--- id: zhao26b_interspeech title: F0 realization of prosodic focus across adulthood in Jianghuai Mandarin authors: - Xinxian Zhao - Xiaohu Yang year: 2026 doi: 10.21437/Interspeech.2026-255 isca_url: https://www.isca-archive.org/interspeech_2026/zhao26b_interspeech.html pdf_url: https://www.isca-archive.org/interspeech_2026/zhao26b_interspeech.pdf session: Prominence, Stress and Focus topics: - prosody - phonetics - evaluation category: phonetics-linguistics institutions: - Shanghai Jiao Tong University - National Research Center for Language and Well-being - Tongji University funding: - China Postdoctoral Science Foundation code: url: "" license: "" open_to_collaboration: false wiki_frontmatter: id: zhao26b_interspeech category: phonetics-linguistics institutions: - Shanghai Jiao Tong University - National Research Center for Language and Well-being - Tongji University updated: 2026-09-29 confidence: full-paper digest: v2 source: https://doi.org/10.21437/Interspeech.2026-255 pdf: https://www.isca-archive.org/interspeech_2026/zhao26b_interspeech.pdf wiki_url: https://interspeech-2026-wiki.vercel.app/papers/zhao26b_interspeech/ markdown_url: https://interspeech-2026-wiki.vercel.app/papers/zhao26b_interspeech/markdown.md --- # F0 realization of prosodic focus across adulthood in Jianghuai Mandarin *Xinxian Zhao, Xiaohu Yang* [PDF](https://www.isca-archive.org/interspeech_2026/zhao26b_interspeech.pdf) · [ISCA page](https://www.isca-archive.org/interspeech_2026/zhao26b_interspeech.html) · [DOI](https://doi.org/10.21437/Interspeech.2026-255) **Category:** `phonetics-linguistics` **TL;DR** — This study investigates age-related variation in the F0 realization of prosodic focus across adulthood in Jianghuai Mandarin, demonstrating that the global tri-zone F0 pattern is preserved across age groups while fine-grained post-focus compression (PFC) is significantly enhanced in older speakers. ## Key contributions - Establishes Jianghuai Mandarin as a canonical '+PFC' language exhibiting the tri-zone F0 pattern (on-focus raising, post-focus compression, and pre-focus stability) across adulthood. - Reveals that global F0 patterning and F0 excursion are robustly preserved across young, middle-aged, and older adult speakers. - Uncovers age-sensitive fine-grained F0 adjustments, showing that older speakers exhibit significantly greater post-focus mean F0 compression than younger and middle-aged adults. - Provides acoustic datasets and linear mixed-effects (LME) modeling analyses tracking lifespan trajectories of prosodic focus production using 1,200 controlled utterances. ## Problem While fundamental frequency (F0) modulation is known to signal prosodic focus, prior research has predominantly focused on young adults, leaving the impact of adult vocal aging on focus realization poorly understood. Furthermore, languages and regional dialects vary widely in whether they exhibit post-focus compression (PFC), and Jianghuai Mandarin—spoken by over 70 million people—has lacked empirical investigation regarding its focus-related F0 typology. Understanding whether age-related physiological degeneration alters or preserves speakers' command over prosodic structures is critical for both theoretical linguistics and speech technology applications. ## Method The study utilized a production task with a Subject-Adverbial-Verb-Object (SAVO) sentence structure across five answer frames representing the five lexical tones of Jianghuai Mandarin. Four elicitation question types elicited neutral focus (NF) and three narrow focus conditions: initial focus (IF, subject), medial focus (MF, adverbial), and final focus (FF, object). Speech signals were recorded at 22.05 kHz using xRecorder, annotated in Praat via ProsodyPro by extracting nine equally spaced points per syllable, and converted from Hz to semitones using sex-specific references (Fr = 64 Hz for females, 55 Hz for males). Dependent variables included mean F0 and F0 excursion (maximum minus minimum F0) for key constituents, alongside narrow-minus-neutral focus difference scores to capture fine-grained F0 adjustment. Statistical