Lost in Translation? Exploring the Shift in Grammatical Gender from Latin to Occitan

2026-06-25 · Source: Paper Index on ACL Anthology · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Computational Linguistics · Depth: Expert, quick

Summary

A study explores the diachronic evolution of grammatical gender from Latin to Occitan, focusing on the shift from a tripartite (masculine, feminine, neuter) to a bipartite (masculine, feminine) system in Romance languages. Researchers introduce an interpretable deep learning framework to investigate this phenomenon at both lexical and contextual levels. They found that conventional tokenization strategies are inadequate for low-resource historical language settings, proposing a new tokenizer that significantly improves performance over baselines. The work evaluates the contribution of morphological features to gender prediction at the lexical level and quantifies the impact of different part-of-speech categories on grammatical gender prediction at the contextual level. These analyses collectively characterize how gender information is distributed between a word's lemma and its surrounding sentence context. The codebase, datasets, and results are publicly available.

Key takeaway

For NLP Engineers or Research Scientists working with historical or low-resource languages, this research highlights the critical need for specialized tokenization. You should prioritize developing or adapting robust tokenizers, as conventional methods are insufficient and hinder model performance. Consider integrating interpretable deep learning frameworks to analyze linguistic phenomena, allowing you to quantify the contributions of features like morphology and part-of-speech to better understand language evolution.

Key insights

Deep learning can reveal historical linguistic shifts, especially with robust tokenization for low-resource languages.

Principles

• Historical language analysis requires specialized tokenization.
• Morphological features inform lexical gender prediction.
• POS categories contribute to contextual gender.

Method

An interpretable deep learning framework investigates grammatical gender shifts by analyzing morphological features and part-of-speech contributions at lexical and contextual levels.

In practice

• Develop custom tokenizers for historical texts.
• Analyze feature importance in linguistic models.
• Release research code and datasets publicly.

Topics

• Grammatical Gender
• Diachronic Linguistics
• Latin to Occitan
• Deep Learning
• Tokenization
• Low-Resource NLP

Best for: AI Scientist, NLP Engineer, Research Scientist

Related on AIssential

• Modelling the Morphology of Verbal Paradigms: A Case Study in the Tokenization of Turkish and Hebrew — Tokenization strategy critically impacts Transformer models' ability to represent complex morphology, especially in non-concatenative languages.
• Gender Disambiguation in Machine Translation: Diagnostic Evaluation in Decoder-Only Architectures — Decoder-only MT models exhibit gender bias, but post-training improves contextual awareness and reduces masculine "Prior Bias."
• How Surprising Is Historical Italian to Language Models? Tokenization Tax, Comprehension Tax, and a Simple Mitigation — LLMs exhibit high surprisal and tokenization costs with historical Italian, yet retain robust semantic understanding, mitigable with temporal context.
• The NLP Landscape: From 1960s to 2020s. — Natural Language Processing has evolved from rule-based systems to advanced deep learning, enabling machines to understand and generate human language.
• A corpus-assisted metaphor analysis in portraying businesswomen: Diachronic changes in Hong Kong English news — Corpus-assisted metaphor analysis reveals diachronic shifts in media portrayals of businesswomen.
• Robust Language Identification for Romansh Varieties — The paper titled "Robust Language Identification for Romansh Varieties," authored by Charlotte Model, Sina Ahmadi, and Jannis Vamvas, was presented at the 11th Edition of the Swiss Text Analytics Conf…

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Paper Index on ACL Anthology.