While psychologists often use a combination of physiological and self-reported data to examine the dynamic effects of stress on performance, the impact of affective states on Foreign Language (FL) speaking performance has almost exclusively been assessed using self-report methodology (e.g., questionnaires, interviews). In fact, studies that correlate physiological data with self-report measures in a classroom context are extremely rare due to both cost and logistical restraints. This study set out to address this gap in language learning research by employing Fitbit smart watches as a tool to unobtrusively collect heart rate (HR) response data. Participants in this study were undergraduate Japanese language students (5 males and 5 females, mean age = 19.7 years, SD = 0.95) at a private university in Japan. Over three sessions, students wore Fitbit smart watches and performed three different class-observed dialogs (with randomized partners and performance order) while seated at their desks. Students were also asked to report their affective state (to index their feelings in the moment) across three intervals within each class session: class start, pre-performance, and post-performance. Using multi-level modeling statistical analysis, elevated self-reported state feelings of distress and embarrassment were found to be significantly positively related with elevated HR response. To further understanding of how affective states unfold in classroom environments, researchers should consider both physiological and self-report measures. With advances in wearable technology, similar research designs to this study may become more commonplace.