Elamur Research⌕
☰
home / papers / 2026 / Speaker or Language? Explaining Variance in Charismatic Prosody Across Luxembourgish and French

Speaker or Language? Explaining Variance in Charismatic Prosody Across Luxembourgish and French

Paper · 2026 · Nina Hosseini-Kivanani, Nafiseh Taghva, Peter Gilles et al.

Summary

The paper asks whether the vocal cues tied to charisma depend more on who is speaking or on which language they use. The setting is bilingual public speech in Luxembourg, where Luxembourgish signals local identity and French carries institutional prestige. The authors took spontaneous speech from ten politicians (five women, five men) from the RTL Archive. Each politician contributed 20 sentences in Luxembourgish and 20 in French, recorded in closely matched public settings, for 400 utterances in total. Transcripts came from LuxASR with manual correction, and boundaries were aligned with WebMAUS and then checked in Praat. ProsodyPro extracted 41 acoustic–prosodic measures. After z-scoring, the authors ran a PCA and fitted mixed-effects models to separate speaker variance from language variance. Speaker identity explained the largest share of variance. Language still had smaller but consistent effects. French showed more shimmer and higher phrase-final F0, which the authors read as a polite, deferential voice. Luxembourgish showed stronger mid-frequency spectral energy, suggesting a more present voice. These results matter because charisma studies usually assume one language, and here personal vocal signatures largely override language-specific modulation.

Key points

  • Data: 10 Luxembourg politicians (including Luc Frieden, Xavier Bettel and Lydie Polfer), each with 20 Luxembourgish and 20 French spontaneous sentences, 400 tokens in total. Only unscripted speech was used, and segments with overlap, noise, laughter or coughing were removed.
  • Pipeline:
    • audio converted to 16 kHz mono;
    • LuxASR transcription, corrected by hand;
    • WebMAUS forced alignment, manually adjusted in Praat against waveform and spectrogram;
    • ProsodyPro features covering F0, intensity, duration and voice quality (H1–H2, H1–A3, Hammarberg index, CPP, jitter, shimmer, HNR, band energies).
  • Analysis: skewed features were log-transformed and all 41 features were z-scored, so model effects read as standard-deviation differences. PCA gave an overview of the feature space, and mixed-effects models split speaker variance from language variance.
  • Main finding: speaker identity accounted for most of the variance, while language accounted for less.
  • Language effects: French had higher shimmer and phrase-final F0, and Luxembourgish had stronger mid-frequency spectral energy. The authors link this to French's prestige role and Luxembourgish's identity role.

Interesting ideas

  • I1. Fit a mixed-effects model with speaker and language as separate sources of variance, so you learn how much a factor matters relative to individual quirks, not just whether it is significant.
  • I2. Use a within-subject design in which the same people produce both conditions, so individual differences cannot pass for a condition effect.
  • I3. Z-score every feature before modeling, so effects on very different scales become comparable standard-deviation shifts.
  • I4. Generate drafts automatically (ASR, forced alignment) and then have a human correct only the visible errors, which makes careful annotation affordable for a low-resource language.
  • I5. State two competing, directional hypotheses up front (prestige alignment vs. identity alignment) so that either outcome is informative.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

  • I1 (learn): Evaluation could model parser or reader errors with author/document and genre as separate components. This would show whether a regression comes from a few odd sources or from a real genre problem.
  • I2 (learn): Negative controls could compare the same authors or the same story across registers (e.g., a fable retold formally and informally). Genre effects would then be measured with the source held constant.
  • I3 (learn): Explainability reports could z-score diverse metrics such as LAS, frame accuracy and answer accuracy before comparing ablations. A one-point change on each scale would then not be treated as equivalent.
  • I4 (learn): Silver trees agreed by witness parsers could be spot-corrected only where they disagree. This yields a gold-like layer for underrepresented constructions at low cost.
  • I5 (frames): When inferring politeness or formality as a speech-act property, the prag crate could test rival hypotheses explicitly. For example, it could check whether hedging marks deference or distance, and record which one the evidence supported.

