To me it comes down to what the users understand. As long as your conceptual model matches the users mental model. If they dont match then the conceptual models needs to be very intuitive.
One approach to making things easier to grasp is to borrow from the real world. Metaphors can be of great help when trying to introduce a concept or interact. Used in the correct way metaphors can be very powerful, but like any tool they can also be applied in the wrong way or in the wrong situation.
Personally I find a lot of user experiences that borrow from the real world are implemented quite poorly. Because they use a metaphor I expect them to work in a certain way and it frustrates me when they don't.
Creating a new language takes a lot of time and effort and I imagine a lot of iterations to get it right. But if done in the correct way can be moch more suitable for the specific task than "shoe horning" in existing real world metaphors.
So to sit firmly on the fence, I believe they both have their place. Implementing in the correct way and in the correct situation and they can both be successful.