ThaiTrees is a 342‑million‑token corpus of Thai text spanning news, Wikipedia, spoken transcripts, and social media, automatically parsed under the Universal Dependencies framework. The authors provide a reproducible pipeline for cleaning, processing, and parsing the data, and release the resulting CoNLL‑U files and a frequency lexicon in machine‑readable formats. This resource enables researchers to search grammatical relations and study syntactic distributions at scale.
arXiv:2609.37883v1 Announce Type: new Abstract: Pre-trained language models (PLMs) with encoder-based architectures have shown impressive capabilities in zero-shot cross-lingual transfer for various...
arXiv:2606. 12708v1 Announce Type: cross Abstract: Despite their linguistic diversity and global significance, African languages remain underrepresented in research and resources to support NLP.
The paper introduces AraSEG, a new Arabic sentence segmentation corpus covering eight genres and diverse punctuation and document structures. Experiments using AraSEG evaluate large language models, lightweight encoders, and dependency parser-based models, revealing that lightweight encoders and parser-based models outperform LLMs under the most challenging conditions. The study also shows that increasing training data size and genre diversity eventually saturates performance, that cross‑genre generalization remains difficult, and that accurate sentence segmentation significantly improves downstream dependency parsing.
arXiv:2608.23448v1 Announce Type: new Abstract: This paper presents new Cantonese ParGram resources and evaluates LLMs for knowledge-driven grammar engineering within a controlled experimental paradi...
arXiv:2608. 06179v1 Announce Type: new Abstract: Preference optimisation has proven effective for improving large language models but typically relies on costly human preference annotations.
The paper presents a pipeline that leverages large language models to extract grammatical rules, example sentences, and lexicons from descriptive grammar books, producing synthetic parallel corpora for fine‑tuning machine translation models. Evaluated on three low‑resource languages—Kalamang, Tuatschin, and Mandan—the synthetic data improves translation quality over seed‑data baselines in 75% of configurations for Kalamang and 59% for Tuatschin, achieving up to +8.8 ChrF++ gains. A factorial study across 96 configurations identifies which combinations of target part‑of‑speech, retrieval granularity, and sample volume drive performance gains and where they fail, demonstrating that static linguistic documentation can be repurposed for practical translation tools for severely under‑resourced languages.
arXiv:2608. 10137v1 Announce Type: cross Abstract: Grammar Constrained Decoding (GCD) forces Language Models (LMs) to produce syntactically valid outputs by masking out non-conforming tokens at each step.
Preference optimisation has proven effective for improving large language models but typically relies on costly human preference annotations. Extending these methods to morphologically rich, low-resource languages remains challenging because such annotations are scarce.
arXiv:2606. 18856v1 Announce Type: cross Abstract: Sequence labelling, a core task of Natural Language Processing (NLP), consists in assigning each token of an input sentence a label.
arXiv:2606. 02991v1 Announce Type: cross Abstract: We introduce TypewriterLM, a 7.
arXiv:2603.01243v3 Announce Type: replace Abstract: Large language models (LLMs) are powerful tools that have found applications beyond human-machine interfaces and chatbots. Beside free-form generat...
The paper proposes a modular tokenizer framework for multilingual large language models, allowing the creation of language‑specific subtokenizers that match monolingual compression quality. It introduces a pretraining strategy that samples these subtokenizers to limit predictions to relevant vocabularies, enabling efficient training and inference. This approach reduces memory usage and speeds up inference without compromising performance.