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

 Institutions

🎓University of LuxembourgLuxembourg · Luxembourg27 works · 337 citations🎓University of Southern DenmarkOdense · Denmark8 works · 111 citations🎓Shiraz UniversityShiraz · Iran2 works · 5 citations

 Fields, topics and concepts

📚 Shelves 4The field of science this work belongs to (OpenAlex fields).Arts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksPsychology · 1,552 worksExperimental and Cognitive Psychology · 870 works

🧭 Topics 2Research subjects and languages the work is about.Speech & spoken language · 480 worksFrench · 70 works

Cite

Cite as Nina Hosseini-Kivanani, Nafiseh Taghva, Peter Gilles and Oliver Niebuhr. 2026. Speaker or Language? Explaining Variance in Charismatic Prosody Across Luxembourgish and French. arXiv:2609.16275.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2609.16275 · arXiv 2609.16275

@misc{hosseinikivanani2026speaker,
  title = {{Speaker or Language? Explaining Variance in Charismatic Prosody Across Luxembourgish and French}},
  author = {Nina Hosseini-Kivanani, Nafiseh Taghva, Peter Gilles and Oliver Niebuhr},
  year = {2026},
  doi = {10.48550/arxiv.2609.16275},
  eprint = {2609.16275},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2609.16275}
}

