem optimal (with a minimum of 100 or so). Such a system could name each syllable and treat it as a primitive. It can also give the same syllables different names, depending on their function. If it's a very common Kanji read as "ka", it can use the same name for that reading only, and other names for other occurrences of "ka".

What is important about this is that we can keep the set of primitives at the exact size we want it, at any stage in the learning process: not too low, to make the stories people use too confusing, and not too high to make remembering the names too hard.

There would of course be issues: verb stems and other single syllable words would require special attention, this system wouldn't work for them. They could a lot of times simply be treated as primitives (the same way Heisig designates certain Kanji as primitives), but there is too many of them for them to all be primitives. At some point, there would need to be mnemonics to help derive them from other, identical sounding primitives.

I would be interested in any critique of this system. I am considering doing some preliminary work on this, so, if it's a bad idea, I'd appreciate brutal honesty (backed up by arguments, of course). It will save me some time.

So your primitives would be either 1 or 2 syllables?

I can see that working, especially for many 2-syllables units, but there are just so many words using 1-syllable units... I think it would be possible though, in fact that's probably the way most learners intuitively do it (those who don't feel like beginning by learning kanji). A systematic approach (RTK-style, with no kanji) would be cool! I'm not sure though if many people would use such a method. It seems to me that people really into learning Japanese and ready to use this kind of long-winded method would rather learn the kanji first (because kanji is cool!).

comeauch Wrote:So your primitives would be either 1 or 2 syllables?

Oh yeah, I forgot about the two syllable readings. Yeah, I suppose some would have to be two syllable.

comeauch Wrote:long-winded method

Another fair criticism, thanks. The order in which the vocab is learned would indeed depend on what primitives it's made up of, rather than how common the words are.

So, just like RtK, it would be an all or nothing approach.

Im one of the people who never (or just tried it for a week) learned the kanji, something like RTK.
I don't have any system when learning vocabulary. I just pick the first 30 that are on my list and start learning. So I can't rely on the kanji as hints. And I don't think, that reling on kanji really helps when you want to read fast, because your brain just looks at the shape of the words. And after I learned the vocab I can reproduce the most of them after they appeard a few times in Anki.

But besides that, learning vocabulary without kanji isn't a good idea. I did the same when I first started learning japanese and regreted it.
Don't learn it the hard way like I did. I had to connect like 1.000 vocabulary from new to the kanji because I didn't used them before. And what I got was only wasted time. (and the experince that it was a bad idea)
Just learn them and then read some textes and do some listening, or in your case, just speak. Do something like japanesepod while learning to use your new learned vocabulary or better find some real japanese people or do skype conversations. This way you will boost your reading, listening and speaking at the same time.

And reading an text without kanji..is just a pain. It's for me even harder then listening, because with hiragana you have like no idea where something starts and where it ends.

There's a blog I came across months ago called something like "learn Japanese through anime." In it, the person talks about how he began learning Japanese through watching anime, finding transcripts online of the anime he liked, and putting the transcripts through some type of program that revealed the most commonly used words in them.

He then learned about 300 of those vocab words through mnemonic devices (without knowing kanji I think). But the definitions were translated into English, and the vocabulary words were in romaji. That sort of follows the lines of learning vocab without kanji just for speaking.

Another article I saw was on the Fluent in 3 Months blog where a person used an immersive method alongside learning the top 1,000 most common Japanese words. I think that was done without kanji and was done primarily with speaking in mind. He even did something comical called "Fake it until you make it" where he basically acted like he knew Japanese and acted like he was Japanese (maybe it came off as offensive, I'm not sure).

I thought about using one of the above methods in conjunction with watching misc Japanese media, but I'm less focused on speaking at the moment, so I intend to learn kanji first, and learn vocab with kanji afterward.

I don't really see any need to use mnemonics for vocabulary. It might work, but I think it's straight-up unnecessary.

I haven't learnt all the kanji - although I know a fair few purely from exposure and that can be useful, most of the time when learning vocabulary I'm not relying on them. Given audio, a sentence, and Anki, it all works out fairly well. It's not really different from learning vocabulary from a language without kanji. I doubt my card failure rate is significantly worse than people relying on piecing together meanings with Heisig's keywords.

