Papers by Michael A. Hedderich
Do We Know What LLMs Don’t Know? A Study of Consistency in Knowledge Probing (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for probing knowledge gaps in large language models are inconsistent and inconsistent. |
| Approach: | They propose a process based on input variations and quantitative metrics to evaluate probing methods that are inconsistent on knowledge gaps. |
| Outcome: | The proposed process exposes two dimensions of inconsistency in knowledge gap probing. |
Semantic Component Analysis: Introducing Multi-Topic Distributions to Clustering-Based Topic Modeling (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Existing methods for topic modeling fail to scale to large datasets or assume one topic per document. |
| Approach: | They propose a topic modeling technique that discovers multiple topics per sample . they evaluate SCA on Twitter datasets in English, Hausa and Chinese . |
| Outcome: | The proposed technique outperforms the LLM-based TopicGPT on Twitter datasets with similar compute budgets. |
Probing LLMs for Multilingual Discourse Generalization Through a Unified Label Set (2025.acl-long)
Copied to clipboard
| Challenge: | Existing work on discourse understanding is constrained by framework-dependent discourse representations. |
| Approach: | They examine whether large language models capture discourse knowledge that generalizes across languages and frameworks. |
| Outcome: | The proposed model can generalize discourse information across languages and frameworks. |
Large Reasoning Models Are (Not Yet) Multilingual Latent Reasoners (2026.findings-acl)
Copied to clipboard
| Challenge: | Recent work shows that large reasoning models arrive at the correct answer before completing textual reasoning steps, indicating the presence of latent reasoning. |
| Approach: | They conduct a systematic investigation of multilingual latent reasoning in large reasoning models across 11 languages. |
| Outcome: | The proposed model arrive at the correct answer before completing the reasoning steps, indicating the presence of latent reasoning. |
Linear Script Representations in Speech Foundation Models Enable Zero-Shot Transliteration (2026.findings-acl)
Copied to clipboard
Ryan Soh-Eun Shim, Kwanghee Choi, Kalvin Chang, Ming-Hao Hsu, Florian Eichin, Zhizheng Wu, Alane Suhr, Michael A. Hedderich, David Harwath, David R. Mortensen, Barbara Plank
| Challenge: | We show that script information is linearly encoded in the activation space of multilingual speech models . modifying activations at inference time induces script change even in unconventional pairings . |
| Approach: | They propose to add script vectors to activations at test time to induce script change . they also show that script information is linearly encoded in the activation space of multilingual speech models . |
| Outcome: | The proposed approach can induce script change even in unconventional language-script pairings. |
Evaluating Robustness of Large Language Models Against Multilingual Typographical Errors (2026.acl-long)
Copied to clipboard
| Challenge: | Large language models (LLMs) are increasingly deployed in multilingual, real-world applications where user inputs introduce typographical errors. |
| Approach: | They propose a multilingual typo generation algorithm that simulates human-like errors based on language-specific keyboard layouts and typing behavior. |
| Outcome: | The proposed model can generate the correct answer ("500") under typos in English, German, and Russian. |
A Survey on Recent Approaches for Natural Language Processing in Low-Resource Scenarios (2021.naacl-main)
Copied to clipboard
| Challenge: | a growing body of work is focused on improving performance in low-resource settings . a goal of this study is to explain how these methods differ in their requirements . |
| Approach: | They propose to analyze data-lean scenarios across different dimensions of data availability to understand which approaches are effective in a specific low-resource setting. |
| Outcome: | The proposed methods enable learning when training data is sparse. |
A Comprehensive Evaluation of Multilingual Chain-of-Thought Reasoning: Performance, Consistency, and Faithfulness Across Languages (2026.findings-eacl)
Copied to clipboard
| Challenge: | Recent work has examined final-answer accuracy in multilingual settings, but the behavior of thinking traces, i.e., the intermediate steps that lead to the final answer, remains underexplored. |
| Approach: | They propose to measure language compliance, answer accuracy, and answer consistency when LRMs are explicitly instructed or prompt-hacked to think in a target language. |
| Outcome: | The proposed model improves in English and other high-resource languages while relying on traces to varying degrees. |
Charting the Landscape of African NLP: Mapping Progress and Shaping the Road Ahead (2025.emnlp-main)
Copied to clipboard
| Challenge: | African languages are often left behind in state-of-the-art natural language processing systems and large language models. |
| Approach: | They analyze 884 research papers on NLP for African languages published over past five years . they identify key trends shaping the field and outline promising directions . |
| Outcome: | The findings identify key trends shaping the field and outline promising directions . the authors analyze 884 research papers on NLP for African languages published over the past five years . |
