<?page no="0"?> www.narr.de Buchreihe zu den Arbeiten aus Anglistik und Amerikanistik ISBN 978-3-381-12821-1 Why do people in online forums say who they are not rather than who they are - especially when offline identities are unverifiable or irrelevant? And how do such linguistic moves reflect wider struggles over belonging and representation in digital publics? This book explores how negative self-identifiers - variants of the template “I’m not a…” - function as a linguistic strategy in Englishlanguage web forums, revealing how users present themselves while negotiating shared interests and epistemic positions. It starts from the premise that identity is something we do - a linguistic performance shaped by context - and takes a critical interest in how the interactional micromanagement of identity relates to its broader sociopolitical dynamics. Adopting a corpus-pragmatic approach, the study examines how negative self-identification positions speakers in relation to their claims and foregrounds identity features that matter in the moment, critically connecting micro-level linguistic practice with wider debates on discourse, identity, and the politics of digitally mediated representation. Triebl The Pragmatics of Negative Self-Identification: I Am What I’m Not Eva Triebl The Pragmatics of Negative Self-Identification: I Am What I’m Not A corpus-based study of UK web forums <?page no="1"?> The Pragmatics of Negative Self-Identification: I Am What I’m Not <?page no="2"?> Buchreihe zu den Arbeiten aus Anglistik und Amerikanistik / Agenda: Advancing Anglophone Studies Herausgegeben von / edited by Alexander Onysko, Ulla Ratheiser, Werner Delanoy Mitherausgeber: innen / Editorial board: Sibylle Baumbach, Marcus Callies, Marta Degani, Alwin Fill, Walter Grünzweig, Sarah Herbe, Walter Hölbling, Julia Hüttner, Allan James, Cornelia Klecker, Ursula Kluwick, Benjamin Kremmel, Andreas Mahler, Christian Mair, Georg Marko, Frauke Matz, Simone Pfenninger, Peter Siemund, Ute Smit, Laurenz Volkmann, Max von Blanckenburg, Werner Wolf, Libe García Zarranz Erstherausgeber / Founding editor: Bernhard Kettemann Band 29 <?page no="3"?> Eva Triebl The Pragmatics of Negative Self-Identification: I Am What I’m Not A corpus-based study of UK web forums <?page no="4"?> Ö S T E R R E I C H I S C H E F O R S C H U N G S G E M E I N S C H A F T ÖFG DOI: https: / / doi.org/ 10.24053/ 9783381128228 © 2025 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Alle Informationen in diesem Buch wurden mit großer Sorgfalt erstellt. Fehler können dennoch nicht völlig ausgeschlossen werden. Weder Verlag noch Autor: innen oder Herausgeber: innen übernehmen deshalb eine Gewährleistung für die Korrektheit des Inhaltes und haften nicht für fehlerhafte Angaben und deren Folgen. Diese Publikation enthält gegebenenfalls Links zu externen Inhalten Dritter, auf die weder Verlag noch Autor: innen oder Herausgeber: innen Einfluss haben. Für die Inhalte der verlinkten Seiten sind stets die jeweiligen Anbieter oder Betreibenden der Seiten verantwortlich. Internet: www.narr.de eMail: info@narr.de Druck: Elanders Waiblingen GmbH ISSN 0939-8481 ISBN 978-3-381-12821-1 (Print) ISBN 978-3-381-12822-8 (ePDF) ISBN 978-3-381-12823-5 (ePub) Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http: / / dnb.dnb.de abrufbar. The appendices are available at https: / / files.narr.digital/ 9783381128211/ Appendices.pdf <?page no="5"?> For Mira—my reason to be my best. <?page no="7"?> 11 13 19 1 21 1.1 21 1.2 23 1.3 24 2 27 2.1 27 2.1.1 28 2.1.2 29 2.1.3 30 2.1.4 31 2.1.5 33 2.2 34 2.2.1 35 2.2.2 41 2.3 43 2.3.1 44 2.3.2 48 2.4 51 3 53 3.1 53 Contents Conventions and Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Background to and motivations for the study . . . . . . . . . . . . . . Research questions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Structure of the book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Identity, (Negative) Self-identification, and the Social World . . . . . . . Conceptualizing identity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Identity as an individual project . . . . . . . . . . . . . . . . . . . Identity as a product of the social . . . . . . . . . . . . . . . . . . Identity as a discursive construction . . . . . . . . . . . . . . . From linguistic (non-) identification to the material social world . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Conceptualization of identity in this study . . . . . . . . . . Conceptual struggles around negative self-identification . . . Contemporary notions and representations of expertise Expertise, (dis-)trust and populism . . . . . . . . . . . . . . . . . The discourse context: Web forums as sites for studying negative self-identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Researching digitally mediated discourse . . . . . . . . . . . Studying identity and (non-)identification in web forums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Theoretically Contextualizing Negative Self-Identifiers . . . . . . . . . . . Studying forms, meanings and functions . . . . . . . . . . . . . . . . . <?page no="8"?> 3.2 54 3.2.1 54 3.2.2 56 3.2.3 60 3.3 65 3.4 67 3.4.1 68 3.4.2 71 3.5 73 4 75 4.1 75 4.2 82 4.2.1 84 4.2.2 85 4.2.3 90 4.2.4 95 4.2.5 98 4.2.6 100 4.3 103 4.4 108 5 111 5.1 111 5.2 114 5.2.1 115 5.2.2 117 5.2.3 118 5.2.4 125 5.3 126 Theorizing negatives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Negation in formal semantics . . . . . . . . . . . . . . . . . . . . . Formal realizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Scope and presupposition . . . . . . . . . . . . . . . . . . . . . . . . . Theorizing the identifying noun phrase . . . . . . . . . . . . . . . . . . The structure “I + copula + NOT + identifying NP” in British English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . A written discourse completion task on negative self-identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The use of the structure in the Spoken BNC2014 . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functionally Conceptualizing Negative Self-Identification . . . . . . . . Negative self-identifiers as speech acts . . . . . . . . . . . . . . . . . . . Negative self-identifiers as context-dependent meaning potentials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pragmatic context models . . . . . . . . . . . . . . . . . . . . . . . . Context in (interactionally oriented) sociolinguistics and variational pragmatics . . . . . . . . . . . . . . . . . . . . . . . . Context in conversation analysis . . . . . . . . . . . . . . . . . . Context in systemic functional linguistics . . . . . . . . . . . Context in corpus pragmatics . . . . . . . . . . . . . . . . . . . . . Context in studies of digitally mediated discourse . . . . Negative self-identifiers and (corpus-based) critical discourse studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Empirical Study: Data and Method . . . . . . . . . . . . . . . . . . . . . . . . Theoretical and methodological principles . . . . . . . . . . . . . . . Data selection and collection . . . . . . . . . . . . . . . . . . . . . . . . . . . . Creating a derived corpus of negative self-identifiers . Creating a corpus with Google searches . . . . . . . . . . . . Sampling criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Ethical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Contents <?page no="9"?> 6 129 6.1 129 6.1.1 129 6.1.2 138 6.2 139 6.3 153 6.4 153 7 157 7.1 157 7.2 162 7.2.1 163 7.2.2 170 7.2.3 171 7.2.4 180 7.2.5 181 7.2.6 184 7.2.7 187 7.3 188 7.3.1 189 7.3.2 197 8 201 8.1 202 8.1.1 207 8.1.2 211 8.1.3 213 8.2 213 8.2.1 216 8.2.2 224 8.2.3 239 8.2.4 249 8.3 254 The Empirical Study: Analytical Approach and Annotation . . . . . . . A formal-functional framework of negative self-identifiers in use . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relations with the co-text . . . . . . . . . . . . . . . . . . . . . . . . Analytical model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Corpus annotation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Data presentation and transcription conventions . . . . . . . . . . Querying and analyzing the annotated corpus . . . . . . . . . . . . . Conceptually Profiling Negative Self-Identifiers . . . . . . . . . . . . . . . . . Theoretical considerations and methodological principles . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Preference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Habits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Expertise and Professionalism . . . . . . . . . . . . . . . . . . . . . Activities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Usage/ consumption/ ownership . . . . . . . . . . . . . . . . . . . . Summary and implications of results . . . . . . . . . . . . . . . . . . . . . Conceptual landscape of identifying noun phrases . . . Implications: “I might not be an expert, but I know what I do not want” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Negative Self-identifiers and Their Immediate Co-Texts . . . . . . . . . . General co-textual profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Clause-internal modification . . . . . . . . . . . . . . . . . . . . . . Negative self-identifiers in their clause-external co-text Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Negative self-identifiers and their formally related co-texts . Profiling co-textual elements: method and categorization principles . . . . . . . . . . . . . . . . . . . . . . . . . Contrast and concession . . . . . . . . . . . . . . . . . . . . . . . . . Cause and consequence . . . . . . . . . . . . . . . . . . . . . . . . . . Addition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contents 9 <?page no="10"?> 9 259 9.1 260 9.1.1 260 9.1.2 278 9.1.3 285 9.1.4 296 9.2 300 9.2.1 302 9.2.2 306 9.2.3 313 9.3 313 9.3.1 313 9.3.2 316 9.3.3 320 10 323 10.1 324 10.2 326 10.3 328 10.4 334 10.5 336 11 339 373 375 The Discourse Functions of Negative Self-identifiers . . . . . . . . . . . . Functional analysis of formally related co-texts . . . . . . . . . . . Functional categorization of mental processes with I as the senser . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functional categorization of relational processes with I as the carrier/ token . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functional categorization of material processes with I as the actor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functional analysis of co-texts preceding negative self-identifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Functional categories of preceding co-texts . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Qualitatively studying negative self-identifiers in context . . . Linking micro-level functions and macro-conceptualizations . . . . . . . . . . . . . . . . . . . . . . . . . Disclaiming expertise . . . . . . . . . . . . . . . . . . . . . . . . . . . . Disclaiming preference . . . . . . . . . . . . . . . . . . . . . . . . . . . Discussion and Conclusion: I Am What I’m Not . . . . . . . . . . . . . . . . . From patterns to profiles: A functional overview . . . . . . . . . . . Self-representation as epistemic positioning . . . . . . . . . . . . . . . Between expertise and opinion: Authority and identity in online forums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Contributions: Method, theory, and social insight . . . . . . . . . . Looking forward: Reflections and research horizons . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Contents <?page no="11"?> Conventions and Abbreviations Throughout this book, corpus examples are reproduced verbatim as far as possi‐ ble. To preserve data integrity while ensuring clarity, the following conventions are used: General notational conventions • Italics — original text under analysis or highlighted linguistic items. • Bold — in corpus examples, highlight negative self-identifiers (NSIs). • Bold italics — in tables, the lemma heading when listing multiple attributes for the same verb (e.g., be: difficult, pointless). • verb: attribute/ value — attributes or values listed after the verb (e.g., be: 0 degrees). • (n) — token frequencies in tables (shown in italics). • (n = X) — token frequency in table captions (shown in italics). • [NAME] — anonymized usernames or identifying details. • [word] — editorial corrections of typos or minimal additions. • [function] — bracketed function tags after pronouns (e.g., it [impersonal]) label grammatical function and are not editorial insertions. • […] — omitted material. • [QUOTE], [PICTURE] — non-authorial text or non-verbal elements. Chapter 8 (transitivity analysis) • get [become] — verb-sense disambiguation. • these [caves] — referent supplied from proximate co-text; ([anaphoric ref] for distal). • it [impersonal], there [impersonal] — impersonal uses of pronouns. Analytical abbreviations • CA = Conversation Analysis • NP = noun phrase • NSI = Negative self-identifier • SFL = Systemic Functional Linguistics • To. = token (individual occurrence in the corpus) • Ty. = type (unique lexical item or construction) • sb. = somebody (used in glosses) • sth. = something (used in glosses) <?page no="13"?> Foreword Negative self-identifiers (NSIs) of the type “I + copula + NOT + indefinite NP” (noun phrase) are a fascinating starting point for studying the links between linguistic micromanagement of identity in interaction and broader questions about the performance and conceptualization of the self in digitally mediated contemporary social life. By mobilizing categories of negative self-identification in anonymous, disembodied online encounters, users indicate which identi‐ ties they perceive to be situationally relevant. Thus, NSIs can be seen as a conversational resource (Ekström & Stevanovic 2023) that offers a reflexive comment on the speaker’s perception and stance toward what is interactionally accomplished. Doing, rather than having, an identity could also be seen as part of a broader shift in contemporary social life, challenging traditional notions of selfhood and its relation to language. United by topics of shared individual in‐ terest, people interacting online linguistically perform identity through affective alignment with community norms (Little et al. 2003; Androutsopoulos 2014; Tagg et al. 2017). Since the study presented in this book was conducted, forums such as Reddit have revived and gained significance, and community norms are increasingly materializing in platform architecture, abstracting and reflexively shaping forum-specific interaction even more, and deemphasizing or renegotiating the role of local, material boundaries such as nationality, language, and gender (Page et al. 2014). Thus, forum interaction simultaneously foregrounds the importance of (shared) individuality (Lüders et al. 2022) and platform-specific perceptions of authenticity - that is, being different and unique (Giddens 1991; Beck & Beck-Gernsheim 2001; Leppänen et al. 2015). Against this background, the question that initiated this study appears even more relevant from today’s perspective: Which identification categories are felt to be relevant enough to be disclaimed, in which contexts do these occur, and what are their functions? The answer provided by this study also paves the way for research into a further question that has since gained significance: How do forum users strategically position themselves to authenticate their claims (Sierra 2023) and connect with others in digitally enabled conversations online? In light of recent sociopolitical developments, the link between identity, authenticity, and credibility has become increasingly salient. The confrontational nature of anonymous, digitally mediated interaction and the heightened risks associated with identification - not just who is writing, but whose authority their <?page no="14"?> 1 danah boyd stylizes her name in lowercase. This typographic choice is respected in this book, following her own usage (boyd 2011). voice is taken to represent - only reinforce the importance of exploring this connection. This book develops a theoretical and methodological framework to qualita‐ tively and quantitatively identify the multiple functions of NSIs, a context-de‐ pendent meaning potential (Ochs 1996; Aijmer 2013). It exemplifies how a corpus-based pragmatic study of a single formally defined structure can reveal patterned linguistic choices and provide windows onto speakers’ metapragmatic awareness, that is, their evaluations of what is pragmatically appropriate or ‘sayable’, and by whom (Silverstein 2003; Spitzmüller & Warnke 2011), in online contexts (Marwick & boyd 2011). 1 Thus, it represents an approach to discourse analysis that links meticulous linguistic analysis to broader questions about language and the social world, assuming that the “big things reside in the small things, and the most inconspicuous and uniquely situated social action is, in that sense, ‘systemic’ and ‘typical’” (Blommaert et al. 2018: 5). The data for this study is a corpus of 936 instances of NSIs in English, used on UK-based web forums. A web forum is defined as a “[w]eb application for publishing user-generated content under the form of a discussion” (Morzy 2013: 1) that “brings people together with [a] shared interest and mind-set” (Biriyai & Emmah 2014). At the time of data collection in 2019, traditional forums were already considered a dying medium (Dee 2017; Sterling 2017), and the now-exploding visibility of reimagined forums like Reddit - which, at the time of writing, boasts over 108 million users and more than 100,000 active communities (Reddit 2025) - was not yet on the radar. Importantly, the study also predates a series of major sociopolitical and digitally amplified crises that have since aggravated epistemic uncertainty, un‐ dermined trust in traditional institutional sources of knowledge, and intensified ideological polarization driven by algorithmic recommendation systems (Chen et al. 2021; Muñoz et al. 2024). At the time these (now largely obsolete) forums were studied, bots did not participate in interactions, algorithms did not yet filter or distort participation, and research still relied heavily on manual coding rather than artificial intelligence (AI)-supported analysis. In this sense, the present book offers a snapshot of corpus-pragmatic research on digital discourse just before everything changed. However, this does not diminish the significance of its results or design. On the contrary, when viewed against the backdrop of paradigmatic shifts in public discourse at the broadest societal level, the study’s findings - modest in 14 Foreword <?page no="15"?> statistical significance though they may be - appear all the more noteworthy. Shortly before a state of “permacrisis” (Musolff 2023; Bradford et al. 2024; Edmiston et al. 2025) came to define a “new normality” and began to undermine consensus on a “shared reality” (Wehrle 2023; Farkas & Schou 2023), this study already revealed discursive struggle (Fairclough 2010) over epistemic certainty and credibility in anonymous expert communities (Au & Eyal 2022). Two paradigms of identification were found to stand out in the data: expertise on the one hand, and preference on the other. Functional analyses revealed that both types of NSIs serve tightly intertwined epistemic and social functions. The formulaic phrase I’m not an expert was found to be routinely used to metapragmatically frame subjective opinions as non-absolute, thus allowing speakers to perform their individually accrued expertise while formally cancel‐ ing the power differential implied by metadiscursive processes of explaining, rationalizing, and assessing information (Silverstein 2003). Conversely, negative self-identification with highly specific preferences emerged as a powerful way of indexing awareness of choice, indicating that performing, rather than claiming expertise through linguistically demonstrating experiential knowledge could be a socially preferable strategy of enacting expertise while aligning with fellow users. Thus, the performance of expertise in online forums appears to be contingent on users indexing their ability to navigate the pluralizing and ever more specialized systems of knowledge (Grundmann 2017). Beyond presenting novel insights about the pragmatics of online interaction and linking them to questions of key societal relevance, this book offers a transparent account of how a research project situated at the interface between pragmatics ( Jucker et al. 2018) and critical discourse studies (CDS) (Unger et al. 2021) can be conceptualized and implemented. Building on research first developed in my doctoral thesis, it traces a trajectory from theorizing a structure with meaning potential, through the compilation of a corpus and its software-assisted analysis, to the critical contextualization of results. Conducted by a single researcher within relatively tight temporal and financial constraints, the project is presented in a way that does not shy away from acknowledging the caveats and obstacles that may arise when independently carrying out original research. Thoroughly grounded in relevant linguistic and discourse-theoretical frameworks, the book synthesizes theories about the social world, identity, and social media to explain why negative self-identification is a pragmatically noteworthy choice that can be examined to learn about conceptualizations that structure meaning-making in online interaction. The study contributes methodologically by demonstrating how micro-level analysis can be combined with explicit reflection on rigour and limitations. It qualitatively analyzes Foreword 15 <?page no="16"?> a self-compiled, secondary corpus of web language, employing a range of concordance programs and showing how this kind of data can be manually and automatically annotated to quantify insights about semantic meanings and pragmatic functions. Since this study was conducted, and at the time of writing this foreword, there has been enormous technological progress, especially with major advan‐ ces in artificial intelligence. Collecting and analyzing comparable social media data today would involve different methods - and probably more effectively so. For example, large language models like ChatGPT (OpenAI 2025) can write code to enable much faster and more systematic data collection, and the latest concordance software such as SketchEngine (Kilgarriff et al. 2014) allows for much easier and more automated processing and visualization of linguistic data. Despite these advances, this study may be of particular interest to readers seeking to expand their knowledge of methodological decisions that lie beneath analyses seemingly effortlessly performed by software. For example, the book discusses challenges involved in systematically and transparently establishing criteria for categorization of natural language, highlighting that language - including metalanguage and analyses based on it - is a choice, which in turn is affected by contextual factors. Just as different researchers would code data slightly differently, automated linguistic analyses are not nec‐ essarily unbiased. The study could also be of interest to those interested in how texts can be systematically analyzed for communicative accomplishments, as it draws together notions that may be employed to categorize and measure discourse units at and beyond the clause level and their formal-functional relationships. By offering meticulous linguistic analyses of discourse around NSIs and situating the results in their immediate and wider social contexts, this book makes a methodological and conceptual contribution to critical linguistic research on the relationship between language, identity, and knowledge as claimed and negotiated on social media. It raises questions about the semiotic strategies that legitimize expert identities in the ‘post-truth’ era (O’Callaghan 2020) and the social practices through which opinions are formed (Cano et al. 2024). In this sense, this study not only captures a particular moment in the evolution of digital discourse but also offers conceptual tools for understanding how language practices both shape and respond to shifting sociotechnical conditions. While the forums analyzed may no longer exist in the same form, the dynamics they reveal - around identity, expertise, and epistemic positioning - remain central to contemporary online life. As digital platforms continue to 16 Foreword <?page no="17"?> evolve, the questions posed here gain new urgency, and the analytical approach developed may inform future investigations into the discursive negotiation of selfhood, credibility, and community in a rapidly changing communicative landscape. Vienna, September 2025 Eva Triebl Foreword 17 <?page no="19"?> Acknowledgments Writing a book is never the work of one person, just as research is always as much about people and friendships as it is about discovery. I have been incredibly lucky to work with unique, knowledgeable, and kind people in the process leading up to this publication. While it is impossible to mention everyone I am grateful to, I would like to express my gratitude to the Österreichische Forschungsgemeinschaft (ÖFG) for their financial support of this publication, and to their anonymous reviewer for their helpful and constructive feedback on the submitted manuscript. I would also like to thank Alexander Onysko, who encouraged me to turn what was originally a PhD thesis into a book, and my best teacher, Georg Marko. It is scholars like them who truly support early career researchers. I am also deeply thankful to my students, whose curiosity, trust, and enthusiasm not only sustain my interest in digitally mediated discourse but continually remind me that research itself is born in dialogue and comes to life through teaching. My sincere thanks go to Carol Bennett for her careful proofreading of the manuscript, and to the publisher, Gunter Narr, and especially Kathrin Heyng and her team, for their support throughout the publication process. Working with them has been both professional and rewarding. <?page no="21"?> 1 Introduction 1.1 Background to and motivations for the study This study explores how speakers use language to discursively manage their social identities in informal communicative situations. More precisely, it inves‐ tigates the discursive and social functions served by variants of one particular linguistic structure - NSIs of the type “I + copula + not + indefinite NP” - in English language as used by writers in UK online discussion forums. Formally, the structure is defined as “I + copula + NOT + indefinite NP”; functionally, the indefinite NP serves as an identifying expression, a perspective developed further in Chapter 3.3. Representing informal written conversations and, thus, highly interactive discourse, these forums are a promising discourse site for studying NSIs. The data for this study is a randomized corpus of 936 variants of the structure in their co-text of usage, annotated for formal, semantic, and functional aspects. This corpus was examined both qualitatively and quantita‐ tively, the latter with the help of the concordance software WordSmith 5.0 (Scott 2008). NSIs are pragmatically interesting structures because they serve to defeat explicit or implicit identity claims present in the immediate co-text, the situa‐ tional context, or the wider cultural context of the utterance, and thus “display significant contextual and interpersonal meanings in addition to their ideational sense” ( Jordan 1998: 706). For instance, while the utterance I am not a doctor could be a direct response to the question Are you a doctor? , it also has the potential to interact with and modify its co-text and context, guiding the interlocutor’s interpretation of the speaker’s utterance. To take two examples from my corpus, the NSI in Sounds like asthma to me, but I’m not a doctor reduces the epistemic status of the speaker’s diagnosis; in I’m not a doctor and this is just my experience, the NSI serves as background against which the (preceding or following) text referred to by this is represented as ‘just experience’, and thus as epistemically inferior, and hedges the clause coordinated by and. Being or not being a doctor - and the motives for making this relevant in the context of a forum discussion - might also have implications beyond the local discourse context: certain discourse functions of the structure might transtextually correlate with certain contexts of usage, pointing to patterns of meaning-making or conceptualizations underlying how speakers routinely manage the interpretation of their utterances on web forums. These forums allow individuals with ever-more specific interests to informally interact across <?page no="22"?> spatial distance, irrespective of factors that might separate them in real life. In these discourse contexts, marked by similarities rather than differences between participants, the question of what functions negative self-identification serves becomes particularly interesting. In other words, in certain communicative situations, ‘not belonging’ might be more important than ‘belonging’ to a particular social group, and this might relate to higher-order conceptualizations of the social situation enacted and the social world represented in these communicative situations. Difference and differentiation, rather than sameness and identification, appear to be key paradigms of contemporary social life, characterized by a reconceptualization of identity toward individualization (Giddens 1991; Beck & Beck-Gernsheim 2001; Bauman 2012; Rosa 2013). Next to globalization, marked by an ‘outsourcing’ of, for example, governance functions of the state to supranational bodies, individualization is the second constitutive feature of late modernity. Individualization can be defined as a process by which the individual becomes the center of social agency, that is, an ‘insourcing’, foregrounding individual choice and responsibility, as well as backgrounding the role of membership in more traditional social categories and the constraints they impose (Beck & Beck-Gernsheim 2001). In this context, the categories people choose to identify with - and conversely, the pool of categories they choose not to identify with - might be proliferating, as social life, just like the free market, is increasingly diversified, that is, fragmented into ever-smaller fields of interest, expertise, and consumption choices. As social life and relations are also increasingly fast-paced, identity might also become an increasingly unstable, fluid, and continually transforming category, leading people to identify with a plurality of concepts and to highlight different aspects of their self-identities from one situation to the next, self-con‐ sciously weighing the risks and benefits involved in identity representation and negotiation. According to Rosa (2013: 4), “social changes can be either analyzed ‘mac‐ rosociologically’ as alterations in ‘objective’ social or systemic structures or investigated ‘microsociologically’ from the viewpoint of a subject-centered social science as a transformation of logics of action and self-relations”. By microlinguistically examining the pragmatic functions of NSIs, this study seeks to contribute insights into how persons interacting on web forums linguistically manage their utterances and, thus, their social relations, in informal written conversations. It also seeks to find if particular functions served by particular variants of the structure can be observed across texts, and if so, what underlying conceptualizations structure such functional patterns. Thus, my study aims 22 1 Introduction <?page no="23"?> to bridge the gap between studying local instances of language in use and addressing questions on the level of the macrosociological context from a critical discourse analytical perspective. 1.2 Research questions The research questions guiding this study were developed to explore both the linguistic patterns and the broader conceptual and sociopolitical implications of NSIs in online forums. While each question is addressed in detail in the chapters indicated below, they are closely interconnected, all grounded in the assumption that language use is shaped by - and in turn helps shape - the social world in which it occurs. Accordingly, the question of broader sociopolitical implications runs through all three research questions and is addressed, where relevant, across Chapters 7 to 10. RQ1 (Chapter 7) A. With which nouns and noun phrases do forum users negatively identify? B. To which conceptual categories can these nouns and NPs be assigned, and how prominently (in frequency and lexical variation) are these categories represented? RQ2 (Chapters 8 & 9) A. What are the formal-functional relations of NSIs in my corpus and their clause-internal and clause-external co-texts? B. What are the meanings and functions of co-texts with certain formal links to the NSI structure, and how frequently are they represented in the corpus? C. Do co-texts with specific meanings and functions occur together with particular conceptual categories of NSIs in patterned ways? RQ3 (Chapter 10) A. What patterned relations can be observed between conceptual categories of NSIs and types of discourse contexts? B. What do these reveal about underlying conceptualizations of self-repre‐ sentation in web forums? 1.2 Research questions 23 <?page no="24"?> 1.3 Structure of the book To answer these questions, the book follows an hourglass-shaped structure: it begins with a broad theorization of identity and the self, then narrows its focus to a detailed linguistic analysis of NSIs, proceeding through increasingly fine-grained levels of co-text and context. In the final chapters, the perspective widens again to critically interpret the findings in light of their sociocultural and sociopolitical embeddedness. This structure allows the study to move between micro-level linguistic analysis and macro-level questions about discourse, iden‐ tity, and contemporary social life. Chapter 2 contextualizes the study and the social research interests it pursues. It reviews literature on different conceptualizations of identity and the self, focusing on how various linguistic approaches understand the relationship between language, mind, the self, and the social world. It also situates this book in a broader contemporary social context, raising questions about late-modern identity conceptualizations that can be addressed in a critical study of NSIs in use. The chapter further discusses web forums as the local discourse context in which negative self-identification is examined, explaining what makes them particularly well-suited sites for exploring how people discursively manage their identities in informal written conversations. Chapter 3 theorizes negation in general - and NSIs specifically - from the perspectives of formal semantics, grammar, and pragmatics. It lays the theoretical foundation for studying the effects realized by the meaning potential of the structure “I + copula + NOT + indefinite NP” as used in authentic discourse contexts. The chapter also explains the criteria for selecting structural variants as data for analysis. Chapter 4 theorizes negative self-identification from a functional perspec‐ tive, explaining how structures like NSIs can be approached as multi-indexical meaning potentials whose functions depend not only on the value of the indefinite NP, but also on the textual, situational, and broader social contexts in which they are used. To explicate possible relations between individual utterances and (what is considered to constitute) context, the chapter reviews various linguistic theories that approach language-in-use and different levels of context. These range from theories that consider minimal textual and/ or social context, such as traditional speech act theory and conversation analysis (CA), to those that treat context as a central departure point, such as systemic functional linguistics (henceforth SFL) and corpus-based approaches that study paradigmatically compiled discourse samples to quantitatively examine micro‐ linguistic choices across texts. I then explain how a corpus-based pragmatic study of negative self-identification can be embedded in the broader framework 24 1 Introduction <?page no="25"?> of corpus-based CDS, which seek to uncover conceptual configurations that structure meaning-making in discourse as part of social practice. The chapter concludes by presenting the discourse model for studying NSIs developed in this study. While Chapter 4 considers negative self-identification in relation to different conceptualizations of textual and non-textual context - seeking to theoretically bridge the gap between the linguistic structure in focus and the contexts in which people use language to contrast themselves with particular concepts - Chapter 5 discusses how the corpus of 936 instances of negative self-identifi‐ cation used in web forum discussions was collected, and what metatheoretical and methodological principles guided the process of data collection and analysis. Chapter 6 elaborates on the analytical approach, introducing the for‐ mal-functional framework that provided the conceptual basis for annotating the data and analyzing the meanings and the meaning relations of and between NSIs and their co-texts. It also presents the annotation scheme applied prior to - and refined during - the analysis. Chapters 7, 8, and 9 report and discuss the results of the analysis. Chapter 7 presents a conceptual profile of the indefinite NPs used in all instances of the NSI structure in the corpus, addressing RQ1. It concludes with a discussion of the implications of these findings and formulates expectations about the conceptual strategies that appear to guide how NSIs are used on web forums. Chapter 8 focuses on how NSIs relate to their immediate co-texts. It presents both qualitative and quantitative analyses of the formal links between the structure and its surrounding sentences or clauses, and explores the experiential functions these co-texts perform in discourse. Chapter 9 builds on this by examining how particular co-texts contribute to the meaning and function of specific NSIs. It also zooms in on two illustrative examples, analyzing their functions in their textual and situational context and relating them to broader questions about social meaning, epistemic stance, and discursive positioning. The final chapter, Chapter 10 addresses RQ3, summarizing the study’s findings and critically discussing their sociocultural implications. 1.3 Structure of the book 25 <?page no="27"?> 2 Identity, (Negative) Self-identification, and the Social World While this study is primarily a linguistic project - analyzing authentic language use in particular contexts to explore the meanings conveyed by a structure with meaning potential - the impetus for analyzing negative self-identification in the first place is the assumption that notions of belonging and not belonging are undergoing transformation in the contemporary social world. In the words of Rosa (2013: 5), “[s]ocial-structural processes of modernization cannot occur without some correspondence in the construction of subjective senses of self, in other words […], social-structural transformation through modernization must necessarily go hand in hand with a transformation of identity”. To lay the groundwork for positioning this within the wider sociocultural context about which it ultimately seeks to provide insights, Section 2.1 first presents different conceptualizations of identity and the self as found across disciplines, with a focus on issues of identity most relevant in the current sociopolitical climate. It also explains how different linguistic approaches view the relationship between language, mind, the social world, and identity. Section 2.2 then discusses specific identity-related concerns that are central to the research questions of this study. Section 2.3 moves from the broader social context to the specific context of linguistic interaction examined here - namely, web forums. It sketches the field of research on online communication and explains what makes web forums a particularly fruitful site for studying identity and negative self-identification. 2.1 Conceptualizing identity It is possible to roughly distinguish four conceptualizations of the self. First, there are essentialist cognitive conceptualizations based on the notion of rational social actors reflexively creating their identities. These assume a prediscursive, ‘true’ (sense of) self. Second, there are psychoanalytic theories of the self, for which Freud (1923/ 1961) laid the cornerstone by being the first to challenge the essentialist notion of a single ‘entity’ constituting the self. This perspective conceptualizes the self as primarily a psychological, but also a social phenomenon. Third, social/ collective identity models mainly define identity through social group membership (Tajfel & Turner 1986). Finally, there are discursive approaches, which are interested in (collective rather than <?page no="28"?> individual) identities as discursively constructed and negotiated socio-cognitive representations (Koller 2012). In the following section, I briefly outline these conceptualizations, focusing on those most relevant to the present study. 2.1.1 Identity as an individual project Contemporary popular representations of identity are largely dominated by the notion of authenticity, which can be understood in two (related) senses. First, authenticity can refer to a ‘unique’ self in the Romantic sense, which one should seek to express (Boas 1964: 1). Second, and particularly salient in late modernity, it can refer to a perceived coherence between online and offline persona. Historically, this idea of a self-created, authentic identity goes back to Enlightenment thinkers such as Locke and Descartes, whose dictum cogito ergo sum (“I think, therefore I am”) sums up the view of the self as cognitively constituted. However, in the contemporary context of late modernity, the individual - less and less constrained by traditional social norms and regulations based on collectivity - is becoming the central unit of social life. This phenomenon has been discussed as individualization. Individualization involves the reflexive ‘design of the self ’ with an emphasis on individual agency and represents one of the key global transformations of social life, alongside risk, pluralizing choices, and perpetual change (Giddens 1991; Beck 1992; Beck & Beck-Gernsheim 2001). Today, two features of the contemporary social world are particularly influ‐ ential in how people constitute a sense of authentic self and manage their identities in social interaction: the internet and the notion of choice between (mostly consumption-related) lifestyles. Owing to the internet, meanings are no longer anchored in place but are de-localized and negotiated paradigmatically on a global scale by “users who do not share a previous offline connection” (Androutsopoulos 2014: 63) - a phenomenon Giddens (1991) calls “disembed‐ ding mechanisms”. According to Dawson (2013: 87), choice represents “the first theme of late modernity”: online communication offers increasingly complex repertoires to choose from when reflexively managing identity in superdiverse settings (Vertovec 2007, 2015; see also Creese & Blackledge 2018). In these online settings, “[the] seamlessly carried out identification process in face-to-face interactions is challenged” (Bahri et al. 2018: 1). As a result, authenticity in this second sense has become increasingly important, serving as a marker of trustworthiness (Seargeant & Tagg 2014: 7). At the same time, late-modern self-representations are often marked by a foregrounding of paradigmatic affiliations, whereby the paradigms of belonging 28 2 Identity, (Negative) Self-identification, and the Social World <?page no="29"?