Phonetic convergence and speech intelligibility
Presentation Type
Abstract
Faculty Advisor
Jennifer Pardo
Access Type
Event
Start Date
25-4-2025 12:00 PM
End Date
25-4-2025 1:00 PM
Description
Phonetic convergence is the phenomenon in which listeners naturally adjust their speech patterns to match those of another talker. It has been hypothesized that phonetic convergence might enhance intelligibility to the target of convergence, the person whose speech the listener is mimicking, due to a self-speech advantage. This is the idea that people understand their own speech better because they are more familiar with their articulatory patterns, which can make speech more intelligible to themselves. To explore this hypothesis, the study was conducted in two phases. In the first phase, a speech intelligibility task was performed, where 20 model talkers were shadowed by 20 listeners to determine if shadowed speech was more intelligible than non-shadowed (baseline) speech. In the second phase, an AXB perceptual similarity task was conducted, where listeners heard three versions of the same word: two from the shadower (A and B), one from the model (X), and then judged if the shadowed speech was more like the model talker's speech. The results showed that a model talker’s own speech was more intelligible than other talkers, supporting the self-speech advantage hypothesis. Shadowing talkers converged phonetically to the model, but no positive relationship between intelligibility and phonetic convergence was found. Instead, a negative relationship was observed, where greater convergence was linked to lower intelligibility. These findings suggest that phonetic convergence may not enhance speech intelligibility, as it could be driven by a social alignment strategy, where listeners adjust their speech to build a social connection with the speaker.
Phonetic convergence and speech intelligibility
Phonetic convergence is the phenomenon in which listeners naturally adjust their speech patterns to match those of another talker. It has been hypothesized that phonetic convergence might enhance intelligibility to the target of convergence, the person whose speech the listener is mimicking, due to a self-speech advantage. This is the idea that people understand their own speech better because they are more familiar with their articulatory patterns, which can make speech more intelligible to themselves. To explore this hypothesis, the study was conducted in two phases. In the first phase, a speech intelligibility task was performed, where 20 model talkers were shadowed by 20 listeners to determine if shadowed speech was more intelligible than non-shadowed (baseline) speech. In the second phase, an AXB perceptual similarity task was conducted, where listeners heard three versions of the same word: two from the shadower (A and B), one from the model (X), and then judged if the shadowed speech was more like the model talker's speech. The results showed that a model talker’s own speech was more intelligible than other talkers, supporting the self-speech advantage hypothesis. Shadowing talkers converged phonetically to the model, but no positive relationship between intelligibility and phonetic convergence was found. Instead, a negative relationship was observed, where greater convergence was linked to lower intelligibility. These findings suggest that phonetic convergence may not enhance speech intelligibility, as it could be driven by a social alignment strategy, where listeners adjust their speech to build a social connection with the speaker.
Comments
Poster presentation at the 2025 Student Research Symposium.