2.5 Year UpdateWell, it's time for another update. The main change over the last six months is that I've been spending a lot more time practicing reading. I've been reading a lot on Satori Reader, but also did some reading on
Shousetsuka ni Narou. I also started studying non-WK vocab on JPDB, replacing my previous attempts at Anki. And I've continued spending a lot of time on Wanikani and listening to podcasts in the background for hours a day, as I've been doing for the last year.
In addition to continuing to practice reading, one goal I have for the next six months is to start learning how to
speak Japanese as well. Since I'm self-studying and only concerned with receptive skills, I haven't bothered to practice productive skills at all over the last 2.5 years, apart from going to a weekly Japanese meetup for a while last year. But I do think it'd be nice to get conversational in Japanese as well, so I intend to try out Italki lessons starting in August.
My main problem is that I have a very low tolerance for making mistakes or not knowing how to say something, which makes it difficult to start speaking. I know that there's some Youtube Polyglots who film themselves harassing random strangers as soon as they learn how to say "hello", but I'm the opposite. I'm not outgoing at all, and even the thought of taking private lessons 2.5 years in is very intimidating. Anyone have any advice for this?
JLPTSix months ago, I took the 2012 N3 practice test
here and did surprisingly well on the test, except for taking much too long on reading and not even attempting the grammar section. So I assumed that after six months of further vocab study and reading practice, I'd easily be able to ace the practice N3 test. Unfortunately, the exact opposite happened, which was a huge blow.
Six months ago, my scores were as follows:
Vocabulary: 28/33
Reading and Grammar: N/A
* Grammar: skipped
* Reading: 13/16
Listening: 17/27
* Task-based comprehension: 4/6
* Point-based comprehension: 5/6
* Summary: 3/3
* Utterance expressions: 3/4
* Quick response: 2/8
When I took
the exact same test yesterday, my new scores were:
Vocabulary: 23/33
Reading and Grammar: 25/39
* Grammar: 12/23
* Reading: 13/16
Listening: 14/27
* Task-based comprehension: 3/6
* Point-based comprehension: 5/6
* Summary: 1/3
* Utterance expressions: 2/4
* Quick response: 3/8
I just don't understand how I could do so much worse on Vocab after six additional months of study. It seems like I must have never known it well in the first place and just gotten much luckier on guessing the first time around.
As for listening, I've always done worse on the practice JLPT listening tests than I would have expected, given the amount of listening immersion I did. Sometimes it feels like the question format is a bit artificial, particularly "utterance expressions". However, I am definitely disappointed by the Summary section, since that's the one place where I would expect to do best, as it involves understanding an entire conversation (so more context) and then answering a question about it. However, this time around, I only managed to fully understand the dialog on question #2 (plus the practice question) and just had to guess on the other two (and guessed wrong both times). I wish I knew whether I understood them better six months ago or just got lucky on guessing. One thing that feels really punishing about the listening section is that even a slight lapse in concentration or a single unknown phrase can make it impossible to answer the question.
The new section was
grammar, which I didn't even bother attempting last time. Grammar has always been my weak point on the practice JLPT tests. I wish I knew a way to study it. I was hoping that I would just naturally pick it up from immersion, and that did happen somewhat - I was actually surprised how many questions I got right! I felt like I was basically just guessing on the whole thing and there were only one or two questions I was confident about until the end.
The final section of the grammar test was interesting - it consists of an essay with five blanks you need to fill in. This seemed more up my alley since there is more context, although it was still grammar words that needed to be filled in and it was still hard to guess. However, I did manage to get four of the five right in this section anyway. (Which means I got 8/18 on the rest of the grammar section.)
The one clear sign of improvement was in reading speed, although even there, I was disappointed that I didn't improve more. The real JLPT gives you a combined 70 minutes for the grammar and reading sections. However, when I took it six months ago, I skipped grammar entirely and spent 81 minutes just on the reading section. This time, I spent 39 minutes on grammar and 60 on reading. So I did speed up a fair bit, though less than I expected. I was also disappointed that I didn't get more questions right on reading - I got the exact same
score as before. I don't have my answers from before, but I wouldn't be surprised if I even got the exact same questions wrong with the exact same answers. One of the downsides of taking the same test multiple times is that you tend to get misled by the same question in the same way each time, so scores are a lot more resistant to improvement than you might naively expect.
WanikaniIn my first year of Japanese, I raced through Wanikani as fast as possible and hit level 60 in only 353 days, then abandoned it when my subscription ended. However, I soon came to realize that I had forgotten a lot of what I had learned, and that I never learned most of the higher level material properly in the first place, so in May 2021, I resubscribed and began my second Wanikani challenge.
I resurrected everything and then set myself to slowly whittling down the resulting pile of ~6500 reviews a second time. Initially, I tried to do it without any resets, but after a month or two of not getting anywhere, I gave in and reset to level 46, where I've been stuck ever since.
