For a recent example of this unclarity in action, consider Sam Harris:
Definitions matter. And in science we are always in the business of framing conversations and making definitions. There is nothing about this process that condemns us to epistemological relativism or that nullifies truth claims. We define "physics" as, loosely speaking, our best effort to understand the behavior of matter and energy in the universe. The discipline is defined with respect to the goal of understanding how matter behaves.
Of course, anyone is free to define "physics" in some other way. A Creationist physicist could come into the room and say, "Well, that's not my definition of physics. My physics is designed to match the Book of Genesis." But we are free to respond to such a person by saying, "You know, you really don't belong at this conference. That's not 'physics' as we are interested in it. You're using the word differently. You're not playing our language game." Such a gesture of exclusion is both legitimate and necessary. The fact that the discourse of physics is not sufficient to silence such a person, the fact that he cannot be brought into our conversation about physics, does not undermine physics as a domain of objective truth.
I don't think Harris really means what he says here. He's described a situation in which mainstream scientists and creationists don't really disagree! If the creationist really defines "physical truth" to mean "claimed by the Bible", then his sentence "It's a physical truth that the Earth is only 6000 years old" really expresses the claim that the Bible claims that the Earth is only 6000 years old -- which may well be a true claim, and in any case doesn't contradict scientists' claims that that the Earth has been around for several billion years.
Of course, real-world creationists are not making such weak claims. They aren't just using language in a different way from us, they're making substantively different claims about how the world is. They presumably hold that religious texts should guide us in "our best effort to understand the behavior of matter and energy in the universe." This is not a definitional truth, but a substantive (and false!) epistemic claim about how we can most reliably uncover the truth about the world.
Likewise with the morally misguided. If religious fundamentalists used the word 'morality' to just mean 'commanded by the Bible', then we might well agree with their claims. For when they assert the sentence, "Morality requires us to do what is commanded by the Bible," this would just mean that it is commanded by the Bible to do what is commanded by the Bible -- which is as vacuously true as any tautology. In particular, it has no implications for what we ought, or have reason, to do.
But of course such people are not really doing anything so innocuously pointless as stipulatively redefining the word 'morality'. Rather, I take it, they mean to talk about the same irreducibly normative property that we do -- they just hold substantively different views from us about what things have this property.
The take-home lesson is that interesting philosophical questions do not turn on mere definitions. Granted, we will want to use clear terminology so that others will correctly understand what it is that we're trying to claim. And we will want to choose useful terminology that enables us to formulate interesting claims. (It would be unhelpful, for example, to redefine all moral terms so that we could no longer pick out normative properties for discussion.) But our claims, though communicated via language, are generally not about language. They're claims about matters of (non-linguistic) fact. And these facts are what they are, quite independently of the language we use to describe them.
Think of it this way: there are a bunch of things and properties in the world, and words that we use to pick these out. Which words pick out which things is a merely semantic matter, of no great interest that I can see, so long as each of us understands what the other means to be picking out by their words. What should really interest us is the substantive question of which things have which properties. If moral claims are to be substantive, there must be some moral property that we are attributing to acts when we call them 'right'. Let's call this property X. Normative ethics then consists in coming up with a theory about which acts have this normative property X, and why (i.e. which non-moral properties are the right-making features). Different people, having grasped this normative property, might have different views about which acts possess it. We would then have a substantive disagreement. Others might change the subject by redefining the word 'right' to pick out some other property besides X, but then we would merely be talking past each other.
Now, Harris may still be right that it's simply not worth engaging with religious fundamentalists in science or ethics. But if so, this is not because they're speaking a different language. It would be because they're using our shared language to make such substantively misguided claims that including them in our epistemic community is unlikely to help bring us closer to the truth.
* Is Normativity Just Semantics?
* Parfit's Triviality Objection