Applying Linguistics in Illness and Healthcare Contexts: Contemporary Studies in Linguistics 9781350057654, 9781350057685, 9781350057661

Citation preview

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Contemporary Studies in Linguistics Series Editor: Li Wei, Chair of Applied Linguistics, University College London, UK The Contemporary Studies in Linguistics series presents state-of-the-art accounts of current research in all areas of linguistics. Written by internationally renowned linguists, the volumes provide a selection of the best scholarship in each area. Each of the chapters appears on the basis of its importance to the field, but also with regards to its wider significance either in terms of methodology, practical application or conclusions. The result is a stimulating contemporary snapshot of the field and a vibrant reader for each of the areas covered the in series.

Titles in the Series: Applying Linguistics in Illness and Healthcare Contexts, edited by Zsófia Demjén Contemporary Applied Linguistics Volume 1, edited by Li Wei and Vivian Cook Contemporary Applied Linguistics Volume 2, edited by Li Wei and Vivian Cook Contemporary Computer-Assisted Language Learning, edited by Michael Thomas, Hayo Reinders and Mark Warschauer Contemporary Corpus Linguistics, edited by Paul Baker Contemporary Critical Discourse Studies, edited by Christopher Hart and Piotr Cap Contemporary Linguistic Parameters, edited by Antonio Fabregas, Jaume Mateu and Michael Putnam Contemporary Media Stylistics, edited by Helen Ringrow and Stephen Pihlaja Contemporary Stylistics, edited by Marina Lambrou and Peter Stockwell Contemporary Task-Based Language Teaching in Asia, edited by Michael Thomas

CONTEMPORARY STUDIES IN LINGUISTICS

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS Edited by Zsófia Demjén

BLOOMSBURY ACADEMIC Bloomsbury Publishing Plc 50 Bedford Square, London, WC1B 3DP, UK 1385 Broadway, New York, NY 10018, USA 29 Earlsfort Terrace, Dublin 2, Ireland BLOOMSBURY, BLOOMSBURY ACADEMIC and the Diana logo are trademarks of Bloomsbury Publishing Plc First published in Great Britain 2020 This paperback edition published in 2022 Copyright © Zsófia Demjén and Contributors, 2020 Zsófia Demjén has asserted her right under the Copyright, Designs and Patents Act, 1988, to be identified as Editor of this work. For legal purposes the Acknowledgements on p. xvi constitute an extension of this copyright page. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying, recording, or any information storage or retrieval system, without prior permission in writing from the publishers. Bloomsbury Publishing Plc does not have any control over, or responsibility for, any third-party websites referred to or in this book. All internet addresses given in this book were correct at the time of going to press. The author and publisher regret any inconvenience caused if addresses have changed or sites have ceased to exist, but can accept no responsibility for any such changes. A catalogue record for this book is available from the British Library. A catalog record for this book is available from the Library of Congress. ISBN: HB: 978-1-3500-5765-4 PB: 978-1-3502-4712-3 ePDF: 978-1-3500-5766-1 eBook: 978-1-3500-5767-8 Series: Contemporary Studies in Linguistics Typeset by Deanta Global Publishing Services, Chennai, India To find out more about our authors and books visit www.bloomsbury.com and sign up for our newsletters.

CONTENTS

LIST OF ILLUSTRATIONS LIST OF CONTRIBUTORS ACKNOWLEDGEMENTS Introduction Zsófia Demjén Part I The experience of illness 1 ‘One gives bad compliments about me, and the other one is telling me to do things’ – (Im)politeness and power in reported interactions between voice-hearers and their voices Zsófia Demjén, Agnes Marszalek, Elena Semino and Filippo Varese 2 Corpus linguistics in illness and healthcare contexts: A case study of diabulimia support groups Gavin Brookes 3 Using a comparative corpus-assisted approach to study health and illness discourses across domains: The case of postnatal depression (PND) in lay, medical and media texts Karen Kinloch and Sylvia Jaworska 4 Applying corpus linguistics to a diagnostic tool for pain Elena Semino, Andrew Hardie and Joanna Zakrzewska

viii x xvi 1

15

17

44

73

99

vi

CONTENTS

Part II Relating to each other

129

5 Improving HIV/AIDS consultations in Malawi: How interactional sociolinguistics can contribute Rachel Chimbwete-Phiri and Stephanie Schnurr

131

6 Empathy displays in Dutch chat counselling: Showcasing a microanalysis of online data Wyke Stommel and Joyce Lamerichs

159

7 The functions of narrative passages in three written online health contexts Franziska Thurnherr, Marie-Thérèse Rudolf von Rohr and Miriam A. Locher

184

Part III Illness in the mass media

219

8 Fighting obesity, sustaining stigma: How can critical metaphor analysis help uncover subtle stigma in media discourse on obesity Dimitrinka Atanasova and Nelya Koteyko

221

9 A media brew of implied, hidden and unknown risk claims: Cognitive discourse analysis of public health communication Chris Tang and Gabriella Rundblad

242

Part IV Professional practices and concerns

269

10 Effective triaging in general practice receptions: A conversation analytic study Rein Sikveland and Elizabeth Stokoe

271

11 A sociolinguistic investigation of professional mobility and multicultural healthcare communication Olga Zayts and Mariana Lazzaro-Salazar

295

12 Applying corpus-based discourse analysis to enhance understanding of barriers to palliative and end of life care provision in general practice Joelle Loew, Sarah Mitchell, Katharine Weetman, Catherine Millington-Sanders and Jeremy Dale

321

CONTENTS

vii

13 A moment outside time: A critical discourse analytic perspective on dominant constructions of suicide Dariusz Galasiński and Justyna Ziółkowska

349

EPILOGUE BY JONATHON TOMLINSON INDEX

372 379

ILLUSTRATIONS

FIGURES 2.1

Sample concordance of ‘insulin’

4.1

The McGill Pain Questionnaire (Melzack, 1983: 44)

101

4.2

Frequency of patient selections and frequency of co-occurrence with ‘pain’ in the OEC for Group 1 descriptors in the MPQ

116

Frequency of patient selections and frequency of co-occurrence with ‘pain’ in the OEC for Group 2 descriptors in the MPQ

116

Frequency of patient selections and frequency of co-occurrence with ‘pain’ in the OEC for Group 3 descriptors in the MPQ

116

Frequency of patient selections and frequency of co-occurrence with ‘pain’ in the OEC for Group 4 descriptors in the MPQ

117

Frequency of patient selections and frequency of co-occurrence with ‘pain’ in the OEC for Group 6 descriptors in the MPQ

117

Frequency of patient selections and frequency of co-occurrence with ‘pain’ in the OEC for Group 7 descriptors in the MPQ

117

Frequency of patient selections and frequency of co-occurrence with ‘pain’ in the OEC for Group 8 descriptors in the MPQ

118

9.1

Example of a nominal frame

249

9.2

Example of an action frame

249

12.1

Sample of ‘lack’ concordance lines

332

12.2

Sample of negation (‘no’ and ‘not’) concordance lines

335

12.3

Sample of palliative and hospice concordance lines

339

4.3 4.4 4.5 4.6 4.7 4.8

57

ILLUSTRATIONS

ix

TABLES 1.1 1.2 2.1 2.2 3.1 3.2 3.3 3.4 3.5 4.1 4.2 4.3 4.4 4.5 4.6

6.1 7.1 7.2 8.1 8.2 9.1 9.2 12.1 12.2 12.3 12.4 12.5

Demographic Characteristics of Interview Participants Normalized Figures for Clauses Including Enhancement, Maintenance and Attack of Face or Sociality Rights Top 20 Keywords, Ranked by Log Ratio Top 20 Collocates of ‘Insulin’ (5L > 5R), Ranked by MI3 PND Corpus Data Strong Collocations of ‘PND’ in the Three Data Sets Top 30 Lemma Collocates of ‘PND’ by Category in the MUMSNET Corpus Top 30 Lemma Collocates of ‘PND’ by Category in the MED-LAY Corpus Top 30 Lemma Collocates of ‘PND’ by Category in the MEDIA Corpus Frequencies of MPQ Descriptors in the OEC, in Descending Order MPQ Descriptors and Closely Related Words in the New-GLS Collocates of MPQ Descriptors in the OEC Seven MPQ Descriptors and Their Top 50 Collocates that Refer to Physical Damage and Illness Eighteen MPQ Descriptors and Their Top 50 Collocates that Refer to Body Parts Pearson Coefficients for Correlation between the Probability that Each Descriptor Is Chosen and the Strength of the Collocation between the Descriptors in Each MPQ Group and the Word ‘Pain’: Sensory Groups with Coefficients >0.9 Empathy Displays in Relation to Posting Overview of data Narrative functions according to sub-corpora, presented in order of likely sequence of occurrence Metaphor Keywords Frequency of Metaphor Use, by Newspaper Examples of Clauses Coded According to Semantic Roles and Concepts The Seven Most Frequently Occurring Agents Survey Questions Keyword List Based on ‘Keyness’ (Generation Method Log-Likelihood) Top 20 Collocates for ‘Lack of ’ Palliative Care Terminology Results Uses of Palliative Care Terminology

27 29 53 56 80 83 85 89 92 105 108 121 111 112

115 167 190 208 226 229 248 253 329 330 334 339 340

CONTRIBUTORS

Dimitrinka Atanasova is Lecturer at Lancaster University. She is interested in health and science communication, and her research has focused on the topics of obesity, mental health, climate change and sustainability. Gavin Brookes is Senior Research Associate in the ESRC Centre for Corpus Approaches to Social Science in the Department of Linguistics and English Language at Lancaster University. His research interests include corpus linguistics, health communication, (critical) discourse studies and multimodality. He is Associate Editor of the International Journal of Corpus Linguistics (John Benjamins). Rachel Chimbwete-Phiri is Lecturer in Language and Communication Studies at the University of Malawi. Her research interests are identity, professional and medical communication. For her doctoral research, she is investigating how clients and health professionals at a rural community hospital in Malawi reproduce and negotiate their knowledge about HIV/AIDS. She has coauthored a journal article on humour in HIV/AIDS consultations in Malawi (with S. Schnurr), and book chapters on applied linguistics and health education discourse (with M. N. MacDonald) and ethnic identities and multiparty politics in Malawi (with E. Chimbwete and M. J. Boti-Phiri). Jeremy Dale is a general practitioner, health services researcher and professor of primary care with a long-standing interest in palliative and end of life care. His research publications include evaluations of different models of care, as well as primary care workforce issues and clinical decision making in general practice.

CONTRIBUTORS

xi

Zsófia Demjén is Associate Professor of Applied Linguistics at the UCL Centre for Applied Linguistics. Her research interests include health communication, metaphor and the intersections of language, mind and health(care). She is author of Sylvia Plath and the Language of Affective States: Written Discourse and the Experience of Depression (2015), co-author of Metaphor, Cancer and the End of Life: A Corpus-Based Study (2018), and co-editor of The Routledge Handbook of Metaphor and Language (2017). Her work has appeared in the Journal of Pragmatics, Applied Linguistics, Metaphor and the Social World, Communication & Medicine, the BMJ’s Medical Humanities and Discourse Studies among others. Dariusz Galasiński is a linguist and Professor at the Faculty of English of Adam Mickiewicz University in Poznan, Poland. His current research focuses on discursive aspects of suicide notes. His latest book is Discourses of Men’s Suicide Notes (2017). Andrew Hardie is Reader in linguistics at Lancaster University. His main research interests are the theory and methodology of corpus linguistics; the descriptive and theoretical study of grammar using corpus data; the languages of Asia; and applications of corpus methods in the humanities and social sciences. He is one of the lead developers of the Corpus Workbench software for indexing and analysing corpus data, and the creator of its online interface, CQPweb. He is co-author, with Tony McEnery, of the book Corpus Linguistics: Method, Theory and Practice (2012). Sylvia Jaworska is Associate Professor in Applied Linguistics in the Department of English Language and Applied Linguistics at the University of Reading. Her main research interests are discourse analysis and corpus linguistics and the application of both methods to study (new) media, health and business communication. She has published widely on these topics in Applied Linguistics, Journal of Pragmatics, International Journal of Corpus Linguistics, Corpora, Discourse & Society, Language in Society and International Journal of Business Communication. Karen Kinloch is a researcher in health communication and corpus linguistics at Lancaster University and Edge Hill University. Following on from an ESRC funded PhD on the comparative discourses around infertility across a range of text types, her research interests are in corpus-assisted discourse studies, health and identity, human reproduction, health and science communication, and perinatal mental health.

xii

CONTRIBUTORS

Nelya Koteyko is Reader in Applied Linguistics at Queen Mary University of London. Her recent research and publications have focused on social media discourses about diabetes, multimodality in health communication and media coverage of mental health. Joyce Lamerichs is Assistant Professor at the Department of Language, Literature and Communication at the VU University in Amsterdam. In her research, she combines insights from conversation analysis and discursive psychology to study the interactional characteristics of health encounters in different institutional settings, varying from end of life talk, psychological research interviews on trauma recovery and addiction counselling. She examines both face-to-face and online encounters and the interactional challenges health professionals face when using these different modes of communication. Mariana Lazzaro-Salazar is a member of the Centro de Estudios para la Integración Intercultural (CEII – UCM), a member of the Ethics Committee and a lecturer in the PhD Programmes of Education and Psychology at Universidad Católica del Maule, Chile. She is also a research associate of the Language in the Workplace Project, Victoria University of Wellington. Mariana’s research has focused on healthcare communication in intercultural contexts. Miriam A. Locher is Professor of the Linguistics of English in the English Department of the University of Basel, Switzerland. She works on interpersonal pragmatics, linguistic politeness, relational work, the exercise of power, disagreements, advice-giving (in health contexts) and computer-mediated communication. Her publications comprise monographs, edited collections and special issues as well as numerous articles in journals and collections. She has recently co-edited the special issue on Language and Health Online (Linguistics Online) and written a monograph on Reflective Writing in Medical Practice (Multilingual Matters). Joelle Loew is a PhD student at the University of Basel (CH) funded by the Hermann Paul School of Linguistics. She holds an MSc in Intercultural Communication from the Centre for Applied Linguistics at Warwick University and has worked as Research Associate at Warwick Medical School (UK). Her research interests are in professional and medical communication with a particular focus on the construction of identities in the workplace. Agnes Marszalek teaches academic writing and research skills at Glasgow International College. Her interests lie in the fields of stylistics, discourse analysis and pragmatics. Her PhD thesis, completed at the University of Glasgow (2016), focused on the language of written comic narratives, and she has published primarily on narrative humour. She is author of Style and

CONTRIBUTORS

xiii

Emotion in Comic Novels and Short Stories (forthcoming). In 2017, she was the research associate on the Power, Control and the Language of Voice-hearing project at the UCL Centre for Applied Linguistics. Catherine Millington-Sanders is RCGP & Marie Curie National Clinical Champion for End of Life Care and Co-Founder of the Social Enterprise Difficult Conversations. She is a practising GP, clinical commissioner and educationalist with seven years of experience as a specialty doctor in palliative medicine. Catherine supports Public Health approach to developing Compassionate Communities, actively implementing the Compassionate City Charter and facilitating the development of supportive networks within our communities. Sarah Mitchell is a practising GP and a National Institute of Health Research Doctoral research fellow at the University of Warwick, UK (2014–19), motivated to carry out research by experience in clinical practice. She has a number of clinical lead roles and works with policy makers at regional and national level. Gabriella Rundblad is Reader in Applied Linguistics at King’s College London. She is an expert on language and cognition, and has managed a number of research projects in the area of health communication, in particular around water safety. In her research, Dr Rundblad utilizes a wide range of methodologies, for example, psycholinguistic experiments, discourse analysis, surveys, focus groups and interviews. Marie-Thérèse Rudolf von Rohr teaches academic writing and EFL at the School of Engineering at the University of Applied Sciences and Arts Northwestern Switzerland. She completed her PhD in English linguistics at the University of Basel, where she investigated persuasive mechanisms from an interpersonal pragmatic perspective in public health discourse online (published with New Ideas in Human Interaction (NIHIN) in 2018). Her research interests are interpersonal pragmatics, discourse analysis and health interaction. She has been a research member of the SNSF-project Language and Health Online (1432869). Stephanie Schnurr is Associate Professor at the Centre for Applied Linguistics at the University of Warwick. Her main research interests are professional and medical communication – with a particular interest in leadership, gender, humour, identity construction and (im)politeness. Stephanie has published widely on all of these topics in internationally renowned journals. She is also the author of Language and Culture at Work (with O. Zayts, 2017), Exploring Professional Communication (2013), and Leadership Discourse at Work (2009); and she has co-edited Identity Struggles (with D. van de Mieroop, 2017) and Challenging Leadership Stereotypes through Discourse (with C. Ilie, 2017).

xiv

CONTRIBUTORS

Elena Semino is Professor of Linguistics and Verbal Art in the Department of Linguistics and English Language at Lancaster University, and Director of the ESRC Centre for Corpus Approaches to Social Science. She holds a visiting professorship at the University of Fuzhou in China. She specializes in stylistics, medical humanities, health communication, corpus linguistics, narratology, and metaphor theory and analysis. Rein Sikveland is a research associate at Loughborough University, UK. Rein’s expertise is in conversation analysis and phonetics. Rein studies the interactional management in professional conversations between the public and commercial, health and police services. His research is applied to the Conversation Analytic Role-play Method (CARM), and he is currently co-chief investigator on applying CARM to teacher education (CAiTE: Conversation Analytic innovation for Teacher Education). Elizabeth Stokoe is Professor of Social Interaction in the School of Social Sciences at Loughborough University, UK. Her current research interests are in conversation analysis, membership categorization and social interaction in various ordinary and institutional settings, including mediation, police interrogation, role-play and simulated interaction. She is the author of Talk: The Science of Conversation (2018). Wyke Stommel is Assistant Professor of Language and Communication at Radboud University, Nijmegen (Netherlands). In her research, she uses Conversation Analysis for the study of both spoken and mediated institutional interaction including chat, email and telephone counselling, face-to-face and video-mediated medical consultations and police interrogations. Wyke has cofounded the MOOD (Microanalysis of Online Data) network. Chris Tang is Lecturer in Applied Linguistics and International Education at King’s College London. His research interests centre on the development and use of applied linguistic methodology in public health communication, risk and disaster communication and language education contexts, particularly those situated in linguistically and culturally diverse environments. Franziska Thurnherr is a PhD candidate in English linguistics at the University of Basel. Her research focuses on interpersonal pragmatics in online (mental) health discourse. She researches health interaction especially with regard to relational work and identity construction and how these concepts influence the therapeutic alliance or doctor–patient relationships. She is further interested in an applied approach to health communication research. She has been a research member of the SNF-project Language and Health Online (143286) and has coedited the special issue on Language and Health Online (2017).

