Is there a physical reason for why the coupling constant
$\beta < 0$ ? I tried to read the section in your paper but didn't see anything relevant.
Mathematically, it should be clear that as
Physically,
The bend in the
$\ce{CO2}$ determines the change in the coupling constant between the local modes,$\beta$ . The motion of the central carbon atom is the primary motion that couples the two carbonyls. When they are collinear, the motion of one carbonyl directly influences the other. Bending the$\ce{CO2}$ means the carbonyls are no longer collinear, which means that the projection of one local vibration on the other decreases. This decrease, in turn, decreases the effective coupling constant.
The fit I found was