The circuit shown in post #1 tries to operate a photo transistor with zero bias voltage which can't work. It looks like an erroneously modified version of a linearized analog coupler using photo diodes in the original design.
I see limited sense in fixing the faulty design, instead you should refer to known working analog coupler circuits, e.g. Vishay IL300 application circuits.
I agree to crutschow's point that a current steered LED gives already some amount of linearity, but the photo transistor partly dwarts it by it's current dependent gain.