AI-ASSISTED ACOUSTIC ANALYSIS OF LEGAL DISCOURSE PROSODY: OPPORTUNITIES AND LIMITATIONS

Keywords: legal discourse, prosody, AI-assisted acoustic analysis, artificial intelligence, computational paralinguistics, forensic phonetics, speech processing, legal linguistics

Abstract

The article examines the methodological potential of artificial intelligence (AI)-assisted acoustic analysis for investigating prosody in legal discourse. Prosody is viewed as an essential component of spoken legal communication, shaping the expression of epistemic stance, interpersonal relations, discourse organisation, and persuasive strategies through variations in pitch, intensity, rhythm, pausing, speech rate, and voice quality. The study addresses the limited attention paid to the acoustic dimension of legal discourse and evaluates the opportunities and challenges of integrating AI technologies into prosodic research. The object of the study is the prosodic organisation of spoken legal discourse; the subject is the methodological potential and limitations of AI-assisted acoustic analysis of legal speech. The aim of the study is to provide a criteria-based evaluation of AI-assisted acoustic analysis as a methodological framework for research on legal prosody. The research is theoretical and methodological, drawing on studies in acoustic phonetics, computational paralinguistics, forensic phonetics, legal linguistics, and speech processing. Descriptive, comparative, and criticalevaluative methods are applied to assess AI-assisted acoustic analysis according to six methodological criteria: validity, reliability, scalability, interpretability, ecological validity, and legal and ethical admissibility. The findings demonstrate that AI significantly improves the efficiency, reproducibility, and scalability of acoustic analysis by automating speech preprocessing, feature extraction, and computational modelling, enabling large-scale investigation of authentic legal speech. At the same time, important limitations remain, including reduced ecological validity, limited model transparency, algorithmic bias, risks of overinterpreting acoustic evidence, and legal and ethical concerns. The study argues that AI should complement rather than replace expert phonetic interpretation. It concludes that a human-in-the-loop approach integrating computational technologies with linguistic expertise provides the most reliable framework for future research on legal prosody while ensuring methodological transparency, scientific validity, and legal accountability.

References

1. Рудік І. В., Вихор О. Ф. Просодичні маркери соціального статусу мовця. Записки з романо-германської філології. 2020. Вип. 1 (44). С. 273–279. DOI: https://doi.org/10.18524/2307-4604.2020.1(44).211020.
2. Akkurt V., Korolova T., Popova O. Persuasion Prosody in Prosecutor’s Speech: Ukrainian and English. Revista Amazonia Investiga. 2021. Vol. 10, no. 47. P. 125–141. DOI: https://doi.org/10.34069/ai/2021.47.11.13.
3. Boersma P., Weenink D. Praat: Doing Phonetics by Computer [Computer program]. Version 7.0.02. 2026. URL: https://www.praat.org (дата звернення: 05.09.2026).
4. Eyben F., Scherer K. R., Schuller B. W. et al. The Geneva Minimalistic Acoustic Parameter Set (GeMAPS) for Voice Research and Affective Computing. IEEE Transactions on Affective Computing. 2016. Vol. 7, no. 2. P. 190–202. DOI: https://doi.org/10.1109/TAFFC.2015.2457417.
5. Heffer C. The Language of Jury Trial: A Corpus-Aided Analysis of Legal-Lay Discourse. London : Palgrave Macmillan, 2005. 253 p. DOI: https://doi.org/10.1057/9780230502888.
6. Koenecke A., Nam A., Lake E. et al. Racial Disparities in Automated Speech Recognition. Proceedings of the National Academy of Sciences. 2020. Vol. 117, no. 14. P. 7684–7689. DOI: https://doi.org/10.1073/pnas.1915768117.
7. Korolova T., Zhmayeva N. Prosody of Negative Modality in Translation. Науковий вісник Південноукраїнського національного педагогічного університету імені К. Д. Ушинського: Лінгвістичні науки. 2021. № 32. С. 74–83. DOI: https://doi.org/10.24195/2616-5317-2021-32-6.
8. McAuliffe M., Socolof M., Mihuc S. et al. Montreal Forced Aligner: Trainable Text-Speech Alignment Using Kaldi. Proceedings of Interspeech 2017. Stockholm, 2017. P. 498–502. DOI: https://doi.org/10.21437/Interspeech.2017-1386.
9. Morrison G. S. Forensic Voice Comparison and the Paradigm Shift. Science & Justice. 2009. Vol. 49, no. 4. P. 298–308. DOI: https://doi.org/10.1016/j.scijus.2009.09.002.
10. Schuller B. W. Speech Emotion Recognition: Two Decades in a Nutshell, Benchmarks, and Ongoing Trends. Communications of the ACM. 2018. Vol. 61, no. 5. P. 90–99. DOI: https://doi.org/10.1145/3129340.
Published
2026-11-18
How to Cite
Rudik, I. V., & Ter-Grygorian, M. H. (2026). AI-ASSISTED ACOUSTIC ANALYSIS OF LEGAL DISCOURSE PROSODY: OPPORTUNITIES AND LIMITATIONS. New Philology, (103), 162-169. https://doi.org/10.26661/2414-1135-2026-103-20
Section
Articles