Two conflicting hypotheses exist regarding how oral reading behavior patterns follow from encounter with the text. One hypothesis views them as following decoding (decoding hypothesis), and the other sees them as occurring after comprehension processes have constructed a semantic representation of the message. These hypotheses are further differentiated by comparing "word callers" (decoders) to "dialect speakers" (comprehenders), two types of readers whose miscues seem to emphasize decoding or comprehending strategies. Three experiments were conducted to determine which theory is correct about oral reading. The experiments introduced specific alterations in texts to be read orally, and measured the oral reading errors that were made on the text surrounding the alterations. The alterations were designed to monitor lexical access, syntactic/semantic integration, and intersentence integration. The results support either hypothesis, depending on the context, the type of material, and the type of reader. All three factors interact to determine whether decoding or comprehension requires more attention in reading diagnosis. (Attached are a list of references, authors' notes, and audience response to the paper following its presentation.) (RL)