Documenting sound change with smartphone apps

摘要: Crowdsourcing linguistic phenomena with smartphone applications is relatively new. Apps have been used to train acoustic models for automatic speech recognition (de Vries et al. 2014) and archive endangered languages (Iwaidja Inyaman Team 2012). Leemann Kolly (2013) developed a free app iOS—Dialakt App (DA) (>78k downloads)—to document language change in Swiss German. Here, we present results of sound based on DA data. predicts the users’ dialects: 16 variables, users select their dialectal variant. then tells which dialect they speak. Underlying this prediction are maps from Linguistic Atlas German-speaking Switzerland (SDS, 1962-2003), documents situation around 1950. If predicted wrongly, indicate actual dialect. With information, variables can be assessed change. Results revealed robustness phonetic variables; lexical morphological were more prone Phonetic like lift (variants: /lupfə, lʏpfə, lipfə/) SDS agreement scores nearly 85%, i.e., little Not all equally robust: ladle /xaelə, xaellə, xaeuə, xaeɫə, xaeɫɫə/) exhibited significant We will illustrate using that show details changes at hand.