The role of phonation type in Chinese Jin tones: a study using acoustic metrics

This paper examines how phonation shapes two marginal tonal categories in a Jin dialect and links the findings to an acoustic metric of voice quality. In Huoji Chinese, the historical checked tone and an emergent neutral-like tone are both short and often non-modal, but they arise from different sources. Checked tokens retain stop coda reflexes through glottal stops and creaky spans. Neutral tokens emerge through prosodic reduction and show context dependent glottalisation or added aspiration noise, with an uncertain distinction between breathy and whispery phonation. Recordings from 17 speakers were segmented at vowel nuclei and measured in a single VoiceSauce pipeline (F0, H1-H2, band limited HNR, duration), with parallel auditory coding for modal, creaky and noisy nonmodal voice. Multidimensional scaling of F0 and duration places the checked tone near the level tone. Adding HNR shifts it further, and adding H1-H2 produces clear separation and almost doubles the mean inter-category distance. Neutral tokens form a band of phonation patterns. Affixed forms tend to glottalise, while root reductions range from modal through creaky to whisper, with whisperiness most common when glottalisation is absent. In line with recent work that moderate correlations between HNR, CPP and creakiness, the Huoji data show that low HNR is ambiguous and can reflect noise or aperiodicity. Interpreting HNR with H1-H2 and duration separates noisy breathiness from irregular, creaky like phonation. The same metrics are applied to Laver’s laryngeal settings. MDS places Huoji tones and controlled English phonation types in a shared H1-H2, HNR and CPP space.