Listener-Driven Sound Change in Exemplar Theory

abstract: This paper presents a formal model to derive both gradient and categorical behaviors in listener-driven sound change. As a case study, I model the development of opposing acoustic trajectories of high vowels before labials in SE Bulgarian, where * i > u , and Polabian, where * u > i . By encoding different listener expectations into my model of labial co-articulation, I am able to derive both change, like SE Bulgarian and Polabian, and stability. In order to simulate the observed behaviors, I embed a Harmonic Grammar (Flemming 2001, Katz 2010, Burns 2021) in an Exemplar Theory (Pierrehumbert 2001) framework. My model suggests that while systems where interlocutors are aligned in their exceptions can result in sub-phonemic innovation, systems where they are misaligned result in both sub-phonemic innovation and eventual re-categorization in the preexisting phoneme inventory.