Victor Mair (a linguist specializing in Chinese) had an interesting article on Language Log the other day in which he wrote:

Quote:If I were the czar or god of Chinese and Japanese language pedagogy, I would not teach students a single Chinese character until they were relatively fluent — about two years. I've always said that we should learn languages the way babies do; they learn to speak long before they learn to write.

http://languagelog.ldc.upenn.edu/nll/?p=10554

That was more or less my approach when I started learning Chinese more seriously a couple of years ago, relying heavily on podcasts and not focusing too much on characters. The problem I ran into is that I found it tremendously difficult to get enough low-level aural exposure without taking a class. Once I started focusing on characters more heavily, my reading comprehension got better, and reading is much easier than listening because you can slow it down as much as you need to -- and reading can improve your vocabulary to the point where TV starts being easier to decode in realtime.

I'm still very much in agreement with Mair in theory, but it's difficult to make it work for people studying on their own. (And that's why I'm in a class now.)

I don't agree...
We're not children, we learn things differently.
Written input is extremely valuable, because it lets the learner see the complete sentence and spend on each word exactly the time he/she needs (which also means going faster when you are more confident).
Learning how to read also lets you use subtitles in the same language you're studying, which will both make reading itself an easier task and let you appreciate more difficult content early on.

I don't think I would be able to express things like this in Englih (a L2) if I waited to be fluent before learning how to write, since I become fluent mainly thanks to watching things with English subs (and translating lyrics I liked... I suppose I woudn't have been able to figure out the correct words without having a transcript) and chatting with people on the internet.

I'll go as far as to add that, the more different is the writing system from the one you have in your language, the more important is learning it well early on. English isn't really phonetically written, so it takes a lot of exposure to start writing and reading it correctly (even for native speakers). But I (an Italian) could probably pick up spoken Spanish without caring to much about writing (since I can already read most of it just because it's similar to Italian), and get back to tweak things only after I'm fluent.

Add to this that Japanese makes a lot more sense with kanji! Words like 火山, 花火, 会社, 社会 or like 持って行く have a lot more sense when written. Remembering how to prnounce 火山 is incredibly easy once you now that the "KA" is the same as in 火曜日 and the "ZAN" is the usual suffix for mountains, "SAN" 富士山 who sonorized.
If I didn't know the kanji I could remember the pronunciation wrongly (I don't talk about volcanoes everyday), I could say "GAZAN", "ZAKAN", I would need to repeat again and again the word to be sure I remember it. Thanks to kanji, I had to see it only once and learned it perfectly.
[Notice how the order of the kanji is perfectly logical both for Japanese and Chinese: "Mountain of Fire" becomes "Fire-Mountain"]

I could agree with you if RTK didn't existed: in that case learning 2000 kanji just to get basic literacy would be a burden. But there are people who manage to learn all of them in three monts! And picking up the reading is even easier, if you look for the patterns.

Edit: that said, it's true that languages are not their written form. But I think one can know this and still start to learn how to write them from the beginning. In the Chinese course I'm following we're doing grammar from day 1, you don't need to know hundreds of characters to do that.

Your approach makes sense, in fact, the Chinese who invented the ideograms already did that. A good portion of kanji have two parts, one part conveys the meaning, and the other part conveys the pronounciation. Many (but not all) of those parts are primitives in RTK, and are called 'signal primitives'.
For example, 可 is a primitive that signals that a kanji is read KA. If you know this, you can read a 漢語 word that contains this primitive even if you haven't learned their characters yet.
Unfortunately, things aren't so simple in japanese. For example:
- 奇 is a primitive that signals that a kanji is read KI. Notice how it contains 可 inside of it? Many signal primitives are like that due to the low number of bushu available and the limitations of the stroke system. If you don't know that 可 and 奇 are different sound primitives, you'll probably read the last one wrong.
- 可 is not the only primitive that is read KA. For example, 加 is also read KA. Notice how it contains か? That's because hiragana and katakana were an attempt to simplify the scripture by using using only one kanji/signal primitive per sound. Unfortunately, japanese doesn't click if you remove all the kanji from the equation, or use only one signal primitive for each sound.

