Stability of prosodic boundary cue production in interactive settings

Prosodic boundaries are used to disambiguate structurally ambiguous sentences by signaling how words or phrases group together. However, it remains debated whether prosodic boundary production is situationally adaptive (listener-oriented) or a by product of speakers' own speech planning processes, i.e., situationally independent (speaker-oriented). This study examined whether prosodic cues (fundamental frequency range, final lengthening, and pause duration) are modulated by listener feedback. Thirty adult native German speakers (mean age: 24.2 years) produced coordinate name sequences, either with (e.g., (Moni und Lilli) und Lisa) or without (Moni und Lilli und Lisa) syntactic branching while interacting with a confederate listener. The confederate listener was pre-programmed occasionally to misunderstand the speaker and request a repetition. We expected that if prosodic boundary marking is listener oriented, speakers would shift across speaking styles, from a casual speaking style to a clear speaking style, and enhance boundary cues following perceived listener difficulty. Results showed that speakers reliably marked syntactic boundaries using all three prosodic cues (f0 range, final lengthening, and pause duration), in line with the Proximity/Antiproximity Principle. We found inter-individual variability in cue combination patterns that speakers used to mark boundaries: pause duration was used most frequently, followed by f0 range, while final lengthening appeared only in combination with other cues. In contrast, speakers showed intra-individual stability, maintaining consistent cue use across speaking styles despite listener feedback. Together, these findings reveal that, although there is inter-individual variability in prosodic cue use, disambiguating prosody is speaker-oriented and situationally independent, even in interactive settings involving feedback.