Distinguished Lectures in Humanities: Are We Nearly There Yet? Some Observations from a Corpus (Phonetic) Traveller Across English

Abstract

Language corpora allow linguists to travel across regional varieties of English, tracking language change over time. Previous research provides conceptual maps and research questions to act as routes, while our computational tools determine how much terrain we can actually cover. Until recently, large-scale speech corpus analysis has lagged behind text analysis, partly because of the complexities of processing audio data at scale.

In this talk, I discuss how advances in speech data processing, applied to the substantial SPeech Across Dialects (SPADE) meta-corpus, help us ask: how well do our maps represent the terrain? Three case studies draw on SPADE corpora across space and time — the English Voicing Effect, the Scottish Vowel Length Rule, and Scottish English /r/. I conclude with a detour into new territory, showing how Automatic Speech Recognition can itself become a novel tool for corpus phonetic analysis.

About the speaker

Jane Stuart-Smith is Professor of Phonetics and Sociolinguistics at the University of Glasgow, where she helped develop the Glasgow University Laboratory of Phonetics. Her research explores how speech and society interact, drawing on Scotland's rich linguistic variation, and recently uses corpus phonetics to trace variation across British Isles and North American Englishes. She also works with Scottish colleagues to promote public understanding of phonetics through accessible web resources such as Seeing Speech, Dynamic Dialects and STAR. A Fellow of the British Academy and the Royal Society of Edinburgh, she chairs the REF2029 sub-panel for Modern Languages and Linguistics.