In these discussions, I usually find out that more than one phenomenon is at work, so what I say here is probably not the whole answer. Having said that, no sideslip or yaw angle is required for dihedral to encourage recovery in the roll axis, through the mechanism that BernieG mentioned ( projected area of a wing half ). To convince yourself, do the following thought experiment: imagine that you have a plane with extreme dihedral, say 90 deg., and it finds itself rolled to the left, 90 deg., with no yaw angle, sideslip, or aileron deflection. What will happen? In this extreme case, it is obvious that the right wing is generating no lift ( lift is the force in the opposite direction from gravity ), and the roll will tend to correct until the two wing halves generate equal lift. Before you start chiming in, I realize that the side force generated by the right wing-half will probably generate some sideslip and yaw, but those effects are not necessary for recovery. For those of you with engineering backgrounds, draw a little free-body-diagram with all the forces on it. My impression is that this phenomenon is the primary mechanism by which dihedral gives roll recovery, even though other effects are certainly present. Having said that, I'm overdue to go do some reading on this topic and get a more thorough answer.

Cheers,

banktoturn