CONTRIBUTORS

xv

Jonathon Tomlinson has been a full-time NHS GP in Hackney, London, since 2001 with time out for voluntary work in Afghanistan in 2003–4. He has an MA in Human Values and Contemporary Global Ethics and a PGCert in Medical Education. He has special interests in the social determinants of health, patient advocacy and the relationships between doctors and patients. He is author of the blog A Better NHS (https://abetternhs.net) on how issues such as choice, kindness, shame and loneliness impact on patients and doctors. Filippo Varese is a clinical psychologist, a senior clinical lecturer at the University of Manchester and the director of the Complex Trauma and Resilience Research Unit within Greater Manchester Mental Health NHS Foundation Trust. His research interests focus on the psychological underpinnings of psychotic symptoms, and the impact of traumatic life experiences on the mental health and well-being of people with complex mental health needs. Katharine Weetman is a PhD student at the University of Warwick collaboratively funded by the ESRC and CCGs of South Warwickshire and Coventry & Rugby. Her teaching and research interests are in health communication with a current focus on discharge communication. Joanna Zakrzewska specialized in orofacial pain after obtaining dental (Kings College, London), medical (Cambridge) degrees and specialist training in oral medicine. She has led a multidisciplinary facial pain unit UCLH, London, and is now specializing in trigeminal neuralgia. She has written 4 books, 20 chapters and over 100 papers. Olga Zayts is Associate Professor at the School of English, the University of Hong Kong. She has published widely on various issues of intercultural professional and medical communication, including decision making, risk communication, informed consent, leadership, identity and gender. She co-authored Language and Culture at Work (with Stephanie Schnurr). Justyna Ziółkowska is a clinical psychologist and Associate Professor at the II Faculty of Psychology at the University of Humanities and Social Sciences in Poland. Her research interests concentrate on experiences of mental illness and discourses of psychiatric care. Her latest book is Samobójstwo. Analiza narracji osób po próbach samobójczych (Suicide. An Analysis of Suicide Attempters’ Narratives), 2016.

ACKNOWLEDGEMENTS

‘The Functions of Narrative Passages in Three Written Online Health Contexts’ by Franziska Thurnherr, Marie-Thérèse Rudolf von Rohr and Miriam A. Locher is reprinted here as Chapter 7 with the kind permission of De Gruyter Open Ltd. This collection would not have been possible without the dedication of all of its contributors. I am immensely grateful for the enthusiasm they brought to this project from the start, and for all their hard work, which included peerreviewing each other’s chapters in addition to their myriad other duties. I am especially thankful for their patience and understanding when things had to slow down for a little while. I’m grateful to Li Wei for encouraging me to actually be the one to get this project going, when I was still only thinking about how useful such a collection would be, and to my editors at Bloomsbury, Andrew Wardell and Becky Holland, as well as Gurdeep Mattu and Helen Saunders, for their support throughout the process.

Introduction ZS Ó FIA DEMJ É N

1 CONTEXT Most aspects of illness and healthcare are mediated by language (cf. Sarangi, 2012). This is true of experiences of and beliefs about illness, death and healthcare provision, which are talked and written about; it is true of diagnostic tools and processes as well as treatments, which are (at least partly) conducted, negotiated and discussed verbally; it is true of public health communications and medical education, where communication skills are explicitly honed; and it is, of course, true of interactions between various stakeholders in different healthcare settings, whether they are face-to-face, at a distance or online. These are all inherently linguistic in nature. How we talk about things, people, experiences and how we talk to, about and for each other have consequences for our relationships, our sense of self, our understanding of and reasoning about the subject at hand, our success at achieving our goals and indeed for our health (e.g. Dowell, Jones and Snadde, 2002). Yet linguistic analysis, both as a source of evidence and as a varied toolkit for making sense of the complexities of healthcare and the vast amount of verbal data now available, has been conspicuously absent from the mainstream of medical education, communication training and even medical humanities. The fact that communication plays a crucial role in healthcare contexts is not controversial and hasn’t been so for a long time. The 1950s and 1960s saw a growing number of often interdisciplinary studies explore how language and communication might shed light on the doctor–patient relationship, for example (see Collins, Peters and Watt, 2011 for an overview). While most of these studies did not fall within linguistics, there is also a long tradition of linguists working on communication especially in the context of doctor–patient interactions (e.g. Heritage and Stivers, 1999; Roberts et al., 2005). However,

2

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

the impact of this more linguistic work on healthcare practice has at times been somewhat limited (Collins, Peters and Watt, 2011; Roberts and Sarangi, 2003). The chapters in this volume address this disconnect by highlighting the practical implications of a range of linguistic methods and tools applied in different illness and healthcare contexts. Applied linguistics has as its central mission to connect ‘knowledge about language to decision-making in the real world’ (Simpson, 2011: 1) and, when appropriate, to investigate ‘real-world problems in which language is a central issue’ (Brumfit, 1995: 27). Its definition as an interdisciplinary field of research and practice dealing with practical problems of language and communication that can be identified, analysed or solved by applying available theories, methods and results of Linguistics or by developing new theoretical and methodological frameworks in Linguistics to work on these problems (AILA, 2017) embodies these aims. And while these objectives are not uncontroversial (cf. Candlin and Sarangi, 2004; Li Wei, 2014; Roberts and Sarangi, 1999; Sarangi, 2005), its focus on language’s role in practical issues and decision making makes it ideally suited for the investigation of healthcare contexts where language and communication can, when they work well, improve information-provision, self-esteem, support and even diagnosis and self-management but, when they work badly, can also be a source of confusion, isolation, anxiety, stigma and even misdiagnosis (Semino et al., 2018). This volume presents research from different approaches in linguistics and demonstrates how linguistic analyses can ●

●

●

●

●

●

●

improve understandings of the lived experience of different illnesses (Demjén, Marszalek, Semino and Varese; Kinloch and Jaworska), feed into communications training (Stommel and Lamerichs; Chimbwete-Phiri and Schnurr), contribute to illness prevention (Chimbwete-Phiri and Schnurr; Tang and Rundblad), improve and illuminate diagnostic categories (Semino, Hardie and Zakrzewska; Demjén, Marszalek, Semino and Varese), improve understandings of issues with self-management (Brookes; Thurnherr, Rudolf von Rohr and Locher; Atanasova and Koteyko), provide insights into problems with public health messaging (Atanasova and Koteyko; Tang and Rundblad), increase access to appropriate care (Sikveland and Stokoe; Loew, Mitchell, Weetman, Millington-Sanders and Dale),

INTRODUCTION

●

●

3

illuminate issues of professional mobility (Zayts and Lazzaro-Salazar) and explain the implications of professional terminology (Galasiński and Ziółkowska; Loew, Mitchell, Weetman, Millington-Sanders and Dale; Semino, Hardie and Zakrzewska).

This is applied linguistics at its best. Importantly, all insights are based on authentic data from actual language, interactions and practice, rather than hypothetical scenarios, role play or simulations (cf. Sarangi, 1994). The collection sets itself apart from other texts on this topic (e.g. de Silva Joyce, 2016; Galasiński, 2013; Gygax and Locher, 2015; Hamilton and Chou, 2014; Harvey and Koteyko, 2013; Pickering, Friginal and Staples, 2017; Locher, 2017; Rudolf von Rohr, 2018; Semino et al., 2018) in its range of linguistic approaches, in its focus on different healthcare contexts and, most importantly, in its emphasis of the practical implications of the current research included in each chapter. A number of chapters are also co-authored by healthcare practitioners, who, along with Tomlinson in the Epilogue, add much needed insights and understandings regarding the settings, pressures, constraints and practical considerations of contemporary healthcare. I completely agree with Tomlinson when he says that for linguistic health research to have the biggest impact, it needs to be grounded in such intimate knowledge of what is already known, what is assumed and what the immediate needs and issues are at a particular point in time. This is the difference between applying theoretical and methodological tools to a real-world setting and focusing on exactly how such a study will be practically relevant to professionals in that setting (Roberts and Sarangi, 2003). Finally, the volume also covers a range of international settings: United Kingdom, Netherlands, Hong Kong, Chile, Malawi and United States.

2 APPLYING LINGUISTICS It is a recurring theme of internet memes that a linguist, upon declaring their discipline, is asked: ‘So, how many languages do you speak?’ (see also Li Wei, 2014). There is often an assumption that communication is simple, transparent and neutral, if people know how to do it ‘properly’, and that therefore there isn’t anything much beyond additional languages to study. It takes quite some explanation to convince people that there is more to language use than immediately meets the eye (or ear). The analysis that linguists do, particularly those focusing on how language is used in real-world texts and interactions, is designed to get below the surface of what is said by focusing on how people choose to put things into words. As Chafe explains, …think of linguistic form as if it were located in a pane of glass through which ideas are transmitted from speaker to listener. Under ordinary

4

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

circumstances language users are not conscious of the glass itself, but only the ideas that pass through it. The form of language [seems] transparent, and it takes a special act of will to focus on the glass and not the ideas. Linguists undergo a training that teaches them how to focus on the glass, but fluent users of a language focus their consciousness only on what they are saying. People use language to organise and communicate ideas without being at all conscious of how their language does it. It is undoubtedly this [apparent] transparency of language that makes it so difficult for more people to understand why language should have a science devoted to it. Still, there are many aspects of language a person can learn to be conscious of. Linguists do that professionally, and the experience of becoming conscious of previously unconscious phenomena is one of the principal joys of linguistic work. (1994: 38, my additions in square brackets) While this explanation somewhat simplifies the communicative process, relying on the so-called conduit model (cf. Reddy, 1979), it nevertheless makes clear that linguistic analysis is very different from simply analysing content or themes (what is said). To a linguist, language is never simple or transparent or neutral, and the theoretical, methodological and analytical tools they use surface and deconstruct this complexity. Linguistic analysis relies on systematic, replicable and theoretically based methods of looking at what choices (e.g. in pronouns, metaphors, grammatical form, etc.) are made (consciously or not) in contrast with other choices that could have been made, how such choices pattern systematically, and what the implications might be. In short, linguistic analysis encourages us to think about the seemingly obvious and rediscover it as profound (cf. Skelton, 2002); it exposes ‘beliefs and practices that might be taken for granted or overlooked altogether’ (Harvey and Koteyko, 2013: 2). As such, it has a particularly important role to play in exploring, understanding and improving what goes on in healthcare contexts (cf. Hunt and Carter, 2011; Sarangi, 2005). With its range of methodologies, applied linguistics straddles the humanities and (social) sciences and can operate in the paradigms and frameworks of both – in fact, it can often be used to bridge the gap (Crawford, Brown and Harvey, 2014). This is particularly important in illness and healthcare contexts, where intensely subjective, personal, human experiences co-exist, and sometimes collide, with the more abstract, impersonal and scientific world of biomedicine. The emergence of fields such as medical humanities and narrative medicine (e.g. Charon, 2006; Greenhalgh and Hurwitz, 1999) speak to this disconnect, as well as to the recognition that both are essential for improvement and progress (Greenhalgh, 2016). However, even the medical humanities and narrative medicine, while adopted and appreciated by many, have a hard time bringing the human and the biomedical together on a large scale. The chapters in this volume demonstrate that linguistics is a useful partner in this endeavour:

INTRODUCTION

5

it has robust and reliable tools that allow it to make sense of data on a large, statistically significant, scale (cf. Crawford, Brown and Harvey, 2014), while also treating people as individuals and taking their words and stories seriously. Thus, it provides the conceptual and methodological glue to connect lived experiences, practices and texts with medical science. 2.1 Methods and tools All chapters in this volume focus on how language is used in particular realworld texts or interactions, to what end and why (always bearing the specifics of the context in mind). This broadly falls within the discourse analytic tradition of applied linguistic analysis. In fact, a number of chapters refer to their approach as a type of discourse analysis. ‘Discourse analysis’ (DA) in this sense refers to a collection of different tools and techniques to make sense of social life through language (Potter and Wetherell, 1987). The chapters in this volume, for example, explore different types of language data (face-to-face, telephone, email and online conversations, technical definitions, research interviews, media articles, diagnostic tools, open-ended responses to questionnaires), using a range of specific methods: ●

●

●

●

●

●

●

General, cognitive or critical discourse analysis: these focus on the content and form of stretches of text (semantics, syntax and function) and look at the social actions that are accomplished (general DA); the relationship with power and ideology (critical DA); the mental representations of information and experience formed or triggered by language (cognitive DA). Conversation analysis (CA), micro-analysis of online data: these focus on how the organization of talk (CA) or online or mediated interactions (micro-analysis of online data) help to achieve social goals. Interactional sociolinguistics: focuses on the micro details of an interaction as well as contextual background in which the encounter takes place to examine how social goals are achieved. Narrative analysis: focuses on how narratives are constructed and performed to achieve interpersonal and discourse goals. Rapport management, relational work and (im)politeness theory: these focus on how people use language to manage competing wants and needs among each other while maintaining (or disrupting) social harmony. Critical metaphor analysis: focuses on metaphorically used words or expressions to raise awareness of their underlying or implied meanings and their implications with regard to ideology and power. Corpus analysis: this is in itself is a collection of tools for making sense of large quantities of data.

6

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Although not an exhaustive toolkit, the variety of methods used in this volume showcases the possibilities that linguistics can offer,1 including several quantitative techniques, which can speed up different kinds of analyses, cope with large amounts of data and increase systematicity, replicability and reliability – a particular advantage in the age of big (and ever increasing) data. At the same time, the methods are not mutually exclusive and none of them are applicable only to the contexts in which they are demonstrated. A number of chapters draw on more than one analytical tool to make the most of their data and all tools are applicable to other healthcare contexts as well.

3 OVERVIEW OF THE BOOK The volume is organized into four parts, representing crucial roles language can play in healthcare contexts. Part I, ‘The Experience of Illness’, focuses on how patients use language (or how they do not use it, in the case of Chapter 4) to describe their symptoms and experiences of illness among themselves or to researchers and healthcare professionals. Chapter 1: Focusing on (im)politeness phenomena in reported interactions between people who hear voices and their voices, Demjén, Marszalek, Semino and Varese examine the relationship between voice and hearer, including relative power, as constructed through language. They link different patterns of (im)politeness, especially on the part of the voice, to different levels of distress experienced by voice-hearers. The authors argue that impoliteness in particular can be one of the ways in which the voices exercise control over the hearers, thereby limiting how they can live their lives. Demjén et al. propose that it might be possible to alter distressing, antagonistic relationships between voices and hearers by changing the way voice and hearer relate to each other linguistically. Chapter 2: Given the number of chapters in this volume that use quantitative corpus methods to make sense of large data sets, Brookes provides a step-by-step introduction to how such an analysis might be conducted and what researchers must consider in the process. Analytic steps and decisions are demonstrated on a data set of online forum contributions around the topic of diabulimia, a contested eating disorder. Brookes uses three standard corpus techniques (keywords, collocations and concordances) to explore how diabulimia, in particular the restriction of insulin intake, is constructed by those with the condition. He shows that insulin restriction is mostly negatively evaluated in responses from other group members, even in a diabulimia support group. This, however, means that most of the contributors who describe restricting their insulin do so in relatively non-specific ways, making their practices more difficult the challenge.

INTRODUCTION

7

Chapter 3: Examining and comparing the collocates of ‘PND’ (i.e. postnatal depression) across three data sets, Kinloch and Jaworska discuss the ways in which mothers on Mumsnet manage and justify experiences of perinatal mental distress and/or PND. They show how distress as well as the diagnosed disease itself are associated with internalized stigma, which needs to be discursively managed, or mitigated. They track where this stigma might stem from by comparing Mumsnet data to media reports and patient information documents on PND. They suggest that healthcare practitioners need to be more aware of the social pressures and aetiological uncertainties that stop mothers recognizing or accepting they may have PND and from seeking help. Chapter 4: Semino, Hardie and Zakrzewska use corpus linguistic tools differently to other chapters in this volume. Their focus is on the linguistic descriptors included in the McGill Pain Questionnaire (MPQ), but rather than analysing their data text (i.e. the questionnaire) with corpus tools, they instead use these tools to explore the usage of the linguistic descriptors in general English. In this way, they are able to show that the seventy-eight descriptors in the MPQ vary greatly in terms of frequency in everyday English (some also being associated with specialist genres). They are therefore not equally likely to be familiar to the patients responding to the questionnaire. Other descriptors are only rarely, if at all, used in the context of pain and may therefore be avoided by patients. By way of a pilot, the authors test whether these aspects might correlate with patient choices of descriptors and finds that this is in fact the case. Part II, ‘Relating to Each Other’, focuses on how the rapport and relationships so crucial for health and effective healthcare are built and negotiated in different healthcare contexts. Chapter 5: Using interactional sociolinguistics, Chimbwete-Phiri and Schnurr unpick the language used by HIV/AIDS counsellors in a prevention clinic in Malawi that has particularly high adherence/attendance rates. They show that the particular ways in which advice and questions are phrased – how the counsellors say what they say – are likely to be contributing factors to the clinic’s success. They draw out the discourse strategies that are most successful – using questions, drawing on local knowledge, metaphors and narratives – and suggest that these could form the basis of a kind of best practice in similar group counselling settings. Chapter 6: Stommel and Lamerichs use a micro-analytic process to document the different strategies that counsellors use in online chat counselling for displaying empathy, to better understand the ways in which such interactions work. They contrast the actual practice of online counselling, with guidelines that advise explicit empathy displays, showing that these might be less important in themselves in the online context. They argue that empathy has to be performed in a way that is appropriate to the affordances of the medium in

8

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

which counselling takes place and that guidelines therefore need to draw more from current successful counselling practice. Chapter 7: Thurnherr, Rudolf von Rohr and Locher explore the functions that narratives perform in three healthcare contexts. Their chapter emphasizes that narratives always fulfil multiple functions simultaneously and cannot therefore be taken at face value. There is no question that narratives play a crucial role in how patients organize, understand and communicate their experiences and that these have clinical value, but the authors argue that the multiple other functions that narratives serve in any given social context influence what is being said and how. In particular, the authors focus on the relational work that narratives do, that is, the interpersonal goals they are used to achieve (e.g. positioning oneself and one’s audiences in particular ways). These goals influence how narratives are told and therefore how they should be understood. Part III, ‘Illness in the Mass Media’, explores language in the context of public communications, where the diffuse nature of the target audience and the framing function of language make getting the message right particularly difficult, and particularly important. Chapter 8: Using critical metaphor analysis, Atanasova and Koteyko expose how news reporting on obesity typically uses War metaphors and show that the ways in which these metaphors frame the issue can contribute to stigma and unfavourable views of obese individuals. For example, War metaphors typically require that there is an enemy to be fought. However, in the case of obesity (as with other chronic conditions), there is no obvious external entity such as a virus: the enemy to be fought are the patients themselves. This leads to a kind of othering that may explain why obese individuals increasingly report feeling stigmatized and receiving unequal treatment in clinical encounters. Chapter 9: Tang and Rundblad use cognitive DA to explore media reporting on the presence and risks of contaminants in drinking water as an example of a ‘health scare’. They show how the media represented the health risks differently from official sources (water industry, governments) and how such skewed representations – in particular of who is responsible for what kinds of actions and what uncertainties exist – can prompt behaviours that have unfavourable outcomes. In the case of potential contaminants in drinking water this may simply be undue anxiety, but in other cases such as the MMR vaccine controversy, it could lead to an increase in measles outbreaks. While their recommendations – for greater explicitness about what exactly is uncertain and who is responsible for different actions – are particularly important for public health communication (with its wide audience) about health risks, it also applies on the more micro level of healthcare professional to patient communication. Finally, Part IV, ‘Professional Practices and Concerns’, deals with the impact that language used by professionals working in healthcare contexts can have on their own practices as well as on their patients.

