@ryotat Thinking about tokenization for Japanese search. One question: do you normally search for kanji only or also for inflected verbs?