Hi @ChristopherFritz! This works for me:
I wonder if a script is interfering
I suspected it might be a script, so I disabled everything except the undo script, refreshed everything, and it still didn’t work for me (leading up to the screenshot yesterday).
I’ll be sure to update the script, just in case. Sorry to make you check into it over something on my end!
Hmm, that is strange! No worries, and thanks for letting me know. I hope the shake works properly for you next time
@rfindley could this be caused by any of Double-Check’s features?
Considering that the Double Check script hijacks a lot of the shake code for its own behavior, I think it’s highly likely.
Update: We’ve added another scenario to this extension! Now if you accidentally enter the reading into the meaning field, you’ll get a shake and a message. If the meaning and the reading are the same (think tsunami or sushi!) then it will simply be marked correct as usual.
You just reduced the number of forum posts complaining about WaniKani by about 30%.
Now this I totally support!
Clarifying question: how does the system detect that the user tried to enter the reading? Does it just scan for the romanization of the reading of the item, or is it more involved than that?
To use an extreme example, if you get the vocabulary item 注釈, “chuushaku,” “tyuusyaku,” and “tixyuusixyaku” would all be converted to ちゅうしゃく for the reading entry. Does the system account for all of that if the user tries to enter any of those for the meaning?
I’m guessing the system was designed with all that in mind, I’m just curious as to how it works.
(No, I don’t know why someone would choose to enter the reading like that last one, I just wanted to illustrate how weird a reading entry could come out if put in the meaning field.)
The reading is detected based on WaniKani’s IME wanakana, so all of the transliterated readings you mention will produce a shake for 注釈 - and “tilyuusilyaku” too! (Though yes, I’m guessing that doesn’t come up a lot!)
The exception is when the reading matches or is close enough to a meaning (visible, allow list, or user synonym) to be marked as correct. So for example, “futon,” “huton,” and “futonn” are all still accepted for 布団, with the standard “Your answer was a bit off” message for those last two. But “hutonn” will receive a shake, because it would previously have been marked as incorrect (and is almost definitely the reading, not a typo of the meaning).
We’ve tested pretty thoroughly, but there are likely to still be quirks, so feedback is very welcome as always!
Neat! Thanks for the explanation.
Does it accept the answer if it’s a user synonym? I know there were previously some issues with this for the blacklist, so I figured it can’t hurt to ask.
Yes, user synonyms override all shakes/messages! The only meaning that can override a user synonym is a block list meaning, as you mentioned (That’s because if it’s on the block list we feel strongly about its being an unhelpful meaning to learn.)
Do you have a shake for that too? Might be confusing if your synonym is rejected
We don’t at the moment. I agree it could be confusing, but we looked at the data, and it’s actually very rare for a user synonym and a block list item to overlap, so we wanted to focus on more common scenarios first. We’ve also been improving the explanations on problem items, so maybe that will be enough. We are keeping an eye on it
Would it really? I thought in WK typing l (L) = r.
I just checked in a lesson and tilyuusilyaku produces ちりゅうしりゃく instead of ちゅうしゃく.
Bless you for adding the reading-in-meaning-field shake!!
At this point in my WaniKani adventure it’s the only “typo” I make, but I do it REALLY often.
Ah yes, it seems to depend on the keyboard you use to input romaji. On wanakana though, “tilyuusilyaku” definitely produces ちゅうしゃく, so you would get a shake for this. I don’t recommend getting into the habit though, since different IMEs render it differently!
That’s for the latest version of wanakana, but WaniKani is using a very outdated version that has bugs like this (and worse). I really would recommend updating to the latest version.
An L=R thing definitely sounds like a feature-not-a-bug in a completely literal sense.