Given the setup on http://www.robotroom.com/BipolarHBridge2.html for an example. It looks to me like the green wires next to the pager motor provide more than one path back to the transistors. What would stop current from flowing back along on of the non-diode protected paths?
Yeah, too bad that Dave Crook cannot draw a regular schematic, or else it would be clearer to you, I'm sure.
Each side of the motor goes to a node common to both transistors in that side plus to 2 diodes which is reverse biased under normal conditions (they're there to catch inductive fly-back from the motor).
It's a very rudimentary H-bridge, as there's no control at all, just the 4 bare transistor switches with base resistors.
Anyway... Your subject line indicates something different from the content of your post, so I'm not sure whether this answers your doubts.