An approach for associating signals from frequency-hopped (FH) emitters has recently been proposed that leverages polarization-frequency characterizations obtained from dual-polarized antenna architectures. An effective extension of these methods is considered here that incorporates multiple spatially separated dual-polarized antennas. Correlations in polarization-frequency responses, space-frequency responses, and hybrid combinations are leveraged to associate FH signals. The extended method is investigated using empirically informed simulations and over-the-air experiments to illustrate the efficacy of the approach. The over-the-air experiments involved four FH signals in a laboratory environment. A two-second interval was collected involving 2000 frequency hopping dwells. Results for a single manifold were found to experience misassociation errors, even at high SNR, due to similarities between polarization-frequency responses associated with the sources. When all manifolds are exploited, perfect association is achieved over the interval. Similar trends occur at low SNR, where increasing the number of available manifolds improves misassociation rates. It is concluded that additional diversities provided by a space-polarization architecture improve association performance.