Plenarvorträge zum Rahmenthema "Tracing Patterns – Sprachlichen Mustern auf der Spur"

Keynote 1: February 25, 2026, 09:30-10:30, Audimax 

The question of how intonation encodes linguistic meaning has two components: What kind of variation in intonational form plays a role in encoding meaning? What meaning distinctions are encoded via variation in intonational form? Traditional approaches to these questions assume a form-meaning mapping based on discrete categories in both domains: Categorical tone features marking prominence (pitch accents) or prosodic phrase boundaries (edge tones) associate with categorical distinctions in meaning related to information structure (e.g., focus, givenness) or speech acts (e.g., question, assertion). Yet a mounting body of empirical evidence paints a more complex picture. In this talk I draw connections between research findings on the variable robustness of intonational distinctions in production and perception [1]; the variable intonational encoding of information structure [2]; and the variable influence of intonational tunes on the interpretation of pragmatic meaning [3, 4]. I argue that when viewed together, these findings call for a dynamical approach to intonation that relates discrete and continuous distinctions in intonational form to graded and/or hierarchically ordered distinctions in pragmatic meaning. This approach is illustrated in the dynamical systems analysis of American English intonation [5] and its mapping to a dynamical system of meaning [6].

[1] Cole, Steffman, Shattuck-Hufnagel, Tilsen. 2023. Laboratory Phonology, 14(1).

[2] Im, Cole, Baumann. 2023. Laboratory Phonology, 14(1)

[3] Sostarics, Cole. 2023. Proc. Interspeech

[4] Sostarics, Ronai, Cole. 2024. Proc. Experiments in Linguistics Meaning

[5] Iskarous, Cole, Steffman. 2024. Jn Phonetics

[6] Cole, van Elswyk, Iskarous, Vigneaux. In progress.

Keynote 4: February 27, 2026, 10:00-11:00, Audimax

Popular ideas and academic hypotheses suggest that there are distinct patterns of linguistic aging. These range from 'disfluent' features in discourse to a decline in the ability to process complex syntactic structures. However, the expression of individual features linked to these patterns shows high variability between speakers. Based on French data from the LangAge corpus, we discuss how linguistic aging is represented in individual trajectories among same-generation speakers. The use of statistical models, such as hierarchical clustering, enables us to sketch distinct individual or group profiles within the corpus, with their respective patterns of frequent or infrequent variants.

Next, we take a closer look at the stable elements in a series of interviews with the same individuals. Stable elements can be identified in retold stories, presenting segments of formulaic language. Using as the basis for our findings up to nine interviews with a panel of speakers, we will discuss the patterns of formulaicity versus variability in these repetitions. Special attention will be given to prosody, particularly the intonational contours that form patterns with underlying linguistic units. Supported by recent research findings, we will suggest that these patterns enhance memorability.

Keynote 3: February 27, 2026, 09:00-10:00, Audimax 

Evaluation is a complex practice that involves assigning a value as well as justifying and legitimising this assignment, and can take different forms depending on the constellation of actors and media involved. On the linguistic side, evaluation practices are extremely patterned. Established patterns of evaluative language use and their underlying interpretative schemata create routines and ensure the social acceptability of evaluative judgements and their consequences.

Due to the patternedness of evaluation practices, corpus linguistic approaches are particularly well suited to their study. Theoretical concepts from the field of the sociology of (e)valuation (Lamont 2012; Nicolae et al. 2019) can thus be operationalised and made fruitful for linguistic empiricism and theory. In my presentation, I would like to demonstrate this in a series of case studies on review practices in various social domains, such as (open) peer review in academia (Falk Delgado, Garretson & Falk Delgado 2019) or customer reviews on ticket agency websites (Meier-Vieracker 2022). Finally, I would like to explore the question of whether and to what extent large language models, with their fine sensitivity to patterned language, can or cannot play along with the language game of evaluation (Schneider 2024).

References

Falk Delgado, Alberto, Gregory Garretson & Anna Falk Delgado. 2019. The language of peer review reports on articles published in the BMJ, 2014–2017: an observational study. Scientometrics 120(3). 1225–1235. doi.org/10.1007/s11192-019-03160-6.

Lamont, Michèle. 2012. Toward a Comparative Sociology of Valuation and Evaluation. Annual Review of Sociology 38(1). 201–221. doi.org/10.1146/annurev-soc-070308-120022.

Meier-Vieracker, Simon. 2022. Between consumers and fans: Writing fan reports as a multifunctional evaluation practice. Journal of Cultural Analytics 7(2). 4–31. doi.org/10.22148/001c.33570.

Nicolae, Stefan, Martin Endreß, Oliver Berli & Daniel Bischur. 2019. Soziologie des Wertens und Bewertens. In Stefan Nicolae, Martin Endreß, Oliver Berli & Daniel Bischur (eds.), (Be)Werten. Beiträge zur sozialen Konstruktion von Wertigkeit (Soziologie des Wertens und Bewertens), 3–20. Wiesbaden: Springer Fachmedien. doi.org/10.1007/978-3-658-21763-1_1.

Schneider, Jan Georg. 2024. Intelligible Texturen. Welche Rolle kann ChatGPT bei der Aufsatzbewertung spielen? VK:KIWA. zenodo.org/records/10877034.

Keynote 2: February 25, 2026, 11:30-12:30, Audimax

Finding patterns is one of the favourite pastimes of corpus linguists. These patterns can take various forms, e.g. collocations, recurrent n-grams, Part-of-Speech patterns, or constructions at various levels of abstraction. The toolsets that have evolved over the past 40 years or so include taggers, lemmatisers, parsers, semantic annotation tools, … – and all of them have in common that they work on text. The automatic analysis of audio and video data is much less mature and is often based on (computer-assisted) manual analysis. In this presentation, I will present opportunities and challenges of big data analysis in multimodal corpus linguistics, based both on audio and video analysis tools. In addition to relevant datasets, both semi-automatic and fully automatic analyses will be discussed, e.g. on phonetic reduction and of aspects of co-speech gesture. Finally, I will give a brief outlook into the application of Large Language Models (LLMs) for research on multimodal phenomena.