On the sources of lexical variability in early Slavonic translations from Greek: The method of “bilingual doublets”