Any low-current low-voltage signal running parallel to the phase wires in the same cable will be just as vulnerable as any other.
A current-loop signal (instead of a voltage signal) might be less vulnerable; you'd have to test that. (MIDI signals in music equipment are sent over a current loop, if you want to look up how they do it to get an idea of how to create such a signal).
A higher voltage signal would be less vulnerable, as it would have a higher signal-to-noise ratio. Typical bipolar hall sensors used in motors can often handle up to 25-30V on their outputs, so if you were to make an interface board on the controller end that provides as high a pullup voltage as possible (depends on the specific halls in the motor) on the motor hall signal wires, and the controller side of the interface uses typical open-collector outputs so the controller's standard pullups operate with them as if they were hall sensors' OC outputs, then you get the best signal-to-noise ratio possible without actually redesigning the controller for some other type of input.
The more current the OC outputs in the halls can sink, the more solid the "zero" is, and the higher the pullup voltage at the controller end, the more solid the "one" is.
I don't know what kind of encoder you' want to know about performance-wise; you'd have to link to something that shows how it works. (encoder is just a term for encoding data; there are numerous things using that term; even the hall sensors in the typical motor are encoders, as are SIN/COS and other common motor sensor types).
But if their outputs are open-collector with a pullup inside the controller like UVW halls are, then they work the same way and have the same advantages and disadvantages.
If their outputs are voltage-producing types, so they actually create a signal (analog or digital) on their own, with no controller-side pullups, then the signal is probably more vulnerable to induced currents, or other interference, dependign on how strong the signal is (how much current the outputs can create).
Any position sensor / encoder that does not use the rotor magnets, but has it's own and so the sensor is not near the motor's magnetic fields / stator (unlike typical UVW halls), will be less vulnerable to magnetic interference from motor operation itself (but this doesn't include interference in the wiring if that wiring runs parallel to high-current wiring).
