Yes, using a properly oriented mild elliptical-shaped bellcrank would deliver a softer neutral, much like certain styles of devices in vouge today. Two large questions come readily to mind concerning such devices.
The circular bellcrank has been used in the past, most notably by Ted Fancher in the '70s. For several reasons, he, and others, stopped using them. They added a bit of complexity and concern should the cable jump from the groove. At best, they were very linear as long as the handle and bell crank spacing was the same. You mentioned the ratios that occur when the spacings differ. The exponential systems, as well as your suggested elliptical bellcrank will deliver some form of differential movement near neutral. The question in my mind is, why is that desired? What are we really trying to cure with soft neutral systems? The elliptical bell crank was considered and rejected by myself and others, back in the 80s. for most of the same reasons the circular bellcrank was dropped earlier.
Modern control systems start with the handle. For competition use, it should be adjustable, and lately, solid point attachment rather than flexible, adjustable cable, has become the favored style. The handle should be lightweight so you barely feel it in your hand. Heavy handles can make a poor showing as they will slow down your responses. The lines must be sized to prevent the stretchy soft feeling that can cause problems. Line length is used to set and maintain lap time, along with the throttle whether IC or E. Next comes the bell crank itself. As discussed it can be where some real help begins in our desire for help and better patterns.
First, let me address the thoughts of mounting the bell crank aft of the wing and routing the lead-outs through some form of pulleys to make the turn in direction. Way back in the early days of control lining, a similar system was produced to get around Jim Walker's patent for the bell crank. At the time, a royalty had to be paid to Walker when his (?) bellcrank was used in a kit, or even on a drawing. A side note: Many old plans do not show a bell crank. Where one would be there was text to the equivalent, "Use your favorite control here." Back to the aft of the wing discussion.
The design in question routed the lead-outs through pulleys, tubes created too much drag, making a 90-degree turn and then attaching to horns mounted top and bottom of the elevator thus eliminating the bell crank completely. It apparently didn't work well as it's the only one I've ever seen that used this workaround. Imagine what would happen if the lead-outs, either singly or both, jumped from the pully.
For modern PA, the use and understanding regarding longer 4-inch bell crank over the then standard 3-inch lengths was a turning point of some significance. The reasons for and use of "Corrected geometry" control horns were known as far back as the 50s. Knowledge of them was held in secret by most of those who were aware of them. I've come across at least one old set, of plans, drawn by one o0f the early greats, that clearly show such corrected horns from the 50s. Back in the day, it was common to keep some elements of design secret from general knowledge. ("Detroit Stunter" comes to mind.) Getting rid of the small dia. (1/8") flexible pushrods in favor of larger dia. dowels, and later, CF tubing along with the use of adjustable ball links helped make our control systems more "bulletproof".
Add to the list, adjustable lead-outs, tip weight boxes, openings in the tail to allow adjustable elevators neutral and throw, neutral stable, and unstable bellcranks. All developments to tune our control systems.
We've come a long way since it all began.
The questions at this time, at least for me, is. What are we attempting to achieve when making it slower around neutral? I have some ideas, but I would like the opinions of others before offering them.