Elamur Research⌕
☰
home / papers / 2026 / Tools for Estimating the Perceived Level of Phonetic Reduction

Tools for Estimating the Perceived Level of Phonetic Reduction

Summary

Phonetic reduction, meaning less careful articulation, is frequent in casual conversation and helps signal pragmatic functions such as holding the turn or positive assessment. Until now there has been no practical tool for measuring it automatically at the word level. Ward and colleagues define the task as estimating, from audio alone, how reduced speech sounds to listeners, either per frame or per word. They train on previously released dialog recordings of bilingual El Paso college students speaking American English and Northern Mexican Spanish, labeled by one annotator on a four-point scale. They also add new annotation sets to check inter-annotator agreement and test other corpora, including Switchboard. The model is deliberately simple: 12th-layer HuBERT Base embeddings feed a linear regression, evaluated leave-one-dialog-out. For English, word-level estimates correlate up to 0.55 with averaged human ratings, well above an earlier prosody-based model (0.243). The authors judge this good enough for statistical corpus studies, which they demonstrate with turn-yielding and prominence-marking analyses. The tools and annotations are openly released and could also support reduction-aware loss functions for speech synthesis.

Key points

  • Training data: 31 stereo minutes of English (3051 regions, 27% reduced or highly reduced) and 25 minutes of Spanish (1614 regions, 31%). Each region was rated from 0 (enunciated) to 3 (highly reduced).
  • Three new evaluation sets were added:
    • 10 minutes of English triple-annotated by the authors to build a gold standard.
    • 4 minutes of Switchboard telephone speech.
    • 9 minutes of an adolescent speaker from a task-oriented dialog corpus.
  • Features are HuBERT Base layer-12 vectors (768 dimensions per 20 ms frame). Linear regression was chosen to limit overfitting on the small dataset.
  • Word-level English predictions reach r = 0.55 with average human judgments, compared with 0.243 for the earlier pitch, spectral tilt and harmonicity model (Ward and Ortega, 2024).
  • Annotations and open-source tools are available at www.cs.utep.edu/nigel/reduction/.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.63317/3qachmx9chqu

@inproceedings{ward2026tools,
  title = {{Tools for Estimating the Perceived Level of Phonetic Reduction}},
  author = {Ward, Nigel and Vazquez, Javier and Boushka, Emma (Danny) R. and Niebuhr, Oliver},
  booktitle = {Proceedings of the Fifteenth Language Resources and Evaluation Conference},
  year = {2026},
  doi = {10.63317/3qachmx9chqu},
  url = {https://doi.org/10.63317/3qachmx9chqu}
}

