Is Ude Garami a Shoulder Lock or an Elbow Lock?

I am currently working on the joint-locking chapter in my book on the science behind fighting techniques. I've come across this rather interesting situation with regards to ude garami (arm entanglement).

The image to the right is the ude garami as taught by the late Shihan Jan de Jong OAM 9th Dan and his descendants. It is considered by them to be a shoulder lock.

Before we proceed, let's have a precise definition of joint-locks/kansetsu waza. Joint-locks/kansetsu waza are techniques where forces are applied to an opponent's body to move a joint towards or beyond the limits of its range of motion.

Ude garami, as the above parties would suggest, is a joint-lock where forces are applied to move the shoulder joint towards or beyond the limits of its range of motion.

Ude garami is most often shown in judo as being applied when both parties are grappling on the ground (see right). Many (e.g., see Jigoro Kano, Kodokan Judo) explain that ude garami targets the elbow. This is why ude garami is allowed in judo competition because only kansetsu waza that target the elbow joint are permitted in judo.

Does ude garami target the shoulder or elbow joint? Does the 'ground' technique differ in the consequences of the applied forces, i.e., targets the elbow rather than shoulder joint?
What are the anatomical consequences of the applied forces in both of those techniques?

Even though some distinguish between ude garami and gyaku ude garami in judo, more often then not the two techniques are referred to as ude garami. Should they be considered different techniques and not the latter a variation of the former if the anatomical consequences are different in both?

There are many more anomalies once you actually begin to study joint-locking/kansetsu waza techniques. Anomalies that throw up confusion and debate in martial arts circles but cannot be answered due to the lack of information and informed discussion.