The need for improving learning at deeper levels has been recognized in the national and international arena because language literacy, numeracy, and other core skills will not go the distance in handling 21st-century tasks. The challenge is aggravated by major gaps in school curricula and professional development of instructors. There is inadequate training of strategies for comprehending technical material, generating inferences, metacognition, integrating information from multiple documents, detecting misinformation, self-regulated learning, communicating through different media, collaborative problem solving, and other 21st-century skills. Some of these gaps can be filled with advances in discourse science and technology. Discourse scientists have identified the mental representations and processing mechanisms associated with different discourse levels. These levels stretch beyond word decomposition, vocabulary, and syntax (called basic literacy skills) into the deeper realms of semantics, situation (mental) models, rhetorical structure, and pragmatics. Computer tools automatically scale discourse on many levels so that individuals can get immediate multilevel feedback on text complexity, essay quality, and answers to questions. Discourse scientists have also identified effective strategies for training students how to improve comprehension of difficult material and mastery of a broad array of 21st-century competencies. These strategies have been implemented in intelligent, adaptive computer technologies, including those with conversational agents that take on different roles (e.g., peers, tutors, mentors). Future challenges lie in creating a curriculum for 21st-century skills that embrace discourse science and in scaling up discourse technologies that facilitate deeper learning.