A sample of 150 Japanese English-as-a-Second-Language students at Southern Illinois University, Carbondale was given a reading comprehension test containing three levels of questions: factual, generalization, and inference, to measure comprehension effects at different proficiency strata. The results indicated that there were significant differences among the proficiency strata for the factual questions, but no significant differences among the generalization and inference levels of questions. An explanation for the compacted scoring distributions and resultant lack of significant differences among the proficiency strata is that the generalization and inference levels of questions required more short term memory ability, the attention and recall of more textual material, and more elaboration and rehearsal than the samples' target language competence could accommodate. In other words, the questions presented a processing overload. Sample test passages and reading probes are appended. The document contains 16 references and 7 figures. (Author/GLR)