Mark correct but still correct the user?

Hi, short time WK user (2.5 years), long time WK forum ****stirrer (5 years).

So you guys have the ability to detect slight mispellings and let things slide but notify the user. Why not do the same with synonyms and just tell the user “Hey it’s actually this but you were close enough!” I think the harshness of WK is largely why people give up especially like me in the later levels. Like "no this vocab has and never will come up again, but it is taking up more of my time because I got it kind of wrong.

Anyways, that’s all. I probably won’t deactivate vacation mode anytime soon but just a thought.


I’ve been using a script to let me mark something correct when I haven’t set a synonym up yet on something. I can still see that the answer I put was wrong, but can decide whether or not I was close enough to pass the vibe check.

Edit: just realizing you posted this in feedback, not necessarily looking for feedback yourself. My mistake. Gonna leave this up though in case anyone stumbled across it later and would be helped by it.

1 Like

Because the elves who are actually marking your reviews would need to know a word is a synonym, and teaching them every possible synonym, every regional variation, every “aww, I reckon that’s close enough” option for every single item would be… infeasible.

1 Like

I think that’s a perfectly valid answer but might be a bit old. Today even with an open source LLMs it would be quite simple to detect synonyms. Of course that’s still not super simple to integrate it into the tech stack, but if WK team wanted to, they could add it in maybe 3-6 months.

1 Like

WaniKani has a feature for this already: each item has a hidden “allow list” with synonyms that are accepted by the review elves but not shown on the item’s page. Similarly there’s a “warning list” of answers which play the “shake” animation and ask you to try again, but don’t penalize you.

The items seem to be handwritten, and you can contact support if you have suggestions. (Content Updates posts almost always include a few allow list changes.) But it might be fun to try using AI to generate them!

1 Like

WaniKani has the warning message, Please check if answer is correct, but doesn’t really show all correct answers, unless Show Info is pressed. Also it doesn’t allow turning Right to Wrong.

A script to auto-show correct answers and synonyms would be safe, but it doesn’t really tell why the answer is unexpectedly warned but right.

If WaniKani has found which right answer is the nearest, it can still normalize to the standard counterpart, no problem here.