In real-time programming, correct timing is as important as correct computational results. However, most real-time programming languages focus on concurrency rather than timing, leaving timing as an indirect result of task priorities and process synchronization. The resulting systems often have high complexity and tend to obfuscate the intentions of the programmer when it comes to timing. In this paper it is demonstrated how the integration of a timing construct into a programming language can improve on this when supported by synchronous communication and an earliest deadline first scheduler. It is argued that the scheduler should be lazy, and not execute statements without deadlines even if they are ready and the system is otherwise idle. This leads to a heightened awareness of the timing requirements associated with the different functionality in the system, and ultimately to the potential use of timing responsibilities as a new way of structuring a program.