***EDIT: AJATT works, what I'm writing about is what I did differently to make AJATT work to supercharge my listening skills. I'm saying this cause a couple of posts criticized this post without them actually reading it***

Ok, I’m throwing this out there even though I’ve not done it for long. This will be the longest post I’ve made on RTK, so I apologize. However, I think it’s important.

**Why I say AJATT doesn’t work (for my listening skills)**

I’ve ripped audio from Dubbed Movies, Blogs, Newscasts, and J-Dramas, put them on my iPod and have played them fairly consistently for about two years. Aside from very basic benefits, I have not seen it increase my listening comprehension. Didn’t matter if I liked the show or songs. I’d remember the scene, but could not follow the actual dialogue. Something is wrong in the AJATT.

Thinking on it, it’s akin to wall papering with a bunch of Japanese magazines, books and newspapers that you like. If you don’t know the kanji, the words and a bit of the grammar, it’s a mess. For those that have studied, you begin to get patterns from the chaos. RTK brought out the kanji, Genki/Tae Kim/UBJG brought meaning the various particles and modifiers, KO2001/iKnow/KiC brought meaning to various words. But with that, the pages of text on my walls still contain garble mixed with bits of clarity.

Now, imagine if instead of random magazines and books that you like, you papered your walls with print outs of Kanji from RTK, and various sentences and words from your Vocabulary deck? Would everywhere you look be something familiar that you knew? I’d argue your mind will stop your gaze at the familiar here. The big problem is all that crap on your walls is soulless sentences with no connection to each other.

Ok, take that one step farther: You mined a manga or a short novel and figured out each and every sentence. You SRS to ensure retention of that knowledge. Now paste the pages of that manga or short story (in big print) on your walls instead. If you happen to scan the wall, you’re seeing stuff that makes sense and follows a narrative. Your brain remembers not only the sentences due to the SRS, now it’s getting training in following a larger story.

Now imagine consistently adding more and more pages as figure them out via your SRS. Not only that, but you paper not just your walls, but every wall you walk by through out your day. I think you’ll find that reading skills will improve via passive AJATT part. Problem is it’s completely impossible outside of your own walls (and iPod touch mangas).

**How to make AJATT work (for my listening skills)**

One simple program – subs2srs.

Yeah, people are using it to get sentences to mine for vocabulary. People are having it automatically trim out sentences that are all kana, aren’t long enough, contain known words, etc. People throw it in the SRS and delete sentences that look too easy (when read), don’t have new words, don’t seem fun, etc. I am not doing that. What I started doing 80 days ago has changed my thinking, or maybe more along the lines of refined my thinking on the matter.

**EDIT: I made edits to these steps to reflect changes since I posted. There replies in this thread that might mention things I changed after the fact**

Step 1: Get an episode I like and convert it via subs2srs but I don’t use English subtitles (might be useful if starting out though). I think the best type starting out is a Drama with a dramanote script or a anime that follows a manga fairly close for later reading benefits.

Step 2: Import into Anki. I set up so the fields are: Japanese Expression, Japanese Furigana, Definitions, Audio, Image and Notes for drama/card number.

Step 3: Display cards as Question: Japanese Expression (without furigana), Image and Audio (*1). Display card as Answer: Japanese Expression (with furigana), Audio again, Definitions, notes..

Step 4: Line by line, card by card, go through Anki and figure out the drama. I’m using EBwin and the Kenkyusha dictionary J-J and J-E myself. It’s not just knowing what the words mean. It’s about knowing why that sentence is being said. That means figuring out the phrases. I put this information on the definition field. If I were starting out, it’s likely all J-E definitions, but no big deal. (*2)

Here’s the kicker: I don’t just delete a card because it READS easy. If I can’t really understand what the actor is saying, it’s probably better to keep the card. I do delete um’s, yeah, and other small inconsequential stuff that I can use yourself with no problem. I am wary though, and ask myself “if I heard this randomly, would I understand it”.

