@elotroalex You are hopefully right in that successful LLMs will be based on more specialized data, but why not create that with colleagues at your university? I argue in my upcoming book exactly this: we must get to know our corpus first, and then create relevant word lists to use either with #corpuslinguistic or #NLP tools or models.
The procedure of using someone else's, esp. a commercial company's, data for historic corpora does normally not give very useful results.

📌#IPrA2023 panel about #corpuslinguistic approaches to #epistemic and #evidential marking in #talkininteraction organized by @jeromejacquin and myself.
Panel contributions:
- Elena Battaglia & Paola Pietrandrea, Incremental and interactive evidential constructions in spoken Italian;
- Jan Fliessbach & Johanna Rockstroh, The placement of wh-objects in French partial interrogatives as a stance-taking device – a challenge for sociolinguistics;
- Mercedes González Vázquez, Interactional uses of evidential markers in Galician;
- Gerda Hassler, French markers of evidentiality in interaction;
- Henrike Helmer & Silke Reineke, Uncertainty vs. dispreference of German "ich weiß nicht": A mixed-methods workflow of a corpus-based interactional linguistic study;
- Ana Keck & Jérôme Jacquin, Expressing a low degree of certainty in interaction: the French epistemic adverb "peut-être" ‘perhaps’ in a corpus of political debates and work meetings;
- Costanza Lucchini & Andrea Rocci, Managing knowledge in financial dialogues: corpus annotation of epistemic modality and evidentiality in earnings conference calls;
- Johanna Miecznikowski, Elena Battaglia & Christian Geddo, An utterance-centered corpus-pragmatic approach to the categorization of information source in spoken Italian;
- Ozan Mustafa & Gustav Kaltenböck, Evidential "last I checked" expressions;
- Flavio Pisciotta, Distribution and function(s) of parenthetical "mi sembra/mi pare" in spoken Italian;
- Clotilde Robin & Jérôme Jacquin, Reportive evidentiality in French-in-interaction: distribution and pragmatic functions of reportive evidentials;
- Stefan Schneider, "J'imagine" as deverbal discourse marker in spoken French. Position in turns and interactions;
- Tanja Trebucchi & Serena Coschignano, Discourse functions of "look" and "see" as evidential and epistemic markers in Italian political discourse: a corpus based study.
Meet you all in Brussels next year 🎉 https://pragmatics.international/page/Brussels2023
@corpuslinguistics