It is very unlikely that the constant compression on the joint would result in any turning of the barrel adjuster. Two factors affect this:
firstly the friction between the threads and the friction between the adjuster against the cable sleeve has to be overcome. The way friction works is that until a minimum threshold force is reached there is no movement. Multiple tiny jolts do not add up over time, they have no effect at all.
Secondly there is a negative leverage effect when turning the threads. When tightening a bolt the bolt applies a much greater force on the job than the spanner applies on the bolt. In this case you are working in reverse, so the compression force on the threads results in a very much weaker turning force on the barrel.*
It is extremely unlikely that the return spring on a brake caliper or gear shifter could provide anything like enough force to overcome the friction and turn the adjuster barrel.
Any turning effect that did happen would probably come from mechanical vibration or something banging or rubbing against the adjuster.
* one way to look at this is that the flats on a bolt with a 5mm head travel 3.14 x 5mm in each turn. The bolt moves forward maybe 0.5mm. So the force gets amplified approximately 30 times. In reverse the pressure on the bolt is weakened by a factor of 30 when it tries to push the flats around.
For anyone interested in how lubrication makes a difference:
There are two types of friction, static friction and dynamic friction. Static friction states what the resistance to movement will be for two stationary surfaces in contact. For any two surfaces the sideways force needed to make them move is proportional to the force pressing them together. Typically for dry metal-metal contact this coefficient of friction is about 1, i.e. a 1 Newton force pressing the surfaces together means a 1 Newton sideways force is required to make them move.
Once the surfaces are moving then the coefficient of friction generally falls to a lower value. For dry metal-metal surfaces the static and dynamic coefficients are about the same, getting the surface moving does not change things much.
Lubricating surfaces typically lowers the coefficients of friction. However, the lowering is not equal: for typical metal-metal contact lubricating lowers the static coefficient a bit, it lowers the dynamic coefficient a lot.
So, lubrication does not make the barrel adjuster much easier to turn, but once you have got it turning lubrication does make it easier to keep on turning.
Finally: we need to consider the difference between force and impulse. If I tap my car with a pencil it applies a high force to the car (listen to the sharp explosive sound it makes). The car does not move because the duration of the force is transitory, the impulse imparted by the pencil is minimal. If a tow-truck rolling very slowly gently nudges the fender the car moves along the road. The force is low (no sound, no damage) but the duration is long and the impulse is considerable.
Imaging a pencil tapping the barrel adjuster. It may well have enough force to make the adjuster move, but the duration of the force is minimal. Theoretically it will get the adjuster to move, but in a random direction and for an infinitesimal distance.
On the other hand the return spring, a twisted cable etc might be able to apply a smaller force for a long time in a consistent direction. It can't get the adjuster to start moving, but if it could then theoretically it might be able to keep it moving one-way for a non-negligible distance.
However, lubricating the adjuster still won't allow the spring/twisted cable to get the adjuster moving because lubrication does not change the static friction much.
Add lubrication and the pencil tap together and they might make a difference. The pencil tap adds a negligible movement in a random direction. Even so it is a movement, and the lubrication has lowered the dynamic friction so that might allow some other force (low but in a consistent direction) to keep the adjuster moving for a while in one direction.