Step 5: Review as they come up. It’s a matter of knowing what is being said/written. Yeah, I'm looking at the Kanji expression, and probably replaying the audio but that’s okay. What is passing here is comprehension of the sentence. If something doesn’t feel right, like there’s more to the sentence, research it more. Maybe there’s a phrase that if I did a “full search” it’ll come up and give deeper meaning to the sentence (another reason I should not be so quick to delete a card). I only mark wrong when I mess up the meaning of the sentence or don't know individual words. I mark it difficult if I don't get the pronunciation of words correct (this can happen often even with audio).

Step 6: When I made the subs2srs, I ripped the full audio of the show. With audacity (I can also use subs2srs), I like to rip the audio into 3min 30sec segments. I put it on my iPod on sequential or random. (*3)

What I noticed with the above:

1. I watch the show after doing this, and had close to 100% comprehension without subs.
2. I read the dramanote script with pleasure and speed.
3. It was fun listening to it on my iPod.

Now, back to my title. For months I’ve always had my iPod playing dozens of hours of drama rips either in my room or on the move. Usually, I just phased it out. Even at night, I’d go to sleep and wake up hearing Japanese. Yeah, I’d recall the scene, but understanding of dialogue was not there. AJATT was not working….

Except in the last 3 months. Audio from the dramas I did with subs2srs are now in the mix. When these come up my mind notices. My ears perk up and I’m following along. I can remember the next line about to come up. I can at times quote at length portions of the dialogue. Even better, dialogue from other audio is beginning to make sense. It wasn’t over night, but it becomes noticeable very fast.

AJATT does work, but it works best with stuff you understand at 80%+ comprehension. It’s was not about reading comprehension, as I think I read ok. I don’t think it was just SRS’ing the sentences. Hell, I’ve read dramanote scripts with good comprehension, yet audio from that same show on my iPod was still white noise at times. It wasn't about vocabulary, as I have about 3000 words well in memory. No, it feels that it’s having the writing and the audio on the subs2srs deck that allowed me to train the bits (sentences), which were put into a bigger whole on the iPod (the dialogues).

Case in point: Rookies. That dialogue was fast, full of slang, and kind of got tedious (I stopped at the hour mark on the first episode due to this). But oddly enough, a month goes by with me reviewing it on in the SRS and playing in my iPod and this show is getting results. I can recall lengths of dialogue. I can understand that annoying 川藤先生 and his odd fashioned hot-blooded way of talking.

By using subs2srs, you train comprehension ability. The listening is trained on the iPod, where the comprehension from the srs is getting reinforced. You get by-product training in reading, typing, pacing, accent during this though.

Listening is by far the most important skill in a language. To get the most from the immersion portion of AJATT, that immersion material is best that you almost fully understand. To prevent a plateau, you must consistent add from a variety of shows and anime. To get better benefit, let it be a show or anime with a written version (dramanote, scripts, manga, etc.). Hell, paste your walls with these pages and walk around with them at the ready. In addition, with trained listening, it’s easier to produce correct audio because you know it sounds right. You’re shadowing the drama perfectly at the pace of the drama (not just one sentence at a time in a SRS). Heck, try doing J-Drama Karaoke with the audio down and you reading the sub-titles out loud. If you can think at the pace, then it’s no stretch to be able to read and type at near that pace in the near future.

As the hours build up, you mind has a firm grasp on the basic to produce the unknown with equal ease.

Ok, this has been lengthy. It’s poorly written and probably missing useful notes to further explain it.

*1 – Use anki to display the kana field as Furigana to neaten up the question field.

*2 – I use separate decks for Kanji, Vocabulary, Grammar and this Sentence mining. If new words (and kanji) come up in sentence mining, I add them should they exist in RTK1 and 3, or in the Core 6k and Tanuki corpus. Yeah, not every new word is added but so far no problem. That sentence mining deck is there ONLY to get comprehension of the drama. **EDIT - Due to recent releases of sorted Core 2k/6k sentences, I stopped activating words from subs2srs in my vocabulary deck. I now just use the definition field if needed**

*3 – I’m hoping the iPod will have a random feature soon where songs with higher stars get played more often. This will allow me to play random audio of both drama’s I have not SRS’d and those that I have. **EDIT - I now just create a new playlist for shows I've subs2srs in addition to a show I haven't on a 1 to 1 basis.**

So in summary, all you're really saying is, if you don't understand what you're listening to, it's meaningless to listen to it? (Or rather, you won't learn much from it).

