Natural Language-based Assessment of L2 Oral Proficiency using LLMs

TL;DR

This paper investigates using large language models with natural language descriptors to assess second language oral proficiency, demonstrating competitive performance and enhanced interpretability in a zero-shot setting.

Contribution

It introduces a natural language-based assessment approach using LLMs with can-do descriptors, showing its effectiveness and generalizability compared to specialized models.

Findings

Achieves competitive performance with LLMs using textual descriptors.

Outperforms BERT-based models trained specifically for assessment.

Effective in mismatched task settings and across languages.

Abstract

Natural language-based assessment (NLA) is an approach to second language assessment that uses instructions - expressed in the form of can-do descriptors - originally intended for human examiners, aiming to determine whether large language models (LLMs) can interpret and apply them in ways comparable to human assessment. In this work, we explore the use of such descriptors with an open-source LLM, Qwen 2.5 72B, to assess responses from the publicly available S&I Corpus in a zero-shot setting. Our results show that this approach - relying solely on textual information - achieves competitive performance: while it does not outperform state-of-the-art speech LLMs fine-tuned for the task, it surpasses a BERT-based model trained specifically for this purpose. NLA proves particularly effective in mismatched task settings, is generalisable to other data types and languages, and offers greater interpretability, as it is grounded in clearly explainable, widely applicable language descriptors.