Tens, maybe hundreds of models and theories of reading exist that explain the same observed facts about reading, but in radically different ways. Also, even the most basic facts about reading vary considerably as function of reading context, reader properties, or language. This has lead to the proposal that reading behavior can ultimately not be decomposed into different components aspects (‘component-dominant-dynamics’ – CDD), but that it is rather a function of the relations between cognitive, language, and task-components (‘interaction-dominant-dynamics’ – IDD). The current state-of-the-art in reading research seems to be well captured by this proposal, but it is unclear how to productively advance a line of research within that framework. Making things even more complicated, the IDD framework itself seems imply the invalidity of some of the most basic research tools in the social sciences, such as the experimental method.

In this presentation, I will outline a general research strategy for reading research that a) is consistent with the IDD framework, i.e., that reading behavior is not (fully) decomposable; b) is centered around some minimal assumption about meaning, which I argue must be a central invariant characteristic for a theory of reading; and c) that can be operationalized by combining traditional research methods with methods from the field of nonlinear complex systems. At the center of the proposed strategy will be a version of Wittgenstein’s concept of language game and I will try to show how language games can be used to understand reading behaviors at a general level and that they also provide a conceptual bracket for the CDD and IDD framework in reading research.