We address the problem of computing the expected reversal distance of a genome with $n$ genes obtained by applying t random reversals to the identity. A good approximation is the expected transposition distance of a product of t random transpositions in S_n. Computing the latter turns out to be equivalent to computing the coefficients of the length function (i.e. the class function returning the number of parts in an integer partition) when written as a linear combination of the irreducible characters of S_n. Using symmetric functions theory, we compute these coefficients, thus obtaining a formula for the expected transposition distance. We also briefly sketch how to compute the variance.