STA 290 Seminar: Shahin Tavakoli

STA 290 Seminar Series

Tuesday, November 15th, 4:10pm, MSB 1147 (Colloquium Room)

Refreshments at 3:30pm in MSB 4110 (Statistics Lounge)

Speaker:          Shahin Tavakoli (University of Cambridge, UK)

Title: "Spatial modeling of Object Data: Analysing dialectal sound changes across the UK"

Abstract: Dialectal variation is of large interest both in linguistics and in social sciences. It has long been studied using textual and semantic differences or phonetic transcriptions, the output usually being a map of isoglosses, i.e. a discrete partition of the region of interest based on dialectal homogeneity. However, dialectal variation is thought to vary continuously in space and this lines of separation have more than one degree of arbitrariness. We introduce here novel statistical techniques to analyse geolocalised speech recordings and to explore the spatial variation of pronunciations continuously over the region of interest.

We apply the proposed method to data coming from the spoken part of the British National Corpus, deposited at the British Library, London, UK and we produce maps of the dialectal variation over England, Wales and Scotland.

This is joint work with Davide Pigoli & John Aston (Cambridge) and John Coleman (Phonetics Lab, Oxford)