I've said that all along.

It's nice that you found a way "around" it though. Question is, will this new way really help you all that much? It sounds more like you memorize a show then you can listen to it... won't that mean that you will never be able to listen to new shows, just shows you've already memorized?

Kind of an insulting summary. Where are you getting meaningless from? Maybe my problem stemmed from depending heavily on ripped audio for my listening training instead of watching more dramas with Japanese sub-titles. Maybe it was reading a lot of English forums and websites more than the less comprehensible Japanese ones. I doubt it, but it's there.

The immersion is important. Equally important it seems to me now is getting more and more of immersion material you comprehend into the mix (and earlier than what I did). Cause, although I didn't say it above, I'm noticing better comprehension of the non-srs'd audio. Now, is this because I built up my vocabulary? It helps, but I've been building my vocabulary quite a while with little headway on listening and speaking.

So, I'm not really memorizing the story. Just the nature of the system brings that about in parts. It does not mean I won't be able to watch new shows. It means that as I get more skilled at comprehending shows I know (accents, slang, pacing) that will carry over to shows I don't know. I hope it carries over in how I talk, and how I write. By adding to my Kanji and Vocabulary decks, it will even carry over to how I read (and further improve how I write).

My first one hour show took up about 250 lines out of 500. It took about 10 hours (I posted about this in the milestone thread). My second show took about the same effort spread over a longer period due to other factors. Similar with my third show.

So that's 3 hours only. I'm wondering what it'll be like six or ten or twenty shows down the line. I won't have to keep as many sentences so the SRS will begin growing slower. I'll have more random audio and shows that I fully comprehend but coming from different aspects of life, gender, occupations and social levels. Plus I can listen to my iPod a hell of a lot easier than sitting at my computer and watching a TV show (I'm still in Africa after all). Time will tell, hopefully for the better.

Thats cool nukemarine, you found a setup that works for you. I use subs2srs in a slightly different way which suits me better. It is great that more people are noticing the awesome possibilities of subs2srs

I prefer editing the tsv in notepad++ instead of using anki, because if I remember correctly anki would run slowly on my laptop when I fiddled with whole a shows subs2srs output.

This is an interesting thing to do with a show you really love, because it'll just be ridiculously easy/fun to go through the whole damn thing and gain 100% comprehension.

Nukemarine Wrote:Kind of an insulting summary. Where are you getting meaningless from?

Dont be surprised at this dude, remember the thread I made where I detailed how I use subs2srs? same insulting assumptions..

Nukemarine Wrote:Kind of an insulting summary. Where are you getting meaningless from?

From the fact that you said you've been doing it for 2 years and it isn't working?

Tobberoth Wrote:Question is, will this new way really help you all that much? It sounds more like you memorize a show then you can listen to it... won't that mean that you will never be able to listen to new shows, just shows you've already memorized?

This implies that it is useless to SRS sentences from one source because it is impossible to use them in another context. Is that what you are trying to say?

nonpoint Wrote:

Tobberoth Wrote:Question is, will this new way really help you all that much? It sounds more like you memorize a show then you can listen to it... won't that mean that you will never be able to listen to new shows, just shows you've already memorized?

This implies that it is useless to SRS sentences from one source because it is impossible to use them in another context. Is that what you are trying to say?

No, not really. I'm just thinking that since the problem here is listening comprehension and the point of listening comprehension is to pick stuff up from audio as it goes, can it not be quite ineffective to simply train using sources where you already know exactly what is being said? Does that really train the area where the problem is?

I don't know, I personally don't have any real problems with listening comprehension anymore so I can't say for sure... if I understand the sentence being said, I can usually hear it perfectly on my first go, and from what I understand, this is the skill Nukemarine wants to train up. And if that's the case, I'm just wondering if this kind of training will work.

My guess is that this kind of training will help to improve general listening comprehension. The more you listen to the same understandable dialog the faster you get at processing it. I can't help but think that if you cover a wide enough range of audio like this that eventually you will just have a good all around listening ability.

