Quine is right only if Quine is wrong.

“…necessity resides in the way in which we say things, and not in the things we talk about.”

–  W.V. Quine, “Three grades of modal involvement”

(note: in what follows I use single-quotation marks to name words and ignore their meaning and double-quotation marks to name a word with emphasis on its meaning.)

Suppose I ask you if the statement

(1)           Brutus killed Caesar.

could have been false, and you say: “Yes, for the word ‘killed’ could have meant “married”, and since Brutus did not marry Caesar, had (1) meant “Brutus married Caesar”, (1) would have been false.”

            Utterly dumbfounded by your response, I present you with the following statement

(2)             7 is prime. 

and ask if (2) could have been false, and you respond in kind: “Yes, for the word ‘7’ could have meant “6”, and since 6 is not prime, had (2) meant “6 is prime”, (2) would have been false.”

            At this point exactly half of my dumbfoundery has vanished because I see that when I asked you whether (1) and (2) could have been false, you responded by mentioning the words I used when I was using those words to talk about things.

            Suppose I make the semantic ascent to your level and ask you if you think it’s true that:

(3)             The word ‘killed’ could have meant “married”. 

and

(4)             The word ‘7’ could have meant “6”.

If you answer “no”, then you’re being inconsistent.  If you answer “yes”, then you disagree with Quine, for (3) and (4) are true only if necessity resides in at least some of things we talk about, for (3) and (4) are equivalent to

(3’)       The word ‘killed’ possibly means “married”.

and

   (4’)        The word ‘7’ possibly means “6”.

and (3’) and (4’) are equivalent to

(3’’)      The word ‘killed’ doesn’t necessarily not mean “married”.

(4’’)      The word ‘7’ doesn’t necessarily not mean “6”.

Ergo, etc.