References

32 references, 0 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. J. Antonakis, N. Bastardoz, P. Jacquart, and B. Shamir, “Charisma: An ill-defined and ill-measured gift,” Annual review of organizational psychology and organizational behavior, vol. 3, no. 1, pp. 293–319, 2016. ↗
  2. C. Adair-Toteff, “Max weber and the sociology of charisma,” in Routledge international handbook of charisma. Routledge, 2020, pp. 7–17. ↗
  3. M. Brecher, “Political leadership and charisma,” Nehru, Ben-Gurion, and Other 20th Century Political Leaders: Intellectual Odyssey I, 2016. ↗
  4. R. Signorello, F. Derrico, I. Poggi, and D. Demolin, “How charisma is perceived from speech: A multidimensional approach,” in 2012 International Conference on Privacy, Security, Risk and Trust and 2012 International Confernece on Social Computing. IEEE, 2012, pp. 435–440. find
  5. A. Rosenberg and J. Hirschberg, “Acoustic/prosodic and lexical correlates of charismatic speech,” https://academiccommons.columbia.edu/, 2005, columbia University Academic Commons. ↗
  6. A. Rosenberg and J. Hirschberg, “Charisma perception from text and speech,” Speech communication, vol. 51, no. 7, pp. 640–655, 2009. ↗
  7. E. Strangert and J. Gustafson, “What makes a good speaker? subject ratings, acoustic measurements and perceptual evaluations.” in Interspeech, vol. 8, 2008, pp. 1688–1691. find
  8. F. Biadsy, A. Rosenberg, R. Carlson, J. Hirschberg, and E. Strangert, “A cross-cultural comparison of american, palestinian, and swedish perception of charismatic speech,” in Speech prosody, vol. 37, 2008, pp. 579–82. find
  9. O. Niebuhr, J. Thumm, and J. Michalsky, “Shapes and timing in charismatic speech-evidence from sounds and melodies,” in 9th International Conference on Speech Prosody 2018. International Speech Communication Association, 2018, pp. 582–586. find
  10. O. Niebuhr and R. Skarnitzl, “Measuring a speaker’s acoustic correlates of pitch-but which? a contrastive analysis for perceived speaker charisma,” in 19th International Congress of Phonetic Sciences: Endangered Languages, and Major Language Varieties. Australasian Speech Science and Technology Association Inc, 2019, pp. 1774–1778. find
  11. O. Niebuhr and R. Skarnitzl, “Those who shout the loudest–do they sound more charismatic?” in Proc. 1st International Conference on Tone and Intonation (TAI), 2021, pp. 1–5. find
  12. O. Niebuhr, R. Skarnitzl, and L. Tylečková, “The acoustic fingerprint of a charismatic voice-initial evidence from correlations between long-term spectral features and listener ratings,” in 9th International Conference on Speech Prosody 2018. International Speech Communication Association, 2018, pp. 359–363. find
  13. O. Niebuhr and K. Fischer, “Do not hesitate!—unless you do it shortly or nasally: How the phonetics of filled pauses determine their subjective frequency and perceived speaker performance,” in Proc. Interspeech 2019, 2019, pp. 544–548. find
  14. I. Mennen, “Bi-directional interference in the intonation of dutch speakers of greek,” Journal of phonetics, vol. 32, no. 4, pp. 543– 563, 2004. ↗
  15. E. De Leeuw, “When your native language sounds foreign: A phonetic investigation into first language attrition,” Ph.D. dissertation, Queen Margaret University, 2008. find
  16. E. Passoni, E. de Leeuw, and E. Levon, “Bilinguals produce pitch range differently in their two languages to convey social meaning,” Language and speech, vol. 65, no. 4, pp. 1071–1095, 2022. ↗
  17. C. M. Esposito and S. u. D. Khan, “The cross-linguistic patterns of phonation types,” Language and Linguistics Compass, vol. 14, no. 12, p. e12392, 2020. ↗
  18. M. A. Sicoli, “Shifting voices with participant roles: Voice qualities and speech registers in mesoamerica,” Language in Society, vol. 39, no. 4, pp. 521–553, 2010. ↗
  19. P. Gilles and C. Moulin, “Luxembourgish,” in Germanic standardizations: Past to present. John Benjamins Publishing Company, 2006, pp. 303–329. find
  20. K. Horner and J. J. Weber, “The language situation in luxembourg,” in Language planning in Europe. Routledge, 2016, pp. 180–240. find
  21. K. Horner and J. J. Weber, Introducing multilingualism: A social approach. Routledge, 2017. find
  22. F. Nolan, “Speaker identification and forensic phonetics,” Handbook of phonetic sciences, pp. 519–533, 1999. find
  23. S. Berger, ““like, comment, subscribe”: Perception of acoustic-prosodic features of content creators’ charismatic speech on youtube,” Ph.D. dissertation, 2024. ↗
  24. P. Gilles, L. E. A. Hillah, and N. Hosseini-Kivanani, “Asrlux: Automatic speech recognition for the low-resource language luxembourgish,” in 20. International Conference of Phonetic Sciences (ICPhS). Guarant International, Prague, Unknown/unspecified, 2023. find
  25. P. Gilles, N. Hosseini-Kivanani, and L. E. A. Hillah, “Lux-asr: Building an asr system for the luxembourgish language,” in 2022 IEEE Spoken Language Technology Workshop (SLT) SLT 2022, 2023. find
  26. T. Kisler, U. Reichel, and F. Schiel, “Multilingual processing of speech via web services,” Computer Speech & Language, vol. 45, pp. 326–347, 2017. ↗
  27. P. Pollák, J. Volı́n, and R. Skarnitzl, “Phone segmentation tool with integrated pronunciation lexicon and czech phonetically labelled reference database.” in LREC, 2008. find
  28. Y. Xu, “Prosodypro—a tool for large-scale systematic prosody analysis.” Laboratoire Parole et Langage, France, 2013. find
  29. E. Delais-Roussarie, B. Post, M. Avanzi, C. Buthke, A. Di Cristo, I. Feldhausen, S.-A. Jun, P. Martin, T. Meisenburg, A. Rialland et al., “Intonational phonology of french: Developing a tobi system for french,” Intonation in romance, pp. 63–100, 2015. ↗
  30. L. Brown and P. Prieto, “(im) politeness: Prosody and gesture,” in The Palgrave handbook of linguistic (im) politeness. Springer, 2017, pp. 357–379. ↗
  31. K. Idemaru, B. Winter, L. Brown, and G. E. Oh, “Loudness trumps pitch in politeness judgments: Evidence from korean deferential speech,” Language and Speech, vol. 63, no. 1, pp. 123– 148, 2020. ↗
  32. C. Gaffney, The effects of intelligence, personality traits, L1 fluency, and L2 proficiency on L2 spoken fluency. University of Toronto (Canada), 2021. find

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/hosseini-kivanani-2026-speaker-language-explaining-variance-charismatic/