My Complete Journey, Reflection, and Advice for Achieving a High Reading Level in Japanese

Providing picturesque examples with enough quantity gleaned through real media that perfectly matched the target vocabulary and grammar of a lesson would be an exhausting process, and the end result would still be boring and yield less reward than the effort required.

I was thinking more along the lines of custom PRODUCED media that was tailored around target vocabulary and grammar. Voluminous graded reader style material in an appealing and engaging format, perhaps with an in-built SRS functionality, instead of tacking on SRS as an afterthought.

That sounds awfully similar to what Satori Reader is attempting.

4 Likes

And Lingq and a growing number of services. Custom produced level-oriented reading/sentence pattern engagement is where the future of language learning is heading, IMO. Many of the people I’ve directly spoken with in the industry, and many other services that discuss their philosophy mostly mirror the sentiment that sentence patterns are the most natural and efficient method for acquiring grammar and vocabulary. While immersion in-the-wild also achieves this, it isn’t optimized for second language acquisition. So you’ll see more and more of these “hybrid study-oriented native immersion services” popping up.

4 Likes

I think this is a very good idea, but I don’t actually think it would be more efficient, and you specifically asked for something that doesn’t sacrifice efficiency.

If you’re going to try to create content entirely based around content for beginners, the fact that you have to make all the sentences fit together into one coherent and (preferably interesting) story is a very big constraint. You already don’t have a lot of freedom with what you can do from a language perspective, so the result would be something looking like a very slow toddler aimed show. Considering downtime, necessary compromises for consistency, and general lack of interest in the result for a lot of people…I think a highly motivated learner would outpace it significantly just going through sentences at their own pace.

3 Likes

If you were going by JLPT or 教育漢字 levels, certainly. Their levels are somewhat arbitrary and don’t match the actual experience of a native Japanese language learner (Japanese children already have an extensive vocabulary and are just learning to associate Kanji with it). Most children have a pretty strong understanding of basic potential and causative statements well before they enter the 1st grade, but those are deferred to N4 in the JLPT for some reason, despite being critical components of speech. So you’d basically be needing to combine everything from absolute beginner to pre-intermediate in order for it to have the expressiveness needed to be compelling and provide enough challenge to not immediately lose value upon the first 1-2 uses.

The まるごと series does a fairly good job at this, and that may be because it’s based on CEFR levels, but the first two volumes do also suffer the “immediately obsolete” problem.

I was just pointing out an observation I had made that there seems to be an intersection between research and results that many instructional materials/services are converging on that is based on a philosophy centered around sentence patterns. I think we’ll continue to see optimization of 2LA as new theories emerge seeking to better integrate and implement the techniques that we know do work (and I think there is still MUCH room left for optimization).

1 Like

Hey, I came back to this thread because I remember you mentioning an Anki deck in the video, and I have a couple of questions (1 related 1 not)

About the Anki deck, I just got it imported and… the first 20 lessons had stuff like する,こと,いる… uh, how much of the content do I already know do I need to slog through to get to the good stuff? Am I wasting my time with this deck? I’m sure there’s bound to be stuff in there I don’t know but, damn I might have to get through like 500+ cards by time I get there.

About leaving WK at level 40, say I go along with your stance on WK and how the last 20 levels are worthless, what am I to fill the void of time with? If WK from 40-60 is inefficient, what is the efficient replacement? Am I using this as more time to read and study my personal Kitsun.io deck?

IME it’s faster to go to the edit-the-deck UI in the desktop Anki app, and then you can scroll through the list pretty quickly and shift-select all the already-known very easy words and suspend those cards, assuming the deck is sorted in frequency order. Then when you’re using the deck and encounter way-too-easy cards you missed, suspend those too so they never appear again.

There’s probably also a script or plugin somewhere that can cross-reference two decks and suspend all the cards you already have elsewhere.

3 Likes

Just a small tip with Anki, the stuff is ordered, right? Just browse the deck and select all the stuff you know then delete it. I don’t know exactly how much stuff you will or won’t know because I don’t really know your level apart from the wk level 35.

Assuming you have done the core deck and know basic grammar:

As for what you do, I mean it really just depends on what you want to do content wise, but the core idea will be consume->mine->srs on repeat. I personally think visual novels are a great way to get started with that process and prime material for learning for reasons stated in the video, but the important part is that you’re engaging with what you like. If you like anime, I think that’s perfectly fine too and would recommend starting with JP subs. If you are going to use anime, though (or even vns…or anything really) if yomichan doesn’t have kitsun integration somehow I would consider not using kitsun. You can make the mining and srs process incredibly smooth, fast, and high quality thanks to all the amazing tools people have made for Anki. I’m not sure what kitsuns like on that front, but if they aren’t compatible I would suggest reconsidering. You’re going to be making countless thousands, possibly tens of thousands of cards, and doing hundreds of thousands of reviews in your journey to becoming a high level reader. Opting to do those things less efficiently for a significantly lower quality result is a big sacrifice.

