PASCAL and DPA: A Pilot Study on Using Prosodic Competence Scores to Predict Communicative Skills for Team Working and Public Speaking

Oliver Niebuhr, Jan MichalskyInterspeech 2019

TL;DR

This study explores whether acoustic features of speech can predict a person's effectiveness in public speaking and team collaboration. Researchers found that "prosodic competence" scores, derived from speech analysis, significantly correlate with students' grades in oral presentations and their self-assessed team performance, suggesting that how we speak can be a strong indicator of how well we communicate.

What Problem Does This Paper Address?

Effective communication is crucial in both individual endeavors like public speaking and collaborative efforts like team projects. While we intuitively associate good communicators with certain speaking styles, quantifying this connection and using it for prediction has been challenging. This paper addresses the gap in understanding how specific acoustic-prosodic features of speech relate to measurable outcomes in public speaking and team working. The authors hypothesize that analyzing these acoustic patterns can provide a predictive measure of communicative competence, moving beyond subjective assessments.

Methods

The study introduces two acoustically-based assessment methods:

  • PASCAL (Prosodic Analysis of Speaker Charisma: Assessment and Learning): This method assesses charismatic and persuasive public speaking. It analyzes speech recordings for specific acoustic features (like pitch register, pitch variation, speaking rate, and voice quality) that are known to contribute to a "strong delivery." The PASCAL score is derived from a subset of these features, recognizing that charisma isn't linearly related to these features but rather has an optimal "sweet spot."
  • DPA (Dynamic Prosodic Adaptation): This method measures prosodic entrainment, which is the ability of speakers to adjust their acoustic features (like pitch) to match their interlocutor. Participants listened to speech stimuli with varying pitch and were asked to respond naturally while adapting their own pitch. The DPA score is calculated based on the correlation between the stimulus and response patterns, focusing on the mean fo (fundamental frequency) of the speech.

These scores were then correlated with students' performance in a real-world engineering project.

Datasets

The study used speech recordings from 22 first-semester Electrical Engineering students at the University of Southern Denmark. These students participated in a 14-week project to design and build a weather station. The recordings were made at the beginning of the project for the PASCAL assessment and during a specific task for the DPA assessment. The performance data used for correlation came from the students' final oral project presentations, written reports, and self-assessments of team performance.

Key Insights and Ideas

  • "How you say it" matters as much as "what you say": The study strongly suggests that the acoustic delivery of speech (prosody) is a significant predictor of communicative success, particularly in public speaking.
  • Charisma is acoustically detectable: The PASCAL score, derived from speech acoustics, showed a strong correlation with oral presentation grades, indicating that features associated with charismatic speaking can be objectively measured and predict performance.
  • Prosodic adaptation signals team competence: The DPA score, measuring the ability to adapt prosody, correlated with self-assessed time management and team work, as well as the objective quality of the final product. This suggests that subtle acoustic adjustments in conversation are linked to effective collaboration.
  • Potential for objective assessment: These findings open doors for developing objective, acoustically-based tools to assess and potentially train communication skills in various professional contexts.

Key Results

  • PASCAL Score: A strong positive correlation (R² = 0.56) was found between individual PASCAL scores and the grades for the final oral presentation. A weaker but significant positive correlation (R² = 0.24) was observed between PASCAL scores and the grades for the written report.
  • DPA Score: Significant positive correlations were found between DPA scores and self-assessed time management (R² = 0.20) and product quality (R² = 0.20).
  • Overall: Higher PASCAL and DPA scores were associated with better performance in the weather station project, both individually and in teams.

Limitations and Future Work

  • Pilot Study: The authors acknowledge this is a pilot study with a relatively small sample size (N=21).
  • Self-Assessment Bias: The DPA score's correlation with time management and team work was based on self-assessment, which might be influenced by subjective perceptions rather than objective reality. Anecdotal evidence suggests that higher DPA teams might perceive their work as better, even if their focus sometimes strayed from the task.
  • L2 Speakers: The study involved L2 speakers of English, and while no L1-specific effects were found, cross-cultural differences in charisma and entrainment warrant further investigation.
  • Speaker Sex: While the models accounted for speaker sex, the sample was skewed, and this aspect needs more thorough investigation.
  • Future Directions: The authors suggest further research to directly measure team work competence as a social and communicative skill and correlate it with DPA. They also propose investigating how these prosodic measures might differ across professions, given their observations in business consulting.

Summary updated

BibTeX
@inproceedings{niebuhr2019pascal,
  author = {Oliver Niebuhr and Jan Michalsky},
  title = {{PASCAL and DPA: A Pilot Study on Using Prosodic Competence Scores to Predict Communicative Skills for Team Working and Public Speaking}},
  booktitle = {Interspeech},
  year = {2019},
}
Figure