The English Dialects App: The creation of a crowdsourced dialect corpus. (2018)

• Record Type: Journal Article
• Authors: Leemann, Adrian; Kolly, Marie-José; Britain, David
• Is Part Of: Ampersand. Volume 5(2018)
• Page Start: 1
• Page End: 17
• Publication Date: 2018
• Subjects: API application programming interface -- BKA German Federal Criminal Police Office -- BNC British National Corpus -- EDA English Dialects App -- EDAC English Dialects App Corpus -- FRED Freiburg English Dialect corpus -- ICE International Corpus of English -- NORMs non-mobile, older, rural, male speakers -- ONS Office for National Statistics -- SED Survey of English Dialects; Linguistics -- Periodicals; 410.5
• DOI: 10.1016/j.amper.2017.11.001
• Languages: English
• ISSNs: 2215-0390

Abstract

Abstract: In this paper, we present the English Dialects App (EDA) and the English Dialects App Corpus (EDAC). EDA is a free iOS and Android app, launched in January 2016 that features a dialect quiz and dialect recordings. For the quiz, users indicate which variants of 26 words they use and the application guesses their local dialect; for the recordings, users can record a short text. The result is EDAC which includes metadata on mobility, ethnicity, age, educational level, and gender. More than 47, 000 users from across the UK have indicated dialect variants for these 26 words, and more than 3, 500 users have provided audio recordings. Unavoidably, EDAC does not successfully reflect distributions of age, ethnicity, qualification levels, and other parameters found for the UK population given that smartphone-based research reaches a specific stratum of the population. Yet there are also clear benefits to the sampling strategy used – benefits and pitfalls are discussed in this article. Future analyses will provide the most comprehensive understanding of English regional dialect variation since the work of the traditional dialectologists. We showcase two such analyses in this article. EDAC should, we demonstrate, be of interest to researchers in dialectology but also in forensic phonetics. Highlights: In this contribution, a new way of collecting English dialect data is described. The corpus collected contains dialect data from more than 47, 000 speakers. We showcase two analyses based on the collected data. The framework and corpus presented have extensive implications for sociolinguistics.