The effectiveness of two types of adjunct questions, standard embedded questions and "why" questions (elaborative interrogation), was investigated for readers differing in structure-building ability (Gernsbacher, 1990). Participants read a textbook chapter either with or without the adjunct questions. Learning was assessed with typical classroom testing methods (multiple choice, short answer). Also, relatedness ratings were used to assess the coherence of learners' representations. High structure builders generally outperformed low structure builders. However, embedded questions but not elaborative interrogation improved the low structure builders' test performances on information targeted by and related to the adjunct questions. Neither study method improved test performance for the high comprehenders. Embedded questions also stimulated more coherent representations. Results indicate that embedded questions are an effective study method for low comprehenders.