A study proposed a sequenced process for inferring word meaning from context in order to investigate experimentally where, within the process, differences between learners of varying skill occur, and to discover the kinds of information students need in order to acquire vocabulary most effectively. Six five-step tasks, each designed around an artificial word, were administered individually to 30 urban fifth graders, 15 with high vocabulary ability and 15 with low. These tasks were at the level of contexts typically encountered in school reading, thereby yielding a realistic picture of where difficulties in processing ordinarily occur. Scores obtained from each step of the task each represented an aspect of the meaning acquisition process: selection of constraints from context; testing a meaning choice within constraints; use of two contexts to constrain meaning choices; evaluation of a meaning choice given the use of two contexts; use of additional contexts to refine word meaning; identification of word meaning given contexts that contain direct meaning clues; and discrimination between sentences that use and those that misuse the newly learned words. Findings indicated that the low ability group misunderstood the relationship between word and context; that both groups demonstrated a semantic interference when considering two contexts simultaneously; and that the process is a complex one. Teacher modeling of important concepts and strategies may be well-suited to improve children's context skills. (Twenty-eight references are attached.) (SR)