Why not horns?

Phase shifts can occur because the drivers are different distances from your ears. This can be corrected physically with driver placement. That's one reason you see speaker boxes that slope back to get the tweeter voice coil aligned with the woofers. That's pretty easy to do or you can do the digital delay. If you look at various front loaded horns you almost always see that they are built with the mouths of the horns different distances away from the listener to align the voice coils.

The harder problem is phase shift caused by crossovers. In a simple first order crossover as you get close to the cutoff frequency you also start getting a phase shift. At the cutoff frequency there is a 45 degree phase shift. Different frequencies get shifted by different amounts. There are other filter configurations that have less or more phase shift but each is a compromise in some way.

I use a digital crossover and delay that they say has no phase shift but I tend to believe it must be screwing with the signal in some other way. I don't know enough about digital filters to prove that but I'm in the "you don't get something for nothing" camp.

.

Herman, I leave the design of speakers to someone else to prevent headaches. I was just pointing out that the crossover can be used to correct time alignment (sometimes this results in the need to reverse the phase of certain drivers as a result), and in fact a crossover will introduce some time alignment issues that might have to be otherwise addressed. I think its a bad idea to rely on that exclusively- doing some physical alignment is important too.

Phase coherency, or lack thereof, is best evaluated by a speaker's ability to reproduce a square wave. The only speaker that I know of that does this well is the original full range Walsh driver.

Eldartford, I believe that some speakers from Dunlavy, Green Mountain, Quad, Vandersteen and Thiel can do a fairly good job of reproducing a square wave too.

Personally I love my ribbons on my Apogees, they just sound right