They probably could, though it seems a little overboard
WK still uses an old version of WanaKana, as they probably don’t wish to risk upgrading (if it ain’t broke…).
Though they wouldn’t need to for this, like you suggested they can just add a regex check on the answer submission after conversion anyway (since they may as well do the same for english/meaning answers too).
For reference, these (plus english punctuation) could be used to throw the “yellow invalid input shakey shake”.
Unicode JA punctuation ranges (excluding
ー [長音符] &
const ZENKAKU_PUNCTUATION_1 = [0xff01, 0xff0f];
const ZENKAKU_PUNCTUATION_2 = [0xff1a, 0xff1f];
const ZENKAKU_PUNCTUATION_3 = [0xff3b, 0xff3f];
const ZENKAKU_PUNCTUATION_4 = [0xff5b, 0xff60];
const ZENKAKU_SYMBOLS_CURRENCY = [0xffe0, 0xffee];
const KATAKANA_PUNCTUATION = [0x30fb, 0x30fc];
const KANA_PUNCTUATION = [0xff61, 0xff65];
const CJK_SYMBOLS_PUNCTUATION_1 = [0x3000, 0x303b];
// exclude 0x303c for WK = `〜`
const CJK_SYMBOLS_PUNCTUATION_2 = [0x303d, 0x303f];
I can still see an argument against this if for any reason new vocab could contain symbols.
They wouldn’t want to block a valid answer.
Personally, I don’t really think it’s worth adding.