Computational thinking (CT) involves a broadly applicable and complex set of processes that are often explained by way of the knowledge, attitudes, and general practices that they entail. However, to become facile with CT, learners require instruction that is grounded in concrete, relevant experiences. This paper examines teacher practices that are intended to promote CT skill acquisition through instruction that takes place in two framing contexts. The phenomenological context, which is based on observable patterns of object interaction that recur in games and simulations, is particularly valuable for developing the capacity to think abstractly. Abstraction is the key to recognizing analogous conditions, an ability that is the basis for transferring learning to new situations. The disciplinary context describes areas of application within and across subject areas, including computer science, that can foster proficiency with data representation, problem decomposition, and other CT skills. Using the Scalable Game Design curriculum as a lens to examine classroom practices, we find that teachers both plan and enact CT instruction in these contexts.