INTRODUCTION

9

Chapter 10: Sikveland and Stokoe show how CA can be used to demonstrate and train a more patient-centred approach to triaging in General Practice surgeries. They analyse telephone interactions between patients and receptionists and show clearly that when an institutionally relevant category such as ‘routine appointment’ (and in some cases ‘urgent’) is presented explicitly to patients calling for appointments, it tends to elicit resistance on the part of the patient and results in a more fraught triaging process. Instead, they suggest that in most cases patients automatically give clues about the urgency of their request in their opening turns and therefore do not require an institutionally prescribed and worded triage process. Chapter 11: Zayts and Lazzaro-Salazar use the tools of interactional sociolinguistics to examine how migrant medical professionals reflect on the intersections of healthcare systems, cultures and practices, and the challenges they face when transitioning to work in a country other than the one they trained in. Their chapter touches on issues of professional mobility as well as multicultural communication and provides a more nuanced understanding of (cultural and linguistic) adjustment and adaptation processes. The authors show that, contrary to much discourse-oriented research, semi-structured interviews can be useful data for linguistic analysis as long as they are treated not as de facto representations of the interviewee’s thoughts, but as social processes in their own right, in which social actions are accomplished, and stances and identities are negotiated. Chapter 12: Loew, Mitchell, Weetman, Millington-Sanders and Dale outline known barriers to the provision of good end of life care such as fractured continuity of care and inadequate support for patients and their families. Using corpus-based DA, they re-analyse existing data to show that there is a lack of clarity around key terminology such as ‘end of life’, ‘palliative’ and ‘hospice’ and that this may contribute to fractured continuity of care. In addition, they find that GPs construct barriers to good PEOLC in terms of an absence (what there is not) of various abstract entities such as resources and time. In this way, they (a) emphasize the importance of these (they maintain focus on absences and do not discuss enablers) and (b) construct barriers in a way that often backgrounds any agency, including their own. Yet, while structural issues are undoubtedly important, the authors argue that more could be done on the part of GPs too, especially when it comes to preparing patients and their families for the end of life. Chapter 13: Galasiński and Ziółkowska use the tools of critical DA to critique standard definitions of suicide used in research and healthcare practice. Examining thirty definitions, they show how these systematically construct suicide as an abstract act, without temporality, and exclude or background those who take their lives. The authors argue that these patterns have implications for how suicide is researched (e.g. prioritizing measurables rather than the lived

10

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

experience), how policies are established and indeed how people are treated. They propose an alternative template for formulating definitions of suicide. The final chapter of the book, the Epilogue, reflects on the role of language in context of primary care in the UK. It highlights some of the points raised, insights gained and questions yet to be explored in the preceding research chapters from the perspective of a general practitioner. 3.1 Intersections While the organization of the volume imposes one particular order on its chapters, there are a number of intersections between different parts and chapters and the collection covers a multitude of illnesses and settings, as well as types of language data. This reflects the increasingly dynamic, complex and interacting contexts within which illness and healthcare are situated and the different communicative practices among and between its range of stakeholders – healthcare professionals, patients, families/carers, medical researchers, journalists and more. These communicative practices go well beyond doctor– patient encounters and include mass media reporting, public communications and increasingly interactions in online patient communities (e.g. Harvey and Koteyko, 2013). As Tomlinson argues in the Epilogue, face-to-face consultations are being displaced and speech is being replaced by text with online forms, forums and emails. For this reason, the chapters in this volume similarly move beyond a focus on healthcare professional–patient interactions, which has traditionally been the staple of health communications research (Ong et al., 1995; Sarangi and Roberts, 1999). Such interactions are clearly important and are the focus of Chapter 3. Equally important, however, are chapters on dialogic communication between healthcare practitioners and patients mediated by technology (Chapter 6) and chapters on mediated communication involving less prototypical professionals such as receptionists (Chapter 10), or not involving healthcare professionals at all, but taking place between peers (Chapters 2, 3 and 7). There are chapters looking at research interviews with healthcare practitioners (Chapter 11) or patients (Chapter 1) and ones focusing on questionnaire responses from doctors (Chapter 12). Several chapters explore monologic communication with the public (Chapters 8 and 9), or with healthcare practitioners and academics (Chapters 4 and 13). This reflects the increasingly diverse contexts and settings within which communication related to health and illness takes place and the wider range of participants playing important roles in healthcare interactions. As noted above, interactions taking place in computer or telecommunications mediated environments, in particular, are increasingly used in healthcare (Harvey and Koteyko, 2013), and this kind of data in large quantities is also more readily available for linguistic analysis (Lupton, 2017). This explains

INTRODUCTION

11

the prevalence quantitative corpus methods applied in various chapters of this volume, whether used alone or in combination with general DA. On the one hand, such ‘big data’ represents a fantastic opportunity and with more data comes more reliability and generalizability. On the other hand, manual, qualitative linguistic analysis is extremely time consuming and therefore difficult to use with extensive data sets. The techniques of corpus linguistics enable researchers to combine the two: statistical reliability and in-depth analysis. Chapter 2 provides an introduction to corpus analysis, which Chapters 3, 4 and 12 build on while also each introducing the specific corpus tools they use. In addition to the above, there is an even split between chapters looking at mental health (Chapters 1, 2, 3, 6, 13) and physical health (Chapters 4, 5, 8, 12, 13), in so far as the two can be considered separate. There are chapters on General Practice (Chapters 10 and 12), specialist clinics (Chapters 4 and 5) and general health and well-being (Chapters 7, 8, 9, 11). There are chapters on treatment or management of illness (Chapters 1, 2, 3, 6, 7, 12), prevention of illness (Chapters 5, 9), and also on the processes surround these such as professional (Chapter 11 and 13) or public attitudes (Chapter 8), gaining access to care (Chapter 10), and diagnostic processes (Chapter 4). 3.2 How to read this volume The collection has been put together with different audiences in mind: (applied) linguists, medical humanities scholars, as well as medical practitioners and educators. To cater for our non-linguistic audiences, authors have provided introductions to the various methods they use in a way that does not assume any previous knowledge. There are clearly marked summaries of the key implications of their findings for healthcare practice at the end of their chapters, which might be most interesting to healthcare practitioners and educators. Medical humanities scholars may find the chapters on lived-experience accounts and illness narratives most enriching, while perhaps also being interested in the range of methods linguistics can offer. Finally, students and practitioners of linguistics will likely want to focus on the overviews of the healthcare contexts and details of the original research showcased in each chapter, perhaps skipping introductions to methods with which they are already familiar.

4 CONCLUDING REMARKS As some of our most significant interactions will take place in a healthcare context, the importance of better communication cannot be overstated. Better communication means not just using different words to say the same things but better understanding how language functions in different contexts.

12

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Linguistics is a psychosocially aware, yet empirical method, with an appreciation for the cultural, contextual and personal pluralities of experience at its core. Of course, as Collins, Peters and Watt (2011) point out, linguistics can learn a lot from healthcare research, and Tomlinson’s Epilogue offers pointers in this vein for future research. They also point out, however, that linguistics has a huge amount to offer healthcare – something that this volume makes explicit in the range of insights it provides.

NOTE 1. The range of approaches is limited to verbal communication: not because other modes of communication, such as the visual or kinetic, are considered unimportant, but because such approaches, to be taken seriously, deserve volumes in their own right.

REFERENCES AILA (Association Internationale de Linguistique Appliquée). Available online: http://www.aila.info/en/about.html (Accessed December 2017). Brumfit, C. J. (1995), ‘Teacher professionalism and research’, in G. Cook and B. Seidlhofer (eds), Principle and Practice in Applied Linguistics, Oxford: Oxford University Press. Candlin, C. N. and Sarangi, S. (2004), ‘Making applied linguistics matter’, Journal of Applied Linguistics, 1 (1): 1–8. Chafe, W. (1994), Discourse, Consciousness, and Time: The Flow and Displacement of Conscious Experience in Speaking and Writing, Chicago: University of Chicago Press. Charon, R. (2006), Narrative Medicine: Honoring the Stories of Illness, Oxford: Oxford University Press. Collins, S., Peters, S. and Watt, I. (2011), ‘Medical communication’, in J. Simpson (ed.), The Routledge Handbook of Applied Linguistics, 97–110, Abingdon: Routledge. Crawford, P., Brown, B. and Harvey, K. (2014), ‘Corpus linguistics and evidencebased health communication’, in H. Hamilton and W. S. Chou (eds), Routledge Handbook of Language and Health Communication, 75–90, New York: Routledge. de Silva Joyce, H. (2016), Language at Work: Analysing Language Use in Work, Education, Medical and Museum Contexts, Newcastle: Cambridge Scholars Publishing. Dowell, J., Jones, A. and Snadden, D. (2002), ‘Exploring medication use to seek concordance with “non-adherent” patients: A qualitative study’, British Journal of General Practice, 52: 24–32. Galasiński, D. (2013), Fathers, Fatherhood and Mental Illness: A Discourse Analysis of Rejection, Basingstoke: Palgrave. Greenhalgh, T. (2016), ‘Cultural contexts of health: The use of narrative research in the health sector’, Health Evidence Network (HEN) Synthesis Report 49, Copenhagen: WHO Regional Office for Europe.

INTRODUCTION

13

Greenhalgh, T. and Hurwitz, B. (1999), ‘Why study narrative?’ British Medical Journal, 318: 48–50. Gygax, F. and Locher, M. (2015), Narrative Matters in Medical Contexts across Disciplines, Amsterdam: Benjamins. Hamilton, H. and Chou, W. S. (2014), The Routledge Handbook of Language and Health Communication, Abingdon: Routledge. Harvey, K. and Koteyko, N., eds (2013), Exploring Health Communication Language in Action, Abingdon: Routledge. Heritage, J. and Stivers, T. (1999), ‘Online commentary in acute medical visits: A method of shaping patient expectations’, Social Science and Medicine, 49: 1501–17. Hunt, D. and Carter, R. (2011), ‘Seeing through The Bell Jar: Investigating linguistic patterns of psychological disorder’, Journal of Medical Humanities Journal of Medical Humanities, 33 (1): 27–39. Locher, M. A. (2017), Reflective Writing in Medical Practice: A Linguistic Perspective, Bristol: Multilingual Matters. Lupton, D. (2017), Digital Health: Critical and Cross-Disciplinary Perspectives, London: Routledge. Ong, L. M. L., de Haesa, C. J. M., Hoosa, A. M. and Lammes, F. B. (1995), ‘Doctorpatient communication: A review of the literature’, Social Science and Medicine, 40 (7): 903–18. Pickering, L., Friginal, E. and Staples, S. (2017), Talking at Work: Corpus-based Explorations of Workplace Discourse, Basingstoke: Palgrave. Potter, J. and Wetherell, M. (1987), Discourse and Social Psychology: Beyond Attitudes and Behaviour, London: Sage. Reddy, M. J. (1979), ‘The conduit metaphor: A case of frame conflict in our language about language’, in A. Ortony (ed.), Metaphor and Thought, 284–310, Cambridge: Cambridge University Press. Roberts, C. and Sarangi, S. (1999), ‘Hybridity in gatekeeping discourse: Issues of practical relevance for the researcher’, in S. Sarangi and C. Roberts (eds), Talk, Work and Institutional Order: Discourse in Medical, Mediation and Management Settings, 363–90, Berlin: Mouton de Gruyter. Roberts, C. and Sarangi, S. (2003), ‘Uptake of discourse research in interprofessional settings: Reporting from medical consultancy’, Applied Linguistics, 24 (3): 338–59. Roberts, C., Moss, B., Wass, V., Sarangi, S. and Jones, R. (2005), ‘Misunderstandings: A qualitative study of primary care consultations in multilingual settings, and educational implications’, Medical Education, 39 (5): 465–75. Rudolf von Rohr, M.-T. (2018), Persuasion in Smoking Cessation Online: An Interpersonal Pragmatics Perspective, Freiburg i. Br.: Albrecht-Ludwigs-Universität Freiburg. Available online: https://freidok.uni-freiburg.de/fedora/objects/freidok: 16755/datastreams/FILE1/content. Sarangi, S. (1994), ‘Intercultural or not? Beyond celebration of cultural differences in miscommunication analysis’, Pragmatics, 4 (3): 409–27. Sarangi, S. (2005), ‘The conditions and consequences of professional discourse studies’, Journal of Applied Linguistics, 2 (3): 371–94. Sarangi, S. (2012), ‘Practising discourse analysis in healthcare settings’, in I. Bourgeault, R. Dingwall and R. De Vries (eds), The SAGE Handbook of Qualitative Methods in Health Research, 397–416, London: Sage.

14

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Sarangi, S. and Roberts, C. (1999), ‘Introduction: Discursive hybridity in medical work’, in S. Sarangi and C. Roberts (eds), Talk, Work and Institutional Order: Discourse in Medical, Mediation and Management Settings, 61–74, Berlin: Mouton de Gruyter. Semino, E., Demjén, Z. Hardie, A., Payne, S. and Rayson, P. (2018), Metaphor, Cancer, and the End of Life: A Corpus-Based Study, New York: Routledge. Simpson, J. (2011), The Routledge Handbook of Applied Linguistics, Abingdon: Routledge. Skelton, J. (2002), ‘Commentary: Understanding conversation’, BMJ, 325: 1151. Wei, Li (2014), ‘Introducing applied linguistics’, in Li Wei (ed.), Applied Linguistics, Chichester: Wiley-Blackwell.

PART I

The experience of illness

CHAPTER ONE

‘One gives bad compliments about me, and the other one is telling me to do things’ – (Im)politeness and power in reported interactions between voice-hearers and their voices1 ZS Ó FIA DEMJ É N , AGNES MARSZALEK , ELENA SEMINO AND FILIPPO VARESE

1 INTRODUCTION Hearing voices, also known as auditory verbal hallucinations (AVHs), involves the perception of verbal content in the absence of an appropriate external stimulus. While AVHs can occur for people without a mental health condition,

18

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

they are seen as a characteristic symptom of schizophrenia-spectrum disorders, with approximately 70 per cent of individuals with such diagnoses reporting hearing voices (McCarthy-Jones, 2012). AVHs also cut across diagnostic groups, as voice-hearing can occur in bipolar disorder, depression, post-traumatic stress disorder, dissociative disorders and obsessive compulsive disorders, among others. While for some voice-hearers these experiences are a source of extreme distress and impairment, leading to a need for care, the actual phenomenology of voice-hearing is extremely heterogeneous, in terms of both form (sensory and conversational qualities, such as loudness, pitch) and content, with variations in terms of their affective impact. As a result, a sizable minority of voice-hearers cope well with their voices (Jenner et al., 2008). This is because distress is generally not caused by the mere presence of voices, but depends on a number of factors, including what the voices say, and how; the relationship that voicehearers establish with their voices; how they make sense of their voices; their perceived control over the voices; and their ability to live the life they want to live, particularly in relation to control over their goals in life (e.g. Mawson et al., 2010; Varese et al., 2017). In this chapter, we present insights from a pilot study exploring what a particular linguistic approach, namely the study of (im)politeness in interaction, can contribute to understanding the relationships between voice-hearers and their voices, and how these might relate to issues of power, control and therefore distress. Factors determining the degree of distress that voices cause are, of course, attended to in clinical psychological research and practice. Interventions often aim to bring about change in power and control dynamics, for example, by increasing the hearers’ coping and perceived control, changing the relationship with voices, challenging beliefs about the power of voices, etc. In fact, assessments of voice-related distress and the determinants of distress are often generally language-based (e.g. clinical interviews such as the auditory hallucinations subscale of the Psychotic Symptom Rating Scales, or PSYRATS-AH, Haddock et al., 1999). However, linguistic analysis is not typically used to explore individuals’ descriptions of their experiences, and most current approaches to understanding and treating distressing voices are limited in a number of ways. First, they involve methods that capture explicit processes, that is, questionnaires, interviews and psychometric assessments (e.g. PSYRATS) that rely both on voicehearers’ conscious awareness of the nature of their relationships with voices and their willingness to disclose them. Second, current approaches often lack a consideration of the phenomenology or ‘lived experience’ of voice-hearing (Thomas et al., 2014; Woods et al., 2014). Finally, the ways in which the relationship between voice and hearer is established and maintained (including power dynamics related to control) remains poorly understood.

