Sociophonetic Variation in Urban Mandarin: A Field Study of Tone Realization

Thesis Abstract

This study examines sociophonetic variation in tone realization among speakers of Urban Mandarin, addressing how social factors and regional mobility shape tonal patterns in contemporary speech. The central problem is understanding how urban multilingual and fluid speaker populations influence tone realization, auditably affecting phonetic targets and phonological categories in real-world communication. The aim is to characterize variation in four tone realizations (tones 1–4) across speaker groups differentiated by age, gender, educational background, migration status, and listening proficiency, and to evaluate how social indices mediate tonal timing, contour shape, and pitch target values. Specific objectives are (1) to quantify cross-speaker variation in fundamental frequency (F0) trajectories for each tone using high-resolution pitch tracks; (2) to assess the effect of social factors on tonal phonetic realization via mixed-effects modeling; (3) to test whether individual speaker mobility and exposure to other Mandarin varieties predict systematic deviations in tone realization; (4) to contrast urban Mandarin tone realization with a rural baseline to ascertain urban-specific sociophonetic patterns; and (5) to integrate findings within a theoretical framework that connects sociolinguistic variation to tone perception and production. The study adopts a cross-sectional field design combining quantitative acoustic analysis with targeted sociolinguistic interviews. Population comprises native Urban Mandarin speakers aged 18–55 residing in Shanghai for at least five years, with a rural-urban comparison sample drawn from adjacent suburban communities. A stratified random sample of 120 participants is selected to ensure balanced representation across age cohorts (18–29, 30–39, 40–55), gender, educational attainment, and migration status. Speech data consist of 20-minute conversational interviews and a 10-minute reading task designed to elicit natural tone production, yielding approximately 4 hours of transcribed speech per participant for a robust acoustic corpus. Instrumentation includes a high-sampling-rate audio recorder and a head-worn microphone, with F0 extracted via Praat and custom preprocessing scripts to remove voicing errors and normalize for speaker-specific factors. Sociolinguistic interviews capture variables such as language background, exposure to other Mandarin varieties, social networks, urban usage contexts, attitudes toward prestige varieties, and self-rated listening proficiency. Data analysis proceeds in three strands. First, acoustic analysis computes F0 mean, standard deviation, pitch range, and contour metrics (rise/fall slopes, tonal alignment to target contours) for each syllable carrying tones 1–4, with error-checking by manual verification on a subset. Second, a series of linear mixed-effects models (LMMs) assesses the influence of fixed effects—age, gender, education, migration status, exposure to other Mandarin varieties, and listening proficiency—on tone realization, while random effects account for speaker and word-level variability. Third, a discriminant analysis and Bayesian hierarchical modeling test the robustness of tone realization as a function of social predictors, and a subset of data undergoes perceptual validation through listening tests to examine how production variation maps onto listener judgments. The study also employs a conceptual model integrating Labovian sociolinguistic theory with Tone and Voice Perception frameworks (e.g., JASA-inspired perception-production links) to interpret findings. Expected findings anticipate that urban Mandarin tone realization exhibits systematic sociolinguistic variation higher-pitched targets and steeper contours for tone 3 among younger, mobile, and highly educated speakers; reduced tonal contrast for tones 2 and 4 in populations with extensive exposure to other Mandarin varieties; and context-dependent modulation of F0 and contour shapes in spontaneous speech compared to reading tasks. Rural-urban comparisons are expected to reveal convergence toward urban patterns in the urban cohort, with residual rural influence among older speakers or individuals with limited mobility. The study contributes to knowledge by detailing how urban sociolinguistic factors shape tonal phonetics in a high-variability environment, advancing theoretical integration of sociolinguistics and phonetics, and offering methodological rigor through mixed-effects modeling and cross-context validation. Practical implications include informing speech recognition systems for Mandarin in urban settings and contributing to language policy discussions on dialect contact and maintenance. The main conclusion posits that tone realization in Urban Mandarin is not monolithic but is dynamically modulated by social identity and exposure, with measurable effects on acoustic targets and perceptual salience. Recommendations include expanding longitudinal monitoring of tone change trajectories in urban centers, integrating sociophonetic measures into Mandarin speech technologies, and conducting cross-city replication to assess regional differences in sociophonetic patterns.