Using a text semantic framework, a two-year study examined the process by which readers construct meanings from standardized reading comprehension tests used as texts. The text analytic system described "ideal reader" interpretations of texts, allowing comparisons with real reader interpretations. Specific analyses described the ideal reader's hypotheses, expectations, and questions about the genre, content, and linguistic aspects of the text. A notation system was developed that provided a map of the changing envisionments that result from interpretations of earlier text segments and take new shapes as the text continues. Intensive interview procedures, including process probes, were used to compare real readers' and ideal readers' behaviors in constructing meaning for a text, as well as a reader's construction of meaning and performance on passage-related multiple choice questions. Readers of a variety of ages participated in the study, with the major analyses focusing on 26 third grade students responding to one text. The findings permitted identification of certain aspects of test language that can interfere with comprehension, including density of ideas, overreliance on assumptions and hypotheses, and assertions contrary to readers' beliefs. The findings also indicated that caution should be used when interpreting standardized test results. (RL)