Abstract

Anyone can use Google ngram-viewer to run a text search across thousands of books, and view the resulting chart in a few seconds. The situation is very different for speech -- mining spoken language requires significant technical skill. SPADE (SPeech Across Dialects of English) unites researchers from the UK, Canada and the US to develop and apply user-friendly software for large-scale speech analysis of multiple diverse speech datasets. We will analyse 43 existing public and private speech datasets of Old World (British Isles) and New World (North American) English, across an effective time span of over 100 years. We will gain an unprecedented understanding of stability and change in English speech across space and time. Our findings will be made accessible using dynamic visualization of speech variation in an interactive sound mapping website. Our tool and approach lays the groundwork for wide-ranging, cross-disciplinary research and speech mining of spoken language collections.