In our next LangSci talk, Amanda Yuile, Research Fellow in the Department of Communication Sciences and Disorders at the Mass General Hospital Institute of Health Professions, will give a talk on „The role of event knowledge in children’s reference comprehension„.
The hybrid talk will take place in building A 2.2, room 2.02. and on MS Teams November 20th at 16:15!
As part of its Emmy Noether program, the German Research Foundation (DFG) is funding Michael Hahn’s work with 1.4 million euros. The project entitled „Understanding and overcoming architectural limitations in neural language models“ will develop new machine learning architectures that make AI systems more robust and logically capable. Current large language models (LLMs) such as ChatGPT are based on the Transformer architecture, which can fundamentally limit their logical reasoning abilities. The project will design model architectures that overcome these limitations and lay methodological foundations for more capable AI systems.
Many congratulations to Michael Hahn on this great accomplishment!
In our first LangSci talk of the semester, George Walkden, Professor of English Linguistics and General Linguistics at the University of Konstanz, will give a talk on „Implicational universals and a self-actuating gradient model of morphosyntactic change„.
The hybrid talk will take place in building A 2.2, room 2.02. and on MS Teams November 6th at 16:15!
A paper by members of project B7 has been published in SKASE Journal of Translation and Interpreting, Special issue: Empirical translation and interpreting studies: „Surprisal explains the occurrence of filler particles in simultaneous interpreting“ by Christina Pollkläsener, Maria Kunilovskaya and Elke Teich.
Bernd Möbius will give a keynote talk on „Information Density and Phonetic Variation“ at the 28th International Conference „Text, Speech and Dialogue“ in Erlangen/Nürnberg, Germany, Aug 25-28, 2025. You can find the abstract below:
In this talk I will take an information-theoretic perspective on speech production and perception. I will explore the relation between information density and phonetic encoding and decoding. Information density of a linguistic unit is defined in terms of surprisal (the unit’s negative log probability in a given context). The main hypothesis underlying our experimental and modeling work is that speakers modulate details of the phonetic encoding in the service of maintaining a balance of the complementary relation between information density and phonetic encoding. To test this hypothesis we analyzed the effects of surprisal on phonetic encoding, in particular on dynamic vowel formant trajectories, plosive voicing, syllable duration, and vowel space size, while controlling for several basic factors related to the prosodic structure, viz. lexical stress and major prosodic boundaries, in the statistical models that accounted for phonetic effects of changes in surprisal (e.g. Malisz et al. 2018, Brandt et al. 2021). Our findings are generally compatible with a weak version of the Smooth Signal Redundancy (SSR) hypothesis (Aylett & Turk 2004, 2006, Turk 2010), suggesting that the prosodic structure mediates between requirements of efficient communication and the speech signal. However, this mediation is not perfect, as we found evidence for additional, direct effects of changes in predictability on the phonetic structure of utterances. These effects appear to be stable across different speech rates in models fit to data derived from six different European languages (Malisz et al. 2018). Moreover, we investigated effects on subword (segmental and syllable) levels and in local prosodic structures (at phrase boundaries), in acoustically clean and in noisy conditions. Our recent findings suggest that speakers make an effort to increase the difference between syllables in high vs. low surprisal contexts in the presence of noise. No interaction was found between noise and surprisal, suggesting that noise-related modifications may be independent of those induced by surprisal. If so, speech production models should include channel-based as well as message-based formulations: although channel coding is not part of linguistic representation (message formulation) during speech planning, it does shape the phonetic output.
In our next LangSci talk, Kyle Mahowald, Assistant Professor at The University of Texas at Austin College of Liberal Arts, will give a talk on „Linguistic Insights from Language Models„.
The hybrid talk will take place in building A 2.2, room 2.02. and on MS Teams July 3rd at 16:15!
We are very happy to announce that within this last call of our Small Grant Scheme within the CRC 1102 we were able to fund 5 projects. We received excellent submissions that convinced reviewers and submission board. Congratulations to all 5 project and thanks to all the reviewers.
Benedict Krieger – Modeling retrieval (N400) and integration (P600) processes in human language comprehension with LLM representations: A research visit at Tilburg University
Isabell Landwehr und Carmen Schacht – CoBra – Compound Branching Resource
Koel Dutta Chowdhury – Beyond English: A Multilingual Pragmatics Benchmark for Evaluating Pragmatic Comprehension in Large Language Models
Kate McCurdy – When Speakers Ignore Linguistic Structure: Modeling German plural production
Diego Alves und Torsten Jachmann – Investigating Factors Influencing the Realization of Overt vs. Null Direct Objects in Brazilian Portuguese
We are very proud to announce that the paper by Uliana Sentsova, Debora Ciminari, Josef Van Genabith and Cristina España-Bonet won the best paper award at the 21st Workshop on Multiword Expressions (MWE 2025) at NAACL 2025.
You can find the paper here.
Many Congratulations!
In our first LangSci talk of the semester, Micha Heilbron, Assistant Professor of Cognitive AI at the University of Amsterdam (UvA), will give a talk on „Language models for cognitive neuroscience: Two success stories and a warning„.
The hybrid talk will take place in building A 2.2, room 2.02. and on MS Teams May 8th at 16:15!
Today Annemarie Verkerk was invited as keynote speaker at the 47th Annual Conference of the German Linguistic Society (March 4–7 2025) to give a talk on „The evolutionary dynamics of language: universals, rates of change, and explanations„. You can find the abstract below:
In the past decades, a shared quantitative framework for typology and historical linguistics has arisen: linguistic phylogenetics. For historical linguistics, phylogenetics can serve as a means to test hypotheses on family relations in a quantitative manner. In its wake, other quantitative tools have emerged, such as computer-assisted cognate detection (List et al. 2018). This talk is concerned with the use of phylogenetics for typology, which involves modeling how linguistic features change on the branches of a phylogeny (family tree). This allows us to test hypotheses about correlation (Dunn et al. 2011), the speed at which typological features change (rate of change, Greenhill et al. 2010) and inferences on what typological features may have looked like in the past (ancestral state estimation, Phillips & Bowern 2022).
In this talk, I will present three case-studies that speak to these three aspects of the evolutionary dynamics of language: correlations, rates of change, and ancestral states. The first case study is on universals, presenting evidence for a large set of Greenbergian implicational universals using phylogenetic methods. The second case study is on indigenous numeral systems, uncovering whether frequently attested base types are favoured evolutionarily through increased rates of change. The third case study is an assessment of reconstructing word order variation in different branches of Indo-European. Throughout, the topic of explaining feature distributions is central: how can we marry functional explanations, for example those rooted in processing efficiency or economy, with linguistic variation and change?