I’m writing this post to recommend resources and approaches that I have found valuable (or believe would be valuable) to those who can relate to my interests and preferences as a learner.
In December 2023, I passed the JLPT N1 with a score of 172/180. I offer this fact as a gauge of my qualifications, but neither my learning journey nor my recommendations here are optimized for performance on the JLPT. Put another way, I would have done little differently had the JLPT not existed.
# My profile as a learner
To put into context my recommendations below, you may want to have in mind my profile as a learner. This section is to outline that.
## Motivation
My motivation for learning Japanese can be broken down into two main aspects:
1. Intellectual reward: To learn a language is to learn a new mode of thinking, especially for a language like Japanese that is constructed so differently from the other languages I know (English, Spanish, learning Mandarin). Trying to figure out a sentence and succeeding is fun.
2. Entertainment: Many of the works I consider masterpieces were originally written in Japanese, and experiencing them in Japanese makes them even better. Experiencing new content without having to pass through another language is fun too.
Both of these aspects are invaluable to me, but then again the time it takes to learn Japanese is ridiculous too, so all in all, I think the two aspects together were necessary to get me to put in the investment.
Since the intellectual aspect of Japanese is one of my primary motivations, I do not enjoy consuming content that I can hardly comprehend. In addition, since entertainment value is another one of my primary motivations, the more highly I think of a work, the more I insist on having the patience to experience it only once I can satisfactorily comprehend it. By comprehension, I don’t just mean having a good idea of what a sentence means, but also understanding why the words, grammar, and context come together to create that meaning. In other words, if an oracle gave me the meaning but I could not explain why the Japanese source carries that meaning, I would not be satisfied. (Note that input can be made comprehensible by external means, such as a tutor or a lookup. Learning can also occur even if you understood without consulting external resources, because you accumulate exposure to the distribution of acceptable usage.) Some incomprehensibility is inevitable when learning a language, but I prefer to minimize it. At the same time, I do not want to be bored. Without a tutor constantly at my side, a big part of my journey and what I’ve written in this post is choosing resources and ways of learning that are efficient and enjoyable.
## Input and output
My interests and circumstances led my learning to be centered on input much more than output. I only started working on output after becoming proficient at understanding Japanese. I would recommend this order if you, like me, don’t have a good use case for output but would still find it nice to be able to communicate when visiting Japan and such. (More recently, though, my plans to study in Japan for a semester have turned my focus toward output.) I don’t believe there to be any particular synergy between learning input and learning output, so it won’t be less efficient in the long run to start by focusing on input and then use that proficiency to accelerate your output.
I speculate that it can be easier to stay motivated if you first focus on input, because if we think about input and output as passive and active ability (cf. passive and active vocabulary), it’s natural for your passive ability to be stronger per time spent, which means you’ll feel more rewarded for your efforts. Regardless, I recommend of course doing what you feel motivated to do.
## Formal learning vs. self-study
I prefer self-study to classroom learning; I self-studied Japanese. After I had become proficient at understanding Japanese, I signed up to take a 4th semester Japanese course at my university (skipped the first 3). However, the pacing and content of the course led me to conclude that my time was more productively spent self-studying, so I dropped the class. A class can encourage you to study, but I think as long as you have the motivation, self-study lets you personalize your learning to be the most fun and productive. My main reason for signing up for the class was to get feedback for my output, but even in this aspect I think what a teacher can offer you is limited in comparison to a 1-on-1 tutor or a language exchange partner.
## JLPT
Since my motivation for studying Japanese is intrinsic and not for practical purposes, my learning journey has been steered by my interests and not by the JLPT. You may then wonder why I took the JLPT. The reason is that my own proficiency goals aligned well with the skills required for the test: the linguistic maturity to comprehend general-domain Japanese. So I took it on as a fun challenge to see how high I could score. My main preparation was taking the two official practice tests online. I realized after registering for the test that the certification would come in handy for qualifying to take classes taught in Japanese in Japan.
# Resources
* [**Genki**](https://www.amazon.com/GENKI-Integrated-Elementary-Japanese-English/dp/4789014401) 1, 2: Beginner textbooks.
