Many factors affect the intelligibility of synthetic speech. One aspect that has been severely neglected in past work is hearing loss. In this study, we investigate whether pure-tone audiometry thresholds across a wide range of frequencies (0.25--20kHz) are correlated with participants' performance on a simple task that involves accurately recalling and processing reminders. Participants' scores correlate not only with thresholds in the frequency ranges commonly associated with speech, but also with extended high-frequency thresholds.