Reflections on learning Japanese after three years of study

It has not been quite three years, but I am getting there. I started learning Japanese in July 2018, and I have consistently studied almost every day. It has also been almost a year since I wrote the last post on this topic. So I might as well write another post now. I still don’t consider myself fluent, but I’ve come a long way. I still haven’t been able to take that trip to Japan due to COVID, and I am not sure this year will be different. Maybe next year.

Around this time last year, I was trying to implement a major overhaul to my study strategy. I wanted to jettison Anki and try a more conversation and passive listening based approach. My major innovation was to commit to weekly 1 hour italki lessons (approximately ~$60/mo). I have done that for almost a year now, and I can say confidently if you are starting to learn any language, you should schedule some kind of regular speaking practice and make yourself do it, potentially from the first month. With a paid instructor, you don’t need to know much if anything to start because they are paid to teach you, and they will be patient and help you out. If you choose to do a language exchange, which I have not really tried, it might be best to wait until you know enough basic constructions to keep a conversation going at some level.

I think language learning is essentially three tasks: speaking, understanding spoken language, and understanding written language (if the target language uses a different script from your native language). The tasks are synergistic, but I don’t think you can focus on one to the exclusion of the other. When it comes to speaking, if this is your first new language (L2), as in my case, you have to learn a skill that you will not learn via any other means than practicing speaking. That skill is the ability to convert in real time the complex sentence structures and rich vocabulary of your native language into concise, simplistic expressions with basic vocabulary using the bits of the target language that you know. This skill is different from learning vocabulary or grammar.

For example, during conversation, if the thought that you want to express is, “My brother is a financial analyst.” You will get stuck and say to yourself, “I don’t know the term for financial analyst”. You need to find an alternative that you can say quickly, such as “My brother works for a bank. He works with numbers,” or something roughly equivalent. Then your listener might offer, “You mean an accountant?” and so on. Many times I get hung up on a complex English sentence structure, unable to directly translate it. For example, “It’s rough taking my dog out now that I’ve moved to the city because he tries to bark at every dog he sees.” This could become “Walking my dog is hard. He barks at other dogs. Before, I was in the country. Now, I am in the city. There are more dogs here.” I think this is a challenging skill to develop, and there’s no way to do it without trying to have real conversations.

I don’t really remember what I was doing on the days between online conversation lessons last calendar year, but it was minimal–probably around 30 minutes a day. I believe I was working through the KanjiDamage Anki deck again. I was listening to native Japanese podcasts but not really understanding much. Definitely there was some level of burnout after two years. My principal insight from this period is that passive absorption of language doesn’t really work–or rather the time frame may be abysmally slow. That is, casually watching and listening, hoping to intuit new word meanings doesn’t seem to work. It probably requires understanding at least 95% of what you’re hearing, as I have read somewhere before. If you are a beginner or intermediate, this means you cannot get much out of a material that you would actually be interested in. You can only watch so much Peppa Pig as an adult learner and stay motivated.

My second insight is more time studying–specifically–more active studying equals more progress. I started studying more intensely, usually around two hours a day this February, and progress has noticeably accelerated. Thirty minutes to an hour a day for Japanese is probably not enough if you are hoping to be fluent in two years–at least, not for me. Perhaps it would be if 50% of that time was conversation. I would like to increase my conversation time, but it’s fairly out of character. I run out of things to say to a stranger online even with once a week lessons. If I could bring myself to do it, I think this would increase my progress substantially.

As for the second task of listening comprehension, I believe this is largely dependent on your vocabulary. This may be the biggest challenge of learning a language. There are just so many words. It seems to take so long, and it never feels like you know enough. Fortunately, the more you know, the easier it gets to add new words due to shared roots of various kinds. For instance, I originally learned 警察 (keisatsu–police) a long time ago. Adding 警官 (keikan-police officer), 警備 (keibi–security), 警戒 (keikai–vigilance) as I came across them in the wild was much easier. Each word tends to become part of a system. This is also a big reason why trying to learn another language with Latin roots like Spanish or French is so much easier. Consejo (advice) in Spanish is similar to counsel in English, for example.

