ISCA Archive Interspeech 2015
ISCA Archive Interspeech 2015

Analysis of spatial variation with app-based crowdsourced audio data

Marie-José Kolly, Adrian Leemann, Florian Matter

Crowdsourcing technologies such as the use of web or smartphone applications enable the collection of large linguistic corpora. Such corpora allow for an analysis of various linguistic features. In the present contribution we analyzed data of >1500 speakers that were crowdsourced with the smartphone application `Dialäkt Äpp' to examine one such feature: the articulation of /kh/ as [k͡x] in Swiss German. We compared our results with findings from historical dialectological surveys. Results showed evidence for sound change in progress in Northern Switzerland with [k͡x] diffusing towards Zurich but retreating from the Northeast. We discuss the findings against the backdrop of methodological caveats of crowdsourcing audio data through smartphone applications.