Projects

LANXIET-IA : Linguistic Expression of Anxiety Across Languages and Artificial Intelligence

The LANXIET-IA project is positioned at the intersection of linguistics, psychology, and artificial intelligence, with the aim of systematically analyzing how anxiety is expressed linguistically across a wide range of languages and cultural contexts. It is grounded in the assumption that anxiety, while sharing common psychological dimensions, manifests differently depending on the lexical, syntactic, pragmatic, and discursive resources specific to each language, as well as the sociocultural frameworks in which those languages are embedded.

The study relies on artificial intelligence–based language models to process large-scale multilingual textual corpora drawn from diverse sources, including spontaneous productions, personal narratives, online interactions, and media discourse. These corpora are examined in order to identify linguistic markers of anxiety, such as lexical choices, grammatical structures, enunciative modalities, recurring metaphors, and discursive strategies used to convey worry, negative anticipation, or uncertainty. The use of language models enables large-scale comparison of these phenomena while preserving sensitivity to language-specific features.

Particular attention is given to themes that generate contemporary forms of anxiety, especially geopolitical anxiety (Guidere, 2025), understood as concerns related to conflicts, international tensions, security threats, and transformations of the global order. The project investigates how such concerns are verbalized differently across linguistic and cultural contexts, revealing distinct framings of risk, fear, and collective responsibility. This approach highlights the ways in which languages structure the subjective experience of anxiety in relation to global issues.

LANXIET-IA ultimately aims to provide a global overview of anxiety in the contemporary world, not as a uniform measurement, but as a mapping of its linguistic and cultural expressions. By combining fine-grained linguistic analysis with computational methods, the project contributes to a deeper understanding of the relationships between language, emotions, and sociopolitical context, while opening methodological perspectives for the comparative study of affects on an international scale.

CRIMENTAL-IA : Linguistic Expression of Violence Across Languages and Artificial Intelligence

The CRIMENTAL-IA project is embedded in an interdisciplinary research framework that brings together linguistics, social sciences, psychology, and artificial intelligence in order to examine how violence is expressed linguistically across languages and cultural contexts. It is based on the premise that violence, whether physical, symbolic, institutional, or verbal, is not manifested solely through actions, but also through specific linguistic forms that vary according to language, social norms, and historical frameworks.

The study relies on the analysis of large-scale multilingual corpora, including testimonies, judicial narratives, media discourse, digital interactions, and narrative productions. Artificial intelligence–based language models are used to identify and compare linguistic markers of violence, such as lexical fields, syntactic constructions, enunciative modalities, forms of reference to perpetrators and victims, and discursive processes of justification, normalization, or denunciation. This approach makes it possible to reveal both regularities and variations in how violence is explicitly stated, narrated, or left implicit.

Particular attention is paid to cultural differences in the conceptualization and verbalization of violence (Guidere, 2024). The project explores how some languages tend to make violent acts explicit, while others favor indirect, euphemistic, or metaphorical forms. It also analyzes how normative and legal frameworks shape discourse by determining what can be named, mitigated, or rendered unspeakable. In this context, artificial intelligence enables the articulation of micro- and macro-level analyses, combining linguistic precision with large-scale comparison.

CRIMENTAL-IA ultimately aims to produce an international mapping of the linguistic expressions of violence, showing how languages and cultures structure the perception, understanding, and narration of violent phenomena. By linking language, power, and social norms, the project contributes to a deeper understanding of the discursive mechanisms of violence and opens new perspectives for the comparative analysis of criminal and social discourse in a globalized context.