Speech planning influences the acoustic realization of phonological alternation: Evidence from morpho-syntactically complex Mandarin Tone 3 sandhi

Speech planning operates at multiple levels of production and shapes the acoustic realization of speech. Phonetic gradience observed in phonological alternation may arise from both articulatory-level planning and higher-level morphophonological planning. Using the degree of incomplete neutralization in morpho-syntactically complex Mandarin Tone 3 sandhi (T3 + T3 → T2 + T3) as a test case, we conducted a production experiment that manipulated syntactic boundary strength and speech rate to assess the contribution of morphophonological-level planning to acoustic realization. Specifically, we analyzed Time × Tone × Syntactic Boundary and Time × Tone × Duration interactions to disentangle the effects of phonological planning from those of articulatory processes. Results showed a greater phonetic difference between Sandhi T3 and Nonsandhi T2 tones (i.e., a larger effect size of incomplete neutralization) under stronger syntactic boundaries and longer durations. These findings support a planning-based account: stronger boundaries and longer durations delay access to upcoming phonological context, allowing more information from underlying forms to cascade into articulation. This pattern is not predicted by the incremental planning hypothesis or by articulatory-level explanations such as pre-boundary lengthening or anticipatory tonal coarticulation reduction. We argue that morphophonological-level planning shapes acoustics through domain-general cascading activation, leaving residual traces of unselected forms in the output. These findings refine our understanding of how speech planning interfaces with phonological alternation and contribute to ongoing debates over symbolic versus gradient representations in speech production theory.