Oh shit! Good for them. That’s an awesome feature
2 Likes
Looks incredibly promising! I’m eager to see how it will turn out.
1 Like
This really should be their regular feature, not supplement in cram IMO. I went through BunPro as best I could but haven’t touched it in years…I ended up making a deck for compression for both listening and reading, I found it far more efficient….basically taking any English out of my study routine as much as I could, learning apps should be shooting for this as level difficulty increases. I also made a deck similar to this proposed idea, but I used a hint system and you have the AI bunpo check if need be…but I’ve used it very selectively to troubleshoot grammar production and hardly a full-time grammar approach.
The fundamental issue with English que approach and what I see on this thread’s proposal is that you are basically training students to be translators, not a goal for everyone…WK has this problem too to some extent asking for precise English translations, and why SRS is not always a great method to learn vocab, just look at anyone’s leech count. That is fine in the beginning, otherwise, you can break the language code…but less so once you are in an intermediate period. When interacting with the language, whatever it is (listening, reading, speaking…) you can’t buffer two languages nearly quick enough, so I’m not sure why learning apps use the same learning method for all levels of difficulty; their method to teach is exactly the same for N5 to N1 and beyond…your needs evolve as your language acquisition evolves, or you just go towards native material only and grind it out, which I think is what most end up doing. And that is fine but I just don’t see many learning apps attempting to do this more efficiently or tackle the intermediate plateau better. And really the good way to learn vocab is eventually just context, and not in isolation.
I thought 日本語の森 is a better product, they teach course N3 and above all in Japanese and use Japanese to define grammar points, their explanations are simple so you build confidence in your listening. One of the few resources that uses Japanese to teach Japanese by natives that I know of.
The reading review type is available for both regular reviews and cram. Therefore, the reading review type is a regular feature (not just supplement). However, cram does have more options such as listening (which may or may not be similar to your self-made decks).
Here are the review settings:
Here are the cram settings:
1 Like
Thanks for sharing that update, I’ve been out of the loop, the fill-in was a mainstay for a while but that is good to know. Can I assume the SRS method is the same? That was another item I was not a fan of…you can’t proceed to the next sentence until the first is passed, but the more angles you can see a grammar point the better, especially at the same time. I’m more or less using an interval method similar to WK, where each. sentence pass 8 times, rather than pass it the first or second time and never see it again. It’s not bad because reviews are a lot faster rather trying the translation fill-in method.
Regarding the OP proposed method, though I’m not a fan of translation exercises, productive grammar is still a helpful method (or at least makes it more fun IME). I’d be more interested in scenario-based production, where I can interpret how I please but use the intended grammar point (and keep the questions in Japanese), so it’s not bridging an English connection and letting the AI correct it.
Could you detail your scenario based production method out of curiosity? We feel as if our grammar point and situation combo method is most balanced for learning but also usability.
I think it could be anything; Japanese audio, picture, response to a question…I don’t have a specific scenario but anything that would simulate productive Japanese in the real world would be my thought…as long as it gets away from a translation exercise from an English subtext. I’ve also made decks that use TTS and audio shadowing, a decent threshold to help with pronunciation independently, which can also be used for production. I get a translation exercise probably easier to manage…or not, some of the auto AI translations I’ve come across for media are not even usable.
As mentioned, I get beginner learners need a translation bridge because you have to start somewhere, but haven’t seen an app yet that dynamically adapts to a learner’s progression to be equally useful from beginning to end…in other words, using the same exercise method from beginner through advanced doesn’t work well IMO, and why I see diminishing returns on both WK and BP the more it gets used. I get that a balanced approach helps with design architecture, you pick a method and fill it with content but can also be a ‘one-size fits no one’. A caveat is that a solid beginner base seems for better for business…makes sense you will have a lot more raw beginners than capturing life long learners, which probably explains why the aforementioned beginner content is solid and less for advanced for successful apps, and many just end up consuming media more due to diminishing returns. But food for thought: what was interesting on the 日本語の森 site was seems a majority of their customer base are Vietnamese and no one is using English on their site or board…no one cares to use English and their team is Japanese so they actually widen their audience and make their app more practical at the same time.
2 Likes
Excited for this new srs for grammar!
I used bunpro for one year, from N5 to half N1 and actually from N2 it became very boring and no explanation for the differences in the answers.
Going to their forum the only explanation for “this answer instead of that answer” was always ‘nuance’. (table flip feelings)
I think the SRS method is the same as you experienced before. There are approximately 10 or more sentences per grammar point. You will only see one sentence per review of a particular grammar point. If you get that sentence correct, you’ll move up one SRS stage and see a different sentence during the next review of that grammar point. If you get it wrong, you’ll go back an SRS stage (I’m not sure if you can jump back multiple stages like on WaniKani) and that sentence becomes a ghost (sentence repeats). Then, after the lower SRS stage’s next review time (see review table below), you’ll see a new sentence.
Ghost sentences allow for repetition of the exact same sentence. If the Ghost Review setting is on, then after you review the same sentence correctly 4 times (progressing through the 5 ghost SRS stages in the ghost review section of the review table below), the ghost is slain. Some users like ghosts, and others do not, so there are other ghost review settings shown in this image:
Also, if you run out of sentences before mastering a grammar point, I think you’ll see non-ghost sentence repeats. This will happen for grammar points you struggle with and/or grammar points with less than 12 sentences (because it takes 12 reviews to progress through the 13 SRS stages on the chart below from the old Bunpro FAQ Post → What is Review SRS? → Learn more about SRS intervals → Bunpro SRS Chart):
The stages (Review SRS 0 to 12) correspond to Beginner 1/2/3, Adept 1/2/3, Seasoned 1/2/3, Expert 1/2/3 and Master. For ghosts, the stages (Review SRS 0 to 4) correspond to Ghost 1/2/3/4 and Ghost Exterminated (or Ghost Slain).
Edit: formatting, correction of numbers, added stage names for reference
2 Likes