(This question involves semantics, epistemology, and empirical studies of speaker judgements - wasn't sure if I should post in philosophy or linguistics, hope this is an OK place for it.)
I've been researching Allan Hazlitt's claim that knowledge attribution isn't factive, i.e. the claim that (in ordinary language) sentences of the form "A knows p" can be true even if the proposition p is false (see "The Myth of Factive Verbs" and "Factive presupposition and the truth condition on knowledge").
People (ordinary English speakers, not philosophers) do speak that way sometimes, or show patterns of judgement that are aligned with a non-factive interpretation for the word "know". The most prominent counter-argument that I'm aware of says that when people speak that way or make those judgements, it's based on a non-literal interpretation of "know". In particular, the claim is that it involves "protagonist projection" (see e.g. Richard Holton's "Some Telling Examples: A Reply to Tsohatzidis"), where someone adopts the perspective of the protagonist. If this is the case, then when someone says "A knows p" in a situation where p has been established as false, then what they literally mean is that "A thinks they know p".
There have been a couple studies that have been trying to tease apart whether ordinary English language speakers judgements support a semantically non-factive reading of the word "know", or if they only make those judgements on a non-literal interpretation. These studies (e.g. "Factive verbs and protagonist projection" by Buckwalter) make an unexamined assumption: That inserting the word "really" forces a literal reading of that sentence. That is, if speakers accept the sentence "A knows p" but reject the sentence "A really knows p" it means that the sentence "A knows p" is only being accepted on a non-literal interpretation.
Is that right? Is there any published research on the semantics of the word "really"? (I haven't been able to find any.) There seem to be sentences where "really" is used as an intensifier for the degree of the predicate that a verb encodes. E.g. we'll interpret "the army really destroyed the city" as implying a greater degree of destruction than "the army destroyed the city." Similarly, we can imagine that "really knows" simply has a higher level of justificatory standard that has to be met.
