The Gesture-Field-Register (GFR) framework for modeling F0 control

In this study, we introduce an F0 modeling framework – which we refer to as the Gesture-Field-Register (GFR) framework – in which F0 production involves joint control of relatively generic intentions and how those intentions are mapped to physical F0 values. Building on Articulatory Phonology (AP) and Task Dynamics (TD), the GFR framework considers F0 gestures to be the fundamental units of F0 control. It further holds (i) that the dynamic target F0 state of a speaker is determined by the blending of F0 gestural targets in a planning field and (ii) that the gestural targets and dynamic targets are represented in normalized values which are converted to F0 in Hz via dynamic control of F0 register. We show how this framework accounts for a variety of empirical F0 patterns, and we present a case study that uses parameter optimization to analyze empirical F0 contours into a time series of gestural activation and register states. In doing so, we demonstrate that the framework allows for gestural targets to be invariant within an utterance, despite the fact that the surface contours are highly variable. Model code and examples for generating and fitting F0 contours are publicly available in Github and OSF repositories. Overall, the GFR framework provides a novel way of conceptualizing and modeling F0 control under AP/TD and further expands the AP/TD by incorporating the mechanisms of a planning field and dynamic register control.