We propose a formal language for describing and explaining statistical causality. Concretely, we define Statistical Causality Language (StaCL) for specifying causal effects on random variables. StaCL incorporates modal operators for interventions to express causal properties between probability distributions in different possible worlds in a Kripke model. We formalize axioms for probability distributions, interventions, and causal predicates using StaCL formulas. These axioms are expressive enough to derive the rules of Pearl's do-calculus. Finally, we demonstrate by examples that StaCL can be used to prove and explain the correctness of statistical causal inference.