> frequently reflect highly semiotized consumption choices. Some have even argued that products have become “consumers’ extended selves” (Mittal 2006: 550). In marketing, for instance, “the idea that consumers use brands to express their identities has led many companies to reposition their products from focusing on functional attributes to focusing on how they fit into a consumer’s lifestyle” (Chernev et al. 2011: 66). There has long been a tendency for companies to rebrand themselves as ‘authentic’ employers. Numerous studies (e.g., Kaneko 2005; Mautner 2005; Mautner 2010; Teixeira & Dill 2011; Tomlinson 2017) have examined the now-established climate of marketization in public institutions - including higher education - where differentiation through corporate identity strategies has become increasingly significant. This means that belonging to a global, life‐ style-based community can be intertwined with social difference and inequality. As Giddens succinctly puts it: “Modernity fragments; it also unites” (1991: 189). 2.1.2 Identity as a product of the social The foregrounding of differentiation and ‘authentication’ through lifestyle choices - rather than affiliation via more traditional social categories such as class, gender, or race - has been critically discussed by theorists who adopt a more social or collective perspective on identity. For example, Dubrofsky and Wood (2014: 282) criticize “the privileging of self-reflexivity and seemingly authentic displays in a context marked by post-racism (i.e., racism no longer exists; we can ignore race altogether) and post-feminism (i.e., gender equality has been reached; there is no need for feminist activism)”. Benwell and Stokoe (2006: 19-20), in a similar vein, argue that “contemporary, late modern and populist notions of the ‘true’, ‘authentic’ self, enshrined in a thousand self-help books and magazines”, corroborate “the ideal of self-fulfillment at the expense of political engagement often deemed to be a feature, or even crisis, of late modern society”. Movements such as Black Lives Matter - which has developed “from a social media post to a global phenomenon” (Maqbool 2020: n.p.) - demonstrate that these ‘traditional markers’ of identity have by no means become obsolete. The idea that identity is primarily a social construct is, among others, represented by social identity theory (Tajfel & Turner 1986), which assumes that affiliation with an in-group and differentiation from out-groups is central to one’s sense of self and, therefore, to one’s actions and behavior. In linguistics, the assumption of social group identities as variables to be correlated with other linguistic variables underlies variationist sociolinguistic approaches to language study. For example, in variational pragmatics, language use has primarily been 2.1 Conceptualizing identity 29 <?page no="30"?> studied in terms of its relation to social communities viewed as relatively homogeneous (Schneider & Barron 2008). This perspective on identity is not without its critics, either. Cameron (1990: 85), for instance, describes such communities as “descriptive generalizations”, arguing that correlating them with particular linguistic variables “in fact do[es] not explain anything”. Similarly, Sealey and Carter (2001: 3) contend that “social categories cannot be regarded as given, independent variables to be deployed in the way that correlative models imply”. In the context of today’s globalized, highly mobile society - marked by large-scale migration - we are confronted with “an exceptional demographic situation characterized by the multiplication of social categories within specific localities” (Wessendorf 2014: 2). This situation has been studied through the lens of superdiversity (e.g., Blommaert & Rampton 2011; Arnaut 2012; Androutsopoulos & Juffermanns 2014). The internet is a key driver of superdiversity, “both by enabling migrants to maintain links with their home countries and by bringing people together into groups shaped by common interest or purpose, regardless of geographical distance or of ethnic, cultural and linguistic difference” (Tagg & Lyons 2018: 312). Superdiversity challenges “groupism” (Brubaker 2004) insofar as it seeks to explore and critically question the centrality of “certain bases of social differ‐ entiation which cross-cut national origin or ethnic group affiliation (especially class and race) […] in super-diverse settings” (Foner et al. 2019: 3). This means that it is becoming increasingly difficult to even unproblemat‐ ically establish social groups as variables to be correlated with particular varieties of language use. To address this perceived explanatory gap between abstract social categories and concrete linguistic practice, approaches such as interactional sociolinguistics focus on the study of language as part of situated social interaction. These approaches assume that shared common ground is not a prediscursive given but is constructed in speech events that are embedded in, and conditioned by, the broader social context (Auer et al. 2014). 2.1.3 Identity as a discursive construction Just like the (critical, social constructionist) concept of discourse, the view of identity as discursively constituted is mainly associated with the work of Foucault (1972) and reflects the broader linguistic turn across the humanities and social sciences at the end of the 20 th century. As Fox (2023: 361) explains, this turn entails a shift toward understanding reality not as something fixed and given, but as something shaped and produced through language. An important prerequisite for the adoption of a discourse-based conceptualization 30 2 Identity, (Negative) Self-identification, and the Social World <?page no="31"?> 2 Key works regarding the view of the self as shaped by both the inner workings of the human psyche and societal factors include Freud’s psychoanalysis and Lacan’s social psychology. Freud (1923/ 1961) sees the ego as constituted through interaction between three layers of the self: the id, the ego, and the superego. The ego is “that part of the id which has been modified by the direct influence of the external world” and serves to identify realistic ways of satisfying the needs of the id, which stands for the primitive, innate part of the psyche. According to Lacan (1966/ 2001), the individual becomes socialized by entering and recognizing itself through the so-called symbolic order - that is, the social world that constitutes its relations, knowledge, and norms in and through discourse. Foucault’s account of the subject as a product of discourse - and thus ideology - has been read as influenced by Althusser’s (1971) theory of interpellation (see Villadsen 2024). These perspectives posit that identification inherently involves subjectification, both in the sense of participating as a subject in social interaction and being subjected to ideologies, since social interactions and the discourse they involve are shaped by dominant social forces. of identity was its theorization not as innately ‘given’, and modifiable by rational subjects, but as a socio-cognitive concept. In other words, an individual subject with conscious and unconscious mental processes is socialized and, thus, becomes part of a society and a participant in social and discursive practices. 2 Foucault’s concept of the self is inextricably linked to his theory of discourse and based on the assumption that what one can be, at a particular point in time, corresponds to what can be said and, thus, known, which means that subjects are “unfinished product[s] of discourse” (Benwell & Stokoe 2006: 30). This view is, in principle, shared by critical theories of discourse. According to Fairclough (2003), styles - defined as ways of being - can be considered inculcations of discourses as particular representations of the world, serving as frameworks for identification, constitutive of and reflected in people’s identity representations and enactments. Similarly, socio-cognitive approaches theorize identity as representations of ‘ways of being’, which are socially constructed and shared in and through historically and culturally situated discourse (cf., e.g., van Dijk 2003; Koller 2005). In reference to Hall (1996/ 2011), Benwell and Stokoe (2006: 31) explain how radically constructionist views of identity that see subjects as mere products of discourse fail to recognize that a subject “must have some kind of psychic coherence and existence prior to discourse” to actively perform an identity. 2.1.4 From linguistic (non-) identification to the material social world In linguistics and approaches based on it (to be discussed in Chapters 3 and 4), the relation between language in use, conceptualizations, power, and the 2.1 Conceptualizing identity 31 <?page no="32"?> material world is far from being unanimously agreed upon. Different approaches to studying language in use have different positions on the significance of local (negative) self-identification with respect to questions of identity beyond the immediate communicative situation or even the wider social context. For example, both SFL (Halliday 1978; Halliday & Matthiessen 2014) and CDS (e.g., van Dijk 2009) adopt a social constructivist perspective on language use. This perspective views identity as socially and thus linguistically constituted and considers any linguistic choice as conditioned by and serving a function in the situational context, embedded into and shaped by the larger socio-historical context. Within this shared perspective, CDS in particular takes an interest in inequalities between social groups and assumes that language features in action and interaction, social relations, and the material world as ways of acting, representing, and being (Fairclough 2003: 25). Such approaches therefore a priori consider the sociopolitical context of discursively enacted and negotiated identities. In contrast, linguistic approaches rooted in ethnomethodology, such as CA and membership categorization analysis (Sacks 1972; Schegloff 1988), have traditionally rejected the consideration of questions of identity beyond what is actually articulated in the conversations studied. While contextual variables such as gender or class are acknowledged to potentially influence talk, it is a requirement in CA to prove that these features are actually made relevant by participants as they interact on their own - not the researcher’s - terms (Schegloff 1992). The question of whether or not it is possible (and fruitful) for a linguistic study of talk to take a strictly empirical and apolitical stance, instead of ‘imposing’ categories on the data that might only figure in the researcher’s understanding of the conversation, has been debated among CA and CDS practitioners. For example, Billig (1999) argues that it is questionable whether there can be an entirely neutral perspective on studying language in use that does not make any reference to speakers’ identities and orientation toward particular social orders beyond what is evident in the talk. Seeking to reconcile the empirical grounding of categories such as identity in CA with the view that concrete textual events are informed by broader conceptualizations of the social world, the genealogical approach combines CA’s attention to interactional detail with an orientation to macrostructures and cultural-historical contexts. It thus recognizes discourse as an interpretative framework and a point of orientation in particular interactions (Wetherell 1998: 403). In a similar vein, in narrative theory, interactionally enacted identities are considered as meaningful in relation to narratives, which are embedded in the broader context of cultural narratives (Page & Thomas 2011). Beyond knowledge 32 2 Identity, (Negative) Self-identification, and the Social World <?page no="33"?> and views informing how speakers manage their linguistic self-representation in particular situations, critical realist practitioners of CDS such as Sealey (2012) stress the relevance of non-textual aspects of speakers’ identity, such as concrete, material obstacles shaping their experience of the world. Also, to a certain extent, it is necessary to recognize the existence of non-textual features of identity for CDS to be able to study language used by particular social groups (e.g., examining how marginalized groups discursively construct their identities requires assuming commonalities between texts produced by people sharing a particular subject position, cf. Koller 2013: 573). 2.1.5 Conceptualization of identity in this study My study sees identity work as primarily accomplished in discourse (Benwell & Stokoe 2006) and shares the assumption held by interactional sociolinguistics (Gumperz 1982; Coupland 2014) that meaning-making in interaction can be studied to explore “speakers’ perceptions of the relationships between linguistic forms and social realities” (Tagg et al. 2017: 30). Against this background, it considers the micro-functions of NSIs to learn how speakers use them to position themselves in relation to their utterances, and thus socially frame their interpretation in potentially patterned ways (Goffman 1981). These are considered to point to assumptions structuring how speakers reflexively use language to enact and explicitly contrast themselves with particular identities in forum discussions, which are embedded into the wider context of contemporary social life (Blitvich & Bou-Franch 2019: 4). Being interested in the categories speakers make relevant when linguistically managing their self-representation in online interaction, the analyses conducted for this study do not consider speakers’ real-life identities in the sense of “bundles of demographic character‐ istics” (Eckert 2012: 88). Consequently, it does not matter whether an NSI was uttered by a 60-year-old male person from China on a diabetes forum or by a 10-year-old Austrian girl on a gaming forum. Like variationist studies of pragmatic markers (Aijmer 2013), my study takes a quantitative approach to studying what functions NSIs, as meaning potentials, are routinely used to fulfill. This means that the main focus of this study is not, at least not primarily, to present a “situated, ethnographic investigation of unfolding interactions” (Tagg et al. 2017), but rather to conduct a microlinguistic, corpus-based analysis of instances of the focal structure. To bridge the gap between such microlinguistic analysis of a sample of NSIs in their proximate co-texts and questions about identity in relation to wider macrosocial concerns, I first situate the discussion in Section 2.3, where web forums are introduced as contexts in which linguistic 2.1 Conceptualizing identity 33 <?page no="34"?> strategies of (non-)identification become a particularly interesting phenomenon to investigate. Section 9.3 then develops this by discussing two instances of NSIs in more detail, considering participants’ online identities as represented on two particular forums. 2.2 Conceptual struggles around negative self-identification Given the context outlined above, the question remains as to what makes it interesting, in social research terms, to study negative self-identification rather than affirmative self-identification. Moreover, what conceptualizations charac‐ terizing the representation of identity in contemporary discourse might be considered worthwhile studying - or even critically questioning? The grounds for studying negatives linguistically are mainly pragmatic: negatives make presuppositions about the textual and non-textual context of usage. However, the prominence of the notion of authenticity and, thus, differentiation in late modern discourses of identity also suggests that not belonging - rather than belonging - might figure among the conceptualizations underlying people’s (online) self-representations. I have already explained that one large-scale social trend that appears relevant is the increasing centrality of the individual as the primary unit of social life, increasingly less constrained by traditional social norms and collective regula‐ tions - a phenomenon referred to as individualization (Beck & Beck-Gernsheim 2001). Individualization can, according to Vandenberghe (2015: 117), be seen as inextricably linked to globalization, in that the two form a “dual process” in late modern society. Individualization represents one of four main trajectories of social change in contemporary life, which is also marked by consumerism (as a consequence of capitalism as the dominant sociopolitical paradigm since modernity), scientization (a development based on rationalization), and subpo‐ litization (emerging from modernity’s democratization) (Kelly & Charlton 1995; Wagner 2012; Marko 2015a). Two trends along the trajectories of scientization and subpolitization appear especially noteworthy regarding the issue of identity and identification: the role and status of (different kinds of) experts against the background of increasing fragmentation and democratization of authority and knowledge. These will be discussed in more detail in Section 2.2.1, and related tendencies toward anti-elitism and populism in Section 2.2.2. 34 2 Identity, (Negative) Self-identification, and the Social World <?page no="35"?> 2.2.1 Contemporary notions and representations of expertise This section defines the concept of expertise and discusses its role in late modern society. Because expertise emerges in my data as one of the most salient conceptual domains of NSIs, it is discussed in greater detail than other aspects of identity, with particular attention to how expertise and lay expertise have been theorized in late modern contexts. Firstly, the role of expert systems in the context of a society marked by heightened sensitivity toward perceived risks is outlined. Then, I explain how experts can be distinguished from non-experts - a differentiation necessary to discuss issues such as power differentials between the two groups, as well as phenomena like the increasingly important role of lay expertise in online contexts. My review also refers to constructionist views of expertise, which are a prerequisite for explicating the role of language in establishing expert identities and, thus, trustworthiness in interaction. 2.2.1.1 Expertise and expert systems in risk society Risks, which are “socially manufactured and predicated on the awareness of a potential threat” (Vandenberghe 2015: 144), have come to feature prominently not only on a global scale (e.g., in discourses of terrorism or global warming) but also in individuals’ everyday lifeworlds. They are key factors even in very private decision-making situations - from major life choices such as whether or not to marry, to mundane decisions such as which brand of cereal to buy (Lash & Wynne 1992: 3). In this context of decisions being seen as determining factors in individuals’ ‘self-designed’ biographies, the role of expert systems (Giddens 1993: 29) has gained momentum. These are systems of technological, professional, and scien‐ tific knowledge that are simultaneously insourced and outsourced, according to Lash (2001: 11, in his foreword to Beck & Beck-Gernsheim 2001). Specific knowledge is increasingly centered within the individual, who may be an expert in one highly specific area and, due to flatter hierarchies in access to knowledge and greater freedom of choice, can access and use information more easily to make reflexive life decisions (insourcing). At the same time, knowledge is lifted out of its original context and circulates globally as abstract (or expert) systems (outsourcing). This means that while the individual may possess deep knowledge in one area of life, they are likely to lack expertise in many others - and thus need to consult external sources of knowledge. However, these external sources - that is, individual experts or institutions that produce and disseminate knowledge - are themselves subject to the dynamics of economic, cultural, and technological change, and therefore compete with one another. This makes 2.2 Conceptual struggles around negative self-identification 35 <?page no="36"?> it more difficult to assess the status of the various expert systems available. Possessing and using (the right kind of constantly evolving) knowledge has therefore become a continuous struggle between different groups believed to hold or share particular kinds or amounts of knowledge. On the one hand, Beck (1992: 58) argues that there is a contrast between, simplistically put, experts (“the technical people” who define what the risks are) and non-experts (“the public” who are informed about risks and thus placed in a passive role). Experts, in this sense, are conceived of as those capable of rational, evidence-based reasoning and, thus, are often equated with the natural sciences. Williams (2014: 1), too, refers to the notion of “professional experts armed with science, technology, and unquestionable authority”. The relationship between experts and non-experts has been discussed in terms of power asymmetries, face management, and mutual (mis-)understanding (Bromme & Jucks 2018), especially in health contexts (e.g., Roter & Hall 2006; Wirtz et al. 2006), but also in legal counselling (Godden & Walton 2006). These discussions have also been part of a larger trend toward increased client-centeredness (Bigi 2011: 67). On the other hand, risk society has amplified “a struggle among rationality claims, some competing and some overlapping” (Beck 1992: 59), that is, there is growing competition among, and increasing skepticism toward ‘accredited’ experts. As Walton (1997) argues, an expert’s authority rests on mutual recog‐ nition of their expertise - a status that can be critically questioned not only by peers, but also by non-experts (Dear 2004: 206; Stehr & Grundmann 2017). In this context, Beck (1992: 61) highlights the importance of the social perception of rationality - that is, what is perceived as rational is no longer exclusively for the sciences to determine, especially since science itself, in their view, has contributed to the creation of many present-day risks. As laypeople go about their daily lives, they “themselves become small, private alternative experts in risks of modernization” (ibid.). It should be noted that associating expertise with scientific knowledge and non-expertise with the absence of such knowledge is an oversimplification. First, it is possible to conceptually distinguish between knowledge and expertise. Williams (2014: 1) argues that expertise “refers to something more instrumental and pragmatic”, whereas knowledge may be seen as “more ‘meaningful’ and hermeneutic”. Elaborating on this distinction, Grundmann (2017), drawing on Bauman (1987), defines experts as interpreters rather than legislators, whose primary skill is advice-giving - that is, the ability to “mediate between the production of knowledge and its application; […] define and interpret situations; and […] set priorities for action”. From this perspective, the notion of expertise implies a “pragmatist take on knowledge” and suggests that knowledge without 36 2 Identity, (Negative) Self-identification, and the Social World <?page no="37"?> experience - or not gained through experience (i.e., tacit knowledge) - does not qualify as expertise. Second, expert identities - like any other identities - are best seen as con‐ struals of discourse rather than as a binary presence or absence of knowledge. In other words, participants must linguistically enact expert identities in order to be perceived as credible sources of advice or help, particularly when they lack formal qualifications (Armstrong et al. 2011; Harvey & Koteyko 2013; Rudolf von Rohr et al. 2019). As Rudolf von Rohr et al. (2019: 242) point out, discursive credibility management is just as important for experts as it is for laypeople, since “being perceived as a credible expert is connected to establishing trust” (ibid.: 221; cf. also Mackiewicz 2010a, 2010b). 2.2.1.2 Lay (online) expertise Representing “a core pillar of the modern information society” ( Johnson 2021: n.p.), the Internet has become a vital source for retrieving, sharing and negoti‐ ating information for experts and laypeople alike. KhosraviNik and Unger (2016: 207) use the term “participatory internet” to refer to a culture on the Web 2.0 that is marked by changing relations of communicative power (KhosraviNik 2018). Indeed, as Herring (2004: 26) puts it, “it has become a truism that computer-mediated communication (CMC) systems, as compared with previous communication technologies, are cheap, fast, and democratic”. In this context of technological advance and democratization of knowledge, there has been a growing tendency for skepticism toward experts (in the abovementioned sense of those with formally accredited, technical knowledge), and the relationship between experts and laypeople has been observed to undergo change by, for example, Kerr et al. (2007). In fact, the very notion of what constitutes expertise has been called into question (e.g., by Williams 2014), with the role of lay experts, sharing their knowledge and experience on various platforms, having become more important, especially since the early 2010s. Lay experts have been defined as ordinary people with knowledge in a specific field of expertise that is based on their personal and thus subjective experience rather than scientific evidence (Rueger et al. 2021: 7). The concept of lay expertise challenges the conceptualization of laypeople as “empty vessels waiting to have their vacant heads filled” by experts (Williams 2014: 1). This can be related to developments in present-day web communication, in which “previous dichotomies such as author/ audience and amateur/ professional are becoming porous” (Seargeant & Tagg 2014: 3). In fact, the opposition between ‘ordinary people’ and ‘technical people’ and, respectively, between knowledge 2.2 Conceptual struggles around negative self-identification 37 <?page no="38"?> gained from experience and knowledge gained from formal education, appears to be based on a rather static, one-dimensional concept of expertise that is increasingly being challenged. As shown by the example of expertise created by people commonly referred to as ‘nerds’ and ‘geeks’ (Tocci 2009; Coleman 2017; Postill 2018), the knowledge and skills of certain groups of people who may not be accredited professionals working for authorized institutions can nevertheless be extremely technical. In fact, this expertise, considered by scholars such as Coleman (2017: 94) to be based on collectivist beliefs and exchanged in “free spaces” (such as forums) on the web, can ‘overpower’ the systems based on expertise held by official experts and become an instrument for undermining corporate monopolies (e.g., by providing free alternatives to programs, games and operating systems) or civil disobedience (e.g., in the case of hacker activism). Rudolf von Rohr et al. (2019: 220), too, reject the assumption that particular groups of people can be linked to particular types of expertise, pointing out that different interactional situations may require access to and bring forth different types of expertise created not only by individuals, but also by groups. As forum users try to jointly solve problems by relating their experiences with the issue of concern, for example, they act as ‘swarm intelligence’, with their individual lay accounts co-constructing expertise. This expertise, by virtue of being ‘tailored’ to a particular user’s problem, might indeed be even more useful than formal expertise available, for example, on an official website or in expert literature on the subject. Research on lay beliefs in the field of health has shown that the beliefs of ordinary people represent complex systems of knowledge that not only reflect societal conceptualizations about health and illness but also influence how people affected by certain health conditions cope with them (Lawton 2003; Taylor & Bury 2007; Barker & Galardi 2011). Online communication about health and illness on platforms such as forums can be considered to empower laypeople because it enables them to connect across geographic boundaries to share their experiences and generate novel forms of expertise. These potentially move beyond strictly scientific, medical explanations. It has been found that the health advice given by lay experts is highly appreciated among peers, and seen as useful (e.g., Mattson & Hall 2011), emotional support (e.g., Bar-Lev 2008, McBride 2011), complementing information by and facilitating communication with doctors (e.g., Rupert et al. 2014), but also representing a potential challenge to formal medical expertise (e.g., Crooks 2006; Radin 2006; Rueger et al. 2020). It has been observed that lay experts often tend to orient toward the same explanations for health conditions as medical professionals, emphasizing, for example, the role of 38 2 Identity, (Negative) Self-identification, and the Social World <?page no="39"?> 3 I use the term extextualized here (following Seargeant & Tagg 2014: 40) to emphasize the sense of discourse being lifted out of its original context, which differs slightly from the more common term entextualized (Leppänen et al. 2014). lifestyle choices for health and ill-health and, thus, the importance of individual decision-making (Sosnowy 2014; Williams 2014). Nevertheless, the subjective nature of lay beliefs and the potential inaccuracy of information provided by peers online can pose a challenge in consultations with medical professionals insofar as this information might not be consistent with, or may even contradict, doctors’ views (Rueger et al. 2020: 2). This means that the need to epistemically frame information is likely to be higher when discussing delicate topics such as health, as there is a risk of inaccurate information causing damage to participants’ real-life physical existence. Thus, by contrasting themselves with medical expertise to hedge the information they share, speakers interacting on medical forums seek to prevent their advice having counter-productive effects. As my discussion of an expertise disclaimer in Section 9.3.2 shows, though, the construction and negotiation of lay expertise is highly relevant not only in online discussions on health issues, but also appears to play an important role in forums devoted to discussing information technology (IT) issues on an advanced level. Irrespective of the topic discussed on a particular forum, another reason why speakers may generally be more careful when presenting their (lay) knowledge and views online could be fears of their postings becoming “extextualized” - that is, lifted from their original context and reinterpreted by different, unpredictable audiences (Seargeant & Tagg 2014: 40) - and damaging their reputation on the forum (and, potentially, beyond). 3 The notion of lay expertise has also been extensively discussed in relation to other genres of CMC, for example, online consumer reviews, also referred to as eWOM (electronic word-of-mouth) (Hennig-Thurau et al. 2004). As Vásquez (2014: 65) explains, consumer reviews are a ‘digitally native’ genre providing a novel form of lay expertise previously unavailable, namely user-generated content about product experiences. As a form of expertise, emanating from laypeople one does not know personally, eWOM has rendered expertise “distrib‐ uted, geographically dispersed, and interpersonally removed” (ibid.: 66). Just as lay medical expertise may “lessen the information asymmetry” between patients and doctors (Rueger et al. 2020: 5), online consumer reviews help consumers make informed, and thus less risky, purchasing decisions. Particularly in the final stage of the decision-making process, product experience reviews - such as those shared on Amazon - may exert a key influence on whether people decide to make a purchase (Ghose & Ipeirotis 2011; Yoo & Gretzel 2009; Phillips et al. 2016). In the field of travel and tourism, for instance, Vermeulen and 2.2 Conceptual struggles around negative self-identification 39 <?page no="40"?> Seegers (2009) found that whether a review was written by an expert or by a layperson made little difference. What these studies highlight is that expertise is discursively constructed rather than a pre-discursive ‘given’ - an issue taken up in the following section. In Section 2.2.2, I turn to a closely related issue: distrust in certain forms of expertise and its connection to broader sociopolitical trends such as populism. 2.2.1.3 Linguistically constructing expertise Expertise is an aspect of discursively performed identities and plays a central role in forums, whose key purpose is knowledge exchange and advice-giving. As Carr (2010: 20) argues, expertise can be enacted by “establishing a deliberate stance in relation to a set of culturally valued or valuable objects”, which highlights the importance of language in constructing expertise. Negative self-identification can be seen as a particularly strong case of such deliberate stance-taking: for example, self-positioning as not a fan of […] or not a believer in […] explicitly contrasts the speaker with the objects or ideas under discussion, thereby projecting credibility and alignment through disalignment. The way people communicate in these spaces therefore has important effects on how they are perceived. If forum users are implicitly aware of what works in designing contexts for their postings, it can be assumed that they use language to present themselves as reliable and, thus, valued members. Indeed, studies have shown that the identities linguistically constructed by forum users affect how their contributions are evaluated by others (Vásquez 2014: 69), which, as Bennett (1999: 4, cited in Mackiewicz 2010a) observes, reflects the fact that online messages are often judged via the question “Who is telling me this? ” Credibility, then, is inseparable from how the speaker’s identity is presented, and local linguistic choices that serve credibility management are not arbitrary but reflect wider notions of what counts as valid, appropriate, and effective communication in a given community (Tagg, Seargeant & Brown 2017; Bucholtz & Hall 2005; Leppänen et al. 2015). Several studies have sought to identify the linguistic strategies by which people position themselves as experts in online interaction. Mackiewicz (2010a) identifies a set of assertions reviewers use to convey their expertise. These include positioning oneself as someone with product-specific experience, in‐ voking familiarity with related products, and making assertions of a relevant role - strategies also reflected in my data, for example in postings that combine disclaimers with experiential claims (e.g., I’m no expert, but in my experience…). Vásquez (2014) likewise shows how reviewers on Amazon use 40 2 Identity, (Negative) Self-identification, and the Social World <?page no="41"?> narrative positioning and hedging to make their accounts authoritative yet relatable. Metadiscursive devices such as definitions, too, have been found, as Marko (2017: 33) puts it, to “play a crucial role in interactively negotiating claims to knowledge and expertise and thus indirectly relations of power in […] lay discussion forum[s] - which does not predetermine these aspects”. Such definitional practices can also take playful forms, for instance when users coin pseudo-technical acronyms (such as OM to refer to married and thus “off-the-market” users in my data) to lend ordinary experiences an air of special‐ ized knowledge. By contrast, acronyms such as PWP (“person with Parkinson’s”) serve to establish authority and common ground, functioning as shorthand that indexes both shared knowledge and insider affiliation in discussions of illness experience. As I have argued elsewhere, the negotiation and enactment of expertise can also be studied to reveal the ideologies of communication that structure meaning-making in online interaction (Spitzmüller 2015; Fairclough 2010; van Dijk 2017; KhosraviNik 2022). Whereas prior research has focused on such cues for enacting expertise, my study explores what people tell about themselves by explicitly disavowing cer‐ tain identities through NSIs, thereby making a pragmatically marked linguistic choice. At the same time, NSIs interact in revealing ways with their co-texts. These co-texts can show information about speakers’ identities and beliefs (whether or not this is intended) (Vásquez 2014: 68), and reflect speakers’ awareness of and orientation to the interactional context of the forum. Unlike the studies noted above, then, my research approaches the question of how identities such as “the expert” are linguistically construed online from the opposite direction: rather than tracing linguistic cues strategically employed to enact a given identity, it takes as its point of departure a form that asserts a speaker’s non-identity and may serve a range of functions in coand context. One such function could be credibility management: negative self-identification with particular groups or roles may project identities that peers interpret as cues for trustworthiness, and NSIs may interact with co-texts that have been shown to index expertise. 2.2.2 Expertise, (dis-)trust and populism As the foregoing discussion has shown, in the contemporary sociopolitical context in which knowledge and risk constitute key paradigms, the issue of trust and the question of who has privileged access and the power to validate and disseminate particular kinds of knowledge and beliefs looms large. The impact of digital transformation on “different aspects of political legitimation, 2.2 Conceptual struggles around negative self-identification 41 <?page no="42"?> societal trust, scientific evidence and public discourse” and the issue of “control and ownership of the Internet” is by some considered to be “one of the most important battles of our time” (All European Academies 2019). Indeed, knowledge seems to be simultaneously becoming (perceived as) more democratic and what could be called an elitist privilege. More people now have access to and may profit from knowledge that has not necessarily been validated by traditional authorities ( Jensen et al. 2012: 2). ‘Ordinary’ people now not only have more information at their disposal, but may also become what Mehlenbacher (2019) has theorized as citizen scientists. Citizen scientists manifest increasingly blurred boundaries between experts and non-experts, but also highlight distrust in experts who are perceived as isolated from the ‘normal’ population. Discussing the reasons for this skepticism, Stehr (1992: 108) explains that groups of experts in financially strong and politically defining fields of knowledge, like cutting-edge science and technology, are “often narrowly located at the apex of the social hierarchy” and considered to be “in the employ of the already powerful and influential”, monopolizing these fields of expertise (cf. also Reed 1996). Again, it should be highlighted that dichotomies such as ‘bottom-up’ versus ‘top-down’ do not suffice to explain the contemporary sociopolitical context, in which political, corporate, and social institutions are intricately connected with - and mediated and reshaped through - social media institutions and their users (KhosraviNik & Unger 2016). These dichotomies are, however, necessary to approach struggles around expertise and knowledge in relation to the global upsurge of populism (or, as Mudde 2004 put it, the “populist Zeitgeist”), which has become a much-re‐ searched topic across disciplines (Rooduijn 2014; Wodak 2015; Moffitt 2016; KhosravNik 2017; Ekström et al. 2018). Populism, rather than constituting an ideology itself, can be viewed as an underlying conceptualization about relations between what Mudde and Kaltwasser (2012) define as three core concepts, viz. “the elite, the people and the general will”, with “elitism and pluralism” as their “direct opposites” (9). These concepts, and assumptions about their relations, are “ideologically flexible” in that the concrete meanings with which they are imbued depend on the “host ideologies” to which they attach. The (perceived) opposition is between a knowledge monopoly held by a top-down elite and a process of knowledge democratization driven by bottom-up movements (cf., e.g., van Dijk 1993) or, more recently, movements representing “motivated resistance to expert consensus” (Merkley 2020a: 24). There are both anti-intellectual and hyper-intellectual yet anti-authoritarian movements that should be considered in this context. On one end of the spectrum, we find a tendency toward anti-intellectualism, which Merkley 42 2 Identity, (Negative) Self-identification, and the Social World <?page no="43"?> (2020b: n.p.) describes as “a fundamental mistrust of intellectuals and experts”. In its extreme form, anti-intellectualism manifests as science denial and conspiracy theories, often “aimed at challenging the status quo and those in power” (Atkinson et al. 2017). For instance, Douglas et al. (2019: 19) note that “extreme climate skeptics assert that climate scientists are involved in data faking and fraud so that they ensure that they keep receiving research funding”. An even more - perhaps the most - extreme case of radical epistemic dissidence is flat-earthism, described by Pannofino (2024: 1) as “a dissident narrative that, if on the one hand rejects the knowledge of institutionalized and organized science and religion, on the other hand accredits an alternative and anti-intellectualist path of knowledge, accessible to the ‘common man’”. Thus, “anti-intellectualism is directly connected populism, a worldview that sees political conflict as primarily between ordinary citizens and a privileged social elite” (Merkley 2020a: 24). At the other end of this spectrum of epistemic distrust, we find anti-elitist movements driven by individuals with a high degree of interest and expertise in science and technology - for example, hackers. These movements have often been interpreted through the lens of antiauthoritarianism, “which manifests itself as a profound skepticism toward institutions and other forms of entrenched power” (Coleman 2017: 93). Against this background of digital media being the primordial site for ex‐ changing and (re-)negotiating expertise, the next section discusses web forums, which allow people from potentially diverse social backgrounds to discuss their views and knowledge on topics of shared interest. It embeds the present study in the research on online communication and discusses the possible relations between NSIs, the communicative context of web forums and questions of identity on a wider scale. While debates about expertise and trust operate on a macro-social level, they also permeate everyday digital communication. Online forums, where individuals negotiate belonging and credibility in peer-to-peer interaction, provide a productive site for observing how such broader dynamics intersect with microlinguistic practices such as negative self-identification. 