With the growing availability of technology, teachers are becoming significant co-designers of the curriculum--resequencing, editing or supplementing a primary textbook with publicly available digital learning objects. This places a grave responsibility on teachers, who must have a clear view of the rationale and didactic intentions of the textbook in order to "re-source" it while maintaining its coherence. Yet, teachers are seldom provided with means to understand textbooks' structure and underlying design principles, particularly at the grain size of individual learning objects. We are developing tools to express the "voice" of a textbook--one tool for assigning specific metadata to learning objects (i.e. "tagging", sometimes referred to as "coding") and another for analysing the coherence of the textbook based on the tagged metadata. In the current study, the 3 pre-calculus chapters of a high school textbook were tagged. Based on the tagged metadata, we used the second tool to reveal the relative prominence of different "types" of tasks, analysed by the categories of metadata. We focused on commonalities across taggers to explore the nature of the textbook. Triangulating our findings in an interview with the textbook author, we show that some patterns in the metadata reflect explicit design principles, such as avoiding symbolic representations in tasks that open a topic, while others reflect tacit principles--making sense to the author though not explicitly intended--such as eliciting non-technological student justifications in the topic of derivative. We conclude that this methodology offers novel opportunities for the study of textbook design.