There is a well-known phenomenon in ancient manuscripts of MT, namely a diversity in word division. Scholars proposed different answers to this question. In the present article, I analyze four occurrences of that type in the Aleppo and the Leningrad (Saint Petersburg) Codex: Ps 9,1; 102,4; Job 24,6; 2 Kgs 6,25. I suggest that the variations are due not only to grammatical or phonetic reasons. They are also motivated by the midrashic tendency of scribes who try to solve in this way real or probable ambiguities in the consonantal text.