As long as our society has no received or generally-shared ontology (in which a theory of epistemology can be rooted), there is a real sense in which legal definitions MUST remain vacuous.
But arguably there are hints in the law of evidence of some still-generally-held views about the general nature of things in the cosmos and (consequently) about legitimate and illegitimate sources of evidence in trials in American courtrooms.
For example, the American law of evidence uniformly imposes a "personal knowledge" requirement for viva voce witness testimony. Does this demonstrate that American law rests on the premise or general belief that "direct personal perception" is the only legitimate source of testimonial evidence? I think careful analysis shows that the answer to this question is "no."
But it is nonetheless interesting and perhaps important that today's law of evidence still draws a basic and pervasive distinction between testimonial evidence and non-testimonial evidence. The law of evidence still takes the view that there is a fundamental distinction between non-human objects as a source of evidence and the thoughts, judgments, and beliefs of human beings as a source of evidence about events in the world. So while the modern law of evidence does not say that testimonial evidence is better than non-testimonial evidence or vice versa, the law of evidence still seems to see a basic difference between testimonial and non-testimonial evidence. Even this distinction, however, is beginning to fray here and there. For example,it is now generally agreed that non-human organisms such as birds "think" -- in the sense that the brains of organisms such as birds and dogs engage in complex calculations -- and it therefore follows that the "beliefs," or inferences, of organisms such as birds about some events in the world (e.g., rain or not-rain, night or not-night, direction of travel) sometimes definitely should serve as evidence.