Due to extremely poor vocabulary, japanese had to import 80% of vocabulary from chinese (which uses very short words), and due to extremely poor phonology, those words all sound the same. So, in order to make sense, the brain of a japanese speaker associates phonemes with kanji/primitives EVEN when they're talking instead of writing. Scientifically speaking: as hearing a known word (like 火山) 'fires' the neurons in the regions of the brain that store 火 and 山, which are different from the ones that store the signal primitives of close-sounding words, they are able to keep them apart even if they sound almost the same. That's also why japanese don't like reading texts in hiragana, as かざん doesn't fire the same regions in the brain as 火山 does, forcing them to consciously 'decode' the words.

Learning the japanese vocabulary without the kanji might work for small sets of words, but as the words add up and all of them start sounding the same, you'll confuse and mix them to no end. Unlike what Ephel has stated, I will argue that Japanese and Chinese cannot be separated from their written form.

Until quite recently, most speakers of Chinese or Japanese have been illiterate.

And blind people in China and Japan don't seem to have a deficit when it comes to spoken language.

Fillanzea Wrote:Until quite recently, most speakers of Chinese or Japanese have been illiterate.

And blind people in China and Japan don't seem to have a deficit when it comes to spoken language.

For that matter, deaf people in Japan still can write and read Japanese, even if they use Japanese Sign Language at their schools. Should we stop learning how to listen to sounds because you can learn a language just by reading it?

Also, are you really comparing the way a blind Japanese child learns his mother tongue to what we should do? Do you have Japanese parents willing to correct your mistakes? Because the most important part of immersion in learning a language is correct feedback, and that's what that blind child has and we're missing (obviously we're also missing the incredible brain plasticity we had when we were toddlers).
Do you really think we should forfeit one of the things that make learning easier?
You don't get anything in exchange for not learning to write.

Is it our aim to be illiterate in Japanese? If that's your aim, go on and don't learn Kanji (or even kana), but if you're aiming at learning writing sooner or later, it's better to learn it at the beginning, when you can use it to expose yourself more to the language, look at complete sentences with all the time you need to analyze them, and have an easier time learning vocab.

I must say, I hate the way RTK-fans push the 'learn all the kanji first if your aim is to be FLUENT and LITERATE. Or if you want to suck forever just go and do it some other way' line. Like there are no other options and the way you happened to do is the only and greatest way to achieve these things.

It's all very well and good to support a way of studying Japanese, but these overly strongly-stated ways of doing it that belittle other ways methods is too far unless you can back it up with some really solid evidence (i.e. scientific studies) that show that learning all the kanji first is the best and greatest way and/or that other paths lead to illiteracy.
Obviously no one is aiming to be illiterate. Your way of becoming literate is not the only way of becoming literate, and the 'learn all kanji before anything else' approach is just as radical as the 'learn the spoken language before kanji' approach, so if you're going to make such strong claims about one being superior you should present evidence that's just as strong.

I wish you guys didn't make this thread about "Kanji first vs. Kanji last". Let's just accept that both have their merits, or at least that there are people who would prefer to do Kanji last. There are people who don't like to read a lot, or who just prefer to learn a language by watching and listening to popular media (btw., I'm not one of them, I started with RtK).

Why not, for the purposes of this thread, just accept that fact, and focus on whether my specific suggestion would help the people who choose this "Kanji last" option.

To DrJones: Thank you for your thoughtful response, first of all. I'm however not suggesting that one should try to master the Japanese language without Kanji. Obviously, eventually, one has to learn the Kanji. Not gonna understand academic language or literature without it.

I'm just suggesting postponing the Kanji until one already speaks on a 3rd or 4th grader's level (understands common words, used in everyday speech), for the sake of making the journey more enjoyable. Studying Kanji early on, without understanding the fun stuff (which for most people is popular media, not reading) isn't very entertaining or motivating.

I'm in the kanji-last camp and I read a lot. It's false to say that it's only a strategy for those who aren't interested in reading - so long as you're not trying to read too high above your level. For me, kanji is still not the bottleneck to understanding - vocabulary is, so my focus is on that.
But when my vocabulary is sufficient to read the things I'm interested in that don't have furigana and the only thing that's stopping me is a lack of kanji knowledge, that's when I'll tackle kanji in an organised manner. At the moment, picking them up by exposure is sufficient for what I'm trying to do. And if not knowing all of the kanji first is impeding vocabulary acquisition, that's not reflected in my Anki statistics or any other way I have of measuring progress.

