Scientists are concerned with definitions, hypotheses, papers, and teaching in a way more than they are with experiments as such. We must be able to state our general truths and even describe our measurements or else there is no science or data in any meaningful sense.
Consider what actually happens after we obtain an experimental result. Someone must say what was counted, in what units, and under what definition of the thing in question. “Significant,” “control,” “error,” “cause,” “species,” “energy”: each of these words and many others like them are presupposed in the sciences. Someone who treats words as secondary will still use them, since he has no choice; and he may well find that he sometimes uses words without noticing when they have stopped meaning what he assumed they meant or, even worse, when he has lost the precise sense of what they do or should mean in a given context.
If we treat words as secondary we will not find that science completely fails to produce data or theories. Rather, we will find that we cannot say precisely what the result was, cannot tell one kind of definition from another (for example, a descriptive definition from an essential definition), and cannot easily see when a mathematical expression or a graph is somehow doing the implicit work of an argument.
Without the arts of speech, which include logic (as we will see), any scientific method becomes something like a knack for doing some particular task and it cannot give us any sense of what the knack really consists of or why it works.