For decades, higher education has been shaped by large-class lectures, which are characterized by large anonymous audiences. Well known issues of large-class lectures are a rather low degree of interactivity and a notable passivity of students, which are aggravated by the social environment created by large audiences. However, research indicates that an active involvement is indispensable for learning to be successful. Active partaking in lectures is thus often a goal of technology- supported lectures. An outstanding feature of social media is certainly their capabilities of facilitating interactions in large groups of participants. Social media thus seem to be a suitable basis for technology-enhanced learning in large-class lectures. However, existing general-purpose social media are often accompanied by several shortcomings that are assumed to hinder their proper use in lectures. This thesis therefore deals with the conception of a social medium, called Backstage, specially tailored for use in large-class lectures. Backstage provides both lecturer- as well as student-initiated communication by means of an Audience Response System and a backchannel. Audience Response Systems allow running quizzes in lectures, e.g., to assess knowledge, and can thus be seen as a technological support of question asking by the lecturer. These systems collect and aggregate the students' answers and report the results back to the audience in real-time. Audience Response Systems have shown to be a very effective means for sustaining lecture- relevant interactivity in lectures. Using a backchannel, students can initiate communication with peers or the lecturer. The backchannel is built upon microblogging, which has become a very popular communication medium in recent years. A key characteristic of microblogging is that messages are very concise, comprising only few words. The brief form of communication makes microblogging quite appealing for a backchannel in lectures. A preliminary evaluation of a first prototype conducted at an early stage of the project, however, indicated that a conventional digital backchannel is prone to information overload. Even a relatively small group can quickly render the backchannel discourse incomprehensible. This incomprehensibility is rooted in a lack of interactional coherence, a rather low communication efficiency, a high information entropy, and a lack of connection between the backchannel and the frontchannel, i.e., the lecture’s discourse. This thesis investigates remedies to these issues. To this aim, lecture slides are integrated in the backchannel to structure and to provide context for the backchannel discourse. The backchannel communication is revised to realize a collaborative annotation of slides by typed backchannel posts. To reduce information entropy backchannel posts have to be assigned to predefined categories. To establish a connection with the frontchannel, backchannel posts have to be stuck on appropriate locations on slides. The lecture slides also improve communication efficiency by routing, which means that the backchannel can filter such that it only shows the posts belonging to the currently displayed slide. Further improvements and modifications, e.g., of the Audience Response System, are described in this thesis. This thesis also reports on an evaluation of Backstage in four courses. The outcomes are promising. Students welcomed the use of Backstage. Backstage not only succeeded in increasing interactivity but also contributed to social awareness, which is a prerequisite of active participation. Furthermore, the backchannel communication was highly lecture-relevant. As another important result, an additional study conducted in collaboration with educational scientists was able to show that students in Backstage-supported lectures used their mobile devices to a greater extent for lecture-relevant activities compared to students in conventional lectures, in which mobile devices were mostly used for lecture-unrelated activities. To establish social control of the backchannel, this thesis investigates rating and ranking of backchannel posts. Furthermore, this thesis proposes a reputation system that aims at incentivizing desirable behavior in the backchannel. The reputation system is based on an eigenvector centrality similar to Google's PageRank. It is highly customizable and also allows considering quiz performance in the computation of reputation. All these approaches, rating, ranking as well as reputation systems have proven to be very effective mechanisms of social control in general-purpose social media.