|
Phonetic bias in sound change and its magnification through speaker interaction| old_uid | 12681 |
|---|
| title | Phonetic bias in sound change and its magnification through speaker interaction |
|---|
| start_date | 2016/12/09 |
|---|
| schedule | 14h-15h30 |
|---|
| online | no |
|---|
| summary | Historical sound change develops out of synchronic phonetic variation (Ohala, 1993). Sound change also infects a community through imitation (Trudgill, 2008). But how are these two aspects of sound change connected? We have addressed this issue by constructing an agent-based computational model (ABM) designed to test a hypothesis about the outcome when two different groups of speakers with different phonetic variants come into contact with each other. In contrast to other ABMs, agents exchanged dynamic signals based on data from real speakers. Word classes were defined as statistical generalisations over remembered signals (Pierrehumbert, 2003). The uptake or not of perceived items into an agent's memory depended on the probabilistic distance to the agent's phonological categories (Blevins & Wedel, 2009). When the model was run with agents representing two groups of speakers from Harrington et al (2008) with fronted and retracted /u/ in Standard Southern British, the directional change was asymmetric along the path of coarticulatory variation towards fronted /u/. The general conclusion is that phonetic variation is likely to be magnified resulting in sound change if some of the interacting speakers densely populate the space along the periphery of a phonological category's major axis of synchronic variation. |
|---|
| responsibles | Hueber |
|---|
| |
|