Congrats erlog!! You're amazing ^^! I'd love to know how you stripped the info from your rom into anki too! Kinda actually makes me want to go download the rom now instead of using my actual game lol...

Do you think you're going to continue working on 準1級 and 1級 now or take a little break from it? I've been trying to push my way through some of the prep books I got a little while ago but I just can't seem to find the same motivation for them as I had for the stuff up through 2級. Hearing the progress you're making though sure does make me want to try a bit harder myself here . Thanks again!! And looking forward to hearing about your results come Monday~!

I was surprised how simple it was to extract the questions from the game data, and was kicking myself for not trying it sooner. I typed about 14,000 of those questions into Anki manually over the past 2 years.

The data format is pretty simple. There's a Shift-JIS-encoded binary file that's about 2 and a half megs in the game data that contains all the regular test questions. The beginning of the file is an index. First there's a list of how many sections there are along with how many questions they contain. Then following that there's a big list of pointers to all the question records.

The question records are a bit wonky, but not difficult to figure out. There's some metadata to denote question type, the question, and then a pointer to the answer. The data's not encrypted, but it is lightly compressed. The questions that all have the same answer all point to a single instance of that answer to cut down on file size. As long as you're following the pointers around it's not really something you have to worry about.

It's also pretty easy to figure out which section is which because they are in the order the game displays them in for the Study section. So it starts with 10級, and then goes on up.

The only complicated piece is figuring out how to massage the data you get back to make appropriate Anki cards. I ended up using Mecab there to add automatic furigana since the only pronunciation the game gives you most of the time is just the hiragana for the target kanji or compound.

The game also does some of the questions slightly differently depending on the test level, and so some stuff you think would be the same question type ends up being split into multiple types for use in the game. So there's some trial and error going on where you need to pull out the raw records, look at them to see what's going on, and then write a handler for that question type.

There's also 2 answer pointers that lead to slightly different answers in some cases because for instances where a kanji is repeated the game will accept either 時々 or 時時. Which pointer is the one you would want to use for Anki isn't really consistent, and so that requires some trial and error for each question type.

That weird bug with the game not telling you if you should be putting in the synonym or the antonym extends all the way to this data, though. At this point, I haven't discovered any place in the record for those questions where that might be denoted. So when I added those 190 synonym/antonym pairs I had to sort them manually into the proper category so I could use them in Anki.

The 読み and 書取 questions are pretty trivial to pull out, and that's probably what most people need to drill anyway.

Thank, I figured out how to get the data file, but I can't figure out what program to oppen it with. Notepad++ shows a bunch of unreadable gibberish before after and inside the questions. What did you use to read the file to pull the data?

It's a big ball of binary Shift-JIS data/metadata. You're not really going to be able to just open it up in Excel or something. If you have a hex editor that's encoding-aware you'd be able to glimpse some of the questions from the Shift-JIS, but byte length of Shift-JIS characters vary from character to character. So the hex editor isn't going to be able to discern what's a Shift-JIS character and what is metadata since it's all just a big ball of binary data.

I wrote a python script to pull out the questions/answers using the pointers in the question index, and used the section index to sort which questions were which type/which level.

It wasn't super difficult to do, but it was a bit time consuming. I'm not going to post my code because I haven't bothered to refactor it now that I've figured out all the question types, and it's like 500 lines of spaghetti.

If someone else wants to write a tool for other people to use based on my description of the data here then I would be okay with that. I just don't have the time to clean up my code, create something that's convenient for other people to use, and deal with all the support questions that would generate.

I also have some moral misgivings about releasing something directly because my script relies on the user having a rom. I know I purchased the game legitimately, but I can't be sure that everyone using my tool has done the same.

Well let me just say I would be VERY interested in the results of such a project

Ah. Alright, thanks anyway. I don't currently know python, so I don't know enough to code up a version for myself, and I already have my plate full with tasks I'm currently learning.

Which Kanken level is closest to/a little higher than JLPT N1? The test site is too far for me to take the real thing, but I have some Kanken quiz apps, and I thought it might be a fun thing to go through, since there are no more JLPT levels I can take.

Um, KanKen is a different skill set altogether. So while the level of vocabulary for JLPT might be KanKen 3級 level, you'd probably have a lot of trouble trying to take that. JLPT is primarily a reading and literacy test. KanKen is a vocabulary test.

There's only like 5 kanji pronunciation questions on JLPT1. On KanKen there's 30. There are 0 kanji writing questions on JLPT. On KanKen you're looking at about 50 questions where you're writing kanji based on a sentence and a reading given to you. On JLPT there's zero questions about kanji radicals. On KanKen there's usually about 10 of those. On JLPT there's no questions about okurigana. There's about 5 of those on each of the KanKen levels. On JLPT there's very very few 4-character idioms. On the higher levels of KanKen you're looking at 10 of those along with 5 more questions where you have to match the meaning. On JLPT there's no questions about stroke order. On KanKen there's about 10 of those on the lower levels of the test.

