Herbo Wrote:

kazelee Wrote:

Herbo Wrote:I cant get neither sounds nor images to work with the spreadsheet. Ive been searching for whatever streaming plugin you were talking about but I havent been able to get it to work.

First make sure there is a *deck name*-media folder. Try comparing the entry in the "audio" and "image" fields of your deck with the names of the audio and image files you have in your media folder. If they don't match, what you might have to do is use the plugin to re-download the items.

OK using that method Ive managed to get the sounds working but it isnt possible with the images. They arent ripped from iKnow. Anywhere I can get them?

This has been covered before, but the images are no longer included in the plugin because resolve checked and the iKnow folks said they licensed the images from a stock photo company, and thus third parties should not develop software that procures these images alongside the audio and sentences.

For people that have had trouble with Anki crashing on OS X with iKnow's mp3 files, this has been fixed in the development branch, and a working version can be downloaded here.

I love how Geroge W. Bush's picture is next to this sentence:
お前の言うことは信じられない

I have been on iknow for less than two weeks but so far I am crazy about it. The only gripe I have is a kana keyboard can not be used for dictation. Other that that, I think I have found 日本語 nirvana.

bodhisamaya Wrote:The only gripe I have is a kana keyboard can not be used for dictation. Other that that, I think I have found 日本語 nirvana.

Please complain about that. I'm trying to get as many people as I can to complain about it. I wanted to touchtype as well, but they told me 'nobody uses that'. -sigh- If we can get enough complaints, they might implement it.

wccrawford Wrote:I'm trying to get as many people as I can to complain about it. I wanted to touchtype as well, but they told me 'nobody uses that'.

Yeah, I saw that. That was kinda lame. I think it would be a great idea, not only to implement it but to encourage people to learn and use it.

I sent an email heaping praise on the site but mentioning the one curious flaw. This was their response:
"Thank you for the suggestion! Your feedback is invaluable to us. We will discuss this as a team, and consider it for upcoming releases."

I hope others do the same. My speed using the kana keyboard has slowed considerably when I chat with Japanese friends as I make mistakes remembering which input I am using.

That the site has 200k Japanese users and utilizes only Romaji input points to the idea that Romaji input is a good choice.

However, like everything else, I like the "option" to choose. Options are great, as not everyone is Nukemarine (and sighs of relief goes out from mothers everywhere).

Do Japanese people actually prefer romaji keyboards? Why would that be if it is the case?

bodhisamaya Wrote:Do Japanese people actually prefer romaji keyboards? Why would that be if it is the case?

They do. It's far and away the most popular input method. It's really just standard. If it wasn't, I'd probably convert to kana input, but as it is, I don't feel a great need to be more Japanese than the Japanese As for why... who knows? Why do *we* prefer qwerty? Suppose it's just been done that way long enough.

QuackingShoe Wrote:

bodhisamaya Wrote:Do Japanese people actually prefer romaji keyboards? Why would that be if it is the case?

They do. It's far and away the most popular input method. It's really just standard. If it wasn't, I'd probably convert to kana input, but as it is, I don't feel a great need to be more Japanese than the Japanese As for why... who knows? Why do *we* prefer qwerty? Suppose it's just been done that way long enough.

Well, I don't prefer Qwerty (damn layout was designed to slow you down). I use Dvorak, which is almost perfect for Japanese input. Let's face it, only needing to know the layout 15 keys versus 50 keys is just easier to type. Not to mention, your fingers travel the keyboard less.

Hmmm. I guess I will go with it then. It must be the most efficient way if the majority of Japanese themselves use it. It feels like I am backtracking but perhaps I just wandered down the wrong path to begin with.

'We' was a general term. There are individuals in Japan who use kana input and even other methods as well, but that's not really the point

Anyway, qwerty wasn't designed to slow typing down and I really wish people would stop saying it was. It was designed to prevent typebar clashes, which is not the same thing; you may argue it's slowER than other methods, but the intention was not to be slow. The original keyboard layout planned was in alphabetical order, and it would have been used if not for the particular problems it caused with the system of the time, and it would have most likely been slower than qwerty. So the problem actually made a faster keyboard layout, if not necessarily the fastest. Seeing as I type around 130 wpm anyway, I'm not inclined to complain

I don't think that mass use = most efficient (unless you're into morphogenetic fields or something ;p), I think it's a matter of the history of hardware and software in Japan being often dominated by 'Western' manufacturing and usage. I've only read bits and pieces of the history of PCs and the Web in Japan, but my impression is that in regards to the kinds of development that would influence input methods, QWERTY/romaji would have been the de facto standard?

