Modelling the actuation of sound change: a phonetic investigation into /s/ to /sh/ retraction
Funding, application period
This project is concerned with the actuation of sound change, that is, how and why ubiquitous phonetic variation occasionally accumulates in a population (or part of a population) to cause the shared language to change permanently. Recent advances in computational modelling have made it possible to simulate interactions between real speakers and to explore in a controlled environment the conditions under which fine-grained phonetic variation is most likely to be converted into a permanent sound change. The proposed project takes as a test case the understudied sound change whereby /s/ changes to /sh/ before consonants (e.g. Old High German 'slafen' > German 'schlafen' (‘to sleep’)). If attested cases of this /s/-to-/sh/ sound change have their origins in a universal phonetic tendency (as has been shown for many other attested sound changes) then such a tendency should be measurable in languages in which the sound change has not yet taken place. This hypothesis will be tested based on a combination of acoustic, physiological and perception data for sibilants in two languages (Australian English and standard Italian). The second major hypothesis to be tested is that fine-grained pronunciation differences between individuals are sufficient to transform a shared phonetic bias into a permanent sound change, when the individuals interact with each other. This hypothesis will be tested with computational models for English-speaking and Italian-speaking populations in which real speakers will be represented by agents who interact with each other by exchanging (parameterised) dynamic speech signals. The results of the proposed project will shed new light on the ways that micro-variation can accumulate such that one language may eventually split in two.