Nukemarine, the title of this post is quite inviting of a flame war
Listening a lot works pretty well for me, I sometimes even pick stuff up from context (which always feels so damn good)

How many hours per day do/did you listen to japanese?
Its not really AJATT you are saying didnt work for you(wtf is ajatt anyway ), it is the input hypothesis you are having trouble with, I think.

Tobberoth Wrote:I'm just wondering if this kind of training will work.

Good point, neither you nor I can know what will work for nukemarine. I guess he'll just have to try and see, whatever works == awesome.

Quote:Its not really AJATT you are saying didnt work for you(wtf is ajatt anyway ), it is the input hypothesis, I think.

Well, I think the input hypothesis says you need to get understandable input, so it seems Nuke's approach is pro input hypothesis. AJATT advises just getting a lot of input even if you don't understand much.

AJATT also advises listening to stuff and picking out stuff you want to understand and then learning it with SRS, and by doing so you increase what you can understand. IIRC Nuke, you were mainly SRSing sentences from learning resources, right?

If I'm clear on what you're saying, it's that you're finding positive results from 'planned redundancy' of subs2srs materials? http://www.supermemo.com/english/ol/ks.htm#Redundancy - Well, to be honest, what I took from this article, selectively interpreting it, was that it was possible to study familiar/recently learned materials outside the SRS with augmenting effect rather than negatively impacting the spacing algorithm, by shifting the frames of reference.

(Other good links here: http://forum.koohii.com/showthread.php?p...7#pid38977)

I pretty much don't bother with passive immersion (despite the token recommendation of getting used to hearing the language) outside of 'watching media that I want to watch that just happens to be Japanese'. I have a kind of attentional 'porous membrane' where if something clicks, that's fine, but I only 'plan' non-SRS input based on whether it primarily consists of materials I've studied in the SRS.

Basically this is a sort of generalized variation of when we prime ourselves for an episode or film by studying it in subs2srs first, no? Actually, you've just reminded me of possible iPod/playlist/Audacity uses, thanks--I think I don't need to wish for card-chains in Anki anymore.

Also, are you really finding good resuls from typing in Anki? While good practice for typing Japanese, since there's not bottom-up gestural memory at work for kanji-kana, I don't see how it could be too useful and haven't tried it yet. I either just write stuff out by hand or if I know it well I don't bother.

vosmiura Wrote:

Quote:Its not really AJATT you are saying didnt work for you(wtf is ajatt anyway ), it is the input hypothesis, I think.

Well, I think the input hypothesis says you need to get understandable input, so it seems Nuke's approach is pro input hypothesis. AJATT advises just getting a lot of input even if you don't understand much.

Wikipedia:"...one of the most prominent modern theories in the field of Second Language Acquisition (SLA) is the Comprehensible Input Hypothesis, developed by Stephen Krashen. This theory was originally called the Input Hypothesis."

We need to stop calling it input hypothesis and start calling it the Comprehensible Input Hypothesis. The old name is makes one discount just how important i+1 is.
I wonder what I can do to go much more i+1, this seems to be the most important thing. Can't believe I never actually looked up the the Comprehensible input Hypothesis on wikipedia... this is so embarrassing.

Nuke - I desperately want to try this out, but my technical skills are nowhere. is there any possible way I could get my hands on an example deck here?

koyota Wrote:Nuke - I desperately want to try this out, but my technical skills are nowhere. is there any possible way I could get my hands on an example deck here?

The thread: http://forum.koohii.com/showthread.php?tid=3236
has a bunch of decks on that wikia wiki-page. I liked the nobuta deck, I f-ing LOVE nobuta wo produce.

Tobberoth Wrote:So in summary, all you're really saying is, if you don't understand what you're listening to, it's meaningless to listen to it? (Or rather, you won't learn much from it).

I've said that all along.

It's nice that you found a way "around" it though. Question is, will this new way really help you all that much? It sounds more like you memorize a show then you can listen to it... won't that mean that you will never be able to listen to new shows, just shows you've already memorized?

"I've said that all along."

LOL. Classic. Mr Tobberoth you crack me up.

I suggest removing all kana later and trying to transcribe what's being said.

This is essentially taking the sentence method and applying it mostly to the ear. So yeah, it will help (better than passive listening). The problem I see is taking chunks that are too large, then you've only got a general idea of the passage with little knowledge of the individual components. Also, you should randomized the cards to boot.