References

55 references, 1 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Bistra Andreeva, William J Barry, and Jacques C Koreman. 2014. A cross-language corpus for studying the phonetics and phonology of prominence. In LREC, pages 326–330. find
  2. Tomás Arias-Vergara, Elizabeth Londoño-Mora, Paula A Pérez-Toro, Maria Schuster, Elmar Nöth, Juan Rafael Orozco-Arroyave, and Andreas Maier. 2023. Measuring phonological precision in children with cleft lip and palate. In Interspeech, pages 4638–4642. find
  3. Judith Bauer, Frank Zalkow, Meinard Müller, and Christian Dittmar. 2025. Explicit emphasis control in text-to-speech synthesis. In Speech Synthesis Workshop (SSW), pages 21–27. find
  4. Grégory Beller, Nicolas Obin, and Xavier Rodet. 2008. Articulation degree as a prosodic dimension of expressive speech. In Speech Prosody, pages 681–684. find
  5. Jason Bishop and Boram Kim. 2018. Anticipatory shortening: Articulation rate, phrase length, and lookahead in speech production. In 9th International Conference on Speech Prosody. find
  6. Mara Breen, Evelina Fedorenko, Michael Wagner, and Edward Gibson. 2010. Acoustic correlates of information structure. Language and cognitive processes, 25(7-9):1044–1098. find
  7. Rachel Steindel Burdin and Cynthia G Clopper. 2015. Phonetic reduction, vowel duration, and prosodic structure. In International Congress of the Phonetic Sciences. find
  8. Kaleigh Susan Cammenga. 2018. The Effect of Vocal Fry on Speech Intelligibility. Ph.D. thesis, Michigan State University. find
  9. Francesco Cangemi and Oliver Niebuhr. 2018. Rethinking reduction and canonical forms. In Francesco Cangemi, Meghan Clayards, Oliver Niebuhr, Barbara Schuppler, and Margaret Zellers, editors, Rethinking Reduction, pages 277–302. de Gruyter. find
  10. Lei Chen, Chenglin Jiang, Yiwei Gu, Yang Liu, and Jiahong Yuan. 2022. Automatically detecting reduced-formed English pronunciations by using deep learning. In Proceedings of the 17th Workshop on Innovative Use of NLP for Building Educational Applications (BEA), pages 22–26. find
  11. Bronya R Chernyak, Talia Ben Simon, Yael Segal, Jeremy Steffman, Eleanor Chodroff, Jennifer S Cole, and Joseph Keshet. 2022. DeepFry: Identifying vocal fry using deep neural networks. In Interspeech, pages 3578–3582. find
  12. Jennifer Cole, José I Hualde, Caroline L Smith, Christopher Eager, Timothy Mahrt, and Ricardo Napoleão de Souza. 2019. Sound, structure and meaning: The bases of prominence ratings in English, French and Spanish. Journal of Phonetics, 75:113–147. find
  13. Maureen de Seyssel, Antony D’Avirro, Adina Williams, and Emmanuel Dupoux. 2024. EmphAssess : a prosodic benchmark on assessing emphasis transfer in speech-to-speech models. In Empirical Methods in Natural Language Processing (EMNLP), pages 495–507. find
  14. Patrick Cormac English, John D Kelleher, and Julie Carson-Berndsen. 2023. Discovering phonetic feature event patterns in transformer embeddings. In Interspeech. find
  15. Mirjam Ernestus and Natasha Warner. 2011. An introduction to reduced pronunciation variants. Journal of Phonetics, 39(SI):253–260. find
  16. John J Godfrey, Edward C Holliman, and Jane Mc-Daniel. 1992. Switchboard: Telephone speech corpus for research and development. In Proceedings of ICASSP, pages 517–520. find
  17. Joakim Gustafson, Eva Székely, Simon Alexandersson, and Jonas Beskow. 2023a. Casual chatter or speaking up? Adjusting articulatory effort in generation of speech and animation for conversational characters. In IEEE 17th International Conference on Automatic Face and Gesture Recognition. find
  18. Joakim Gustafson, Éva Székely, and Jonas Beskow. 2023b. Generation of speech and facial animation with controllable articulatory effort for amusing conversational characters. In Proceedings of the 23rd ACM International Conference on Intelligent Virtual Agents, pages 1–9. find
