Well, of course maths, programming, and natural languages are different, but all of them involve a symbolic language that models something and expresses that model. Kun seems to be focusing just on the differing degrees of precision these symbolic languages employ.
People tend to think of natural language as only a medium of communication, but it is also the way the human mind models whatever it perceives or imagines, and despite the fact we do that mostly instinctively, it is by far the hardest thing about language.
Programming is not math, it is language - a programming language is a language which defines, describes, and expresses an algorithm. Useful programs are frequently express something mathematical, but that is a function of the application, not programming itself.
Maths is just a whole lot of symbolic language. Learning maths is language learning, but it is also learning to describe things with precision and clarity and algorithmically. Some natural language learning is like that (e.g. advanced Latin grammar) and some is not (e.g. introductory conversational courses).