Mandarin Chinese Prosodic Phrase Grouping and Modeling—Method and Implications

One major feature of the prosody of Mandarin Chinese speech flow is prosodic phrase grouping [1, 2, and 3]. Phrasal and sentential intonations are governed by a prosody framework that structurally group phrases into a larger/longer and identifiable unit. An overall prosody pattern of such phrase grouping with prosodic specifications is superimposed on phrase group. In turn, individual phrasal intonation under prosody grouping has to adjust in accordance with structural specification from the prosody framework. The output is then seen as derived outcome. The aim of the present paper is to experiment how to simulate prosodic phrase grouping using the Fujisaki intonation model that originally specifies only phrasal or sentential intonations, and how such an intonation model can be further enhanced by incorporating prosodic specifications such as boundaries and breaks, prosody levels/layers and phrase positions under the notion of phrase grouping. The experiments began with aligning the phrase command of the intonation model to boundaries of breaks in the speech flow, then examining prosodic characteristics such as relative position of the target phrase within a prosodic phrase group. Finally, using a linear regression model to predict prosody output from prosodic words upward, predictions of an overall pattern for prosodic phrase grouping was derived. The pattern matched with a prosody base form aimed at prosodic phrase grouping; it also accounted for how and why phrasal intonations were modified in relation to prosody organization. Hence, phrasal intonation is seen as components of prosodic phrase grouping.