Analyzing Polish Mandarin Learners’ Tones Using Generalized Additive Mixed Models

Abstract

The study investigates the monosyllabic tonal acquisition of 10 female Polish-speaking beginner learners of Mandarin and compares their patterns of tone production with 9 Taiwanese Mandarin native speakers. The dynamic data were analyzed using a generalized additive mixed model approach with 11 evenly divided frequency bands (in Hz) extracted from the tonal-bearing unit and converted into T-values. The results show that the L1 Polish learners displayed a relatively lower initial point in producing T4, while T1, T2, and T3 were all well-produced in terms of register and contour. It is suggested that L1 Polish female beginners practice setting up the starting point of T4 as T1. The generalized additive mixed models approach has been shown to be effective in describing Chinese tone patterns, including those of native speakers, in a scientific way, by applying a smoothing procedure to items read out by participants so that they can be successfully compared.