2.3 The discourse context: Web forums as sites for studying negative self-identification Analyses of language in use range from close attention to micro-level linguistic choices to broader concerns with their role in macro-level societal issues, with the context of a discourse event (discussed in Chapter 4) serving as the interface between these levels. The context of interest in this study are online forums, where NSIs are used to position speakers in interaction and index the 2.3 The discourse context: Web forums as sites for studying negative self-identification 43 <?page no="44"?> wider social world in which these interactions are embedded (Bou-Franch & Blitvich 2019: 3). Accordingly, the project is conceptualized as a pragmatically oriented, socioculturally motivated and thus critical analysis of digital discourse, specifically web forum interaction, which warrants an overview of research on digital discourse and on identity in digitally mediated interactions. 2.3.1 Researching digitally mediated discourse In the 1990s, the beginnings of online communication marked “a radical departure from what came before” (Herring 2004a: 26) in terms of influencing social practices. Ever since, online communication and online communities have been extensively studied from the perspectives of language-based ap‐ proaches grounded in sociology, anthropology, and ethnography, as well as from interdisciplinary approaches that draw on concepts from both linguistics and other disciplines interested in exploring social processes (Sindoni 2019: 74). This scholarly interest coalesced into what Susan Herring (2001) termed computer-mediated discourse analysis (CMDA), a field that evolved out of earlier studies of computer-mediated communication (CMC). At its inception, CMC was defined as the study of “the social, communicative and linguistic impact of communication technologies, which have continually evolved in connection with the use of computer networks (esp. the Internet)” (Beißwenger & Storrer 2008: 292; see also Herring 2004b; Thurlow et al. 2004). Among the early landmark publications are Ferrara et al.’s (1991) study of “interactive written discourse” and Herring’s (1996) edited volume on CMC, which explored linguis‐ tic, social, ethical, and cross-cultural perspectives on online interaction. The launch of the Journal of Computer-Mediated Communication in 1995 signaled the consolidation of this new field, publishing early studies of interaction in platforms such as Icp OnLine (Hutchison, 1995). The more explicit focus on language on the Internet - and thus the beginnings of linguistic and discourse-analytical approaches to online communication - dates to the early 2000s: during this time, CMC research became more focused on linguistics, acknowledging, like Crystal, that “what is immediately obvious when engaging in any of the Internet’s functions is its linguistic character”, and that [consequently, the Internet] is likely to be a linguistic revolution” (viii). The first wave of CMDA, then, was largely text-focused, describing the language of online genres and their socially conditioned varieties (e.g., Herring 1998; Baym 1996; Cherny 1999). Increasingly, CMC research sought to explain the social impact of newly emerging and fast-evolving communication technologies. The label CMDA, mainly associated with Herring (2001: 626), marked this shift in 44 2 Identity, (Negative) Self-identification, and the Social World <?page no="45"?> CMC studies toward engaging with the “discursive negotiation and expression of social relations in cyberspace”. Subsequent waves addressed online discourse more explicitly as social practice (e.g., Androutsopoulos 2006; Danet & Herring 2007; Bloor 2016) and increasingly turned toward multimodality (e.g., Kress & van Leeuwen 2001; O’Halloran 2013; Georgakopoulou & Spiliotti 2016; Ng 2018). Research in CMDA, not tied to a single method or linguistic theory, has been grouped into five strands (Herring 2014). The first strand encompasses studies where the genre is medium-constituted, such as Tagg’s (2012) work on text messages, as well as analyses of discursive practices associated with particular, but potentially transcending, media, such as YouTube eating shows (Rüdiger 2022) or online restaurant reviews (Vásquez & Chik 2021). These examples illustrate that, as Belcher (2023) notes, it is difficult to disentangle genres from the affordances of new media - a reason why she recommends emphasizing “prominence in contemporary communication” (Hafner & Pun 2020, p. 4) as a criterion in research on digital genres. For my study, this strand is relevant in showing how particular online environments shape recurrent pragmatic structures, a perspective I apply to forums and NSIs. The second strand comprises research on structural features of digital discourse, such as emoji (Riordan & Kreuz 2010; Danesi 2016), hashtags (Zappavigna 2018), and kineticons (Lyons 2018). This strand is relevant to my study in highlighting the pragmatic role of structural features in digital discourse, a perspective that informs my approach to analyzing NSIs in forums. A third strand focuses on linguistic diversity, including studies of digital multilingualism (e.g., Danet & Herring 2007; Leppänen & Sultana 2023). This work underscores how online interaction complicates traditional sociolinguis‐ tic variables, a challenge that resonates with my treatment of NSIs beyond demographic categories. The fourth strand concerns pragmatic phenomena such as addressivity (De Oliveira 2013), performativity (Virtanen 2013), and interactional dynamics - for example, on YouTube (Benson, 2015) and more recently on TikTok (Dainas 2025). This research also examines specific resources mobilized in digital discourse, including punctuation (Androutsopoulos & Busch 2021) and intensification strategies (Cotgrove 2025), as well as discourse-specific practices such as polyvocal scorn in conspiracy communication (Dynel & Zappavigna 2023) or metalinguistic density in digital representations of protest (Deschrijver 2024). This is the strand most closely aligned with my own focus, since NSIs represent a pragmatic strategy for stance-taking and identity work in forum interaction. Finally, a fifth strand is constituted by CMDA ‘proper’, which, like my study, is concerned with the relationship between local practices in digital discourse and the social world more broadly (Herring 2001, 2004, 2019). 2.3 The discourse context: Web forums as sites for studying negative self-identification 45 <?page no="46"?> However, the boundaries between these categories are increasingly porous. Platform infrastructures and affordances are shaped by social interactions, and vice versa (Lüders et al. 2022), so that structural, pragmatic, and social dimensions of digital discourse can no longer be easily disentangled. This is especially relevant for a study like mine, which moves across pragmatic, structural, and social concerns. In the meantime, acknowledging that computers are no longer the sole technologies mediating digital interaction (Kopf 2025: 2), CMDA scholars have begun to adopt the label digital discourse analysis (DDA; Vásquez 2022). This move shifts the focus beyond “computer” as the primary reference point and instead foregrounds the wider ecology of devices and platforms through which communication is mediated, integrating multimodal and cross-platform practices under a more inclusive umbrella. As the overview above suggests, the fifth strand of CMDA/ DDA, concerned with the social and societal ramifications of digitally mediated discourse, overlaps with the research objectives of CDS (see Section 4.3). Although not traditionally focused on online discourse, CDS underwent a “digitally mediated turn” (Unger 2025: n.p.) in the 2000s, when the rise of the participatory web was understood to entail profound social change. Early critical scholarship already anticipated these concerns, with Graham (2000), for example, interrogating the role of information technologies in hyper-capitalism (see Unger 2025 for a comprehensive overview). Since then, a growing body of work in the CDS tradition - often under the labels social media critical discourse studies (SM-CDS; KhosraviNik 2023) or digitally mediated discourse analysis (DMDA; Unger 2020) - has explored the entanglement of discourse, platform affordances, and sociopolitical power. Platform affordances are here understood as “possibilities for action” that both enable and constrain behavioral outcomes (Evans et al. 2017: 36). Social media, in turn, are defined as “web-based and mobile services that allow individuals, communities, and organizations to collaborate, connect, interact, and build community by enabling them to create, co-create, modify, share, and engage with content (useror bot-generated)” (Nau et al. 2022: 15). Research in DMDA can be grouped into three broad orientations: (1) macro-level analyses of ideology and infrastructures, (2) meso-level analyses of genres and practices of digitally mediated communication, and (3) micro-level analyses of linguistic and semiotic strategies. At the macro level, scholarship has examined how digital infrastructures intersect with broader ideological and political formations. This includes studies on the ideological implications of user-led content production (e.g., Ritzer & Jurgenson 2010; Kopf 2020), the entanglement of social media with right-wing 46 2 Identity, (Negative) Self-identification, and the Social World <?page no="47"?> populist discourse (e.g., KhosraviNik 2017, 2018), and the circulation of misog‐ ynistic or extremist ideologies in online subcultures such as the manosphere and incel communities (e.g., Ging 2019). Critical work has also highlighted how social media enable new forms of activism, for instance in digital feminism (Dobson 2015). At the meso level, research has focused on the communicative forms and practices characteristic of digital environments. This includes work on platforms such as Twitter (e.g., Zappavigna 2012) and genres such as text messages (Tagg 2012), shared small stories (Georgakopoulou 2014), selfies and captions (Page 2019), and Instagram practices (Ghaffari 2020). Other studies have examined pragmatic and interactional dynamics, for instance offence-taking on social media (Tagg et al. 2017) or relational work in text messaging (Spilioti 2012). Broader practice-based and ethnographic perspectives have foregrounded users’ mediated actions and everyday meaning-making (e.g., Androutsopoulos 2008; Barton & Lee 2013; Jones, Chik & Hafner 2015; García-Gómez 2017). At the micro level, research has zoomed in on the linguistic and semiotic resources that constitute digitally mediated interaction. Examples include anal‐ yses of emoticons (e.g., Skovholt et al. 2014), punctuation (Androutsopoulos & Busch 2021), intensification strategies in YouTube comments (Cotgrove 2025), and humor and irony in online reviews (Vásquez & China 2019). Other work has highlighted multimodal resources for stance-taking and affiliation across platforms (e.g., Zappavigna 2018). Taken together, this body of work highlights both the diversity of approaches within CMDA/ DMDA and their shared concern with linking digitally mediated forms of discourse to broader social processes. Within this broader trajectory, my project aligns most closely with the fourth and fifth strands identified by Herring (2014) - research into pragmatic phenomena and CMDA proper, concerned with the relationship between online discourse and the social world (Herring 2001, 2004b, 2019). At the same time, it exemplifies the blurred boundaries between micro-level pragmatic research, meso-level analyses of communicative practices, and macro-level questions of identity and society. Rather than centering on platform-specific features or on large-scale ideo‐ logical formations alone, my analysis of NSIs highlights a pragmatic structure that cuts across media, while also foregrounding forums as paradigmatic sites of interaction. In doing so, the study contributes to pragmatic and discourse-ana‐ lytic understandings of online communication and opens a perspective on how processes of (non-)identification unfold in digitally mediated contexts. Having outlined these wider developments in the study of digital discourse, the next 2.3 The discourse context: Web forums as sites for studying negative self-identification 47 <?page no="48"?> section turns more explicitly to the issue of identity and (non-)identification in online forums, which is central to my analysis. 2.3.2 Studying identity and (non-)identification in web forums The influence of web communication on people’s social lives - and, by ex‐ tension, their identities - has been enormous (boyd 2011; Seargeant & Tagg 2014). Between 2019 and 2025, the number of users of social networking sites rose from 3.31 to 5.44 billion, corresponding to just under two thirds of the world’s population (Statista 2025). Social media - broadly defined as “online environments which enable social interaction” (Leppänen et al. 2014: 113) - have thus transformed the web into a space where everyday sociality unfolds, reshaping interaction through linguistic as well as multimodal practices (Seargeant & Tagg 2014: 2-3). Scholars have theorized this shift in different ways. Parsell (2008: 41), for instance, highlights the Internet as a powerful force in “the manufacture of identity”, enabling new forms of participation, fragmentation, and meaning-making beyond traditional genres of conversation or narrative (cf. Herring et al. 2013: 9). Merchant (2006: 235) emphasizes the role of digital media in connecting dispersed networks and generating new forms of social identity, while Hodkinson (2007: 625) stresses their contribution to “fragmented, fluid patterns of individual identity”. From a CMDA perspective, identity is understood as discursively constructed in interaction (Rudolf von Rohr et al. 2019: 219). Vásquez (2014: 67) describes it as an “interactional achievement”, inseparable from processes of belonging and differentiation (Benwell & Stokoe 2006: 35). Leppänen et al. (2014: 112) similarly underline the dynamic character of online identity as emerging through “active processes of identification and self-understanding, seeking or eschewing commonality, connectedness and groupness”. These insights have motivated a wide range of empirical studies. Research on virtual communities (Preece et al. 2003) has shown how forums allow individuals to experiment with self-presentation in ways often unavailable offline (Vásquez 2014: 67; Sims 2016), while also adhering to community-specific codes of practice. This dual orientation makes forums and related platforms compelling sites for studying discursive (non-)identification. Given the sheer volume of research, only a brief sketch of key strands is possible here. Gee (2004) develops a theory of identity through participation in online games, linking digital and offline selves. Page (2012, 2018) and Georgakopoulou (2022) highlight the role of small, digitally mediated narratives in shaping how users position themselves. Work on gender has been particularly 48 2 Identity, (Negative) Self-identification, and the Social World <?page no="49"?> prominent: broad accounts trace how social media reconfigure gendered iden‐ tity performance (Marwick 2014; Seargeant & Tagg 2014; Herring & Kapidzic 2015), while more focused studies examine specific genres or practices, such as Vásquez and China’s (2019) study of gendered product reviews or Maaranen and Tienari’s (2020) analysis of hypermasculinity on social media. Critical discourse studies have also attended to gender-based conflict and online harassment (Diabah 2023; Maruenda-Bataller, Mercé & Castellano-Ortolà 2024) as well as digital feminist activism (Mendes et al. 2019). Health-related identities constitute another major focus: Pavlova and Berker (2020) examine mental health discourse on Twitter, and Hunt (2015) investigates self-representation on diabetes-related Facebook pages. The linguistic strategies used to enact identity have been extensively re‐ searched. For example, Deumert (2014) and de Lange et al. (2015) examine playful self-representation, while Page (2019) explores self-denigrating humor in so-called “ugly selfies”. Another key pragmatic strategy studied in digital discourse is sarcasm (e.g., Bamman & Smith 2015; Sykora, Elayan, & Jackson 2020). In the context of ‘post-truth’ digitally mediated public discourse (Farkas & Schou 2023) and increasingly polarized debates around competing “versions of truth” (Hutchby & Wooffitt 1998; Wehrle 2023), questions of how users take a stance in epistemological debates (Deschrijver 2021) have gained prominence in CDS. As for the digital platforms of interest in discourse studies, scholarship has engaged with social networking sites (SNS) such as Facebook and LinkedIn (Seargeant & Tagg 2014; Tagg et al. 2017; Garzone 2018; Petroni 2019), as well as Twitter (now X) (e.g., Zappavigna 2012; Page 2014; Ross & Rivers 2017), Weibo (Zeng et al. 2017; Yu et al. 2023), and TikTok (Zulli & Zulli 2020; Zeng & Yan 2024; Tanner & Gillardin 2025). While forums were often described in the 2010s as being “eclipsed […] by social networking sites such as Facebook” (Pendry & Salvatore 2015: 211), they continue to serve as spaces for anonymous, peer-led interaction and contribute to individual and social well-being by fostering a sense of belonging among users (ibid.). Indeed, while traditional message-board-style online forums were in decline at the time of data collection for this study, the evolution of platforms such as Reddit - which had over a billion monthly users in 2025 (Saini 2025) demonstrates that forums remain relevant as sites of “rich social interaction and identity development” (Seering et al. 2019: 1417). Forums - both old-school and newer variants - are particularly valuable for studying NSIs. They bring together people from diverse backgrounds to exchange knowledge and (lay) expertise on topics of shared interest. Authen‐ 2.3 The discourse context: Web forums as sites for studying negative self-identification 49 <?page no="50"?> ticity - understood as personal uniqueness and coherence between online and offline persona - plays a key role in constructing expertise online. This can be linguistically indexed, among other things, by eschewing commonalities with particular social groups. At the same time, forums represent communities with their own interactional norms and sense of groupness, which means it may also be important for users to orient themselves toward their peers and highlight shared traits to create a sense of “shared individuality” (Lüders et al. 2022). Negative self-identifiers can function as linguistic indices of both individual authenticity and, thus, credibility, as well as group affiliation and belonging - two functions I will discuss in more detail below. As for the first notion, authenticity, forums are online contexts in which lay expertise on topics of shared interest can complement, challenge, or even substitute traditional expert knowledge. To be perceived as credible, it is particularly important for users “to provide information about their identities, especially with respect to establishing credibility, and to give readers reasons for trusting the information offered” (Vásquez 2014: 68). As I have argued, credibility can be constructed on various grounds and linguistically indexed through multiple means. Interaction in forums is primarily text-based, more anonymous, and simultaneously less private than face-to-face communication. It tends to be more informal and allows users to display aspects of their identities that they might not emphasize offline. At the same time, it carries a heightened “potential for misunderstanding” (Delahunty 2012: 407). Within this context, the use of NSIs may reflect a greater need for a person to explicitly distance themself from particular social groups - to say who they are not - in order to construct authenticity and, by extension, trust (Page 2014: 46; Seargeant & Tagg 2014: 7). Against the broader sociocultural backdrop of trust struggles in contemporary knowledge society, distancing from certain group affiliations may enhance perceptions of trustworthiness. Such moves reflect underlying assumptions about what constitutes a credible identity. This, in turn, allows the analysis of NSIs to be situated within a wider sociopolitical context marked by distrust and heightened perceptions of risk. As Page puts it, “[t]he ambiguity associated with online representation of the self sits within a wider complex of debates about the nature of authenticity, trust and reputation” (2014: 47). As for the second notion, groupness - a shared sense of commonality - is crucial for the achievement of identity, which, as Leppänen et al. (2014: 112) put it, is a “contractual achievement”. It shapes how speakers in “collapsed contexts” (Marwick & boyd 2011: 10) strategically manage discursive self-representation by stylizing their utterances with regard to their audience - both the immediate online community and, particularly in public forums, a potentially unlimited 50 2 Identity, (Negative) Self-identification, and the Social World <?page no="51"?> audience of anonymous readers (the ‘contractual party’, so to speak). Negative self-identifiers may be used to explicitly disalign with particular groups and implicitly align with others, allowing speakers to manage their face in front of their peers by pre-empting certain interpretations of their utterances. If speakers across online forum communities stylize their discourse to position themselves in relation to specific groups and perform particular identities to serve context-specific interactional functions - thereby achieving a sense of belonging - this may reflect their assumptions about what constitutes appropriate linguistic behavior in the online environments they inhabit. In this sense, the use of NSIs can be understood in light of the conceptual frameworks that structure meaning-making in discourse. This includes van Dijk’s (1997) notion of cognitive context models - expectations about social situations that guide discourse production and interpretation. It also resonates with Keane’s (2018) concept of semiotic ideologies - that is, assumptions about the potential pragmatic effects of employing particular semiotic resources in particular ways. 2.4 Summary In this chapter, I contextualized the study of NSIs by outlining key issues related to the notion of identity and showing how these expressions - as assertions of ‘non-identity’ - can be examined across texts to reveal conceptual strategies underlying forum users’ self-representation and discourse management. I also sketched conceptualizations of the self as discussed in contemporary social theory, highlighting tensions between individualist differentiation and a sense of belonging within disembedded, potentially superdiverse digital communities. Additionally, I addressed struggles around trust and conflicting notions of credibility. By reviewing literature on digital discourse, I explained why forums are particularly well-suited for studying how people perform aspects of their identity, especially in light of audience perceptions and the collapsed contexts of digitally mediated interaction. What makes NSIs of the type “I’m not a…” particularly interesting from a linguistic perspective is that negation constitutes a distinctive linguistic choice - one that offers a unique lens for examining the relationship between the micromanagement of identity and contextual features. 2.4 Summary 51 <?page no="53"?> 3 Theoretically Contextualizing Negative Self-Identifiers This chapter lays the theoretical foundation for studying NSIs in English with the aim of creating a formal-functional profile of instances of the structure in use. First, a brief discussion is provided on how the notions of form, function, and meaning are understood in this study. Then, NSIs are theoretically contextualized by reviewing theories on negatives. 3.1 Studying forms, meanings and functions This study adopts a functional perspective toward studying variants of a formally defined structure, namely NSIs of the type “I + copula + NOT + indefinite NP”. It assumes that “meaning resides in systemic patterns of choice” (Halliday & Matthiessen 2014: 23) and that choice is dependent on the functions language is employed to serve in its context of use. This means that, in this study, NSIs are viewed as lexicogrammatical choices in particular co-texts in certain social situations, embedded in larger social contexts. The study takes a corpus-based approach to analyze the discourse-pragmatic functions this structure serves in forum discussions, focusing on the (potentially patterned) communicative effects of NSIs. Corpus-pragmatic studies, as Aijmer (2018: 555) explains, are “form-based, […] mapping words or constructions onto a range of functions”. It is in this sense that the term formal is used in this book (except in the established collocation formal semantics). The functions of NSIs crucially depend on the ideational meanings of the nouns and NPs with which speakers contrast themselves as well as on the linguistic co-texts in which these identifiers occur. The ideational meanings in question are the semantic meanings of lexemes, understood as meaning potentials encoded by particular words, used to differently carve up speakers’ experience of the world (Allwood 2003: 16). To differentiate between NPs in the instances of the structure examined here, these were qualitatively analyzed and annotated in the corpus (Chapter 6) and conceptually profiled through quantitative analysis (Chapter 7). As for the co-textual elements providing the resources for interpreting their discursive functions, NSIs may be preand postmodified within the clause, and they can also be syntactically linked to other clauses. These co-texts, representing particular (formal) grammatical categories (e.g., conjunctions, adverb phrases, <?page no="54"?> or coordinated clauses), were annotated in the corpus so that they could be studied functionally. The clause-internal co-texts of the structure were analyzed in terms of their discourse-pragmatic functions (e.g., adverb phrases serving to index stance). The clause-external co-texts of NSIs were first analyzed in terms of their experiential functions, drawing on the transitivity framework (Chapter 8). Then, they were studied in terms of their overall communicative function (e.g., to provide advice) (Chapter 9). Since NSIs can also functionally interact with units above the sentence level, such units of text were identified by means of qualitative analyses, and their (potentially patterned) relations with particular conceptual categories of the focal structure were examined. 3.2 Theorizing negatives In the following section, I will first define negatives in general and discuss NSIs from a formal semantic perspective (Section 3.2.1). I then explain how negation can be formally and syntactically expressed, the meaning differences between these different realizations of negation, and their frequency of usage (Section 3.2.2). The next section briefly discusses the issues of presupposition and scope, addressing what negatives negate (3.2.3). In Section 3.3, I characterize NSIs of the type “I + copula + NOT + indefinite NP” more specifically by explaining why negative self-identification with NPs, rather than with, for example, adjectives, was explored in this study. Section 3.4 presents the results of a pilot study investigating whether speakers, when presented with a number of prompts, would actually use the structure in focus (3.4.1). It also discusses the use of the structure in the spoken component of the BNC 2014, an 11.5-million-word corpus comprising transcripts of recorded informal conversations in English collected between 2012 and 2016 (Love et al., 2017) (3.4.2). Chapter 4 then explores how NSIs can be studied to learn about their functions in interaction. 3.2.1 Negation in formal semantics Negation, as a logical operator, reverses the truth value of statements. This means that if a proposition P is true, then -P is false. From a formal semantic viewpoint, P and -P cannot both be true (Law of Non-Contradiction) and cannot both be false (Law of the Excluded Middle). A third classical “law of thought” is the principle of Double Negation (--P = P) (Horn & Wansing 2020). Accordingly, negation as a logical operator can cancel itself, so that, for example, impossible is equivalent to possible. 54 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="55"?> Negation poses a problem for truth-conditional semantics insofar as it is not possible to specify necessary and sufficient conditions under which negative sentences are true. For instance, it is possible to define such conditions for the sentence Eva is a snake. Taking a compositional approach to meaning ( Jönsson 2008), we could claim that the meaning of snake consists of the components [reptile] and [legless], and that the sentence is true if both propositions (She is a reptile ∧ She is legless) hold. However, when we try to set up similar conditions for negative sentences, complications arise. The sentence Eva is not a snake (-snake(Eva)) could be true for several reasons: she might not be an animal at all, she might not be a snake but a dinosaur, or, from the speaker’s perspective, she might not just be a snake but the snake (i.e., the most special of all snakes). Formal semantics relies on De Morgan’s law to solve this problem, which postulates that the negation of a conjunction is equivalent to the disjunction of the negations: -(P ∧ Q) ≡ (-P ∨ -Q) This means that the negation of a conjunction is true if at least one of the conjuncts is false (Horn & Wansing 2020). Accordingly, the entire sentence is false if both propositions hold, but true if at least one does not (cf. Kempson 1977: 119-20). For instance, the sentence Eva is not a snake (-snake(Eva)) would be considered true if Eva has legs. While truth tables can perfectly explain sentences if viewed in terms of truth values, they cannot account for the kind of utterances examined in this study, which is concerned with the functions of negative self-identification for speakers’ identity management in discourse. Think, for instance, of the utterance I’m not an expert. The ‘truth’ of this statement depends less on logical conditions and more on what is taken to represent expertise in the particular communicative situation. Moreover, it may not only serve to make a descriptive statement about the speaker, but also perform a range of rhetorical functions in relation to its coand context (see Chapter 4 for a detailed discussion of these relations). That discourse-oriented approaches are more apt to studying NSIs in actual textual events also shows in the fact that, while contradictions are logically impossible, they can make perfect sense in authentic situations of language use. For example, I could say that my dad is not my dad to convey that the person who has been taking care of me all my life is actually not my biological father. In terms of Gricean Maxims, this could be explained as flouting the Maxim of Quality (Grice 1975), which involves saying something obviously false to generate additional meaning. Similarly, relevance theorists (e.g., Sperber & Wilson 1986) explain the acceptance of contradiction in natural language by arguing that hearers seek a meaningful interpretation by assuming one of the 3.2 Theorizing negatives 55 <?page no="56"?> 4 It could be argued that a fifth category is constituted by no, which can function as an elliptical response to a yes/ no question (e.g., Do you like George Clooney? - No.); as an interjection expressing the polarity of a clause following an explicit proposition (e.g., Well, the Conservatives are always telling us how the poor are poor because they want to be… not because they have no opportunity to lift themselves up, no, that couldn’t be the reason); or as an interjection cataphorically marking the polarity of an upcoming clause (e.g., And for the last time, no, I am not a shill for Miles or Morath; So, no, if an American band did the same thing, the SWAT team would NOT be called). All examples are drawn from the Corpus of Contemporary American English (COCA, Davies 2008). 5 The usefulness and acceptability of working with invented sentences in linguistics (e.g., by generative grammarians such as Chomsky 1987) can be subjected to critique: arguably, introspectively generated examples are neither objective nor empirical (Talmy: n.d.), but they can be useful as illustrations, serving to facilitate the reader’s comprehension (Cook 2002: 264). propositions to be the one actually intended as true. In my example, this would be the proposition that the subject in question is not my biological dad, which in turn prompts the hearer to reinterpret my use of dad as meaning something like “the person who acts as or appears to be my father” (Hidalgo-Downing 2000: 216). Both approaches highlight that listeners work to extract coherent meanings even from utterances that contradict logical principles. 3.2.2 Formal realizations Negation can also be viewed in terms of the various ways in which it can be formally realized and interact with other operators (Horn & Wansing 2020). This means looking at negation from the perspective of (formal) grammar, that is, grammar considered as the “outward form taken by systemic choices” construing particular meanings (Halliday & Matthiessen 2014: 23). In basic terms, there are four types of negation in English, viz. not-negation (or VP-negation), no-negation (also referred to as NP-negation), morphological negation and inherent negation (Givón 1993: 202). 4 According to Payne (1985), the most prototypical case of negation is a negative as opposed to a positive verb phrase (e.g., I’m studying a lot vs. I’m not studying a lot). Negation can also refer to NPs (e.g., I’m no girly girl), to pairs of antonymic lexemes, where the use of the marked form suggests a negation of the unmarked form (e.g., I’m happy vs. I’m unhappy), and to lexemes which have ‘negative’ as part of their semantic meaning (e.g., I’m happy vs. I’m sad). To begin with the most frequent form, not-negation is formed by inserting the negative particle not or its contracted form n’t after the operator, that is, the verb used for clause negation or for forming interrogative clauses, which can be an auxiliary verb, the copula be, or the auxiliary do as a dummy operator. The following (invented) 5 examples serve as illustrations: 56 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="57"?> 3.1. He can’t do it. 3.2. He’s not a good writer. 3.3. He doesn’t like writing. In spoken language, negation usually takes the shape of a contraction being attached to the word preceding it. Depending on what word precedes the con‐ traction, we can distinguish between verb contraction and negative contraction, which are complementarily distributed, that is, they can never co-occur in the same clause. Verb contraction occurs with the verbs be, have, will and would, which attach as contracted forms to a host - a pronoun, proper name or wh-word - preceding them (e.g., She’s not listening, Eva’s not coming, Why’s he leaving? ). In contrast, in negative contraction the contracted negator attaches to the operators be, have, do or a modal verb preceding it (e.g., He just can’t cook, there wasn’t enough beer). While verb contraction is favored with first and second pronoun subjects (You’re not being nice rather than you aren’t being nice), negative contraction is preferred with the verbs have, will and would (He won’t do it rather than he’ll not do it) (Varela Peréz 2013: 260). There are six verbs which, unlike lexical verbs, do not need a dummy do because they can serve as operators themselves, viz. have, and the semi-modals have to, need, dare, used to and ought to. For example, both you don’t need to cry and you needn’t cry are possible, albeit with differences in style and frequency of use (the latter being used very infrequently and exclusively in British English though, according to Biber et al. 1999: 484). In contrast to not-negation, the negation marker no introduces an NP as a determiner, e.g.: 3.4. I’m no lady. 3.5. He gave her no choice. Alternatively, negation can be incorporated into pronouns or pronominal adverbs such as nothing, nobody or never (Givón 1993: 205) as well as into quantifiers such as few (Horn & Wansing 2020), so for example: 3.6. Nobody came to the party. 3.2 Theorizing negatives 57 <?page no="58"?> 6 The BNC (World Edition) demo refers specifically to the demographically sampled component of the British National Corpus, containing approximately four million words of spontaneous spoken conversation recorded in the UK between 1985 and 1994 (see Xiao, McEnery and Qian 2006: 111). 7 With other negative affixes like deor un-, I think it is more difficult to argue that they realize negation - at least in the sense of ‘the opposite of affirmation’ - because they seem to be used to describe a different kind of process rather than to negate an implied affirmation. For example, while I disagree could also be expressed by I don’t agree, when Joe Cocker sings “unchain my heart”, he is not asking the implied addressee not to Both the negator not and the determiner no can be used for clausal negation and convey the same propositional content (compare, for example, She had no clue and She didn’t have a clue). However, corpus findings by Biber et al. (1999) show that the usage of the two types of negation significantly varies according to register and that, overall, no-negation is used considerably less frequently, especially in spoken communication. They report that only one out of ten conversations, and three out of ten samples of written language in use, feature no-negation (169-70). This finding is also supported by Xiao and McEnery (2010: 111), who examine the Freiburg-LOB Corpus of British English (F-LOB) and the BNCdemo. 6 Additionally, in cases where no-negation and not-negation can be used interchangeably to convey the same propositional content, no-negation has been argued to be the more emphatic choice (compare, e.g., he’s not a head of department, which suggests that the person simply does not have that job vs. he’s no head of department, which could be used to say that the referent does not have the appropriate leadership qualities) (Biber et al. 1999: 169). Sentences with a negative word such as nobody as subject cannot be expressed by not-negation in Standard English. As Martínez (2013) argues, however, double and multiple negation as well as other non-standard variants of negatives, like innit (invariably used as a question tag) and the adverb never (used as a negator in statements in the past tense) are becoming more frequent in spoken modern English. Non-standard negation is also discussed in depth in Anderwald (2002). As for morphological and inherent negation, these patterns, in contrast to syntactic negation, are only possible with particular lexemes, that is, they are more idiosyncratic. Morphological negation means that a negative affix (e.g., the prefix un-) is attached to the unmarked positive form and the sentence is thus marked as negative. An example cited by Givón (1993: 202) is I think she’s unhappy, which, because the adjective unhappy is the morphologically marked form of happy, is perceived as negation of the ‘unmarked’ assumption that she is, in fact, happy. Negative morphemes can, of course, attach not only to adjectives, but also to verbs and nouns. As for verbs, morphological negation is mainly realized by the prefix disas in disapprove; 7 nouns can be turned into 58 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="59"?> ‘chain’ it - in fact, she already has his heart, so expressing the song’s hook line by “don’t chain my heart” would tell a completely different story. So from a logical point of view, it seems that unchaining and chaining cannot both be true (given that it is the same chain and the same body part, etc.), but they can both be false, if no action involving any chain, metaphorical or literal, is performed. 8 This observation was, by the way, a subject of interest for Plato, who noted that being ugly is by no means the same as not being beautiful, in his discussion of negation in The Sophist: “When we assert non-being, it should seem what we assert is not the contrary of being, but only something other” (Taylor 1971: 164). negatives by adding the prefixes dis-, de-, or antias in disadvantage, dehydration or anti-Marxist. A closer look at morphological negation, however, reveals that there are very few cases in which the lexeme marked for negative can be used to actually express the negation of its positive counterpart. While logically it could be argued that He is not happy is equivalent to He is unhappy because both sentences formally express negations of the proposition He is happy, not being happy does not necessarily mean being unhappy in real life 8 (Stubbs 1983: 111). Likewise, not being a Marxist does not necessarily mean being an anti-Marxist. And it would be very difficult to argue that He suffers from dehydration could equally well be expressed by He doesn’t suffer from hydration. Finally, inherent negatives feature lexemes that have an inherently negative meaning, for example, I doubt his story as opposed to the corresponding affirmative proposition explicitly stated or implicitly present in our background knowledge, that is, I believe his story. Like morphological negation, inherent negatives do not represent prototypical cases of negation because, in many cases, the conceptual domain they are part of cannot be clearly divided into a binary opposition between affirmative and negative as, firstly, a doubt does not equal disbelief, and secondly, belief could be expressed by other nouns such as skepticism. Just as lexical words can come in pairs where one is marked for positive or negative, there are paired grammatical operators associated with a positive or negative context. This phenomenon of complementarily distributed, that is, mutually exclusive grammatical forms is called polarity. Examples of such polarity items are adverbs like already vs. yet and too vs. either or determiners like some vs. any. While negative polarity items are sensitive to syntactic neg‐ ation, they do not appear to be sensitive to inherent negation. So, for example, it is fair to say John isn’t happy and Tim isn’t either, but combining either with the morphological negative unhappy does not work (*John is unhappy and Tim is either). In addition, Martin (1981, cit. in Stubbs 1983: 112) points out, polarity is closely tied to syntax: so, for example, only positive polarity clauses can be exclamative (which is why *What a horrible lecture this wasn’t! is 3.2 Theorizing negatives 59 <?page no="60"?