Trunk exoskeletons are an emerging technology that could reduce spinal loading, guide trunk motion, and augment lifting ability. However, while they have achieved promising results in brief laboratory studies, they have not yet been tested in longer-term real-world studies - partially due to reliance on stationary sensors such as cameras. To enable future real-world evaluations of trunk exoskeletons, this paper describes two preliminary studies on using inertial measurement units (IMUs) to collect kinematic data from an exoskeleton wearer. In the first study, a participant performed three activities (walking, sit-to-stand, box lifting) while trunk flexion angle was measured with both IMUs and reference cameras. The mean absolute difference in flexion angle between the two methods was 1.4° during walking, 3.6° during sit-to-stand and 5.2° during box lifting, showing that IMUs can measure trunk flexion with a reasonable accuracy. In the second study, six participants performed five activities (standing, sitting straight, slouching, 'good' lifting, 'bad' lifting), and a naïve Bayes classifier was used to automatically classify the activity from IMU data. The classification accuracy was 92.2%, indicating the feasibility of automated activity classification using IMUs. The IMUs will next be used to obtain longer-term recordings of different activities performed both with and without a trunk exoskeleton to determine how the exoskeleton affects a person's posture and behavior.