A split-DP hypothesis for Latin and Italo-Romance