Advisability of Studying Vocab sans Examples?

[torpedotaiyaki]

TL;DR:

I’m been wondering how advisable it is (long term) to learn vocab without examples/context. This is a pretty much a listening issue, i.e. not a reading issue.

Context:

     I’m increasingly doubting how much of a good idea it is - or indeed, if it is even really feasible at all - to learn vocab without example sentences. I only know about 4.5k words (9k cards according to Anki, at least) and already I’m have cases where homophones are starting to stack up. Fun stuff like 「艦上«carrier borne/aboard ship»・勘定«bill»・感情«emotion»」 or 「描く«to draw»・書く«to write»・かく«to scratch»・各«each»」. I cannot really know what is being said with these words without an example to provide context.
     However, homophones are the least problematic example. What’s actually giving me heartburn is words that I don’t have a good excuse to fail on, but I do anyway. Ironically (or maybe not??), this tends to be shorter words. I’m not sure I’ve ever failed the words 津波避難進路«tsunami evacuation route» or 攻撃型潜水艦«attack submarine»※.
     By contrast, I’ve been struggling with words like 流行«en vogue» or 溜まる«to be saved» for an embarrassingly long time. I usually pause my audio before it gets to any example sentence to really test how well I know a word, actually. I then on words like these I stall out. I only let the sentence play if I can’t recall the word within a reasonable time frame (5-7 seconds-ish, usually based on vibes). I often snap to the definition after hearing the example sentence, so it seems as if I’m learning to identify the sentence, instead the word. A common reply here is ‘just change out the sentence then’. This is only a band aid, though. As it’s usually only a matter of time before the new sentence ends up in the same spot.

I am of two minds on this problem, and my core confliction is I cannot bias towards any solution despite having ruminated on this for quite awhile. Those two minds are as such:

1. Git good. If you need an example to know the word, then you do not know the word. Things like homophones or single mora words (胃) are obviously exceptions to this kind of thing. But a word like “sheet” gives me no problems in English, and thusly something like 溜まる should similarly be child’s play. If there is a data recall failure, there are many ways to try to improve recall.

2. Just listen to the sentences, stupid. “Japanese is a contextual language” is so commonly repeated as to almost be dogmatic, but it is not false. You may be able to get decently far, but you can only make so much progress by learning words in isolation (and un-inflected dictionary forms no less). There are many ways to try to improve recall, but (A=B) in a vacuum is a fundamentally doomed approach.

How doth the forum opine?

※ I know in some sense these examples are not words, but multiple words strung together, but I feel that doesn’t really matter here. …maybe.

[Clerk1856]

What type of review are you doing? What’s on the front of the card and what’s on the back?

[kfrl]

I have been SRSing for about 5 years, with roughly 20,000 words known (granted that number is inflated by what jmdict considers a word). I have pretty much never looked at example sentences. So yes, it’s completely feasible. I’d also go further and say it’s advisable, with the caveat that this approach prioritises being able to get through a lot of reviews quickly at the expense of not knowing the deep nuance of words, but I pick that up from immersion, not SRS. (In fact, I know someone who only SRSs the readings of words and gets the meanings entirely from immersion).

However, I don’t do it this way because I believe “if you need an example to know the word, then you do not know the word.” You worry that “it seems as if I’m learning to identify the sentence,” but I don’t think that’s a problem at all. When you see that word in the wild, it will probably be in a similar context to the example sentence you’ve memorised, and you’ll put two and two together. So, I don’t think relying on examples is bad at all, I just don’t do it to save time. And yes, I have embarrassing leeches like your 流行 and 溜まる: when I’ve identified one of those is becoming a problem then I’ll come up with a mnemonic, or just accept it and hope something I read/watch soon will use it enough that it’ll stick.

[torpedotaiyaki]

@Clerk1856 Describing it via text would probably be laborious; I’ll just upload images. Here’s a busy and simple card back side. The front side of my listening cards are essentially blank, sans the audio replay buttons (so there’s nothing to really screenshot).

