Yes. Alas, most of the automagic type inferences we are used to in FP languages (Hindley-Milner) does not play too nice with OOP's class members.

What if class members had to be typed?

They are (in Scala).

I think it's overloading (punning on static type of the same function, operator, or method name) that makes explicit type declarations more often necessary in Scala than in Haskell or ML.