A study of the constraints on word recognition errors (miscues) readers make when reading a passage printed conventionally and one printed in a regularized orthography such as the initial teaching alphabet (i.t.a.) was undertaken with 50 second-grade students divided equally between children taught according to the orthographies being studied. The passage was deliberately created to set up wrong predictions and to cause readers to consciously confirm these predictions before they continued reading. The students read the selection aloud, and miscues were recorded including pauses, insertions, omissions, substitutions, repetitions, self-corrections, and mispronunciations. The students were asked questions related to comprehension after they finished reading. It was found that the i.t.a. sample had significantly fewer miscues thought to be produced by content and syntax than the conventional alphabet sample. Comprehension was also significantly higher among i.t.a. students. (TJ)