We present a framework for the interpretation of the far-infrared spectra of galaxies in which we have expanded the model parameters compared with previous work by varying the ionization parameter $U$, column density $N$(H), and gas density at the cloud face $n$(H$^{+}$) for a central starburst or AGN. We compare these models carried out with the $Cloudy$ spectral synthesis code to trends in line-to-total far-infrared luminosity ratios, far-infrared fine-structure line ratios, IRAS colors, and OH and H$_{2}$O column densities with trends found in the well-studied sample of ten nearby galaxies from the IRAS Bright Galaxy Sample with infrared luminosities greater than 10$^{10}$ L$_{\odot}$ and IRAS 60 micron fluxes equal to or greater than that of the nearby ULIRG Arp 220. We find that the spectral sequence extending from normal starburst-type emission line spectra to ULIRG-type absorption-dominated spectra with significant absorption from excited levels, can be best explained by simultaneously increasing the hydrogen column density, from as low as 10$^{21}$ cm$^{-2}$ to as high as 10$^{24.8}$ cm$^{-2}$ or greater, and the ionization parameter, from as low as 10$^{-4}$ to as high as 1. The starburst models best reproduce most of the sequence, while AGN models are somewhat better able to produce the high OH and H$_{2}$O column densities in Arp 220. Our results suggest that the molecular ISM in ULIRG-like, absorption-dominated systems is located close to and at least partially obscures the source of power throughout much of the far-infrared, which must be taken into account in order to properly interpret diagnostics of both their sources of power and of feedback.