> 9 ‘Comparative’ might not be the perfect term to use here, since true comparisons in the sense of A is better than B are expressed with positive polarity items: while He’s better than someone else is a comparison, stating that he is better than anyone else means that he is, in fact, the best. ungrammatical). In contrast, negative polarity items are used in questions (Any news about the scholarship? ), in conditional clauses (Call me asap if you have any problems), and in comparative constructions (He’s better than anyone else).  9 3.2.3 Scope and presupposition Describing the meanings of negatives in natural language means taking a pragmatic perspective on negation, considering negative propositions as speech acts (Searle 1969: 10). In this section, I briefly discuss two issues that have been explored by pragmatic approaches to account for the meanings of utterances featuring a NEG-operator, namely the scope of negation and the relation between negation and presupposition. Generally, it can be said that negation differs in scope, which means that it can affect single words or phrases (local negation) or the entire clause. The parts of the utterance that do not fall under the scope of negation are positively presupposed, that is implicitly understood as asserted. As for local negation, optional arguments in the clause (i.e., benefactives, associatives and instrumentals or adverbs of time, frequency and place) attract the focus of negation and thus define what is to be taken as presupposed and as negated information (Givón 1993: 198-9, Biber et al. 1999: 88), as can be seen in the following examples (which were invented for illustrative purposes): 3.7. I’m not a feminist right now. (→ The focus of negation here is on the adverb of time (right now), so the utterance does not mean that I am generally not a feminist.) 3.8. He didn’t insult her on purpose. (→ While it is presupposed that he insulted her, what is negated is that he did it on purpose.) 3.9. I don’t want to be in a relationship with you. (→ Just because the speaker does not want to be in a relationship with the addressee, this does not necessarily mean that she generally does not want a relationship.) The same is true for presuppositions coded in relative clauses, verb complements and adverbial clauses: 60 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="61"?> 3.10. I love the guitarist playing barefoot in the square. 3.10’. I love the guitarist playing barefoot in the square. (→ Whether I love her or not, she’s playing barefoot in the square.) 3.11. He knew that she was at home. 3.11’. He didn’t know she was at home. (→ But it is still true that she was at home.) 3.12. Not feeling well, I cancelled the meet‐ ing. 3.12’. Despite not feeling well, I didn’t can‐ cel the meeting. (→ I still wasn’t feeling well.) Finally, negation can be located in adverbials, in which case it does not affect the rest of the clause. For example, Biber et al. (2003: 245) mention adverbials like unexpectedly, the negation of which does not influence the interpretation of the rest of the clause: 3.13. Not unexpectedly, he turned up late for the meeting. (→ He did turn up late for the meeting, but it was not unexpected.) The question of the scope of local negation is relevant for the research conducted here insofar as there are instances of NSIs where the negation only applies to adjuncts, like the adverb right now in the example above, in which case the NSI can be classified as a merely ‘temporally valid’ one, serving different com‐ municative functions than instances where the entire proposition is negated. The fact that presuppositional inferences are not affected by not-negation poses a problem for truth-conditional semantics when dealing with sentences containing non-referential NPs presupposing the existence and uniqueness of their referent. See Russell (1905), Atlas (1974), and Karttunen and Peters (1979) for the long-standing debate on sentences such as, notoriously, The (present) King of France is not bald, which is ambiguous between a narrow scope reading, where the negation only applies to the claim that the mentioned king is not bald, and a wide scope reading, where the presupposed existence of the king also falls under the scope of the not-negation. As my study considers utterances with first-person subjects, that is, utterances whose referent is clear from the situational context, the questions of the scope of negation in clauses containing non-referential NPs does not represent a problem, apart from the mentioned cases where an NSI contains a locally negatable element, with functional implications. Not considered either, for reasons of space and relevance, are cases of presupposition-cancelling negation such as I haven’t stopped smoking - I’ve never smoked in my life (but see Horn 1985 and Carston 1988 for discussions 3.2 Theorizing negatives 61 <?page no="62"?> of metalinguistic negation). Even if marked uses of negatives such as the one just cited occurred in my corpus, approaching them by discussing the semantics of the NEG-operator would not explain what they are used to do in the actual written conversations examined here. The notion of presupposition does matter for the present study, though. From a cognitive pragmatic perspective, negatives can indeed be considered in terms of what they presuppose, namely an expectation which they serve to defeat (Givón 1993: 188). Another topic relating to the scope of negation is negation in complex clauses (e.g., Bond 2011). Complex clauses display two patterns of negation, according to whether the NEG-operator is in the main or the complement clause. Depending on the complement-taking verb (the verb introducing the complement clause), these two negation patterns can either yield two distinct senses or not (Givón 1993: 201). This can be seen from the (invented) examples below: 3.14. a. I didn’t know you were a feminist. vs. a’. I knew you weren’t a feminist. - b. You didn’t ask me to help you. vs. b’. You asked me not to help you. - c. I don’t want you to leave. vs. c’. I want you not to leave. - d. You couldn’t greet him. vs. d’. You could not greet him. The distinct senses of examples 3.14a and a’ above can be explained in terms of logic, since the negated proposition is the main clause (~ I knew) in one case and the complement clause in the other (~ you were a feminist). With non-implicative verbs such as want, the two senses are similar from a logic viewpoint, but structurally different and used to perform different speech acts: while in I don’t want you to leave, (3.14c), the speaker is the subject of the negated verb want and the focus is thus on their wish, the negation in I want you not to leave (c’) takes scope over the verb leave, whose subject is you, and the utterance thus conveys a stronger directive sense. The same difference in scope of negation can be observed, with an even bigger meaning difference, when comparing 3.14d and d’: while the NEG-operator attaches to the modal could in d, conveying the meaning that the speaker was unable to perform the greeting, semantically, it governs the interpretation of greet in d’, meaning that the speaker has the option of choosing not to greet. To sum up, the complement-taking verbs in the above examples yielded two different senses of negation depending on whether the NEG-operator is in the main or complement clause. In English (and other languages, of course), there are also so-called implicative verbs (Karttunen 1971). These are verbs which carry presuppositions about their complements in that the truth of a proposition they express implies the truth 62 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="63"?> of the proposition stated by their complement. An example of an implicative verb is manage. While the affirmative use of this verb in 3.15a implies that the essay was finished, negating the predicate in 3.15b implies that the speaker did not finish the essay. As 3.15c shows, the ‘swap’ of negation illustrated in 3.14 above is not possible with implicative verbs without effecting a complete change in meaning. As the positive semantic prosody of manage, managing not to do something appears as an odd linguistic choice except, perhaps, in ironic uses - for instance, I could utter example 3.15c ironically to highlight, say, that despite the vast amount of time he had to write the essay in question, he still did not do it. 3.15. a. He managed to finish the essay. - b. He didn’t manage to finish the essay. - c. He managed not to finish the essay. For the present study of NSIs, however, the relation between complement-taking verbs and negatives does not play a role, as only NSIs in the first-person singular are considered. This means that the only possibilities of NSIs occurring as complements in complex clauses are the invented examples I know I’m not a maths genius or complex sentences with a non-first-person subject in the main clause, that is, of the type You know I’m not a metal fan. Even if such cases occurred, the question of different senses of the structure examined depending on the syntactic position of the NEG-operator does not arise, and therefore will not be discussed any further. Another issue that should briefly be mentioned is negative raising (NEG-rais‐ ing), which can be defined as a meaning-preserving transformation that moves the NEG-operator out of its position in the complement clause of a so-called NEG-raising predicate, where it is interpreted, to a higher clause, where it is pronounced. From a semantic perspective, this means that certain sentence-em‐ bedding predicates, if negated, imply an affirmative sentence in which negation extends into the dependent clause they introduce (Fillmore 1963; Gajewski 2007: 289). Compare, for example, the NEG-raising verb think in 3.16a. with the non-NR predicate say in 3.16b. below (both examples are invented). 3.16. a. She didn’t think [he would be home]. → She thought [he wouldn’t be home]. - b. She didn’t say [he would be home]. → She said [he wouldn’t be home]. 3.2 Theorizing negatives 63 <?page no="64"?> While the negative sentence in 3.16a. can be turned into a corresponding affirmative sentence in which the negation has taken scope over the dependent clause introduced by think, this is not possible for example b. because here, the affirmative has a different meaning: just because you do not say A, this does not automatically imply that you said ~A. This is just possible with a small number of verbs. Horn (1978) provides a list of NEG-raising predicates, grouped into five semantic fields: 1. [OPINION] think, believe, suppose, imagine, expect, reckon, feel 2. [PERCEPTION] seem, appear, look like, sound like, feel like 3. [PROBABILITY] be probable, be likely, 4. [INTENTION/ VOLITION] choose, plan 5. [ JUDGMENT/ OBLIGATION] be supposed to, ought, should, be desirable, advise, suggest. Regarding the present study, NEG-raising could play a role if self-identification in the affirmative, occurring as dependent clauses over which a NEG-operator in the superordinate clause has scope, were considered in the analysis. So, for instance, the corpus compiled for this study could theoretically also include utterances like I don’t think I’m a feminist. However, it makes a conceptual and functional difference whether identification with a particular concept is only epistemically modified by a superordinate clause, or whether the copula linking a subject and an indefinite NP is directly negated, to the effect that the negative is part of the proposition expressed, rather than only part of the sentence-embedding predicate: I’m not a feminist is, clearly, a stronger case of negative self-identification. Apparently, self-identification (whether negative or not) has a different communicative effect depending on whether it is used in an independent sentence or as a complement clause after a verb of opinion, perception etc. So, to create another example with a verb from Horn’s list, It appears that I’m not a feminist can be argued to represent a weaker instance of negative identification than the unembedded and thus unmitigated I’m not a feminist. Also, with most verbs in Horn’s list, it is not even possible to invent sentences featuring (negative) identifiers with first-person subjects because they combine with infinitive constructions (e.g., I choose not to be a carnivore). While utterances such as this one - which arguably serves to contrast the speaker with the group of carnivores - could, theoretically, be analyzed as part of this study, they are not included in the corpus examined here for reasons of feasibility. An even clearer case can be made against considering NSIs in clauses dependent on matrix clauses with a different subject, that is, direct quotations (e.g., in He said “I’m no liar”), representations of mental processes of different subjects (e.g., 64 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="65"?> He thinks that I’m not a liar), because obviously, these represent other people’s identity ascriptions and not the speaker’s. To conclude, this section has characterized negatives and NSIs from the perspectives of formal semantics, (formal) grammar and (micro-) pragmatics, explaining why they are relevant for the study of NSIs, or why they do not need to be considered in the analysis. It was outlined which types of negation exist, how negatives can be approached in terms of formal semantic truth conditions, and how negation can be theorized to take scope over certain parts of propositions or negate them in their entirety depending on formal and semantic aspects. As formal accounts of negatives lack explanatory value when it comes to exploring their effects in authentic discourse contexts, the subsequent chap‐ ters will adopt a functional perspective on negative self-identification. Before discussing approaches to account for the functions of NSIs, in the next section, I will briefly formally and conceptually characterize the second, syntactically defined variable of NSIs examined here, namely nouns and NPs. 3.3 Theorizing the identifying noun phrase In this project, the element most decisive for an NSI’s function is what the speaker contrasts themselves with - namely, the noun or NP. The reason for the decision to examine negative self-identification with nouns and thus to omit negatives with similar functions such as I’m not normally jealous, is the special ontological status of nouns in contrast to other word classes and their distinctive conceptual and linguistic properties: nouns are conceptually more stable and autonomous than verbs, adjectives, adverbs, prepositions and conjunctions, which code temporally limited and dependent relations. Therefore, it makes a conceptual difference whether I say that I’m not active or I’m not an active person: while negating being active is likely to be interpreted as description of a more or less temporary state (e.g., I’m not active on social media at the moment), negative identification with the compound active person suggests that the speaker negatively identifies more generally and permanently with the concept of being active, in the sense of regularly engaging in physical activity (Givón 1979: 321, Langacker 1987: 58). What is more, the combination of (negated) copula and noun allows for the closest material (in terms of linguistic signs) and cognitive proximity (in terms of association between concepts) between the subject and a potentially relatively complex conceptual unit. The following invented examples - for illustrative purposes - show this: 3.3 Theorizing the identifying noun phrase 65 <?page no="66"?> 3.17. I’m not a musician. 3.18. I’m not working as a musician. 3.19. I’m not interested in music. Saying I’m not a musician construes a different identity for the speaker than saying I’m not working as a musician or even I’m not interested in music. This, of course, relates to the different process types represented by these examples and the different participant status of the nouns musician and music. In example 3.17, the NSI is an identifying relational process (Halliday & Matthiessen 2014: 2019) negating membership of the speaker in the conceptual category musician. In example 3.18, what is negated is not category membership but the (non-)engagement of the speaker in a material process - working - the field of which is specified further by a circumstantial, viz. as a musician. This means that the conceptual prominence of the noun’s meaning in relation to the speaker’s identity is lower than in the first example (which also shows in the fact that even though the speaker might not currently be working as a musician, they might still identify as such). In example 3.19, the speaker talks about what they are (not) interested in - namely music, which could not even be used to refer to their identity and, in terms of thematic progression, is only of secondary importance in this representation. Thus, the linguistic and conceptual relevance of music for the speaker’s identity is highest in the first example, which is the reason why this study focuses on negated identifying relational processes (specifying be as the copula of choice) with the speaker in the role of the token and an indefinite NP in the role of the value. Regarding the types of nouns examined here, this study deals with count nouns, which are conceptually distinguishable from mass nouns. This is reflec‐ ted in the fact that count nouns can be referred to by means of determiners - the indefinite article a(n) in the case of the structure “I + copula + indefinite article + NP” - and that they can be quantified, that is, counted, and qualified, that is, ascribed properties by means of preand postmodifiers (e.g., a nice doctor, a doctor with good online rankings) (Radden & Dirven 2007: 65). Syntactically, a noun may constitute or - if modified by dependents - serve as heads of phrases, that is, syntactic constituents. The following variants of indefinite NPs are possible and of relevance for the analyses that were conducted for this study: 66 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="67"?> • Premodifier + noun: a real fan • Noun + postmodifier: a fan of the new Audi TT • Noun + (reduced) relative clause: a fan who attends every game/ attending every game • Noun + Noun: an Audi fan This distinction matters because meaning is componential, that is, the meaning of the identifying phrase is created by the way different meaning components are structurally related. For example, it makes a difference whether I negatively identify with fan of the new Audi TT or the compound Audi fan. In the first case, I primarily contrast myself with the noun fan, only specifying the kind of fandom in the noun’s periphery by using a postmodifier, while in the second case, I combine the meanings of the two nouns into a compound with fan as head and Audi as modifier, thereby creating a conceptual unity which is assumed to have a special status in our mental lexicon. When two concepts are combined, their meanings are not just added to each other, but they interact based on a particular relation established by the speaker, which in turn is shaped by, reflects, and influences their world knowledge (Gagné & Spalding 2010). For instance, the compound TF evangelist (taken from my corpus of NSIs) is more than just a merger of the meanings of the nouns TF (standing for Teach First, an educational program in which graduates are placed in schools as trainee teachers) and evangelist (i.e., someone who dogmatically believes in something). The latter is used in its metaphorical sense here, which I take to imply a relation between an actual evangelist and a TF evangelist. Negative identification with compounds like this one appears interesting in the context of this study, because informally and effortlessly contrasting oneself with a relatively complex conceptual combination might suggest that the relation between the two concepts is taken as natural, based on one’s knowledge of the world. In other words, my analysis considers not only the overall meaning of the NPs that speakers negatively identify with, but also their formal appearance and, thus, different meaning structures. 3.4 The structure “I + copula + NOT + identifying NP” in British English To test if NSIs of the type “I + copula + NOT + indefinite NP” in fact represent structures used by native speakers of English to contrast themselves with particular concepts, and to shed light on negative self-identification from a practical, usage-based perspective, I carried out a small-scale pilot survey with 3.4 The structure “I + copula + NOT + identifying NP” in British English 67 <?page no="68"?> 10 The full survey is included in Appendix 3. native speakers of English in the initial stages of this project. The results of this web survey are presented in the next section, which represents a first step toward bridging the gap between formal theorization of negatives and the functional study of authentic cases of negative self-identification. The subsequent section, then, explores the use of the structure in focus in the BNC 2014, to find which formal variants of the structure appear most frequently in a different dataset than the one studied here, and which conceptual categories of identifying NPs feature most prominently in this corpus (a detailed conceptual profile of identifying NPs of the corpus used for this study is presented in Chapter 7 of this book). 3.4.1 A written discourse completion task on negative self-identification The goal of this pilot survey 10 was to ensure that no relevant formal variants of the “I’m not a”structure would be left unconsidered in the corpus. The questionnaire used for this survey was a written Discourse Completion Task (DCT) (Blum-Kulka 1982; Jebahi 2011: 650), in which I presented 17 participants with three different kinds of prompts destined to elicit variants of the “I’m not a”-structure. The first type were seven questions asking respondents whether they iden‐ tified with a particular category, to which they had to provide answers of more than just one word (to avoid simple yes/ no answers). To make sure that respondents’ answers would not be influenced too much by the wording of the questions posed, I varied the identifying verbs, not always using the structure “Are you a(n) (+ noun)”, but also verbs like identify, consider etc. The goal of this question type was to find whether said structure would be used by speakers to directly position themselves in relation to the respective NPs, explicitly referring to the noun in question, or whether they would react more indirectly. A concrete example of a more indirect response I obtained is I eat animal products on a regular basis as a response to the question Are you a vegetarian? . Secondly, I presented them with sentences making affirmative assertions about their identity, such as You’re a pessimist, asking them to react to these claims reusing the indefinite NP in their statement. Thirdly, I asked them to form sentences of prompts merely consisting of a first-person pronoun and a noun in brackets, for example, “I (philosopher)”. The aim of prompts of the second and the third type was to establish whether constructions with be, or other verbs, would be 68 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="69"?> used by speakers, and whether being directly confronted with a claim about their self-identity or being free to form sentences featuring an NP would make a difference. It should be mentioned, though, that this pilot represents an informal, exploratory research survey and has several limitations: firstly, the sample of 17 L1 English speakers is relatively small. Secondly, although I included a variety of nouns from different semantic categories in the prompts, assum‐ ing that they might spark different reactions and expressions of (negative) self-identification (nouns referring to particular professions, e.g., gardener, philosopher; health identifiers, e.g., diabetic, ideological identifiers, e.g., social‐ ist, preference identifiers, e.g., football fan, stamp collector, etc.), these terms cannot be considered representative of identity categories potentially more relevant to the respondents. Thirdly, the answers to these prompts do not represent natural language in use, as they are certainly influenced by the fact that they are elicited, constituting (often, apparently, willingly humorous) responses to my questions rather than instances of language in use as they appear in naturally occurring conversation. The results of the survey are represented in Table 3.1 below. Prompt type Type 1 “Are you a/ n…? ” “Do you consider yourself a/ n…? ” “Do you identify as…? ” Type 2 “You are a/ n…” Type 3 “I (noun)” Number of questions 7 5 9 Number of responses 104 75 161 Use of be (% of respon‐ ses in this prompt type) 31% 60% 48% Use of other cop‐ ula-like predicate constructions (e.g., consider myself, label myself, characterize my‐ self) 10% 9% 8% Use of non-copulative verb to (negatively) identify with noun or concept it designates (e.g., I have some social‐ ist beliefs, I am develop‐ ing a passion for flowers) - 4% 11% 3.4 The structure “I + copula + NOT + identifying NP” in British English 69 <?page no="70"?> Prompt type Type 1 “Are you a/ n…? ” “Do you consider yourself a/ n…? ” “Do you identify as…? ” Type 2 “You are a/ n…” Type 3 “I (noun)” One-word affirma‐ tion/ negation (e.g., nope, yes, absolutely) 47% 14% 7% Statement about the noun or the concept it designates - 9% 25% Other 12% 4% 1% Table 3.1: Results of the pilot discourse completion task by prompt type. Overall, the pilot survey revealed that about half of the constructions used by respondents to self-identify or negatively self-identify with the noun of the prompt featured the verb be (155 of 340, that is, 46% of responses). The highest proportion of responses to type-1 prompts (i.e., questions about the respondents’ identity) were one-word affirmations or negations, often followed by explanations (e.g., “yes, women are the dominant sex” in response to the question “Are you a feminist? ”). I attribute this to the nature of the prompt - questions invite a simple yes/ no answer, and many respondents ignored my request not to provide one-word replies. This highlights that eliciting NSIs is a difficult task: while questions are the most direct way of eliciting statements about participants’ identities, they also risk being answered with a mere yes or no rather than a longer reply featuring an NSI. This result is still interesting, as it suggests that NSIs of the type “I + copula + NOT + indefinite NP” are not frequently used assertively in response to questions. Regarding prompts of the third type (I + noun), almost half of the responses represented constructions of the type “I + copula [+/ - NOT] + indefinite NP”. Twenty-five percent of responses to this prompt constituted statements not about the speaker but about the concept designated by the noun in question (e.g., I (politician) resulted in the sentence “I distrust most politicians”). This suggests that while NSIs featuring the copula be appear to be the default response to prompts of this kind, respondents also felt invited by these prompts to express their opinions on the concepts in question. As for (negative) self-identification using other copula-like verbs besides be, 22 out of 340 responses featured other predicates, namely consider myself (14 instances), become (2), label myself (2), identify as (2), characterize myself, 70 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="71"?> 11 CQPweb is an open-source, web-based corpus query and analysis interface (Hardie 2012). and describe myself. In eight cases, the copula-like predicate of the question was implicitly taken up in elliptical responses such as “No, I don’t”. It can be concluded that while other copula-like verbs are used by speakers to negatively self-identify with particular concepts, they occur rather infrequently compared to the structure “I + copula + NOT + indefinite NP”. Also, the use of some of these verbs by respondents might relate to the fact that they appeared in the prompts. This supports the decision to focus on NSIs expressed by variants of this structure in my study and not to consider variants with other copula-like predicates and verbs. Exactly which formal variants of the structure were examined in this study, and how and from what kind of data source they were collected, is explained in Chapter 5. 3.4.2 The use of the structure in the Spoken BNC2014 As will be discussed in more detail, I created a corpus of instances of the structure “I + copula + NOT + indefinite NP” in English, as used in UK web forums, for three main reasons. First, web forum discussions are considered the “quintessential form” of authentic interactional data (Meredith 2019: 242) (see 4.2.3). Second, forums are sites where individuals from across the world can informally discuss and exchange expertise on issues of common interest in “collapsed contexts” (boyd 2014; Heyd 2014; Pendry & Salvatore 2015) (see Chapter 2). Third, existing corpora of CMC were found to offer no clear advantage over a self-compiled corpus focused specifically on the structure under investigation. To gain insight into how this structure is represented in a different type of data, I also consulted the Spoken BNC2014. Using the corpus analysis tool CQPweb, 11 I searched for the same formal variants of the structure that were included in the corpus compiled for this study. Table 3.2 below presents the results. It displays both the raw output from the corpus search - based on the specified variants (with optional elements in brackets) - and the actual frequencies of valid instances after sorting the data and removing false positives and duplicates. 3.4 The structure “I + copula + NOT + identifying NP” in British English 71 <?page no="72"?> 12 False positives/ duplicates removed. Formal variant Example from BNC Raw frequency (before sorting) Fre‐ quency per mil‐ lion words Fre‐ quency after sorting 12 I’m (adv) not (adv) a/ n (adv) (adj) + NP I’m not an expert on Chinatowns of course we don’t have a Chi‐ natown here 301 matches in 225 texts 26.351 246 I (adv) am (adv) not (adv) a/ n (adv) (adj) + NP I am not naturally an athlete or a gym‐ nast… 9 matches in 9 texts 0.788 9 I’m (adv) no + NP I’m no expert skier… 10 matches in 10 texts 0.875 10 I (adv) am (adv) no + NP I mean I am no ex‐ pert… 2 matches in 2 texts 0.175 2 I’m (adv) a/ n (adv) (adj) + NP I’m an introvert but I’m an introvert that flits between… 716 matches in 393 texts 62.683 - Table 3.2: Frequencies of variants of negative self-identifiers in the Spoken BNC2014. Optional elements (adverbs, adjectives) are shown in brackets. Figures are based on a search of 11,422,617 words from 1,251 texts in the Spoken BNC2014. The final variant (“I’m a/ n…”) is included for contrast and was not sorted, as it does not express negation. While these results represent only a first step toward studying the use of NSIs in authentic British English, they show that the most frequently used variant of the structure in spoken English is “I’m not a/ n + indefinite NP”. In contrast, the uncontracted (I am not) variant and the no-negated forms are used extremely rarely. Since interpreting the frequency of a particular phrase is difficult without a reference value, I also searched the corpus for non-negated variants of the structure - that is, affirmative self-identifiers. These occur approximately three times as often as NSIs. The findings support previous research on register differences in the use of no and not negation (Biber et al. 1999; see also earlier in this chapter), as well as on the differing status of affirmative and negated utterances, with the latter representing the marked option - a point discussed in more detail in Chapter 2. The set of NSIs retrieved through this corpus search is revisited in Chapter 7, which presents a conceptual profile of such expressions as used in web forum discussions. 72 3 Theoretically Contextualizing Negative Self-Identifiers <?page no="73"?> 3.5 Summary This chapter has contextualized negation by first discussing its treatment as a logical operator in formal semantics, and second by examining how it is approached in formal grammar and pragmatic accounts concerned with the conventional meanings of negative utterances. The aim was to provide a theoretical backdrop for the phenomenon examined in this book - namely, negatives of the type “I + copula + NOT + identifying NP” - to clarify which forms of NSIs are considered in this study and to outline the types of questions that can be asked about their effects. As shown, however, what NSIs do in real-life language use often exceeds what can be explained on purely formal pragmatic grounds. Conversely, topics typically discussed in formal pragmatics often focus on marked or invented examples that are unlikely to appear in a corpus of authentic usage. This chapter also justified the focus on negative self-identification using nouns, explained the formal variants of identifying NPs, and outlined the conceptual - and thus functional - differences these variants entail. In addition, it presented the findings of a written DCT, which offered an alternative perspective on negative self-identification, as well as corpus data from the Spoken BNC2014. The DCT results suggest that the structure “I + copula + NOT + indefinite NP” is frequently used and therefore represents a promising candidate for corpus-based analysis. The corpus search confirmed that in spoken British English, the variant I’m not a/ n + identifying noun is by far the most frequently used. The finding that affirmative self-identifiers occur roughly three times as often supports the interpretation of NSIs as a marked linguistic choice. The next chapter provides a functional account of structures such as NSIs, exploring how the relationship between a linguistic form and its meaning potential can be theorized in relation to textual and contextual factors. 3.5 Summary 73 <?page no="75"?> 4 Functionally Conceptualizing Negative Self-Identification This chapter approaches NSIs from a functional, discourse-based perspective and provides the conceptual foundation for the theoretical and analytical model used in the empirical part of this book. Negative self-identifiers are considered both as situated microlinguistic interactions and as a transtextually recurring interactional type that can be interpreted in light of broader macrosocietal issues. In alignment with Hasan (2009), the chapter starts from the premise that “although the formal mechanisms which enable the construal of meaning are inherent to language by virtue of its being a semiotic system, the actual semantic reservoir of a language […] is in fact responsive to the speech community’s uses of language in the contexts of their social life” (259). Accordingly, I begin by demonstrating why NSIs are best approached as speech acts, rather than analyzed solely through formal grammatical or semantic frameworks. I will then turn to a more detailed discussion of the notion of context. This is essential for understanding what they are used to do and for accounting for linguistic structures like NSIs, which can serve different functions depending on the textual, situational, and wider cultural contexts in which they occur. I will thus move from a pragmatic characterization of NSIs as speech acts toward a broader exploration of approaches to language in use that theorize the relationship between linguistic structures and varying notions of context. 4.1 Negative self-identifiers as speech acts So far, I have reviewed key issues related to negation as they have been examined in grammar and formal semantics. However, one question that has not yet been addressed - but is vital to the goals of this study - is: why do people use negation in the first place? Traditionally, negation has been viewed as a kind of ‘second choice’. Leech (1983: 100), for instance, considers negative assertions “pragmatically less favored than positive ones” because they are “less informative than their positive counterparts”. For Leech (ibid.), negation is a marked option within the framework of Grice’s (1975) Cooperative Principle, as it is typically avoided in favor of a positive statement. This view is based on what Leech (ibid.) terms the “Sub-Maxim of Negative Uninformativeness”, in combination with the Maxim <?page no="76"?> of Quantity: when considered in isolation, negative utterances are seen as less informative than positive ones and, due to their more complex syntax, more difficult to process. For this reason, it would seem odd for B to respond to A in the following way: 4.1. A: What’s new? - B: I didn’t win the lottery. This view of negation is, in principle, shared by Givón (1993: 191), who argues that negative utterances are pragmatically less likely because they are cognitively less salient than positive assertions. Drawing on the figure-ground distinction from Gestalt psychology, he suggests that non-events can be im‐ agined as cognitively non-salient ground - inertia - against which positive assertions, that is, events or changes, stand out as figures. By presupposing the corresponding affirmative as situationally relevant, shared knowledge - and thus as a kind of norm - negative assertions, such as the structure examined in this study, function to make a counter-normative statement. Rather than introducing new information, negative assertions, in Givón’s view, serve to “correct[ ] the hearer’s mistaken beliefs” (1993: 190). Due to this function - what he sees as “a robust discourse-pragmatic component” - Givón (2018: 188) argues that negative assertions should be distinguished from the three traditionally recognized speech acts, and instead be seen as a type of speech act in their own right. Negatives represent a ‘breach’ of the norm not only conceptually, but also statistically: they are used considerably less frequently than affirmatives. For example, Jordan (1998: 714) cites a corpus study by Huddleston et al. (1968), which found that only 4.5 percent of all clauses examined were grammatically negative. Similarly, Halliday and James (2005) report that in the 18-million-word written component of the COBUILD corpus, the ratio between positive and negative polarity in finite clauses is 0.9 to 0.1. Another indication that negation is an exception to default communicative norms is its somewhat delicate status in social interaction. As Givón (1993: 193) notes, disagreement is often mitigated through softening devices such as modals or adverbials: speakers are more likely to say Maybe that’s not the best idea or That might not be a good idea than to assert bluntly That’s not a good idea. Similar observations are made by Stubbs (1983: 177-78), who refers to studies on the elliptical forms yes and no, which found that refusals are rarely expressed 76 4 Functionally Conceptualizing Negative Self-Identification <?page no="77"?> 13 It should be noted that the relative frequency and directness of negative responses vary across languages and the cultural contexts in which they are used. For example, Wierzbicka (2003) and Ogiermann (2009) show that some languages (e.g., Polish, Russian) permit more direct refusals than English, while others (e.g., Korean) tend to prefer highly indirect forms (see Kwon 2004). through a simple no - suggesting that negation, disagreement, and refusal are culturally less favored than affirmatives. 13 It is precisely this ‘problematic’ status of negation that makes negative assertions - particularly those involving the speaker’s self-identity - worthy of closer examination. Why make a counter-normative statement about yourself by saying what you are not, when you could simply assert what you are? As I have argued, approaching NSIs solely through the lens of formal semantics - that is, in terms of truth conditions - is insufficient to explain the communicative functions they perform in real, situated language use. As Jordan (1998) points out, while the utterances My cook is not a man and My cook is a woman are ideationally equivalent, the use of the ‘norm-reversing’ negative assertion implies additional textual and/ or interpersonal meanings that emerge in interaction with the surrounding co-text and non-verbal context (Halliday 1973). This claim can easily be supported with examples: introspectively playing around with the NSI I’m not a doctor by placing it in different linguistic contexts shows that its meaning seems to change depending on its textual surroundings. Even if there is no co-text, the NSI seems to evoke a situational or even wider cultural context allowing for its interpretation: 4.2. A: Are you a doctor? - B: No, I’m not a doctor. 4.3. A: Do you think I can take another Aspirin? - B: I’d say yes, but I’m not a doctor. 4.4. I’m not a doctor, can I still join the forum? 4.5. Tonight, I’m not a doctor - let’s get this party started. 4.6. I’m not a proper doctor, just a linguistics PhD. 4.7. No doctors. While in 4.2, the NSI is used assertively, uttered in response to a question, it is used in 4.3 to mitigate the epistemic status of speaker B’s answer. In 4.4, negatively identifying as a doctor triggers background assumptions about the 4.1 Negative self-identifiers as speech acts 77 <?page no="78"?