Core vocab (top center), metadata box (top left), 0-2 example sentences (middle), and any notes/annotions (bottom). I do not put pictures in cards. To be explicit, the only thing I’m testing on is the core vocab data at the top (in concert with the idea of card atomization). Everything else is just context, expository, or for clarity. Well, except for verb transitivity. I will fail myself if I mix up verb transitivity, as IMHO they’d be two different words.

@kfrl

the caveat that this approach prioritises being able to get through a lot of reviews quickly at the expense of not knowing the deep nuance of words,

Would you mind elaborating on this rationale? I try to answer cards within 5-ish sec, but this is mostly for the purposes of keeping total flashcard time to a minimum. I think perhaps my rationale for doing so may be different than what you’re describing. Unless that’s precisely what you meant lol

[kfrl]

Yep, that is precisely what I meant!

Some general advice that might help you come to a decision:

1. Define your goals: you haven’t said what exactly you’re trying to achieve. You’ve said it’s a listening issue, but does that mean you’re aiming towards everyday/holiday conversation? Watching anime without subtitles? Business communications? A clear goal might help settle the path.
2. Just try it. Spend a week, a month or whatever listening to the whole example sentence every time. If you’re worried about learning the example, not the word: this will be a tiny blip in your lifetime of SRS.
3. On that note, remember that SRS is only a small part of language learning, one that most learners have never even heard of, and your SRS approach doesn’t need to be perfect to be useful.

[torpedotaiyaki]

That’s worth mulling over.

SRS is only a small part of language learning

Indeed. I’ve always considered it as ‘synthetic’ study - like the scaffolding to the language study building. Real immersion, media, and conversation is the real brick and mortar of the building. It’s just harder for me to stay consistent at the latter.

[SilentRequiem]

You’re actually missing the biggest issue: the words you think you know, but actually don’t.

Here’s the thing, the dictionaries and textbooks lie. Dolly-sensei got that absolutely right, and more Japanese I learn, the more right I realize she was.

Dictionaries don’t tell you what a Japanese word means, they tell you what an English speaker might say when they want to say something similar. And that’s not the same thing at all.

This manifests in two ways. First, is poor grammar. If you look at jisho.org, for example, 分かる means ‘to understand’. As soon as you see an example sentence, however, you know that this is not the case. 答えが分かる。You can see that 答え (‘answer’) is marked with が, which means 答え is the subject of the verb, not the object. So 分かる actually means something closer to ‘to be understood’.

Digging a little deeper, we find that 分かる is a punctual verb, meaning that in marks a moment of change or transition. So, in fact, 分かる means ‘to become understood’. Which explains why the Japanese say 分かった when something is explained to them. It has ‘become understood’.

Second, dictionaries lack nuance. This word I learned yesterday, 了承 【りょうしょう】, for example is acknowledgement; acknowledgment; understanding; approval; consent. But this meaning is far, far too broad. 了承 is specifically approval that moves downward in a power structure.

A boss can 了承 (approve) a request of his underlings, but you would never 了承 (acknowledge) the actions of a co-worker, or (understand) the decisions of your boss, and you don’t 了承 (consent) to having sex. None of this is apparent from the dictionary entry; only by seeing it in context do you come to understand the nuance of the word.

TL;DR If you’re not learning Japanese with a wide variety of Japanese sentences, (ie, ‘in context’), you’re not really learning Japanese at all. You’re just playing a Japanese themed memory game.

Fortunately, Bunpro makes seeing words in multiple contexts really easy to do.

[torpedotaiyaki]

@SilentRequiem Man, you don’t know how much I still bemoan Cure Dolly’s (possible/postulated) passing. I’m unsure if her material is quite as relevant to modern learners as it once was, since the resources for Japanese learning are so much more numerous and high quality today than 20 years ago, but I was absolutely one of those people who got started off on Genki/Japanese for Busy People, etc, which so much her material was aimed at correcting the deficiencies of. Her framing of grammar constructs and rejection of certain dogmatic maxims were invaluable to me. I still think of the に particle in the way she framed it (archer aiming a bow at something). I still think of her explanation of how 赤い doesn’t mean red - it means is red. I don’t think anyone has really usurped her yet. The presentation was terrible, but the content was absolutely GOATed. Kaname Naito is one the the few contemporary JP language teaching Youtubers I’d put in the same hemisphere.

