Contextually dependent cue weighting for a laryngeal contrast in Shanghai Wu

Jie Zhang, Hanbo YanICPhS 2015

Overview

Shanghai Wu features a complex laryngeal contrast traditionally described as a distinction between "voiced" (murmured) and "voiceless" (modal) consonants. This study investigates how this contrast is signaled across different consonant types—stops, fricatives, and sonorants—and how listeners adapt when primary acoustic cues are neutralized by phonological processes like tone sandhi.

From Phonation to Pitch

While historical descriptions of Shanghai Wu emphasized vocal quality (breathiness) as the defining feature of the laryngeal contrast, contemporary production data from younger speakers suggests a significant shift. Pitch (fundamental frequency, f0) has emerged as the primary signal for the contrast in isolated words. However, the reliance on pitch is not absolute; it is a dynamic "trading relation" that varies significantly depending on the linguistic context.

The Impact of Tone Sandhi

A major challenge for maintaining this contrast occurs in "tone sandhi" contexts, where the pitch of a syllable is altered when it appears in a compound word. The study found that for stops and sonorants, this process largely neutralizes the pitch differences that normally distinguish voiced and voiceless onsets.

Normalized F0 for stops in the disyllabic sandhi context, showing the neutralization of the pitch cue.

As a result of this neutralization, listener identification accuracy for stops and sonorants drops significantly in sandhi contexts, with sonorants falling to chance levels. This suggests that some phonological contrasts may only be "real" or functional in specific prosodic positions.

Contextual Flexibility and Cue Trading

The most striking finding is the robustness of fricatives. Unlike stops and sonorants, fricatives maintain a significant contrast even under sandhi. Listeners achieve this by shifting their attention to secondary cues, such as consonant duration and residual phonation differences, when the primary pitch cue becomes unreliable.

Consonant duration for stops, fricatives, and sonorants, highlighting the significant difference in fricatives.

Statistical analysis (CART) confirmed that while f0 is the sole significant predictor for stops and sonorants in isolation, listeners integrate f0, duration, and phonation for fricatives, demonstrating a sophisticated and flexible perceptual strategy.

Conclusion

The study demonstrates that speech perception is not "hard-wired" to single acoustic features. Instead, cue weighting is highly context-dependent, allowing for the preservation of phonological contrasts through dynamic trading relations. These findings highlight the importance of considering both segmental and prosodic context in phonetic research and suggest an ongoing evolution in the phonological system of Shanghai Wu as it moves from a phonation-based to a tone-based contrast.

Summary updated

BibTeX
@inproceedings{zhang2015contextually,
  author = {Jie Zhang and Hanbo Yan},
  title = {{Contextually dependent cue weighting for a laryngeal contrast in Shanghai Wu}},
  booktitle = {ICPhS},
  year = {2015},
}
Figure