(IM)POLITENESS AND POWER IN VOICE-HEARING

19

We begin to address these shortcomings here by exploring how ten voicehearers with schizophrenia-spectrum diagnoses use language to describe their interactions with their voices, in the course of interviews with one of the authors (FV) (see Varese et al., 2016; Varese et al., 2017; Demjén et al., 2019). This means that we take descriptions of the lived experience of interacting with voices as our starting point. Linguistic analysis generally, as the chapters in this volume demonstrate, relies on systematic and theoretically based ways of investigating people’s linguistic choices when they describe their experiences. In very general terms, this involves looking at what linguistic choices are made (consciously or not) in contrast with other choices that could have been made, how such choices pattern systematically and what the implications of these choices might be. This makes it possible to develop an approach to the phenomenology of voice-hearing that also includes implicit processes, that is, processes that are less amenable to conscious monitoring and manipulation. Evidence from other areas of the psychological sciences (e.g. experimental psychopathology) suggests that such implicit processes can be influential determinants of mental health and well-being (Franck, de Raedt and de Houwer, 2007). In linguistics, relationships between individuals, including power dynamics, are known to be reflected, negotiated, maintained and challenged in interaction (see also Brookes; Chimbwete-Phiri and Schnurr; Kinloch and Jaworska; Stommel and Lamerichs; Thurnherr et al.; Zayts and Lazzaro-Salazar, all in this volume). We assume that this, to an extent, also applies to interactions between a person who hears voices and the voices that they hear, as in the following extract from our data: Excerpt 1 it could be one minute past midnight, a brand new day and all of a sudden the voices will say “you worthless bastard”. I am used to it now and they’ll say “you worthless bastard and it’s another day and we are gonna kick the fucking shit out of your head” and then they start on you2 then. They call you all sorts of things “useless, pathetic, hopeless, fat, evil, nasty bastard” and all that lot and they swear at you and they say “where’s your fucking nurses now to support you and your mental health team they don’t give a fuck about you” they say “why don’t you teach the bastards a lesson and kill yourself blow the flat up, set yourself on fire, go out there later on today and walk under a bus” (Participant 6, our emphases) Interactions between voices and voice-hearers are clearly not the prototypical kinds of interactions that linguists normally analyse. Analysts have no direct access to these interactions except as reported by the voice-hearer. Moreover,

20

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

information about voices’ identity, status or context is mediated through the hearer, or not available at all. As with reports of interactions more generally, it is highly likely that these accounts are not complete and precise records of all that the voices (or hearers) say, but simply a recollection of a particular selection deemed relevant by the hearer as a response to, in this case, research interview questions. This restricts the specificity with which linguistic frameworks can be applied. Nevertheless, voice-hearers’ reports of interactions with their voices, as in Excerpt 1, are the closest we can get to the lived experience of those interactions, and should be taken seriously as people’s attempt to share those experiences in the context of the interviews we conducted. We therefore take voice-hearers’ reports at face value, and apply to them linguistic tools known to surface relationships and power dynamics. One set of such tools includes various frameworks concerned with rapport management and linguistic (im)politeness in pragmatics (e.g. Brown and Levinson, 1987; Locher and Watts, 2005; Spencer-Oatey, 2008; Culpeper, 2011a), a sub-discipline of linguistics, as we will introduce in more detail later in this chapter. Broadly, these frameworks explore how language is used to maintain or disrupt social harmony as individuals go about achieving their (interactional) goals. Excerpt 1, for example, is rife with instances of language use (underlined) that appear to be designed to disrupt the possibility of a supportive and balanced relationship between Participant 6 and the voices he hears. The voices employ prototypical impoliteness formulae (such as insults and challenging questions; see later in this chapter for more detail) to achieve this disruption. The mere fact that it is exclusively the voices producing these utterances, not the hearer, begins to point towards the nature of the relationship and power dynamic at play. The particular linguistic expressions that are attributed to the voices provide further evidence to this effect. In this chapter, we therefore explore the extent to which concepts and ideas from research in pragmatics may contribute new insights into the relationship dynamics between voice-hearers and their voices, in particular where issues of relative power are concerned. As we will set out in this chapter, power in the linguistic sense is related to control both in the sense of perceived control over voices and people’s ability to live the life they want to live. In what follows, we begin with an overview of relevant linguistic literature on power and (im)politeness, providing an introduction to our methodology, before presenting our data and specific methods. This is followed by our analysis of patterns of (im)polite behaviour in interactions between voices and hearers, as reported by the latter. While we focus on a very specific context in our analysis, the concepts we introduce from (im)politeness theory will be at play in any interaction in healthcare contexts and can therefore be used to shed light on the dynamics at work in other healthcare relationships as well.

(IM)POLITENESS AND POWER IN VOICE-HEARING

21

2 INTRODUCTION TO THE METHODOLOGY: POWER, RAPPORT MANAGEMENT AND (IM)POLITENESS In this section, we introduce the components of the analytical framework that will be applied in our analysis of reported interactions with voices in our interviews. This analytical framework is formed of concepts drawn from linguistic approaches to power and interpersonal relationships in interaction, particularly from the branch of linguistics known as pragmatics. Research in pragmatics focuses on ‘meanings that arise from the use of communicative resources in context, and in particular, the meanings implied by speakers, inferred by hearers, and negotiated between them in interaction’ (Culpeper et al., 2018: 3). For reasons of space and accessibility, we cannot do justice to the debates associated with these concepts in the specialist literature, but these debates can be accessed via the studies we refer to in the course of this section. 2.1 Power Linguistic choices in communication are one of the key means to exercise power. They can be used to influence people and states of affairs, and to maintain or alter power relationships between individuals and groups (Locher, 2004). The investigation of power dynamics has at least two distinct pedigrees in linguistic research focusing on language in use: one sits within the framework of critical discourse analysis (CDA; see Galasińksi and Ziółkowska, this volume, for an overview) and the other in the sub-discipline of pragmatics. In brief, the former tends to focus on institutional, ideological, political and other forms of power embedded in texts and practices, while the latter tends to focus on interpersonal power dynamics within interactions. In the CDA tradition, Fairclough (1989) makes an important distinction between ‘power behind discourse’ and ‘power in discourse’. Power behind discourse is related to social and institutional settings in which people take on roles that have implications for who has power over whom (e.g. workplace hierarchies; parent–child relationships in families; expertise in discussions of scientific issues). These are relatively static and dependent on pre-existing settings or contexts. Power in discourse, on the other hand, is to do with how power relationships are enacted and negotiated through choices made in communication, for example, whether and how people address or refer to each other, tell each other what to do, criticize each other, etc. Power in and power behind discourse are seen as interconnected, since how one interacts with other people constructs and affects the power dynamics, even within institutional settings involving specific hierarchical constraints. But the fewer constraints there are behind discourse (as in the case of friendship groups), the greater the impact of the language used, because the absence of institutionally

22

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

established hierarchies leaves more scope for constructing and negotiating power relationships in discourse. The pragmatics tradition focuses on power as one aspect of the relationship between interlocutors. It is concerned both with how power behind discourse influences the choices that people (can) make in interaction, and with how these choices may themselves be used to construct and negotiate power relationships. Power is therefore not seen as static, but rather as highly dynamic, fluid and negotiated in interaction (Locher and Bousefield, 2008). 2.2 Rapport management and (im)politeness Power is one of several aspects of social relationships that make it a rather delicate matter to decide what to say and how to say it in interaction with other people. Pragmatics highlights in particular the balancing act we often have to engage in between achieving our personal or professional goals on the one hand and maintaining social harmony on the other. For example, asking a friend to borrow their car for the weekend involves an imposition on that person, and providing critical feedback to a work colleague has the potential to cause offence. There are, however, a range of linguistic strategies at our disposal to try to protect the personal and professional relationships with the people involved, while still achieving the goals of asking for the favour or delivering the feedback. For example, one might check whether the friend needs the car that weekend before asking to borrow it, or tell the colleague that their performance is ‘in need of improvement’ rather than ‘unsatisfactory’. In contrast, there are also circumstances in which interlocutors make linguistic choices that directly undermine social harmony, as when insults are exchanged among neighbours during long-standing disputes over noise levels or parking. Which goals we set ourselves in interaction, and how we attempt to achieve them via language, is closely connected with power relationships. For example, asking someone to speak more quietly is likely to be linguistically realized rather differently by a teacher speaking to their pupils versus an office worker speaking to a new colleague sitting at a nearby desk. And insults can sometimes be used to affect people’s self-esteem and control their behaviour, for example in abusive relationships. The pragmatics literature, broadly conceived, provides a set of concepts for making sense of these phenomena that are relevant to our data, including particularly ‘face’, ‘sociality rights’, ‘rapport management’, ‘relational work’ and ‘(im)politeness’ (Goffman, 1967, Brown and Levinson, 1987, Locher, 2004, Spencer-Oatey, 2008, Culpeper, 2011a, b). We will now introduce each of these concepts in turn. The concepts of face and sociality rights capture different aspects of people that need to be attended to in interaction to avoid jeopardizing social

(IM)POLITENESS AND POWER IN VOICE-HEARING

23

harmony. ‘Face’ is related to common-sense notions like reputation, prestige and self-esteem. In colloquial English, it is reflected in the idiom ‘to lose face’, which means ‘that one’s public image suffers some damage, often resulting in humiliation or embarrassment’ (Culpeper, 2011b: 398). In a more technical sense, ‘face’ is the positive social value a person effectively claims for himself [sic] by the line others assume he has taken during a particular contact. Face is an image of self delineated in terms of approved social attributes – albeit an image that others may share. (Goffman, 1967: 5) Brown and Levinson (1987) more specifically describe what they call ‘positive face’ as ‘the desire to be ratified, understood, approved of, liked or admired’ (Brown and Levinson, 1987: 62), and introduce the concept of ‘face threatening act’ for communicative acts that can go against this desire, such as criticism. Spencer-Oatey (2008) defines face as ‘people’s sense of worth, dignity, honour, reputation, competence and so on’ (Spencer-Oatey, 2008: 13), and breaks it down into three components: ●

●

●

Individual identity face (sometimes ‘quality face’): ‘a fundamental desire for people to evaluate us positively in terms of our personal qualities; e.g. our competence, abilities, appearance etc.’ (Spencer-Oatey, 2002: 540). Relational identity face: ‘the self-concept derived from connections and role relationships with significant others’ (Spencer-Oatey, 2007: 641). Social identity face: ‘the value that we effectively claim for ourselves in terms of social or group roles, and is closely associated with our sense of public worth.’ (Spencer-Oatey, 2002: 540).

To return to our previous examples, providing critical feedback to a colleague is problematic for social harmony because it undermines one or more aspects of their face, such as their individual identity face if their competence is criticized, or their relational identity face if they are told that colleagues do not want to work closely with them. The notion of ‘sociality rights’ is used by Spencer-Oatey (2008) to capture social expectations of appropriate behaviour, fairness, mutual consideration, contractual/legal agreement, etc. (as an extension and specification of what Brown and Levinson (1987) rather confusingly call ‘negative face’). Sociality rights can be broken down into ‘equity rights’ (being treated fairly, not being imposed upon or exploited) and association rights (being entitled to involvement and association with others) (Spencer-Oatey, 2008). Asking a friend to borrow their car is problematic for social harmony in terms of equity rights, as it involves an imposition on their ability to dispose of their car as they

24

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

wish. In Brown and Levinson’s (1987: 62) terms, this kind of request threatens the addressee’s ‘want’ that their ‘actions be unimpeded by others’. Different overarching terms have been used to capture what people do in interaction to construct their relationships with other people. Locher and Watts (2008: 78) talk about ‘relational work’ to refer to ‘the work people invest in negotiating their relationships in interaction’. Spencer-Oatey (2002, 2007, 2008) uses the notion of ‘rapport management’ to refer to how interlocutors orient to face and sociality rights, and proposes three types of orientation that are particularly relevant to our data: ●

●

●

Rapport maintenance orientation: ‘a desire to maintain or protect harmonious relations between the interlocutors’ (e.g. expressing a request via an interrogative rather than an imperative). Rapport enhancement orientation: ‘a desire to strengthen or enhance harmonious relations between the interlocutors’ (e.g. complimenting somebody’s cooking). Rapport challenge orientation: ‘a desire to challenge or impair harmonious relations between the interlocutors’ (e.g. humiliating the interlocutor).3

The general notions of relational work and rapport management include the more specific phenomena that have been captured by theories of ‘politeness’ and ‘impoliteness’. Politeness theory, as proposed by Brown and Levinson (1987), focuses on linguistic strategies aimed at mitigating face threats, which provided the basis for Spencer-Oatey’s rapport maintenance orientation (see Kerbrat-Orecchioni (1997) for the use of the notion of ‘face enhancing acts’ in politeness theory). The study of impoliteness, on the other hand, focuses on the use of language to attack face and cause offence (cf. Spencer-Oatey’s rapport challenge orientation), and is particularly relevant to our data. Among the ways in which impoliteness can be achieved in language, Culpeper (2011a) identifies the following set of conventionalized formulae: ●

●

●

insults in the form of negative vocatives (e.g. ‘you fat moron’), negative assertions (e.g. ‘you can’t do anything right’), negative references (e.g. ‘your stinking mouth’), third person negative references within earshot of the target (e.g. ‘the daft bimbo’) criticisms/complaints (e.g. ‘that was absolutely rubbish’) challenging or unpalatable questions or presuppositions (e.g. ‘why do you make my life impossible?’)

●

condescensions (e.g. ‘that’s childish’)

●

message enforcers (e.g. ‘listen here’ as a preface to an utterance)

(IM)POLITENESS AND POWER IN VOICE-HEARING

●

dismissals (e.g. ‘shove off ’)

●

silencers (e.g. ‘shut the fuck up’)

●

threats (e.g. ‘I’m gonna beat the shit out of you’)

●

curses and ill-wishes (e.g. ‘go to hell’).

25

(examples from Culpeper, 2011a: 135–6) Whether the use of any formula actually results in offence, however, depends on how the hearer perceives it. 2.3 Impoliteness and power As we have already mentioned, power is an important aspect of relational work as it affects the way in which interlocutors orient to one another’s face and sociality rights. Impoliteness, in particular, is often discussed in relation to the negotiation and exercise of power in interaction (see also Bousfield, 2008; Locher and Bousefield, 2008). Other things being equal, a powerful participant has more freedom to use fewer strategies to maintain others’ face and sociality rights, and more freedom to be impolite, that is, to attack these. Culpeper (2011a) uses the term ‘coercive’ impoliteness, for the instances of impoliteness which seek ‘a realignment of values between the producer and the target such that the producer benefits’ (Culpeper, 2011a: 226). This includes causing damage to an interlocutor’s face and reducing the power or status they have. As Bousfield and Locher (2008) put it, impoliteness is an exercise of power as it has arguably always in some way an effect on one’s addressees in that it alters the future action-environment of one’s interlocutors … because an interlocutor whose face is damaged by an utterance suddenly finds his or her response options to be sharply restricted. (Locher and Bousefield, 2008: 8–9) Culpeper (2008) elaborates on this restriction of options stating that it arises because a hearer or addressee is pressured into action (see also García-Pastor, 2008): self-preservatory action, for example, to maintain face and challenge the power dynamic, or not reacting and thereby accepting damage to face and ceding power to the speaker. In our introductory excerpt, Participant 6 does not respond to the voices’ impoliteness, but instead says that he is ‘used to it now’, suggesting that he has accepted the power imbalance between himself and the voices. Of course, as indicated above, different kinds of impoliteness suggest different kinds of power. Insults, for example, make the speaker seem superior, while dismissals or silencers suggest that the speaker has the power to influence others’ actions within or beyond the interaction (Culpeper, 2008, drawing on Beebe, 1995).

26

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

3 DATA AND METHODS Our analysis in this chapter is a secondary analysis of semi-structured interviews with ten voice-hearers with diagnoses of schizophrenia-spectrum disorders. The semi-structured interviews included questions from the auditory hallucinations subscale of the Psychotic Symptoms Rating Scale (PSYRATS-AH; Haddock et al., 1999), sections of the Cognitive Assessment of Voices interview (CAV; Chadwick and Birchwood, 1994) as well as questions developed to assess and quantify additional features of voices (e.g. self-report ratings of voice-related distress and pleasantness of voices; Varese et al., 2017). The interviews were recorded and fully transcribed and anonymized for the original study, but to avoid preempting the linguistic analysis, one of the authors (FV) removed all references to specific quantitative ratings of voice-related distress before the transcripts were shared with linguistic researchers (ZD, ES, AM). Interview transcripts ranged from 1,125 to 6,526 words (average = 3,326; median = 2,885) and totalling 33,257 words, after these exclusions. Our ten participants were drawn from a larger sample of clinical and nonclinical voice-hearers recruited for a previous mixed-method investigation (Varese et al., 2016; Varese et al., 2017). Participants for our secondary analysis were selected from the original pool of participants according to two criteria: diagnostic homogeneity (to avoid likely variation due to the diversity of clinical features of participants) and ‘spread’ of voice-related distress scores (to ensure ability to contrast participants with differing levels of distress in the present analyses). All participants were also aged above sixteen; had experienced voices in the two weeks prior to participating in the study; had history of voice-hearing for a minimum of six months; and their experience of voices was not due to organic illness (e.g. brain injury), hypnagogic/hypnopompic states or alcohol/ drugs intoxication. Table 1.14 displays the participants’ basic demographic characteristics (age, gender, education level), the number of years they have heard voices for, their scores on the amount and intensity of distress items of the PSYRATSAH (ranging from 0 to 4), their self-reported ratings of voice-related distress and voice pleasantness, and their scores on a measure of severity of anxiety, depression and stress symptoms in the week preceding the interview (the DASS21; Lovibond and Lovibond, 1995). The original study received full NHS ethics approval (REC reference: 13/ NW/0290) and R&D approval from the participating NHS Trusts, which allowed us to conduct our secondary, linguistic, analysis on the data set. As the audio recordings were no longer available, our linguistic analysis was conducted on the transcripts from the original study (see also Demjén et al., 2019). This involved dividing the transcripts into clauses (minimal structures consisting of a verb and surrounding words that are closely linked to it grammatically), and

Secondary

Secondary (GCSE)

Higher Education

Secondary

Secondary

Secondary (GCSE)

University degree

Secondary (GCSE)

Secondary (GCSE)

P11 (51, Female)

P13 (37, Male)

P14 (20, Female)

P17 (44, Male)

P19 (31, Male)

P22 (40, Female)

P24 (46, Male)

P29 (21, Female)

P33 (47, Female)

12

16

19

12

10

10

18

12

12

9

13

1

20

33

3

27

1

10

36

19

Education Years since years voice onset

0

2

3

3

3

4

3

1

4

4

PSYRAT-AH amount of distress

0

2

3

3

3

3

2

1

3

4

PSYRATS-AH intensity of distress

2

7

7

8

8

8

6

2

8

10

Self-reported voice-related distress

7

10

3

2

0

0

0

8

0

0

Self-reported voice pleasantness

10

26

34

46

41

45

43

0

57

58

DASS21

Notes: DASS21 = Short version of the Depression Anxiety Stress Scale; PSYRATS-AH = Auditory hallucinations scale of the Psychotic Symptoms Rating Scales

Secondary

P6 (59, Male)

Participant ID (age, gender) Education level

TABLE 1.1 Demographic Characteristics of Interview Participants

28

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

manually coding each clause for a number of different linguistic dimensions. Among other things, we identified the clauses in which the voice-hearer reports speech exchanged between themselves and the voice, and coded them for the three main rapport management orientation categories we introduced earlier, that is, maintenance, enhancement and challenge/attack in relation to face or sociality rights. Where relevant, we also coded for the presence of the (im) politeness formulae presented earlier in this chapter. We begin Section 4 with an examination of patterns of rapport management, reflecting relationships that might cause distress, followed by examples of patterns that are likely to reflect more amicable relationships. Specifically, we begin with a focus on one participant, Participant 6, and use patterns in other interviews to demonstrate alternative, less distressing ways in which the relationship with voices can manifest. We begin with some overall trends, then look in more detail at the specific aspect of the person’s face or freedom of action that is attacked, maintained or enhanced, and what strategies are used to achieve these effects. The analysis is followed by an overview of the implications of our findings for clinical practice.

