We analyze the kinematics of six Virgo cluster dwarf early-type galaxies (dEs) from their globular cluster (GC) systems. We present new Keck/DEIMOS spectroscopy for three of them and re-analyze the data found in the literature for the remaining three. We use two independent methods to estimate the rotation amplitude (V rot ) and velocity dispersion (σ GC ) of the GC systems and evaluate their statistical significance by simulating non-rotating GC systems with the same number of GC satellites and velocity uncertainties. Our measured kinematics agree with the published values for the three galaxies from the literature and, in all cases, some rotation is measured. However, our simulations show that the null hypothesis of being non-rotating GC systems cannot be ruled out. In the case of VCC1861, the measured V rot and the simulations indicate that it is not rotating. In the case of VCC1528, the null hypothesis can be marginally ruled out, and thus it might be rotating although further confirmation is needed. In our analysis, we find that, in general, the measured V rot tends to be overestimated and the measured σ GC tends to be underestimated by amounts that depend on the intrinsic V rot /σ GC , the number of observed GCs (N GC ), and the velocity uncertainties. The bias is negligible when N GC 20. In those cases where a large N GC is not available, it is imperative to obtain data with small velocity uncertainties. For instance, errors of 2 km s −1 lead to V rot < 10 km s −1 for a system that is intrinsically not rotating.