> mentioned forum as the situational context of language use (namely, that it is normally only used by doctors). In 4.5, the NSI is used to deny assumptions about doctors implicitly present in the wider context of language use - perhaps that doctors are not normally expected to celebrate. In 4.6, a modified NSI introduces an evaluative partition within the category (‘proper’ vs. ‘not proper’ doctor), positioning the speaker on the non-medical side while still leveraging the category’s social meanings. The last example, 4.7, presupposes a normative affirmative in which doctors are allowed (to wherever) and, incidentally, also illustrates that the “mutual contextual beliefs” (Bach & Harnish 1979: 5) activated by and relevant for the interpretation of this utterance include knowledge about the wider cultural context: most people would probably expect “No doctors” to be written on a sign on the door of a building, for example, and wonder why just doctors are forbidden to go there (for similar examples, cf. Jordan 1998: 713). These examples demonstrate that the contexts which NSIs evoke and interact with could be anything from linguistic elements directly surrounding them to the broader cultural context in which they are used and against which they are interpreted. Depending on their contexts of usage, NSIs realize different meaning potentials, thus serving different communicative functions. According to Givón (1978: 22), “all it takes for the linguist to discover these communicative functions of negation is to probe the discourse context within which negative clauses are used in natural communication”. Indeed, exploring NSIs from a usage-centered perspective appears a promising endeavor, yielding new insights into how the grammatical and lexical properties of negatives in use interact with their textual and contextual environments to create meaning in interaction. As Jordan puts it: Studies of direct and implicit negations in their true textual and contextual environ‐ ments are needed to determine the ways by which negations display significant con‐ textual and interpersonal meanings in addition to their ideational sense. No amount of scholarly debate regarding the ‘meaning’ of such negative and positive statements can be complete without considering their contextual, textual and interpersonal roles in natural language. And for this we need to study actual communicated examples of negation within their contexts of situation (Malinowski 1935) and use. Thus, truly pragmatic studies are required. ( Jordan 1998: 706) A core assumption in pragmatics is that utterances do not have fixed, situa‐ tion-independent meanings, but that “meaning-making is a dynamic process, involving the negotiation of meaning between speaker and hearer, the context of utterance (physical, social and linguistic) and the meaning potential of the utterance” (Thomas 1995: 22). The aim of pragmatic analysis of language is to 78 4 Functionally Conceptualizing Negative Self-Identification <?page no="79"?> find out what makes language use effective in a particular context, or, to put it in van Dijk’s (2008: 6) words, to study “appropriateness, that is, the rules that adapt text and talk to the constraints of their social environments”. To answer this question, we can look at language from essentially two viewpoints: we can either look at what forms can be used to successfully, or appropriately, realize a particular function, asking questions of the type “how do people use language to do X? ”, or we can explore what functions a particular form fulfills (Cameron 2001: 72). My study falls into the second category, because it examines the functions of a formally defined linguistic phenomenon. As first proposed in John Austin’s speech act theory (1962/ 1975), language is used to do things in the real world, and explicating the relationship between what is linguistically encoded and what is pragmatically accomplished is a central concern in pragmatic research. Seen as speech acts rather than merely constatives, NSIs are less about making propositional claims (Huang 2015: 128) than about accomplishing communicative effects in interaction. Because they negate rather than assert a category membership, NSIs carry little propositional information in themselves; their relevance lies instead in how they manage alignment, stance, and identity in context. How precisely this is achieved depends on both formal and contextual aspects, whereby the relationship between form and function is in most cases not one-to-one, but many-to-many: a form might realize many different functions, and one and the same function might be expressible by different forms (Cameron 2001: 73). As for formal features, the relationship between what is formally expressed (the locution), what is intended by the speaker (the illocution), and what is understood by the hearer (the perlocution) is - or appears to be - relatively straightforward in the presence of so-called performative verbs. These are verbs which perform the illocutionary act they denote. Thus, the utterance containing the explicit performative I promise not to cheat can easily be interpreted as a promise. Here, form and function are directly interpretable without much contextual information. However, explicit performatives do not invariably succeed - after all, what looks like a promise could be used as a threat. Thomas (1995: 46), discussing the collapse of the performative hypothesis, cites cases where the performative promise does not hold because the conditions for felicity are violated, for example when I promise is uttered as a threat rather than a genuine commitment. Just as with explicit performatives, the performative effect of NSIs appears tense-sensitive. For example, the past tense I promised to come over neither functions as a promise nor as a threat, but merely as a descriptive statement (Condoravdi & Lauer 2011: 1). The same applies to NSIs: while lacking a performative verb, they seem to function as self-representations 4.1 Negative self-identifiers as speech acts 79 <?page no="80"?> only in the present or present perfect. When shifted to the past (e.g., I was not a member back then), they serve a descriptive rather than self-representational function. In this study, therefore, I restrict the analysis to NSIs with a present focus (i.e., in the present and present perfect tense), since only these carry the performative force of projecting an identity claim in the here-and-now of discourse. What also complicates matters is that utterances may be used to “do things without performative verbs” (Thomas 1995: 46). For example, the utterance I’m not going to use Google Translate could be used to perform the same speech act as the explicit I promise not to cheat [on this assignment]; however, the speaker may not strictly speaking commit to not cheating by committing to the truth of the claim that they will not use one particular tool - eventually, what they mean to commit to depends on their intention, and not just the words used. In fact, whether speech acts serve a particular intended function or not, and are thus “felicitous” (Austin 1962/ 1975), depends not only on the words used, but on a number of contextual requirements, which Searle (1969) theorized as “felicity conditions”. Thus, the NSI in I’m not normally one to brag may fail to achieve its intended mitigating effect if uttered by someone well known for self-aggrandizement. In such cases, as with Donald Trump’s dismissal of his remarks as mere “locker room banter”, the illocutionary force is undercut by the hearer’s perceptions, which are as crucial as the speaker’s intentions. As we have seen, with NSIs, it does not seem to be possible at all to relate form to function without knowledge of the coand context in which they occur. The only propositional content they have is that the speaker does not associate themselves with the category specified by the indefinite NP, which means that without knowing the context, the utterances I’m not a murderer, I’m not a ticket inspector, and I’m not a carrier of the coronavirus are all the same in terms of speech act theory - they are negated assertives whose illocutionary force and perlocutionary effect can only be guessed on the basis of the meaning of the indefinite NP. We might, for instance, assume that in all three examples the perlocutionary effect may be relief (albeit with different stakes in each case). Grice’s (1975) Cooperative Principle and its four maxims do not seem to work as an initial approach to NSIs, either: this theory posits that speakers are cooperative when they interact verbally, in the sense that they implicitly draw on rational and logical principles that enable purpose-oriented, rather than random and pointless, communication. This means that we assume, per default, that our interlocutors communicate rationally and in a meaningful way. As mentioned above, NSIs (and negatives in general) always constitute a problem in terms of these maxims if viewed in isolation, because asserting what you are 80 4 Functionally Conceptualizing Negative Self-Identification <?page no="81"?> not, technically violates the Maxim of Quantity by providing content that is non-informative: thus, if someone asks me what I do for a living and I answer by saying I’m not a lawyer, the person would probably think I am playing games (except in certain contexts, e.g., a situation where being a lawyer is the default option and the negative self-identification points to a second option evident from the context). In fact, the only case in which this structure is to be taken as a direct speech act observing all maxims and intended to provide information is when it is used as an answer to questions of the type Are you a/ n X? . While speech act theory and Gricean Maxims are of little use in analyzing decontextualized instances of negative identification, they can help to account for certain cases of NSIs as used in (real or imagined) textual and situational contexts. For instance, it would be possible to argue that the utterance I haven’t been a member of this forum for long counts as an apology for having violated certain forum rules. Gricean Maxims have explanatory power in cases where NSIs are used to flout these maxims and hence create conversational implicatures. For example, the following hypothetical exchanges illustrate how NSIs can be analyzed with reference to Gricean maxims: 4.8. A: Do you want to take a bet on the outcome? - B: I’m not a betting person. (Implicature → No, I don’t want to.) 4.9. A: Will I succeed at the exam? - B: I’m not a clairvoyant. (Implicature → I don’t know.) 4.10. A: Did you understand what I was trying to say? - B: I’m not an idiot. (Implicature → This is a stupid question, of course I have understood.) In example 4.8, B flouts the Maxim of Quantity by providing an answer which contains less information - though expressed in more words - than necessary. While a simple “No” would suffice as an answer to A’s invitation, B negatively identifies as a betting person, thus avoiding a clear answer and coming across as either more polite - indirectness might be perceived as friendlier than direct rejection - or, indeed, as rude - declining by negatively identifying as a betting person implies that A can be classified as such. In 4.9, B’s answer is not directly relevant to A’s question, but from their background knowledge, A probably knows that clairvoyants are people who (pretend to be able to) predict the future and could thus answer her question and can thus arrive at the implicature created by B’s negative self-identification. In 4.10, just like in 4.8, an NSI is used 4.1 Negative self-identifiers as speech acts 81 <?page no="82"?> instead of a simple yes/ no answer, flouting the Maxim of Quantity and thus constituting an answer that expresses B’s annoyance at A’s question. While applying Gricean Maxims to account for cases such as these works quite well, it should be kept in mind that real-life instances of language in use do not normally consist of otherwise decontextualized question-answer pairs, but are surrounded by written or spoken text (as we have seen, this also represented the main problem with the written DCT presented in Section 3.4.1). This co-text, in turn, is shaped by and indexes aspects of the situation in which language is used, which is conditioned by who is talking to whom about what and with which goals as well as by, ultimately, what can be said, known, and done at a particular moment in time and space. In fact, the question of what precisely represents the ‘context’ of an individual utterance with which it is deemed to interact is hard to answer once we depart from examples prototypically used to explain speech act functions, like the ones above. This is why context and related questions - such as the possibility of explicating the context-sensitivity of linguistic expressions and the role of social factors in epistemological matters - are not only the subjects of philosophical debates (cf., e.g., Cappelen & Lepore 2005; Goldman & O’Connor 2019), but also present a practical problem for linguistic analysis of discourse meanings created by forms with a meaning potential. As Akmana and Bazzanella (2003: 322) put it, while “context appears to be crucial both on the theoretical and on the applied levels”, it is “difficult to analyze scientifically and grasp in all its different demeanors”. Yet, if NSIs in authentic language use obtain their meaning in relation to their textual, situational and perhaps wider sociocultural context, the question arises how precisely we can describe, collect and analyze this context. According to Aijmer (2013: 12), “if meaning is selected in the communication situation, we need a description of the contextual factors which interact with the meaning potential of the lexical item”. The following section turns to how different linguistic theories conceptualize the relation between meaning potential and context. 4.2 Negative self-identifiers as context-dependent meaning potentials This section explains how NSIs, as speech acts indexical of aspects of both the situational and broader cultural context, can be approached at different levels of textual and non-textual context. It provides the conceptual foundation for the theoretical and analytical model applied in the empirical chapters - examining these structures both as situated microlinguistic interactions and as a recurring type of interaction interpretable in light of macrosocietal issues. 82 4 Functionally Conceptualizing Negative Self-Identification <?page no="83"?> Understanding NSIs in this way requires a model of meaning that is not fixed or purely compositional, but responsive to contextual variables at multiple levels. As Lemke (2000: 222) argues, “the meaning of text is made through time, and never wholly predictably: the aggregate probabilities for each choice that are the system are not only re-weighted for each situation and each text, but dynamically shift during the process of text-production itself ”. This dynamic view of meaning-making is particularly useful for analyzing how the same linguistic structure (such as I’m not a linguist) can take on different functions depending on both its immediate co-text and the broader sociocultural setting in which it occurs. The next subsections operationalize this notion of context responsiveness, beginning with pragmatics, sociolinguistics and variational pragmatics, and moving on to CA systemic functional linguistics (SFL), corpus pragmatics, and finally studies of digitally mediated discourse. As has been demonstrated earlier in examples 4.8-4.10, in many cases NSIs are not used to react to what someone else has said, but seemingly arise out of nowhere. Imagine, for instance, a situation at a linguistics conference where someone from the audience utters the following in the Q/ A session: I’m not a linguist, but I’m not sure if I’m convinced by your data collection method. Here, the NSI firstly relates to the rest of the speaker’s own utterance, premodifying the coordinated clause introduced by but, which, viewed in terms of the preparatory condition (in Searle’s sense), presents a contrast to the NSI as the speaker expresses criticism despite mitigating their authority to do so, and which itself is epistemically modified (I’m not sure) and thus very tentative. But beyond that, under this reading, it could be argued that the use of the NSI indexes various aspects that are relevant on the level of the communicative situation, and possibly the wider sociocultural context: by stressing their non-membership with the social group of linguists, the speaker might be acknowledging that it is seen as (socially, professionally) problematic if a non-linguist expresses doubts about a linguist’s work at a linguistics conference. Thus, their utterance might be interpreted as reflecting and reproducing the social conventions shared (or assumed by the speaker to be shared) among the conference participants in this communicative situation and even in the context of academia in general. The example thus shows, once again, that structures like NSIs can only be sufficiently accounted for if considered in relation to their textual and (more narrowly or more broadly conceived) non-co-text of usage. The further away we move from the meaning of the structure in its local co-text, toward particular contextual variables interacting with particular linguistic variables in patterned ways, the more we move away from traditional pragmatics into the realms of sociolinguistics and variational pragmatics and, ultimately, discourse analysis. 4.2 Negative self-identifiers as context-dependent meaning potentials 83 <?page no="84"?> 14 This conceptualization of context as a mere “disambiguator, or accidental helper in interpretation” (Hasan 2009: 255), has been subject to criticism by functionalists. For example, Hasan argues that “language proved so effective in the performance of human practices not because it presents itself as sounds or graphs. Rather, to prove effective, it must have possessed the property of being meaningful: its elements must have been capable of relating to the experiences of the users” (ibid: 254). From a discourse analytical perspective, (potentially) recurrent functions of NSIs with particular meanings in forums as historically situated sites of language use with particular characteristics can be evaluated in terms of their (potential) sociocultural implications. In my study, I adopt analytical concepts from differ‐ ent linguistic and discourse-analytical accounts of meanings in context to be able to relate the local use of NSIs to questions about the social world. These approaches, amongst other things, differ to a lesser or greater extent with respect to what precisely the notion of “context” involves and on what basis we can claim that a particular speech act reflects, constructs, or is embedded into a certain context. In the following, I will discuss some of the issues that arise when trying to answer this question, and review - though by no means exhaustively - approaches to the relation between utterances and their context-dependent and context-altering functions that are most relevant for my study. 4.2.1 Pragmatic context models In pragmatics, as a part of theoretical linguistics, the focus is on abstract principles of language use rather than on concrete instances of communication. Levinson (1983: 9), for instance, defines pragmatics as “the study of those relations between language and context that are grammaticalized, or encoded in the structure of a language”. Accordingly, speech act theory examines the cognitive mechanisms underlying communication, defined by Huang (2015: 49) as “communication which is intended to be recognized as having been intended”. In this view, context refers to “a cluster of actual states of affairs or events of various kinds, related to the issuing of an utterance and to its intended force” (Sbisà 2002: 422). In other words, context in formalist pragmatic approaches represents an abstract set of conditions, the adherence to or violation of which determines the success of a speech act. 14 These conditions are widely considered to be primarily cognitive in nature. They are described variously as “mutual contextual beliefs” by Bach and Harnish (1979: 5); in terms of hearer’s assessment of the relevance of the speaker’s utterances by Sperber and Wilson (1986); while van Dijk discusses them in terms of cognitive context models, conceived of as participants’ subjective representations of what is 84 4 Functionally Conceptualizing Negative Self-Identification <?page no="85"?> pragmatically appropriate in particular communicative situations stored in episodic memory, serving as an interface between discourse and social situations (2008). Studying NSIs from this perspective means identifying the textual and non-textual contextual conditions under which using an NSI is appropriate. It also involves considering what conclusions the identification of recurring utterance-context relationships allows regarding speakers’ representations of contexts in which negatively identifying with a particular category is deemed appropriate. Importantly, the direction of influence is not just one-way: inter‐ locutors have influence not only on the immediate context they are creating, but also on the larger social context, as emphasized in pragmatics, interactional sociolinguistics, and discourse studies (Thomas 1995; Goffman 1981; Fairclough 1992). NSIs can both presuppose existing context models (for example, by disclaiming expertise in situations where it is expected) and recalibrate them (by redefining what counts as legitimate participation in that situation). This dual potential is of theoretical relevance for my study as analyzing how people use NSIs in and across web forum discussions can show how speakers strategically, and potentially routinely, manage the interpretation of what they say in these discourse contexts, which in turn can be seen as pointing toward their underlying assumptions about how best to represent themselves to achieve their communicative goals. 4.2.2 Context in (interactionally oriented) sociolinguistics and variational pragmatics Being interested in the relations between language users and their social identity and language use, sociolinguistic approaches (e.g., Labov 1972; Hymes 1974) the‐ orize context in more social terms, seeking to describe how particular contexts systematically constrain people’s talk, thus characterizing “the communication conduct of a community” (Hymes 1974: 9). Referring back to the pragmatic approaches described in the previous section, where context mainly serves to evaluate the appropriateness of certain speech acts, these approaches could be argued to examine the relation between context and text the opposite way by taking social and situational variables as their starting point to investigate linguistic difference. Seeking to account for how individuals - rather than contextually defined communities of practice - manage their discourse in particular communicative situations, variational pragmatics assumes that speech acts are both contextually influenced and “context-changing actions” (Sbisà 2002: 434). If we conceptualize NSIs as context-changing speech acts and examine their functions in terms of 4.2 Negative self-identifiers as context-dependent meaning potentials 85 <?page no="86"?> their adherence to interpersonal maxims (Leech 1983) and to the management of social distance, we might, for instance, explain how they are used to communi‐ cate politely (Brown & Levinson 1987) or, alternatively, impolitely (Culpeper 2009). For instance, the NSI I’m not a virtuoso like you, but I happen to know what staccato is could mock-politely hint at the arrogance of an overly self-confident fellow guitarist. This contrast illustrates how NSIs can function both as redres‐ sive strategies that mitigate face-threatening acts (as in the conference example) and as aggravating strategies that enact mock-polite disalignment (as in the guitarist example) (cf. Brown & Levinson 1987; Culpeper 2009). Viewing NSIs as speech acts strategically used for face-management implies a perspective according to which these structures are not only linked to their immediate co-text, but also to a range of non-textual aspects specified for the context of the communicative situation. How - or whether - to perform a face-threatening act (FTA) depends greatly on the social identity of the participants, and especially on their relationship. What plays a role here, amongst other things, is horizontal social distance (i.e., how well the interactants know each other), vertical social distance (i.e., the interactants’ respective position in a social hierarchy), and other aspects of social identity such as discursive roles and group membership. All of these aspects are of interest in this study, since it seeks to find whether speakers informally interacting online routinely index non-membership with particular social groups to strategically manage the interpretation of their contributions in online forums. Because such functions are negotiated turn by turn, Section 4.2.3 considers how conversational sequencing shapes the use and interpretation of NSIs. Assuming that structures like NSIs can serve a variety of purposes means adopting a multidimensional view of context and, conversely, a multi-indexical view of structures with a meaning potential. In other words, structures like NSIs are considered to potentially index different aspects that are considered ‘contextual’ - for example, dimensions of identity, stance, activity, or footing. An approach which emphasizes this multidimensionality of text-context interac‐ tion is interactional sociolinguistics (Gumperz 1996; Gumperz & Cook-Gumperz 2008). Criticizing traditional linguistic analyses for viewing context as a “rel‐ atively static, external and determining reference point traditionally added to language analysis as something of an afterthought” (Rampton 2017: 16), interactional sociolinguistics postulates that “context is an understanding of the social world activated in the midst of things, an understanding of the social world that is interactionally ratified or undermined from one moment to the next as the participants in an encounter respond to one another” (ibid.). Interactional sociolinguistics sees speech events and situational and cultural context as 86 4 Functionally Conceptualizing Negative Self-Identification <?page no="87"?> inseparable and normally examines face-to-face interactions, focusing on social differences between interlocutors and concerning itself with “interpretation and understanding and how they are intertwined with the construction of shared common ground” (Auer et al. 2014). Since culture is seen as “constituted in and through situated speaking practices”, it is proposed “to take the speech event as the unit of analysis rather than community-wide linguistic and cultural norms” (Gumperz & Cook-Gumperz 2008: 536; Auer & Roberts 2011: 385). Language, from this perspective - which draws on Silverstein’s (1992, 1993) discussion of indexicality - is “pervasively indexical” (Rampton 2017: 16), which entails that “small-scale interactions” ( Jacquemet 2011: 475) are seen as reflecting and affecting not only the immediate situational, but also the wider sociocultural context. Interactional sociolinguistics puts forward two important concepts to de‐ scribe the processes of interpretation and understanding in linguistic inter‐ action, namely inference and contextualization cues (Gumperz 1982, 1996). Inference refers to the process by which individuals draw on their linguistic knowledge as well as on knowledge gained from previous experiences to make sense of incoming signals in the form of language or other forms of semiosis, while the concept of contextualization cues captures the view that language, produced and interpreted by skilled agents in dynamic interaction, indexes much more than propositional content, reflecting language users’ awareness of contextual aspects and influencing the context. In the case of NSIs, for example, the pattern I’m not a [category], but… (as in the conference example above) can be understood as a cue complex, where the negative identifier and the concessive but jointly signal how the following contribution should be interpreted. This simultaneously acknowledged and interactionally construed and modified context is understood to comprise both aspects of the immediate communicative situation as well as the wider cultural context (e.g., particular norms of behavior of language toward which language users orient themselves in ongoing discourse) and, hence, allows “performing microanalysis of interaction in light of macro-societal issues” (Gordon 2011: 67). In this respect, interactional sociolinguistics draws on Goffman’s (1974: 11) concept of framing, which refers to negotiation among interactants of what is defined as situation, and what this situation means for norms of interaction associated with it. To explain how NSIs respond to and shape the interactional context, the framework by Ochs (1996) is useful. Criticizing studies that “dwell on the relation of linguistic forms to only one situational dimension”, Ochs (1996: 418) names five socio-cultural dimensions that can be considered in linguistic studies interested in the functions fulfilled by the usage of forms with meaning 4.2 Negative self-identifiers as context-dependent meaning potentials 87 <?page no="88"?> 15 Orthographic features are retained, with minor spelling corrections provided in square brackets for clarity (see Section 6.3). 16 This, incidentally, supports claims by scholars of pragmatics such as Sbisà (2002: 426), who argues that “the goals of the conversation determine against which (and how fine-grained) aspects of the same object or event the truth/ falsity of the speech act is to be evaluated”. She further explains that “the situatedness of the speech act goes hand in hand with the delimitation of its context”. Accordingly, the utterance I’m a robot, when used in a real-world context in which people like the speaker cannot (yet) be robots, flouts the maxim of quality and is likely to be interpreted metaphorically. When negated, however, the utterance (I’m not a robot) becomes literally true - except in hypothetical or fictional contexts (e.g., Blade Runner) where robots try to pass for real people. In such cases, the utterance presupposes either an explicit or implicit identity claim in the textual or non-textual context, by which the speaker is either identified as a robot or ascribed robot-like qualities. An example of the first would be Google’s prompt requiring users to confirm that they are real people and not robots. An example of the second could involve a response to the question “Have you answered all the emails yet? ”, where I’m not a robot serves as a metaphorical objection to unrealistic expectations. An example of the second could involve a response to the question “Have you answered all the emails yet? ”, where I’m not a robot serves as a metaphorical objection to unrealistic expectations and thus as a stance of resistance to imposed conformity. potentials: (1) participants’ social identity, (2) the social act performed, (3) the activity speakers engage in, (4) the affective stance, and (5) the epistemic stance. Regarding the question of how a linguistic form with a meaning potential and, more concretely, NSIs, can be claimed to be indexical of these aspects, the example of stance illustrates quite well how meaning potentials can be used to modify the interpretation of what is said and, thus, have an effect on the situational context (Aijmer 2013: 15). This can be illustrated by reference to examples from the corpus, such as 4.11 below. Incidentally, the switch from invented examples to corpus examples to illustrate this point in itself supports the argument for considering NSIs as multi-indexical meaning potentials. It is quite easy to explain certain pragmatic phenomena using made-up sentences which, under certain, equally easily imaginable circumstances, can be interpreted in a certain way. However, the effects of real-life language used in real communicative situations and surrounded by co-texts cannot be explained in terms of simple form-function mappings and are more difficult - if not impossible - to make up. Thus, from now on, all examples cited come from my corpus unless indicated otherwise. In example 4.11, 15 taken from a TV fandom forum discussion, the NSI I’m not a robot negates an identity claim that is unlikely to be presupposed, as A presumably does not consider B to be a robot. 16 Its relevance, therefore, depends on a metaphorical interpretation. What this utterance accomplishes 88 4 Functionally Conceptualizing Negative Self-Identification <?page no="89"?> interactionally can only be understood in relation to its co-text - specifically, A’s preceding turn and the surrounding parts of B’s utterance. 4.11. A: No one likes change, so no matter who they put in there, the avid fans were going to protest to get the old back. And I acknowledge my post does not have any spoilers either, but I would like to read about some if anyone has them. Thank you : ) - B: [NAME], I love TG, but I have a mind of my own, a sense of what is good and what is bad, and I don’t think the Seaver character is good. I’m not a robot that someone can program into liking this or that […] In this exchange, B averts A’s indirect relativization of their dislike of a new character in a TV show (“Seaver”) by saying that B has their own sense of what is good and what is bad. The NSI here intensifies B’s rejection of A’s relativization (and generalization about “avid fans”) and is thus an example of meaning potentials which, in Aijmer’s (2013: 15) words, have “a rhetorical function allowing the speaker to take up a stance of alignment or disalignment to the hearer or to what is said”. The NSI thus simultaneously interacts with A’s turn and with B’s own utterance to index their stance, namely their unwillingness to be convinced by A’s relativization of their dislike. Stance, according to Ochs (1996), can in turn be seen as linked to social identity - in the example, the NSI could also be seen as interacting with A’s generalization about what no one does, and serve to distance B from the general crowd of avid fans and thus, perhaps, to position her as an individual with her own opinion. Moreover, the metaphorical identifier “robot” evokes resistance to algorithmic conformity, a stance that anticipates the critical-discursive dimension of later chapters, where NSIs are interpreted in relation to broader sociocultural pressures on self-representation. Regarding the relevance of variational pragmatic approaches to the present study, studies of discourse markers according to the framework just described focus on identifying “conventionalized meanings that are part of the speaker’s grammatical knowledge” (Aijmer 2013: 30). This is not possible or aimed for in the present study, as NSIs of the type “I + copula + NOT + indefinite NP” contain a semantic variable, viz. the indefinite NP, and thus cannot be examined for conventional meanings. What is analyzed, however, is the extent to which negative identification with the same, or semantically related, NPs can be found to realize similar functions across a sample of local interactions. Accordingly, my unit of analysis is the NSI construction token together with its co-text, rather than the string “I’m not a …” in isolation. To sum up, the main contributions from sociolinguistics and interactively oriented pragmatics to theorizing the relation between language use and social 4.2 Negative self-identifiers as context-dependent meaning potentials 89 <?page no="90"?> context presented in this section are, firstly, sociolinguistics’ theorization of con‐ textual parameters to describe how contexts systematically constrain people’s talk, and secondly, the view of context and text as mutually influencing each other, whereby context is seen as multidimensional and linguistic structures are believed to reflect and index more than a single dimension of the communicative situation. 4.2.3 Context in conversation analysis As the example above has shown, meaning potentials often obtain their function not only in interaction with the rest of what the same speaker says, but also in relation to contributions by other speakers, for example, when they are used to signal disagreement. As Thomas (1995: 196) argues, “almost all speech acts are collaborative” because “collaboration is necessary for the speech act to ‘succeed’”; force, she claims, is in fact also up for negotiation. Although written, online forum discussions form a domain well suited to CA research: their informality and low social distance make them conceptually oral rather than scriptural ( Janich 2017: 44; see also Darics & Koller 2018: 11 for a discussion of why the oral/ written distinction may be less relevant for contemporary communicative practices, which they argue can better be studied in terms of their purpose). For this reason, they can be productively approached from the perspective of CA ( Jefferson 1972; Sacks 1984a, b, 1992; Schegloff 1988; more recently Gibson 2009; Meredith 2017). In what follows, I outline how my project draws on CA insights while also diverging from prototypical CA approaches in its treatment of context. Authentic linguistic data is explored in CA and the role of language in interac‐ tion is considered the “primordial site of sociality” (Meredith 2019: 241). Having its roots in ethnomethodology, CA is interested in how social order is created as speakers interact linguistically, sequentially structuring their talk and thus collaborating to achieve successful communication. Conversation analysis is centrally concerned with recording and analyzing naturally occurring language data, seeking to study language as it is used with as little researcher intervention as possible during data collection, processing and analysis. While traditionally, CA has mainly recorded and transcribed oral face-to-face interaction (Mondada 2013: 35), recently it has increasingly focused on written online conversations, like the forum discussions explored in the present study, considering them to represent data ideally suited for studying authentic interaction. As for the role of context in CA, the general view among CA practitioners is that “contextual features should only be taken into account if observably 90 4 Functionally Conceptualizing Negative Self-Identification <?page no="91"?> relevant for specific interactions” in order to avoid “treating participants as ‘puppets’ of socio-cultural forces” and to minimize the danger of imposing the analyst’s interpretation on the talk (de Kok 2008: 886, cf. also Cameron 2001: 88). In other words, CA treats context as participant-relevant rather than researcher-imposed, recognizing that meaning emerges from participants’ own displayed orientations rather than from analytic assumptions. Being interested in the linguistic and thus conceptual categories language users routinely contrast themselves with in their everyday interactions, structuring their experience of social life, my study shares with CA an interest in members’ own categories, tackled in CA by membership categorization analysis (MCA) (Wooffitt 2005). Studying which categories are referred to across interactional situations on forums, my study is also interested in the relation between potentially patterned functions of NSIs in talk-in-interaction and the wider sociocultural context. Where participants’ orientations toward particular con‐ texts recur across interactions, such patterned relevance can itself be taken as indicative of links to the wider sociocultural domain, which is also the locus of interest in critical approaches. Contextual features are also of relevance in CA, and for the present study, insofar as it is assumed that the technological specificities of the mediums through which interaction takes place “can both afford and constrain the interactional potential” (Meredith 2019: 243). The effects of such “affordances” (Hutchby 2001; Evans et al. 2017; Tagg et al. 2017) are particularly accentuated in written online communication (see Section 4.2.6 for a discussion of other aspects of online contexts). For example, online forums typically display a different organization of turn-taking than spoken conversations. As Meredith (2019: 245) explains, individual postings in forum discussions tend to contain several turn constructional units (TCUs), for example, and speakers may choose to respond to all or just some of them in their response; certain turns may not even be answered at all. This is likely to relate to the tendency for forum discussions to be asynchronous instead of quasi-synchronous (as would be the case with chatroom discussions), which means that more time passes in between contributions by different speakers, leading participants to write longer postings including more TCUs. Transition relevance places (TRPs), too, play a different role in forum discussions, as they are normally simply represented by the end of a posting and not by pauses or completion of particular speech acts, as would be the case in synchronous conversations. In other words, whereas spoken interaction makes next-speaker entry relevant at points of syntactic or pragmatic completion, online forums bind turn-transition to the technical boundary of a posting. 4.2 Negative self-identifiers as context-dependent meaning potentials 91 <?page no="92"?> These contrasts illustrate how the technological affordances of forums shape interactional sequencing - a theme taken up again when discussing digitally mediated contexts more broadly. In terms of turn management, speakers in forums mostly self-select, responding, for example, to a question asked in a thread-initial posting. Speakers may also be selected by others, for example, through mentioning or through tagging their names, depending on the design of the forum. As will be explained in more detail in Chapter 5, considerations of the se‐ quential organization of forum conversations had an impact on the way the data for this study was collected and annotated. In terms of data collection, co-textual postings were included based on their relevance for interpreting the posting containing the NSI. Regarding the way the data were annotated, I marked up turns by different speakers, acknowledging the different conceptualization of what constitutes a turn in online written conversations. Example 4.12 below illustrates some of the features just discussed. In this exchange, the speaker, using the NSI I’m not a great fan of textbooks (speaker A), asks three questions (shaded in grey) in their initial posting. In a spoken conversation, these would probably constitute separate turns. In their response, speaker B does not directly respond to them, but instead replies with a question which appears more as a critical comment on their interpretation of the questions asked (namely “Are you expecting the maths book to do the teaching for you? ”), implying that A might have problematic expectations about teachers’ books. While in a spoken conversation, A would probably answer this question right away, thus using this TCU as a TRP, to fend off the criticism implied. This, however, is not the case in this written conversation because B’s posting continues beyond completion of this first TCU. 4.12. A: -- [SUBJECT] Anyone familiar with Target Maths books who could help me? Just started in a new school and the class are working from ‘Target Maths’ books with which I am unfamiliar. Trying to get my head round how to introduce the topic of work for the page. I like the idea of having 3 levels all on one page but often the LO doesn’t differentiate between the 2 levels (easy to adapt but why put it there then? I ask as [a-non]-specialist Maths person.) Is there a Teacher’s book which provides some starters/ ideas to introduce the concepts? I’m sure once I get used to the books it will be fine, I just feel at the moment I and the children are struggling.Any help would be welcome. I have used Maths Sphere in the past and am thinking of subscribing. Would this fit alongside, do you think? - B: I’m a bit confused by this. Are you expecting the maths book to do the teaching for you? I’ve been using Target Maths/ Maths on Target for years and usually use them to provide pupils with practice questions once we’ve gone through something. My middle group tend to work with them perhaps twice a week. It’s 92 4 Functionally Conceptualizing Negative Self-Identification <?page no="93"?