4 ANALYSIS Participant 6 (P6) is a man in his late fifties with a long-standing history of mental health difficulties; he has been hearing voices for almost twenty years. From the interview, it is clear that he values the support he is receiving from mental health services, and he is very concerned that he might no longer receive this support in the near future. As Table 1.1 shows, P6, according to the PSYRATS and CAV measures, is among the most distressed individuals among our interviewees. The interview was also one of the longest in our dataset at just over 6,000 words. 4.1 Participant 6 and distress At the highest level of generality, we can provide an overview of the frequencies of different types of (im)politeness strategies per 1,000 clauses. The interview with P6 includes 1,002 clauses uttered by the interviewee and among these clauses 124 were coded as voices’ attacks on some aspect of the hearer (approx. 124/1,000 clauses) and 16 were coded as hearer’s attacks on some aspect of voices (16/1,000 clauses), that is, instances of impoliteness. There were no examples of enhancement in this interview and only two examples of potential maintenance, that is, politeness. Table 1.2 shows how this pattern of rapport management compares with other interviews. Of course, such high level quantification is mainly useful as a way into the data and only begins to point towards the nature of the relationships between voices and hearers. We therefore now turn to exactly how these patterns are realized,

0

Maintenance per 1000 clauses

0

0

124

2

0

1002

P6

0

0

114

23

0

132

P22

Attack per 1000 clauses 18 16 0 a These two clauses are negated, that is, reports of enhancement not happening.

0

Enhancement per 1000 clauses

(Im)politeness directed from Hearers to Voices

177

2

Maintenance per 1000 clauses

Attack per 1000 clauses

0

435

Enhancement per 1000 clauses

(Im)politeness directed from Voices to Hearers

Clauses overall

P11

0

0

0

84

32

10

311

P24

12

8

0

82

8

0

503

P14

0

0

0

72

1

2a

691

P17

1

31

10

42

21

28

287

P13

0

0

0

36

9

42

336

P29

0

8

0

27

90

0

256

P19

TABLE 1.2 Normalized Figures for Clauses Including Enhancement, Maintenance and Attack of Face or Sociality Rights

4

0

0

12

8

30

732

P33

30

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

addressing specifically the ways in which they might indicate a reduction in P6’s ability to control his life and maintain his own self-esteem. Given the proportion of relational work in this interview that involves some form of impoliteness, we will focus on this in the discussion of P6. The examples in this section are P6’s reports of what his voices say. Starting with impoliteness directed at ‘face’, P6 reports hearing the kinds of insults, in the form of negative vocatives, that have been highlighted in the psychological literature as characteristic of voice-hearing and related distress. In the introductory excerpt, ‘you worthless bastard’ and ‘useless, pathetic, hopeless, fat, evil, nasty bastard’, uttered by the voices to P6, are typical examples of this. They constitute attacks on P6’s individual identity face, including his appearance, character and general worth. However, interestingly, the majority of the voices’ attacks on P6’s face are not negative vocatives but rather utterances such as the following: (1) ‘it’s gonna be a disaster, you’re not gonna get on with the person, they’re gonna plot against ya, they’ll want to get rid of ya as soon as they meet ya, they’ll-they’ll do everything they can to rid of you off the books, it’s gonna be a disaster, they’re gonna be nasty, horrible people.’ (2) ‘they are all gonna get rid of you’ (3) ‘it’s a lie they do know [who the next healthcare coordinator will be] and you’re not getting one they want you out.’ (4) ‘they don’t give a fuck about you’ These examples seem to fall between typical impoliteness formulae: they are negative assertions to an extent, but they are also similar to warnings or threats as they refer to or have implications for the future. Warnings overlap with threats (which are among the standard impoliteness formulae) in their future orientation, but tend to be designed to inform the person of some negative future eventuality for their own benefit. They are particularly interesting here because these are statements about someone else in relation to the hearer, not the hearer himself. The voices warn P6 by making negative assertions that refer to either the healthcare visitors or the nature of their relationship with P6. In this way, the voices attack the hearer’s relational identity face, that is, his ability to have good relationships with others, and particularly his sense of being valued and supported by them. This is particularly important for P6. As outlined above, P6 not only values the support he is receiving from services but also worries that the support will be withdrawn. Because warnings are generally intended to benefit the hearer, they are perhaps more difficult to recognize as impoliteness/face attack and therefore potentially more difficult to resist. While the main effect of threats tends to be intimidation (Culpeper, 2011a), the same cannot be said of warnings, where

(IM)POLITENESS AND POWER IN VOICE-HEARING

31

the effect, in this case at least, seems to be to create anxiety. Specifically, these attacks both reflect and feed P6’s most intense fears. The voices also employ challenging or unpalatable questions to attack P6’s relational identity face such as the following: (5) ‘If you’re getting on so great how come you never get any positives from the last visit? How come he’s not here to support you now? Where is he when we’re kicking the fuck out of you?’ Similarly to the previous set of examples, these extracts are questions about someone else in relation to the hearer, not the hearer himself. They insinuate that P6’s relationship with the healthcare visitor is not good, thereby undermining his confidence in a relationship that might be helpful for coping with the voices, and indeed the main social relationship that P6 has with anyone. Turning now to impoliteness directed at P6’s sociality rights – the desire for fairness, to be unimpeded and to associate with others – the voices attack P6 most frequently using commands and threats. The latter make violent predictions about things the voices will do to the hearer: (6) ‘we are gonna kick the fucking shit out of your head’ (7) ‘and we’re gonna fucking have you for this’ (8) ‘you aren’t gonna get no peace until you’re dead’ These attack the person’s right to be treated fairly, as well as their right to not be imposed upon, and are among the types of examples that clinicians are generally already well attuned to. What is interesting here is that, in the impoliteness literature, threats are generally known to come with conditions (Culpeper, 2011a). Conditional threats can potentially be counteracted by ensuring that the condition placed on them is not met. For example, if someone threatens to ‘kick the fucking shit out of your head if you do that again’, then, arguably, the addressee can avoid the ‘kicking’ by not doing ‘that’ again. The threats in our data, however, are non-conditional. As noted by Culpeper et al. (2017), non-conditional threats offer no opportunities to prevent the aggression in the threat being realized. They leave the addressee with no way out. This is likely to contribute to the distress experienced by voice-hearers whose voices employ these strategies. It is worth pointing out that Culpeper et al. (2017) were looking at threats in the context of religiously aggravated hate crime – such is the severity of the threats in our examples. The commands P6’s voices employ, which challenge P6’s sociality rights, are similarly extreme:5 (9) ‘hang yourself ’ (10) ‘kill yourself blow the flat up, set yourself on fire, go out there later on today and walk under a bus’

32

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

These kinds of commands are not accounted for particularly well with prototypical impoliteness formulae: some can be interpreted as partly ill-wishes, but this category does not capture the full thrust of the utterances. They are most similar, in fact, to what Culpeper et al. (2017) describe as ‘incitement’ in the context of religiously aggravated hate crime. However, the individual being incited to violence is identical with the future victim, namely the voice-hearer. Culpeper et al. (2017) describe the effect of incitement as similar to threats, that is, intimidation. In our case, however, the entity that P6 needs to fear is in fact himself. Such commands are well documented in the clinical literature on voice-hearing and schizophrenia and are known to be predictors of distress. In our data as well, this kind of impoliteness, aimed at the person’s right not to be imposed upon, are quite typical of interviewees who are distressed by voices (e.g. P6, P11, P17). Given the complex way in which such commands induce fear, this is hardly surprising. The voices are also reported as challenging P6’s sociality rights in additional ways: (11) ‘your fucking [health] visit’s winding down’ (12) ‘you’ve had 25 minutes of your visit’ These utterances could be seen as simple statements of fact. However, in the case of P6, they are examples of context-driven implicational impoliteness (Culpeper, 2011a), as they state things that P6 wishes would not happen. Because of their future orientation, these are similar to the warnings and threats discussed earlier: the voices repeatedly warning of a visit coming to an end can be seen as for P6’s benefit, as he might want to know how much more time he has to get the support he needs (though he might be aware of this anyway). However, these statements are actually still indicating harm about to come to him – the time he has left to get support from nurses is coming to an end and he will then be at the mercy of the voices again. There isn’t enough research to determine whether such implicational impoliteness causes as much or more offence than prototypical formulae (Culpeper et al., 2017), but in the case of P6, they are likely to be the source of more distress because these particular utterances get under his skin: they once again reflect and reinforce his particular fears. Overall then, even based on the information in Table 1.2, one can say that, in the interview with P6, the hearer and his voices have a rather asymmetrical and antagonistic relationship. The voices are reported as performing almost eight times more attacks on face and sociality rights than P6. This is indicative of the power dynamic at play, since asymmetrical patterns of impoliteness tend to occur in situations where there is a more powerful individual targeting the less powerful (Culpeper, 2008). Of course, quantification at this general level is only a starting point; however, we did find that, across our ten interviews, the highest levels of distress tended to correlate with the highest numbers of attacks from

(IM)POLITENESS AND POWER IN VOICE-HEARING

33

voice to hearer (around 100 or more per 1,000 clauses) (P6, P11, P22). Similarly, we observed that distress correlated with the absence of face enhancement strategies and low numbers of face maintenance strategies (P6, P11, P14, P17, P22), although the latter were less frequent in general. In other words, distress was associated with more antagonistic relationships between voices and hearers. The more detailed examination of impolite utterances employed by P6’s voices revealed much more about why this might be the case. The voices claimed the power both to influence P6’s sociality rights and to undermine his confidence or sense of self, that is, face, and they did so using a variety of, often extreme, impoliteness formulae in combination, and at times designed to damage those aspects of P6 that were most vulnerable (his relational identity face, for example). Most of the time, P6 does not challenge his voices. As already indicated in the introductory excerpt, he has rather gotten used to things as they are, including the power dynamic constructed through the voices’ verbal behaviour. This is especially interesting: given that P6’s voices are not entities that exist within the shared external reality, at least not at the time of the reported interactions, there is no pre-existing power behind discourse; any power that exists only comes into play within discourse. In other words, any power that the voices have is claimed and attributed via language in interaction and is not legitimated in any other way. In theory, this could make it easier for P6 to challenge the power dynamic suggested by the voices and makes P6’s acquiescence all the more marked. We find explanations for this in the instances where P6 does respond to the voices. P6 explains that if health visits go well, and he feels that ‘they [health visitors] believed me’ then this gives him ‘sticks to hit [the voices] with’: (13) ‘you gotta get as much goodness and feedback out of that [health] visit as you can, so that when they’ve gone you’ve got some positives you can say “well nurse [NAME] said this and that and that, work that one out you bastards, sort that one out”.’ However, when the visit does not go well, responding to or challenging the voices becomes difficult: (14) ‘and you can say “nothing” cos nothing did. You can't say well [NAME] said this but he didn’t [NAME] said that, “no he didn’t.” [the voices contradicting the hearer]’ At other times, the voices are simply too powerful: (15) ‘You try to over-shout them, but you can't and that’s what it’s like with the voices you try to shout back or fight back or anything like that and you can't.’ Some of these examples include elements of what Culpeper et al. (2003) describe as ‘offensive counters’, that is, impolite responses to received

34

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

impoliteness (e.g. ‘you bastards’ and presumably also ‘fight back’). However, as these examples show, in contrast with the voices’ attacks on P6’s face, any responses are generally difficult to execute (as indicated by ‘try’ and ‘can’), and sometimes frustrated or non-realized (as indicated by ‘can’t’). They are also contingent on circumstances beyond P6’s control: the healthcare meeting and his relationship with healthcare professionals. This provides further evidence of the extent of the asymmetry between voice and hearer and the inevitable acceptance of the attacks this results in further damages P6’s face and sociality rights (cf. Culpeper et al., 2003). In Culpeper’s (2011a) terms, the kind of impoliteness described in this section can be classified as ‘coercive’ (Culpeper, 2011a: 226), as it ‘involves coercive action that is not in the interest of the target’. Drawing on Tedeschi and Felson (1994), Culpeper (2011a) understands coercive action here as action designed to impose harm or force compliance. The intended harm need not be physical, but can also include damage to social identity, that is, damage to face (cf. Spencer-Oatey, 2008). This also reduces the target’s power. In theory, coercive impoliteness is risky, because there is a high probability of retaliation, either immediately or in future (Culpeper, 2011a). However, the range of formulae used by the voices and the range of face and sociality rights attacked can be seen as pre-emptive (and preventative) of any such retaliation. In fact, it is worth noting that ‘impoliteness’ as an umbrella term tends to cover quite mild impoliteness and might even be seen as inadequate for the relational behaviour we describe. The voices’ overall behaviour towards P6, and indeed some other participants in our data, might therefore warrant Rudanko’s (2006) ‘aggravated impoliteness’ label, which Culpeper et al. (2017) suggest for instances of religiously aggravated hate crime. The asymmetrical relationship and severity of the impoliteness behaviours (from voice to hearer), combined with the multiple aspects of the person attacked, using a range of formulae, goes some way towards explaining why P6 (and others) are distressed by their voices. Some of these impoliteness formulae are obvious (e.g. self-harm commands and threats) and are adequately addressed by existing psychiatric assessments and interventions. Others, however, deserve more attention. Some forms of impoliteness (e.g. threat/warnings, unpalatable questions) are, on the surface at least, meant to be helpful to the voice-hearer, which means that, when they are used to undermine, they can be harder to identify and do something about. This in turn can result in further reduction of hearers’ sense of control over their lives, which leads to distress. 4.2 Other participants and patterns Not all participants in our cohort reported such confrontational relationships with their voices. Participants 13, 29 and 33, for example, showed very different

(IM)POLITENESS AND POWER IN VOICE-HEARING

35

patterns in terms of (im)politeness, and these differences help to explain, at least in part, why these participants were less distressed by their voices. The more symmetrical or reciprocal nature of the relationship between voices and hearers, and the lower degree of antagonism, are already suggested by the figures in Table 1.2. The interviews with P13, P29 and P33, in particular, display lower frequencies (per 1,000 clauses) of impoliteness, or rapport challenge orientation, and higher frequencies of instances of rapport maintenance orientation, which captures linguistic strategies used to (a) minimize any threat to one’s interlocutor and to ‘maintain’ each other’s face and sociality rights; and (b) actively bolster, or ‘enhance’ a counterpart’s face or sociality rights. Because such strategies promote social harmony, they result in more amicable relationships and can therefore be linked to lower levels of distress. The precise ways in which social harmony is upheld in these relationships with voices is, of course, also of interest, not least because it might reveal strategies that can be usefully employed to make more antagonistic relationships with voices less so. A crucial difference between, for example, P6’s experience and that of both P13 and P33 is that the latter report or refer to the voices enhancing their individual identity face. In the following extracts, the voices actively praise or encourage the hearers, for example, for past or future achievements or their general abilities (note: some of these are reported in indirect speech): (16) ‘You have done OK’ (P33) (17) ‘You will soon be writing that book’. (P33) (18) ‘You are getting better’ (P13) (19) ‘Most times they praise me.’ (P13) (20) ‘they try to support me’. (P13) Similarly, P13, in particular, returns the politeness and encourages his voices: (21) ‘And I try and support her’ (22) ‘supporting them in many ways’ While the indirect way in which these utterances (and Examples 26 and 27) are reported means that we cannot comment on any specific linguistic strategies, or indeed on whether they relate to face or sociality rights, the ways in which P13 labels them (praise, support) suggest enhancement. The relationship between power and face/rights enhancement has received very little attention in the linguistics literature. However, it could be argued that, for praise and encouragement to be meaningful, the speaker needs to be in a more authoritative position in some way (e.g. having a better idea about the abilities of the hearer than the hearer him/herself). But, crucially, any power that the speaker has is used for the hearer’s rather than the speaker’s benefit.

