Previously, five phonon energies were used to explain 16 peaks in the measured lattice absorption spectrum of GaAs. It is shown here that the five phonon energies are not unique and that a different set of values is also compatible with the available infrared data both with respect to the absorption peak positions and the temperature dependence of the absorption. A similar difficulty is also noted for some previously published work on CdS. It is suggested that difference band absorption measurements might show which sets of phonon energies are correct.