Open Access

Johanna Vogel

• How do scientists speak when addressing a non-specialist audience? Science communication has become a widely discussed topic in the public sphere and a prominent area of academic research (e.g. Janich/Kalwa 2018). While several handbooks provide practical guidance for communicating with lay audiences (e.g., Falkenberg 2021; Wagner/McKee 2023), systematic analyses of spoken external science communication at the lexical-syntactic level remain scarce. In contrast, internal science communication, especially in written form, has been extensively explored, notably by Weinrich (1989) and Kretzenbacher (1992). For spoken academic discourse, the GeWiss corpus (Meißner/Slavcheva 2014) offers a valuable comparative framework. This presentation is part of a larger academic project investigating the linguistic structures of spoken external science communication, with a focus on social media formats such as Science Slams and TikTok videos. The poster addresses methodological and technical challenges involved in constructing a corpus of spoken social media content and shares preliminary findings on the linguistic construction of orality in Science Slams. The reference corpus is being built from audio material, mainly from YouTube and TikTok, pre-segmented and automatically transcribed using aTrain. Transcriptions are manually corrected and annotated according to the cGAT conventions. Further processing is conducted using the corpus annotation software EXMARaLDA, and morphosyntactic tagging is carried out with TreeTagger based on STTS 2.0 guidelines, enabling a systematic, lexical-syntactic analysis. The corpus will continue to grow over the coming years. The current dataset comprises approximately two hours of audio from the German Science Slam Championships (2021–2024). The initial analysis concentrates on markers of orality, with a focus on modal particles (e.g., also, halt), which, following Stein (2003: 439), can serve as indicators of spoken discourse. These findings are compared with recent analyses of internal scientific speech (cf. Schwendemann & Wallner 2023) to contribute toward defining spoken external science communication as a distinct linguistic register.