* My journey began with Genki, back when I wasn’t sure whether to commit myself to learning Japanese. In my taste, these books had the right balance between laying a thorough foundation to Japanese grammar and staying approachable to beginners. I was drawn in by the way Japanese is constructed, at least at the surface I was exposed to, which led me to decide to continue with learning the language.
* [**Satori Reader**](https://www.satorireader.com/)
* I dare to say that Satori Reader is the most irreplaceable resource for learning Japanese. With a foundation from Genki, Satori Reader took me to a point where I was comfortable learning from native media. The stories are voiced, translated, manually parsed into words, and annotated with teaching moments. The stories are interesting too (some favorites were 恋人 and 隣人), compared to most native media and especially compared to what is comprehensible at this stage. This is the ideal environment to aim high and fall back on translations and explanations for comprehensibility and learning.
* Since there is audio to go with the text, if you’re using the \[text + audio → meaning\] cards described later, Satori Reader works perfectly as your immersion (i.e. consumption of Japanese content) before entering native media.
* The full version is paid at $9/month, but the free version is extensive enough for you to decide whether to get the full version.
* [**JPDB**](https://jpdb.io/): Vocab [SRS](https://en.wikipedia.org/wiki/Spaced_repetition), known for its media decks.
* The media decks (and frequency decks) are the characteristic feature of JPDB. Basically, a bunch of media has been mined for you (but without context/images/audio). However, the reason I didn’t make much use of these decks is that they’re best used if you tell JPDB about all the words (JMDict entries, to be exact) you know, since otherwise you’ll get cards you already know. JMDict has lots of entries that I would not consider words that are worth SRSing, so I made little attempt to give JPDB a complete listing of what I know, and this gap only grew as time went on. Many people seem to use the media decks though, so it may be less tedious than I make it sound to mark any words you don’t want in your rotation as “never forget.” Another cool feature (but again one that is most useful if JPDB knows what you know) is coverage stats (i.e. percentage words known) on media decks, to help you find comprehensible input.
* JPDB has a feature to attach i+1 sentences to your cards when available, and even to randomize the sentence. Nearly all words and translated sentences have audio, generated by a good TTS. Saying the word aloud also helps for remembering the reading. However, for complete beginners, it may be better to use a premade Anki deck instead of JPDB. JPDB’s sentences may contain misparses or be confusing. These account for \~1% of all sentences, but I believe that at the beginning, being able to count on simple, clear sentences with direct translations and human recordings is important for getting the basic usage of a word and getting used to how Japanese sounds. JPDB lets you [import](https://jpdb.io/anki-import) your progress from Anki decks, but it has limitations so you should try it on your deck and see what happens.
* A Patreon feature is the mpv plugin, which lets you look up words and mine the context when watching a show. There is a [browser extension](https://github.com/max-kamps/jpd-breader) that lets you mine from the web.
* It used to be widely attested that JPDB’s scheduling algorithm is superior to Anki’s, but now with Anki’s FSRS I think it’s no longer clear.
* What I myself have gotten the most out of JPDB is a convenient way to do **audio cards** as a building block to listening ability. Audio cards don’t seem to be that popular, so let me explain them. As opposed to “text cards,” which test \[text → meaning, reading\] to train reading recognition, that most people use, audio cards test \ to train listening recognition. (I will not specifically address sentence cards, but what I say pertains to them too because people who use sentence cards usually retain the format \[text → meaning, reading\].) For this discussion, I will assume that you value reading and listening ability about equally. (If you hardly care about listening ability, I suggest considering the switch to \[text → meaning\] cards without testing the reading.) Consider a person who only does audio cards. People might say that she’s neglecting her reading recognition by not doing text cards. The thing is, the reverse can be said about the majority of Japanese learners, who only do text cards – that they are neglecting their listening recognition. Building on this, there are three approaches to vocab SRS that I’d like to advocate for. I’m also interested in hearing your experience if you’ve tried any of them.