I see your point, and agree to a large extent.

Listening to Japanese audio input doesn't improve my comprehension if I don't understand most of what is being said. However, if I do understand most of the dialogue, I find that I pick up on new things.

That is what I have modified the audio-input method to use a combination of J-E input. Think about it, if you take a difficult lecture that you are completely unfamiliar with, that is spoken in your 1st language, and listen to it passively for 5 hrs, 10hrs, 25 hrs, etc., you will eventually learn/memorize/understand it.

That is why I use Japanesepod101 Intermediate/Upper Intermediate lessons. Most lessons follow the same format, dialogue at native speed, dialogue w/ line-by-line translations, explanation/commentary. I put 2-3 lessons on replay, listen passively, and after a while I understand most of what is being said.

Every now and then, I mine an old lesson for sentences.

Anyways, my point is that I use Jpod for listening comprehension NOT reading comprehension. I don't learn the meaning via sentences. By the time I make sentences I already know what 90 or so percent means, but I have no idea how to read the new kanji I encounter. However, when ever I hear similar spoken patterns/vocab I usually recognize it in almost any context.

Nukemarine Wrote:AJATT does work, but it works best with stuff you understand at 80%+ comprehension.

Personally, I disagree. Because when I started to watch movies and listen to Japanese music I scratched my head so often I was getting headaches. Now I do not.
Well, people are different.

Guys, try to remember we're all coming from different backgrounds, mindsets, job settings, home settings, prior Japanese exposure, etc. I'm doing what is currently working for me. I was just impressed at how well it was working for. Hence the slightly inflammatory title with parenthetical caveat.

See, I'm in the military, so that limits my ability to take a year off to go to a school in Japan like Tobberoth. I can't walk around with an iPod 24/7 like Khatzumoto. I found jPod boring unlike oregum. I'm not fully fascinated with anime like Masamo. However, I'm able to play my iPod a lot, and I've been playing these on speakers in my room which is similar when I was in Japan.

I've been listening to rip audio from Dramas I watched, as I did enjoy the shows. So there was a little comprehension on what was going on. It just wasn't having the effect despite it being pleasurable. And I'm not the type to re-read the same damn thing over and over despite having easier understanding reading. So, I've no problem listening to the same stuff over and over, but won't read it over and over. The way above practices a little reading and listening to a certain point (during SRS), but the main point is I know what that whole sentence means and why. It's that I get LOTS of listening practice via the iPod. So Kazelee, it's not the sentence method for audio. My listening comprehension didn't improve with audio, cause I didn't notice it with my iKnow sentences.

What I am doing is four different decks. I don't need to go straight audio to typing because I do audio to handwriting on my Vocabulary deck. I don't need to worry about Kanji accuracy cause I have my RTK deck. I was thinking of removing the Kana later on down the road like Kazelee was suggesting, but now I'm doubting that will be necessary. The real training will come from the iPod with the SRS in the supporting role.

The typing portion is just a step up in addition to my writing training via the vocabulary decks. I also do that on my Grammar deck, but that's down to like 20 reviews a week. Even my wife is noticing an increase in my typing speed (though my grammar is horrible to be sure).

The speaking portion may come from shadowing or reciting the dialogue at length, which is something that used to come only from singing Karaoke (compare how I did reading iKnow sentences versus singing Ashita Hareru Kana back in March). In fact, just to set a reference point, I'll do the intro to Zettai Kareshi and post on the link on the pronunciation evaluation thread (hmm, may use J-Drama karaoke as a term from now on).

I don't think anyone can claim this is an intuitive idea and always been out there. Before it was mine sentences from dramas or books. Or use audio of the sentences. Or use SRS for dramas to find new words. I think I'm doing something different, or at least what others haven't said they're doing. With AJATT, he says listen or watch it till it gets boring. This ain't that either, as I could have watched with subtitles lots of times (and have). It's the audio equivalent of telling someone to sentence mine the first 2 chapters of Harry Potter, then constantly re-read them while you sentence mine first book of 20th Century Boys, then re-read those while, etc... But I just cannot see doing that with a freaking book. It's a process easier to do with listening with an iPod and it enforces the listening skill which is most important and most lacking for me.

