Research on miscommunication has not ceased to grow since the early 1980s, especially in connection with the analysis of cross-cultural communication. Yet, this interest has not stimulated a critical debate on the theoretical models underpinning most miscommunication research. This study aims at encouraging scientific discussion by advocating an empirical treatment of communicative conflict, that is one which is grounded in the detailed examination of linguistic data. Through the fine-grained analysis of participants' sense-making processes in a corpus of real-life intercultural data, we seek to unveil the linguistic and conversational strategies that speakers put to work in the handling of miscommunication. Our empirical analysis of understanding difficulties, based exlusively on observable trouble in talk, suggests that analysts can legitimately focus only on those stretches of talk which are experienced as problematic by speakers themselves. It also shows that miscommunication cannot be attributed to an individual speaker, but that it is jointly constructed by interactants through the ways in which they assess and respond to each other's conversational contributions. 1. INTRODUCTION
