Im sure what Wayne is referring to is possible unwanted phase changes caused by irregularities in the surface of a rough horn wall, causing frequency response abberations. As to foam in the waveguide, the foam would be absorbing the high frequencies (depending on how high you use them), and changing the frequency response more or less to your liking. In designing a horn you generally assume smooth wall's as it would not be easy to accurately model horn surfaces with random rough walls. On that note rough walls could be used to ones benefit but would probably take a lot of measurements and trial and error, and theres usually better ways to fix a problem such as rough frequency response than this.