> not a maths scheme to just go through page by page. We’ve just got MyMaths (have missed it for the past few years and used it frequently at my previous school). Again, I wouldn’t use that to ask the children to teach themselves with it. They use it to practise and consolidate content we have covered. - A: Good point [NAME of B]. No I’m not expecting the book to do the ‘teaching’, but just personally I find the ‘reminders’ (teaching point) at the top of the page, not incredibly helpful to the children when they settle to the exercise.and I always have a raft of hands up.Really I feel I would prefer to ditch the books and do my own exercises, based on my own input, but with reports a new job/ class I’m just not finding the time to prepare my own. I’m not a great fan of textbooks and prefer to dip in and out of a variety, but at this school I just have this one and feel I’m expected to use this the previous teacher did and I’m trying to porvide [provide] continuity for the children. There are good points about them and I do like the differentiated exercises. It’s possibly because I’m used to teaching in themes/ topics for a a week at a time, moving on depending on how the class do and this book seems to move on quite rapidly from one aspect to another. It probably is because I’m not used to them and once I get my head round it will be fine.Thanks for your help everyone. In addition to considering individual postings by different speakers as turns in written online conversations, this study refers to CA concepts such as adjacency pairs to explicate the local functions of NSIs. Adjacency pairs are turns by different speakers that are related, in that particular first-pair parts (FPPs) are usually followed by corresponding second-pair parts (SPPs). The very structure of forum discussions means that question-answer adjacency pairs are particularly relevant in the examined data: forum conversations are usually started with a thread-initial posting, referred to as topic initiation (Meredith 2019: 251) and normally featuring a subject line. This posting prototypically contains one or several questions or can be considered as such by virtue of seeking response from other speakers, but it may also (and simultaneously) represent a FPP of other types of adjacency pairs such as greeting-greeting. An example of an NSI featuring in a response in a question-answer adjacency pair is 4.13 below. Here, the NSI contained in B’s turn not only modifies their own utterance (epistemically mitigating their answer to A’s question), but also responds to the presupposition contained in this question, namely that there are “legal experts” around in that forum: 4.13. A: Pretty grim stuff. Any of the legal experts know if he’s likely to be facing a custodial sentence? - B: I’m not a legal expert, but from that I would suggest that the answer to your question is ‘yes’. 4.2 Negative self-identifiers as context-dependent meaning potentials 93 <?page no="94"?> This illustrates the NSI’s role as a presupposition-challenging preface, marking both disalignment from the question’s premise and mitigation of the forthcom‐ ing answer. It should also be noted that adjacency pairs in online conversations may need to be approached differently than the ones occurring in offline conversations in that FPPs and SPPs of adjacency pairs may be further apart, separated, for instance, by other speakers’ postings. A case in point is 4.14 below, where the posting by speaker C, containing the NSI and serving as the answer to A’s initial posting, is preceded by a contribution by another speaker (B), who does not answer A’s question (but critically comments on it instead). 4.14. A: [SUBJECT] Account Hacked---AGAIN I have used a professional to virus scan my computer, have updated all the security software on my computer, now use the 2 step verification sign in process. […] However, within a couple of hours the listings were made ‘active’ again, which suggests to me that the 2 - step verification process is not working, that security on my computer is not the issue and the hackers are able to access my account from within Amazon. If anyone is able to shed any light, I would be hugely grateful. I have had a good run with Amazon, but if it takes another month to resolve this I am going to have to review selling my products on Amazon. [NAME] - B: Sorry, but that is utter rubbish. Anti virus programs are effective. If they weren’t, why would anyone bother using them in the first place. […] - C: Hi,It is listed in other Threads that Amazon Buyer accounts have been used to send unsolicited emails to Sellers (Turbo your listings etc), as well as Bogus Security alerts.The source of this intrusion is not necessarily your PC, but could be from a bogus buyer message. I’m no expert (so will bow to superior knowledge), but wouldn’t this bypass AVG etc, clicking on this “amazon” message? Just a thought….. This illustrates the NSI’s role as a hedged expertise disclaimer, allowing the speaker to contribute an answer while simultaneously downplaying their epistemic authority. The possibility of adjacency pairs being further apart in online conversations than in face-to-face interactions is relevant for this study because NSIs may occur in SPPs responding to relatively remote FPPs. These FPPs needed to be recovered and annotated as such in the data collection and annotation phase. Summing up, this study adopts analytical concepts from CA as it examines NSIs used in written, informal interaction on web forums, which means that the sequentiality of talk - that is, the co-text in the form of contributions by speakers other than the one using an NSI - is relevant for their interpretation. This means drawing on concepts such as turn-taking and adjacency pairs, which, as 94 4 Functionally Conceptualizing Negative Self-Identification <?page no="95"?> has been argued, need to be conceived of slightly differently when studying online conversations. While my project does take the impact of the affordances of web forums into account in the way data were collected and analyzed, its main focus is to explicate the use of NSIs in the context of these discussions and not to conduct a fully-fledged CA of the data (in this respect, my study can be considered to take into account less of the situational conversational context than a prototypical CA project). At the same time, analyzing what NSIs are used for and in which co-texts across written online conversations, my study also seeks to identify patterns of using NSIs. Its findings are interpreted with a view to the wider cultural context, which might not be considered in the same way by CA projects. 4.2.4 Context in systemic functional linguistics The SFL approach to language in use is one in which context plays a central role at all levels of theorization, and language is seen as having evolved according to the functions it serves in the social world (Halliday 1978: 4; Halliday & Matthiessen 2014, first mentioned in Chapter 2). What distinguishes the SFL perspective on context from that of interactionally oriented sociolinguistics (4.2.2), which, is also centrally concerned with context, is that it sees itself as an integrated theory of language (Bartlett 2017: 375); SFL constitutes what Hasan refers to as an exotropic theory of language, because it “embed[s] its object of study in a context where the processes of its evolution, stability and change can be seen to originate in the interaction of the object of study with other universes of human experience” (2005: 155). It considers language as systemic potential used to realize particular functions in context, an idea also put forth by Vygotsky (1978). Context in SFL plays a role at all levels (or strata) of theorization, from the social context in which language is used (SFL is socially oriented and historically informed by material conditions of language use) to meanings expressed by lexicogrammatical choices in the context of particular communicative situations. The idea is that, depending on the metafunction language serves in a particular situation, distinct lexicogrammatical choices with contrasting meanings are made to realize this function. For SFL, “[w]hat humans do with language in our social lives, and how language is itself organized, are two sides of the same social-semiotic coin” (Lukin 2016: 143). Assuming every linguistic choice to serve a function in context and thus seeing language use as inextricably linked to the material conditions of language users, SFL constitutes the theoretical foundation of critical approaches to linguistic analysis (see Section 4.3 below). 4.2 Negative self-identifiers as context-dependent meaning potentials 95 <?page no="96"?> My study of NSIs draws on the concepts of transitivity, mood and information structure to describe the interaction of NSIs with their context. Referring to the ideational metafunctions as one of three metafunctions of language distinguished in SFL, it is amongst other things of relevance whether NSIs occurring in complex clauses are textually related to material, mental, verbal or relational processes: for instance, if a disclaimer of expertise such as I am not a doctor is coordinated with the clause but I think this is eczema, representing a mental process by which the speaker provides a diagnosis, then the NSI could be interpreted as simultaneously disclaiming expertise and providing a diagnosis. The interpersonal metafunction, too, matters for the analysis. Imagine, for instance, the same disclaimer was coordinated with the clause but you should definitely see one - then this coordinated clause functions interpersonally as advice by virtue of having the addressee as its theme and containing the deontic modal should. Regarding information structure, it makes a difference whether an NSI occurs as an independent clause, making the speaker the theme of the utterance, or as part of a subordinate clause, in which case it has a different informational status. The three metafunctions of language proposed by SFL are seen as associated with the context of situation, which is both defined by language, and conditions it. For Hasan (1995: 219) context of situation in SFL comprises all those aspects of the non-linguistic surroundings of a particular speech event that are made relevant by language use. More precisely, it is assumed that a situation can be conceived of in terms of three aspects, viz. field (what is being talked about), tenor (the relationships negotiated by the interlocutors) and mode (what role language plays in the interaction and whether it is spoken or written). Each of these three aspects, in turn, is related to one of the three metafunctions of language (Halliday & Hasan 1989: 12; Bartlett 2017: 381). Tenor, field and mode can configure to form situation types, typical in a society, such as lectures or casual chat among friends. Situation types have an impact on register. For instance, a medical consultation might be characterized as 1) a highly specialized field, 2) involving experts talking to laypeople and, thus, a power difference, social distance and formality, and 3) a spoken, (largely) unprepared face-to-face interaction. The notion of register as a particular configuration of language functions linked to particular situation types is inseparably linked to the concept of genre, which can be defined as a text type defined by its purpose and associated with a particular structural organization and register (Bhatia 2002; Derewianka 2016). While the present study does not primarily aim to examine UK web forum discussions as a genre, but rather explores the functions of only one particular formally defined speech act as used in the context of these forums, genre-based approaches to linguistic analysis are relevant when accounting for NSIs inter‐ 96 4 Functionally Conceptualizing Negative Self-Identification <?page no="97"?> acting not only with their immediate textual surroundings, but framing, or preor postmodifying the interpretation of textual components above sentence level. One useful concept to explain how NSIs may fulfill functions with respect to the macro-organization of a text is that of moves. A move is a discourse segment serving a communicative function, that is, a semantically coherent stretch of text that can be related to the writer’s purpose (Swales 1990). It can be theorized as having a particular macro-theme (e.g., most explicitly the heading of a posting), providing the point of departure for a larger stretch of text, as well as hyper-themes, that is, the “packaging of information within phases of a text” (Forey & Sampson 2017: 134). As shown by the following example from my corpus, where the NSI interacts with the entire story preceding it, it is necessary to consider the macro-structure of postings to web forums containing NSIs to explain how they preor postmodify discourse segments, rather than just individual clauses: 4.15. My neighbour has a poxy little Dog… It appears she has invited a friend to stay tonight with a similarly poxy little Dog, only knowing her own poxy Dog doesn’t mix well with other Dogs they’ve had the bright idea of shutting the similarly poxy Dog upstairs in the room backing onto our bedroom while they go and drink themselves (even more) stupid.The poor f ’ucker has been barking all evening and now all I can hear is it whimpering and crying in between it’s [its] attempts to bark despite essentially barking itself horse [hoarse]. - I am no Dog lover but the whole thing has royally pissed me off. The NSI I am no Dog lover in example 4.15 follows a lengthy account of a situation whose theme - the speaker’s neighbor’s dog - is introduced by the heading (or macro-theme). By negatively identifying as a dog lover, the speaker postmodifies the interpretation of the entire story, thereby expressing criticism of the dog owners (whose “bright idea” is ironic, just like the fact that they drink themselves even more stupid) and sympathy for the neglected dog (the “poor f ’ucker”). This act of negative self-identification also underlines the severity of the case, suggesting that it constitutes an exception to the normal state of affairs. Systemic Functional Linguistics not only theorizes functional components above sentence level, but also provides the theoretical basis for discussing local instances of language in use on a ‘supra-situational’ level, that is, in relation to social values on a wider scale. Defining the context of situation as the “environment in which meanings are being exchanged” (Halliday & Hasan 1989: 12), SFL sees language use in particular situations as “specific instance[s] of a larger system, the context of culture, which comprises the set of systemically contrastive behaviors possible within that culture” (Bartlett 2017: 4.2 Negative self-identifiers as context-dependent meaning potentials 97 <?page no="98"?> 381). Thus, if particular forms (here, NSIs) are routinely used in particular textual environments to fulfill certain pragmatic functions, this could indicate speakers’ orientation toward orders of indexicality, which, defined as “systematically reproduced, stratified meanings” (Blommaert 2005: 73), influence what can be said in particular situations and how (a theme I return to in Section 4.3). To find how particular NSIs interact with their co-texts across forum discussions, this study adopts a corpus-based perspective to studying the structure in focus. The next section will therefore be concerned with corpus linguistics and, more specifically, corpus pragmatics. Summarizing key concepts that explain the relationship between text and context in SFL, this theory views language as intrinsically functional and as both construing and being conditioned by context. Context, from this perspective, is considered on the level of the social system, the level of the communicative situation (differentiation of situation types associated with context-dependent configurations of particular language functions), and the level of language in use (serving three metafunctions realized by the lexicogrammatical choices availa‐ ble to the speaker). Integrating the dimensions of language system, language function and social context, SFL is an important theoretical and methodological basis for the project presented here. The next section reviews corpus pragmatics as an approach that provides a framework for studying linguistic forms and their functions across texts, both in relation to specific features selected for analysis and in connection with the situational and wider cultural context. 4.2.5 Context in corpus pragmatics In corpus pragmatics (and, indeed, in any corpus-based approach) context has a special methodological and theoretical status. Unlike approaches seeking to explicate the relations between (real or invented) linguistic forms and (actual or theoretically assumed) contexts, corpus-based approaches necessarily predefine aspects of the textual and non-co-text within which the language they examine is embedded: thus, while traditional pragmatic approaches explore the relation between the formal and semantic properties of an utterance and the social situation in which it is appropriate (which implies an ex ante perspective on language, focusing on language competence), corpus-based approaches examine situated instances of language use as (presumably appropriately) used in particular contexts, thereby seeking to find out what they have in common (which implies an ex post perspective on language use, focusing on performance as finite realizations of infinite possibilities of expression and the rules that are established as speakers communicate effectively). In this respect, 98 4 Functionally Conceptualizing Negative Self-Identification <?page no="99"?> corpus-based approaches can be considered an inversion of the generative grammatical approach to linguistic inquiry (Spitzmüller & Warnke 2011: 29). Corpus-based approaches study language use transtextually and quantitatively by examining instances of particular linguistic elements or structures across texts. This means that while corpus analysis, by definition, studies contextually situated language use, at the same time it always involves decontextualization by taking a ‘vertical’ or paradigmatic perspective on language use. This perspective stands in sharp contrast to the ‘horizontal’, syntagmatic reading traditionally employed in pragmatics, where “texts are received and interpreted in the same temporal order in which they were produced” (Rühlemann & Aijmer 2014: 3) - a distinction I return to in Chapters 5-9, where close analysis of horizontal sequences is set alongside vertical patterns across the corpus. Due to this difference in perspective, corpus linguistics and pragmatics were long regarded “as parallel but often mutually exclusive” (Romero-Trillo 2008: 2). However, since “the impact of corpora has been such that observers speak of a ‘corpus revolution’” (Rühlemann & Aijmer 2014: 4), there has been an increasing number of pragmatic studies employing corpus linguistic methods over the last two decades, with the result that corpus pragmatics is now a firmly established approach within pragmatics. Corpus pragmatics is particularly fruitful when applied to studying the functions of linguistic forms with context-depending meaning potentials, as a “corpus-driven, bottom-up” approach allows “the discovery of more or less different functions […] depending on the linguistic and discourse context” (Aijmer 2013: 29). Corpus-pragmatic studies focus on “either a discourse particle with a fixed form that can easily be retrieved from a large corpus, or a speech function that is generally realized in a small number of variant patterns” (Jucker et al. 2009: 4). My study, too, has a relatively fixed linguistic structure as its starting point and combines qualitative analysis (closely examining the co-text in which the various conceptually differentiated NSIs occur), with quantification (counting the instances of identifying nouns pertaining to a particular conceptual category and cross-categorizing them with particular functions realized by elements of the co-text, cf. also Marko 2015b). It differs, however, in that it focuses on a constructional pattern that is neither limited to a single discourse particle nor reducible to a small set of fixed variants. Here too, it is important to stress that the analytical unit is not the string “I’m not a …” abstractly conceived, but each NSI construction token in relation to its local co-text and discourse function. In this sense, NSIs can also function as pre-emptive addressivity moves, anticipating and managing an imagined heterogeneous audience. Exploring meaning-making in online interaction by taking a corpus-based approach, according to Knight (2015: 4.2 Negative self-identifiers as context-dependent meaning potentials 99 <?page no="100"?> 20), is “ideally situated to contribute to the investigation of digital discourse” as it can reveal “patterns of language use in large-scale bodies of” text. Relying predominantly on forms to explicate function and selecting only a few elements of the co-text of a particular linguistic structure to be examined across texts, instead of looking at the full local co-text of usage, is not entirely unproblematic. According to Archer and Culpeper (2018: 496), for example, approaching instances of language in use this way usually means not paying enough attention to “the (situational, social and cultural) dynamics of context, particularly at the local micro level”. This criticism is, at least to a certain extent, also justified with respect to the present study. The reason why a form-based, corpus-pragmatic approach is still chosen - at the expense of fine-grained analysis of the local discourse contexts of the structure in focus - is the present study’s interest in potentially patterned relations between particular types of NSIs and particular types of discourse context. To close the gap between fine-grained corpus analysis of NSIs in their immediate co-texts and the interpretation of these functional patterns in light of macro-sociopolitical issues, Section 4.2.6 below reviews literature that accounts for the specificities of online contexts influenced and designed by speakers’ linguistic choices in digital interactional settings such as web forums. As outlined in Chapter 5, the methodological design also allows for returning to selected stretches of co-text to contextualize corpus-based findings - an approach taken up in the analyses in Chapters 7-9. 4.2.6 Context in studies of digitally mediated discourse Within studies of digital discourse (reviewed in Chapter 2), the unprecedented specificities of online contexts have received much attention. A key idea about social networking sites, such as Facebook, is that they “collapse diverse social contexts into one” (Marwick & boyd 2011: 10) by merging potentially very different social networks in one single virtual realm. In such collapsed online contexts - bringing together participants from (super-)diverse social backgrounds - users “imagine and respond to a particularly complex set of contextual variables as they design their posts and interactions” (Tagg et al. 2017: 20). Not only are the audiences of what is posted online widely unknown to the speaker, hard to delimit and potentially growing and changing, they are also likely to be extremely diverse. This has important implications for online self-representation, affecting, 100 4 Functionally Conceptualizing Negative Self-Identification <?page no="101"?> 17 It should be mentioned, though, that web forums, representing older forms of online social platforms, differ from social networking sites usually discussed with reference to context collapse. Unlike social networks such as Facebook, which are “purpose-built to facilitate social interaction” (Seargeant & Tagg 2014: 2) among people who have already known each other in the offline world, forums “tend to bring together users who do not share a previous offline connection” (Androutsopoulos 2014: 63). Also, the communities hosted and represented by forums have been considered as a somewhat special space in the vast virtual realm because of their high potential for enabling more permanent and intimate online communities to build (Ahuja & Galvin 2003; Andresen 2009; Bateman et al. 2011). 18 The bold initial letters of the aspects this framework jointly forms the mnemonic POSTING, which alludes to Hymes’ (1974) SPEAKING model on which it is based and from which it departs by studying written texts - that is, mainly postings - in online contexts. for example, what people decide to share about themselves and how they design these postings with a view to their imagined audience. 17 As Tagg et al. (2017: 25) point out, context not only shapes and can be indexed by a text, but is construed and manipulated as people, aware of the contexts within which they interact, use language to serve particular social functions. Indeed, in line with Lemke’s view quoted at the beginning of this chapter, according to which “the meaning of text is made through time” (Lemke 2000), Tagg et al. (2017: 27) argue that “it is not necessarily always possible or desirable to distinguish the text (language and its co-text) from its context”. Consequently, their theory of context design provides an analytical framework to account for how speakers engaging in online conversations reflexively stylize their utterances and, thus, discursively construe their identities, in a complex process of “active construction and negotiation of context as part of the communicative exchange” (ibid.: 32). The linguistic strategies employed to manage the anticipated response of what is thought to potentially represent the audience of content shared online have been referred to as “addressivity strategies” by Seargeant et al. (2012). This concept is relevant for the study of negative self-identification because speakers’ use of the structure might reflect their considerations about the potential effects of their utterances on other forum participants (or even wider audiences on the web). In other words, negative self-identification serves not only to make a negative assertion about one’s identity in order to eschew alignment with particular categories and thus project a particular self-identity into an already existing communicative situation, but also to construct a context within which speakers’ linguistic performance is expected to be most effective (Androutsopoulos 2014: 64). In Table 4.1 below, I have summarized the aspects of the original framework (Tagg et al. 2017: 37-38) 18 to become operationalized in the study of NSIs 4.2 Negative self-identifiers as context-dependent meaning potentials 101 <?page no="102"?> used in web forums. The contextual feature “identification” apparently has a particularly important status in the present study of NSIs as linguistic micro-realization of negating identification. By explicitly contrasting speakers with “particular ideologies, discourses and individuals […] as well as ascribed social roles and particular concepts” (ibid.), their use can be seen as being shaped by and simultaneously construing the interactional context in which particular identity ascriptions are linguistically made relevant. They can index, for example, how forum users perceive the relationships with other members of the site and the common ground they share (or do not share). The use of NSIs can reflect “people’s awareness of the competing norms and sources of authority” (ibid.: 41) that matter in interactional situations on and across forums, and point to conceptualizations of belonging and not belonging to particular social groups which “cut across traditionally perceived offline (and online) contexts” (ibid.: 35). These cross-cutting conceptualizations, in turn, can be discussed in relation to the wider sociopolitical context. Negative self-identification with particular social groups in forum discussions may also indicate speakers’ awareness of the medium’s affordances. For example, NSIs can reflect a heightened need for explicit disalignment from certain groups, given speakers’ awareness that their postings might be extextualized (see Section 2.2.1.2), read, and potentially misinterpreted by different, unpredictable audiences (ibid.: 40). Features of Online Contexts Participants Forum users’ knowledge of their audi‐ ence, that is, their familiarity and com‐ municative history (Tagg et al. 2017: 40) with other people interacting on the same forum and their contributions in the im‐ mediate communicative situation Online media ideologies Forum users’ conceptions about the pur‐ pose of the forum Site affordances Forum users’ conceptions about how fo‐ rum interaction works in terms of techno‐ logical affordances Text type (or mode) of the communication Web forums as quasi-asynchronous, typed interaction between physically distant in‐ terlocutors Identification process • Using NSIs as explicit disalignment with particular concepts • Using NSIs as stance framing for ad‐ vice, disclaimers, or evaluation 102 4 Functionally Conceptualizing Negative Self-Identification <?page no="103"?> Norms of communication Various global norms of communication as well as more local or even site-specific conceptualizations about what constitutes appropriate communicative behavior on (particular) forums Goals The function(s) of the particular posting, possibly indexed through contextualiza‐ tion cues Table 4.1: Features of online contexts (adapted from Tagg et al. 2017: 37-38), with additional elaboration under “Identification process” to highlight the relevance of NSIs. Due to the quantitative orientation of the present study, the level of the situational contexts in which the structure is used by particular speakers in particular online contexts is not addressed in depth in this study - it primarily focuses on investigating the (potentially patterned) micro-pragmatic functions of instances of NSIs in their immediate co-texts across interactional situations on online forums. In this way, the study seeks to employ “empirical, micro-level methods to shed light on macro-level phenomena”, as called for by Herring (2004b: 338). However, to illustrate how NSIs can be analyzed as interacting with online contexts, a qualitative analysis of two instances of the structure in use is provided in Chapter 9 to bring together the micro-pragmatic level of analysis and the macro-level of the wider sociopolitical context within which situated linguistic choices are embedded. This concern with linking micro-level interac‐ tion and macro-level context also underpins the critical-discursive perspective outlined in Section 4.3. 4.3 Negative self-identifiers and (corpus-based) critical discourse studies Having reviewed approaches that theorize how context shapes and is shaped by language use, I now turn to critical discourse studies (CDS). While CDS integrates these concerns into a broader project of linking micro-level analysis to macro-level sociopolitical critique, my discussion begins from the structure itself, treating NSIs as linguistic forms that fulfill specific functions in online forum discussions - understood as particular types of informal conversation - and thus as pragmatic phenomena that occur in particular discourse contexts. This means that my starting point has been the structure as well as the question of how its interaction with elements of the textual and non-textual co-text can be theorized. 4.3 Negative self-identifiers and (corpus-based) critical discourse studies 103 <?page no="104"?> In this section, I shift the perspective from bottom-up to top-down, outlining how my project can be situated within CDS an umbrella term within which critical discourse analysis (CDA) remains a central strand (van Dijk 1993; Wodak & Meyer 2016). While my project can be framed as a critical discourse analytical endeavor, I suggest that this book departs from prototypical examples of CDS as traditionally conceived. Critical discourse studies is interested in the language/ knowledge side of social phenomena and therefore pursues the primary objective of “find[ing] beliefs and attitudes constructed in and through texts” (Marko 2015a: 69). The basic assumption in CDS is that discourse, in the sense of language in use, mediates meanings, that is, certain conceptualizations of the world, which influence our beliefs, values and attitudes and, consequently, the way we act in the world. Thus, these conceptualizations are instrumental in “establishing, maintaining and changing social relations of power, domination and exploita‐ tion” (Fairclough 2003: 9). As for its disciplinary status, CDS would probably be considered most closely associated with linguistics by most scholars (Marko 2015a: 119): Drawing heavily on SFL as an exotropic - that is, socially oriented - theory of language (Bartlett 2017), it views language as fulfilling functions in social practices and links language use to the conditions of those who use it. However, projects in CDS are not typically united by adherence to a specific theory of language. Instead, most representatives view CDS as an interdisciplinary endeavor defined by its critical and emancipatory potential and “therefore not interested in investigating a linguistic unit per se but in studying social phenomena which are necessarily complex and thus require a multidisciplinary and multi-methodical approach” (Wodak & Meyer 2016: 2). Fairclough (2003: 6) likewise stresses the importance of transdisciplinary dialogue between disciplines with different perspectives on language. In brief, as Cameron (2001: 123) explains, in CDS, “the purpose of analyzing discourse in its first sense (language in use) also functions as discourse in its second sense (a form of social practice that ‘constructs the objects of which it purports to speak’)”. This study draws on Marko’s (2015a: 157) theorization of discourse as con‐ ceivable on three levels of abstraction: as socially conditioned and constitutive language in use as opposed to language as an abstract system, as concrete texts (also referred to as textual events, which are part of discourse), and as types of language use with concomitant bodies of knowledge and attitudes (or discourses as a count noun), representing what Reisigl and Wodak (2016: 27) call “a cluster of context-dependent semiotic practices that are situated within specific fields of social action”. These componential levels of discourse 104 4 Functionally Conceptualizing Negative Self-Identification <?page no="105"?> correspond to three analytical levels: firstly, linguistic analysis of particular linguistic forms, secondly, the interpretation of the meanings they realize in particular discourses, and, thirdly, the social effects of these meanings and their sociopolitical significance. It should be mentioned, though, that these levels are not entirely separate analytical steps in a linear research process. On the contrary, CDS adopt a hermeneutic research process, the first step of which is normally the identification of issues on the level of the sociopolitical context, which are translated into researchable questions determining the paradigms to be analyzed and the methods chosen to do so. The results of this analysis are interpreted and critically evaluated in terms of their implications for the sociopolitical context. The conceptualization of discourse in terms of ascending levels of abstraction and, correspondingly, analysis, is based on a key theoretical assumption adopted from SFL (discussed in Section 4.2.4), namely the multi-modular approach to meaning. As has been discussed, according to this approach, the linguistic choices speakers make in particular social situations both serve and depend on the three metafunctions, which means that formal choice is inherently functional, and function is inherently contextual, as it depends on social situations. The process through which language users arrive at the full meaning of a text that “coheres within itself and with the context of situation” (Halliday 2003: 17) involves two levels, namely meanings encoded by the linguistic forms used, and enhanced meanings, which language users arrive at by drawing on coand contextual cues and their background knowledge (Marko 2015a: 168). According to this conceptualization of discourse, NSIs are seen as structures encoding particular meanings depending on their formal appearance and the meaning of the identifying NP. To arrive at their full meaning, speakers draw on the surrounding text, their situationally relevant pragmatic knowledge (see Section 4.2.1) and their background knowledge, that is, the meaning of the negative identifier is enhanced when interpreting it in relation to the meanings of its textual and non-textual contexts of usage. It is generally possible to distinguish between a more syntagmatic and a more paradigmatic perspective on language in use (Marko 2015a: 194), which also has methodological implications: a syntagmatic perspective on language in use is marked by a pre-established interest in particular texts and a focus on their idiosyncratic properties, that is, meanings established by syntagmatic relations between particular forms and structures, as well as their conditions of production, distribution and reception. In contrast, paradigmatically oriented analyses center on meaning relations established by recurrence of particular meanings and meaning relations across texts. The language system and the 4.3 Negative self-identifiers and (corpus-based) critical discourse studies 105 <?page no="106"?> social system in which language is used are related (Halliday 1991); in fact, system and actualizations of the system in communicative processes can be considered to interact. According to de Beaugrande (2008: 43), “a ‘language’ is a potential system; a ‘text’ is an actual system” of possible and actual choices and combinations, and corpora, as samples of actualizations in communica‐ tive processes (‘discourses’), can reveal actualizations dynamically, exerting pressure on the system. Thus, for example, grammaticality is the result of grammaticalization and stems from, and can only be observed in language use. While the corpus examined for this study is, of course, way too small to be representative of how NSIs are used generally and systematically, my study is interested in actualizations of the potential meanings of variants of the structure in focus across UK web forum discussions. Accordingly, the unit of analysis is each token of the formally defined NSI pattern (“I + copula + NOT + indefinite NP”) situated in the UK forum context, with functions derived inductively from co-text and context rather than assumed a priori. Consequently, the present study shares theoretical and methodological principles with corpus-based CDS (Mautner & Koller 2004; Baker 2006, 2013, 2014; Mautner 2009; Marko 2015a). Corpus-based CDS are interested in the linguistic elements and structures realizing general and specific linguistic paradigms (defined by grammatical, pragmatic, textual and contextual features) in a corpus, as a materially ob‐ servable representation of a discourse (Bubenhofer 2008: 1). Regarding the paradigms chosen for analysis, most studies in corpus-based CDS identify a struggle on the level of the wider sociopolitical context, which is assumed to crystallize around particular linguistic items whose collocational profile can reveal higher-level patterns of meaning-making involved in the discourse under scrutiny (Hardt-Mautner 1995). Accordingly, these studies examine linguistic elements that feature prominently in the discourse of interest, with a particular focus on ideologically problematic representations. Mautner (2007: 51), for example, examines large corpora to establish a collocational profile of the word elderly, thereby providing “lexico-grammatical evidence of stereotypical constructions of age and aging”. Another possible reason for selecting particular lexemes for linguistic analysis is that they denote a concept whose representa‐ tion is of interest in light of a particular social research question. For example, Nardone (2018), seeking to shed light on contemporary German and Italian representations of working women, analyzed lexical collocates of concordances featuring the words whose use is assumed to manifest semantic struggle, viz. women, work, men etc., in large general corpora, whereas the paradigm my study investigates is formally and contextually defined as instances of “I + copula + NOT + indefinite NP” used on UK web forums. 106 4 Functionally Conceptualizing Negative Self-Identification <?page no="107"?