Stanley and Quine

The attempt in Professor Jason Stanley's paper("Referential Semantics", Issues in the Philosophy of Language, The 22nd World Congress of Philosophy, Seoul, 2008) to construct a functional formalism structure out of the notion of degree in 3.2 children looks successful. I would believe that anybody can agree that the expression "3.2 children" in a sentence "An average American family has 3.2 children" may be taken to refer to this structural entity. But not only everyday people on the street but also philosophers in this seminar of the philosophy of language may have used expressions of a kind like "3.2 children" long before any functional structure was there suggested.

My question is this. Is semantics descriptive or prescriptive? If it is prescriptive what would its purpose or value be? Is it to be computational or to be ontological of a particular kind like physicalism? If it is descriptive, is it to be descriptive of speaker's language use or of speaker's state of mind when he utters the sentence?

Suppose that Professor Alvin Plantinger constructed a formal function for the word "god" in his system. Would there be a difference between ways to construct a formal functional entity, Quinean and Stanlean?