For awhile, I've thought that my Si4707 breakout board was damaged. This was because it would show up on all I2C addresses after a "normal" boot up sequence.
I recently received a Si4700/Si4735 evaluation set. In experimenting with that, I discovered that the same boot up sequence results in the same output. I've since modified the connections (7-wires vice 4) and the boot-up sequence (reordered steps and added manipulation of the extra pins) and the board now shows up at address 0x63. I'm not yet sure that this normal (not much time to check) but it shows up in some other people's documentation, so I'm hoping... If it works, I'll go back to the Si4707 and retry.