Seminal
The magazine and atlas of EU-funded researchA new publication from the web and AI team at Scott Charnley

Atlas / Israel

European Research Council grant · Running

WordOrigins From Understanding to Generating Meaningful Language: Unravelling the Origins of Our Words

Coordinator
The Hebrew University of Jerusalem, Israel
Runs
1 May 2026 to 30 April 2031
Status
Running
EU funding
€1.5m
Total cost
€1.5m
Partners
2 organisations in 2 countries
Type
European Research Council grant
Grant agreement
101222215

What the project is doing

In the project's own words, from its summary on CORDIS.

WordOrigins will establish an understanding of language generation by integrating ecological neuroscience and computational models. Language is central to human experience, enabling communication and thought expression. However, the brain mechanisms that generate meaningful language, convey coherent thoughts and elicit responses still need to be understood. I will address this gap by leveraging advances in electrophysiology (iEEG) and artificial intelligence, particularly Large Language Models (LLMs).

These technologies offer the opportunity to explore unconstrained language generation and uncover the cognitive processes underlying it. The research combines two key elements: Ecological Electrophysiological Measurements: iEEG data from participants engaged in natural conversations provide insights into how the brain processes language in real-life situations; and Modeling Cognitive Processes with LLMs: LLMs offer a computational framework for modeling and interpreting language generation, generating testable predictions about how language is processed.

This proposal focuses on several fundamental endeavours: Planning Multiple Words in Language Generation studies how the brain plans speech by predicting multiple future words. Modeling the Cognitive Effort of Generating Meaningful Words investigates how the brain produces more effortful, meaningful words. Multilingual Neural Encoding and Language Representation focuses on bilingual patients to understand whether early language processing is abstract or language-specific.

Eliciting Meaningful Interaction Through Neural Synchronization examines whether LLM-based agents can enhance neural synchronization and trust during interaction. Disentangling Synchronization from Shared Input explores how cognitive mechanisms beyond shared stimuli affect synchronization. Together these elements promises us concrete foundations for the challenge of understanding how our ideas are formed and communicated.

Topics

Who is involved

OrganisationCityCountryRole
The Hebrew University of JerusalemJerusalemIsraelCoordinator
The Chancellor Masters and Scholars of the University of CambridgeCambridgeUnited KingdomParticipant

Source: CORDIS, European Commission. Data of 5 Nov 2026, reused under the Commission's reuse policy. View on CORDIS · Grant DOI