Accurate and early predictions of student performance could significantly affect interventions during teaching and assessment, which gradually could lead to improved learning outcomes. In our research, we seek to identify and formalize temporal parameters as predictors of performance ("temporal learning analytics" or TLA) and examine students' temporal behaviour during testing (i.e., in terms of time-spent). The goal is to specify a functional set of parameters that will be embedded in an adaptive assessment system in order to contribute towards the personalization of feedback services. In this paper, we present the motivation and rationale for our work, along with our methodology, initial results, contributions so far, and plans for future work.