I was embarrassed by my original sloppy comment and wanted to rephrase it. As background, I was deeply impressed by Kripke’s “Naming and Necessity” and also influenced by Locke, Berkeley and Hume, as well as Wittgenstein and Quine.
An adult shows a child a leaf and says “leaf.” This an act of naming. With the naming accomplished, if the adult points to the leaf, the child says “leaf.” If the adult says “leaf?” the child points to the leaf. Until the adult calls other things leaves, “leaf” is a rigid designator: for the child, that one leaf is the only thing in the world called “leaf.”
It’s a stretch to say a leaf being named “leaf” is an analytic truth. However, one definition of “analytic truth” is “true by definition” (all bachelors, etc). In this way, a leaf is a “leaf” because “leaf” is ‘defined’ by the adult naming act. The statement “this object is ‘leaf’ is true” because that was what it was named, just as unmarried men are named "bachelors."
When we are using “leaf” for a variety of related but different objects, science is a Quinean network built on the above “rigid designator” foundation. A scientist might say “this object the child calls ‘leaf’ is a member of a class of biological entities, and we can explain how leaves evolved, how they function, and why there are tests to show that that particular ‘leaf’ is a member of that set.”
As you say, science and phenomena are different things, though they both use the word “leaf”:
"So what we know about water is indeed correlated with its underlying molecular composition. Yet, in another sense, you can’t literally correlate water with H₂O with H₂O with water. Thus such correlations must be between what we phenomenally experience and know, and water’s underlying molecular composition."
Kripke talked about how we might see a cat and say “that’s a cat.” The scientist then says “this cat is a biological creature with the following properties.” Then Kripke slices the cat in two with a samurai sword, and we see it is a robot cat, stuffed with metal parts and wires. It is or was a phenomenal cat, but it is not a cat in the scientific sense.
This to me is how science and phenomena correlate. They are both “conceptual schemes” that have so many connections in our experience of the world that they seem to share some ontology, but in a final analysis, that sensed of shared ontology emerges from language, rather than ontology.
