f vs θ

The fricatives f and θ have an asymmetrical relationship: θ is rarer than f typologically; θ is more phonetically variable than f; θ is more confusable with f than f is with θ; θ > f is common diachronically while f > θ is rare or unattested; and θ is acquired later by children than f. But why? The first half of the dissertation is about the f-θ asymmetry at the level of the individual speaker, using artificial language learning and perception experiments to test different explanations of the asymmetry. Chapter 1 lays out the basic empirical facts about f and θ. Chapter 2 discusses an artificial language learning experiment searching for both analytic bias, a mental bias skewing language learners towards particular rules based on their phonetic content, or channel bias, an asymmetry between rates of misperception for the two sounds. Chapter 3 digs deeper into a channel bias explanation with a perception experiment on speakers of English and Greek, discussing crosslinguistic variation in the perception of f and θ. The second half of the dissertation is more theoretical: we ask how individual-level perceptual asymmetries can scale up to the level of the world’s languages. Chapter 4 discusses how individual misperception events by child learners might lead to the existence of a whole new I-language, related to the input by regular sound change. Chapter 5 describes a model of multiple speakers of different ages interacting in a speech community, illustrating how new innovations by individual learners can spread to the rest of the population. Zooming out further to the level of the whole world, Chapter 6 presents a model of multiple languages undergoing generations of sound change, showing how phonetic biases in the actuation of sound change translate into typological patterns across the world’s languages.