  19. Wei-Ning Hsu, Benjamin Bolte, Yao-Hung Hubert Tsai, Kushal Lakhotia, Ruslan Salakhutdinov, and Abdelrahman Mohamed. 2021. HuBert: Self-supervised speech representation learning by masked prediction of hidden units. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 29:3451–3460. find our card
  20. Wen-Chin Huang, Benjamin Peloquin, Justine Kao, Changhan Wang, Hongyu Gong, Elizabeth Salesky, Yossi Adi, Ann Lee, and Peng-Jen Chen. 2023. A Holistic Cascade System, Benchmark, and Human Evaluation Protocol for Expressive Speech-to-Speech Translation. In ICASSP. find
  21. T Florian Jaeger and Esteban Buz. 2017. Signal reduction and linguistic encoding. The Handbook of Psycholinguistics, pages 38–81. find
  22. Yishan Jiao, Visar Berisha, Julie Liss, Sih-Chiao Hsu, Erika Levy, and Megan McAuliffe. 2016. Articulation entropy: An unsupervised measure of articulatory precision. IEEE Signal Processing Letters, 24(4):485–489. find
  23. Daniel Jurafsky, Alan Bell, Eric Fosler-Lussier, Cynthia Girand, and William D Raymond. 1998. Reduction of English function words in Switchboard. In International Conference on Spoken Language Processing. find
  24. Greg Kochanski, Esther Grabe, John Coleman, and Burton Rosner. 2005. Loudness predicts prominence: Fundamental frequency lends little. Journal of the Acoustical Society of America, 118(2):1038–1054. find
  25. Frank Kügler, Stefan Baumann, Bistra Andreeva, Bettina Braun, Martine Grice, Jana Neitsch, Oliver Niebuhr, Jörg Peters, Christine T. Röhr, Antje Schweitzer, and Petra Wagner. 2019. Annotation of German intonation: DIMA compared with other annotation systems. In ICPhS. find
  26. D Robert Ladd and Amalia Arvaniti. 2023. Prosodic prominence across languages. In Annual Review of Linguistics, volume 9, pages 171–193. Annual Reviews. find
  27. Harm Lameris, Joakim Gustafsson, and Éva Székely. 2023. Beyond style: Synthesizing speech with pragmatic functions. In Interspeech, pages 3382–3386. find
  28. Sungbok Lee, Erik Bresch, Jason Adams, Abe Kazemzadeh, and Shrikanth Narayanan. 2006. A study of emotional speech articulation using a fast magnetic resonance imaging technique. In Ninth International Conference on Spoken Language Processing. find
  29. Heike Lehnert-LeHouillier, Susana Terrazas, and Steven Sandoval. 2020. Prosodic entrainment in conversations of verbal children and teens on the autism spectrum. Frontiers in Psychology, 11:2718. find
  30. Guan-Ting Lin, Chi-Luen Feng, Wei-Ping Huang, Yuan Tseng, Tzu-Han Lin, Chen-An Li, Hung-yi Lee, and Nigel G. Ward. 2022. On the utility of self-supervised models for prosody-related tasks. In IEEE Workshop on Spoken Language Technology (SLT), pages 1104–1111. find
  31. John Local and Gareth Walker. 2012. How phonetic features project more talk. Journal of the International Phonetic Association, 42:255–280. find
  32. Nichola Lubold, Stephanie A Borrie, Tyson S Barrett, Megan M Willi, and Visar Berisha. 2019. Do conversational partners entrain on articulatory precision? In Interspeech, pages 1931–1935. find
  33. Matthew Marge, Carol Espy-Wilson, Nigel G. Ward, et al. 2022. Spoken language interaction with robots: Research issues and recommendations. Computer Speech and Language, 71. 101225. find
  34. Devang S Ram Mohan, Vivian Hu, Tian Huey Teh, Alexandra Torresquintero, Christopher G R Wallis, Marlene Staib, Lorenzo Foglianti, Jiameng Gao, and Simon King. 2021. Ctrl-p: Temporal control of prosodic variation for speech synthesis. In Interspeech, pages 3875–3879. find
  35. Jaya Narain, Vasudha Kowtha, Colin Lea, Lauren Tooley, Dianna Yee, Vikramjit Mitra, Zifang Huang, Miquel Espi Marques, Jon Huang, Carlos Avendano, et al. 2025. Voice quality dimensions as interpretable primitives for speaking style for atypical speech and affect. In Interspeech. find
  36. Oliver Niebuhr. 2014. ’A little more ironic’: Voice quality and segmental reduction differences between sarcastic and neutral utterances. In 7th Speech Prosody Conference, pages 608–612. find
