Main Content

AnDy

The BMBF Project "Automatische Analyse der Dynamik dialektalen Sprechens mit Methoden der Künstlichen Intelligenz (AnDy)" is created as an interdisciplinary project between computer science (Lucie Flek) and regional language research (Alfred Lameli). The project deals with linguistic variation in the tension between standard language and dialects in the regional languages of Germany. Previous studies showed that speakers adjust their way of speaking to their interlocutors and thereby adapt or delimit themselves. This can also be observed in regional language variation. There are speakers who vary their speech strongly and switch between the poles of standard speech and dialect (type switcher or type shifter) and speakers who constantly use one speech position (type moveless). 

The aim of this project is to identify different speaker types in the dialect areas of Germany and to analyse the variation behavior in different conversational situations. These data will be used to understand current and historical processes of language change, as well as the strategic use of language in e.g., contexts of proximity and distance. Recordings from the REDE corpus are used as the data basis. The corpus contains speech recordings of male speakers of different generations from all over Germany in different conversational situations. With the help of these recordings, a data-classifying analysis tool is developed that links automatically obtained, dynamically alternating standard and dialect variants with extralinguistic factors, so that dynamic and spatially comparative analyses of speaker types become possible. It is expected that, in addition to insights into variational behavior and different speaker types, new insights into the structuring of speech contributions can be gained.

The project contributes to strengthening the data skills of researchers in the field and to methodological awareness in regional language research. The innovative, easy-to-use research environment will contribute to a new perspective on research.