People don’t read individual letters, we read the word. When there is an ambiguous word, we read the whole sentence, or the whole paragraph. So, I am suggesting something like:
The system should prioritize showing a word with all the kanji the user had learned previously, but I think it will still be fine if the system was not able to do that.
For vocabulary:
Lower level user may not be able to read the whole sentence, but having know at least one other word in the sentence helps a lot right? Just show the furiganas and word-by-word translation after the user attempted an answer, and then the user can learn how to use the word!
The same thing can also be done for radicals, but I’m not giving example because you get the idea already.
Wikipedia can be used for example sentence extractor – it’s even free for commercial use.
Or create some business proposals, with https://www.kanshudo.com/ for example.
What do you think?





