In “Two Dogmas of Empiricism” Quine argues that there is no real basis for the distinction between analytic statements and synthetic statements. In other words he argues that the sharp division between statements that are true purely in virtue of meaning and statements that are true in virtue of both meaning and empirical fact is false. He sets out to give a definition of analyticity in order to show that the task is impossible and thus that the idea of analytic truth as sharply separate from synthetic truth is unfounded.
Quine states that philosophers usually place analytic statement into two categories. The first type of statement belongs to the category of logical truths. The example he give of this type is the sentence “No unmarried man is married”. This type of statement is not only always true but also remains true under all interpretations of ‘man’ and ‘married’. Only the meaning of the logical words and prefixes must remain constant. The key here is that these sentences are true in virtue of their syntactical form rather than their semantic meaning.
The second type of analytic statement can be easily transformed into a logical truth by replacing synonyms for synonyms. Quine gives the example of “No bachelor is married”. This sentence is always true, presumably because bachelor and unmarried man mean the same thing. In other words they are synonyms.
At this point Quine’s task is to attempt to define “synonymy” in order to explain the notion of analyticity. First, he looks into the idea of definition. In the end he comes to the conclusion that definition is grounded in the usage of language and so it really isn’t any guarantee of sameness of meaning. When we look inside a dictionary we find synonyms but this information is only a report of a lexicographer’s empirical observation concerning the usage of the language. Therefore, a dictionary can not be the basis for synonymy. He admits that explicating a definition in the context of scientific discourse is not a mere report on synonymy but says that it still relies somewhat on prior usage of the words. Definitions can not be the source of synonymy because definitions are rooted in empiricism.
If one wants to sharply distinguish analytic from synthetic then one does not want to have empiricism as the basis of analyticity.
Quine moves to a new strategy in order to find a firm ground for synonymy. He suggests that interchangeability of two linguistic forms without the change of truth value could possibly give us some basis for synonymy. Quine concludes that interchangeability salva veritate will not be sufficient. In order for it to work one would need to use an intensional language that contained adverbs such as “necessarily”. The sentence “All and only bachelors are unmarried men.” is not sufficient to ensure sameness of meaning between the terms “bachelor” and “unmarried man”. It is only enough to ensure sameness of reference. There is no guarantee that the sameness of reference between the two terms is do to meaning rather than to the way the world contingently turned out to be. In order ensure synonymy we would have to qualify the sentence by saying “Necessarily, all and only bachelors are unmarried men.” However, in this context the word “necessarily” presupposes analyticity. The conclusion is that these synonymies are grounded in empiricism. Quine says that “In formal and informal works alike, thus, we find that definition— except in the extreme case of the explicitly conventional introduction of new notation--hinges on prior relationships of synonymy.” He contends that even in formal languages such as those of math and logic, definitions rest on prior synonymies. Thus according to Quine’s view, Ayer’s claim that the a priori necessity of math and logic rests on analyticity, could not be true.