> Instances of NSIs cannot, per se, be argued to be ideologically laden, as this depends on the meaning of the noun or NP speakers contrast themselves with, and even more so on the context of usage. Nevertheless, the structure was chosen based on the assumption that there is an ideological struggle around notions of belonging and not belonging, potentially manifesting itself in how speakers use the structure in focus. This means that an interest in the sociopolitical context of potentially changing conceptualizations of the self and self-identification plays a role for the research presented here from the start: it represents the impetus for researching negative self-identification in the first place, and the social context of language use is taken into consideration at all levels of the research - from contextualization of the research and data selection to methods chosen and interpretation of results in light of their potential implications for the broader sociopolitical context. In other words, the project presented here is critical in that it avoids a decontextualized or ahistorical perspective on the meanings of NSIs, and assumes that linguistic choices are inherently informed by the social context of language use - so that, from a functional perspective, ‘non-critical’ discourse analysis is arguably impossible. In this sense, my study not only describes, but also denaturalizes language use by critically questioning the dominance of particular communicative functions of NSIs, referring to theoretical concepts from a variety of disciplines. At the same time, while my study does not adopt an ex-ante interest in a particular social ‘wrong’, it shares with Fairclough (2003: 4) an interest in the language of late modern capitalist society. I cautiously interpret my results with a view to potential relations between what people contrast themselves with and the wider sociocultural context. To some extent, this includes how with reference to Bourdieu (1998), Fairclough (2003: 7) speaks of a “re-structuring and re-scaling of social relations in accordance with the demands of an unre‐ strained global capitalism” and of the reflection of “dominant character types of contemporary societies” in speakers’ self-conceptualization and linguistic self-expression. In line with CDS, my study is not only critical, but also self-critical in recognizing that, given the data examined and the analyses performed, these connections can only remain tentative. For example, while patterns of usage identified for NSIs could resonate with social theoretical observations about particular social phenomena or conceptualizations featuring prominently in contemporary self-representations, this study only considers a miniature aspect of speakers’ identity management in online discourse and can thus only reveal traces of discourses manifesting themselves in speakers’ use of the structure examined. 4.3 Negative self-identifiers and (corpus-based) critical discourse studies 107 <?page no="108"?> Negative self-identifiers, the object of interest in the study presented here, can be approached from two perspectives: bottom-up and pragmatically, to find what discourse functions are served by occurrences of the formally defined structure “I + copula + NOT + indefinite NP”, and top-down, with an interest in how people talk about themselves in the negative and how dominant functions of the structure relate to broader societal issues. The corpus - as a sample of discourse situations in which the structure was used - is defined formally, by the occurrence of the structure “I + copula + NOT + indefinite NP”, and by the type of text within which this structure appears, namely UK web forums. By identifying the meanings of variants of the structure and their linguistic contexts across texts, frequently realized pragmatic functions of NSIs can be determined and critically discussed as patterns of talking about self-identity in the negative with particular communicative goals. In this way, CDS provides the overarching framework within which my corpus-based analysis of NSIs is situated. 4.4 Summary The aim of this chapter was to position this study theoretically within ap‐ proaches that explore the functions of language in use, all of which are relevant to understanding NSIs. Since studying the function of a form in context requires defining what is meant by context and how it interacts with language use, context was chosen as the focal concept structuring this chapter. I began by discussing the notion of context in speech act theory, where it plays a theoretical role, setting conditions for the appropriateness of utterances. I then moved on to sociolinguistic approaches, which focus on the impact of social context on linguistic variation, and approaches interested in the pragmatics of interaction, which account for aspects of the situational context of language use. I also considered CA and corpus pragmatics, both relevant methodologically for this study, which analyses a corpus of written conversations and examines the sequential co-text of NSIs and their use across texts. In addition, SFL is important both theoretically and practically, as it views formal choice as contextual and links linguistic choices to the sociocultural context. Additionally, SFL’s transi‐ tivity framework, that will be used in Chapter 8, enables systematic analysis of meanings in the context of NSIs. I also reviewed theoretical frameworks that account for how linguistic choices are shaped by, and contribute to, digital interactional contexts. Finally, I argued that this project is situated within CDS, an interdisciplinary field concerned with the relationship between linguistic choices and broader conceptualizations of self-identity and the social world. Critical discourse studies aims to identify sites of social struggle by analyzing 108 4 Functionally Conceptualizing Negative Self-Identification <?page no="109"?> how linguistic forms contribute to the construction of meaning across local and transtextual levels within social processes, as represented in particular types of texts. I outlined key CDS concepts and how they are understood and applied in the context of this study. In Chapter 5, I explain how I compiled a corpus of forum discussions featuring NSIs, the considerations that informed this process, and how I prepared the data for analysis, including representativeness and annotation decisions that operationalize the context dimensions outlined here. 4.4 Summary 109 <?page no="111"?> 5 The Empirical Study: Data and Method This chapter presents the data for this study and the theoretical and meth‐ odological principles underlying data selection and sampling, annotation and analysis. After introducing these principles in Section 5.1, I explain according to which initial considerations variants of NSIs of the type “I + copula + NOT + identifying NP” were selected as the structure targeted in this analysis; on what grounds I decided to compile a derived corpus of instances of this structure taken from UK web forums; and how I went about the task of creating this dataset (Section 5.2). I then introduce the formal-functional framework providing the conceptual basis for studying NSIs in their co-texts. This framework puts into practice the theoretical considerations about interaction between linguistic form and its co-text of usage as the observable part of the communicative situation that can be empirically studied by corpus linguistic methods (Section 5.3). Finally, in Section 5.4, this formal-functional framework is translated into an annotation scheme that allows for qualitative and quantitative study of the relations between forms and meaning across texts. Together, Sections 5.3 and 5.4 show how the theoretical framework is operationalized through annotation, linking the methodological principles introduced here to the analyses in later chapters. 5.1 Theoretical and methodological principles My study is concerned with the meanings created by language users’ lexicog‐ rammatical choices in authentic discourse contexts and therefore relies on principled analysis of authentic data. I compiled a corpus of 936 NSIs, conceived as observable material traces of past social and cognitive communicative events. This approach is based on the assumption that functional patterns reflect underlying patterns of conceptualizing the world, as evidenced in specific links between semantically and formally defined classes of NSIs and their co-texts. This means that my study has an interest in “what is usual and typical” rather than in the unique, assuming that “frequency in the corpus is observable evidence of probability in the system” (Stubbs 2007: 130). The “system” here is to be understood as the socio-pragmatic system of using NSIs rather than the language system in general: the corpus compiled for this study is, thus, not statistically representative in any probabilistic sense, but constitutes a purposive sample, compiled on the basis of a specific formal construction to facilitate <?page no="112"?> 19 Inferential quantitative methods explicate causal and linear relationships between two or more variables and calculate the probability that these relationships can be extrapolated from the examined sample (i.e., the corpus) to the whole population. They can, for example, be useful to analyze collocations by determining “the above-chance frequent co-occurrence of two words within a pre-determined span” (Baker et al. 2008: 278). focused analysis of a constrained set of lexicogrammatical realizations. Conse‐ quently, the findings presented here are not intended to support generalizations about the full range of negative self-identification practices across discourse domains or populations. Instead, they are best understood as pattern-based insights into recurrent pragmatic and semantic tendencies within a corpus constructed to foreground a particular discursive strategy. The approach to quantification taken in this study is descriptive, which means that frequencies and relations between frequencies of particular linguistic ele‐ ments occurring in the corpus are counted, providing percentages and rankings. Inferential quantitative methods are not used in this study, although they have proven relevant for (critical) corpus-based discourse analysis. 19 Quantification proceeds in stages of qualitative differentiation among linguistic elements pertaining to a particular formally defined paradigm and quantification, that is, counting the frequencies of items assigned to particular categories and comparing those categories. Of course, there is constant alternation between these stages of analysis, as quantification leads to the identification of salient categories of linguistic elements, which can then be further qualitatively differentiated (Marko 2015a: 205). Like all linguistic research with an interest in social research questions, the research conducted here is staged, involving the stage of contextualization, that is, a discussion of the social or societal relevance of a particular discourse phenomenon to be researched, and the stage of operationalization, where the social research interests are translated into linguistically researchable questions and the analysis is carried out. As already mentioned when discussing CDS in Section 4.3, this project is conceptualized as an iterative, hermeneutic process in which meanings are both understood and produced: “[T]he meaning of one part can only be understood in the context of the whole, but this in turn is only accessible from its component parts” (Titscher et al. 2000: 240). Therefore, in the tradition of Grounded Theory (Glaser & Strauss 1967), “data collection is not a phase that must be finished before the analysis starts but might be a permanently ongoing procedure” (Meyer 2001: 18). Regarding the question of quality control, with a phenomenon as fuzzy as the relation between linguistic forms and meanings, it is not possible to 112 5 The Empirical Study: Data and Method <?page no="113"?> apply the exact same criteria that would be used in other disciplines. Studying the role of language in social processes is partly reliant on the researcher’s interpretations - for example, creating a conceptual profile of a particular discourse involves categorization of linguistic elements by the researcher, whose ways of meaning-making necessarily influence the categorization process. Thus, as Meyer (2001: 29) explains, “the classical concepts of validity and reliability cannot be applied without modification”. According to Agar et al. (2007: 378), “many qualitative researchers have wanted to dismiss these as merely ‘positivist’ concerns”, which is why they cite other concepts better suited to ensure sound qualitative research, viz. the completeness of descriptions, the saturation of categories, the authenticity of the researcher, the consistency of categorization, the credibility of the study’s findings in the sense that the researcher’s interpretations can reasonably be related to reality, and plausibility of the study’s findings in the light of scientifically accepted theories. To ensure credibility and plausibility of the findings of the analysis, the categorization of the data was principled, that is, not merely relying on inter‐ pretation of meanings, but on formal and semantic aspects theorized before data analysis. Based on the principle of hermeneutic research, items of data were not just assigned to a set of categories established before analysis, but the quality of categories was iteratively tested in the process of analyzing more data, potentially leading to modification of the criteria for category inclusion. What is more, the formally defined linguistic paradigms selected for analysis were exhaustively examined and vertical categorization with the goal of quantification was always supplemented by qualitative, horizontal analysis of representatives of the identified categories. In terms of triangulation, that is, approaching the same research question by using multiple datasets, different investigators, theories or methods (Denzin 1970/ 2017), no comparative triangulation of the findings of this study was con‐ ducted. An exception is represented by the findings of the conceptual profiling presented in Chapter 7, which were compared with data from the Spoken BNC2014. To limit the bias of one researcher investigating a phenomenon from only one perspective or with just one method or dataset (Baker & Levon 2015: 223), this study examines several linguistic paradigms (from identifying NPs to entire sentences and larger stretches of text preceding NSIs) both qualitatively and quantitatively, drawing on a variety of theoretical concepts (outlined in Chapter 3). This means that the findings of this study are not based on only a single perspective, since the various parts of the study produce a more complete picture. Another measure I took to triangulate some of my findings was to employ the semantic tagger Wmatrix 4.0 (Rayson 2008) to automatically perform 5.1 Theoretical and methodological principles 113 <?page no="114"?> semantic analyses of my data, thus supplementing my interpretation as a single human researcher by that of an algorithm. The online survey conducted prior to corpus collection described in Section 3.4.1 also served to look at negative self-identification from more than just one perspective in the initial stages of the project. 5.2 Data selection and collection As King (2009: 299) explains, “the usefulness of any given corpus, whether ‘general-use’ or ‘project-based’, ‘raw’ or ‘annotated’ (Beißwenger & Storrer 2008), hinges upon sound corpus construction”. Lew (2009: 290) discusses this “soundness” with reference to several key parameters: corpus size (i.e., the number of word tokens), linguistic representativeness (i.e., of a specified population), balancing (i.e., avoiding overor under-representation of particular text types), noisiness (i.e., the presence of non-standard spelling and grammar in authentic discourse), and the functionality and accessibility of the corpus for analysis. Orienting toward and critically reflecting on these parameters, this section outlines how I selected and compiled a dataset for analyzing the use of my focus structure. As discussed in Chapter 3, the decision to concentrate on variants of the structure “I + copula + NOT + identifying NP” was based on specific theoreti‐ cal-linguistic considerations. The pilot survey presented there further supported this decision by offering insights into how speakers use language for (negative) self-identification. As previously argued, web forums constitute ideal data sites for examining how people use NSIs in informal discussions of shared interest, for both theoretical and practical reasons. In what follows, I explain why I chose to construct a custom corpus of web forum discussions rather than rely on an existing corpus (5.2.1); why I used customized Google searches as my primary data collection tool (5.2.2); the principles and procedures I applied to compile a corpus of formally defined variants of the structure “I + copula + NOT + identifying NP” in their proximate co-texts across thematically varied UK-based, English-language web forums (5.2.3); and how I ensured compliance with ethical standards (5.2.4). Throughout, I also critically recontextualize my methodological approach and dataset in light of more recent developments in corpus construction and analysis. 114 5 The Empirical Study: Data and Method <?page no="115"?> 5.2.1 Creating a derived corpus of negative self-identifiers The decision to compile a corpus myself was taken after scrutinizing existing corpora of CMC regarding their suitability for the research goals pursued here. The advantages of using an existing corpus are that the texts have already been collected, that ethical questions have already been addressed and that larger corpora, habitually designed as parts of bigger projects within linguistics, are usually controlled for contextual features (e.g., year and place of publication of text samples, gender of speakers etc.) and often annotated grammatically and/ or semantically. The corpora considered as candidates for this project in the initial stages of this research were the ukWac corpus (a 2-billion-word corpus constructed from the Web, limited to the .uk domain and using medium-frequency words from the BNC as seeds), the WaCkypedia corpus (an 800-million-token corpus of Wikipedia text), the Westbury Lab Usenet Corpus (a 7-billion-word corpus taken of English Usenet postings), the Westbury Lab Wikipedia Corpus (containing just under one billion words from Wikipedia) and the Reddit Comments Corpus (made up of 1.1 billion comments on Reddit collected between 2007 and 2011). However, none of these corpora proved preferable to a self-compiled corpus of NSIs in their context: the ukWac corpus, while being POS-tagged and lemmatized, is made up of general web language and thus just as unspecific regarding the type of web discourse as the Internet as a whole. While POS-tag‐ ging and lemmatization are generally useful for conducting corpus-based research with concordance software, they are of no particular relevance for the present research. To answer my research questions, I mainly focused on units above word level (e.g., sentences and phrases preceding NSIs) or even above sentence-level and thus needed to segment the corpus myself in specific ways. Also, due to the informality of the data examined here, the probability of faulty categorization by a tagger was considered too high. The WaCkypedia corpus was ruled out because Wikipedia entries are not normally interactive and spontaneous discourse, but monologic, planned text with the purpose of providing information - the same applies to the Westbury Lab Wikipedia corpus. While the Westbury Lab Usenet Corpus would have been potentially suitable because it consists of postings, I found that searching seven billion words of Usenet postings for occurrences of the structure “I + copula + NOT + indefinite NP” offered no clear advantage over customized Google searches to detect the structure in web forums, especially given the fact that the importance of Usenet has diminished with respect to Internet forums today. Finally, the Reddit Comments corpus, while consisting of comments and thus theoretically constituting the text type of interest because of its interactive character, is 5.2 Data selection and collection 115 <?page no="116"?> extremely large, which means that once again, searching it for instances of negative identification would not have yielded any advantages over a targeted web search approach using Google. From today’s perspective, existing corpora of CMC would indeed present viable alternatives to a self-compiled corpus. Notably, the English Web Corpus (enTenTen), in its most recent version comprising 52 billion words of online discourse in English and annotated for genre and topic, allows researchers to limit searches to texts from .uk domains. Searching only for the variant I’m not a under these criteria yields an impressive 10,814 hits (see Kilgarriff et al., 2014, for details of the Sketch Engine architecture). Analyzing this data through manual corpus annotation and categorization of co-texts would be unfeasible at such scale. The decision to “go fully large-scale” thus reflects a more general challenge - and opportunity - for CDS scholars: the unprecedented extent to which discourse data is digitally produced, globally disseminated, and increasingly available for research purposes (Aranda et al. 2021). To harness this potential, approaches such as Structural Topic Modelling (STM) (DiMaggio et al. 2013; Schmiedel et al. 2019) have been fruitfully employed to inductively identify lexical elements that occur in connection with specific topics at a non-random rate, and to discover lexical patterns and meanings characterizing a discourse without imposing interpretative frames a priori. Topic modelling has therefore come to be seen as a suitable method for critical studies of large language corpora (Brookes & McEnery 2019). Yet to generate probabilistic groupings of words that are interpreted as topics, a level of abstraction is required that disregards micro-level linguistic features such as grammatical structure, speaker stance, and pragmatic function. The present study, rather than seeking to reveal large-scale manifestations of discursive struggles through topic modelling, as in Aranda et al. (2021), focuses instead on the forms and functions of a formally defined construction - its syntactic environment and pragmatic effects within interactional context. To learn about these interactional and structural properties, “carefully inter‐ weaving horizontal and vertical methods of analysis” (Haugh 2018: 619) is required, which necessarily implies a reduction of the most impressive datasets for the sake of analytic feasibility. Therefore, this study employed customized web searches to compile a project-specific corpus. One clear limitation of compiling a web-derived corpus is the lack of control over the time frame of the data. From today’s perspective, this presents a meth‐ odological weakness: the absence of temporal metadata hinders the possibility of diachronic analysis, limits contextual transparency, and complicates reproduci‐ bility - especially in light of recent changes to search engine algorithms and the 116 5 The Empirical Study: Data and Method <?page no="117"?> decreasing transparency of platforms like Google. At the time of data collection, however, restricting the dataset to a clearly defined period would have been prohibitively time-consuming and carried the risk of yielding too few instances of certain structural variants. Accepting this disadvantage, the data collection period was set from July to September 2019, and the only constraint was that postings needed to be published after 2015 to secure a sufficient and varied sample of relevant data. As noted above, customized Google searches provided access to relatively recent postings across a wide range of forums. This breadth of source material proved advantageous for the aims of the study, which seeks to examine the pragmatic and syntactic properties of a specific construction across thematically diverse contexts, rather than within a particular discourse domain. 5.2.2 Creating a corpus with Google searches In terms of accessing the textual resources, there was a choice between using conventional search engines such as Google and those such as WebCorp that are designed specifically for linguistic research. For this study, customized Google searches were used, because exploratory searches with WebCorp suggested that linguistic search engines, as Lew (2009: 297) has argued, “do not offer dramatic improvements over the basic search engine functionality, but they do have one significant disadvantage: inferior speed”. The option to limit Google searches to websites with certain abbreviations in their URLs allowed me to automatically control the results of my search for two contextual features, namely the kind of website (forums, marked as such by having the word forum or thread in their web address) and the website’s geographical location (UK websites, indicated as .uk in the URL). Another limitation that deserves closer scrutiny, particularly given recent, highly controversial updates to Google’s search engine algorithm (Germain 2024a) and the growing awareness of the dangers of algorithmic search bias (Canca 2019, 2020; Germain 2024b; Tong 2025) is the dependence on commercial search engines for identifying and retrieving linguistic data. Search engine algorithms such as Google’s are known to personalize and rank results based on various factors including popularity, query phrasing, and user behavior; they are also sensitive to the ideological slant of search queries (Tong 2025). Although the grammatical structure “I’m not a [NP]” is neutral in form, specific instantiations - such as “I’m not a racist” or “I’m not a feminist” - are ideologically charged by indexing a speaker’s stance within broader sociopolitical discourses. Tong’s (2025) study shows that even subtle variations in user input can lead to markedly different sets of 5.2 Data selection and collection 117 <?page no="118"?> search results, raising concerns about algorithmic amplification of biased query formulation. However, since my queries targeted a grammatical structure rather than semantically or ideologically polarizing content, it is unlikely that the search results were significantly skewed in ways that systematically bias the representation of NSIs. Nonetheless, it must be acknowledged that search engine indexation and ranking may have influenced which types of forums and posts surfaced first, and that this influence is inherently difficult to quantify. For this reason, I sampled instances across multiple pages of results to reduce any potential bias stemming from Google’s ranking algorithms, recording forum names and URLs and archiving thread pages locally to ensure accessibility despite search engine churn. In the next section, I explain the concrete criteria for collecting formal variants of NSIs as used on web forums. 5.2.3 Sampling criteria Corpora are “generally assembled with particular purposes in mind and are often assembled to be (informally speaking) representative of some language or text type” (Leech 1992: 116). The structure this study seeks to explore - and therefore needs to represent - is defined both grammatically (with the point of departure being the formally specified structure “I + copula + NOT + indefinite NP” and its variants) and contextually (i.e., as it occurs in online forum discussions). As previously explained, inclusion in my corpus depends on the occurrence of this structure: the corpus is intended to represent an adequate sample of instances of usage of the structure, rather than a sample of discourse on a particular topic or produced by a particular social group. This requirement, however, raises two methodological questions. First, at the level of the focal structure: which formal variants of the structure should be included, and to what extent should the corpus proportionally reflect their actual frequencies of occurrence? Second, at the level of the forums as discourse contexts: which forums should be included, and to what extent should the corpus proportionally reflect the actual distribution of the structure across these forums - or the salience of those forums within the broader category of UK-based English-language web forums? Both questions will be addressed below. 5.2.3.1 Sampling formal variants of negative self-identifiers Data collection for this study was carried out bottom-up and inclusively, meaning that there were no initial constraints regarding the topics of the forums or the demographic characteristics of the language users whose discourse I 118 5 The Empirical Study: Data and Method <?page no="119"?> investigated. However, as discussed in the previous section, search results presented by Google do not constitute a neutral or random representation of web content. They are shaped by algorithmic processes such as indexation, crawling preferences, and ranking, which influence both which websites are included in the search index and how prominently they appear in results. As a result, when searching for a phrase across a wide range of forums, certain platforms are likely to be overrepresented - due to factors such as search engine optimization, forum size or activity, or a genuinely higher frequency of the target structure. Another challenge I faced was that there is no unified search mechanism capable of capturing all formal variants at once - which would be a prerequisite for search results that accurately reflect the proportional distribution of formal variants across web forums. This introduces structural bias and makes it impossible to construct a corpus that proportionally reflects the distribution of NSIs across all UK-based English-language forums. In light of these limitations, I opted for a sampling strategy that ensured coverage of a wide range of formal variants of the target structure (Biber 1993: 244), namely, collecting a predefined sample of formally defined variants. This offered a transparent and practicable method for ensuring structural diversity, while avoiding reliance on volatile or opaque measures such as web-wide usage frequency. After defining a set of search strings, I set an upper limit of 100 tokens per variant and included all available instances for comparatively rare variants. Instances were numbered sequentially as they were collected, and sampling for a given variant stopped once 100 tokens had been reached. While this means that rarer variants are proportionally overrepresented, this does not affect the analysis, since formal variation is not treated as an analytic variable; the focus lies on the semantics of the identifying NP and its discourse functions. To avoid inflating counts, duplicate instances (the same post retrieved by multiple query variants) were identified and removed. Certain variants - such as adverbially modified NSIs (e.g., I’m not actually a professional guitarist) - occur far less frequently than prototypical, unmodified constructions (e.g., I’m not a doctor), with the former appearing fewer than fifty times compared to tens of thousands of the latter. Proportional sampling would therefore have required excluding such rare forms altogether - undermining the aim of ensuring structural diversity in the first place. Importantly, formal variation was not treated as a variable of analysis. The focus of the study is on the semantics of the identifying NP and its discourse functions, not on the relationship between surface form and function. Sampling by formal variation simply provided a consistent and objective method for corpus construction. Unless a systematic relationship between formal realiza‐ 5.2 Data selection and collection 119 <?page no="120"?> tion and pragmatic function is assumed - as may be the case with expertise disclaimers or uncontracted, adverbially unmodified, or no-negated variants - the overrepresentation of particular forms does not compromise the validity of the analysis. It was thus accepted as a necessary and deliberate methodological compromise. As for the formal variants included in the corpus, selection criteria were informed by the formal theorization of negation and by the results of the written DCT presented in Chapter 2. As explained in Section 3.1, only present simple and present perfect constructions were included, as these are the tenses in which NSIs can serve the functions relevant to this study - namely, negating identification with a particular category at or up to the moment of speaking, thereby performing, rather than merely describing, negative self-identification. For example, I wasn’t a feminist in 2001 does not function as an NSI in this sense, just as I declared the session closed is used descriptively rather than performatively (Austin 1962/ 1975; Huang 2015). While it would certainly be of interest to examine variants involving modal or auxiliary verbs -such as future-oriented constructions (e.g., I will never be a quitter) - doing so would require a shift in analytical focus and is therefore beyond the scope of the present study. The instances of “I + copula + NOT + indefinite NP” included in the corpus vary along the following formal dimensions: • Tenses: present simple, present perfect simple (I’m not, I’ve never been, I am not, I have never been) • Contraction: I’m not, I am not • No-negation: I am no, I’m no • Constructions with never: I have never been • Adverbs: e.g., I’m not really, I’m definitely not • Indefinite article: I’m not a/ an NSIs embedded in subject or object clauses governed by a matrix clause with a different subject - and where the matrix clause affects the overall polarity or modality - were not included in the corpus. Clearly, if someone is quoting directly or indirectly what someone else has said about them, it is not the same as if they were performing negative self-identification. Therefore, constructions in which an NSI appears within a subordinate clause controlled by a different speaker’s perspective - such as direct quotations (e.g., He said “I’m no liar”) or indirect speech (e.g., She can’t argue that I am not an expert) - were excluded. 120 5 The Empirical Study: Data and Method <?page no="121"?> 5.2.3.2 Controlling contextual variables To control for context, Google search results were restricted to language used in forums and discussion threads by adding the search parameters inurl: forum and inurl: thread to each query. These specifications were intended to systematically target at least two common URL formats used by discussion platforms, and the search was further limited to forums with a .uk domain in their web address. This restriction may have excluded UK-based forums hosted on other domains (e.g., .com or .org) and, conversely, may have over-included sites optimized for search engines by virtue of their URL structure; this trade-off is acknowledged as a limitation of the sampling design. Negative self-identifiers found in thread titles or headlines were excluded. While such cases (e.g., I’m not an avocado fan) do constitute instances of negative identification, headlines function as separate textual segments and are less embedded in the type of verbal interaction that is of interest in this study. To further constrain context, the search was limited to forums with a .uk domain in their web address. Although posting time may have interactional relevance (Dringus 2010), the forum discussions examined here are asynchronous in nature (Herring 2007). Depending on forum type and norms, considerable time may elapse between contributions, and even years-old posts may still receive replies. Given this asynchronicity - and in light of the study’s focus on the functions of NSIs - temporal factors were only minimally controlled. The sole temporal criterion for inclusion was that the initial post in a thread must have been published in 2015 or later. To avoid skewing the corpus toward a particular type of forum - or at least to mitigate the effect of Google ranking - sampling was systematically randomized. Depending on the number of search result pages for each formally defined query, I collected the same number of instances per page. For example, if a search returned 50 result pages, I sampled two NSIs from each page to reach the target of 100 occurrences. In practice, this usually meant working with the first 40-60 result pages per query. As noted above, if a variant occurred fewer than 100 times in total, all instances were included in the corpus. The type of forum from which each instance was taken was recorded and included in the corpus markup to ensure transparency regarding the composition of the dataset. Table 5.1 below presents an overview of the variants included in the corpus, the search specifications used to retrieve them, and the number of instances collected for each. Negative self-identifiers were grouped according to the variables contraction, no-negation, tense, and indefinite article. Within the present perfect tense category, variants containing never were considerably more frequent than unmodified forms. This likely reflects a tense-re‐ 5.2 Data selection and collection 121 <?page no="122"?> 20 Query strings were entered as quoted sequences (e.g., “I’m not a”, “I am no”). Adverbially modified variants were retrieved using a wildcard slot (e.g., “I’m not * a”), which allowed for a single intervening adverb. Punctuation and line breaks between tokens were treated as permissible variation. lated meaning difference: unmodified present perfect forms (e.g., I haven’t been a basketball player) tend to co-occur with time-bounded prepositional phrases (e.g., for two years). For this reason, variants with and without never were treated as separate data groups, with the latter group being smaller and containing all instances yielded by the search. Similarly, adverbially modified variants were relatively rare, which is why Group 3 is also smaller than the others. In total, the corpus consists of 936 NSIs, each accompanied by its co-text (see below). Data group Variant description URL: forum (.uk) URL: thread (.uk) Total 1 I’m not a/ n 101 100 201 2 I am not a/ n 101 101 202 3 Modified forms of 1 & 2: I’m/ am not “…” a/ n, I’m/ am “…” not a/ n 35 14 49 4 I’m/ I am no 101 99 200 5 Modified forms of 4: I’m “…” no, I am “…” no, I “…” am no 7 4 11 6 I’ve/ I have never been a/ n 100 101 201 7 Modified forms of 6: I’ve “…” never been a/ n, I have never “…” been a/ n, I “…” have never been a/ n, I have “…” never been a/ n 6 5 11 8 I’ve/ I have not/ haven’t been a/ n 47 12 59 9 Modified forms of 8: I’ve “…” not been a/ n, I’ve not “…” been a/ n, I have “…” not been a/ n, I “…” have not been a/ n 2 0 2 Total - - - 936 Table 5.1: Overview of variants of negative self-identifiers in the corpus, with token counts per variant. Searches were restricted to .uk domains via inurl: forum and in‐ url: thread. 20 122 5 The Empirical Study: Data and Method <?page no="123"?> As mentioned earlier, sampling in this study was randomized to avoid skewing the corpus - for instance, by favoring a particular type of forum. The result is a dataset that is highly varied in terms of discussion topics. Certain forum types are nonetheless overrepresented. However, this thematic imbalance does not substantially affect the analysis, as discussions are often only loosely related - or even unrelated - to the forum’s overarching theme. For example, car enthusiasts discuss relationship issues, cancer patients talk about food and work, and users on drug recovery boards exchange business advice. To provide a more structured overview of the forum landscape, the domains from which data were drawn were retrospectively grouped into thematic categories. This classification, while not part of the original sampling strategy, supports a fuller understanding of the corpus context and is discussed in Section 6.2. In sum, with respect to the initial criterion of representativeness, the corpus does not reflect thematic proportionality but constitutes a sufficiently large and formally varied sample of NSIs to support the identification of functional pat‐ terns. At the same time, it captures the diversity and liveliness of contemporary online forum discourse. 5.2.3.3 Capturing relevant co-text and interactional context Beyond the instances of NSIs themselves, what additional textual and contextual elements need to be included to make the corpus fit for analysis? When compiling a corpus of forum discussions based on the occurrence of a formally defined phrase, one must decide how much surrounding co-text and contextual information to include. The scope of context required, in turn, depends on the analytical focus and on how the phenomenon under scrutiny is theoretically conceptualized. In this study, forum discussions are treated as orderly, informal, asynchronous written conversations, with participants taking turns posting within threads on a shared topic. These discussions combine features of spoken interaction (Androutsopoulos 2006: 425) with characteristics of short written narratives (Georgakopoulou 2007, 2013). For this reason, and as mentioned earlier, I refer to forum participants as speakers rather than writers. Web discussion forums exhibit a characteristic tree-like structure. A forum - typically devoted to a broad topic (e.g., headaches) - is organized into threads with titles, which define what Auinger and Fischer (2008) term the “global context” of the discussion. These threads contain posts on a range of related subtopics. Conversations usually begin with a thread-initiating turn - often a question, request, declarative statement, or expressive speech act - that sets the 5.2 Data selection and collection 123 <?page no="124"?> communicative intent for the discussion. These initial turns typically aim to elicit a response, whether in the form of advice, empathy, or commentary. Often, the original communicative goal is itself co-negotiated by the forum community, with other users interpreting, reframing, or expanding the initial contribution in their replies. In this way, forum discussions create a topic-defined situational frame within which participants interact. Negative self-identifiers function within this frame by indexing and negotiating aspects of both the textual and situational context in which they are embedded. The relevance of contextual components for interpreting an NSI depends on its sequential placement within the thread. Such identifiers can appear in thread-initial posts - where they may reference the topic of the forum, the thread title, or the speaker’s own narrative, often simultaneously. They can also appear mid-turn, interacting with adjacent text by the same speaker, or as responses to implicit or explicit identity attributions made by other contributors. This variation necessitates the inclusion of multiple components of the forum discussion in the corpus to enable grounded reconstruction of the communicative processes in which NSIs are used. These components include: • Thread-initial turn: In most cases, the first post of the thread containing the NSI is included in the corpus because: - It contains the NSI. - It is the post to which the NSI responds. - It sets the topical frame of the discussion in which the NSI appears. • Turns by the same speaker: These include: - The speaker’s turn preceding the one containing the NSI. Preceding contributions are often relevant, as NSIs may modify, or be modified by, something the speaker has said earlier. Not all prior turns are included, however - for instance, when the same user participated in an earlier, unrelated part of the discussion. - The turn containing the NSI. As NSIs typically interact with the surrounding co-text (both preceding and following), this turn is always included. • Turns by other speakers preceding the NSI: In multi-party discussions, some NSIs respond to or are triggered by identity claims made by other participants. In such cases, additional preceding turns are included as needed to reconstruct the interactional sequence. The following schematic illustrates the components of the forum discussions included in the corpus for each instance of an NSI, highlighting the relevant turns selected for contextual analysis. 