The oscillatory flow birefringence (OFB) properties of a variety of polymer/solvent systems are considered, with particular emphasis on the phase angle behavior (relative phasing of the birefringence with respect to the shear rate) at high effective frequencies. For polymers with large, optically anisotropic side groups and negative polarizability anisotropy, the birefringence is observed to change sign at high frequency, in contrast to the prediction of straightforward chain dynamics treatments such as the bead-spring model. The systems for which this anomaly is reported include polystyrenes with molecular weights ranging from 2100 to 44 000 000 (both linear and branched), three poly(α-methylstyrenes), and one poly(4-vinylbiphenyl) sample in the viscous solvents Aroclor 1248, Aroclor 1254, and tricresyl phosphate. For these solutions the high effective frequency phase angle behavior is essentially independent of concentration and molecular weight over the rather wide ranges covered; on the other hand, the time scale on which the anomaly is detected is proportional to solvent viscosity and is sensitive to the detailed side-group structure, though not to the overall chain topology. Initial measurements on an additional poly(α-methylstyrene) sample did not display the anomalous sign change, but remeasurement after 5 years revealed OFB properties consistent with those on all other poly(α-methylstyrene) and polystyrene solutions measured. This apparently extremely slow approach to full solvation is attributed to microcrystallinity, which may be eliminated by reprecipitation and freeze-drying. In addition, preliminary OFB results are presented for a tetrachlorinated polycarbonate of bisphenol A in Aroclor 1248; this polymer has no significant optically anisotropic side group, and no evidence of a sign change in the birefringence is observed. Possible sources of a positive birefringence which could account for the observed anomalous behavior are considered. These include internal field effects ("form birefringence"), uncertainty in the solvent birefringence contribution correction procedure, various kinds of polymer-solvent interactions, and quasi-backbone-independent orientation of side groups in the flow field. Only a side-group orientation process or a substantial modification of solvent properties upon the addition of polymer appears to be consistent with the observed behavior.