Let's Discuss Flashcard Types

I’ve been looking around the forums and I’ve found many posts about using flashcard SRSs but not much about how to make them more effective. Specifically I’m asking about different types of cards that you have found to be the most beneficial for your learning.

I’m going to start using Anki again as my main reviewing tool, and I’ve been thinking about how I want to make my cards. Sentence cards are a very popular method, but some time ago I stumbled upon this entry where the author argues against them in favour of word-in-front, context-in-back cards. The main points they give are mainly these:

  1. They take too long to review.
  2. The beginning of the sentence triggers your memory of the situation and therefore its meaning, and it doesn’t guarantee that you actually remember the information in other contexts.
  3. They test too much at once, not only the meaning and reading of the word you go after, but also of all the other words, grammar, etc. This is much better supplied by immersion.

The context at the back of the card is therefore used when you don’t remember the word (and fail the card) and need a refresh; dispensable otherwise. However, it also comes to my mind that the word-in-front cards have one flaw that sentence cards don’t, and that is words with multiple meanings. With sentence cards, the sentence itself lets you know the specific meaning of the word you’re looking for even if the word has many meanings, but this is something that can’t be done with the other type. I’ve been thinking about how to get around this issue. The entry talks about a ‘hint’ field, where you can put little bits of information in the frontside to clear ambiguity, like words that can be read with both kun and on readings, and so on. But I wonder if specifying something like “Word (not x meaning)” could potentially affect the effectiveness of other cards with their other meanings when they come for review, this is, I would already know the answer of a card if I’ve had the other one a few days before and therefore cheat. I think the example cards they put in the entry cover all the meanings in the back with [1], [2] and so on, but I don’t think this is a good solution, considering that it only has one context example.

In both cases, it is much more effective to include anything that could strengthen your links to the piece of information you’re learning, like audio, images, etc, rather than just word in front and meaning + reading in the back.

I’m interested in knowing everyone’s experiences with different card types, especially those of you that have tried multiple methods over time and found what works best for you, using SRS as a complementary tool.

3 Likes

This is the main reason I don’t use sentence cards. I prefer to put the word on the front and the sentence on the back. That way I can easily refresh how the word is used when I get it wrong.

I don’t think this is that common though. I’ve only done this once I think (I usually just avoid studying overly ambiguous words since I have so much to learn anyway). In that case I put the context in Japanese in parentheses: (様子を)窺う versus(機会を)窺う

I think that approach would be ideal, but personally I find it incredibly difficult to add this context succinctly. The example above was a rare case where I got it to work well, and even this required collaborating with a few people to figure out how to do it right.

I haven’t actually tried making cards with sentences/audio/images on the front, so I can’t really say any more on the topic.

8 Likes

This is probably how I would choose to do it myself as well. I guess with many words with multiple meanings, they are very strongly related and could be understood only knowing one. Thanks for your input!

Oh sorry I meant on the back. I guess if you put it on the front it amplifies the issue of point number 2.

Ah, yeah. I have sentences and audio on the back of my cards. I never bother with images though. I’m grabbing the words from context, so sentences are easy, and audio is easy to find too, but images would be much harder.

2 Likes

I’m not sure if you use Anki specifically, but do you know about ShareX? It’s a tool to grab audio, video or images easily, and it has an option to export them to the clipboard where you can paste them to the cards you’re making. The site I linked on the first post has an entry to set it up. It also has an OCD tool which is alrightish.

For images alone you could also just make a cut with Windows + Shift + S if you’re on Windows, no idea on mac.

Just in case you weren’t aware of it and could make use of it :slight_smile:

I have several Anki sets (grouped thematically, like transportation, people, places, etc.), but all of them are Japanese->English with typing. The approach I use is the following:

  1. I read the word aloud and try to type its meaning in English. That way I bring out the emotional association I have with the word based on a verbal cue (me reading).
  2. If I don’t remember the reading, I check the furigana I added as an extra field to the cards.

I’m not sure English->Japanese cards would work, because as far as I remember, Anki sets are 1:1. I also haven’t tried audio or sentence cards, because that would be a ton more work and probably wouldn’t help as much.

Yeah, that would be like counting cards in a card game - trying to use logic and memorization to counter the system.

I think it depends on the way you create associations in your brain - is the mechanism more based on visuals (images) or sounds? The reason I read everything I note aloud is because I build associations verbally. However, there were words I learned through a mixed approach, but mostly because I had a specific image of these items in my head already :slight_smile: .

So anything that would help you personally strengthen the emotional connection between the word and its meaning would work.

1 Like

I pretty much agree with all the points you listed against sentence cards, plus

  1. They’re annoying :laughing: I don’t wanna read the same sentence over and over again, it makes the review process incredibly tedious. I just don’t like it.

I do have the sentence on the front anyway though - as an example, the front side of my cards looks like this:

card

I just usually don’t read it. I automatically don’t pay attention to the sentence if I don’t need it. I only read it if I can’t remember the word otherwise, for a better learning effect(?), and to check whether I remember it with the help of the sentence (I often do, which probably proves point 2)
I usually fail the card anyway in that case, though.
And I also read it in this situation:

Like, if I remember that I have several cards for the same word, and I think something like “this could be X or Y” I read the sentence to make it less ambiguous. I don’t have many cards like that, though.

I don’t really like images on my cards, but agree about the audio. I don’t know for sure but I feel like it helps my retention to put audio on the back.

1 Like

Yeah, I’m not personally interested in SRSing production, only recognition. For production I would rather just write entries for correction and speak with people :slight_smile:

Honestly I have no idea :joy: I guess I also expect this journey of self-discovery through Anki to teach me ways of how I work best. I do think though that the audio for the sentences on the back will be very helpful for pronunciation and listening practice.

