In math you often have a class of objects and functions on those objects (say, groups with group homomorphisms, or simpler, types in a programming language with functions that take one type to another). Then in math one often learns about an object by associating another object with it. For example, a topological space can be associated with a specific kind of group that tells you about the space's structure.
Then you can ask how this "association" behaves with the functions on the object. I'll use notation now, and say that for an object X the "associated" object is G(X). Often G carries over to functions, so that if f is a function X -> Y then G(f) is a function G(X) to G(Y). Such an "association" is called covariant because it preserves the ordering of domain -> codomain. The "association" is instead called contravariant if it reverses the order. E.g., f: X -> Y becomes G(f): G(Y) -> G(X). Of course, I'm using "association" but the real word for G is functor.