When entering a sentence, I had used the word 奇麗 in kanji, but the parser wouldn't read it. It kept splitting the two kanji and wouldn't parse it correctly. Entering it just in hiragana fixed it, and then showed the word in kanji after parsing.
Where specifically are you entering the text (dictionary, text analyzer, etc.), or the regular sentence parser? Can you give the full line of text that it was part of?