I think predictions/hypotheses have to have a different status to observations, and I think the suggestion of a cline between full (prediction) and partial (observation) loses important aspects, although that is part of what is going on. Presumably it is the status that these representations within some discourse of doing science that makes a big difference. Certainly very meta.
So all together, perhaps indeed Peirce might help a bit. When we are making observations (and describing them) we are making use of instantiations of the theory (using those instantiations as a resource to impose sense). And that does not mean that observations are then 'just' instant[i]ations of theory as if that defined what observations are.
So at first glance I like your object : representamen : interpretant ~ data : description : observation proposal...
 To be clear — given the strategically omitted word here 'instantiation' — Bateman previously claimed (here) that both predictions and observations are partial instantiations of theory, while simultaneously asserting that the relationship between observation and theory is not one of instantiation.
 To be clear, since an instance of theory is an actual theory, rather than an observation or prediction, this is not "part of what is going on".
 It is worth pointing out here that this statement is consistent with the notion of instantiation, since instantiations of theory — that is, actual linguistic theories — are used in observations about, and descriptions of, language.
 As previously explained here, these proportionalities are invalid, since they misconstrue an observation as the meaning or ramification of a description.