Feature suggestion: "That looks like a typo" for -u/-i mistakes

WaniKani Feedback
1

It looks like the general “you’re missing visible kana” feature I would like has already been deployed and then subsequently removed about a year ago - that was before I most recently restarted WaniKani, so I was not aware of this until searching the forum just now. It seems like that feature had some issues leading to it being removed. I propose a similar but very limited feature that should be relatively straightforward.

Suggested feature

  1. When giving an answer for the reading of a vocabulary item,
  2. the item in question is a verb,
  3. all but the final kana in the answer are correct, and
  4. the last kana has the correct consonant but the vowel is -i instead of -u

the “That looks like a typo” message should be given instead of marking the answer as wrong.

This case should be fairly simple to detect if it’s known that the item is a verb. If this information is not already available, an item could for example be classified as a verb if

  1. it has at least one meaning starting with the string "To " and
  2. the last character in the item is a u-row hiragana (or more specifically, one of the ones that verbs do actually end with).

Rationale

Accidentally typing “i” instead of “u” is a fairly common mistake, especially on a phone keyboard. In fact, “i” and “u” are adjacent keys on all common keyboard layouts I’m aware of (QWERTY, QWERTZ, AZERTY and even Dvorak). As the correct kana is visible and the rest of the reading is correct, this kind of error is almost always a typo. Most of the common keyboard layouts do not place other vowel keys adjacent to “u”, so the check can be limited to just “i”.

This check should have relatively few false positives, and any false positives would simply benefit from the same typo protection as the intended case. False negatives would simply result in the current behaviour where the answer is marked wrong. The check should also be straightforward to test automatically.

In principle a similar check would be possible for i-adjectives and “nouns that look like the -i form of a verb”, though this would most likely require more work.

3 Likes
2

It’s displayed in the “Word Type” attribute, so it’s certainly part of the database.

Your best chance of seeing this implemented is to ask for a volunteer to make or tweak a userscript, however. I’d personally recommend using an undo script to make typos a non-issue more generally.

5 Likes