evaluations employed linear mixed-effect (LME) models in R with Group, Constituent, and Focus (or Focus Contrast) as fixed effects, and Participant as a random intercept, utilizing Bonferroni-corrected pairwise comparisons to test specific hypotheses about aging and regional intonation patterns. ## Experimental setup Sixty healthy native speakers of Jianghuai Mandarin divided evenly into three gender-matched age groups: young (20–30 years), middle-aged (40–50 years), and older (60–70 years), with 20 participants (10 females) per group. A total of 1,200 recordings (60 participants × 20 prompt questions: 4 question types × 5 lexical tone frames) were analyzed. The primary metrics were mean F0, F0 excursion, and narrow-minus-neutral difference scores evaluated via LME models. ## Results LME models confirmed significant main effects of Constituent and Focus, showing that Jianghuai Mandarin universally employs the canonical tri-zone F0 pattern: on-focus mean F0 increased significantly (e.g., Subject in IF: higher by p < .001), post-focus mean F0 decreased significantly, and pre-focus regions remained unchanged. F0 excursion mirrored this, showing larger excursions in on-focus and smaller excursions in post-focus regions. For fine-grained F0 adjustment, older adults demonstrated significantly more negative mean F0 differences (enhanced PFC) in post-focus regions compared to young and middle-aged groups, specifically in the Adverbial and Object during IF vs. NF contrasts (p < .001 and p < .01) and in the Object during MF vs. NF contrasts (p < .001). Conversely, F0 excursion differences showed no significant group main effects, indicating that age-related modulation is localized specifically to mean F0 post-focus compression rather than global excursion metrics. ## Limitations The study is limited by its moderate sample size per age-gender cohort and does not treat gender as a primary between-subject variable due to participant pool constraints. The analysis relies solely on acoustic F0 measurements without incorporating duration, intensity cues, or subjective speech intelligibility evaluations. Additionally, the scope is restricted to read speech elicited via prompt questions in a controlled laboratory environment rather than natural conversational dialogue. ## Why read this Speech scientists and phoneticians should read this paper to understand how structural intonational invariants interact with physiological aging in tone languages. It offers critical empirical baseline data for modeling prosodic control, speaker normalization, and expressive text-to-speech synthesis across diverse adult age groups. ## Code None released (as of this page's `updated` date). If you are an author with a repo, please claim this entry — see CONTRIBUTING.md. ## Applications Development of age-robust text-to-speech (TTS) systems, clinical speech assessment tools for evaluating vocal aging and prosodic control, and automated speaker state monitoring. ## Institutions / 機構 Shanghai Jiao Tong University, National Research Center for Language and Well-being, Tongji University **Funding / 經費:** China Postdoctoral Science Foundation ## Related - [Prosodic Realization of Focus in Yi-Mandarin Bilingual Speakers: On-Focus Expansion without Post-Focus Compression](zhang26g_interspeech.md) — same problem · relatedness 2.0/3 - [Age-Related Changes in Mandarin Lexical Tone Production: Acoustic Properties and Tonal Distinctiveness](mao26b_interspeech.md) — same problem · relatedness 2.0/3 - [Age-related Differences in Acoustic Realization of Aspirated Fricatives in Shaxi Bai](zhang26q_interspeech.md) — same problem · relatedness 1.9/3 - [Towards an understanding of prosodic cue weighting for turn-end classification in older adults with varying hearing abilities](curetti26_interspeech.md) — relatedness 1.8/3 - [Acoustic Differences Between Citation and Sandhi Tones Across Three Generations in Xiamen Southern Min](xie26d_interspeech.md) — same problem · relatedness 1.8/3 All 950k paper pairs scored by TypeSafe Jev (`scripts/related/`); relatedness 0 = unrelated … 3 = directly comparable.