Abstract
This study investigates the longitudinal acoustic development of lexical Mandarin tones among adult native English speakers during an eight-week intensive language immersion program. Utilizing a microgenetic research design with bi-weekly acoustic recordings across 16 measurement points, we tracked intra-individual variations in fundamental frequency (F0) height, F0 slope, and pitch velocity across the four primary Mandarin citation tones (Tone 1 high-level, Tone 2 high-rising, Tone 3 low-dipping, and Tone 4 high-falling). Time-normalized F0 contours were extracted from target monosyllabic and disyllabic tokens produced in carrier sentences and compared against acoustic baselines established by native Mandarin speakers. Acoustic generalized additive mixed-effects models (GAMMs) revealed distinct developmental trajectories: Tone 1 (55) and Tone 4 (51) stabilized rapidly within the first three weeks of immersion, characterized by early pitch height accuracy. Conversely, Tone 2 (35) and Tone 3 (214) exhibited pronounced microgenetic variability, marked by non-linear regression phases where learners struggled to decouple pitch register from dynamic pitch velocity. Disyllabic context analysis demonstrated that Tone 3 sandhi rule application lagged significantly behind citation tone production. These empirical findings demonstrate that L2 tone acquisition is non-monotonic and governed by distinct perceptual and biomechanical constraints, offering crucial pedagogical implications for visual pitch feedback in technology-enhanced immersion curricula.
