A word doesn't even have a discrete meaning, except locally in relation to other words.
Saying A = B + C looks discrete, just by hiding any potential non-discreteness inside B and C.
For example, a formal proof is a discrete process: it follows step-wise rules that you can assign natural numbers to (this is the first step, this is the second step, this is the third step). A non-discrete process, a continuous one, would have a smooth transition between these steps, which is hard to even imagine.
While I am not convinced it is correct to say that "human reasoning is discrete", human language is definitely discrete. Words don't blend smoothly into each other. If you don't believe me, try to define a function f:[0,1] -> Words, such that f(0) = "red" and f(1) = "blue" and tell me what is f(sqrt(2)/2), or what is df/dx.
There are aspects of cognition that are discrete: a language contains a finite set of phonemes and words, a human mind is capable of (painfully slowly) carrying out purely symbolic algorithms like those a computer performs, etc. My point was that these things are a small subset of cognition, and most of cognition we have no particular reason to think depends on discreteness, which I think is the same point you're making.
Personally I strongly suspect that the "discrete" aspects of cognition are things that have evolved on top of / within a system that is fundamentally continuous (analogue) in nature.
How do you convince yourself that you have thoughts that cannot be accurately written down no matter how many words you use?
> Personally I strongly suspect that the "discrete" aspects of cognition are things that have evolved on top of / within a system that is fundamentally continuous (analogue) in nature.
How do you tell whether things are really fundamentally continuous, or a really high definition pixel art?