The emission-line properties of 45 intermediate redshift (z approximately 2) steep spectrum quasars are presented and compared with those of a sample of flat radio spectrum sources of comparable mean luminosity and redshift. It is found that the steep-spectrum sources have, on average, smaller widths of the C IV lambda-1549 and C III] lambda-1909 line profiles, larger Mg II lambda-2800 equivalent widths, and smaller (He II + O III])/C IV intensity ratios than the flat-spectrum sources. The steep-spectrum sources also display C IV profiles that tend to be asymmetric to the red, whereas those in the flat-spectrum sources are found to be asymmetric to the blue. These line parameters (with the exception of the C IV asymmetry) appear to be weakly correlated with the 6-11 cm radio spectral index in the combined sample of steep- and flat-spectrum sources. The Baldwin effect and the correlation between the Mg II-C IV and C III]-C IV redshift differences reported elsewhere are also present both within the combined sample and among the steep-spectrum objects alone. These results support the proposed unified models of steep and flat radio spectrum quasars. Given the sensitivities of the associated line parameters to luminosity, the senses of the correlations with the radio spectral index suggest that the ionization of the broad-line region (BLR) gas is anisotropic, such that it is most strongly ionized in the direction of the radio axis. This is consistent with the evidence that the continuum emission is similarly anisotropic and that the extended narrow-line regions of Seyfert and radio galaxies align with their radio axes. Such ionization anisotropy can in principle reconcile discrepancies between the BLR sizes implied by variability studies and the expectations of standard photoionization models, although quantitatively such reconciliation is model dependent. The sense of the correlations found between the radio spectral index and the widths of the C IV and C III] profiles is opposite to that implied by the results of previous studies of the H-beta profile in low-redshift quasars. The interpretation of this is unclear, but the effect may be related to the different mean luminosities of the respective samples.
