Abstract
Recent urban varieties of Mandarin among younger speakers include Japanese phonemic loanwords, e.g., yàsāxī ‘gentle’. Unlike most Chinese words, typically disyllabic, many Japanese loanwords are polysyllabic and their source forms carry lexically specified pitch patterns, such as L-H-H-H (e.g., yasashii ‘gentle’). Focusing on trisyllabic loanwords, this study addresses: 1) What pitch patterns do native Mandarin speakers without systematic exposure to Japanese produce? Do they follow Chinese character-based pronunciation (i.e., yàsāxī falling-high-high), the Japanese pitch pattern, or something else? 2) What predicts such patterns?
A real-word production task was conducted with 40 native Mandarin speakers (aged 18–35, 20F). After a language-experience questionnaire (LHQ3) [1] and a 2AFC identification task on lexical familiarity with six frequent trisyllabic loanwords, participants read these words aloud in three context conditions: isolated word, in (contextualized and controlled) carrier-sentences (all declarative, with critical words sentence-medial). The Chinese character-based tonal patterns of these words were dipping-falling-high, falling-high-high, and high-rising-high respectively. 1134 tokens were analyzed after data cleaning.
For question 1, hierarchical agglomerative clustering (Ward’s linkage, Euclidean distance) identified two optimal clusters via silhouette analysis, featuring two patterns (Fig. 1): L-H-L (61.29%) and L-H-H (38.71%), contrasting the strong-weak-strong pattern typical of Mandarin trisyllabic words [2]. For question 2, context, lexical familiarity, and character tonal patterns were expected to influence the produced pitch patterns. Generalized Additive Mixed Models (GAMMs) showed that beyond individual variation and word-specific contours, context was the strongest predictor of F0 contour variability, while familiarity and character tonal patterns had minor effects.
To conclude, this study observed unique pitch patterns (neither Japanese-like nor Chinese-like, e.g., Fig. 2) of the elicited trisyllabic Japanese loanwords and a strong context effect. Furthermore, individual variation and word-specific pitch patterns while producing such loanwords require further investigation.
A real-word production task was conducted with 40 native Mandarin speakers (aged 18–35, 20F). After a language-experience questionnaire (LHQ3) [1] and a 2AFC identification task on lexical familiarity with six frequent trisyllabic loanwords, participants read these words aloud in three context conditions: isolated word, in (contextualized and controlled) carrier-sentences (all declarative, with critical words sentence-medial). The Chinese character-based tonal patterns of these words were dipping-falling-high, falling-high-high, and high-rising-high respectively. 1134 tokens were analyzed after data cleaning.
For question 1, hierarchical agglomerative clustering (Ward’s linkage, Euclidean distance) identified two optimal clusters via silhouette analysis, featuring two patterns (Fig. 1): L-H-L (61.29%) and L-H-H (38.71%), contrasting the strong-weak-strong pattern typical of Mandarin trisyllabic words [2]. For question 2, context, lexical familiarity, and character tonal patterns were expected to influence the produced pitch patterns. Generalized Additive Mixed Models (GAMMs) showed that beyond individual variation and word-specific contours, context was the strongest predictor of F0 contour variability, while familiarity and character tonal patterns had minor effects.
To conclude, this study observed unique pitch patterns (neither Japanese-like nor Chinese-like, e.g., Fig. 2) of the elicited trisyllabic Japanese loanwords and a strong context effect. Furthermore, individual variation and word-specific pitch patterns while producing such loanwords require further investigation.
Publication type
Poster
Presentation
DvdF25_P1_Hou_Lameris.pdf
(350.38 KB)
Year of publication
2025
Conference location
Utrecht
Conference name
Dag van de Fonetiek 2025
Publisher
Nederlandse Vereniging voor Fonetische Wetenschappen