University of Twente Student Theses
Measuring Conversation Quality
Salverda, Jens (2025) Measuring Conversation Quality.
|
PDF
5MB |
| Abstract: | Prior research has identified key factors contributing to enjoyable conversations, including rapport, synchrony and empathy among interlocutors. This study aims to operationalize these indicators within customer-agent interactions by examining both verbal and non-verbal elements of speech, such as pitch range, voice intensity andturn-taking. While also exploring the potential of physiological measures such as heart rate. The objective is to enable an assessment of conversation quality in terms of the amount of rapport the interlocutors have, which holds significant utility for contact centres. The application can empower guidance to customer service representatives, fostering alignment with customers and more harmonious interactions. To achieve this, experiments are conducted involving dyadic conversations between a trained actor and participants (N =6,aged18–28). Each conversation consisted of a stationary phase and a support-breaking phase to capture shifts in interaction dynamics. Multimodal data, including audio, video, and physiological measurements, were collected alongside participants’ subjective perceptions of rapport via questionnaires. Key findings revealed significant differences between the normal and non-rapport phases in synchronized smiling, speech rate, and prosodic features such as the standard deviation of pitch and intensity. Machine learning models (Random Forest and Logistic Regression) achieved 84% accuracy in classifying rapport and non-rapport moments. However, the exclusion of nonverbal features, such as synchronized smiling and head movements, reduced classification accuracy to 58%, underscoring the importance of nonverbal cues in rapport measurement. This study provides insights into measuring and fostering rapport in customer-agent interactions, with implications for improving conversational quality and customer satisfaction in contact centers. |
| Item Type: | Essay (Master) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Interaction Technology MSc (60030) |
| Link to this item: | https://purl.utwente.nl/essays/105179 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page