An evidentially constrained notion of truth has two clauses characterising it and an evidence transcendent notion of truth has only one clause. The first clause which characterises evidentially constrained truth (also called an epistemic notion of truth) specifies that truth is, in a manner of speaking, a status which some of our propositions may obtain when certain conditions have been fulfilled. So this sense of truth characterises truth as a status of a proposition when it has satisfied certain conditions. What exactly are these conditions of which the first clause speaks? Quite simply and relatively uncontroversially, they are a proposition's truth conditions. Truth conditions refer to the content of an assertion which, if the assertion is able to denote accurately, would make the assertion true. Truth conditions are, therefore, thought of as states of affairs in the world which language attempts to represent, and either does or does not successfully do.
An evidentially constrained notion of truth has, in addition to the first clause of this notion of truth, the further specification, or clause, that in order to make a claim to truth, in other words, a claim that the proposition has met such truth conditions, it must be possible for the speaker (user) of the
proposition to manifest how they know that the proposition has satisfied such conditions. We see here a construal of truth which has two interdependent clauses working together to stipulate what truth is. It is only the second clause, regarding the manifestation of knowledge of the proposition having met the requirements forwarded by the first clause, which is essential to an evidentially constrained notion of truth because the first clause it shares with an evidence transcendent notion of truth. (We shall discuss this shortly, though.)
From the above it should be evident that truth, as an epistemic notion, has everything to do with the capacity of the speaker. It cannot be the case, by an epistemic notion of truth, that achieving truth has anything to do with the linguistic community because when a speaker says something they are only warranted in saying this if they have access to the knowledge of what would make such a statement true. When a lay person says "Light can bend under certain conditions" but has no knowledge of the facts that would make this true they are, by an epistemic notion of truth, not warranted in making this claim even if it happens to be absolutely true. It is only the experimenter or scientist (or even informed lay person) who knows the facts that would make such a statement true, who is warranted in making such a claim. An epistemic notion of truth asks for a cognitive component in the definition of truth itself. Truth, in this sense, is not just a status of a linguistic item but is, rather, the status of the speaker's knowledge of the linguistic item and its relationship with certain states of affairs. It is when truth is construed as an epistemic concept of sorts that we think of it as a result of having established when the speaker is warranted in making certain
propositions and when not. In other words, truth is the status of a linguistic construct only when the speaker is able to explain the relationship between the construct and world it refers to and is
therefore warranted in doing so. Truth, therefore, becomes something like warranted assertibility.
Truth being something like warranted assertibility entails that, according to such an epistemic notion of truth, many proposition will remain indeterminate in truth value. This is because an epistemic notion of truth holds that wielding the truth predicate (that aspect of a sentence which either implies or overtly states that the sentence is making a claim to truth) there exists no
appropriate speaker knowledge is illegitimate. And we are not warranted unless we can say how we know the truth predicate bearing sentence to have tracked the intended matters of fact. Note that it is not being claimed that some propositions are indeterminate in truth value for the reason that there are no states of affairs which they can denote- this would be a metaphysical presupposition. The claim, simply, is that if we do not have the right sort of access to the facts and cannot say in what way we do when we think we have access to the facts, there is simply no manner in which we can determine the truth value of certain propositions. Hence, some propositions being indeterminate in truth. And, it seems, this conclusion hinges entirely on a particular construal of truth- that it is evidentially constrained and, therefore, epistemically defined. Revisiting, briefly, our example from above: The reason why a claim such as "There is a maximonis in that box" is indeterminate in truth value is not because we assume there is no such a thing as a maximonis, but rather because we have no way of telling (knowing/determining) when our propositions will be true.
According to Weiss (2002; 107), Dummett lists four criteria for when we can take a sentence to be correctly asserted (true), in other words, when there is warrant for its assertion. The first is that it should be asserted on the basis of inductive evidence. Secondly, it must be capable of being used as a basis for inference, thirdly, that it can be correctly denied in certain circumstances and fourthly, that one should withhold judgement about its truth or falsity in certain circumstances. In short, Dummett will have it that moral claims cannot merely (only) form part of a deductively valid inference, but should be based on inductive (empirical) evidence. We should also be able to falsify
it. Referring back to our discussion about the importance of falsification we can see how much more conclusive evidence has to be in order to provide reasons for falsification. Thinking, then, of what a moral fact would be in order to determine, by Dummett's standards, whether moral
language is warranted: Are moral claims ever asserted on the basis of inductive evidence? And, in the light of much uncertainty about what exactly a moral fact looks like, are ever going to be able to deny moral claims?