About 10% of optically selected radio-quiet quasars have broad high-ionization absorption lines and are known as broad absorption line quasars, or BALQSOs. About 15% of these optically selected BALQSOs also have broad low-ionization absorption lines and are known as Mg II BALQSOs. Although uncommon in optically selected surveys, Mg II BALQSOs appear to be much more common in IR-selected samples and might be significantly reddened by dust. We examine the absorption-line properties of six Mg II BALQSOs from the BALQSO sample of Weymann et al. and find the following trends: (1) Al III absorption is always stronger than Al II absorption; (2) C iv absorption tends to be very deep and broad; and (3) low-ionization absorption tends to be narrower and to lie at the low-velocity end of the C IV trough. Assuming that the absorbing material is in photoionization equilibrium allows us to constrain the ionization parameter and total column density of the absorbing region quite tightly. The Al III/Al ii ratio limits the ionization parameter from below and the column density from above, while the Mg II equivalent width limits the column density from below. The Mg II-Al III absorbing region must lie behind a column of higher ionization material large enough to block the ionizing flux above the He II edge at 54.4 eV. Since the low-ionization gas ought to be optically thick at 13.6 eV, Lyman-limit observations of the quasar continuum can give the covering factor of the low-ionization absorbing region. Soft X-ray observations can constrain the Mg ii BALQSO ionization parameter from above. The fact that the Mg II and Al III absorption features tend to lie at the low-velocity ends of the high-ionization troughs indicates that the outflows in Mg II BALQSOs are not accelerating monotonically. They must be either decelerating or nonmonotonic. We show that it is possible to decelerate a hot cloud-confining outflow by adding thick absorbing clouds if the ionization parameters of these clouds are high. Soft X-ray observations can potentially rule out this possibility. More likely, the outflow velocities arise in turbulent regions in which there are many more clouds at low velocities than at high velocities. Such an absorbing region can be optically thick in Mg II only at low velocities but optically thick in C IV at both low and high velocities. A turbulent absorbing region will have a lower ionization parameter than a mass-loaded decelerating region, will not be so thick to soft X-rays, and will not strain the mass and energy budget of the active galactic nucleus so heavily. If the highly nonmonotonic, turbulent flow picture is correct, and if the heavily shielded regions also contain dust, then photons of resonance lines with fairly high optical depths, such as C IV, can be scattered enough times that they are preferentially destroyed by dust. If so, this would weaken the usual argument made against high covering factors in BALQSOs. We also discuss why the Mg ii BALQSO phenomenon might be a manifestation of a quasar's efforts to expel a thick shroud of gas and dust.
