The following scholars have kindly agreed to deliver plenaries:
Astrid Ensslin
University of Regensburg
Photo by: Julia Dragan
Abstract
Research on language ideologies in video games has shown that speech accents are strategically employed to construct fictional worlds and to position characters morally and socially as well as in line with dominant language ideologies and cultural stereotypes (Ensslin 2010; Goorimoorthee et al. 2019; Stein 2023). Such performed, voice-acted accents reflect real-world biases held by developers and target audiences and often serve to reproduce and reinforce existing social hierarchies (Villanueva & Ensslin 2021). Yet they have also been leveraged by indie game designers in particular to shape socially diverse fictional societies (Kjeldgaard‐Christiansen & Hejná 2022). Meanwhile, in the paratextual sphere, players and influencers use social media platforms to construct and negotiate meritocratic status through metalinguistic signalling and folk linguistic, vernacular theory crafting (Tarnarutckaia & Ensslin 2019; Spowage 2025), placing themselves on a spectrum between lay and expert discourse.
In this talk, I develop a theory of videogame speech accents both as a pragma-performative sociophonetic repository and an object of narrative game design and meritocratic gamer discourse. Based on the distinction between orthogame and paragame discourse (Carter et al. 2012), I zoom in on language ideologies underlying character design, voice acting and accent distribution, players’ metalinguistic discourse and the more recent phenomenon of the “folk” ludolinguist, who acts as “lingfluencer” on social media and particularly streaming platforms in a way that may come to shape public and particularly GenZ understanding of language categories, usage, and linguistic theory more generally.
References
Carter, Marcus, Martin Gibbs & Mitchell Harrop. 2012. Metagames, paragames and orthogames: A new vocabulary. In Proceedings of the International Conference on the Foundations of Digital Games (FDG '12), 11–17. New York: Association for Computing Machinery.
Ensslin, Astrid. 2010. Black and white: Language ideologies in computer game discourse. In Sally Johnson & Tommaso Milani (eds.), Language Ideologies and Media Discourse: Texts, Practices, Policies, 205–222. London: Continuum,
Goorimoorthee, Tejasvi, Adrianna Csipo, Shelby Carleton & Astrid Ensslin. 2019. Language ideologies in videogame discourse: Forms of sociophonetic othering in accented character speech. In Astrid Ensslin & Isabel Balteiro (eds.), Approaches to Videogame Discourse: Lexis, Interaction, Textuality, 269–287. London: Bloomsbury.
Kjeldgaard‐Christiansen, Jens & Míša Hejná. 2022. The voices of game worlds: A sociolinguistic analysis of Disco Elysium. Games and Culture 18(5), 578–597.
Spowage, Kate. 2025. Against the game: Sid Meier’s Civilization and vernacular theories of language. Language in Society 55(3), 502–522.
Stein, Simon David. 2023. Space rednecks, robot butlers, and feline foreigners: Language attitudes toward varieties of English in videogames. Games and Culture 18(8) 1043–1070.
Tarnarutckaia, Elizaveta & Astrid Ensslin. 2019. The myth of the “clarté française”: Language ideologies and metalinguistic discourse of videogame speech accents on Reddit. Discourse, Context & Media 33, 100352.
Villanueva, Emily & Astrid Ensslin. 2021. Divine intervention: Multimodal pragmatics and unconventional opposition in performed character speech in Dragon Age: Inquisition. In Gaëlle Planchenault & Livia Poljak (eds.), Pragmatics of Accents, 205–228. Amsterdam: Benjamins.
Videogame Speech Accents as Performed Identity, Ideology, and Folk (Ludo)linguistics Videogame Speech Accents as Performed Identity, Ideology, and Folk (Ludo)linguistics
Maria Pavesi
University of Pavia
Roman Schneider
Leibniz Institute for the German Language & University of Mannheim
Abstract
Song lyrics are among the most widely circulated texts in most speech communities, yet they entered corpus linguistics relatively late and uneasily. This talk introduces the Songkorpus (Schneider 2026a, 2026b), a large annotated collection of German lyrics spanning seven decades, with particular attention to its curation and the methodological decisions involved. I argue that performed language is not deficient data but a stress test for our analytical instruments.
Two claims organise the talk. First: building the corpus is already an act of measurement. Lyrics look like spoken language by standard register measures (Werner 2021; Schneider & Broll 2026) – direct address, contractions, low lexical density – although none of the situational conditions for spontaneous speech apply: the involvement is staged rather than interactional (Biber 1988; Koch & Oesterreicher 1985). Such material forces decisions – what counts as an error, which tagger to trust, how much a language model may „tidy up“ – that rarely appear in published results, yet determine what subsequent analyses can detect. Three cases illustrate the problem: standard taggers disagree substantially on how much of the vocabulary they recognise; a common sung reduction is misread as an imperative, turning a phonological phenomenon into a speech act; and LLM pre-editing repairs precisely those forms that make the material analytically interesting.
Second: what the instruments detect depends on how phenomena are operationalised. Idiomatic creativity becomes visible only when fixedness and contextual mismatch are measured jointly (Amin et al. 2021). To test this, we generated a control corpus of 10,800 AI-written lyrics under varied conditions. Although the texts reproduce many surface characteristics of the genre, they are not subject to the same formal constraints as human-produced lyrics. The difference is measurable: pure end rhyme occurs in 83–95% of human texts but only 44–51% of AI-generated ones, while rhyme rate itself interacts with lexical diversity, line length, and word length.
Performed language, I conclude, is measurable – provided we report the constraints our measures compete with.
References
Amin, Miriam, Peter Fankhauser, Marc Kupietz & Roman Schneider. 2021. Data-driven identification of idioms in song lyrics. Proceedings of the 17th Workshop on Multiword Expressions (MWE 2021), 13–22.
Biber, Douglas. 1988. Variation across Speech and Writing. Cambridge: Cambridge University Press.
Koch, Peter & Wulf Oesterreicher. 1985. Sprache der Nähe – Sprache der Distanz. Romanistisches Jahrbuch 36. 15–43.
Schneider, Roman / Broll, Sarah (2026): Revisiting the immediacy-distance continuum: Evidence based automatic text type classification. In: Computational Humanities Research, 2:e24, Cambridge University Press. https://doi.org/10.1017/chr.2026.10041
Schneider, Roman. 2026a. Linguistic resources for the study of pop culture. In Valentin Werner, Cecelia Cutler & Andrew Moody (eds.), Handbook of Language and Pop Culture. Berlin/Boston: De Gruyter Mouton.
Schneider, Roman. 2026b. Pop lyrics through time: Challenges in corpus-based modeling of linguistic and emotional dynamics in German pop lyrics. Proceedings of the 12th Workshop on Challenges in the Management of Large Corpora (LREC 2026), Palma de Mallorca.
Werner, Valentin (2021): Catchy and Conversational? A Register Analysis of Pop Lyrics. In: Corpora 16/2, Edinburgh University Press, 237-270.
Rethinking Linguistic Evidence through the Measurement of Performed Language: A Song Corpus Approach Rethinking Linguistic Evidence through the Measurement of Performed Language: A Song Corpus Approach
Raffaele Zago
University of Catania