I have a hunch that Krishnamurthi's book might be a good read in that regard (though not sure).
However I wonder what's left after you remove the issues I mentioned above. There are two cases: typed language and untyped language. In the first case you're left with Type Theory. In the second case you're left with, either a Turing Machine or a Lambda Calculus (e.g., implemented in a syntax free language like scheme). Is this the correct line of thinking? Or is there more to PLT than what I described?
To continue the thought further, in case of scheme, you have the base language on top of which you can build any PL feature or any PL paradigm you want. That means what's left is a survey/exploration of all possible PL features. Here I'm not interested in syntactic features. But in that case, what's left is really a survey/exploration of all possible PL paradigms, or programming paradigms [1]. So I guess what I'm really after is a massive study of the equivalent of all programming languages, but in a syntax-free way, i.e., in scheme syntax! (And I'm looking for a book that can provide that).