There were three classes of predicted knowledge sources: (a) induced sublexical relations (i.e., induction of orthographic–phonological relations from the experience of print words), (b) acrophones from letter names, and (c) transfer from spelling experience.