36

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

The utterances above, for example, resemble parent–child interactions. Additionally, in the case of P13, because the enhancement goes both ways, there is a kind of reciprocity in place: power shifts back and forth between voices and hearer depending on the situation. This is reminiscent of close relationships such as that between siblings or friends. In fact, this is exactly how P13 describes the relationship, saying, ‘I treat her like a sister’ and ‘they treat me as a friend’. The amicable nature of the relationship between P13 and his voices is also evident in extracts such as the following: (23) ‘I will try and correct her or point her in the right direction.’ (24) ‘I am probably more giving them advice.’ P13 reports verbal behaviour on his part that potentially threatens the face or sociality rights of the voices. Correction implies criticism, which can fall under politeness or impoliteness depending on how it is performed. Advice similarly implies some problem, lack or wrongdoing that needs correction, albeit again with the hearer’s interest in mind. However, the way in which these utterances are reported suggests that these potential threats to face and rights are uttered in a non-antagonistic way, even though the interviewee does not provide any specific wordings. Minimally, there is no evidence of any intent to cause offence, and the hedging that P13 uses in his reports (‘try’, ‘probably’, ‘more’) suggests that the advice and correction are delivered gently. Indeed, P13 even reports the voices seeking advice from him. The voices actively defer to his knowledge and experience, which means that even a fairly unmitigated delivery would not disrupt social harmony. Where there is evidence of potential impoliteness in interviews with less distressed participants, it is reported differently from more distressed interviewees such as P6 and P17. In P13’s interview, for example, rather than harmful commands, voices’ attacks on sociality rights mainly consist of challenging questions, which are only potentially undermining, and a dismissal (‘Fuck off ’), which is qualified as potentially ‘jokey’. In the case of P29, the way the voices try to influence her freedom of action is even more mild and falls within the remit of politeness rather than impoliteness: (25) ‘Say I picked something up in the shop – she often says “There is no point in buying that.” The other day, I picked up a jumper in Next, and I heard her voice saying “There is no point in buying that”’. ‘There is no point in buying that’ is indirect, which is one of Brown and Levinson’s (1987) politeness mechanisms. It implies rather than imposes, making it easier for the participant to not comply. It is an attempt at exercising power, but not in a forceful way, ensuring that the hearer is not offended in the

(IM)POLITENESS AND POWER IN VOICE-HEARING

37

process. Similarly, the few commands that his voices issue at P13 even enhance face: ‘Keep up your volunteer work. Keep doing the things that … Keep up the good work.’ When it comes to attacks on face, although P13 reports some insults, they are often milder than those in other interviews, or perceived as light-hearted: (26) ‘Every now and then, they will swear at me. Just rude words and things like that.’ (27) ‘Sometimes they can take the mickey out of me as well.’ P29 reports some very harsh negative assertion insults (e.g. ‘You life isn’t worth living’) but points out that these are infrequent and associated with only one of the voices that she hears. P33 also reports some attacks in the form of negative assertions: (28) ‘you’re sitting around too much’ (29) ‘you’re spiteful’ (30) ‘you won’t be able to do that’ These utterances potentially attack both face and sociality rights but not in a way that forces the participant into action or confrontation. In addition, there is evidence in at least some interviews that people who are less distressed by their voices are also able to challenge them. P13, for example, reports responding to the voices by saying, ‘You’re unfair’, while P33 responds with a possible attack on the voice’s sociality rights: ‘I can tell them to shut up.’ This again suggests a more egalitarian or symmetrical relationship. A word of caution here, however: the fact that we are dealing with data where all this is effectively going on within one individual’s mind means that the interpretation of reciprocal attacks is not quite so simple. P11, one of our more distressed participants, also reports attacking her voices’ face and sociality rights in return: ‘And I am saying out loud “Fuck off, and leave me the fuck alone. I’m sick of hearing you. I don’t want to hear you anymore. Go away!”’ While this may suggest a more balanced power dynamic, the ‘offense/counter-offense’ nature of the exchanges (cf. Culpeper et al., 2003) results in a very antagonistic relationship, potentially akin to civil war. Overall then, participants who are less distressed by their voices report fewer and milder instances of impoliteness on the part of the voices, which means that such voices do not exercise power as much as the voices of more distressed participants. The fact that hearers are able to resist or challenge their voices in return also supports this interpretation to some extent. In addition, there is evidence of politeness strategies to maintain and even enhance aspects of both the hearers and the voices, resulting in an amicable and symmetrical relationship.

38

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

5 IMPLICATIONS AND RECOMMENDATIONS FOR CLINICAL PRACTICE As outlined earlier, voice-hearing is not distressing for everyone. In clinical practice, health professionals attend to the factors that are known to play a role in distress, notably, hearers’ relationships with voices, their perceived control over the voices and their ability to control their life goals. However, there is still a limited amount of research on how relationships between voice and hearer, including power dynamics, are established and maintained, and the variety of ways that the phenomenology of voice-hearing specifically contributes to distress. It is this gap that our analysis has begun to address. As we have shown, there are a number of ways in which particularly impoliteness can cause distress in people who hear voices, and how the use of politeness might mean that the experience of voices is not distressing for some voice-hearers. Patterns that are likely to cause or contribute to distress in voicehearers include ●

●

●

●

●

the presence of large numbers of attacks on a person’s face or sociality rights; multiple types of face being attacked in combination with attacks on sociality rights; voices using multiple impoliteness strategies, including ones that could potentially masquerade as supportive on the surface; imbalance or lack of reciprocity in the relational work being reported; the hearer perceiving themselves as unable to react, respond or resist fully in the way that they wish to do so.

These patterns contribute to distress because they represent limitations on the voice-hearer’s ability to maintain a particular self-image and have the freedom to do things they want and not do things they do not want to do. These are important homeostatic processes that are absolutely linked to the general definition of control, as the ability to lead the life you want to lead (Varese et al., 2016, 2017). Therefore, curtailing these freedoms and rights leads to distress. Linguistic patterns, on the other hand, that are likely to facilitate coping well include ●

●

●

fewer instances of impoliteness; evidence of politeness mechanisms to maintain and even enhance face and/or sociality rights; reciprocity in the (im)politeness being reported (except in cases where the balance is in the form of aggression from both sides);

(IM)POLITENESS AND POWER IN VOICE-HEARING

●

39

the hearer perceiving themselves as able to react, respond or resist at least to some extent.

Our data does not allow us to determine a cause and effect relationship here. However, we know from the literature that interpersonal relationships and power are constructed and negotiated in interaction. This suggests that encouraging voice-hearers to interact with their voices in the ways that P13 and P33, for example, do, can, over time, improve the type of relationship they have. It might, therefore, be useful to continue thinking of the interactions with voices as interactions with other people, as is already the case in some existing relational therapies for psychosis/voice-hearing. In recent years, there has been a surge in interest in therapies that could specifically modify the relationship between hearers and their voices, such as AVATAR therapy (Craig et al., 2018), Relating Therapy for distressing voices (Hayward et al., 2018) and the Voice Dialogue approach developed within the context of the International Hearing Voices movement (Steel et al., 2019; Corstens et al., 2012). Future applied research could explore the value of applying rapport maintenance and enhancement orientations to manage the relationships and rapport with voices (cf. Spencer-Oatey, 2008) in the context of these novel therapies with a growing evidence base. Our analysis, especially of the particularly targeted attacks that voices directed at P6, also suggests that it might be helpful in clinical practice to open up discussions about where an individual’s priorities lie. As Spencer-Oatey (2008) points out, different aspects of face and sociality rights can be more or less important for different people, and the more that voices attack aspects of particular importance to a hearer, the more distressed they are likely to be. This is consistent with best practice in psychological therapies with people with psychosis, which stress the importance of placing the client’s goals, priorities and values at the centre of therapeutic interventions. A related avenue to explore is the issue of less obvious types of impoliteness, such as unpalatable questions and threats/warnings, for example. These might be more difficult to identify as attacks and therefore to resist, thereby meriting particular attention in a therapeutic context.

6 CONCLUDING REMARKS In this chapter, we reported on a pilot project that explored how concepts from rapport management and (im)politeness research can contribute to a better understanding of how relationships between voice and hearer are established and maintained, and how this is related to the degree of distress that voicehearers might experience. These linguistic frameworks explore how language is used to maintain or disrupt social harmony as individuals go about achieving

40

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

their (interactional) goals. They also acknowledge that relationships, including power dynamics, are reflected, enacted and created in interaction. Importantly, linguistic analysis of this kind does not require that people, including voicehearers, are aware of and willing to explicitly describe or assess the nature of their relationships with others. Such analyses can be performed with any kind of data that details what is said in verbal interactions. While we focused on a very specific context in our analysis, the concepts we introduce from (im) politeness theory are at play in any interaction in healthcare contexts and can therefore be used to shed light on the dynamics at work in other healthcare relationships as well. We centred our analyses around the concepts of face and sociality rights, focusing on their enhancement, maintenance or attack in the reported interactions between voices and hearers. The most distressing voices tended to undermine a voice-hearer using a variety of extreme impoliteness formulae (akin to those identified in contexts of religiously aggravated hate crime), designed to damage those aspects of the person that were most important to them. Such attacks tended to go unchallenged by the voice-hearer, which means that the voices successfully claimed power for themselves through language. We showed that such (im)politeness patterns, that is, different types of attacks, can lead to distress because they result in limitations on the person’s ability to maintain a particular self-image and have the freedom to do things they want and not do things they do not want to do. Future research would need to examine the extent to which such antagonistic relationships can be ‘turned around’ by actively applying politeness strategies that less distressed participants used.

NOTES 1. This study was funded by a UCL Institute of Education Seed Grant (2017/18). The work was supported by the Lichtenberg-Kolleg at the Georg-August-Universität Göttingen, Germany. 2. It is worth pointing out that studies of (im)politeness tend to focus on direct speech, rather than direct reports of the kinds of things that are said. Indeed, depending on the level of detail in indirect speech reporting, certain kinds of linguistic analyses are impossible to perform. However, we follow Culpeper et al. (2017) in considering such more marginal instances of (im)politeness, provided that enough information is included in the report for a basic categorization of the orientation of an utterance in terms of attack, maintenance or enhancement. 3. Spencer-Oatey’s framework also includes ‘rapport neglect orientation’, but this is not sufficiently relevant to our data to be included here. 4. The linguistic researchers (ZD, ES and AM) did not have access to the information in Table 1.1 prior to the analysis. The linguistic analysis was done ‘blind’ to avoid creating assumptions relating to distress. In this way, power and distress

(IM)POLITENESS AND POWER IN VOICE-HEARING

41

were linked with (im)politeness patterns on the basis of linguistic evidence and the linguistic assessments were examined in light of psychometric ones only retrospectively. 5. This is, in fact, what makes these commands instances of impoliteness, rather than bold-on-record politeness.

REFERENCES Beebe, L. M. (1995), ‘Polite fictions: Instrumental rudeness as pragmatic competence’, in J. E. Alatis, C. A. Straehle, B. Gallenberger and M. Ronkin (eds), Linguistics and the Education of Language Teachers: Ethnolinguistic, Psycholinguistics and Sociolinguistic Aspects. Georgetown University Round Table on Languages and Linguistics, 154–68, Georgetown: Georgetown University Press. Bousfield, D. (2008), Impoliteness in Interaction, Philadelphia and Amsterdam: John Benjamins. Bousfield, D. and Locher, M. A., eds (2008), Impoliteness in Language: Studies on Its Interplay with Power in Theory and Practice, Berlin: Mouton de Gruyter. Brown, P. and Levinson, S. C. (1987), Politeness: Some Universals in Language Use, Cambridge: Cambridge University Press. Chadwick, P. and Birchwood, M. (1994), ‘The omnipotence of voices: A cognitive approach to auditory hallucinations’, The British Journal of Psychiatry, 164 (2): 190–201. Corstens, D., Longden, E. and May, R. (2012), ‘Talking with voices: Exploring what is expressed by the voices people hear’, Psychosis, 4 (2): 95–104. Craig, T. K., Rus-Calafell, M., Ward, T., Leff, J. P., Huckvale, M., Howarth, E., Emsley, R. and Garety, P. A. (2018), ‘AVATAR therapy for auditory verbal hallucinations in people with psychosis: A single-blind, randomised controlled trial’, The Lancet Psychiatry, 5 (1): 31–40. Culpeper, J. (2008), ‘Reflections on impoliteness, relational work and power’, in D. Bousfield and M. Locher (eds), Impoliteness in Language: Studies on Its Interplay with Power in Theory and Practice, 17–44, Berlin: Mouton de Gruyter. Culpeper, J. (2011a), Impoliteness: Using Language to Cause Offence, Cambridge: Cambridge University Press. Culpeper, J. (2011b), ‘Politeness and impoliteness’, in K. Aijmer and G. Andersen (eds), Sociopragmatics, 391–436, Berlin: Mouton de Gruyter. Culpeper, J., Bousfield, D. and Wichmann, A. (2003), ‘Impoliteness revisited: With special reference to dynamic and prosodic aspects’, Journal of Pragmatics, 135: 1545–79. Culpeper, J., Iganski, P. S. and Sweiry, A. B. (2017), ‘Linguistic impoliteness and religiously aggravated hate crime in England and Wales’, Journal of Language Aggression and Conflict, 5 (1): 1–29. Culpeper, J., Mackey, A. and Taguchi, N. (2018), Second Language Pragmatics: From Theory to Research, New York and London: Routledge. Demjén, Z., Semino, E., Marszalek, A. and Varese, F. (2019), ‘Metaphor framing and distress in lived-experience accounts of voice-hearing’, Psychosis, 11: 1, 16–27. Available online: https://doi.org/10.1080/17522439.2018.1563626. Fairclough, N. (1989), Language and Power, London and New York: Longman.

42

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Franck, E., de Raedt, R. and de Houwer, J. (2007), ‘Implicit but not explicit selfesteem predicts future depressive symptomatology’, Behavior Research and Therapy, 45 (10): 2448–55. García-Pastor, M. D. (2008), ‘Political campaign debates as zero-sum games: Impoliteness and power in candidates’ exchanges’, in D. Bousfield and M. A. Locher (eds), Impoliteness in Language: Studies on Its Interplay with Power in Theory and Practice, 101–26, Berlin: Mouton de Gruyter. Goffman, E. (1967), Interaction Ritual: Essays on Face-to-face Interaction, Chicago: Aldine Publishing. Haddock, G., McCarron, J., Tarrier, N. and Faragher, E. B. (1999), ‘Scales to measure dimensions of hallucinations and delusions: The psychotic symptom rating scales (PSYRATS)’, Psychological Medicine, 29 (4): 879–89. Hayward, M., Bogen-Johnston, L. and Deamer, F. (2018), ‘Relating Therapy for distressing voices: Who, or what, is changing?’, Psychosis, 10, no. 2: 132–41. Jenner, J. A., Rutten, S., Beuckens, J., Boonstra, N. and Sytema, S. (2008), ‘Positive and useful auditory vocal hallucinations: Prevalence, characteristics, attributions, and implications for treatment’, Acta Psychiatrica Scandinavica, 118 (3): 238–45. Kerbrat-Orecchioni, C. (1997), ‘A multilevel approach in the study of talk-ininteraction’, Pragmatics, 7 (1): 1–20. Locher, M. A. (2004), Power and Politeness in Action: Disagreements in Oral Communication, Berlin: Mouton de Gruyter. Locher, M. A. and Bousfield, D. (2008), ‘Introduction: Impoliteness and power in language’, in D. Bousfield and M. A. Locher (eds), Impoliteness in Language: Studies on Its Interplay with Power in Theory and Practice, 1–13, Berlin: Mouton de Gruyter. Locher, M. A. and Watts, R. J. (2005), ‘Politeness theory and relational work’, Journal of Politeness Research, 1, H. 1: 9–33. Locher, M. A. and Watts, R. J. (2008), ‘Relational work and impoliteness: Negotiating norms of linguistic behaviour’, in D. Bousfield and M. A. Locher (eds), Impoliteness in Language: Studies on Its Interplay with Power in Theory and Practice, 77–99, Berlin: Mouton de Gruyter. Lovibond, P. F. and Lovibond, S. H. (1995), ‘The structure of negative emotional states: Comparison of the depression anxiety stress scales (DASS) with the beck depression and anxiety inventories’, Behaviour Research and Therapy, 33 (3): 335–43. Mawson, A., Cohen, K. and Berry, K. (2010), ‘Reviewing evidence for the cognitive model of auditory hallucinations: The relationship between cognitive voice appraisals and distress during psychosis’, Clinical Psychology Review, 30 (2): 248–58. McCarthy-Jones, S. (2012), Hearing Voices: The Histories, Causes, and Meanings of Auditory Verbal Hallucinations, Cambridge: Cambridge University Press. Rudanko, J. (2006), ‘Aggravated impoliteness and two types of speaker intention in an episode in Shakespeare’s Timon of Athens’, Journal of Pragmatics, 38 (6): 829–41. Spencer-Oatey, H. (2002), ‘Managing rapport in talk: Using rapport sensitive incidents to explore the motivational concerns underlying the management of relations’, Journal of Pragmatics, 34 (5): 529–45. Spencer-Oatey, H. (2007), ‘Theories of identity and the analysis of face’, Journal of Pragmatics, 39 (4): 639–56.

(IM)POLITENESS AND POWER IN VOICE-HEARING

43

Spencer-Oatey, H. (2008), ‘Rapport management: A framework for analysis’, in H. Spencer-Oatey (ed.), Culturally Speaking: Culture, Communication and Politeness Theory, 2nd edn, 11–47, London and New York: Continuum. Steel, C., Schnackenberg, J., Perry, H., Longden, E., Greenfield, E. and Corstens, D. (2019), ‘Making sense of voices: A case series’, Psychosis, 11: 1–13. Tedeschi, J. T. and Felson, R. B. (1994), Violence, Aggression, and Coercive Actions, Washington DC: American Psychological Association. Thomas, N., Hayward, M., Peters, E., et al. (2014), ‘Psychological therapies for auditory hallucinations (Voices): Current status and key directions for future research’, Schizophrenia Bulletin, 40 (Supplement 4): S202–S12. Varese, F., Tai, S. J., Pearson, L. and Mansell, W. (2016), ‘Thematic associations between personal goals and clinical and non-clinical voices (Auditory Verbal Hallucinations)’, Psychosis, 8 (1): 12–22. Varese, F., Mansell, W. and Tai, S. J. (2017), ‘What is distressing about auditory verbal hallucinations? The contribution of goal interference and goal facilitation’, Psychology and Psychotherapy: Theory, Research and Practice, 90 (4): 720–34. Woods, A., Jones, N., Bernini, M., Callard, F., Alderson-Day, B., Badcock, J.C., Bell, V., Cook, C.C.H., Csordas, T., Humpston, C., Krueger, J., Larøi, F., McCarthyJones, S., Moseley, P., Powell, H., Raballo, A., Smailes, D. and Fernyhough, C. (2014), ‘Interdisciplinary approaches to the phenomenology of auditory verbal hallucinations’, Schizophrenia Bulletin, 40 (Supplement 4): S246–S54.

