Feature request: Listening Practice

[jeff8v7]

This is completely random... but it would be great to have a quick listening quiz at the end of the day, to improve listening and pronunciation retention. Something like iKnow Rapid Choice (speed learning with simple quizzes).

The audio for each recent vocab would be played and you're given 3 seconds to choose the meaning from 2 choices. Missed items go back in the queue. It takes 4-5 minutes to go through 100 items, so it's minimal extra study time, but really seems to help with retention.

I'm doing this now with iKnow and custom wanikani courses. I've noticed that words I've studied this way more easily pop into my head and “sound right” when I’m doing wanikani reviews. I also notice the words more during Japanese conversation.

Would be a great feature which does above and beyond wanikani. Since wk has audio for vocab, it's just a way of reusing the data to improve retention and listening skills. Could be useful for leeches too.

Here's a screenshot from iKnow. I would go with "listen only" and not even display the vocab kanji.

[UInt2048]

You wouldn't be able to sync this with anything, though, so what happens if you want to pick up where you left off on another device — such as your iPad instead of your phone where you normally do it?

Edit: Just to be clear, I completely support this. After reading the comments below, showing how an implementation would be possible without syncing, I would love for this to be implemented!

[skymaiden]

I would love some kind of extra quizzing on Tsurukame, even if it's local only (no syncing). I recently installed the Self-Study Quiz script on desktop for leech training. But on days where I'm stuck on my phone (happened recently for about 2 weeks), I don't have access to scripts outside the ones included in this app...so no leech training possible 😢

[jeff8v7]

You wouldn't be able to sync this with anything, though

I don't think a sync is needed, because there's no additional state to maintain.

The listening review queue could be derived from their current wanikani vocab Reviews for the vocab they are currently being studied. For example (just a quick idea, not a well thought out or tested algorithm)

All vocab below "Enlightened" sorted by WK level (highest first), then filtered by SRS level, SRS Level 1 - 1:1 chance of being in this review queue SRS Level 2 - 1:2 chance of being in review queue SRS Level 3 - 1:3 chance of being in review queue etc.

This would always give a unique but meaningful review set, even if a user does simultaneous listening reviews on different devices, because the queue is wk progress relevant but random.

Just for fun... I looked at how this would pan out for my current review queue. For me, this algorithm would give about 230 listening review items, or approx 10 minutes (if you don't make many mistakes)... less actually because I counted radicals and kanji, but you only need listening practice for vocab.

You don't need to store correct/incorrect guesses either, because the goal is just to gain listening exposure.

The listening review session continues until you bail out or have answered all the reviews successfully. If you answer incorrectly, that review item is asked again 5 words later. Could pause at 2, 5, 10 min to say "Wow! You're kicking it! 23% done. Want to keep going?"

Anyway... probably way more detail than needed! Point being... this could be totally useful without maintaining or syncing any state. :-)

[jeff8v7]

@skymaiden But on days where I'm stuck on my phone (happened recently for about 2 weeks), I don't have access to scripts outside the ones included in this app...so no leech training possible 😢

That's why I was thinking app instead of script too... Suppose a stand alone app would be fine, but tsurukame is great, so why not just tuck it in there :-)

[bmironer]

Might be a good fit for the Apple Watch app too.

[jeff8v7]

Might be a good fit for the Apple Watch app too.

Definitely - no typing needed - just listen and choose A or B. Easy way to do a little extra practice anytime.

[jeff8v7]

If anyone is interested in developing this, I'd be happy to send you some cash!

[UInt2048]

@jeff8v7 Is that a bribe?

David Sansome does it, and refuses the donation like always

[jeff8v7]

LOL - I just don't have the time/expertise to do it myself, and it's only useful to me while using wanikani (1 year to level 60!).

I'm actually thinking of hiring someone to make a stand-alone app, but it seemed more useful to others if it was in tsurukame.

The listening practice is really helping my retention, so I'll keep using iKnow for this meanwhile. But using iKnow is more cumbersome, because it doesn't know my wanikani progress.

[UInt2048]

@jeff8v7 Definitely agree, I would love this to be within Tsurukame.

Speaking of A or B, where would the other choice come from that's not the correct answer? Obviously which one is the correct answer would be randomized, but there does need to be another option that's not completely ridiculous, and not a homophone where it's not fair...

[jeff8v7]

Good question... might take some tweaking to find the sweet spot.

I think a good 2nd choice word could be selected randomly from the same wanikani level (i.e. Level 5) & SRS level (Apprentice 4). Presumably this would be somewhere nearby in their brain's neural network, even if the meaning isn't close.

If it's too hard, it's ok. They could a different comparison word in their next try. Or just stop asking them after a few missed tries. This is just a listening bonus, not the actual SRS system, so it doesn't have to be precise.

