In this research, we introduce LoRA probing, a lightweight approach for observing how core syntactic abilities emerge during LLM pretraining. Leveraging OLMo-2’s public intermediate checkpoints, we trace learning curves across 24 pretraining stages on 33 Universal Dependencies languages by fine-tuning LoRA with step-by-step parsing instructions and a simple tabular output. To fit the relatively short context length of the OLMo-2, we design a compact 2-step-no-form prompt template and this matches the baseline in average accuracy while halving the context length and substantially increasing throughput, enabling efficient large-scale evaluation. Token Recall surpasses 0.9 within the first 1–2K pretraining steps, indicating that stable output formatting emerges early. Despite OLMo-2-7B’s English-centric pretraining, LAS exceeds 80 points in 29 of 33 languages; however, relations such as iobj and csubj show delayed onset and instability across many languages. LoRA probing thus provides a practical, reproducible lens on the cross-lingual dynamics of syntactic acquisition during LLM pretraining.

omnes flores is an NLP framework based on Universal Dependencies (UD) that utilizes multilingual Large Language Models (LLMs), and its default model is trained on data from 40 UD languages comprising 40 treebanks. For the EvaLatin 2026 Dependency Parsing Tasks, we extended the training data of omnes flores by incorporating six public Latin treebanks from UD and trained a dependency parsing model using the extended training data. The dependency parser of omnes flores normally takes a list of word FORM values as input. However, since the EvaLatin 2026 test data includes an UPOS column, we investigated whether incorporating both FORM and UPOS during both training and inference could improve parsing accuracy. Our experiments show that training using both FORM and UPOS improves performance by 0.5-1.0 LAS points on Prose compared with training using only FORM, but decreases performance by 5 points on Poetry.

Recent advances in large language models (LLMs) have enabled impressive performance in various tasks. However, standard prompting often struggles to produce structurally valid and accurate outputs, especially in dependency parsing. We propose a novel step-by-step instruction strategy, where universal part-of-speech tagging precedes the prediction of syntactic heads and dependency labels, and a simplified CoNLL-U like output format, our method achieves state-of-the-art accuracy on Universal Dependencies datasets across 17 languages without hallucination or contamination. We further show that multilingual fine-tuning simultaneously improves cross-language generalization performance. Our results highlight the effectiveness of explicit reasoning steps in LLM-based parsing and offer a scalable, format-consistent alternative to bracket-based approaches.

In this study, we have developed Universal Dependencies (UD) resources for spoken Japanese in the Corpus of Everyday Japanese Conversation (CEJC). The CEJC is a large corpus of spoken language that encompasses various everyday conversations in Japanese, and includes word delimitation and part-of-speech annotation. We have newly annotated Long Word Unit delimitation and Bunsetsu (Japanese phrase)-based dependencies, including Bunsetsu boundaries, for CEJC. The UD of Japanese resources was constructed in accordance with hand-maintained conversion rules from the CEJC with two types of word delimitation, part-of-speech tags and Bunsetsu-based syntactic dependency relations. Furthermore, we examined various issues pertaining to the construction of UD in the CEJC by comparing it with the written Japanese corpus and evaluating UD parsing accuracy.