Mnemotechnics of Xenia

Let's consider Vaughn Gene's approach to "Basic Fluency in Japanese." Like John (see this post), Vaughn also uses extensive mnemonics, however instead of applying it to the kanji, he recommends aiming for conversational fluency first. This is after all a very natural starting place:

"The Prelude: Learn only the very basics of grammar, because you really want to learn that in context when you learn sentences. And learn the hiragana and katakana syllabaries. However, I would not touch kanji until you can speak and understand the language with confidence. All kanji can be read with hiragana and katakana. I read a lot of Japanese now, but this is a result of sheer exposure and just years of being around the kanji that I needed. For now, focus on learning to understand and speak the language.

Step one: Learn 2,000 to 4,000 of the most common vocabulary words. Do this up front to save years of study time and everything will become somewhat comprehensible. Deliberate practice is key here. This is probably the hardest part of learning a language, but vocabulary in isolation can take you very far [cf. Guy Deutscher]. And it really opens up any kind of content you want to enjoy. How do you remember all of these words? I use mnemonic associations. You can find full spreadsheets of vocabulary lists online, and then maybe you'll just have to add a column at the end to enter a mnemonic association, and write whatever helps you remember the word. Try to do it on your own. This is going to be a deliberate effort, but it's going to save you months if not years of practice, so take your time with it. 

Step two: Practice listening and repeating the words. You can find text to speech websites these days to help you hear it as you study it. That way you can kind of get the input and the output at the same time. Anki decks have maybe 5000 words already laid out. And it has the sounds, it has sentences, has everything pretty much there. So you can just download that and get to using it right away. [Added: Do a lot of retrieval practice. Try to describe pictures, translate sentences, and respond to prompts, all in your target language.]

Step three: Make or find several hundred sentences in every category around your life, and drill those. This means basic things around work, hobbies, interests, anything that involves the same things that you would talk about in your native language. Sentences are frameworks. And the more frameworks you learn, the better. All you have to do is change out the subjects and the vocabulary words to suit any context. So get all the vocab words you need, make a whole lot of sentences, and once you kind of have it down, just move on to the next thing.

A note about translators and LLMs: You can prompt an LLM with "Give me 30 sentences about this subject of interest in Japanese." "Create a conversation that I can practice out loud." "Explain to me the rules and grammar and the ways that it's used." "Give me several sentences with this word." Just get really creative. It can just spit all that out right away. You can leverage AI in such a way that you can get a week's worth of work done in just a few minutes of typing, and now you have practice for the next couple of days right there. So really leverage that to get a lot of the upfront work out of the way. Some people see language learning as impossible. But deliberate practice leads to subconscious mastery."

Vaughn’s follow up video “2.0” elaborated on a few of the above points:

“Where do you find sentences? Shadowing native content is a great way to start. Find a video like “100 Japanese phrases.” Learn to shadow the entire video. And whenever you find a sentence that you like, or just think of one you want to say, put it somewhere in your notes. Make ten different variations. And just drill them over and over again until you can say them without thinking. Start small. “My name is this. What's your name? How old are you?” All that. Learn to say those just effortlessly. Then try more advanced sentences, and so on. But don't skip over subject related things. Whatever you like to do, learn to talk about that and use that confidence to go into the next thing.

I know in the first video I said about 2,000 to 4,000 words. If you want an end goal, 10,000 is good, but 6,000 to 7,000 is like the magic number for vocabulary. I find that after 7,000 most of it's like cherrypicked vocabulary words that I may rarely use, but anything before that is common. And if you really want to flex in Japanese, getting up to 15,000 or 20,000 is great, but you're going to need to watch a lot of TED talks and scientific things and all that.”

