So a beta-alpha-beta structure will have two loops that could be turns - between the first beta strand and the alpha helix, and the helix and the second strand. They might be beta turns, but they might not.
Although there are different definitions of beta-turns, one common one is the presence of a hydrogen bond between the carbonyl of residue (i) and the amino of residue (i+3). I'm not actually sure how frequent this is, but you can see this from a search on PDBeMotif. There is also Motivated Proteins, which gives nice 2D (and 3D) images of motifs.
However, you could get other conformations of these loops that are not beta-turns, especially if they are quite long.