AutoSSD: A System for Automated Detection of Similar Speech Responses in Language Tests

Abstract

Evaluating spoken language proficiency stands as a pivotal component of language assessment. A well-known method for cheating in spoken language tests involves reciting pre-scripted responses. This paper describes a system developed to detect this type of cheating, called the Automated Speech Similarity Detector (AutoSSD). AutoSSD leverages artificial intelligence (AI) transcription to convert spoken responses into texts. It then calculates various metrics of verbatim similarity between the texts, flags pairs whose similarity levels exceed set thresholds, and provides an interactive interface for a subsequent expert review. Our system has proven effective in practical operation, and our findings can help other spoken language assessments bolster their test security measures.

Suggested citation: Fauss, M., Hao, J., Li, C., Palmer, M., & Choi, I. (2026). AutoSSD: A system for automated
detection of similar speech responses in language tests (Research Memorandum No. RM-26-02). ETS. https://doi.org/10.64634/1g0whg02