Cross-Linguistic Prosody: A Comparative Analysis of Intonational Patterns in English and Mandarin

Thesis Abstract

This study investigates cross-linguistic prosody by examining how English and Mandarin articulate intonational patterns across discourse genres, addressing the problem of limited cross-language empirical data informing theories of universal prosody versus language-specific realization. The aim is to compare segmental and phrase-level intonation, focusing on melodic contour, tonal alignment, and durational cues, to identify convergences and divergences that shape meaning, emphasis, and discourse structure in the two languages. Specific objectives include (1) delineating canonical intonational contours in English and Mandarin within narrative, interrogative, and expository genres; (2) analyzing pitch-target realization, alignment with syntactic boundaries, and duration patterns using labeled data; (3) evaluating the role of discourse-pragmatic factors (topic focus, given/new information) in modulating prosodic realizations; (4) assessing whether shared articulatory constraints or language-specific phonological inventories account for observed differences; and (5) testing the applicability of theoretical models such as the Autosegmental-Metrical (AM) framework and the ToBI annotation scheme across both languages. A mixed-methods design combines corpus-based acoustic analysis with perceptual evaluation. The population comprises native English speakers (N=40) and native Mandarin speakers (N=40) aged 20–35, balanced for gender, with similar education levels and no reported speech or hearing impairments. Data collection utilizes two primary instruments (a) a controlled elicitation corpus consisting of 60 short dialogues per language (20 narrative, 20 questions, 20 expository passages) produced by participants, and (b) a spontaneous speech corpus of 30 minutes per language drawn from contemporary media transcripts annotated for discourse structure. Acoustic measurements are extracted with Praat, including f0 mean, f0 range, contour slope, peak alignment with punctuation, syllable duration, and boundary lengthening. Perceptual data are gathered through a 5-point Likert scale evaluation by a separate panel of 12 trained listeners assessing naturalness and perceived emphasis for sampled utterances. Data analysis employs a hierarchical linear modeling approach to examine cross-language effects on prosodic variables while controlling for genre and discourse focus. ANOVA is used to identify significant differences in contour types and tonal realizations across languages and genres. Regression analyses investigate the predictive power of discourse focus on pitch alignment and duration cues, with separate models for English and Mandarin to elucidate language-specific mechanisms. The AM phonological framework informs the coding of intonational patterns, while a modified ToBI labeling scheme is applied to both languages to ensure cross-linguistic compatibility. A qualitative component analyzes instances where listeners’ judgments diverge from acoustic measures, applying thematic analysis to perceptual data and linking findings to perceived emphasis and discourse coherence. Reliability of annotation is established through inter-annotator agreement (Cohen’s kappa), targeting ? ? 0.75. Expected findings include robust cross-language differences in tonal alignment English tends to exhibit more phrasal boundary-driven pitch resets and higher variability in focal pitch attainment, whereas Mandarin displays more syllable-toned, lexicalized pitch movements with clearer syllable-level alignment to discourse boundaries. Genre effects are anticipated, with narrative and expository styles showing stronger boundary cues in English and Mandarin, respectively. Perceptual results are expected to corroborate acoustic distinctions, though some cross-language listeners may rely more on contextual cues than on pitch in certain genres. The study anticipates that some universal prosodic tendencies—such as increased boundary lengthening at clause boundaries—will emerge, alongside language-specific realizations shaped by phonological inventories and segmental timing. Contribution to knowledge includes (i) an empirically grounded cross-linguistic map of English and Mandarin prosody across genres; (ii) methodological refinement for cross-language ToBI annotation and cross-language acoustic analysis; (iii) empirical evidence informing AM theory and prosody-syntax interfaces in a bilingual context; and (iv) practical implications for second-language teaching, speech technology, and clinical phonology by clarifying how discourse pragmatics interface with prosody in typologically distinct languages. The study concludes that while certain prosodic functions are shared across languages, language-specific phonological constraints and discourse strategies produce distinct realizations that must be accounted for in theoretical models and applied tools. Recommendations include adopting language-aware prosody models in speech synthesis and recognition systems and extending the cross-linguistic framework to additional tonal and non-tonal languages to validate the generalizability of observed patterns.