You walk 3 steps, then you walk one more step.
This is the same as walking 4 steps from the start.
You grab 3 Oreos from the bag, then you grab one more Oreo.
You have the same number of Oreos as if you grabbed 4 Oreos from the get go.
3 + 1 = 4 ... in the strongest sense of "equals"
These are truths learned very very early in life, sometime between the onset of self mobility but well before actual talking.
And these truths are not learned independently ... the further step, the creation of the abstract notion 3+1=4 from the many different instances that we encounter that fact, this is also something learned very very early in life by people of all cultures, of all classes, of all intelligence levels (possibly there is a threshold on that last category).
It is not a matter of discourse.
The only question is whether it is a "necessary" truth or a "contingent" truth. How we come to know it is certainly a contigency ... many mathematical minds believe the abstract notion is necessarily so ... that there are really existing out there entities we call numbers that have necessary properties, their mathematical properties.
I won't go that far, but i certainly have no capability to explain the phenomenon that such a "realism" can support and that seems to be true of the world.
What is important here is the notion of a class of truths that are beyond discourse. They can be discussed, of course, and caught up in the web of language and the power structures there-of ... but they need not be.
And there-in lies the force of the notion that certain of the sciences are really different, because they tie into this type of truth ... physics being the best example. The fact that the formulas, the mathematical formulas come through paradigm changes intact, that is evidence of what we inply in the claim that the truths of physics "at its best" are outside discourse.
There are many other truths in physics which are not outside discourse. Pretty much, in fact, whenever we describe in mechanical or other terms what it is a certain mathemtical model "means", there we are back in the tabgled web of language. This is not such a problem for the oldest truths of science where the terms have been long walled-off with a technical meaning. Occasionally there is a change (eg. post-Einstein F=ma ... the "m" becomes a variable ... a classic example), ususally more slight than this example, but most of these terms stay in their formulas and do their work there-in.
Stepping forward to current science, there things become more problematic. Quantum theory, string theory, etc ... it is impossible to talk about these things with any accuracy in languge. We are best off just leaving the math to be the math.
I have used the "at its best" term. This usually means that the truths of the mature science (physics, chemistry, some biology) are best thought of as just the math formulas, not our description of the formulas. Take something like Universal Gravity. So long as we stick with the formulas we can really talk about knowing something about the phenomenon ... we can predict with stunning accuracy where a certain meteor will be at a certain time and hit it with Deep Impact. We know within a few 100 meters where Pluto will be in 10,000 years (barring other major force chages, like the sun exploding) relative to Earth. Stuning. But if you ask an ontological question about Universal Gravity, such as asking for a mechanical (or otherwise) explanation of what UG is (what UG really is) ... there the answers of science have not changed much since Newton's discovery of UG. [Essentially ... a shoulder shrug.]
The question at hand, though, is whether there is some knowledge outside discourse ... not whether all knowledge can be so. I think the answer is a definite "yes". The question then becomes, where is that line drawn, how much knowledge of this type is possible and what are the limits.
Not surprisingly, then, i work in epistemology (naturalized) and philosophy of science.
One last note ...
... science, at its best, is still very likely to say that certain kinds of questions are bad questions. It is very different from sociology or philosophy, though, in that it doesn't just tell you that you are asking the wrong kinds of questions and refusing to participate in that line of questioning ... it just states that its not very good with those kinds of questions (ontological) or that you are thinking of the phenomenon in a wrong manner (wave/partical duality) ... but it can take a crack at answering even those issues so long as you are happy with a partial answer. It can talk of gravity as a "rubber sheet" on which balls (planets) are rolling and get across, that way, some basic notions. It can talk about wave/partical duality in a way that is not arbitrary by noting the conditions under which, say, light, will always be measured as a wave and others under which it must be measuered as a partical. The duality is not ever arbitrary.
Again, such answers are not fully satisfying, but one does walk away with something.
(The problem of "Other Minds" has a related approach, even, too ... we can say something of what it is like to be a bat ... but we can't say "what it is like to be a bat" unqualified-ally. But if one is mostly a nihilist on what knowledge can actually do for us one will have some joy in the discoveries.)
[Maybe this is like the Augustinian-Calvinist notion of salvation ... we ALL deserve to burn. Because of God's grace some can be saved ... but none deserve it.]
I'll stop now.