The thing that is preserved is "meaning" or "referent" -- the term Quine uses (hence, "reference transparency") -- not "value". The distinction between referent/meaning and value is the most important aspect of distinguishing between pure FP programming languages and others, and yet that's the thing that is so commonly confused by FPers using the terminology, which makes it quite pointless.

In most programming languages the reference or meaning of a term is not a value; in pure functional languages the meaning is a value and that's what makes them special, not their referential transparency which they share with imperative languages.

Here's an example from C:

    int global_x = 0;

    void f() { x++; }
    void g() { x++; }

f and g have the same meaning in C (but the function `void h() { x += 2; }` does not) yet `m(f)` and `m(g)` will not have the same meaning if M is defined as:

    #define m(x) #x

However, f and g are interchangeable anywhere else (this is not actually true because their addresses can be obtained and compared; showing that a C-like language retains its referential transparency despite the existence of so-called l-values was the point of what I think is the first paper to introduce the notion referential transparency to the study of programming languages: https://github.com/papers-we-love/papers-we-love/blob/main/l... You may be surprised to see that Strachey also uses the word "value" but his point later is that value is not what you think it is)