Process models play a key role in business process management. While control flow aspects are well understood, data aspects received less attention. In process models, data is commonly represented by data objects, and data object states are employed as abstractions from concrete properties of data objects. However, there is little agreement on how those states can actually be defined. This paper proposes a framework for the definition of data object states. The framework generalizes and consolidates the dimensions that make up data object states in related work. It supports concurrent states to capture different perspectives on a given object. The work is evaluated by (i) mapping the state definitions of existing approaches to the framework to demonstrate its unifying nature, and (ii) a prototypical implementation based on decision tables showcasing the applicability of the framework.