Brajot. ISSP 2020. Sources and interactions of long-term phonatory instabilities François-Xavier Brajot Communication Sciences and Disorders Ohio University. Athens, OH 45701. Long-term phonatory instabilities are slow oscillations of vocal fundamental frequency (f o ) or intensity ranging from 0.1 to upwards of 25 Hz. They are posited to originate from various sources, including neurological, biomechanical, aerodynamic and acoustic. I review recent empirical and theoretical work that focuses on three such sources: tremor, wow, and a vocal “beat” due to cardiovascular pulsation. I argue that these phonatory instabilities provide a unique window into speech sensorimotor mechanisms and are of significant clinical value. I take as starting point the reflex resonance model of vocal vibrato proposed by Titze et al. (2002). The model consists of two negative feedback loops that alter descending cortical activation to cricothyroid and thyroarytenoid muscles based on changes in vocal fold tension (a laryngeal reflex). The principle control parameters are the gain and conduction times of sensory feedback. When gain is large enough, vocal f o oscillates around 4 to 7 Hz. The model accounts for a number of observations, including differences in modulation depth and frequency associated with practice in singers, changes with age, or pathology resulting in vocal tremor. In contrast to tremor, the 0.2-2 Hz wow has been associated with auditory feedback (Ternström & Friberg, 1989). Support for this is provided in Brajot and Lawrence (2018), in which f o wow extent and period could be reliably increased by delaying auditory feedback (Figure 1). Figure 1. Example f o time series for different delays of auditory feedback (DAF) in one subject. Incorporating an auditory feedback loop into the reflex resonance model described above, a colleague and I show that wow can be elicited from otherwise overdamped dynamics as auditory feedback parameters reach critical values described by an Andronov-Hopf bifurcation (Brajot & Neiman, forthcoming). As with tremor, modulation depth and frequency depend primarily on feedback gain and delay, respectively. We also successfully derive a parametric formula to estimate delays in the feedback loop from wow frequency. Evaluating the expanded model with both laryngeal reflex and auditory feedback loops in place, we find that the two systems interact. Increasing reflex gain, for example, can lower the threshold at which auditory feedback gain may induce wow. As reflex gain approaches the bifurcation line, f o oscillations become quasiperiodic (tremor superimposed on wow) for certain intervals of auditory delay. For reflex gains well above bifurcation, tremor predominates. Such interactions have only been identified by numerical analysis and have yet to be verified experimentally.