* **Text cards and audio cards**: This is the approach I took. The reason is the same as why most people do text cards – so that I could recognize words (whether by reading or listening) as they appear in books, shows, etc. Consider JPDB’s anime decks. Most people would study them with text cards, but if you think about it, this is a bit silly because you don’t read anime. The ability to recall meaning and reading from text is quite different from the ability to recall meaning from audio.
The main reason people would be reluctant to use this SRS approach is that it takes too much time. First of all, your SRS time would not quite double because reviewing text cards would reinforce the matching audio cards (and vice versa) to some degree – the degree to which the two abilities in the last sentence of the previous paragraph are correlated, if loosely. Still, it’s true you would spend more time on SRS, but your recognition of spoken words would improve that much – it’s the same reason why people do text cards.
In fact, I contend that listening recognition plays a more crucial role in listening ability than reading recognition in reading ability, because, especially in non-interactive settings, speaking speed is controlled by the speaker and it’s often inconvenient to pause/rewind, whereas you can read at the pace that you’re comfortable and lookups are more convenient. Of course, the ability to parse real-speed speech is another skill that must be developed by authentic listening practice, but understanding the individual words is a prerequisite (again, same for reading). SRS is not necessary to achieve this prerequisite, but it adds efficiency.
A justification for only doing text cards is that reading content that has accompanying audio (such as anime with subs) allows you to lean on reading ability and gradually but painlessly raise your listening ability. From this perspective, only doing text cards isn’t so much neglecting listening recognition as it is preparing your reading ability to transfer more easily to listening ability. This point has merit, and it brings me to the next approach.
* **\[text + audio → meaning\]**: I’ll call these “easy cards” because they’re trivial in comparison to either text cards or audio cards. The benefit of easy cards is that you spend embarrassingly little time on SRS (with an adjustable/adaptive scheduler, review load can be traded for difficulty) while still learning to recognize the words. The concern of course is that you rely on both text and audio to be present to reliably understand. The solution is to choose your early immersion to include both text and audio, such as Satori Reader, show/YouTube with subs, visual novel with voice, book with audiobook, podcast with transcript. This way, your reading and listening abilities rise until you reach a point where you’re comfortable in one medium alone. My geometric analogy is that if the improvement speed of a single ability under isolation is 1, then the speed of each ability done together is 1/√2 ≈ 0.7.
* **None**: When motivating both text cards and audio cards, I brought up the concern that adding audio cards would take too much time. It’s a valid point, but it also prompts the question, “From the perspective of someone who does not SRS vocab at all, wouldn’t it take too much time to add in text cards?” In other words, if you are omitting either text cards or audio cards, you should consider omitting both.
There seems to be an implicit assumption that a serious Japanese learner must use SRS. The no SRS approach is a step down from easy cards in terms of efficiency, but it lets you focus on immersion and eliminates with SRS the parts of it that are hard to get right but that we forgetfully bear with. To name a few: What is the best order to learn words in? There are premade beginner decks, but they can’t be personalized to the content you will consume and be in an order conducive to intuitive world building (e.g. “wallet” and “keys” appearing together; this aspect is not addressed by JPDB). How do you determine whether you know a word to decide whether to pass it? Words are ambiguous. Requiring yourself to recall multiple meanings without context is an artificially more difficult task than real usage. Testing whether you could understand a word in a sentence taken from a limited sample of sentences risks inability to generalize.
There are probably a lot of people who would enjoy their time learning Japanese (or another language) but are kept off by the impression that it requires using SRS, and in a specific way. I hope to have advocated against this impression.
* Let me overview the technical side of implementing audio cards in JPDB. I can follow up with details if there is interest.
* Many people mine sentences with audio in Anki, but I wanted something I could SRS directly without having to mine. JPDB doesn’t officially support audio cards, but you can implement them yourself without much difficulty by hiding the text from the card front using local overrides in Chrome or with a userscript. (The sloppy way is to just not look at the screen.) The audio is generated from a high-quality TTS (you can choose from two voices, plus two more and the option to randomize if you’re a Patreon sub), with correct pitch accent for nearly all words (with an indication when automatically generated).
* One thing that becomes a problem at some point is homonyms (same reading). JPDB only implements a “do not confuse” section for homographs (same spelling). I wrote a script to load my review history and detect when the current word has the same reading as another and if so to display the other words of the same reading. Another option is to handwave it and pass it if you recalled any word with the reading.
