The Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) has established the Priority Programme “Robust Assessment & Safe Applicability of Language Modelling: Foundations for a New Field of Language Science & Technology (LaSTing)” (SPP 2556). This program aims to foster interdisciplinary research at the intersection of cognitive language sciences and language technology. The program is set to run for six years, divided into two phases, with this call specifically for proposals for the initial three-year funding period.

Programme Goals and Focus

The overarching goal of the Priority Programme LaSTing is to promote cross-disciplinary efforts dedicated to the understanding, testing, and safe application of modern language technology, particularly focusing on language modelling. The program seeks to address foundational and methodological issues in this rapidly evolving field.

This initiative targets researchers in the interdisciplinary field of cognitive and computational language sciences, including linguistics, psychology, neuroscience, computational linguistics, artificial intelligence, philosophy, and computer science. The program encourages contributions that bring concepts and methods from the theoretical/empirical language sciences to bear on core issues.

Individual projects are expected to relate to at least one of the Priority Area’s core issues: robust assessment, safe applicability, and foundational questions. The Priority Programme especially encourages contributions that seek to address these core issues by bringing to bear concepts and methods from the theoretical/empirical language sciences.

Core Research Areas

Robust Assessment involves developing standards for the methodology of testing and assessment of language models. Methodology should be generalisable, transferable, and reproducible, remaining relevant to future models and datasets.

Safe Applicability focuses on ensuring language technology is conceptually sound, validated, ethical (bias- and harm-free, privacy-respecting), and economical (minimising data requirements and energy consumption). This is particularly important in high-stakes applications, including scientific research.

Foundational Questions explore the core mechanisms of language models, the impact of training data and objectives, and their role in scientific research into human language. This includes addressing the nature of language models and their potential limits.

Examples of Research Questions

The programme encourages research into various questions, including:

• Behavioural Assessment: Developing robust methods for experimentally assessing the capabilities of language models.

• Representations & Mechanisms: Investigating the information retrievable from language models' latent representations.

• Training & Optimisation: Understanding language models in terms of their optimisation and comparing this with human language learning.

• Task Decomposition Models: Establishing best practices for using language models as part of a larger composition of tasks.

• Resource Efficiency: Solving problems of data-hunger and computational costs.

• Alternative Models: Exploring how language science can benefit from models beyond text-to-text language models.

• Ontological Status: Examining whether language models are models or theories of language.

• Explanatory Potential: Investigating how novel language technology can be used in support of explanations.

• LM capabilities: Identifying the limits of LM capabilities and why.

Work focused mainly on improving system performance or merely seeking new areas of application with established tools is outside the scope of this Priority Programme.

Eligibility and Submission Details

The Priority Programme encourages broad and deep interdisciplinary collaboration and implements measures to support diversity, networking, and the success of early career researchers. Early career researchers are explicitly encouraged to submit their own proposals.

Proposals must be submitted to the DFG via the elan portal by 30 September 2025. Registration in the elan portal is required by 23 September 2025 for first-time applicants. When preparing your proposal, please note the Guidelines Priority Programme (DFG form 50.05) and the Proposal Preparation Instructions – Project Proposals (DFG form 54.01).

By submitting a proposal, applicants agree to the DFG sharing necessary information with the Priority Programme coordinator after the call deadline. Funding decisions are expected by March 2026.

Equity, Diversity, and Good Research Practice

The DFG strongly welcomes proposals from researchers of all backgrounds, genders, and career stages. The DFG encourages female researchers in particular to submit proposals.

DFG funding may only be awarded to research institutions that have implemented the guidelines laid down in the Code of Conduct for Safeguarding Good Research Practice. Verify implementation within your institution to avoid delays in funding disbursement.

The DFG takes the protection of your personal data and its confidential treatment extremely seriously. Therefore, please refer to the DFG’s Privacy Policy.

Contact Information

For scientific enquiries, contact Professor Dr. Michael Franke at michael.franke@uni-tuebingen.de. For programme-related questions, contact Dr. Helga Weyerts-Schweda at helga.weyerts-schweda@dfg.de. For administrative questions, contact Melanie Klein at melanie.klein@dfg.de.

More information is available at www.lasting-spp.org.