Logging in just to tell you that today I found the fifth example of a very rare but apparently regular sound change in North Halmahera

Proves again that the best method in historical linguistics is still staring at the data for a really long time