* If you do both text cards and audio cards, you’ll need two JPDB accounts since only a single card exists for each word. You’ll want to stagger the two accounts to make the two cards for a single word likely to be due for review at spaced apart times. I also had a script to help add words I added in one account to the other, although there’s no reason you need to learn the exact same words on both accounts, given that words may not have the same frequency in text as in speech.
* An approximation to audio cards that you could do without JPDB is simply \[reading → meaning\]. Even though this is pretty much equivalent, audio cards are a better test of recognizing words in actual speech.
* [**DoJG**](https://www.amazon.com/Dictionary-Basic-Japanese-Grammar/dp/4789004546)
* A Dictionary of {Basic, Intermediate, Advanced} Japanese Grammar (DoJG) is a book series of in-depth grammar explanations and comparisons along with syntax rules and examples. As it’s a dictionary, it’s in alphabetical rather than pedagogical order, so I would recommend reading a beginner textbook or grammar guide before even the basic volume. I read it on my phone during moments like riding public transport or using the restroom. I didn’t SRS grammar with the DoJG deck or Bunpro or anything because I never found a satisfying way to SRS grammar. After Genki, I just read DoJG and the next resource listed.
* Reading over a grammar resource is an efficient way to get most of the benefit that you would gain out of SRS or more involved study, since as long as you can recognize it during immersion, you can use context to reinforce your understanding of the grammar or choose to look it up again. This pays off because while there’s less grammar than vocab to learn at a given level, it’s usually harder to identify and look up unknown grammar than it is for vocab, due to the former’s fundamentality and nonlinearity, although there are Yomichan dictionaries to help you do it.
* [**Handbook of Japanese Grammar Patterns**](https://www.amazon.com/Handbook-Japanese-Patterns-Teachers-Learners/dp/4874246788)
* This dictionary is more comprehensive than DoJG in terms of the number of grammar points but has shorter explanations. There are still plenty of examples. This is not split into volumes by level, so I wouldn’t recommend reading it until you know a good amount of grammar.
* Another thing about grammar is that, while reading dictionaries is beneficial and all, whether (and how fast) you can comprehend something often depends less on how many grammar points you know and more on your parsing ability. The latter is based less on hardcoding and more on intuition, and the way you improve it is by immersion, hopefully that is comprehensible. This is a big reason why I consider Satori Reader such a valuable resource, for being a comprehensible bridge from a beginner foundation to the comprehensibility of native media.
# More resources
* **Yomichan**: Browser extension for word lookups: [guide](https://learnjapanese.moe/yomichan/).
* **Writing**: If you’re interested in writing, the [kanken deck](https://ankiweb.net/shared/info/759825185) is in my opinion the best way to learn. You learn to write words (rather than isolated kanji) given context and the reading. The effect of learning this way is that if you can speak something then you’ll be able to write it.
* **Pitch accent**: Just being aware that pitch accent exists is probably enough for gradual acquisition for the most part. Darius has a (https://www.youtube.com/watch?v=I-dRbTnLmBY) on strategies for acquiring pitch accent. [Dogen’s course](https://www.patreon.com/dogen) is good for learning more.
