If I'm interpreting your notation properly, you can read off the marginal distributions using the parameters of the bivariate distribution. If X and Y are jointly bivariate normal, then each of X and Y is normally distributed. You simply need to specify the mean and variance for each. If I'm reading this properly, X has mean 3 and variance 2, and Y has mean 5 and variance 4. I'm guessing the 0.5 is either the covariance or correlation, but this doesn't impact the marginals.