Lost in Pronunciation: Detecting Chinese Offensive Language Disguised by Phonetic Cloaking Replacement

TL;DR

This paper introduces a taxonomy and dataset for Chinese phonetic cloaking replacement (PCR), revealing current detection methods' limitations and proposing a Pinyin-based prompting strategy to improve robustness in toxicity detection.

Contribution

It provides the first comprehensive taxonomy of Chinese PCR, a realistic benchmark dataset, and a lightweight Pinyin-based prompting technique to enhance detection accuracy.

Findings

State-of-the-art models achieve only 0.672 F1-score on PCR detection.

Zero-shot chain-of-thought prompting reduces detection performance.

Pinyin-based prompting significantly recovers detection accuracy.

Abstract

Phonetic Cloaking Replacement (PCR), defined as the deliberate use of homophonic or near-homophonic variants to hide toxic intent, has become a major obstacle to Chinese content moderation. While this problem is well-recognized, existing evaluations predominantly rely on rule-based, synthetic perturbations that ignore the creativity of real users. We organize PCR into a four-way surface-form taxonomy and compile \ours, a dataset of 500 naturally occurring, phonetically cloaked offensive posts gathered from the RedNote platform. Benchmarking state-of-the-art LLMs on this dataset exposes a serious weakness: the best model reaches only an F1-score of 0.672, and zero-shot chain-of-thought prompting pushes performance even lower. Guided by error analysis, we revisit a Pinyin-based prompting strategy that earlier studies judged ineffective and show that it recovers much of the lost accuracy. This study offers the first comprehensive taxonomy of Chinese PCR, a realistic benchmark that reveals current detectors' limits, and a lightweight mitigation technique that advances research on robust toxicity detection.