On the Interplay Between Fine-tuning and Sentence-level Probing for Linguistic Knowledge in Pre-trained Transformers (2020.findings-emnlp)
Copied to clipboard
| Challenge: | linguistic knowledge encoded in pre-trained contextual embeddings is poorly understood . fine-tuning can be used to investigate the representations of pre-train models . |
| Approach: | They propose to investigate fine-tuning of contextualized embedding models through sentence-level probing. |
| Outcome: | The proposed method improves probing accuracy for three pre-trained models. |
Handling Noisy Labels for Robustly Learning from Self-Training Data for Low-Resource Sequence Labeling (N19-3)
Copied to clipboard
| Challenge: | In low-resource environments, self-training is less effective due to unreliable annotations . we combine self-teaching with noise handling to clean the self-labeled data . |
| Approach: | They propose to combine self-training with noise handling to clean unlabeled data . they propose to model clean and noisy labels separately to improve performance . |
| Outcome: | The proposed method performs better than baseline methods on Chunking and NER. |
Transfer Learning and Distant Supervision for Multilingual Transformer Models: A Study on African Languages (2020.emnlp-main)
Copied to clipboard
| Challenge: | Recent studies show that results from high-resource languages cannot be easily transferred to realistic, low-resourced scenarios. |
| Approach: | They analyse performance of multilingual transformer models using available resources for Hausa, isiXhosa and NER and topic classification. |
| Outcome: | The proposed models can achieve with as little as 10 or 100 labeled sentences the same performance as baselines with much more supervised training data. |
MAKIEval: A Multilingual Automatic WiKidata-based Framework for Cultural Awareness Evaluation for LLMs (2025.findings-emnlp)
Copied to clipboard
| Challenge: | Large language models (LLMs) are used globally across many languages, but their English-centric pretraining raises concerns about cross-lingual disparities for cultural awareness . |
| Approach: | They introduce an automatic multilingual framework for evaluating cultural awareness in large language models across languages, regions, and topics. |
| Outcome: | The framework evaluates open-ended text generation, capturing how models express culturally grounded knowledge in natural language. |
Feature-Dependent Confusion Matrices for Low-Resource NER Labeling with Noisy Labels (D19-1)
Copied to clipboard
| Challenge: | Existing approaches to improve supervised labeling with noisy training data do not take the input features into account or they need to learn the noise modeling from scratch. |
| Approach: | They propose to cluster training data using input features and compute different confusion matrices for each cluster. |
| Outcome: | The proposed model improves on low-resource named entity recognition settings in several languages, compared with other models which do not take the input features into account or need to learn noise modeling from scratch. |
What’s the Difference? Supporting Users in Identifying the Effects of Prompt and Model Changes Through Token Patterns (2025.acl-long)
Copied to clipboard
| Challenge: | Existing evaluation methods for prompting for large language models have limitations such as being labor-intensive or lacking insights. |
| Approach: | They propose a new approach that automatically distinguishes between random variations and systematic differences in language model outputs by using token patterns. |
| Outcome: | The proposed method combines both automation and human analysis to provide new insights into established prompt data. |
AfriqueLLM: How Data Mixing and Model Architecture Impact Continued Pre-training for African Languages (2026.acl-long)
Copied to clipboard
Hao Yu, Tianyi Xu, Michael A. Hedderich, Wassim Hamidouche, Syed Waqas Zamir, David Ifeoluwa Adelani
| Challenge: | Continued pretraining (CPT) is a practical route to language adaptation, but improvements on demanding capabilities such as mathematical reasoning are limited. |
| Approach: | They propose to use CPT to adapt large language models to African languages . they use math, code, and synthetic translated data to analyze their models . |
| Outcome: | The proposed models improve on multilingual benchmarks and document-level translation. |
From Weights to Activations: Is Steering the Next Frontier of Adaptation? (2026.acl-long)
Copied to clipboard
Simon Ostermann, Daniil Gurgurov, Tanja Baeumel, Michael A. Hedderich, Sebastian Lapuschkin, Wojciech Samek, Vera Schmitt
| Challenge: | Pre-trained large language models are the basis of a wide range of NLP tasks. |
| Approach: | They propose to use parameter updates and parameter-efficient adaptation to modify behavior of large language models. |
| Outcome: | The proposed method enables local and reversible behavioral change without parameter updates. |
Persistent Personas? Role-Playing, Instruction Following, and Safety in Extended Interactions (2026.eacl-long)
Copied to clipboard
| Challenge: | Persona-assigned large language models are used in education, healthcare and sociodemographic simulations. |
| Approach: | They propose a protocol that combines long persona dialogues and evaluation datasets to create dialogue-conditioned benchmarks that can robustly measure long-context effects. |
| Outcome: | The proposed protocol can measure persona fidelity, instruction-following, and safety in long conversations. |