CHAPTER TWO

Corpus linguistics in illness and healthcare contexts: A case study of diabulimia support groups GAVIN BROOKES

1 INTRODUCTION Corpus linguistics is a group of methods that use specialist computer programs to analyse large collections of naturally occurring texts (McEnery and Hardie, 2012). These data sets are known as corpora (singular corpus, Latin for ‘body’) and are assembled with the aim of representing a language or linguistic variety on a broad scale. Due to the ease with which they can be analysed using computational methods, corpora are often much larger and more representative than the types of data that tend to be analysed using manual, purely qualitative methods of linguistic analysis. Indeed, corpora typically run into millions, and sometimes billions, of words in size, importantly providing researchers with the opportunity to base their analyses on more representative bodies of text and so to make their findings more generalizable to the social group, context or linguistic variety under study. The aim of this chapter is to provide a detailed introduction to corpus linguistics, in preparation for subsequent chapters in this volume, and to

CORPUS LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

45

demonstrate how corpus methods can be used to approach language in particular illness and healthcare contexts, showcasing some of the main benefits of applying corpus methods in this area. These aims will be met by way of a case study examining the language used in messages posted to an online support group concerned with a contested eating disorder known as ‘diabulimia’. The next section of this chapter provides an overview of the topic of diabulimia, detailing its medical status, prevalence and health implications for those who experience it. Following this, a more detailed introduction to corpus linguistics is given, before the focus of the chapter turns to the diabulimia case study, first introducing the data and then the analysis, which makes use of three established corpus techniques (keywords, collocation and concordance). The findings are then discussed in terms of their possible implications for medical practitioners caring for people with diabulimia in the future. The chapter concludes with a reflection on the utility of corpus methods for health(care) communication research, considering both their strengths and drawbacks in relation to this burgeoning field of linguistic inquiry.

2 DIABULIMIA IN CONTEXT Diabulimia is a contested eating disorder whereby people with type 1 diabetes deliberately reduce their insulin intake in order to shed calories and control their body weight. Although some people who experience diabulimia might conceive of and talk about it as if it were a medical disorder, the reality is that it is not recognized as such by practitioners and medical authorities, who instead tend to view it as an inappropriate compensatory behaviour, as well as a marker of deviance from prescribed diabetes self-management regimen (Sharma, 2013). The Diagnostic and Statistical Manual of Mental Disorders (DSM-5) (American Psychiatric Association, 2013: online), an authority on the classification of mental disorders, does not recognize diabulimia as a legitimate mental condition, but instead offers the following labels under which it might be classified: ‘inappropriate compensatory purging behaviour’, ‘misuse of medications for weight loss’, ‘bulimia nervosa’ and ‘eating disorders not otherwise specified’. The moniker diabulimia is therefore not a medically legitimate label but is a portmanteau of the words ‘diabetes’ and ‘bulimia’, invented by people experiencing diabulimia as a means through which to share their experiences of it and seek advice about it online (Goebel-Fabbri et al., 2008). Due to its contested status, it is not possible for a person to receive a diabulimia diagnosis from a medical practitioner. This notwithstanding, as many as 30 per cent of people with insulin-dependent diabetes are estimated to have intentionally restricted their insulin to control their body weight at some point in their lives (Goebel-Fabbri et al., 2008), with adolescents (Colton et al., 2009) and women

46

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

(Shih, 2011) understood to be the most affected groups. Diabulimia can lead to a number of negative health consequences for its sufferers, including diabetic neuropathy, kidney disease, diabetic retinopathy and increased susceptibility to heart attack and stroke (Mathieu, 2008). Research into the long-term effects of diabulimia suggests that life expectancy in those affected by it could be reducible by as much as thirteen years (Shih, 2011). As a consequence of its contested status, diabulimia is under-researched from both medical and social scientific perspectives (Hughes, 2010). Of those studies which have sought to provide insight into diabulimia, the majority has approached the topic from a decidedly positivistic perspective (Thorne, 1997), seeking to provide prevalence figures and to understand its biological consequences and potential causes. Although such studies provide an undoubtedly useful resource for health professionals seeking to gain an understanding of what diabulimia is, they reveal little about individuals’ subjective experiences and understandings of this emerging health phenomenon (Balfe, 2007). Meanwhile, that limited body of research which has explored lived experiences of diabulimia has based its insights on anecdotal evidence and researcher-invented accounts (Shih, 2011), meaning that, in empirical terms at least, we know very little about the perspectives of people who have lived experience of diabulimia. The case study reported in this chapter will help to address this knowledge gap by examining the language that people use to disclose and discursively construct their experiences and understandings of diabulimia in the context of three online diabetes support groups, at the same time demonstrating the usefulness of corpus linguistic methods for examining this type of healthrelated language data. Although language should not be treated as providing a transparent window into the minds and experiences of writers/speakers, for people can ‘design’ and even filter their linguistic output to suit their objectives, audiences and particular contexts of interaction (Bell, 1984), it seems fair to assume that the linguistic choices that the support group contributors make when talking about diabulimia do at minimum offer representations of their and others’ experiences and understandings of it. There is clear value in analysing such representations, not least because they have been produced by contributors for potentially significant purposes in this context, but also because they are consumed, reproduced and challenged by other members of the support groups under study, and so have the potential to shape those other members’ own understandings and experiences of this health issue. For the purposes of this analysis, then, there is clear value in utilizing corpus techniques, which will allow me to base my findings on a large collection of support group messages that is more likely than smaller data sets to represent the potentially diverse range of diabulimia-related perspectives shared by the members of these online communities. These advantages, and more besides, will be explored in more detail in the next section.

CORPUS LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

47

3 OVERVIEW OF CORPUS LINGUISTIC METHODOLOGY The data sets on which corpus linguistic analyses are based, corpora, come in two main types: general and specialized. General corpora, designed to represent language use on a broad scale, tend to be very large. For example, the British National Corpus (BNC), which was designed to represent general British English as used during the late 1980s/early 1990s, consists of 100 million words (90 per cent written and 10 per cent spoken texts) (Aston and Burnard, 1998). Specialized corpora, on the other hand, are designed to represent language in more specific contexts and so tend to be smaller than general corpora, though even specialized corpora can still amount to millions of words. For example, in their study of patient feedback about the National Health Service (NHS) in England, Brookes and Baker (2017) examined a corpus of 228,113 online patient comments (29 million words) posted to the NHS Choices website over a period of 2.5 years. Even as a specialized corpus, this data set afforded a more widely representative picture of patients’ concerns than would have been possible (or at least practical) had these researchers adopted a purely manual approach to their analysis. As a result, their findings were more generalizable to the patient population overall. Returning to the case study reported in this chapter, corpus methods therefore provide the option of exploring a greater number and wider range of subjective perspectives on the topic of diabulimia than would likely be apprehensible through the analysis of a small number of interviews or illness accounts, for example. In terms of its application to health(care) language, following an initial interest in written and then spoken discourse, recent corpus studies of health(care) communication have tended to focus on language produced in digital contexts, with research exploring online platforms for advice-seeking and -giving gaining prominence in recent years. For example, across a series of studies, Harvey and colleagues examined variations of a two-million-word corpus of emails sent by young people to a health website aimed at adolescents (Harvey et al., 2007; Harvey, 2012). Another popular area of focus among corpus studies of online health(care) interaction is online, particularly peerto-peer, support groups, for example Demmen et al.’s (2015) analysis of metaphors for cancer and end of life care in corpora representing support group interactions involving patients, family carers and healthcare professionals (see also Kinloch and Jaworska, this volume). An emerging area of focus for corpus studies of digital health(care) communication is patient feedback, for example Brookes and Baker’s aforementioned (2017) study of key concerns in a twenty-nine-million-word corpus of online patient feedback about the NHS in England (see also Loew et al., this volume, exploring GPs’ responses to an online questionnaire). This strong focus on digital discourse in corpus studies

48

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

of health(care) communication can be interpreted as reflecting the growing influence of digital technologies over the ways that people communicate – and indeed act – in relation to their health (Lupton, 2017). On a practical level, for the purposes of a corpus analysis, texts that have been produced within digital contexts are also eminently more convenient to collect than spoken and written texts and require minimal processing to be readied for computational corpus analysis. By exploring the discourse surrounding diabulimia in online support groups, the case study reported in this chapter can thus be viewed as harnessing the advantages of collecting and analysing readily digitized language data, all the while contributing to the growing body of corpus-based online health communication research. Once a corpus has been compiled, or an existing corpus selected, it is then analysed with the help of specialist computer programs, such as WordSmith Tools or AntConc (see Semino et al., this volume, for examples of other tools). Such packages, which are often freely available or purchasable for a relatively modest license fee, allow the researcher to search for every occurrence of any word or combination of words, generate frequency information about linguistic phenomena of interest (e.g. words, chains of words, grammatical types), perform statistical tests on those frequencies (i.e. to measure the significance or strength of relationships between phenomena) and present the data in ways that render it more amenable to manual analysis (Baker, 2006), meaning that mixed methods of analysis can be used. Indeed, corpus studies of health(care) language tend towards interdisciplinarity – incorporating perspectives from fields as diverse as medicine, psychology and sociology into their analyses (Brookes and Harvey, 2016). As well as reflecting the interdisciplinary nature of health(care) language approaches generally, this methodological and theoretical diversity also mirrors the diversity of corpus linguistics approaches, whose flexibility provides fertile ground for such multiplicity of methods and theoretical approaches (Hardie and McEnery, 2010). Corpus studies of health(care) communication are thus able to provide both quantitative and qualitative perspectives on data that represent a wider range of communicative routines, lived experiences and understandings of health(care) and illness than would likely be feasible using purely qualitative approaches, producing – by and large – more generalizable findings in the process (Crawford et al., 2014). Computer assistance can also be beneficial for the analysis itself, as it can reveal patterns that run counter to human intuition or which feature sparingly in one or two texts but become significant when considered as part of a larger collection (Baker, 2006). A related advantage of corpus methods is their ability to help analysts guide their analytical focus using more objective criteria, such as frequency information and statistical significance, thus reducing (though not eliminating entirely) the influence of human bias on their analyses. This increased objectivity is also supported by corpus linguistics’ commitment to

CORPUS LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

49

methodological transparency, which is underpinned by two guiding principles: (i) no systematic bias in the selection of texts included in the corpus (i.e. do not exclude a text because it does not fit a pre-existing argument or theory) and (ii) total accountability (all data gathered must be accounted for) (McEnery and Hardie, 2012). Combined, these principles can help analysts to overcome the accusation, often directed at more qualitative approaches to discourse analysis, that analyses are based on convenient texts or examples, cherrypicked by the researcher because they support a preconceived argument or theory (Widdowson, 2004). In this way, by affording the opportunity to examine large quantities of authentic language data, corpus linguistic methods might also be said to go some way towards appeasing the commitment to more objective approaches to large data sets that is commonplace in the domain of empirical health research (Brown et al., 2006), thus having the potential to help researchers to bridge the gap between social scientific and biomedical perspectives on health and illness.

4 DATA AND METHOD The case study presented in this chapter is based on a specialized, purpose-built corpus of messages posted to three English-speaking diabetes support groups. Support groups were sourced through a search engine query using the searchterms ‘diabetes support group’, ‘diabetes forum’ and ‘diabetes message board’. Of the top 100 search results, only three support groups met the following criteria for inclusion in the corpus: (i) English-speaking; (ii) dedicated to diabetes; (iii) hosts peer-to-peer, user-generated content (as opposed to practitionerdirected or practitioner-led content); (iv) not affiliated to a healthcare provider or charity (such sites are typically monitored by practitioners and other specialists); (v) meets ethical criteria (does not require registration to view content, explicitly informs users of the public nature of their contributions and does not explicitly discourage, or state requirement of permission for, the use of content for research purposes). From the three qualifying support groups, individual threads (chronologically ordered chains of messages) were included in the corpus if they contained mentions of the words ‘diabulimia’ and/or ‘diabulimic’ once in the thread title and/or three or more times across the messages contained within. Qualifying threads were included in the corpus in their entirety. The completed corpus is 119,982 words in size, comprising 81 threads and 1,072 messages posted

50

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

between 2007 and 2014. The corpus contains messages from a mixture of contributors who ostensibly do and do not present themselves as having firsthand experience of diabulimia, with the latter group consisting mainly of relatives and advice-givers. With the data available, I have little choice but to accept all the messages at face value. However, I am conscious of the possibility for contributors to falsely present either as having or not having diabulimia (in the former case, for example, to seek advice on behalf of another (Harvey, 2012)). With these caveats in mind, the corpus is most accurately described as representing the disclosure of experiences and understandings of diabulimia by contributors to three online, English-speaking diabetes support groups. Before progressing further, it is perhaps useful to briefly outline some of the ethical considerations that arise when using online data such as the support groups featured in my corpus. Debates concerning the ethics of online data collection began around the turn of the twenty-first century and continue to endure to this day. One area of contestation relates to whether online interactions should be regarded as belonging to the public or private domain (Elgesem, 2002). To this end, I wish to distinguish between those online spaces that can be accessed only by registered members who have ‘logged in’ and those that do not require a login but can be accessed publicly, by anyone. While we might consider websites or support groups which require users to register and log in to view content as constituting private domains of interaction (accessible to the members of those online communities only), those online spaces which do not require people to register and log in to view content might be considered to be more public than private (see also Coulson et al., 2007; Demmen et al., 2015). Thus, as per point (v) of the support group inclusion criteria outlined above, I included in my corpus only those support groups which I regarded as ‘public’ in as much as they did not require registration to view content, did explicitly inform users of the public nature of their contributions and did not explicitly discourage, or state requirement of permission for, the use of the content for research purposes (see also Eysenbach and Till, 2001). Due to the public nature of the messages, I did not seek informed consent from the support group contributors but have removed all mentions of names and any other identifying information (e.g. contact details, locations, etc.) in order to preserve the contributors’ anonymity. Ethical approval was obtained prior to data collection. As well as being suited to investigating a health phenomenon that was effectively born in cyberspace (Sharma, 2013), online support groups offer two significant methodological advantages for this case study. First, the support group threads were readily available in a digital format, which meant that they required minimal processing and could be imported into the corpus with relative ease. Second, as a relatively anonymized platform, online support groups afford their users opportunities for heightened levels of candour compared to

CORPUS LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

51

interactions taking place in offline contexts – what Suler (2004: 322) refers to as the ‘online disinhibition effect’. However, a limitation of sourcing data from anonymizing platforms is that it is not possible to access demographic information about the contributors, which meant that it was not possible to assess the demographic balance of the people who posted the messages included in my corpus. Consequently, my findings cannot be attributed to any specific group. It should also be noted that the contributors to the support groups in my corpus constitute a self-selecting sample, which means that the diabulimiarelated perspectives they provide might best represent the perspectives of people who have access to the internet and who choose to disclose their illness experiences online. It is important to acknowledge that there might be important differences between this group’s experiences and understandings of diabulimia and those of people who, for one reason or another, do not access or contribute to such groups.

5 USING CORPUS TECHNIQUES TO ANALYSE ONLINE SUPPORT GROUP MESSAGES The analysis in this case study was carried out in an exploratory fashion – with the aim of first identifying key themes in the data, before examining how one of those key themes (specifically, insulin) is linguistically constructed across the support group messages. There is no standard approach or set of procedures in corpus linguistic methodology. However, the forthcoming analysis will make use of three well-established techniques in corpus linguistics, namely keywords, collocation and concordance, all of which were accessed using WordSmith Tools Version 7 (Scott, 2016).1 The forthcoming case study will be reported in a way that serves to demonstrate how these techniques can be combined to at first achieve a quantitative overview of key themes in the data before exploring the linguistic construction of those themes in a more qualitative fashion. 5.1 Keywords My examination of the support group messages begins with keywords. Keywords are words which occur with a significantly higher frequency in the corpus being analysed when it is compared against another corpus (Scott, 1999). To generate keywords, we first have to generate a frequency list for all the words occurring in the corpus under study (obtained through the ‘word list’ function of WordSmith and most other corpus software packages). We then compare the word list for the corpus we are analysing – in this case, my corpus of diabulimia support group messages – against the word list for a comparable reference corpus, which typically represents a norm or ‘benchmark’ for the type of language under investigation. Words are deemed to be keywords based

52

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

on statistical comparisons of the word frequency information for each corpus, with keywords representing those words that occur considerably more often in the analysis corpus compared to the reference corpus. The computational method of generating keywords is therefore not based on concepts that are subjectively viewed as important to culture (see Williams, 1983), but allows for potentially any word to be a keyword, provided that it occurs frequently enough in the corpus under analysis compared to in the reference corpus. When selecting a reference corpus, we usually want one that is similar in size to, or larger than, the corpus being analysed. Ideally, the reference corpus should also represent language belonging to the same genre as the texts in the analysis corpus, so that the keywords will flag up what is lexically distinctive about the texts in the corpus compared to others of a similar type. To generate a set of keywords for this case study, I compared the word frequency list for my corpus of diabulimia messages against the word frequency list for the general language BNC (introduced earlier; downloaded from lexically.net). The BNC is a suitable reference corpus in this regard, as its combination of spoken and written registers can be said to represent a standard or benchmark for the partwritten/part-spoken character of computer-mediated communication (Baron, 1998).2 The resulting keywords were therefore those words which occurred with an unusually high frequency in my corpus compared to the BNC (in other words, compared to general British English). Another important decision we have to make when generating keywords regards which statistical test to use. Most corpus analytical software packages offer a number of statistical tests which measure either effect size or statistical significance. Effect size metrics indicate the strength of the difference or relationship we have found (i.e. higher scores indicating stronger differences/ relationships), while measures of statistical significance indicate the level of confidence the researcher can have that the difference or relationship observed is dependable and not the result of a sampling error (Gabrielatos, 2018). In other words, effect size indicates strength, while statistical significance indicates confidence. It is recommended that researchers combine metrics of effect size and statistical significance to make their results more robust. Accordingly, for this case study I generated keywords using Log Ratio (McEnery and Baker, 2016: 23), which combines a test of statistical significance (log-likelihood, Dunning, 1993) with a measure of effect size, which quantifies the strength of the difference between the observed frequencies, independent of the sample size. This statistic will produce as keywords those words whose relative (or ‘normalized’) frequency is significantly higher (beyond the 99.9 per cent threshold level) in my corpus compared to the reference corpus (BNC). Each statistically significant keyword is then assigned a Log Ratio score based on the size of the observed difference in relative frequencies, with bigger differences producing higher scores. This measure therefore uses log-likelihood as a cut-off

CORPUS LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

53

to ensure that keywords are significant but has the advantage that it also allows us to rank those keywords according to how unusually high, or ‘marked’, their frequency is in the corpus being analysed. Since Log Ratio privileges lowfrequency keywords, I also stipulated that keywords should occur in a minimum of 5 per cent of the texts in the corpus (n = 53). The top 20 keywords (an arbitrary cut-off) ranked by Log Ratio are displayed in Table 2.1. One way of approaching these keywords is to group them into semantic categories reflecting the most characteristic themes in the corpus. Thus, we might interpret these keywords as indicating themes such as diabulimia (‘diabulimia’, ‘disorder’), insulin (‘lantus’ (a commercial brand of insulin), ‘insulin’, ‘pump’),

TABLE 2.1 Top 20 Keywords, Ranked by Log Ratio Diabulimia corpus freq. Rank Keyword

BNC Freq.