This is interesting. Thanks for sharing! I wonder if there’s a way to hide text and reveal it with a click or a hover. That way you could have the hint hidden and when you come across those words it forces you to think of both meanings before checking which one it’s asking for.

Edit: there is!

1 Like

There is! Several ways.
The easiest, built-in way is to put this on the card:
{{hint:YOURFIELDNAME}}
Plan B (if you don’t like the design of it or something) is to make something similar yourself with html/css and/or javascript.
Edit: Looks like you’ve already found out yourself :grinning_face_with_smiling_eyes:

1 Like

Thanks for the explanation! :wink:

Sort of inspired by @Scylie I decided to revamp my cards a little bit and found this article: Adding Hints in Anki | ChrisK91

It’s basically just a bit coding, but nothing too fancy :slight_smile:

EDIT:
My cards look like this now:
image

2 Likes

Here’s what I’m currently doing with Anki. I have two decks. Here’s a typical Japanese → English vocabulary card…

155553450_503293877740403_8976528863726230294_n

I adapted Anki’s Japanese card template. The fields: English (copy-pasted from Jisho); the kanji meanings (always the “main” WaniKani meaning); the pitch accent name; a sound file (from Suzuki-kun). I only add words where I’ve encountered the kanji in WaniKani but the words themselves aren’t part of WaniKani (and I almost always stick to words that have the #common tag in Jisho). I know it’s a tremendous amount of upfront work for each card, but in my experience, it’s been a lot less “net” work. I can instantly see any part that’s tripping me up.

Here’s a typical entry from my kanji deck…

155458407_721977275169119_5366348596952174771_n

I made a list of the 156 jōyō kanji that aren’t in WaniKani as well as the five kanji that got pulled from the jōyō kanji in 2010. I’m adding them to my Anki deck whenever my WaniKani reviews and lessons are current. The left-hand column just indicates whether I’ve already added the kanji to this deck. I copy-paste the definitions and readings from Jisho and the meanings of the radicals from WaniKani.

Again, it’s a lot of upfront work but a whole lot less work in the long term.

I tried making a sentence deck once. I ended up stumbling over the same thing that’s been tripping me up in Bunpro: even when I’m trying not to, I’ll memorize the sentence rather than the concept. Someday I’d like to make myself a sort of Mad-Libs SRS that takes common Japanese sentence forms and populates them with nouns and verbs and adjectives at random. But that’s more of a far-future thing.

I tried it years ago and I think it was too stressful for me to gain anything substantial. It ended up with me trying to think of one specific word when it would’ve been better to just pick the word that comes to mind, just as we do with when we speak our native language.

I’m planning to make a new English → Japanese deck once I complete WaniKani. This time, I’ll make it a separate deck and allow for each English word to have multiple Japanese answers. Like for one card that says “situation,” I’ll accept 事態、場合、状況, etc.

3 Likes

That’s actually quite impressive! Did you create these cards by hand?
I’m now thinking that perhaps it would be possible to do something like this programmatically and scrape the results off Jisho.org or build a thin Web API around it. Maybe I’ll try something like that over the weekend in Python.

3 Likes

Not quite what you mean, but for fast and automatic card creation Yomichan is amazing. Probably the most useful vocab tool I know paired with Anki.
Inserting WK kanji meanings like on the card above wouldn’t be possible as far as I know, but apart from that you can automatically create cards with a lot of information with just one click (word, furigana, context sentence, audio, pitch accent, screenshot, …).
Using dictionaries with jlpt, WK level and common-word tags is also possible :smile:

4 Likes

Thanks! I did indeed do every single one by hand. I had the same thought, but I was never good enough at coding to learn how. But I’ve done some stuff with Python in the past (mostly automating some repetitive mesh-making steps in Blender), so I could easily see such a thing working. If you end up giving it a shot, I’d be interested to see what you make!

3 Likes

I’m guilty of ignoring my context sentences on the front because I’m lazy. I only end up reading them if I forget the word.

1 Like

I’ve (tentatively) settled on the word-sentence hybrid on the front.

Usually the word alone is enough for me. Sometimes the sentence helps. Once I have many cards, I figure I won’t mind needing context to help with a few words here and there. The main issue is if I (unintentionally) memorize a sentence.

I typically don’t even look at the back unless I fail a card. On the back I have:

  • Word/sentence from the front.
  • Audio of sentence (if taken from a source with audio).
  • Screenshot of source.
    • Often useless if it’s from an anime.
    • Useful if there’s relevant artwork in the same panel. If the sentence is in a panel with no artwork, I don’t bother making a card for it.
  • Image from web search of the word.
    • Mostly this is to supplement any useless anime screenshots.

Examples:

Manga Sentence Card

Anime Sentence Card

Anime (+ Image) Sentence Card

3 Likes

I’ve been making anime cards for a while now(November last year afair). and i did sentence cards before that for a month. all the points QM(the author of the guide makes) were true. for point 2. I’ve seen people argue that you can add the same word multiple times but it’s just going to take forever. sentence cards are already slow if you’re starting out adding a word multiple times is going to have consequences. my Anki sessions usually take 20 minutes and I add 10 new cards every day.

you’re supposed learn one meaning from the flashcards and the rest through reading. most important thing to remember is that you learn through reading/listening not anki. anki is just there to help you remember hard to remember stuff.

2 Likes

Fair enough. So you still put the other meanings in the same flashcard or only one?

i put all of them since i use yomichan…but honestly an approximate understanding is good enough for now, at least for me.

1 Like