* **Onomatopoeia**: Onomatopoeia tend to be an obnoxious presence when starting to learn Japanese. I think handwaving them is advisable early on, but I’ve come to appreciate the sensitivity and vividness that they offer to the language. If you’re interested in studying the theory of sound symbolisms in Japanese, [here](https://archive.org/details/soundsymbolicsys00hama) is a book. There are also books written in Japanese like [this](https://www.amazon.co.jp/%E3%82%AA%E3%83%8E%E3%83%9E%E3%83%88%E3%83%9A%E3%81%AE%E8%AC%8E%E2%80%95%E2%80%95%E3%83%94%E3%82%AB%E3%83%81%E3%83%A5%E3%82%A6%E3%81%8B%E3%82%89%E3%83%A2%E3%83%95%E3%83%A2%E3%83%95%E3%81%BE%E3%81%A7-%E5%B2%A9%E6%B3%A2%E7%A7%91%E5%AD%A6%E3%83%A9%E3%82%A4%E3%83%96%E3%83%A9%E3%83%AA%E3%83%BC-%E7%AA%AA%E8%96%97-%E6%99%B4%E5%A4%AB/dp/4000296612). Onomatopoeia have building blocks just like other words, but they live in a more abstract space, which while initially makes them more fuzzy to learn, also allows you to use your imagination so that they eventually become intuitive. For learning the words themselves, [this](https://www.amazon.co.jp/%E6%93%AC%E9%9F%B3%E8%AA%9E%E3%83%BB%E6%93%AC%E6%85%8B%E8%AA%9E%E8%BE%9E%E5%85%B8-%E8%AC%9B%E8%AB%87%E7%A4%BE%E5%AD%A6%E8%A1%93%E6%96%87%E5%BA%AB-%E5%B1%B1%E5%8F%A3-%E4%BB%B2%E7%BE%8E/dp/4062922959/) is a dictionary I mined. There’s a lot of nuanced comparison between related words.
* **Listening**
* Listening to something while doing something else, like doing chores or commuting, is efficient. Podcasts are generally what comes to mind for this; [here](https://docs.google.com/spreadsheets/d/17P2dBQHnBnHcG3ua_24IO6sP9RDC-5b3WHV9Ri2N5qU/edit#gid=0) is a good list. It is more difficult to make input comprehensible here because it’s inconvenient to look up words, rewind, or ask a question. At the stage before beginner podcasts are comprehensible, downloading audio from Satori Reader stories you’ve already done can serve as good material. [Speechling](https://speechling.com/listening/japanese) is another option: a Japanese sentence is played, followed by a translation. A more chill option is to listen to Japanese music that you’ve learned the lyrics to.
* It is convenient to have audio files on your phone to listen to offline. To do this, you can record your computer internal audio as you leave a podcast or whatever playing on mute overnight and transfer the file to your phone. Also, [youtube-dl](https://github.com/ytdl-org/youtube-dl) can be used to download videos from YouTube.
* For reading along with an audiobook, it is convenient to be able to rewind the audiobook while keeping your reader on the screen. You can use the previous bullet point and play the audio file in VLC media player. VLC supports global hotkeys, which you can set and use to pause/play, seek, etc. even if the VLC window does not have focus.
* [**KKLC**](https://keystojapanese.com/)
* This is the only kanji study I did, but I definitely recommend it. I generally agree that kanji study is optional, but I think going through KKLC can be very helpful for not getting overwhelmed by the quantity of kanji that exist and for learning to think about kanji in terms of components. I think most other benefits of kanji study are not gained more efficiently than through vocab study.
* The order in which the book presents kanji is quite optimal: graphemes are introduced before the kanji they appear in; kanji that share a major component appear together, allowing the mnemonics for them to form a story; some consideration is given to having common kanji appear earlier. References are made to previous similar looking kanji.
* The keywords are another strongpoint: they assign multiple keywords to a single kanji if they feel the need to, since the reality of kanji is that they can have multiple core meanings. I used a script to replace JPDB’s kanji keywords with KKLC’s.
* The usage of mnemonics is a topic of split preferences, but my philosophy is that they’re fine to use as long as you don’t perpetually rely on them. The purpose of any studying, whether using mnemonics or not, is to prepare you for immersion, where you’ll be able to reinforce your knowledge in context. If mnemonics shorten the time it takes you to remember something, then they have served the purpose of getting you to comprehensible immersion quicker, and over time you’ll grow out of them. The same reason is behind why you learn basic vocab and grammar through translations and explanations in your native language. The intent is not to forever think of Japanese in terms of your native language, but to efficiently hand you the pieces to put together during immersion.
* The way I used the book was just reading through it and writing the kanji, using graph paper. I didn’t use SRS or anything to drill the content of the book: I agree with the opinion that you should eventually move on from thinking of kanji in terms of keywords; I don’t believe memorizing kanji readings happens more efficiently in isolation than by learning vocab they appear in.