Broadly I agree on the jmdict issues. I’m not really ready to fully jump into J>J dictionaries yet, but I’ve long been quite eager to leave jmdict behind.

You’re just playing a Japanese themed memory game.

Huh… That’s a hell of way to put it. Accurate though.

[ka5]

So you have a spoken word on the front of the card, with optional example sentences/context.
And the issue is whether to pass or fail a card if you don’t recognize the word in isolation but do recognize the whole sentence?

I don’t think there’s really a problem with that. I’d pass or fail this situation based on the mood of the day.

There’s value in just remembering example sentences too. The language is a huge set of possible sentences in context.

If you pass the word in SRS and then don’t recognize it in an unfamiliar sentence, why not make a second card with that new sentence. Or go back and fail existing card manually and send it to the review pile again.

To me, a word needs some context to be on SRS rotation. It can be an example sentence aka mining. It can be a more general idea of where it came from or which context it can be used in.

Card layout I personally use has example sentence on the front blurred:

Most of the time I don’t look at it at all to save time. But it gives flexibility to formulate the mental “answer” either immediately based on the word alone, or after un-blurring the example. Pretty similar to your situation, but in text rather than in sound.

Fundamentally, I wouldn’t treat SRS as some sort of objective test that has to be uniform, calibrated, produce a rigorous retention metric, and be approved by internet experts on SRS.
SRS is just a tool for reminding yourself that certain words and situations exist.

[Beiri]

There’s no wrong way to learn so I’ll just post my current workflow. I do 17 new cards of Anki vocab at 80% retention because a year ago I calculated that this would keep it within 30 minutes of Anki per day. I try to do each card within 3 seconds to keep total review time down and I always pass cards I get the reading on and then briefly look at the definition, immersion is a much better way to learn definitions. I follow that up immediately with 10 new cards of the kanken writing deck, this takes me another 15 to 20 minutes and I’ll usually watch anime while I do this. The rest of my free time is allotted to reading which is usually light novels or visual novels for about 2 hours daily but I try to do more when I can.

I never do anything that I don’t enjoy or ever force myself to read when I don’t want to. Learning Japanese without enjoyment is inefficient it’s better to take a break.

I mine every single word that I encounter that I do not know. I use Anki reorder script soeverything is rearranged by frequency, high frequency cards always hit my new pile first.

I have leeches in Anki set to 3 because I found the optimal way to remember things I have trouble with is to not expend any effort on them at all. If the card is bad it gets deleted quickly and then remined after I see it again in a book. I’ve noticed that leeches will sometimes be remembered and not remined at all which significantly cuts down on the amount of anki I have to do total.

If I see a particularly good sentence or example with a voice line in a visual novel I will often delete Anki entries and replace them with that. All of my cards are vocab with example sentences on the back with definitions and readings.

Other than that my job allows me to do a lot of listening immersion so I usually get four to eight hours of audiobooks or music everyday.

Anki should not be something you ever think about, streamline your workflow and make it work for you and don’t get attached to cards because they are replaceable.

You may have noticed I did not mention bunpro at all that’s because until recently it wasn’t really a part of my regular workflow. Bunpro does not have good SRS so it is suboptimal to do it everyday, I do 10 or 15 cards if I feel like it while sitting on the toilet. I’ll do new lessons the same way. Once they optimize the algorithm it might be worth doing again but until then it creates far too many reviews for very little return, I mostly use it as a reference book.

This workflow is the result of 3 years of refinement and studying the language I’ll probably keep doing this until I’m fluent I hope what I’ve written down can help you somehow.