N

%

Texts

200

0.17

146

N

%

Log Ratio

0

ry Nat< speaki:n:g_ (0.2) .hh Hi:,=>I was ↑wondering if I could make an ap↓pointment plea::s:e.hh (.) Is it something urgent or routi::ne. .hhh uh: well:- (.) uh: it’s not urgent, but I could do with seeing somebody if possible. Today:, (0.7) I’ve got a cancellation at ten past twel::ve,

R formulates the triaging as ‘Is it something urgent or routi::ne.’ (line 6), requesting that P choose between the two categories. R’s query is followed by P’s delayed and hesitant response in lines 7–8. In her response P resists the terms of receptionist’s question, by avoiding both a ‘yes’ and a ‘no’, and favouring her own view (with ‘well:’; Heritage, 2015). That is, although P does not regard her inquiry as urgent (‘it’s not urgent, but…’, line 7), she favours being seen sooner rather than later, and in this way treats her inquiry as non‘routine’, that is, as legitimate and non-trivial. Without addressing the level of urgency (or P’s condition) any further, R offers a same-day appointment in line 10, seemingly treating P’s inquiry (like P herself) as urgent.

282

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Excerpt 5 highlights some of the complexities around asking for urgent appointments: P resists non-urgency, and might also resist accounting for her appointment. At least in this case R has established that P considers her own enquiry as urgent enough for a same-day appointment. We examine one final excerpt in this section, in which the patient hears and understands, and rejects, the notion of ‘routine’ appointment. Excerpt 6: GP3 78 1 2 3 4 5 6 7 8 9 10 11 12

P: R: P: P: R: P: R:

Hello. ↑Could I make an ap↓pointment please to see a doctor. You can. Is it a routine appointment you’re requiring, (0.6) No. (.) °no°.= =(is it) something a bit more urgen:t, (0.7) Uh: (p) (.) it’s not urgent, but it needs looking at? (0.3) Okay,=↑Can I ask what the problem is >to see if< I can help where- where to put this appointment please.

P’s appointment request is met with an explicit granting, ‘You can.’ (line 2), leading to further specification of R’s request, this time with a polar interrogative ‘Is it a routine appointment you’re requiring,’ (line 2). Polar interrogatives typically make an account relevant if the answer is ‘no’ (Robinson and Bolden, 2010); however, asking just ‘is it routine’ leads to resistance and no account. P re-completes the sequence with a second ‘no’ in line 6, as a confirmation that no further action is taken on his part to expand on the response (Curl, Local and Walker, 2006; Sikveland, 2012). At this point, immediately following P’s second ‘no’ in line 7, R pursues a response with another polar interrogative, requesting P to confirm whether it’s ‘urgent’. Again, P resists categorizing his needs as ‘urgent’, as well as ‘routine’, but stresses the point (and reason for calling) that ‘it needs looking at?’ (line 9). Now, as P has resisted giving any information or claim of urgency, R goes on to ask more explicitly what P’s needs are (lines 11–12). In sum, we have shown that, when receptions initiate triaging, they often ask about (or presume) the likely routine nature of patient requests. Receptionists initiate triaging by either (a) formulating polar and declarative interrogatives containing the category ‘routine’, which prefer a confirmation or (b) using alternative interrogatives requiring patients to opt for either ‘routine’ or ‘urgent’. While receptionists and patients may be familiar with the categories ‘urgent’ and ‘routine’, patients are reluctant to categorize their needs with

EFFECTIVE TRIAGING IN GENERAL PRACTICE RECEPTIONS

283

either term. We may speculate whether ‘routine’ is not a description patients themselves would use to account for their calling the doctor (i.e. their reason for calling is more important than ‘routine’), and whether ‘urgent’ lies at the other extreme – again not quite fitting with the patient’s reason for calling the doctor (as opposed to phoning 999, the emergency telephone number in the UK). In any case, asking patients to confirm ‘urgent’ or ‘routine’ seems ineffective. So how can triaging be done more effectively? In Section 4.2 we see how a reference to ‘urgent’ can be productive in soliciting an account, which informs the triaging rather than hindering it. 4.2 Effective triaging: Soliciting accounts when patients request a same-day appointment In this section, we report alternative ways in which receptionists initiated triaging without the ‘routine/urgent’ distinction, and instead solicited an account for patients’ needs for a same-day appointment. The first, and most direct way, receptionists did so was by formulating a wh-interrogative, for example, ‘what is it for?’. In Excerpt 7, P is phoning to book an appointment for her child, and R initiates the triaging with ‘May I ask what the problem is today plea:se,’ (line 3). Excerpt 7: GP3 240, 0:07-0:19 1 2 3 4 5 6 7 8 9 10 11

P: R: P: R: P:

P:

Can I make an appointment for today. Please. (0.5) May I ask what the problem is today plea:se, (0.5) Yeah: it's for my: baby.=He's got- (.) really really bad nappy rash. #Oh right#.= =Uh:m we're on about day three now and it just >doesn't seem to be getting any< better. (0.3) Could you do eight fifty.

P responds to R’s wh-interrogative, first by confirming R’s ‘may I’ (‘yeah:’, line 5), then providing ‘what’ her reason is for making an appointment (lines 5–6), which she further supports in lines 8–9 by highlighting the continued presence of her child’s condition. Based on this example, we may argue that wh-interrogatives are an effective way of soliciting an account. In Excerpt 8, however, the nature of the wh-interrogative becomes a problem for the patient. P is phoning to book an appointment on behalf of her husband.

284

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Excerpt 8: GP3 89, 0:03-0:15 1 2 3 4 5 6 7 8

P: R: P:

R: P:

Hiya, I’m wondering if I can make an appointment for my ↓husband plea:se. Okay,=What seems to be the problem [today:.] [Uh:m:- ] He’s a- °ahh af° (.) he’s I want to make it for- I need an appointment- uh a routine appointment as a l- as late on in the day as possible,= =Okay.=Just bear with [me? ] [plea]se_

P delays her response in line 4 (‘Uh:m:-’), followed by two aborted phrasal initiations: ‘He’s a-’ (line 4), ‘he’s’ (line 5), following which P redirects the trajectory of her turn towards the need for an appointment: ‘I want to make it for- I need an appointment-’ (line 5). The continued stops and restarts suggest trouble in producing an answer not fitted to the question. As P proceeds, she ends up specifying her request to time of day, we also note that P categorizes the appointment herself as ‘routine’ (line 6). While patients may resist (‘just’) ‘routine’ as formulated by receptionists, Excerpt 8 provides evidence that patients know what this institutional category means and may use it (here, as a resource to avoid any further pursuit of her husband’s condition). In Excerpts 9 and 10, we show a different way to solicit an account from patients, which avoids contrasting ‘routine’ or ‘urgent’ categories and is less risky than asking an entirely open wh-question as in Excerpts 7 and 8. Excerpt 9: GP3 120 1 2 3 4 5 6 7 8

R: P: R: P: R:

Good afternoon, surgery, Claire speaking. Hiya,=>I was wondering< if you’ve got any appointments=either .hhh (.) like later on today o:r tomorrow morning, ↑Is it something urgent that you’re requiri:ng. (0.3) Uhm I’ve g- I think I’ve got a chest infection and I need an inhaler. N:ot be a second.

P has already indicated ‘urgency’ in her inquiry, by specifying a requested time (lines 2–3): ‘later on today o:r tomorrow morning,’. While R might be required institutionally to initiate triaging, there are also interactional features in P’s inquiry to support doing so: P has explicitly requested a same-day appointment. R asks, ‘Is it something urgent that you’re requiri:ng.’ (line 4). Following a gap of 0.3 seconds (line 5), and an ‘Uhm’ (line 6), P gives details of her condition (‘I’ve got a chest infection’) and needs (‘I need an inhaler.’) in lines 6–7. This is accepted by R as sufficient reason for a same-day appointment in line 8, as she indicates a go-ahead on the appointment booking.

EFFECTIVE TRIAGING IN GENERAL PRACTICE RECEPTIONS

285

Note that, compared to earlier examples, the patient more straightforwardly supplies an account without resisting the ‘urgent’ category proposed by R. And compared to the polar interrogatives in some of the earlier examples (‘is it just a routine appointment’), which requested confirmation, this time the query takes the form of a polar interrogative not simply requesting confirmation. Specifically, the receptionist’s use of ‘something’ seems to work in favour of soliciting an account rather than just a ‘yes’/’no’ (dis)confirmation. This supports previous research, which has shown how replacing ‘any/anything’ with ‘some/something’ in doctors’ questions to patients regarding additional concerns (i.e. ‘Is there (some/any)thing else you would like to address in the visit today?’), significantly increases patients’ expression of unanticipated concerns (Heritage et al., 2007). This is because a ‘yes’ response implies there is an additional concern; and whereas ‘any/anything’ is negatively polarized and thereby grammatically fitted with a ‘no’ response, ‘some/something’ is positively polarized and grammatically fitted with a ‘yes’ response. On this basis we propose that positively polarized features such as ‘some(thing)’ are well suited for soliciting such an account in questions, including ‘is it [something] urgent’. Similarly, in Excerpt 10, the receptionist solicits the patient’s account. Excerpt 10: GP3 256 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

R: P:

R: P:

R:

P: R:

Good morning,=↑Surgery.=Ree speakin’, (0.6) ↑Hiya I'm just wondering if it's possible if I can get an appointment this mornin’. (0.6) Y:e=Is it something ↑urgent for today duck¿=Or is [it- ] [.hhh]hh YEAH I've uh:m: (0.8) .ptk (0.2) I’ve had it for about four days now.=B#ut: I've# got a really bad th:roat and ch:e:st,=>and I've-< I'm asthmatic so I'm not sure if I've got chest infection. Alright duck >let me have a< look for you,=↑Yeah can you get in this ↑morni:n’ ten o'clo:ck¿ (0.6) YEAH that's fine [t h a n k y o u : . ] [What's your name then] plea:se.

As R projects a next turn constructional unit with ‘Or is it’ in lines 6–7, and thus deleting any particular preference to the question in the first part of her turn, P responds to ‘Is it something ↑urgent for today duck¿’ in overlap, with an emphatic confirmation ‘YEAH’ followed by an account: P highlights the duration of the symptoms, and also their severity (‘got a really bad th:roat and ch:e:st,’).

286

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

Excerpt 11 shows a less emphatic response, with a comparatively downgraded ‘He’s not >very well< with his stomach.’ (P is calling on behalf of her partner.) This is followed by expanded triaging queries, and eventually the booking of a same-day appointment. Excerpt 11: GP3 73, 0:03-0:29 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20

P: R: P:

R: P: R:

P:

R: P:

(Hi), Can I make an appointment for this morning please. (0.2) #Yeah# is it something urgent that you requiring? (0.6) Uh:m yeah it’s >for me< partner.=He’s not >very well< with his stomach. (0.3) °Okay°,=Has he seen anyone with it befo:re¿ (0.5) N:o he’s not no. °Right°.=.hh when you say with his stomach, >has he got< tummy pain or is he being ill. (0.7) Uh:m: both really.=↑It’s li:ke- (.) ↓uh how can I say it. I ↑think it’s summat to do with his acids more so because he:’s: uh:m: everything smells quite eggy, (0.5) >No worriesI’m just wondering if< a doctor could ring me ba:ck. (0.3) Something urgent for toda:y? (1.2) N:o I just need some more tablets.=I’m running out.

In sum, we have seen that when receptionists use formats such as ‘is it something urgent for today’, they may solicit a confirmation plus account from patients, rather than the resistance that follows ‘is it just a routine appointment’, and

EFFECTIVE TRIAGING IN GENERAL PRACTICE RECEPTIONS

287

when contrasting ‘urgent’ with ‘routine’ categories. We have shown how this distinction can have implications for the progressivity of the call. We argued for the role of ‘something’ in soliciting a further expansion. A key difference between ‘is it something urgent for today’ and ‘is it just a routine appointment’ is that the first design orients to and affords patient needs, in other words it assumes patients have a legitimate reason for calling (cf. Murdoch et al., 2014). 4.3 Flexible triaging: How patients indicate urgency in their first turn In the previous sections, we have seen that patients may request to be seen on the same day, prior to or following receptionist-led triaging. In this section, we focus on the patients’ first turn pre-triaging, to establish whether and how receptionists may spot urgency based on the patients’ formulation of their inquiry. Putting to one side institutional requirements (i.e. the practice policy to initiate triaging), we examine the interactional evidence in support (or not) of triaging, and what makes sense from the patient’s perspective. We will show first that patients may ask for an appointment ‘today’ in their first turn, in which case asking questions about ‘urgency’ become legitimate. In other cases, however, there is no such indication of urgency in the patient’s first turn. In these cases, patients generally accept a future appointment, especially if that appointment is within the next three to four days. Based on our finding that patients themselves make distinctions of urgency available to the GP receptionist in their first turn, we argue that triaging should be used flexibly. Doing triaging by default is not sufficiently attentive to patients. In Excerpt 13, the patient indicates urgency in their first turn; however, the receptionist goes on to treat the inquiry as non-urgent. Excerpt 13: GP2 146, 0:03-1:00 1 P: Hello.=Can you tell me if there’s any available appointments 2 for this afternoon please. 3 (0.5) 4 R: Just bear with me one second? 5 (15.8) 6 R: And what’s your date of bi:rth, ((5 lines omitted)) 12 R: .tk uh the first free slot I’ve got >coming up< is (0.2) 13 at nine fiftee:n, Is that any good to you?= 14 P: =I need one this uh toni:ght, I think I’ve got a chest 15 infection. #Okay,# so it’s medically urgent for today is it? 16 R: 17 P: Please. 18 (7.2) 19 R: .ptk I’ve got twenty five past four this afternoon with Doctor 20 Taylor,

288

APPLYING LINGUISTICS IN ILLNESS AND HEALTHCARE CONTEXTS

P requests an appointment for ‘this afternoon’ in lines 1–2. R does not initiate triaging in response, but shows that she is looking for availability on the system. R then offers the ‘first free slot’ (which is not ‘this afternoon’) at lines 12–13, and asks P if that is ‘any good’ (note the negatively polarized ‘any’, which prefers ‘no’). P then upgrades her need for a same-day appointment (‘I need one this uh toni:ght’) and supplies an account (‘I think I’ve got a chest infection.’, lines 14–15). R then formulates the upshot of P’s request, which also initiates triaging, with a confirmation request: ‘so it’s medically urgent for today is it?’. For comparison, the patient in Excerpt 14 asks for an appointment without specifying day or urgency. Excerpt 14: GP2 314 1 2 3 4 5 6 7 8

P:

9

P:

R: P: R:

Hello love,=↑Can I make an appointment please love, (0.5) Yeah(p), (.) .pthhh a:::nd just a seco:n:d, (1.3) I've got Friday morning or afternoon on Monday? (0.5) Yeah Friday will do fine love.=That's fine.

P does not specify time in the initial inquiry, but asks, ‘Can I make an appointment please love,’ (line 1). Arguably, there is little evidence of urgency here, and R goes on to offer the next appointment which is Friday on the same week, or Monday the week after. (We do not know what day the patient is calling; however, it seems to be at least two days before ‘Friday’.) Thus, if patients do not ask for an immediate appointment, offering the next available one is the appropriate thing for receptionists to do next – there is no need to initiate triaging. Excerpt 15 is similar. Excerpt 15: GP1 75 1 2 3 4 5 6 7 8 9

R: P: R: R: P:

Good morning,=Limetown Surgery:, (0.5) #Ah# goo’ mor’ing. (.) A#h#, I’m ringing to make an appointment with a nurse please, to have my ears syringed. Ohkhay, (0.5) Bear with me a mo’#e#nt, #le#t me just find the next appointmen:t, .mh[h ] [‘K] you.

EFFECTIVE TRIAGING IN GENERAL PRACTICE RECEPTIONS

289

Here, P’s request does not specify a time or that she requires this service today, and appears happy with R’s offer to ‘just find the next appointment’ (note her truncated ‘thank you’ at line 9). Finally, in Excerpt 16, P designs her request as non-urgent from the start of the call. Excerpt 16: GP2 21 1 2 3 4 5 6 7 8 9 10 11

R: P: P: R: R: P:

Good afternoon, Reception,=Melanie speaking? (0.9) Hello. Uh: (.) ↑can I make an appointment plea:se, (.)