Question about vocabulary learning

So today while reading I came across the word 抱える “kakaeru”, to hold. I had to look it up.

I thought, I’ll add it to my list of words I had to look up, but what can I do right now to try to force myself to remember? The meaning isn’t so hard (because of the kanji) but what about the pronunciation? A crazy mnemonic? (“I don’t want kaka in the air so HOLD your sneeze”)?

Any suggestions?

No good suggestions, except to note that you might want to think of something that distinguishes it from 抱く(だく) and 抱く(いだく) while you’re working with the word…

If it was me, I’d either add the sentence I found it in to Anki or find the weirdest/most interesting sentence I can with it and add that instead.

1 Like

I imagine that you came across this word while reading a light novel or something. In that case, you can act along the lines of what Bokusenou was suggesting. Check if the word is level-appropriate—⌈抱える⌋ is N3 or N2 level, so it passes that test—and then add a card that includes some context, anywhere from a fragment of a sentence up to a couple of sentences, to your SRS. Highlight the target word on the front, and put the meaning in English on the back.

You could do that… but ultimately, it doesn’t really work, it’s not efficient for learners until around post-N2 level, because it’s far easier for the brain to recall something it hears, than something encountered through reading text alone. Professor Cornelius Kubler, who has taught East Asian languages for decades, said something to the effect of, “You can learn the spoken language without the written language, but not the other way around.” The principle applies even if your end goal is to learn to read.

Now that I’ve finally realized this, and really actually internalized this reality, I’ve re-designed my studying entirely around listening, and I’ve loaded up on stuff with ja/en transcripts. I do still study the kanji, but I see that as more of a side hobby and a personal interest not directly related to language study.

While listening, when you get to a passage with an unknown word, check the meaning and repeat that short segment five or ten times on loop. If you like, add it to SRS. This is what people are doing when they talk about “mining” media with subs2srs or Voracious.

Gains come from listening-first practice, and challenges like “hard words” take care of themselves. For whatever reason it’s harder for learners to pick up Japanese verbs than nouns, but the answer isn’t to use mnemonics. The correct choice is to change your routine to make learning easy.


I stopped trying to create mnemonic for new learned information a long time ago. Mnemonics are great for quickly storing something for short term memory, but I don’t think it’s that much faster to convert it to long term. And long term (or instant access to the information) is really what is useful for language namely for listening and speaking.

Personally, I find repetitively hearing the new words using sentence cards(and then later associate the kanji/reading to the sound for reading purposes) is the easiest and most “natural” way to acquire words. From my experience, it takes about 3 days of SRS of hearing the same word/sentence for the brain to recognize it easily. Though, it still takes tons more usages and contexts of the word to make it truly my own. SRSing just enables the brain to start picking out that word from the wild and creating key intuitions.

Also another thing about learning vocab is that quantity is better than quality(pretty common theme for language). The fact is that to acquire a language, you’ll need tons and tons of vocab. It’s a never ending process. So instead of spending tons of time to learn each word perfectly, a better strategy is to half ass learn a lot of them. The ones that matters, you’ll get quite good at very quickly since you see it over and over. The ones that you don’t, well who cares and you’ve planted some seeds somewhere deep in your brain. In general, spending too much time on each individual word is really not worth it. Maybe, using some elaborate mnemonic system to learn real well might not be the best strategy overall as it encourages more time in front loading ways to hack the brain to learn abstract concepts. But the key piece is that you’ll still need to do immersion to bake into long term memory.


This is starting to resonate with me as well. What does your immersion process/deck building process look like exactly?

I’m currently “immersing” by working through audio from books for learners.

There’s an app called WorkAudioBook that is helpful for working with audio lessons, but it’s not an option for me so I’m using a subtitle editing app so that I can highlight interesting or unknown expressions and play them on loop. I’m not using SRS or reviewing those highlights right now, although I am saving the resulting timing data so I could do that later if I want. I am also working on automating my process, including media organization, audio playback/looping, mining, and later review.

For more ideas, see the “How to Immerse: Listening” video by MattVSJapan.

Can I ask which series?

