Conversational Success in Twitter Webcare Dialogues

Social media have radically changed the communication between organizations and their customers. Organizations are faced with the questions whether they should be active in social media, and if so, how they should respond in a simultaneously efficient and effective manner. This project addresses these questions by investigating the relationship between the course, outcome, and linguistic characteristics of social media conversations.

VU researchers from Communication Science and Linguistics have previously addressed these issues, but from different angles. Communication researchers have studied how webcare messages from organizations are perceived by users in terms of brand evaluation by means of experiments. Recently, 2600 Twitter webcare dialogues from a Dutch organization have been collected. This dataset has been analyzed in a quantitative way using automated text analyses, such as LIWC, in order to determine the course and the outcome of conversations. Linguists have followed a more qualitative approach using insights from conversation analysis to study online interaction between organizations and customers in depth. A dataset of 100 Twitter webcare dialogues from Dutch hospitals has been collected. These dialogues are part of so-called Twitter consultation hours in which users ask questions to a health professional on a particular medical topic.

The aim of the current project is to combine expertise to further analyze these Twitter conversations in three ways:

  1. By using micro analyses of the course of the interaction, the linguistic characteristics of the interaction (e.g., personal vs. impersonal style), and the outcome of the interaction (positive vs. negative) automatic text analysis will be enriched.
  2.  Social media managers will explore their Twitter webcare conversations using the discursive action method3 to become aware of how they talk to their customers.
  3.  Audience effects of different Twitter dialogues between organization and customers will be investigated by means of experiments.



  • Sander Wensink
    MSc Communication Science
  • Joyce van Oosterhout
    MA Communication and Information studies