Since starting WK the second time, I've done
39096 reviews in a little over a year. For the first few months of the challenge, there was little sign of visible progress. However, in the last six months, my review count has fallen increasingly rapidly. My review count is now below 2000, down from a post-reset high of over 5300!
There are multiple reasons for this acceleration. Part of it is that I'm actually getting better over time and starting to remember the words more. The increased accuracy also feeds into increased review speed, since I spend a lot less time looking up missed answers and trying to figure out how to distinguish words/kanji/etc that I keep confusing.
However, the most significant cause of this pattern is a natural consequence of the SRS levels. At first, most everything was at Apprentice, and even reviewing it correctly would only punt it eight hours into the future. However, over time, with tireless reviewing every day, more and more of the items have reached the higher SRS tiers, and thus take longer to come up for review again (or eventually stop coming up at all), and so my review count would go down increasingly quickly, even with no other changes. Here's a graph of the number of items in each SRS category over time:
Satori ReaderI actually started SR way back in fall 2021, but I've been putting more emphasis on it lately. I set myself the goal of reading every story on SR, and then ending my subscription and moving on to reading free web novels on syosetsu.com. I've already finished Kona's Big Adventure, Akiko's Foreign Exchange, Hole in the Wall, Meditation, Sakura and Suzuki's Long Distance Relationship, and The Jam Maker, and am currently working on Oku-Nikkou and Kiki-Mimi Radio.
That being said, I do think that SR is really important as a bridge. I've learned a lot from the grammar and culture explanations, and being able to instantly see the (manually curated) definition and reading and translation of everything is really nice.
JPDBLast December, I experimented with Anki for a while, but Anki is a pain to use. Fortunately,
JPDB fills the same space while being much nicer and having better features, etc.
I started JPDB in March, and initially went through the N5, N4, and N3 decks, before starting the いつも通りの日常で、、君からすれば deck because that was the web novel listed on JPDB with the fewest unknown words. I've been very slowly going through that ever since. I was at 75% words known when I started and am currently at 91%, although a lot of that was just from blacklisting words. I've also started occasionally adding new words I encounter in the wild to my "General" deck on JPDB (63 to date).
Despite being much better than Anki (talk about low bars!), JPDB was pretty miserable at first. After aggressively blacklisting words (particularly anything grammatical, but also sometimes homographs, words that seem too hard, etc.) reducing my max review session to six minutes, it has gotten better, but it is still a slog. One complaint I had early on was that there were a lot of misparses in the example sentences, but I haven't noticed any recently, so either they fixed that or else I just started blacklisting the kind of word that is likely to have misparses (like grammar stuff).
In a six minute review session, I typically manage to go through 14-15 cards. It's actually remarkable, just how consistent it is, since you'd expect it to vary a lot based on luck when a lot of easy or hard cards come up at once (though my last two sessions were just 12 each). Everybody else says they're doing hundreds of reviews a day on JPDB - I'm wondering how much of it is just them pointlessly reviewing easy cards a lot. A common suggestion I've seen is to auto-fail any card which you can't recall instantly, which seems like a recipe for just endlessly churning leeches. Besides, theoretically you're learning best when something is hard to recall.
I've noticed a pattern where typically, I'll get my review pile down to 0 on JPDB and do a few new cards for a day or two, and then suddenly get overwhelmed by the reviews and not manage to get it down to 0 again for many days (and thus do no new cards). Most recently, I haven't done any new cards in the last week, and I don't think I'll manage to get the review pile down today either.
Meanwhile, my
new card pile is up to 500 from a low of 481, thanks to a) not doing any new cards in the last week, and b) adding some new words that I encountered in the wild. (Most recently, I added しっくり , which I learned from watching Komi Can't Communicate on Thursday night. I actually added several new words from the movie Bubble last night, but then removed them again because I don't want my new card pile to go above 500
).
Edit: In my second JPDB session today, I must have failed a kanji or something, because the new count went down to 499, so I went ahead and added uzu from Bubble to bring it back up.
Anyway, I think JPDB has a lot of promise if it continues to improve in the future, although I do have a few gripes. One is how you can't control the readings that appear on kanji cards. For example, I added 白杖 (after seeing it on Satori Reader - The Jam Maker), but the kanji card for 杖 on JPDB only lists つえ and づえ as readings, despite it being more natural to learn onyomis first anyway. Also, the kanji cards mostly don't have mnemonics, but I guess I can't complain much about that, since that would require WK-style curation and you'd probably have to come up with your own mnemonics for them to be any good anyway.
Another gripe is that JPDB doesn't give much sense of progress when doing reviews, so it is hard to stay motivated. Going from say, 90% to 91% words known can easily take weeks, so there's no sign of progress in the interim. It would be nice if a) values were shown in 01.% increments instead and b) have a little green popup like WK does whenever a word becomes known during reviews.
Lastly, I wish it was easier to get them to add new decks for web novels. Apparently, there's a process set up to add most forms of media, but not web novels, so they almost never get added.