I am still not settled on the best way to learn vocabulary. One thing I have learned is that trying to memorize words out of context, say from someone else’s “deck of useful words” or a Genki word list is probably one of the less effective strategies. The way I think of it now, there are two types of words: “first class” words, which include all your native words and L2 words which you can use like native words–that is, without translating, and “second class” words, which you sort of know but have to stop, reflect, and translate. Second class words must be converted to first class words, but that requires a contextual, conceptual understanding. What I mean is, 犬 (inu) doesn’t mean “dog”; it means that thing that lives with you, that you have to feed and take on walks. 発信 (hasshin) doesn’t mean “departure”; it’s what you say right when you launch your Evangelion. 偶然 (guuzen) is what you say when your nemesis unexpectedly gets onto the same elevator as you.

Memorizing a suggested vocabulary list just creates a bunch of second class words. You still have to get them into first class somehow. And if you don’t, you will struggle to remember them. I believe it’s more or less as difficult as memorizing random nonsense patterns. If you build up too many second class words, you will end up spending much of your time forgetting and rememorizing them.

I’m still undecided on the best way to create first class words. Clearly, you need an associated, interesting context. On the other hand, furiously writing down every unknown word in an article, from a scene on TV, or even a chat log seems excessive and not maximally productive. I currently take words from scenes in shows I am watching by pausing and looking up words I hear clearly. From online articles, I try to be judicious, picking around 5-10 that seem most relevant, otherwise the count will spiral out of control.

Last time, I wrote about my disappointment overusing Anki. Naturally, after a hiatus, I am back to using Anki a lot, but with some changes. I have two decks. One is my daily deck, which I use mostly as Anki is defaulted, except with a leech threshold of three. I have a timebox of 7 minutes (yes, only 7!), and I ensure my deck never takes more than that to review by severely limiting the new words. I use this mostly for Kanji study.

The other deck is my I-just-can’t-help-it deck, which I add tons of words to by semi automatic processes (Yomichan [Firefox extension] and Aedict [Android app]). For this deck, I don’t feel obligated to keep the green reviews under control. I also tried setting my leech threshold to one. That meant, as soon as I couldn’t remember the word after the first time I looked it up, it was gone. Now I have my leech threshold at two, as the previous setting caused me to lose too many words I kind of knew. The reasoning for the low leech threshold is if I can’t keep a word in mind, I will need more context to learn the word well. Similarly, if I added a word, but it’s taking too long to memorize in Anki, I just delete it. Most importantly, I do not allow total Anki time in a day to exceed Netflix/Youtube/reading time. The low leech threshold lets me keep Anki time down, and if it starts getting too high I stop adding as many words. That said, the more aggressively I add new words, the more progress I feel.

Lastly, for Japanese or a flavor of Chinese, you have to learn the Kanji to be able to read. This is proving to be a monster challenge. I ignored Kanji during my first year of study. I learned to recognize abstract English meanings of the Jouyou Kanji the year before last, and I repeated that exercise last year. Yet even now I am still feeling quite illiterate. Being able to recognize 車 (kuruma) as “car” is useful to some extent, but it is more likely to be part of a compound like 馬車 (basha). In this case, it is more important to recognize it as しゃ (sha). In this case, a carriage is sort of like a car, but you really need the phonetic association, not the English meaning. Most Kanji readings are like this or even more abstracted, and I don’t know most of them.

I can read online text with the assistance of Yomichan. Beyond that–PDFs, signs, etc–text is completely incomprehensible. This is a major problem, as I feel for me the best way to improve grammar and vocabulary would probably be by reading as I could find more interesting material than I see on Netflix. To this end, I have very recently bought Kanji in Context (KIC) and its workbooks. Unfortunately, this brings me back to the complaint just a moment ago–that memorizing vocabulary lists without context is ineffective or at least inefficient. The KIC workbooks do have repetition and sentence level contextual use built in, so hopefully I can succeed with this method.

If you are trying to learn Japanese, my advice is to start learning to recognize Kanji readings immediately by memorizing them in the context of words. It feels impossible to communicate without them in the modern era since so much is done by text messaging these days. I regret not starting earlier. On the other hand, it is a bit of a chicken and egg problem. I feel like you need a substantial Japanese vocabulary to effectively learn the Kanji.

So where am I now as I approach three years? I might give myself low intermediate. Fluency will still take at least another year, if not two, I imagine. I can have a conversation with you in Japanese, but I might have trouble understanding a lot of the words you use. I can basically express whatever I want, at least in a round-about way. I can watch TV shows, but I get lost in dialogue heavy scenes. I don’t understand the nuances, but I generally understand what is going on. I still understand Podcasts only at a topical level–that is, not very well. 本当に知りたいなら、話しましょう!