Again, it's early. I'm just throwing all this out there. I'm the type of guy that likes to measure my progress, and I noticed this happening in the process.

Also, sorry for being so hyper on the boards lately. I'm flying back to Japan this Monday and seeing my wife for the first time in six months. Lots of energy right now, lots of ideas going through my head and no other place to vent.

You guys are looking at the sky and missing the big picture. How the hell do you think children learn? They take classes, rip dramas to their Ipods?

If you think listening to a target language is a waste of your time. You should really have your head examined. Cause its the same logic as saying actually Going to the country will do NOTHING for your language acquisition.

Yes Tobberoth you are a linguistics master. [rolleyes]

Draak Wrote:You guys are looking at the sky and missing the big picture. How the hell do you think children learn? They take classes, rip dramas to their Ipods?

If you think listening to a target language is a waste of your time. You should really have your head examined. Cause its the same logic as saying actually Going to the country will do NOTHING for your language acquisition.

Yes Tobberoth you are a linguistics master. [rolleyes]

I'm not a child, I don't want to spend 10 years to learn the language decently. It's a HUGE fallacy that it's a good idea to learn a language exactly like a child.

And thank you for calling me a linguistics master, i'll just ignore the rolleyes comment since you seem to lack reason to be sarcastic, especially since I didn't say listening to a target language is a waste of time.

i read as much of this post as i could handle (about 30 seconds) and i`ld like to recommend a couple of books that has helped my listening skills.. these books are called Read Real Japanese. One of them contains contemporary writings (essays) and the other contains short stories. The best part about them is the CD included with them that reads through both books in about an hour each at full speed Japanese. You might want to read the book at night or whenever and listen to the cds on an mp3 player or something when you are out and about... when i 1st started listening to them i couldn`t understand hardly anything and now since i listen to them every week i notice more and more my ability to follow the stories welll... it might help you out some...

http://www.amazon.com/Read-Real-Japanese...798&sr=8-1

http://www.amazon.com/Read-Real-Japanese...798&sr=8-2

Hashiriya Wrote:i read as much of this post as i could handle (about 30 seconds) and i`ld like to recommend a couple of books that has helped my listening skills.. these books are called Read Real Japanese. One of them contains contemporary writings (essays) and the other contains short stories. The best part about them is the CD included with them that reads through both books in about an hour each at full speed Japanese. You might want to read the book at night or whenever and listen to the cds on an mp3 player or something when you are out and about... when i 1st started listening to them i couldn`t understand hardly anything and now since i listen to them every week i notice more and more my ability to follow the stories welll... it might help you out some...

http://www.amazon.com/Read-Real-Japanese...798&sr=8-1

http://www.amazon.com/Read-Real-Japanese...798&sr=8-2

Hey Hashiriya, if you're interested in learning Japanese via text + audio, you should look into a site called smart.fm--formerly known as iKnow. There's also something called 'shadowing', try using a forum search here.

Yeah, ah, most of those here aren't children. I know the whole 'learn like a child' thing is popular around here, and there is truth to it when speaking of certain things, but honestly, children are so ridiculously different in learning style. This becomes most apparent when you try to teach children XD


As Nukemarine mentioned, and tried but I guess failed to emphasize sufficiently in his posts (and title), he's largely talking about himself and not you. Everyone has their own circumstances. I know people who have achieved success through rather wildly varying methods.
His post is placed up for consideration, and that's what it warrants.

is that a joke nestor? (as i put together the majority of the KO.2001 lists on smart.fm) i was referring to those books because they cover complete stories not just random sentences as on smart.fm

Hashiriya Wrote:is that a joke nestor? (as i put together the majority of the KO.2001 lists on smart.fm) i was referring to those books because they cover complete stories not just random sentences as on smart.fm

Yes, a joke. Seemed like you ignored Nuke's lengthy comments just to recommend some stuff the 'veterans' like Nuke have discussed before, repeatedly. I know I get frustrated when ppl pop in to ignore what I said and recommend something to me like that, so I figured I'd vicariously vent for Nukemarine w/ my joke to you. Nothing personal. ^_^