I started seriously studying for KanKen a few months before I passed JLPT1. The first level I had to study for outside of KanKen DS3 in order to pass was 5級, and then I started taking the official test at 4級. Even that required a few months of study, though, despite the fact that it's only about 1300 different kanji that I had previously already studied many times through RTK/Kanji in Context/reading practice.

To give you an idea of how different KanKen is from JLPT1. It took me a few hours of study to finally pass the lowest level of KanKen, the one aimed at first graders, because I was unfamiliar with the kun-yomi readings for some of those basic kanji.

KanKen tests a much deeper and thorough understanding of kanji and related vocabulary than anything you would ever find on the JLPT.

So I would suggest starting at the very lowest level of KanKen, and working your way up using something like KanKen DS3. If you're as solid as you think you are then you should be able to pass through those "easy" levels no sweat. If you're like me, though, then you're probably going to be surprised how different it is.

I cannot recommend KanKen DS3 enough. It is pretty much exactly the same as taking the real paper test. Every time I have taken the test my scores on practice tests in the game have mirrored my real score almost precisely for all the levels I have taken. In cases where I wasn't passing the practice tests, I also failed the real tests with a similar score. In cases where I was passing the practices tests, I also passed the real test with a similar score.

It's a really great value, and a really great way for those of you outside Japan to experience KanKen.

Congrats on passing ２級, erlog! Your posts about the test have been very helpful and inspiring. I took ７級 in person yesterday and I'm pretty sure I passed, in part due to the great info in this topic.

If no one else is working on it, I'm interested in making a tool to dump all the questions from the DS3 game. I'll put the source on github or something so others can see the progress or modify it for their own purposes.

I've done this sort of work before but if erlog could post some info about the compression format and the hex locations of pointers etc it would be very helpful so I don't have to figure it all out from scratch.

I don't know if I passed 2級 yet. I should get the results today, and I'll post again when I do.

The NDS cartridge format has already been solved, and individual files can be dumped out using 7zip or something else. So once you're looking at the files it's pretty clear which one is the file you want based on the hints I gave before.

There is no compression on that file. Beyond the section indexes and question indexes it's all just flat Shift-JIS strings that are prepended with their length.

Thanks, your information was enough for me to make a very basic program to dump parts of the text, which is mostly good enough for my purposes. Here's a sample of the output and the program itself, for any programmers who would like to improve it. http://pastebin.com/pPKwymMA

I passed 2級! My 2.5 year KanKen adventure is over! I might eventually try to tackle 準1級, but at the moment I really need to concentrate on business vocabulary and grammar usage. Next goals are getting level J2+ of the Business Japanese Test and trying to get 150+ on the N1 JLPT.

I'd like to thank Mebzup and all the others in this thread for their support. I literally could not have done it without their advice, and most of all their attitude. Seeing people who talk about taking/passing KanKen like it's no sweat motivated me quite a lot. It took something that felt like a dream, and made it seem graspable.

The support I got in this thread from people reading my posts also spurred me on. I know this probably seems a bit melodramatic, but I feel really great. This level of the test is described as being for Japanese high school graduates, Japanese university students, and general Japanese adults. This is a big thing for me since I have always tried to measure my language progress against that of actual Japanese people instead of the skewed scale that gets offered to foreigners in Japan where being able to have even minor conversations is seen as some kind of big victory.

This really feels like something finally worthy of the respect I sometimes receive from Japanese people.

Wow, congratulations ! Now go frame that paper or something :p

The certificate it came with is huge. It's like 15 inches diagonal, and bigger than all the certificates for the previous levels. They couldn't fit it in my mail slot, and so I had to go pick it up directly from the post office.

Seeing the "Failed Delivery Notice" in my mailbox from the KanKen Foundation with the note "non-standard size" made me pretty certain I had passed, but man it's nice to have the score sheet.

Congratulations! I'm curious about the scoring...it says passing score is 155+. I took 2-kyu on Friday, so I won't know for about a month how I did, but I was under the impression passing is 160+ (80%)...is it really just 155, or is it weighted for each test or something?

Congrats on taking 2級!

It is 155. That rounds off to 80% in your head pretty easily, though. I thought it was 160 for a long time too.

Oh, I forgot to thank Flamerokz in my big post up there. He helped me ruin my life with Anki by convincing me it was a good idea to do 1000+ reps in a day to catch up.

