Validation of interpretative analysis investigating the methodology: Voice quality in call centre discourse

Abstract

Voice quality features can be viewed as a paralinguistic and multimodal resource to convey meanings (Leijssen, 2006, Martin, 2007). A number of studies establish a strong association between voice quality and interpersonal meanings (c.f. Leijssen, 2006, Stokoe and Edwards, 2007), especially a change in voice quality leads to a change in interpersonal meaning (Buchbinder, 2008). This is particularly true in telephone interactions which, because of the absence of visual cues，results in the communicators relying heavily on ‘differences in voice quality to assess emotional impact’ (McCoyd and Kerson, 2006:402). Voice quality features are significant for expressing attitude, but the meaning making process is often considered to be unclear and less systematic in the existing literature. Descriptions of voice quality have traditionally comprised ‘qualitative terms’ or ‘adjective’ such as warm, rough, creaky, breathy and dull (Titze and Story, 2002). Some call centre trainers suggest the customer service representatives (CSR) （to）‘sound’ positive and friendly, or use ‘appropriate tone’ when communicating with customers. However, typically training materials do not address the relevant voice quality feature before they come up with a conceptual meaning or interpretation. The voice quality features that result in a ‘positive and friendly voice’ have not yet been to be fully explored. Drawing on Systemic Functional Linguistics, van Leeuween (1999) has developed the Systems Network of sound, voice and music, a highly theorized Sound Quality framework to model the meaning making systems. van Leeuwen’s (1999) System Network is the starting point for the current voice quality study. This presentation will show examples of how voice quality features create meaning potential in call centre context, and the methodology of identifying such features. Bateman (2008) indicates that multimodal meaning has to be interpreted more rigorously with transparent validation. Following a qualitative research approach, a preliminary inter-rater agreement has been conducted to identify obvious voice quality changes that infer a change in interpersonal meaning. The results of this initial study of voice quality validation processes can be considered as a new challenge to text and multi-modal analysis. These findings are crucial to investigate the development of validating voice quality features in call centre interactions and help to strengthen the credibility of the presented findings. The current study will be of great benefit to training in the call centre industry and the validation of multimodal analysis.