Aikynaro Wrote:I'm in the kanji-last camp and I read a lot. It's false to say that it's only a strategy for those who aren't interested in reading - so long as you're not trying to read too high above your level. For me, kanji is still not the bottleneck to understanding - vocabulary is, so my focus is on that.
But when my vocabulary is sufficient to read the things I'm interested in that don't have furigana and the only thing that's stopping me is a lack of kanji knowledge, that's when I'll tackle kanji in an organised manner. At the moment, picking them up by exposure is sufficient for what I'm trying to do. And if not knowing all of the kanji first is impeding vocabulary acquisition, that's not reflected in my Anki statistics or any other way I have of measuring progress.

Do you use mnemonics at all to learn vocab? What do you think of my idea?

I don't use any mnemonics for vocab (I replied earlier in the thread to the OP).
I think your idea is fine - I'm sure it would work for memorising vocabulary in isolation - but I think it's unnecessary and too fiddly. I find learning vocabulary using Anki (in sentences with audio) to be effective, simple, and fast. Your method doesn't sound simple or fast.
Perhaps it might have a role for absolute beginners though where i+1 sentences don't exist, but I think it would quickly become inefficient compared to other methods. Or possibly if there's one particular word that you always end up forgetting and you really need to remember it, something like this could work well.

I'm not a big fan of mnemonics in general though. I found trying to do RTK painful.

It sure is possible to become a fluent speaker without having ever touched the Kanji seriously. I stumbled over facts like "meeting must be かい" long before i met 会.
That being said, Kanji are a big help. So much so, that I'd recommend to learn them first, because vocabulary is the biggest thing to learn and will take years no matter what method, but with Kanji it's so much easier, because you have them as mnemonic tool.

I don't think anyone really thinks that you can't learn Japanese without Kanji, but nobody would recommend it, as it makes the task a harder one to achieve.

Aikynaro Wrote:I don't use any mnemonics for vocab (I replied earlier in the thread to the OP).
I think your idea is fine - I'm sure it would work for memorising vocabulary in isolation - but I think it's unnecessary and too fiddly. I find learning vocabulary using Anki (in sentences with audio) to be effective, simple, and fast. Your method doesn't sound simple or fast.

I agree that the simplest way of learning something (anything) is repetition. That much is obvious.

However, I find the claim that this simplest way is also the fastest dubious. I think introducing a level of complexity (by identifying patterns and connections between the individual building blocks) usually helps speed up the process. It certainly did with the Kanji, for me.

That's just the nature of human memory: we're not very good at remembering disconnected facts, connections help us. Even when you are learning with repetition, you are unconsciously forming connections. While I have performed no experiments to back this up, I find the position that identifying patterns overtly rather than ignoring them is what speeds things up, more consistent with my past experiences.

I think this is especially true for Japanese, in which there are so few syllables to compose words with. That makes it harder to remember the words by ear, and establishing connections between words all the more effective. Knowing the Kanji ahead of time solves that problem well (it creates connections between the words through the Kanji readings, and between the Kanji readings through the signal primitives and a few other means), but this thread is about people who choose not to learn the Kanji first.

Also, please note that I never suggested memorizing vocabulary in isolation. I simply suggested a system of mnemonics for memorizing it. That is in no way incongruent with using Anki and/or example sentences as well.

Aikynaro Wrote:I don't really see any need to use mnemonics for vocabulary. It might work, but I think it's straight-up unnecessary.

I haven't learnt all the kanji - although I know a fair few purely from exposure and that can be useful, most of the time when learning vocabulary I'm not relying on them. Given audio, a sentence, and Anki, it all works out fairly well. It's not really different from learning vocabulary from a language without kanji. I doubt my card failure rate is significantly worse than people relying on piecing together meanings with Heisig's keywords.

I think you're wrong about that. I think a well thought out mnemonic system would lower failure rates quite a bit, for the reasons I listed above.

Let me put it this way: Kanji are a lot harder to remember than words. And yet, my Kanji failure rates were always much lower, and my review intervals much longer, than for my vocab. And this despite my vocab deck containing words I might already be somewhat familiar with, or get exposed to while I'm also learning them with Anki, despite the fact that I use context (closed deleted sentences) to produce the words, and despite the fact that I do rely on remembering Kanji and its readings rather than the word, to help produce many of them.