  1. やさしい日本語: This is a new textbook series that covers N5 & N4 material in four (1, 2, 3, 4) relatively short volumes. I like the many example mini-dialogues with recorded audio. I’m starting from the first volume, and most of it is familiar to me, but I don’t mind because there is material I don’t know.
  2. 日本語単語 Speedmaster: This series has four volumes covering all JLPT levels. Here’s N5/N4 and N3. The audio for the example sentences is good. The second volume (N3) has more example sentences and therefore more useful audio.
  3. Speak Japanese! series: Each page has a mini-lesson on a grammar and conversation topics, with about ten examples with audio. The “easiest” book has en and jp audio, you might skip to any of the five books (here’s one) listed as “upper beginner through upper intermediate” level, which have ja audio only.
  4. Shadowing: Let’s Speak Japanese: I have the first volume but I haven’t used it yet. I previously linked to the “preview” on YouTube here.

I used to recommend NihongoShark for their great lessons and many thousands of example sentences (with audio) but they are transitioning to a new “walled garden” style platform and won’t be providing Anki content to new customers. I learned the sentence「次の文を普通形の文に直しなさい」from them several months ago, and thanks to the magic of listening, I still remember each mora, even though I couldn’t write the words (in kanji) past 次の文。

Thanks this is great. I am filing these for future potential use. I am actually working through Shadowing: Let’s Speak Japanese now. I am only a few days in so I can’t speak to it too well, but so far it is really getting me into the habit of daily shadowing which I feel like I was really lacking.

What you says is even truer for the Japanese language, where the written form is not a direct phonetic rapresentation of the spoken language, but it adds another layer of indirection with kanji.

To be honest, once you know the kanji, they’re such a help to infer the meaning, that often you’ll end up attaching the meaning of a word to its kanji form, and once you remove the kanji (ie when you’re listening without a transcript) you’ll be unable to recall the meaning of a word you were supposed to know.

This is why in those days I read everything in hiragana, and it’s unbeliveable how much it helped me with listening comprehension.
When I want to read something, I put it inside Anki, which will auto-generate furigana. Then with the help of CSS I hide the kanji, and use the furigana as hiragana but with a different color, to help me see the beginning and end of words:

As the Anki Japanese plugin isn’t perfect and it tends to make mistakes, I tend to read only things for wich I have the audio.

For example things like this:

【怪談】77話つめあわせ -2019年度怪談朗読セレクション総まとめ-【投稿話/ホラホリ図書館/怖い話/都市伝説/朗読つめあわせ】

This is 11 hours of Japanese audio and, while transcript aren’t immediately available, it suffices to search a small section of text within Google in order to find the transcripts. I’ll then import them inside Anki and I’ll read them in hiragana, after I’ve listened to the audio.

1 Like

This approach looks helpful for boosting listening comprehension without “cheating” too much.

MeCab tokenizes たいけんだん as two separate words, and such inconsistencies with the Anki Japanese plugin always annoyed me. I think I can improve on the tokenization and maybe also enable かん on mouseover. I’ll try to implement this functionality outside of Anki.


TBH, after 1+ year of KanKen study, I’ve noticed the opposite effect. I’m able to deduce the kanji I don’t know for words just by hearing them. My brain can use a combination of the sentence context, its knowledge of kanji meanings, and knowledge of word construction (熟語構成 - synonym pattern, antonym pattern, decoration, action, denial) to figure out what the kanji ought to be. That usually gives me the meaning for the word as well.

This supposes a lot of kanji study, I think, and knowing the various meanings that can attach to a given kanji. I use the 漢字辞典 to look up individual kanji meanings.

1 Like

I haven’t tried this method yet, but in my copy of “A Frequency Dictionary of Japanese”, the authors go into some detail of which corpora and morphological analysers they used, etc. One interesting part:

First, a morphological analyzer called MeCab with a dictionary called UniDic, specially developed for the BCCWJ project, were used for SUW analysis. These SUWs were then filtered by a tool called Comainu and made into LUWs.

@fkb9g just to let you know, I’ve now had a chance to test out Comainu and it works!

$ echo "体験談" | mecab
体験	タイケン	タイケン	体験	名詞-普通名詞-サ変可能			0
談	ダン	ダン	談	名詞-普通名詞-一般			1

And then with Comainu which adds extra fields:

$ echo "体験談" | plain2longout
B	体験	タイケン	タイケン	体験	名詞-普通名詞-サ変可能			名詞-普通名詞-一般	*	*タイケンダン	体験談	体験談
	談	ダン	ダン	談	名詞-普通名詞-一般			*	*	*	*	*	*

There wasn’t any documentation on the output format, but the last field seems to indicate the detection of a LUW (long unit word) which is what I think you’re after.

1 Like

About Anki audio cards, what format do you think is the best?

  1. audio of sentence only, for example: パラシュートを装着する。
  2. audio with target word and sentence, for example: 装着。パラシュートを装着する。