Inhalte

This X-Student Research Group investigates mental imagery as a novel linguistic marker in psychiatric research on schizophrenia. Building on the ongoing project 'Detecting Linguistic Markers of Schizo-Bipolar Disorders Using NLP,' the study aims to extract imageability values from linguistic corpora to explore their diagnostic potential. Schizophrenia affects 1% of the population, yet objective biomarkers remain scarce. Since language disturbances are characteristic of the disorder, NLP provides a promising tool for automated analysis. Prior research from our group (Nenchev et al., 2024) suggests that imageability declines post-onset of schizophrenia. Students in this interdisciplinary seminar will analyze existing corpora, including texts by Unica Zrn and Robert Walser, as well as patient interviews, to assess imageability using linguistic norms, large language models (LLMs), and diffusion models. They will also explore whether image generation models, such as stable diffusion, can provide new insights into the alignment between language and mental imagery. Key research questions include whether imageability serves as a reliable linguistic marker for schizophrenia and how computational models can enhance its measurement.

Kontakt

ivan.nenchev@charite.de, benjamnin.wilck@mail.huji.ac.il

Link zum Vorlesungsverzeichnis

folgt