Studies on the genetic relationship of Japanese with other languages have usually been based on lexical evidence foremost, which obviously poses problems if indeed each word has its own history. The search for external cognates of Japanese etyma cannot be limited to the comparison of attested written forms but requires a thorough reconstruction of the internal history of the languages involved as a preliminary step. I will propose a more comparative approach to Japanese etymology and give some more examples of how the Ryukyuan data, and to a lesser extent the data from Ainu borrowings, can clarify our knowledge of the (pre)history of the Japanese language.