Yeah, you could do that. You could probably reroute the audio output to the input, but I'm not totally sure how to do that, and you'd probably lose quality that way (EDIT: one way to do it might be to put an audio cable from your line out jack to your mic in jack). What I'd do instead would be to use memblocks to mix the WAVs together. If you don't know, memblocks let you access the raw data from different types of media.
If you create a new memblock (MAKE MEMBLOCK) that's big enough to store one whole loop, you can take smaller sounds and convert them to memblocks with MAKE MEMBLOCK FROM SOUND, and then take the data from those and stick it into the new memblock at appropriate points. If they overlap you can mix them together by simple addition, I think (as long as you make sure the value doesn't go above the maximum "clipping threshold").