Two postdoctoral researcher positions in ERC project on computational modelling of language in (...)
Barcelona
EUR 50.000 - 70.000
The two positions form part of work package 1 in the ERC Synergy grant “DELTA-LANG – The delta of language” (no. 101118756, 2024-2029), which is EU-funded with 9,939,329€ and spread over four European sites (Barcelona, Zürich, Groningen, and Tromsoe). The overall project’s aim is to map clinical fluctuations in symptom states onto language states and to then explore how these language states map onto brain states, with a view to enhance mechanistic understanding. Our ultimate clinical goal is to track and predict transitions between symptom states in psychosis.
The focus of this specific work package at Barcelona lies in developing a suite of computational linguistic measures of spontaneous speech that will be most promising in tracking symptom state fluctuations. Several longitudinal datasets containing both speech and fMRI data will be generated in the course of the project, and we are already working with a wide range of cross-sectional datasets in a large cross-linguistic database. We currently focus on semantic metrics, specifically using embeddings from large language models (LLMs) to understand the structure, geometry, dimensionality and dynamic properties of the semantic spaces that we navigate when we speak. Both appointed candidates will develop independent research lines within this project and are encouraged to make specific suggestions in this respect in their application letters. We value a diversity of scientific backgrounds in this cross-disciplinary project. Appointed candidates will benefit from an extensive network of clinical collaborators, the resources of the consortium as a whole, and a rich and diverse local research environment at the UPF, particularly in the areas of computational linguistics and cognitive and computational neuroscience.
Please direct any prior inquiries to the PI at **************@upf.edu.
Tasks:
* Play leading roles in different research lines dedicated to defining language states that will map onto symptom and brain states in a longitudinal setting.
* Creatively develop new theoretical concepts, methodological tools, and analytical pipelines for this purpose.
* Have an interest or skills in neuroimaging data analysis, particularly MRI, and apply them in multimodal datasets comprising both fMRI and speech, across different languages. EEG and MEG datasets may also be used to relate metrics extracted from LLMs to brain dynamics.
* Help to establish and co-lead an interdisciplinary team of other researchers in the work package.
* Collaborate closely with research and clinical teams at other project sites.
* Be lead author on high-level publications.
#J-18808-Ljbffr