  37. Oliver Niebuhr. 2016. Rich reduction: Sound-segment residuals and the encoding of communicative functions along the hypo-hyper scale. In 7th Tutorial & Research Workshop on Experimental Linguistics, ExLing, pages 11–24. International Speech Communication Association (ISCA). find
  38. Oliver Niebuhr. 2017. Clear speech — mere speech? How segmental and prosodic speech reduction shape the impression that speakers create on listeners. In Interspeech, volume 18, pages 894–898. find
  39. Oliver Niebuhr, Karin Gors, and Evelin Graupe. 2013. Speech reduction, intensity, and F0 shape are cues to turn-taking. In SigDial, pages 261– 269. find
  40. Benjamin Picart, Thomas Drugman, and Thierry Dutoit. 2014. Analysis and HMM-based synthesis of hypo and hyperarticulated speech. Computer Speech & Language, 28(2):687–707. find
  41. Leendert Plug. 2005. From words to actions: The phonetics of eigenlijk in two communicative contexts. Phonetica, 62(2-4):131–145. find
  42. Leendert Plug. 2011. Phonetic reduction and informational redundancy in self-initiated self-repair in Dutch. Journal of Phonetics, 39(3):289–297. find
  43. Sebastião Quintas, Mathieu Balaguer, Julie Mauclair, Virginie Woisard, and Julien Pinquier. 2023. Can we use speaker embeddings on spontaneous speech obtained from medical conversations to predict intelligibility? In Automatic Speech Recognition and Understanding Workshop (ASRU), pages 1–7. IEEE. find
  44. Simon Roessig. 2024. The inverse relation of pre-nuclear and nuclear prominences in german. Laboratory Phonology, 15(1):1–43. find
  45. Ming Tu, Anna Grabek, Julie Liss, and Visar Berisha. 2018. Investigating the role of L1 in automatic pronunciation evaluation of L2 speech. In Interspeech, pages 1636–1640. find
  46. Petra Wagner, Jonas Beskow, Simon Betz, Jens Edlund, Joakim Gustafson, Gustav Eje Henter, Sébastien Le Maguer, Zofia Malisz, Éva Székely, Christina Tånnander, et al. 2019. Speech synthesis evaluation: State-of-the-art assessment and suggestion for a novel research program. In Proceedings of the 10th… find
  47. Nigel G Ward. 2019. Prosodic Patterns in English Conversation. Cambridge University Press. find
  48. Nigel G Ward, Raul O Gomez, Carlos A Ortega, and Georgina Bugarini. 2025a. Phonetic reduction is associated with positive assessment and other pragmatic functions. Speech Communication, 175. 103305. find
  49. Nigel G Ward, Divette Marco, and Olac Fuentes. 2025b. Which prosodic features matter most for pragmatics? IEEE ICASSP. find
  50. Nigel G Ward and Carlos A Ortega. 2024. Preliminaries to a study of the pragmatic functions of reduced articulatory precision in dialog. Technical Report UTEP-CS-24-23, University of Texas at El Paso, Department of Computer Science. find
  51. Nigel G. Ward, Steven D. Werner, David G. Novick, Tatsuya Kawahara, Elizabeth E. Shriberg, Louis-Philipp Morency, and Catharine Oertel. 2013. The similar segments in social speech task. In MediaEval Workshop. find
  52. Duane G Watson. 2010. The many roads to prominence: Understanding emphasis in conversation. In B. Ross, editor, Psychology of Learning and Motivation, pages 163–183. Elsevier. find
  53. Shu-Wen Yang, Heng-Jui Chang, Zili Huang, Andy T Liu, Cheng-I Lai, Haibin Wu, Jiatong Shi, Xuankai Chang, Hsiang-Sheng Tsai, Wen-Chin Huang, et al. 2024. A large-scale evaluation of speech foundation models. IEEE/ACM Transactions on Audio, Speech, and Language Processing, 32:2884–2899. find
  54. Margaret Zellers, Barbara Schuppler, and Meghan Clayards. 2018. Introduction, or: Why rethink reduction? In Francesco Cangemi, Meghan Clayards, Oliver Niebuhr, Barbara Schuppler, and Margaret Zellers, editors, Rethinking Reduction, pages 1–24. de Gruyter. 12. find
  55. Language Resource References John J. Godfrey and Edward C. Holliman and Jane McDaniel. 1992. Switchboard: Telephone speech corpus for research and development. Linguistic Data Consortium, Release 2, ISLRN 988-076-156-109-5. find

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/ward-2026-tools-estimating-perceived-level-phonetic/