We have been trying to find a solution to this for many years and have come to a conclusion that we won't be able to implement this in a generalized way.
First of all, we do not do dynamic translation of words and phrases. As such, we cannot support translated or localized random attribute names. If we were able to support user defined alternative names, we have to work through the search experience. If one were to search for abc*, we have received differences of opinion on how that should work. Should be return both the the base and alternative words? What if the user entered list is incomplete and we end up with a mix of base and alternative words? What if there are duplications between the base and alternative words used in different places? Lastly there are user specified operations, especially in Asset Analytics, that depends on exact words like path. If such a database was moved and alternative words are used, then it's likely many things would break.
Given there are so many pitfalls and uncertainties, rather than providing an incomplete and unstable solution, we have decided against implementing this idea.