Of course, maybe kitsun as added something and it does support all the tools.

5 Likes

Thanks.

I think I’ll stick with kitsun, honestly. It has a feature to pull and create entire cards from Jisho.org. I used to use Anki and it was a nightmare, hence why I switched. Very happy with that choice, and have already made 400+ cards since I started. I don’t mind creating a card taking 10 extra seconds, lol. How many lessons do you think is preferable per day? Since watching the vid I went with the 20 you talked about, but I don’t recall you mentioning ever doing more than 20 a day? So is 20 good? Or do you recommend doing more? Of course if I ditch WK I’ll double those lessons regardless.

It’s not so much about the time as it is about the quality of your cards. Like I said, youre spending more time on something of “significantly lower quality”. Even ignoring stuff like images, the sentence where you got it, isolated word audio, sentence audio, pitch accent graphs, and etc., Being able to use monolingual definitions is a big advantage. Not to mention tying a single English word or phrase to japanese words when you’re training your recall of those words is… suboptimal to say the least. It’s a big reason I kept asking @Raionus for an Anki style mode on floflo. Not sure if it ever got added tho.

But yeah I did 20 cards a day for the most part. For some parts of my journey I did more. Recently, even if I read for awhile it’s not a guarantee I’ll even find 20 words Ive never seen/seen and forgotten so nowadays it’s less than 20 on average. It just depends on what’s sustainable given your level of effort, time, and general ability to learn new info I guess. I wouldn’t recommend doing more really, but I dont think it’s a bad idea to do less. You can learn words outside of srs too, so you can add 10 cards a day but still be learning more than that on average if you’re immersing plenty.

And fwiw, I used to use Anki too and I hated it and quit. Then I took time, set it up, and copied an experienced users interface and setup. It was much better after that.

4 Likes

I think a better question would be how many lessons/reviews per day did you find yourself doing during the pre-advanced stage where you’re just trying to sponge everything to break through to advanced?

1 Like

Again…advanced is like an iffy word since to me, advanced means that you can read your books without a dictionary but other people have different ideas. However, luckily in my case I was so consistent that the answer doesn’t change really. 20 words a day.

I had maybe between 150-200 reviews per day?

3 Likes

What is your opinion on doing many more reviews than that (if there is genuinely a large enough knowledge gap to justify it)?

1 Like

I don’t think there is an issue with doing more reviews necessarily. When I did all the kanken 1 words I was doing like…maybe 500 a day. It’s just a matter of if you can handle it mentally.

The problem is what doing more reviews implies. If you have 300 reviews a day, it’s probably because one or both of the following

You’re doing more than 20 new cards a day
Your retention rate is significantly lower than mine was.

Both of those are not good imo. 25, honestly sure whatever, but like more than that per day seems like a lot. Poor retention is also a problem and it’s only going to get worse. Now, of course remembering 50% of 10000 words is still better than remembering 99% of 5000 words technically, but just the time you’re putting into it is not ideal.

If you’re ok with your review quantity and time (and ffs put on back to back reviews please), then I see no issue with having more reviews provided you’re retaining your information.

4 Likes

As an experiment I tried doing 55 new words per day. Reviews were exhausting, retention was 45% at the start, but interestingly, doing that every day, the retention rate started slowly increasing. The brain adjusts somewhat. And like you say, x% at a higher volume can be better than a higher y% at a lower volume, if time efficiency isn’t a concern.

1 Like

I dunno anything about that addon, I use 10reader (because I’m on Firefox). While it (Yomichan)may not have Kitsun integration, Kitsun has Anki integration. It’d be a bit of a thing, but, it should work out to create the same quality cards, though with a slight increase in time taken. Also, you mention pitch accents n’ stuff. Uh, those help you read? Or did you assume I had a interest in speaking/listening? My fascination with JP is strictly with reading, tbh, though if I ever wanted to I don’t think I’d be at all opposed to learning the other aspects.

Update:

Just added that core deck to Kitsun. Can confirm it’s literally a 1:1 match of what Anki would of provided with that same deck.

I mean short term retention is important, but I’d say the real important factor is how well you know those words after maybe a month. Once you’ve done it for a month, if you haven’t already, maybe you could report back.

2 Likes

I’m not talking about the core deck, I’m talking about making your own cards. Do the 2k on whatever you want.

Pitch accents are how the word is pronounced. You don’t have to add pitch accent graphs if you don’t want to. There are only pros though, and really no cons. As for audio, it will still help your reading. Multimodal learnings effectiveness has been proved well with science. Besides, I also was only interested in reading for several years until I mover to Japan. Things can change. Even if it doesnt, it’s really just pros all around and no notable cons.

3 Likes

That actually sounds like a fun experiment. I’ll give it a try. I’ll start the N4 JTango deck. Do you have recommendations for a simple % correct Anki stats display addin? I tried a couple but they weren’t working for me.

I don’t, actually. I don’t really track those stats anymore, but I’m sure any old addon will do if it’s just a stats display. Even anki has built in stuff.

1 Like