This study reports preliminary results on the characteristics of covariational reasoning in an augmented reality-rich environment. In contrast to the research literature that has hitherto focused on levels of covariational reasoning, this study addresses its linguistic characteristics. Two groups of three 15 and 17-year-old students participated in this study. Each group carried out two activities (Hooke's law & Galileo) using AR headsets. The students' interactions were video-recorded. Based on inductive analysis of the students' covariational reasoning language, we identified five linguistic categories of covariational reasoning: direct-covariation, mediated-covariation, conditional covariation, multi-variable covariation, and second-order covariation.