Diachronic development of text types

Universität des Saarlandes (Germany)

Objectives: The project will use interpretable computational approaches to model how narratives are linguistically constructed, distributed, and transformed across text types over time. Rather than treating narratives as fixed thematic units, the project understands them as evolving patterns of linguistic choices: recurring lexical associations, framing devices, evaluative expressions, semantic shifts, and changes in the predictability of particular formulations. These patterns can differ substantially across text types, for instance between institutional communication, news media, public commentary, and social media, while also moving between them through processes of uptake, adaptation, simplification, or contestation. Applying corpus-based and probabilistic measures to diachronic and cross-text-type data, Anastasiia will be able to: 1) identify linguistic features through which narratives are expressed in different text types; 2) trace how these narrative patterns change over time; 3) assess how narratives circulate across text types and whether particular formulations become stabilized, contested, or recontextualized; 4) model semantic shifts in key narrative terms and expressions; and 5) develop a transparent computational workflow for the analysis of narrative dynamics, made publicly available where possible. The project thereby contributes to computational linguistics, corpus-based discourse analysis, and the study of language variation and change by showing how narratives emerge, stabilize, and transform across communicative contexts.

Expected Results:
1) A data-driven analysis of narratives across text types.
2) A computationally supported methodology for diachronic analysis of narratives across text types. Anastasiia will present research at D7.3 and D7.4.

Planned secondments: Ukraine (via Remote secondment) (Year 2-3. 6 months. Working on large-scale digital analysis of lexicalization and the evolution of meaning across text types , Also visiting: UCC (stylistics and text reuse); UHELSINKI (modelling the Corpus of Early English Correspondence)