Perhaps in your vocab deck you do recognition rather than production? That would explain the low failure rate, but don't think it helps you learn the vocab as well as if you did production.

andikaze Wrote:It sure is possible to become a fluent speaker without having ever touched the Kanji seriously. I stumbled over facts like "meeting must be かい" long before i met 会.
That being said, Kanji are a big help. So much so, that I'd recommend to learn them first, because vocabulary is the biggest thing to learn and will take years no matter what method, but with Kanji it's so much easier, because you have them as mnemonic tool.

I don't think anyone really thinks that you can't learn Japanese without Kanji, but nobody would recommend it, as it makes the task a harder one to achieve.

Nobody? I've even seen examples of people recommending exactly that in this thread.

Aikynaro Wrote:I must say, I hate the way RTK-fans push the 'learn all the kanji first if your aim is to be FLUENT and LITERATE. Or if you want to suck forever just go and do it some other way' line. Like there are no other options and the way you happened to do is the only and greatest way to achieve these things.

That's a standard debate "tactic" on here that you see all the time; I absolutely hate it. Since I've come here, I've probably encountered 10 separate reasons why I shouldn't be fluent or literate based on the way I learned Japanese.

Quote:Due to extremely poor vocabulary, japanese had to import 80% of vocabulary from chinese (which uses very short words), and due to extremely poor phonology, those words all sound the same. So, in order to make sense, the brain of a japanese speaker associates phonemes with kanji/primitives EVEN when they're talking instead of writing. Scientifically speaking: as hearing a known word (like 火山) 'fires' the neurons in the regions of the brain that store 火 and 山, which are different from the ones that store the signal primitives of close-sounding words, they are able to keep them apart even if they sound almost the same. That's also why japanese don't like reading texts in hiragana, as かざん doesn't fire the same regions in the brain as 火山 does, forcing them to consciously 'decode' the words.

Citation needed

I've read about how sometimes with acting that actors will use nothing but phonetic pronunciation from their own English language to mimic knowledge in a foreign language if their role in a movie, play, or TV show requires it. Both in America and in Japan. I'm assuming Americans would use romaji alongside a native speaker to help with accurate pronunciation, and in Japan they would use katakana alongside a native English speaker.

So learning a language without kanji wouldn't seem unheard of. I'd imagine a particular element would just involve learning vocab with romaji, consuming a lot of Japanese audio, and having conversations with a Japanese speaker or tutor.

Stansfield123 Wrote:Nobody? I've even seen examples of people recommending exactly that in this thread.

Well, they're wrong then, and I'm a walking talking example.

There is no wrong or right way of learning.
Some learn the kanji first, some don't. But knowing them definitely helps. And I'm saying this as one of the people who didn't learnd them.
You could also learn kanji and vocabulary at the same time and if you're not interested in writing the kanji it should be absolutely no problem. (that is what I want to do from now on)

But either way you choose the most important thing is just to learn daily. As long as you do that you will succeed in the end.

For me, personally, I've found that mnemonics don't really work for me in learning vocabulary -- I definitely agree that vocabulary shouldn't be disconnected facts, but for me, the connections come from context and imagination and real-world usage. Like if I'm reading Harry Potter, and I get to the part in the first chapter where all the owls are flying around in the daytime -- I imagine that happening, and that's linked with the word フクロウ in my mind, and that's a stronger mnemonic than a contrived story would be. More abstract words are harder, of course, but I think it's true for abstract words even more than concrete ones that you only learn them by seeing them a hundred times in a hundred different contexts, not just by linking them to an English (or whatever your native language is) definition.

My decks are recognition, yes. They serve my purpose perfectly well (being able to recognise a word when I see it outside of study). Perhaps production forces you to know the word better before you can mature it, but I'm not interested in using Anki for doing that (I've tried and disliked production cards). I can see that mnemonics could work for that purpose though.

For me, the connections you speak about would be the context, meaning, audio/manner that the word is said, and yes, kanji (I might not 'know' them, but the shapes are distinctive even if you haven't gone through and studied them. There are plenty of kanji I do know through exposure too though). Personally, I don't feel there's a need for another thing to connect it to.

Basically, I can't imagine a way that your system would improve on the system I have currently, for me. I already learn vocabulary quite effectively, after all - and I find mnemonics an annoyance.

But sure, theoretically what you have sounds plausible. I'm sure for someone out there it would be an ideal method, but not intrinsically better than other ways of learning vocab without kanji.