Yeah, I definitely wouldn't want to claim it was the most efficient, only that it's the most popular. Popularity rarely manages to have a decent correlation to genuine benefit. If you're already comfortable with kana input and feel that it's quick and effective, I don't really see a reason to change... unless you really want to use that iknow feature

Yeah, who knows how things'll go in the future when more support for East Asian encoding is supported in software. Here's one article on the topic of Japanese domain names:

http://domainnamewire.com/2009/02/24/how...ain-names/
(Actually, here's a two part analysis of IDNs that talks a bit about the sociology of using romaji/kanji as well as issues with web infrastructure: http://www.stippy.com/japan-tech/idn-dou...n-names-1/)

There's also the fact that in 50 years Japanese people will have 6 thumbs for ケータイ use, plus they'll have QR code devices implanted in their eyes.

I can't believe I totally missed the tidal wave of interest in iKnow. Finally did 3 hours/160 items today in Japanese Core 6000.

Some very smart aspects of the design:

1. Time Limits on replies
2. Yes/No before multiple choice - if you can think of the answer before you answer Yes/No, then multiple choice search or peeking is a "non issue".
3. Connecting reading, listening, and production: GREAT!

It certainly is more fun than SRS or mining for cards.

Of course there are some very minor annoyances, but overall, it is as if someone knew what I wanted and implemented the result. Rather than going into flame wars or discussing what is "better" or "more efficient", I'm going back into iKnow and do some more sessions...

The only thing I don't like about iKnow is that once you've finished a lesson the way you're reviewing those studied items from time to time is not very effective.
I know that some people just put the whole stuff into their Anki decks once they're done with a lesson and review there then, but I'd prefer to be able to do that in iKnow somehow. They should have a better review system, that's all I wish for ^^;

I thought I would be picking up vocabulary (I am getting about 2-4 new vocabulary per set of 10), but surprisingly, the real pick-up was in LISTENING.

Also, after doing a few hundred items, I realize now how over-dependent I am on READING. When doing dictation, it's really obvious because I understand much much better when I can see Kanji (in iKnow) vs just seeing Hiragana/hearing (in dictation).

One of the ways of improving listening is, in iKnow, I don't look at the sentence.

Instead, I just listen first to try to understand the sample sentence, and then later, look at the sentence. Seeing the Kanji helps tremendously, but I don't want to use it as a crutch which hinders my listening skills.

--------------------------------------------------------------

I haven't used iKnow long enough to tell if the repetition interval is any good (apparently, many people cite this area as a strong defect in iKnow), but for now, I'm not concerned.

I just want to juice this resource for all its worth!!!

For example, Brainspeed seems more like a game than a learning tool, but in fact, it has helped me sharpen up on the vocabulary I don't know... ...it becomes quite obvious which vocabulary are weak after a session in Brainspeed...

Yes, it has a lot going for it. It's the long term issues that got me to move to Anki plus it looses one of it's strengths in time: listening review. When you get "complete" the review changes to showing the entire sentence in kanji, and you pick out the english meaning of the bold word. Gave way too much away, you could not look away as you would not know which word was being tested unless you blurred the sentence (very annoying).

Anyway, should you decide to import everything you've done into Anki latter, it's very doable. I did iKnow for about 2 months until I imported all 1400 items I had done (1200 complete), spaced them over 2 months (core 1 and 2 got 48 to 64 day spacing while core 6 got 1 week for example). The ability to do vocabulary dictation and reading helped show which ones I were really weak on.

PLUS, in time there may be a way to sync Anki (well, 3rd party applications) with iKnow, but no telling how far away that will be. Should that come about, you can get the best of both worlds: Options and proper SRS with Anki combined with Features and Word progress across any type of item list with iKnow.

I recently imported Core 1-3 into Anki for my long term studying. As I hadn't been keeping up with my old stuff on iKnow, I'm having to re-learn a few words... But they're easy because I've learned them once already. Right now, I just use it from my phone when I have a few minutes.

