A set of linguistic features typical of spoken English must first of all be establised. In this respect, Biber (1988) and colleagues have a well known analytial framework. Using this MF/MD approach might be too demanding, I think the linguistic features positively weighted alongside Dimensions 1 and 3 are important indicators of the "oralness".
See http://www.corpus4u.com/forum_view.asp?forum_id=51&view_id=1225 for discussions in this connection.