The UX for context sentences is not ideal for learning.
At least for me, I always skip them. They are too small to read and it seems just uncomfortable.
How to fix that:
Make font bigger and more pronounced.
Add an audio reading for EACH context sentence.
Allow easy playing of audio feedback for each sentence with J (it would cycle through them each time you press J just like with kanji readings)
IMPORTANT: have a yellow highlight run through the kanji as they are being read in the sentence via audio. That way I can press “J” repeatedly and can see the kanji read out loud real time and highlighted. You can highlihgt normal characters in yellow and the kanji being learned in red or something like that.
Largest amount of work would be getting the audio readings, but its not impossible. Since most likely you are using LLM to automate coding already, why not allocate more time, effort and budget to polishing the readings to perfection?
This, at least, should not be too hard to implement and would be a big improvement.
As for other suggestions – I quite agree it would be great if they were implrmented, though I’m not sure how feasable it would be…
Even simply adding furigana would help a lot…
I remember, my biggest problem with context sentences when I started was that they all contained unknown kanji (some of which were tought later and some are not on WK at all), which I couldn’t even pronounce…
Something I would LOVE to see is to hide the English translations under a “click-to-reveal” type of box. I think it would be great to have a chance to read and try to decipher the Japanese sentences before my brain automatically processes the English text.
Have you tried the script for Context Sentence examples?
It provides audio (via on device TTS). Colours the kanji (differently for the now rare non wk ones) and makes them a hyperlink to their wk page.
Plus a link to ichi.moe analyser.
I think it used to move a highlight through the sentence but it isn’t now on my Samsung tablet.
I guess the quality and accuracy of the TextToSpeech will vary with what is available on your device.
The reason I’m not using any of those extensions because its another barrier and becomes increasingly unsafe to use external extensions, etc.. but thanks, I will check them out for sure
I think ideally wanikani service should integrate anything that could help us optimize learning.
In such cases, I use Winamp, because it can be set to be “Always On Top”, so that it would keep blocking whatever’s behind it even when you switch to another app
Just my opinion, but machine-generated audio reading, and also machine-generated Furigana, can have minor (?) inaccuracies; that make me unsure if the whole plan is ideal for learning. (It might be good for some users at some learning stage, but might not be good to others.)
If the specific inaccuracy is important enough, however few, it might not be that minor.
Nothing against native-human sentence audio and Furigana, nonetheless. I suggest you ignore WaniKani’s, and use something like Satori Reader if you care about audio.
WaniKani context sentences would be better with hide the English translations under a “click-to-reveal”. I want to see as a native feature too.
Not sure how you feel about Web Extensions like Yomitan or 10ten. They are generally safer than User Scripts.
I will use a smartphone app, typing or OCR by myself, if I have to.
My biggest problem with reading before WaniKani wasn’t directly all those unknown Kanji, but rather all those unfamiliar vocabularies, and the amount of time I have to use a dictionary for a page of reading.
tbh, finger writing and radical/Kanji lookup is a problem too. But trust me, diligence will give something in return.
And even after using a dictionary, it surfaces another problem – piecing together all those vocabularies.
Huh. May have to try that userscript. So far I’ve been running WK at 150% size in a somewhat small browser window (Not small enough to trigger phone layout usually but I keep them fairly small in general), so the first english sentance is almost always off the bottom of the page and I can scroll them into view as desired… but it’s a nuisence.
I use the media context feature all the time in Kakehashi. I’m constantly finding errors in the LLM generated furigana where they don’t match the anime sound clip. The TTS for the context sentence also has mistakes constantly. I still appreciate the features because I know it’s an unofficial third-party app but having those sort of errors on a paid educational tool would be unacceptable.
Would it be nice to have real human audio recordings and human generated furigana for all the context sentences? Yes absolutely. But I’m also fine just pasting words into jisho and checking that way. It’s part of studying and reinforces the information.
The larger font and color highlighting should be easy and cheap to implement so I’ll thump the table with you on that.