* A benefit of learning how to write (at a basic level, not necessarily from memory) is that you can make a habit of writing words, in SRS or wherever you encounter them, that are easily confused with another word, look interesting, or for whatever reason. Writing can let you practice your writing and help you remember the composition of difficult words/kanji. You’ll also get better at handwriting on your phone to look up stuff you see irl.
* The book lists a few example words for each kanji, using only kanji presented so far, but it might be more useful to look up the kanji on JPDB and look at the top vocab. This takes advantage of JPDB’s frequency data, and adding these words to your JPDB rotation is an option.
* There are also graded reading sets that go along with KKLC. I didn’t use them, but basically for each kanji they provide sentences that use only the kanji presented so far.
* **japaneseasmr**: Erotic audioworks.
* [**Google IME**](https://www.google.co.jp/ime/): Type in Japanese.
* **Examples**: [Massif](https://massif.la/ja) and [YouGlish](https://youglish.com/japanese) are resources for finding example sentences, from web novels and YouTube, respectively. For Google search, enclosing your query in quotes will find exact matches, and appending とは will eliminate Chinese results.
* [**HiNative**](https://hinative.com/): Ask natives questions.
* [**LangCorrect**](https://langcorrect.com/): Get feedback on your writing.
* **Media**: [Resources](https://learnjapanese.moe/resources/#acquiring-japanese) for acquiring Japanese media. My recommendations for starter media are [また、同じ夢を見ていた](https://www.amazon.co.jp/%E3%81%BE%E3%81%9F%E3%80%81%E5%90%8C%E3%81%98%E5%A4%A2%E3%82%92%E8%A6%8B%E3%81%A6%E3%81%84%E3%81%9F-%E4%BD%8F%E9%87%8E%E3%82%88%E3%82%8B-ebook/dp/B01BLAY1G6), [ひげを剃る。そして女子高生を拾う。](https://www.amazon.co.jp/%E3%81%B2%E3%81%92%E3%82%92%E5%89%83%E3%82%8B%E3%80%82%E3%81%9D%E3%81%97%E3%81%A6%E5%A5%B3%E5%AD%90%E9%AB%98%E7%94%9F%E3%82%92%E6%8B%BE%E3%81%86%E3%80%82%E3%80%90%E9%9B%BB%E5%AD%90%E7%89%B9%E5%88%A5%E7%89%88%E3%80%91-%E8%A7%92%E5%B7%9D%E3%82%B9%E3%83%8B%E3%83%BC%E3%82%AB%E3%83%BC%E6%96%87%E5%BA%AB-%E3%81%97%E3%82%81%E3%81%95%E3%81%B0-ebook/dp/B079FG1TTC), and the Clannad VN followed by the anime. The nice thing about Clannad is that the VN and anime are both good, and reading the VN first will make the anime viewing experience much smoother since you already know what happens.
* [**Amazon**](https://www.amazon.co.jp/) Kindle & Audible: Place to buy books and their audiobooks. [Here](https://docs.google.com/document/d/1n0WLFu68g4ntBcdqSlVP0_92oZOy8VXR4AT2VFl5vkE/edit?usp=sharing) is a guide for how to extract an epub from Kindle books to use with the [ttsu](https://reader.ttsu.app/) reader. [Bookwalker](https://bookwalker.jp/) is extensive but does not support lookups, unless with OCR.
* **VNs**: Extract text from a visual novel: [guide](https://www.reddit.com/r/visualnovels/wiki/vnhooking/). [VNDB](https://vndb.org/) is a database, and [this](https://seiya-saiga.com/game/kouryaku.html) (nsfw) is a list of walkthroughs.
* [**SoftEther VPN**](https://www.softether.org/): Free VPN, useful for accessing Netflix Japan. Choose a server that is in Japan and new (uptime less than like 10 hours).
* [**Subadub**](https://chromewebstore.google.com/detail/subadub/jamiekdimmhnnemaaimmdahnahfmfdfk): Browser extension for making Netflix subs Yomichan-able.
* [**Natively**](https://learnnatively.com/browse/jpn/): Rate difficulty, language learning, and entertainment for media.
by ofthesamename