Ah, well that would seem to move the math more in my favor, then. It sounds like we took the test for about the same reasons...I started a year and a half ago with 4, after passing JLPT N1, and have moved up level by level from there, and it is inspirational to see someone else make it. It's been great for vocab acquisition, and I've relied heavily on the official Kanken books (especially the STEP books) and Anki to advance. I really recommend those books (I actually haven't played the DS game).

@erlog&co:
Congratulations! Thank you very much for your updates, learning strategies and guidance. In the same way that you were once motivated by the progress of others, you now also serve as a motivator for others yourself!

It seems like it'd be interesting to tackle KanKen myself at some point in the future. However, for the time being I feel like there are more essential areas where I'm still severely lacking, so I'll try to address those first. Active output comes to mind...

Congratulations for your success!

Why not keep on moving forward from now on?

Even just 10 minutes a day would take you eventually to beat 準1級 and 1級, unless you're not interested in those at all.

What resources are available online for preparing for 漢検, besides the oficial samples and renshuu.org? I've seen some Anki decks, but the ones that look useful are for 1級 only I think.

Has anyone here beaten 1級?

Congrats erlog!!! If it wasn't for you I wouldn't have had nearly enough motivation to try for kanken myself! I'm sure you've inspired a ton of other people to push themselves and actually get objectively better at Japanese as well! Good luck on all your other endeavours too!!

Sebastian Wrote:Why not keep on moving forward from now on?

In the longer term I will keep moving forward, but in the shorter term(the next 6 months) I really have to stop focusing on Kanji Kentei in order to spend time focusing on studying grammar and business vocabulary. Grammar is a big big hole in my Japanese ability right now. I haven't really studied it since like 2008.

It probably sounded more final in that post up there, but I'm really just saying I'm gonna lay off Kanji Kentei for at least a year after this. It's mostly for R E A L L I F E reasons of needing to finish my résumé before my contract finishes in August of 2015. I'm gonna be job hunting next spring, and I want to have things in order.

I have JLPT1, and now I have KanKen 2級. I need to increase my score on the Business Japanese Test by a lot, and I also need to improve my real ability in that area. Right now I'm J3 level of the BJT. You can read about the levels here:
http://www.kanken.or.jp/bjt/english/abou...guide.html

I want to increase to J2 by the time of the November test, and then J1 next year. I think this is definitely doable if I start applying Anki to grammar and business Japanese vocab. I've spent the last 2.5 years progressing rapidly by throwing Anki at KanKen. Now my kanji skills are fine. My general vocab is fine. My reading is fine.

It's time to start using Anki to fill up the holes in my Japanese skills so that I can finally spin my Japanese skills into a good job that takes advantage of these skills. Once I have a stable job again I'll probably try to tackle 準1級, but right now at this moment it would be a waste of time.

Grammar has been my white whale for a long time because I find it really difficult to study and remember all the finnicky usage stuff. That's kind of why I started studying for KanKen. Studying for KanKen is just flat memorization, and I'm used to that. Studying grammar takes more brain power than that, and that's really annoying.

tl;dr: I want to spend the next 6 months focusing on business Japanese stuff that will help me get a job next year.

Yeah, I think around jun-1 and 1 you start really hitting the law of diminishing returns. I bought the jun-1 book, and while there are many kanji that I think I've seen in mostly older literature, I'm not sure it'd be worth the effort if there's still other more practical areas of ones japanese ability that could be improved.

I love you erlog-sama you are an inspiration.

Got back my 2-kyu results yesterday:

読み (27/30)
部首 (10/10)
熟語の構成 (18/20)
四字熟語 - 書き取り (20/20)
四字熟語 - 意味 (10/10)
対義語・類義語 (20/20)
同音・同訓異字 (18/20)
誤字訂正 (6/10)
おくりがな (10/10)
書き取り (38/50)
TOTAL (177/200 [PASS])

My study method mainly involved SRS-ing the vocab from the Kanken step books. I had three decks - vocab, 4 character compounds, and radicals. After I SRS-ed vocab from a "step" of the book, I would write out each word by hand a few times and then complete the review section for that step. After I completed the step book (took about 3-4 months; I took 2-kyu about 5 months after Jun-2-kyu), I moved onto one of the other official Kanken books for the final month that has everything broken up into sections for each section of the test for review, as well as the previous tests book for benchmarking my progress. I also had an iOS app called "Kanken Puchi Drill 5000" that I used for review. In the final month of intense review I was able to move my score from around the 110s to passing, although I outperformed my practice results on the actual exam. I started with 4-kyu in January 2013, and it's been a completely grueling year and a half of study to reach this point. Think I'm going to relax from testing for a while

Congratulations, you did it faster than I did. You also got a better score. You should rest for a while.