Towards Corpus-Grounded Agentic LLMs for Multilingual Grammatical Analysis
This work addresses the problem of reducing methodological and technical effort in empirical grammar research for linguists, offering an incremental step toward scalable automation of corpus-based grammatical inquiry.
The authors tackled the challenge of automating grammatical analysis from annotated corpora by developing an agentic LLM framework that integrates natural-language task interpretation, code generation, and data-driven reasoning, applied to Universal Dependencies corpora across 13 word-order features and over 170 languages, achieving results that demonstrate feasibility in combining LLM reasoning with structured linguistic data.
Empirical grammar research has become increasingly data-driven, but the systematic analysis of annotated corpora still requires substantial methodological and technical effort. We explore how agentic large language models (LLMs) can streamline this process by reasoning over annotated corpora and producing interpretable, data-grounded answers to linguistic questions. We introduce an agentic framework for corpus-grounded grammatical analysis that integrates concepts such as natural-language task interpretation, code generation, and data-driven reasoning. As a proof of concept, we apply it to Universal Dependencies (UD) corpora, testing it on multilingual grammatical tasks inspired by the World Atlas of Language Structures (WALS). The evaluation spans 13 word-order features and over 170 languages, assessing system performance across three complementary dimensions - dominant-order accuracy, order-coverage completeness, and distributional fidelity - which reflect how well the system generalizes, identifies, and quantifies word-order variations. The results demonstrate the feasibility of combining LLM reasoning with structured linguistic data, offering a first step toward interpretable, scalable automation of corpus-based grammatical inquiry.