I think it's working rather well.

The listening is a major plus for me as well. The vocab in step #1 and #2 is already familiar to me but the listening stumps me. When the example sentences come up I will turn away and listen, repeat, then translate it into English before reading it from the screen. I don't know if any of these sentences repeat later on, but if not, then I will have had practice repeating and translating 6000 sentences by the time I finish core 6000. Tomorrow iknow is shutting down for major improvements from what the message at the top of the page informs. Who knows what iknow will become by the end of the year? It will only continue to improve with the lists users are putting up along side the gradual improvements iknow itself seems committed to.

5AM here in Tokyo and the maintenance on iKnow just started: says it will be down most of the day.

I have been doing:
1. Core 2000 #3
2. Core 6000 #1, #2, #3

Mainichi Shimbun has a list of 5,500 words ordered by frequency of appearance - the Core 6000 seems to be ordered similarly (in the sense that very similar words are showing up in order).

I skipped #1-2 in Core 2000, but #3 in Core 2000 does seem to be much easier than anything in Core 6000.

Someone wrote earlier that Core 6000 seems to be in between JLPT2 and JLPT1 - I totally agree with that assessment.

You can pass JLPT2 without knowing all of Core 6000, but it's useful to know as many as possible. However, it's probably not enough for JLPT1.

If I ever make a frequency list for vocabulary, it would probably be like this:

Scan ch2 for frequency list. I forget where, but there are programs that do that. Now, if someone can think of a better site (or sites) that can be scanned for word frequency, I'm open to it.

Next, bunch the vocabulary lists in groups of 1000.

Now the tricky part. Organizing these bunches into more learner friendly order. The Kanji words I think should follow the KO2001 order. The trick is to share them between kanji's. Example would be 水泳 where there's words for 水 but none for 泳, then 水泳 would go near where 泳 is at. For all kana words or words that don't have ko2001 kanji, they are spaced evenly through the list.

Repeat for each group of 1000 words.

My reasoning: KO2001 does not list full kana words, especially english loan words. In addition, you're having to learn 3000 words based off the first 1000 kanji before learning any other types of words. I think you're missing some very common words due to this.

On the other hand, a straight out frequency list is not user friendly.

By combining the best of the two, I think you'll have the best way of: Learning common words, reinforcing On and Kun readings, learn faster thanks to thematic groupings that KO2001 order provides. Ch2 sounds like the best place to scan for words as it's real people, talking about stuff they know about, using words they actually use.

I'd say 2ch is a bad place, just like 4chan would be a bad place to get an English sample. Lots of slang, lots of specific discussions, lots of idiocy. Maybe if you actually controlled which parts of 2ch and only took samples from the serious boards it could work, but I doubt it. They have a very special way of talking on 2ch, it isn't really comparable to real-life discussions.

Interesting interview with the CEO of iKnow:

They know about SuperMemo's algos - they attempted to improve on it - here's a description that I found:

--------------------------------------------------------

Based on years of applied research, Cerego has built adaptive, web-based applications that accelerate knowledge acquisition. Cerego’s patented core learning engine is driven by algorithms that generate optimal learning schedules for discrete chunks of declarative learning content, called “items”. This intelligent scheduling is achieved by gathering metadata on individual user performance and modeling memory decay patterns at the granular level of every item.

Cerego's unique approach to learning has been to bring together, within a unified system, what has previously been a broad and scattered collection of well-researched and accepted scientific principles in the fields of cognitive psychology, educational psychology, and neuroscience regarding efficient learning patterns. Through this work, Cerego has created a learning methodology that maximizes the effectiveness and efficiency of learning, memory, and performance.

The well-documented Ebbinghaus forgetting curve illustrates the decline in memory retention over time—there is a predictable and measurable rate at which you forget. A method of learning based on this insight, spaced rehearsal, has proven that there is an optimal moment to review what you have learned. Reviewing a piece of knowledge just before you forget it is the best way to remember it. Reviewing something while it’s still fresh in your mind, or after you have completely forgotten it, is inefficient. Cerego’s learning algorithms calculate this moment for the individual and adapt in real time to according to a user’s performance on an item-by-item basis.