Sampling distributions are just like any other distribution, except their "data points" are all possible sample parameters derived from samples of the same size!
Your sample needs to fit certain criteria for the corresponding sampling distribution to be considered normal. (We'll get into this with z-scores vs. t-scores.)