Each Enigma machine setting induces a certain encryption in the sense of a function from the space of plain texts to the set of cipher texts.
The number of different Enigma machine settings can be counted for a given Enigma version (and usage). This is answered here: How many possible Enigma machine settings?.
But do different machine settings necessarily lead to different encryption functions? If not, what is the number of different encryption functions? Or, put differently, what ist the effective key length?
I think that this could be answered using group theory, but I still don't know how exactly this could be done.
Edit: I think I will represent the machine states as elements of a permutation group, provide these representations to a computer algebra system, let the system calculate the order of the group that is generated from all these elements and compare that group order to the total number of machine states. This is not a satisfactory answer because it probably won't give me much insight. It will just be a yes/no - answer.