Recall that John had applied mnemonic associations to learning kanji (the Heisig method), but Vaughn in his approach applies it to vocabulary memorization, which is the most basic area of language learning. He describes mnemonics as a sort of  "temporary scaffolding" that is to be phased out and no longer used as familiarity increases and the associations are no longer needed. In other words, the goal is understanding the language, not mnemonics for their own sake. Only use these as necessary, then feel free to discard them. Indeed, we may think of language in a similar way. If the goal is xenia, then by all means use language, but once the words are no longer necessary they can be disposed of. As Zhuangzi famously said: 

"Nets are for catching fish; after one gets the fish, one forgets the net. Traps are for catching rabbits; after one gets the rabbit, one forgets the trap. Words are for getting meaning; after one gets the meaning, one forgets the words. Where can I find people who have forgotten words, and have a word with them?" — Zhuangzi, Ch. 26 

And yet, viewed another way, Zhuangi is not wholly right in what he says here, because the transcendent can only be glimpsed through the immanent. The embodied kotodama of language (semiosis, more generally) is needed to allow us to see the transcendent meaning and metaphysical associations behind it. And so the kotodama never entirely disappear, but they do transform as needed. And Zhuangzi would affirm this, since transformation is one of three core themes he explores, per Brook Ziporyn, along with perspective and dependence. Using that sort of approach, which is consistent with xenia, we can go almost anywhere. See also the balanced perspective of Olle Linge on the uses of mnemonics, a very useful tool that functions best when used within a broad educational context. 

Another Japanese language learner, "Matt vs Japan" (Matt) advocates the "All Japanese All The Time" (AJATT) full immersion method rather than traditional classroom study. He cites Stephen Krashen and J. Marvin Brown to justify his approach. Krashen, as you may recall, highlights the importance of comprehensible input for learners, but unlike Matt, he also recommends extensive reading (Matt focuses on verbal skills). Basically, “comprehensible input” means consistently exposing yourself to language content that’s just above your current level – close enough to what you already know that your brain can work to fill in the gaps and raise your language level. Matt makes some good points:

"People who are really new to language learning have this idea that language learning is a lot like math. Memorize the words, memorize the grammar formulas, and then you'll just understand and be able to use it yourself. But in reality there's many small nuances, and for most words in Japanese there's not a direct one to one equivalent in English. So you have to grasp the nuance if you want to be speaking fluently, you can't be translating in your head. Japanese people don't speak by placing Japanese words into English sentences, right? So the starting point is listening, active listening, not just passive. Stephen Krashen's philosophy, or conclusion, was basically throw classroom study out the window and focus on getting comprehensible input.

If you are a language learner, LLMs are an amazing tool. They are the best possible dictionary. I have a prompt that I use to generate the definitions for an Anki add-on. It will use the openAI API. Like I said before, a lot of Japanese words don't mean the same thing as any one English word. They have a unique nuance that isn’t equivalent to one specific English word. So a lot of times dictionaries function more like a thesaurus. Believing that every Japanese word has a perfect English equivalent would be actively harmful, because then you project that onto the input that you get. You can also use LLMs to generate comprehensible input. There's this feature where you can upload a PDF and it will make a podcast of two people talking about the content of the PDF you uploaded in everyday Japanese. For language learners, this is amazing."

Notice how Matt's concerns are almost opposite those of Vaughn. But this should bring everyone back down to reality. If languages are replete with nuance, and fluency is a nearly impossible goal, then second language learners must ask themselves: What is a realistic goal? For me, I think reading and writing would be my first goal. And I say this for several reasons. Literacy remains the easiest method for asynchronous communication and vocabulary expansion. To this day, most information is stored as raw text. This is the doorway into the enormous "adjacent possible" that, for a second language learner such as myself, lies latent within foreign cultures, just waiting to be explored. And character recognition is a prerequisite if you are going to use reading to expand your vocabulary. Listening is probably the second skill I would like. But there's a caveat. I may just be leveraging literacy in order to gain listening skills - I suspect that the "inner voice" produced while reading is able to indirectly strengthen listening skills, just as "subvocalization" while reading can assist with speaking skills. And listening is also personally important to me: I want to finally be able to understand my wife in her native tongue. That's something I've always wanted. But speaking is my least important skill, for several reasons. First of all, I do not feel a need to draw attention to myself, at any particular moment. I am here to learn from others. Anything I have to say is better said after some reflection anyway. That being said, conversational ability is nice to have, but since the nuances of fluent speech will be the most difficult to spontaneously reproduce, my expectations (and goals) are correspondingly lowest in this area.