The paper evaluates four metrics—CKA, ANC, GMM dominance per token, and ILO—used to measure cross‑lingual representation sharing in multilingual language models. Across 21 models ranging from 125 M to 14 B parameters, the metrics disagree, and the authors attribute this to anisotropy, where representations cluster in a narrow embedding cone. Only ILO shows a strong, robust correlation with cross‑lingual transfer performance (Spearman’s ρ = 0.90) after controlling for model size, family, and task variation, leading the authors to recommend ILO as the primary metric alongside anisotropy diagnostics.
By Oskar Holmstr\"om, Marcel Bollmann, Marco Kuhlmann
The paper introduces an Encoding Probe that reconstructs language model representations using interpretable features, addressing limitations of traditional decoding probes such as incomparable feature contributions and correlation effects. It evaluates this approach on text and speech transformer models, examining features from acoustics, phonetics, syntax, lexicon, and speaker identity. Findings reveal that speaker-related effects vary with training objectives and datasets, while syntactic and lexical features independently contribute to reconstruction, offering a complementary perspective on model interpretation.
By Gaofei Shen, Martijn Bentum, Tomas O. Lentz, Afra Alishahi, Grzegorz Chrupa{\l}a
arXiv:2602.01161v2 Announce Type: replace
Abstract: The global deployment of large language models (LLMs) has raised concerns about cultural misalignment, yet the linguistic properties of fine-tuning...
By Reem I. Masoud, Chen Feng, Shunta Asano, Saied Alshahrani, Philip Colin Treleaven, Miguel R. D. Rodrigues
arXiv:2605. 27618v2 Announce Type: replace Abstract: Despite the wide use of explainability techniques to attempt to understand the behavior of Artificial Intelligence (AI), the generated explanations may not always be reliable.
By Tom\'as Pereira, Jo\~ao Vitorino, Eva Maia, Isabel Pra\c{c}a
arXiv:2606. 05180v1 Announce Type: cross Abstract: Automated scoring models are increasingly used to assign rubric-based quality ratings to complex language performances, including classroom transcripts, yet they typically provide little insight into why a particular score is produced.
By Ivo Bueno, Babette B\"uhler, Philipp Stark, Tim F\"utterer, Ulrich Trautwein, Dorottya Demszky, Heather Hill, Enkelejda Kasneci
arXiv:2609.24650v1 Announce Type: new
Abstract: Readability assessment is essential for tailoring texts to intended audiences across educational, healthcare, and information retrieval domains. Howeve...
By Rapha\"el Thieffry, Matej Martinc
arXiv:2608.29034v1 Announce Type: cross
Abstract: A wide range of methods have been proposed for interpreting language models, delivering important insights into their inner workings. However, differ...
By Zhang Enyan, R. Thomas McCoy
The paper "When Explanations Cannot Be Read: Measuring and Correcting SHAP and LIME Rendering for Right-to-Left Languages" identifies that standard SHAP and LIME visualizations, designed for left‑to‑right scripts, fail to display attribution values correctly for right‑to‑left languages such as Urdu, Arabic, Persian, and Hebrew. It introduces SHAP‑RTL, a rendering layer that preserves the original attribution values while correcting reading direction, script shaping, and font selection for each language. The authors evaluate SHAP‑RTL on hate‑and‑offensive‑language datasets using TF‑IDF and logistic regression, showing that default rendering yields high character error rates, while SHAP‑RTL maintains correct visualizations across all tested languages.
By Rameesha Zia, Muhammad Shahid Iqbal Malik
arXiv:2605.28190v2 Announce Type: replace
Abstract: Embedding benchmarks like MTEB report a single score per model, implicitly treating robustness as a static, scalar property. We argue that embeddin...
By Manuel Frank, Haithem Afli
The paper investigates multilingual confidence calibration in large language models, revealing that non‑English languages are systematically less well calibrated than English. By analyzing internal representations, the authors find that late‑intermediate layers provide a more reliable confidence signal than the final layer, which is biased by English‑centric training. They propose training‑free methods such as Language‑Aware Confidence Ensemble (LACE) to adaptively select optimal layers per language, aiming to improve global equity and trustworthiness of LLMs.
By Ej Zhou, Caiqi Zhang, Tiancheng Hu, Chengzu Li, Nigel Collier, Ivan Vuli\'c, Anna Korhonen
arXiv:2609.36194v1 Announce Type: new
Abstract: Extracted sentiment directions can vary across samples even when downstream sentiment classification remains accurate. To evaluate direction reproducib...
By Muhammad Abdullahi Said, Abass Oguntade, Elisha Komolafe, Babangida Sani, Fatima Muhammad Adam, Muhammad Sammani Sani
The paper compares generative and encoder-based neural models for multilingual Named Entity Recognition (NER) across the eleven languages of the Naamapadam benchmark. Five classic model families, four decoder-only large language models fine‑tuned with LoRA and 4‑bit NF4 quantisation, and nine generative models in zero‑to‑5‑shot inference were evaluated under strict CoNLL span‑level metrics. Encoder-based models (mBERT and XLM‑R) achieved substantially higher F1 scores—up to 0.675 on Hindi—than any generative architecture, with gaps of 7.5–40 percentage points; the best few‑shot result reached only 28% of the encoder baseline. The study identifies three language clusters (encoder‑dominant, partial‑coverage, and failure‑zone) and offers deployment guidelines based on transfer learning and low‑resource NLP principles.
By Jakkala Mahesh, Jatavath Shravan Kumar, Komalla Shivani, Sujoy Sarkar