Presentation of self and symptoms in primary care consultations involving patients from non-English speaking backgrounds.

This paper draws on the PLEDGE research project (Patients with Limited English and Doctors in General Practice) 1 The Patients with Limited English and Doctors in General Practice (PLEDGE) project was funded by Sir Siegmund Warburg's Voluntary Settlement (2001-2003). The research team was: Celia Roberts, Roger Jones, Becky Moss, Srikant Sarangi and Val Wass. which has a database of 232 video-recorded interactions from GP surgeries in South East London. We focus on the opening episodes-the first opportunity the patient has to report on why they have come to see the doctor-to explore some of the contrasts in self presentation and the interactional work that doctors do when faced with the unexpected. Patients who speak a local London or standard variety of English present three aspects: a description of symptoms, the context in which they occurred, and an affective or epistemic stance. These 'micro discourse routines' are accomplished interactionally through the design of figure/ground relationships, framing and metacommunication and presentation of the 'moral self'. Although some patients from non-English speaking backgrounds use broadly similar 'micro discourse routines', the majority configure the relationship between medically salient facts, adequate contextual information and the stance which conveys the 'moral self' in different and apparently less 'orderly' ways. So openings often become protracted and harder work interactionally for both sides. While conversation analytic studies and communication skills textbooks represent the medical consultations as orderly, we suggest that such apparent orderliness must, at least, be partly the result of ironing out linguistic and cultural diversity. Interactional sociolinguistic analysis is used to shed light on the design of these routines and to provide analytic frameworks for doctors in reflecting on their own practice in ways which challenge patient-centred models.