Learning philosophy historically, one finds a distinction frequently made between analytic and synthetic statements. We want to put declarative statements into two boxes:
Analytic:
“Cats are mammals”
“Every bachelor is unmarried”
Synthetic:
“There has existed a black dog”
“There are three people in this room right now”
How would we characterize analytic statements? One common way is to say that the truth of analytic statements depends purely on what the constituent terms mean, whereas synthetic statements (additionally) depend on the state of the world.
One of the two dogmas of Two Dogmas of Empiricism by Quine is the notion of analyticity, i.e. that the analytic-synthetic distinction can be made at all.
He argues that the type of information required to make us change our opinion of the truth of “cats are mammals” and any synthetic statement are actually not different in kind.
Quine’s conclusion: we should cease to make the meaning-theory distinction / language-theory distinction / meaning-belief distinction — all there is is the use of our expressions, and the usage is what determines both the meanings and what we take to be true.
Quine’s target is Carnap (who is thinking of artificial languages: first you fix the language and then you go into the world to see which are true in virtue of what they mean).