The difference is the path to ground. In the first figure, the design uses npn transistors, and in the second it uses pnp transistor to pull to ground. With bipolar transistors you will probably see using either of these designs, more heat generated in the transistors than if you used a design with MOSFETs. In the top design, the bottom transistors will be driven into saturation with sufficient voltage on A or B (provided there is enough current), this really depends on the specs of the motor and how much current you need to push through the motor. [Saturation is driving the transistor into its most ON state, this drops the least voltage across the CE junctions and passes the most current while limiting the power dissipated (heat) from the transistor.] On the second design the path to ground will be limited to the PN junction voltage from motor to base, so if A is operating from the same supply, you will probably not achieve saturation in design 2 on the ground path from the motor.
Both designs will require A & B voltage probably be about 0.5 to 1 volt above +V to drive the top rail (+V) to the motor such that the upper transistors are in saturation. If these are very small motors and if you are driving A & B with CMOS 8051 outputs direct, then you should limit +V to about 4.0 - 4.25 at most. Otherwise, you will need voltage translation to try and drive the upper transistors into saturation.
If the motor you are trying to control draws much current under normal run conditions, I would recommend searching the web for MOSFET H bridge alternative.