A Cross-Dialect Comparison of Generational Variation in a High-Falling Tone: Evidence from Kunming and Xi'an Dialects

Although intergenerational differences are widely used to investigate language variation and change in apparent time, systematic cross-dialect evidence for comparable tonal categories remains limited. This study examines the high-falling T3 in the urban dialects of Kunming and Xi'an, two Mandarin dialects from different regional backgrounds. The dataset consists of 60 native speakers, divided into older, middle-aged, and younger groups, with balanced gender representation within each group. Fundamental frequency (F0) was extracted using Mini-Speech-Lab and analysed through nine time-normalised sampling points, duration, slope measures, and first-order differences. Within-dialect generational differences were examined using Kruskal–Wallis tests, while cross-dialect differences within corresponding age groups were examined using Mann–Whitney tests, with Holm correction applied for multiple comparisons. The results show that, although T3 in both dialects belongs broadly to the high-falling category, their acoustic implementation and generational patterning are not identical. Kunming shows more robust within-dialect generational differentiation than Xi'an, with significant differences concentrated in the early-to-middle contour and selected dynamic measures, whereas cross-dialect differences are widest in the older group and become more selectively concentrated in the middle and later portions of the contour among the middle-aged and younger groups. These findings suggest that comparable tonal categories do not necessarily exhibit parallel generational patterning across dialects. The study contributes to Chinese sociophonetics by showing that cross-dialect similarity should be evaluated not only through tonal labels, but also through the temporal distribution and structural organisation of acoustic differences.