[skymaiden]

@turtlemaster19 It would be nice if this feature had a typing option as well as multi-choice, to ensure proper recall 😊(I don't like multi-choice, it makes me lazy)

Originally I thought this feature could be grouped with a leech training feature (which I desperately need on mobile), if it was based on the popular Self-Study script. Should I instead create a new issue for leech training?

[UInt2048]

@skymaiden Definitely agree. Leech training, however, is at #332. :wink:

[jeff8v7]

@skymaiden I don't like multi-choice, it makes me lazy

LOL - For memorization & recall training, I totally agree. Typing is important. Multiple-choice doesn't cut it.

For listening & recognition training, however, I totally disagree. Multiple choice is important. Typing doesn't cut it.

This has nothing to do with laziness. I has to do with neurolinguistics, and how our brain processes language.

Listening recognition needs to be fast (listening speed, not typing speed). It doesn't need to be as precise, because recognition is always in context of the surrounding words, sentences, conversations, relationships, and environments – all of which your brain is processing very quickly. On top of that spoken language is rarely precise. (i.e., People talk funny)

Reading and writing are slower and more precise. In my opinion, wanikani is excellent for recall, but not so good for recognition (based on level 60 users who can't understand Japanese conversation).

That's my goal.. add some listening recognition training to wanikani!

You can try some of these ideas out by mixing recall and recognition activities. Try to memorize words with only multiple-choice – good luck, it'll take forever. Trying to transcribe a conversation in real time – good luck, not possible. Or read a book then watch the same movie – very different experiences in speed and precision.

Sorry for my super long answer... In practice I've found that multiple-choice listening recognition practice actually helps with my leeches too. Language is funny that way :-)

[jeff8v7]

@skymaiden Have you tried inventing your own mnemonics for your leeches. That's working pretty well for me. It takes some time, but I tend to remember them well after...

[skymaiden]

@jeff8v7

For listening & recognition training, however, I totally disagree. Multiple choice is important. Typing doesn't cut it. This has nothing to do with laziness. I has to do with neurolinguistics, and how our brain processes language.

I'm not a neurolinguistics specialist. But I do know that when I work on my active audio recall with typing (with the Self-Study Quiz script), I don't need the option of choosing an answer from a random list of potential options. I work on pure recognition when I listen to native Japanese content, which I find more useful and fun personally.

Reading and writing are slower and more precise. In my opinion, wanikani is excellent for recall, but not so good for recognition (based on level 60 users who can't understand Japanese conversation).

I don't know the proportion of level 60 users who can't understand audio content, but I've seen a lot who seem pretty good because they consume a lot of native material alongside WK 😊

Have you tried inventing your own mnemonics for your leeches. That's working pretty well for me. It takes some time, but I tend to remember them well after...

I do make new mnemonics whenever I need to, but the further along I get in WK the more leeches I find myself with – because of similar meanings or graphemes. I'm only at level 17 right now, but working specifically with a leech trainer has helped me immensely since I started using it. The built-in SRS intervals of WK don't fit my brain, I need more repetitions for things to stick.

[jeff8v7]

All good points :-) We enjoy different types of learning.

I like the urgency of a 3 second timer & multiple choice for listening recognition, and the ability cruise through 100 reviews in 4-5 minutes, without stopping for typing or typos.

I like typing for wanikani recall practice too :)

[jeff8v7]

This listening practice request could be simplified to the following:

After each lesson or review session, (optionally) do a listening practice on the vocab just studied.

Some specifics:

• Only use items in this session for comparison (option B)
• If possible, use a similar item (containing matching Kanji) for comparison (option B)

Remember, this is just listening practice on the words you just studied. It's not a "test" or "review" or "lesson", so it doesn't need to be complicated or comprehensive. The goal is simply to give your brain a little more "comprehensible input" to add to your Japanese neural network.

[UInt2048]

@jeff8v7 I beg to differ on the after each lesson/review part. I believe it should be its own category distinct from lessons or reviews, especially since lesson sessions are between 3-10 items, and when you finish a review session, you usually want a break. 🤪 I agree on using items recently in lessons and/or reviews.

[jeff8v7]

I love this as a stand alone feature too. Maybe with an after lesson/review option :-)

Great point about being tired after reviews. Since everything is fresh in your head, it should be quick & easy, but later would be fine too.

I didn't realize 3-10 items is a normal lesson schedule. I do 20-40 every morning :-) A really short listening review is still a great way to reinforce what was learned.

But, yeah, for most people, you might be right. A stand alone "listening" test may be better.

[UInt2048]

@jeff8v7 LOL, I meant that until #13 is implemented - @mzsanford might actually finish soon - it's not possible to leave the Lesson screen with more than 10 new lessons done.

I'm glad that you agree now, though!