124 5 The Empirical Study: Data and Method <?page no="125"?> [Thread-initial turn] ↓ [Other speaker 1: prior turn] ↓ [Same speaker: prior turn] ↓ →→ [Same speaker: turn containing NSI] ←← ↓ [Other speaker 2: subsequent turn, if relevant] Figure 5.1: Schematic representation of turn selection for inclusion in the corpus (arrows = sequential order). 5.2.4 Ethical considerations Before compiling a corpus of authentic language data, a key issue that must be addressed is research ethics (Coffin et al. 2010; Page et al. 2014). To ensure the ethical soundness of this study, I obtained approval from Lancaster University’s FASS-LUMS Research Ethics Committee prior to initiating corpus collection and adhered throughout to the University’s Code of Ethics (Lancaster University 2009). In compliance with data protection legislation applicable at the time of data collection in 2019 - namely, the EU General Data Protection Regulation (Regulation (EU) 2016/ 679) and the UK Data Protection Act 2018 - data were collected exclusively from publicly accessible forums, that is, platforms where posts can be viewed without registration or login, and in line with site access rules (terms of service, robots.txt). One advantage of such data, in terms of participant anonymity, is that most users post under web aliases or usernames that do not correspond to their real names, making offline identification unlikely even without further anonymization. That said, individuals may still maintain stable online identities across platforms. To address this, I additionally anonymized all usernames, replacing them with generic tags to distinguish between different speakers in the corpus. Regarding the inclusion of verbatim forum posts, I decided to retain original phrasings in the final version after weighing the balance of potential benefit and risk (Bruckman 2002; Markham & Buchanan 2012). The analytical value of these examples outweighs the minimal risk of identification. Moreover, the study’s quantitative orientation - its interest in recurring patterns rather than individual cases - offers an additional ethical safeguard: personal narratives 5.2 Data selection and collection 125 <?page no="126"?> and identifying details are not the object of investigation. Thus, data excerpts reproduced verbatim are primarily short, decontextualized utterances that do not contain identifying information. In line with Pihlaja (2017: 224), longer, contextualized posts are included only when necessary to illustrate broader trends or to enable the kind of detailed qualitative discussion that would not be possible using short extracts alone. Where user roles or forum infrastructure are analytically relevant, they are described in abstracted terms. Another consideration was that, while users may not expect academics to analyze their writing, the same applies to most copyrighted material in books, newspapers, or other published media. As Pihlaja (2017: 223) argues, creators may not specifically foresee their content being examined in an academic context, but its use in scholarly research is neither unreasonable nor inherently unethical. The fact that forum contributions are publicly available and posted voluntarily means that, when used with care and sensitivity, they constitute a legitimate and valuable source of authentic language data for discourse analysis. Importantly, the overarching principle I observed throughout was to “act[] and writ[e] in a way that does not target users for ridicule or critical comment” (Pihlaja 2017: 221). 5.3 Summary In the first part of this chapter, I outlined the theoretical and methodological principles underlying this study. I explained that the analysis focuses on how variants of a predefined linguistic structure are typically used across texts and therefore it adopts a descriptive quantitative approach. This involves counting the frequency of analytically determined conceptual categories of NSIs in co-texts with particular meanings. I also argued that the study proceeds in stages and in an iterative manner, following the principle of hermeneutic research. Regarding quality control, I described how this research systematically examines different linguistic paradigms both qualitatively and quantitatively, based on explicitly defined criteria. I further noted that the data’s suitability was assessed prior to corpus compilation, and that results were partly complemented by automated analyses. The second part of the chapter explained why a self-compiled corpus was used rather than an existing CMC corpus. I described the rationale and ethical standards guiding data collection and processing. The corpus was constructed to include a sample of specific formal variants of the target structure from UK-based web forums, with a cap of 100 instances per variant to ensure formal balance. At the same time, no thematic restrictions were placed on forum 126 5 The Empirical Study: Data and Method <?page no="127"?> content, resulting in a broad variety of topics. I also detailed the principles used to include particular stretches of forum dialogue deemed necessary for interpreting the function of the target structure. The result is a corpus of 936 instances of NSIs embedded in their respective co-texts. The next chapter introduces the formal-functional framework used to analyze the relationship between these instances and their contexts of use and explains the annotation procedure based on this model. 5.3 Summary 127 <?page no="129"?> 6 The Empirical Study: Analytical Approach and Annotation 6.1 A formal-functional framework of negative self-identifiers in use This chapter presents a formal-functional framework of NSIs of the type “I + copula + NOT + identifying NP” in use. It provides the conceptual basis for the annotation of the data, to be presented in the next section, and is operationalized by means of qualitative and quantitative analysis of the corpus of NSIs, to be presented in Chapters 7, 8 and 9. Section 6.2 details the annotation; Section 6.4 outlines querying procedures. In the following, NSIs of this type will be considered in terms of their relations with their clause-internal co-text and different elements and units of the clause-external co-text to provide a conceptual grid for analyzing them. The chapter concludes with a model according to which the structure can be systematically approached. While the methods used to analyze the data are schematically explained in this section, more detailed information on the principles according to which the data were analyzed are discussed in Chapters 7, 8 and 9, presenting the results of the respective analyses. 6.1.1 Relations with the co-text As explained in Chapter 3, NSIs are linguistic structures which obtain their discourse meaning in interaction with their co-text, serving to index various aspects of the ongoing discourse situation (and beyond). The meaning of the identifying NP with which speakers contrast themselves is, of course, decisive for the function of the structure in use. In addition to considering the meaning of this NP, analyzing NSIs pragmatically requires accounting for the functional relation between the NSI and co-textual elements. First of all, though, a defini‐ tion is necessary of what co-textual elements are considered. Since my study not only seeks to qualitatively describe, but also to quantify relations between NSIs and their co-texts to identify functional patterns, the co-texts in which the structure appears need to be categorized. The criteria for categorizing these co-texts can be based on linguistic forms representing particular grammatical categories (e.g., it’s easy for me to say this in example 6.1 below is a clause coordinated with the NSI I’m not a parent by the coordinating conjunction so). These grammatically defined categories of co-text can be analyzed in terms of <?page no="130"?> their textual, ideational and interpersonal functions after being identified in the corpus. For example, I could examine all independent sentences following NSIs to find how they functionally relate to these instances of the structure. Or, these co-texts could be categorized according to their pragmatic functions (e.g., a sentence or a couple of sentences may serve as advice). In that case, the categorization criterion is the overall function as interpreted by the researcher, ‘top-down’. This approach contrasts with a ‘bottom-up’ process of analysis, where individual forms are studied to arrive at the functions they potentially jointly fulfill. Both approaches are applied in this study, because NSIs may interact with (a) their immediately adjacent co-texts, (b) with textual elements in the relatively more distal discourse context serving particular pragmatic functions, or (c) with both of these. Example 6.1 from my corpus illustrates this: 6.1. A: Hi all I am looking to research all options including the self funding of treatment. In my head the perfect treatment for my 5 year old newly diagnosed daughter would be to have a CGM and and a tubeless pump. Has anyone done the research to look into annual costs of either if self funding. If the answer to the above is yes do you have the broken down amounts for both? I’m also looking at the option of part funding as we have the potential option of the animas vibe and dexcom CGM fully funded. I would hope if they stump up this amount we could find the rest between the family as a whole. Any help with this would be hugely appreciated - B: I would consider a pump the best tool for good control, but I’d consider the Dexcom CGM more of a safety device. I’m not a parent so it’s easy for me to say this (and I’m on MDI) but if I ever have a kid with diabetes, I’d have them on a CGM as soon as possible. In this exchange, B uses the NSI I’m not a parent to react to A’s query in which they implicitly identify as a parent by referring to their newly diagnosed daughter. The NSI also postmodifies B’s advice, which is formally marked as such by the use of the modal would. Thus, together with the metadiscursive statement expressed by the coordinated clause introduced by so (it’s easy for me to say this), it serves to mitigate the potential face threat of advice-giving (by linguistically enacting authority), performed by the coordinated clause introduced by but (if I ever have a kid with diabetes, I’d have them on a CGM as soon as possible). To account for the functions of this NSI in this example, both elements pertaining to B’s own turn (in different sequential positions and with different syntactic relations with the NSI) and elements pertaining to the previous turn by A need to be considered. In addition, NSIs might themselves be modified by, for example, an adverbial phrase: imagine, for instance, if speaker B above had written Unfortunately, I’m not a parent, expressing their stance toward the identity claim negated by the structure of interest. In that case, the function of the NSI would be 130 6 The Empirical Study: Analytical Approach and Annotation <?page no="131"?> slightly different, as it would not merely serve as a sort of hedge. By commenting on the propositional content expressed by I’m not a parent, the speaker would index their stance toward this proposition and, thus, an aspect of the situational context - their attitude toward being a parent - which might have significance beyond the interaction considered here. This shows that to fully account for the discourse function of the NSI in this (invented) example, it would be necessary to consider both the premodifying adverb and its impact on the NSI as well as the impact of the NSI on the rest of the text. This is due to certain characteristics that NSIs share with discourse mark‐ ers: they can perform a range of textual and non-textual functions - from enhancing discourse coherence to fulfilling interpersonal roles - which may be differentially foregrounded depending on “topic, social situation and sequen‐ tiality” (Aijmer 2013: 31). NSIs can interact with several contextual categories simultaneously, relating not only to nearby linguistic elements but also to more distal discourse units, and not only to individual words or phrases but also to larger textual segments such as sentences or functional moves. In the (invented) example discussed above, the most proximate co-textual element is the adverb unfortunately, which evaluatively modifies the NSI and turns it into an explicit comment on the speaker’s identity. The most distal element is part of the interactional structure - namely, speaker A’s turn containing an implicit identity claim, which the NSI addresses. At the same time, the NSI also fulfills a politeness function in relation to the rest of B’s advice-giving contribution. These observations have two methodological implications for the corpus-based analysis of NSIs as a linguistic structure whose discourse function emerges in relation to its co-text. First, the analysis must consider a sufficient span of horizontal co-text (Collins 2019: 19; see also Section 7.1). Second, because it is not possible to anticipate or systematically search for all the potential functions that NSIs may perform, the analysis requires a principled selection of which formally defined co-textual categories to examine - and in what order. Moreover, the method employed here - manual inspection of a tagged corpus using concordance software - does not allow for the simultaneous analysis of interactions between the structure and multiple co-textual categories. However, this limitation may not be critical. As Pichler (2010: 599) argues, formal-func‐ tional “models with fewer domains might be preferable” for “quantitative purposes”. To identify syntagmatically established functions using vertical methods - that is, to make them countable - this study distinguishes between formally and functionally defined categories of co-text. These categories are differentiated based on (1) their proximity and syntactic relation to the NSI (e.g., whether they 6.1 A formal-functional framework of negative self-identifiers in use 131 <?page no="132"?> are clauses formally connected to the NSI or separate preceding sentences), and (2) their structural scope (ranging from phrases to full sentences to larger textual units). The analysis begins with the most proximate linguistic elements and grad‐ ually ‘zooms out’ to examine how NSIs interact with increasingly larger stretches of text and how they contribute to global discourse functions. This stepwise approach - starting with micro-level lexicogrammatical features using a transitivity framework (see Chapter 8) - allows for a highly systematic and fine-grained analysis. However, such detailed linguistic categorization can ini‐ tially obscure broader communicative effects. To counterbalance this limitation, the analyses are supplemented with qualitative discussions of individual corpus examples, focusing on the more global pragmatic functions of NSIs within their immediate discourse environments. Table 6.1 below schematically presents the formal-functional categories of co-text considered in the analysis. The list begins with phrases that modify the structure clause-internally and ends with textual units above sentence level that may interact with NSIs. While such higher-level units cannot be formally predefined, they can be identified through qualitative analysis, as their functional coherence is established by sequential meaning relations. Accordingly, the highest structural unit of co-text annotated in the corpus is the sentence. The table includes concrete examples from the corpus to illustrate the types of functional relationships these analytical units may have with NSIs. Some categories are included for the sake of completeness but do not occur in the examined data and are therefore not accompanied by examples. In some cases, the form of a co-textual category determines its function in relation to NSIs - for instance, an NSI occurring in a coordinated clause introduced by but typically signals contrast with the preceding clause. However, functional relationships often do not follow directly from structural form. Their interpretation instead depends on the specific linguistic elements involved. For example, to be fair, to be frank, and (just) to repeat share a similar grammatical structure, but their functional impact on the NSI differs based on meaning. Similarly, an NSI might modify the interpretation of advice-giving, which could be expressed in a single word or span multiple sentences. Therefore, rather than predetermining the functions of contextual units surrounding NSIs, the analysis identifies and annotates structural paradigms, the functions of which are then interpreted through qualitative analysis. As such, the functions illustrated in Table 6.1 apply specifically to the examples shown and should not be generalized to the formal categories themselves. 132 6 The Empirical Study: Analytical Approach and Annotation <?page no="133"?> Finally, depending on whether clause-internal or clause-external elements are being considered, the functional relationship can be described in two ways: either with reference to the co-text modifying the NSI, or vice versa, with the NSI modifying the surrounding co-text. This reflects the fact that NSIs are inherently context-dependent structures: only in the absence of any surrounding discourse can they be interpreted as self-contained propositions, modified solely by their internal preor postmodifiers. Co-textual category Category member Corpus example Functional rela‐ tion of co-textual category with NSI Phrases preor postmodifying NSIs Infinitive clause To be honest, I’ve never really been a big eater Expresses stance toward the proposi‐ tion in the NSI - Present participle clause I’m not a fan of fixed track, having last used sectional track when a teenager. Indicates causal re‐ lation to the NSI - Range adverbial phrase (when it comes to, in terms of, as for…) With regard to the retinography, I’m no expert Restricts the truth value of the NSI to a specific domain - Sentence adverbial (luckily, obvi‐ ously…) Unfortunately, I’m not a Mac guy. Conveys epistemic or affective stance toward the NSI - Comparative ad‐ verbial phrase Contrary to how it may seem, I’m not a big fan of supple‐ ments Draws contrast or comparison within the discourse or sit‐ uation - Temporal adverbial phrase Throughout my ca‐ reer, I have not been a teacher Specifies the tem‐ poral scope of the NSI - Sequence/ discourse-manage‐ ment adverb Finally, I’m no doc‐ tor/ As I said before, I’m no engineer Positions the NSI within the dis‐ course context - Metadiscursive phrase Warning: I am not a Windows program‐ mer./ I confess that I am no expert on such matters Provides metadis‐ cursive framing of the NSI 6.1 A formal-functional framework of negative self-identifiers in use 133 <?page no="134"?> Co-textual category Category member Corpus example Functional rela‐ tion of co-textual category with NSI Clauses superor subordinated to NSIs Concessive clause Although I haven’t been a good boy I de‐ serve one of those frames/ I have not been a big user of this although I have “played” with it a few times Presents a conces‐ sive contrast to the NSI - Reason/ result clause I’m not entirely sure why I refer to cash machines as ‘cash-points’, as I’ve never been an LTSB customer/ I’ve never been a fan of Panar‐ acer tyres because of their ultra thin side-walls States a cause or consequence rela‐ ted to the NSI - Relative clause My cousin did a poem on the invite, which I’m not a huge fan of Positions the speaker in relation to someone/ some‐ thing in the main clause Independent sen‐ tences preceding or following NSIs (coordinated or not) Causal connection (because, as) And I really need to lose the Saxo mir‐ rors, [because] I’m not a fan Provides a reason for another state‐ ment - Additive connec‐ tion I’m not so sure early blood tests is a good thing as I live a good 2 hours away and I’m not a morning person Adds emphasis or reinforcement to another statement - Contrastive con‐ nection (however, but, though) I am new to the fo‐ rum, however I am not a newly diag‐ nosed diabetic Presents a contrast to another state‐ ment - Other-speaker ut‐ terance - statement A: And the sheep can now sleep peacefully knowing the Aber‐ donian has returned home B: I’m actually not an Aberdonian Rejects an explicit or implicit identity claim 134 6 The Empirical Study: Analytical Approach and Annotation <?page no="135"?> Co-textual category Category member Corpus example Functional rela‐ tion of co-textual category with NSI - Other-speaker ut‐ terance - question A: Any of the legal experts know if he’s likely to be facing a custodial sentence? B: I’m not a legal ex‐ pert Answers a question directly (adhering to or flouting Gri‐ cean Maxims) - Other-speaker ut‐ terance - exclama‐ tion A: Good photo that! You’d almost think it was another species! B: I’m not a good photographer Rejects an explicit or implicit identity claim in an expres‐ sive Textual units at or above sentence level preceding or following NSIs - Example from cor‐ pus: long preface seeking advice on bike - Table 6.1: Formal-functional framework of NSIs developed for this study. Negative self-identifiers, as mentioned earlier, usually interact with more than one element of the co-text, fulfilling several discourse functions in the situa‐ tional context. Like discourse markers, they can “signal feedback to a preceding utterance and point forwards to the following turn” (Aijmer 2013: 30). In this study, because of its quantitative orientation, only those categories of formally and functionally defined co-text are selected for qualitative functional analysis that turn out to occur frequently in the data. For example, the study might reveal that the category “NSI provides reason” from the table above features quite often in the corpus. According to the methodological approach I adopt, this would result in members of that category - that is, sentences formally marked as consequences of NSIs - being examined by means of qualitative, functional analysis. The aim of this approach is to find high-frequency relations of form and function. Therefore, the focus of the analysis is on a relatively small set of particular, frequently occurring categories of NSIs and types of co-texts, and not all instances of the structure in the corpus are examined in detail (this, of course, might still be undertaken in future projects with different research objectives). As indicated earlier, because discourse components above sentence level have no fixed form, content or function, it is impossible to create a formal-functional framework for their analysis. It is, however, possible to stipulate a set of characteristics prior to analysis based on which a group of individual sentences can be claimed to constitute a functional whole, that is, a (short) text fulfilling 6.1 A formal-functional framework of negative self-identifiers in use 135 <?page no="136"?> a particular communicative purpose which an NSI can be used to modify. As argued in Section 4.2.4, the notion of move as a macrostructural component of a text type (or genre) is useful to differentiate between such functional - rather than only grammatical - categories of co-text. Accordingly, in my study textual passages are considered functional discourse units if they have an identifiable macro theme or topic, a discernible communicative purpose and textual cohesion. Example 6.2 from my corpus below serves as an example of this. In this exchange, speaker B uses the NSI turn-finally to modify the entire piece of advice presented by the preceding text: 6.2. A: I have a bugie [budgie] that started to pant/ squeak. It sits crumpled up with it’s [its] back feathers puffed up and squeaks. It seems to be getting louder. It´s been doing this for the last 1½ days. The squeaking is very quick. It still eats and drinks. The bird is in a cage with two others, who are not showing any signs of the same. I’ve read that a pure seed diet will give lack of iodine and this can lead to respitory [respiratory] problems. They are in our kitchen (large kitchen), maybe that’s the cause too.Does anyone have any suggestions. I will try and take it to the vet tomorrow. I live in the south of sweden and there seems to be a definate[definite] lack of avian vets. - B: hello, Bart my budgie had the exact same problem over a week ago. She is on the mend now but I gave her iodine in her water, treated her with ivermectin to rule out mites and also when I cover her up at night I put a cup of hot chamomile [chamomile] tea outside of her cage under her blanket so that she could breathe in the vapours. It appears that camomile relaxes the airways a bit as it is a natural anti inflammatory and reduces swelling. I did the camomile tea for 2 nights and she greatly improved after this. I would move your bird away from any open windows/ draughts and any smelly things at all like deodrants [deodorants] etc. I have started making my husband spray all his smellies in the bathroom away from Bart! Hope this helps a bit---I’m not an expert but there are many people on here who have been very helpful to me and I’d like to share my experiences after Bart has been unwell.. Speaker B’s posting is a response to A’s request for advice on a health problem of their budgie. The advice B provides is framed as a story reporting on a similar problem with their budgie, Bart, and relating how the speaker successfully treated it. After having told Bart’s story, the speaker concludes their posting with an expressive, metadiscursively commenting on their text (Hope this helps a bit), which is immediately followed by an NSI disclaiming expertise. This epistemically mitigates the entire piece of advice presented in the form of a story. Referring to the features mentioned above, this ‘advice story’ can be analyzed as follows: 136 6 The Empirical Study: Analytical Approach and Annotation <?page no="137"?> • L EVEL O F IDEATIONAL METAF UNCTION : An identifiable macro theme/ topic (e.g., the budgie’s illness and its treatment) - Semantic relations between lexemes allowing identification of a unify‐ ing topic. For example, in the example above: conceptual domains “problem” (problem, on the mend, improve), “body” (breathe, airways, swelling), “treatment” (iodine, ivermectin, chamomile tea, anti-inflammatory) • L EVEL O F I NTERP ER S ONAL METAFUNCTION : A discernible communicative purpose, or “pragmatic aboutness” (Pípalovà 2008) reflected in the relation‐ ship between writer/ reader constructed in the text (e.g., sharing experience with budgie treatment) - Sentence types: mainly declarative, no directives or interrogatives → Providing information - Personalization: focus on writer or reader E.g.: I gave, I did, it appears, I have started, I would → focus on writer - Standing: representation of author as possessing authority/ expertise on the subject ○ Reference to external sources: other people who have been helpful to me ○ Social distance: low (informal expressions used, e.g., smellies) → Low demonstration of authority in the example - Stance: epistemic and deontic modality ○ Epistemic modality: it appears that; otherwise strong ○ Deontic modality: low (I would, otherwise no obligation expressed) • L EVEL O F TEXTUAL METAFUNCTION : Textual cohesion - Organization into macro-, hyperand clause themes; logical thematic progression E.g.: Introduction of budgie with same problem (Bart my budgie…), which is now solved (She is on the mend…), description of what was done to solve problem (I gave her, I did…), final recommendation (I would…) - Lexical cohesion: repetition, meaning relations between lexemes used E.g.: my budgie - she - her cage; gave her, treated her, covered her - Grammatical cohesion: linking adverbials, tense consistency E.g.: consistent use of past tense marking the passage as “story” To sum up, a functional discourse unit is conceived of here as a thematic, cohesive whole with a clear communicative function (e.g., relating a story, giving advice). Another characteristic of such discourse units is that they can normally be metadiscursively labelled as such. For example, in the expressive Hope this helps a bit in example 6.2 above, this refers to the entire discourse 6.1 A formal-functional framework of negative self-identifiers in use 137 <?page no="138"?> unit. Another example from the corpus which illustrates this point are phrases like Just to give you some background info, followed by a longer textual passage, which is then commented on using an NSI. The extent to which reference of NSIs to discourse segments above the sentence level plays a role for their function (and also what kinds of functional units interacting with NSIs can be found in the corpus used for this study (as introduced in detail in Section 5.2) will be determined in the analysis (this chapter and Chapter 7). 6.1.2 Analytical model To conclude, this section has discussed how the relationship between NSIs and their clause-internal and clause-external co-texts can be conceptualized. I have defined categories of co-texts relevant for studying NSI functions across texts, distinguishing them by formal and functional aspects. The following table summarizes the approach to studying NSIs and reiterates RQ1 and RQ2. RQ3, concerning the socio-political significance of the results, underpins all stages of the analysis and is therefore not tied to a single analytic step. Linguistic paradigm analyzed Research question(s) Method(s) Nouns and their preand postmodifiers in identifying NPs 1A. With which nouns and NPs do forum users nega‐ tively identify? 1B. To which conceptual categories can these nouns and NPs be assigned, and how prominently (in fre‐ quency and lexical varia‐ tion) are these categories represented? Conceptual profiling Words and phrases preand postmodifying NSIs Forms linking NSIs to their clause-external co-text 2A. What are the for‐ mal-functional relations of NSIs in my corpus and their clause-internal and clause-external co-texts? Qualitative and quantitative functional analyses of for‐ mally defined categories (parts of speech clause-internally modifying NSIs, and elements formally and functionally re‐ lating NSIs to their clause-ex‐ ternal co-text) 138 6 The Empirical Study: Analytical Approach and Annotation <?page no="139"?> Clauses and sentences formally related to NSIs Sentences not formally related to NSIs (poten‐ tially part of higher-or‐ der functional units) 2B. What are the meanings and functions of co-texts with certain formal links to the NSI structure, and how frequently are they represented in the corpus? Qualitative (transitivity) anal‐ ysis of particular co-texts → quantification; qualitative functional analysis of senten‐ ces constituting particular process-participant configu‐ rations → quantification - 2C. Do co-texts with spe‐ cific meanings and func‐ tions occur together with particular conceptual cate‐ gories of NSIs in patterned ways? Cross-categorization of func‐ tional categories of co-text with conceptual categories of NSIs Table 6.2: Analyses conducted to answer Research Questions 1 and 2. 6.2 Corpus annotation This section outlines the annotation procedures applied to the corpus in order to examine relationships between particular conceptual categories of NSIs and specific features of their textual and contextual co-texts. Drawing on the formal-functional framework just presented, I defined categories that could be marked in the text to enable both qualitative description and quantitative analysis of form-function patterns. Beyond that, the section also offers a snapshot of what was practically and methodologically feasible for a single researcher working without AI-based support in the late 2010s and early 2020s. At the time, the use of corpus methods in discourse pragmatics was still viewed with skepticism - as Archer and Culpeper (2018: 468) put it, “the corpus-based method does not seem to promise much reward for pragmatics research given its typical focus on form”. Pragmatic annotation had received relatively little attention even in key publications in the field (e.g., Rühlemann & Aijmer 2014), as pragmatic meaning was considered too context-dependent or functionally diffuse to be captured through searchable tagging. Indeed, there are methodological intricacies involved in annotating pragmatic data, and in the following, I explain how I approached these challenges in annotating this corpus - through “carefully interweaving horizontal and vertical methods of analysis”, as advocated by Haugh (2018: 619). Revisiting his “call for further research in corpus-based metapragmatic studies” from today’s perspective, it is clear that the field has evolved considerably: corpus-based approaches are now widely established in pragmatics (Landert et al. 2023), 6.2 Corpus annotation 139 <?page no="140"?> and advances in annotation tools - including recent developments in large language models (LLMs) and other natural language processing techniques - have significantly expanded researchers’ methodological options. Against this backdrop, the annotation strategies described here reflect both theoretical choices and the practical limitations of their time. Regarding the intricacies of annotating corpora for pragmatic analysis, one of the first questions concerns the principles by which the data should be segmented and what information about these segments should be encoded. Generally, the more information is included in the annotation, the more interpretive work has already been carried out - allowing results to be obtained more efficiently, but at the cost of reduced flexibility in post-annotation analysis. The less form-based and the more contextand interpretation-dependent the categorization, the greater the potential for annotation to become a source of dispute (Archer 2012). As Sinclair (2004: 191) warns, categorizing data into functional units risks the danger of viewing corpus data “through the tags”, thereby overlooking aspects not captured by the annotation. Once larger discourse units are joined together based on functional rather than formal criteria, they can no longer be examined in isolation, and other elements - perhaps equally or more relevant for functional analysis - may be excluded from quantitative consideration, which typically relies on searchable tags. This, incidentally, is also one reason why discourse annotation - which often involves highly context-dependent categories and is thus predominantly carried out manually - is relatively rare, in contrast to POS-tagging, which can be per‐ formed automatically with near-complete accuracy (but see Culpeper & Archer 2008 for a corpus annotated for discursive features such as specific adjacency pairs). In the present study, as will be explained below, a problem-oriented annotation scheme was applied. Rather than aiming for exhaustive markup of the linguistic context of NSIs, the annotation focused selectively on features that would facilitate addressing the study’s research questions. The stages of markup and analysis were not entirely distinct; instead, annotation and interpretation were carried out iteratively, with functional categories and tags being refined, modified, or abandoned as the analysis progressed. Regarding the structure of interest in this study, the only tag consistently applied across all variants of the pattern “I + copula + NOT + indefinite NP” was one serving as a wildcard. This enabled the retrieval of all relevant instances in the corpus at once. However, annotation became more complex when it came to the semantic content of the indefinite NPs used in these structures. In order to search for particular conceptual categories of NSIs - which serve as a starting point for their functional interpretation - semantic annotation 140 6 The Empirical Study: Analytical Approach and Annotation <?page no="141"?> was indispensable. This necessarily meant imposing an interpretation, albeit a principled and transparent one, on each instance from the outset. Initially, I also considered annotating the functional relations between NSIs and their surrounding co-texts. Specifically, I experimented with tagging whether the structure modified preceding or following elements, requiring the bracketing and labelling of nearby discourse segments as functionally impacted by the NSI. This approach ultimately proved problematic. First, it lacked transparency, as determining whether a given linguistic element was functionally modified by the structure in question often depended on subjective or context-sensitive judgments. Second, it risked obscuring their capacity to point to multiple possible functions or referents in context. As Archer and Culpeper (2018: 499) emphasize, “[p]ragmatic phenomena cannot be reduced to binary choices”. Consequently, these functional interaction tags were dropped in favor of a leaner scheme that tagged only categories with independent discourse-func‐ tional value, irrespective of whether they turned out to be relevant for the function of the NSIs. This more flexible annotation strategy follows Pichler’s (2010: 600) advice to “categorize tokens in ways that allow maximum flexibility in data quantification”. In its final version, the annotation scheme focused on the immediate, turn-in‐ itial co-text of NSIs, marking basic syntactic features (e.g., sentence type: declarative, directive) and select formal relations (e.g., coordination, contrast). A smaller number of clause-internal elements were also annotated where relevant for interpretation, such as discourse markers (e.g., anyway, to be honest), temporal adjuncts (e.g., since December), and metadiscursive elements (e.g., note that). Larger dialogic units - such as advice-giving or experience-relating - were analyzed qualitatively and annotated only if structurally salient. In sum, the annotation prioritizes interpretive openness over exhaustive markup, balancing analytic structure with methodological transparency. Table 6.3 below provides an overview of the semantic categories that in‐ formed the annotation of NSIs in the corpus. These categories were developed to capture different dimensions of meaning associated with the indefinite NP in the NSI structure. They include general characteristics (e.g., social, ideological, or health-related attributes), roles specific to social and digital contexts, habitual behaviors, varying levels of expertise, and expressions of preference. The labels [+ expert] and [expert], as well as [+ fan] and [fan], do not refer to actual levels of knowledge or orientation, but merely indicate the presence or absence of lexical items such as expert, fan, enthusiast, or lover in the self-identifying phrase. The category “roles” refers to identities and social positions that are particularly 6.2 Corpus annotation 141 <?page no="142"?> relevant within the online environments in which the NSIs were produced, such as forum-specific responsibilities or interactional positions shaped by digital discourse norms. The examples are drawn from the annotated corpus. Category and subcategory Tag pattern Representative example Characteristics - - General <nsi-CG> I’m not a parent Evaluative <nsi-CE> I’m not an ‘overprotective’ mother Psychological <nsi-CM> I have never been an anxious person Physical <nsi-CP> I’m not a tall or stocky sort of person Social <nsi-CS> I’m not an isolated person Ideological/ religious <nsi-ID> I’m not a nationalist Health/ illness <nsi-PAT> I’m not a diabetic Metaphorical <nsi-ME> I am not a tortoise Geographic <nsi-GEO> I’m not an EU resident Roles (forum-internal) - - Social role <nsi-RS> I’m not a saint on this forum Membership <nsi-RM> I’m not a member Trade/ profession <nsi-RT> I’m not a bookseller Virtual identity <nsi-RV> I’m not a rich player Forum-specific <nsi-RF> I’m not a forum admin Habits - - Routine <nsi-HR> I’m not a morning person Food/ drink <nsi-HF> I’m not a big drinker Substance <nsi-HS> I’m not an intravenous drug user Product-related <nsi-HP> I’m not a big user of this Activity-related <nsi-HA> I’m not a runner 142 6 The Empirical Study: Analytical Approach and Annotation <?page no="143"?> 21 The category “usage/ consumption/ ownership” is treated as a fringe category, as its subcategories overlap with “preference”, “habits” and “expertise” without fully aligning with any of them (see Section 7.2.7). Category and subcategory Tag pattern Representative example Usage/ consumption/ ownership 21 - - Usage <nsi-USE> I have never been a serious user of the later versions of FTM Consumption <nsi-CON> I’m not a Sky subscriber Ownership <nsi-OWE> I’m not a car owner Expertise (+ expert) - - General <nsi-EX> I’m not an expert Specific <nsi-EXS> I am not a DICOM expert Expertise (expert) - - General <nsi-PRO> I am not a pioneer Specific <nsi-PROS> I am not a native English speaker Profession <nsi-PROF> I’m not a cardiologist Activity <nsi-PROA> I am not a runner Preference [+ fan] - - General <nsi-FAN> I’m not a fan Activity <nsi-FAN-A> I’m not a great lover of positions 2/ 4 Activity-involved <nsi-FAN-AI> I follow Daveo & Boycie Nature/ animals <nsi-FAN-AN> I am not a fan of slow worms Business-related <nsi-FAN-B> I’ve never been a fan of the Scandina‐ via markets Entertainment <nsi-FAN-E> I’ve never been a fan of “So Broken” Ideological/ religious <nsi-FAN-ID> I am not a fan of violence for the sake of violence IT-related <nsi-FAN-IT> I am not a fan of the Instacrew concept Product <nsi-FAN-P> I’m not a big fan of tablets 6.2 Corpus annotation 143 <?page no="144"?> 22 Within each <nsi-…> span, the semantic core - the noun or NP that specifies what is being disclaimed - was additionally marked using <ic-…> tags. Both tags carry a category label as part of the element name (e.g., <nsi-HA>I have never been a <ic-HA>smoker</ ic-HA></ nsi-HA>), which enables a more fine-grained analysis of lexical patterns. The notation is TEI-inspired but not formally TEI-conformant, since TEI normally encodes categories as attribute values (e.g., <nsi type="HA">…</ nsi>). Category and subcategory Tag pattern Representative example Person <nsi-FAN-PE> I’m still not a fan of [NAME] Visual element/ part <nsi-FAN-V> I’ve never been a fan of the mac-style icons Preference [fan] - - Activity <nsi-PR-A> I’ve never been a gym person Activity-involved <nsi-PR-AI> I have never been a TGS kind of rider Animals/ pets <nsi-PR-AN> I’m not a dog lover Entertainment <nsi-PR-E> I’ve never been a Radiohead hater Ideological/ religious <nsi-PR-ID> I am not a slavish adherent to their politics IT-related <nsi-PR-IT> I’m not a database guy Product <nsi-PR-P> I’m not a Mac person Person <nsi-PR-PE> I’m no supporter of [NAME] Food <nsi-PR-F> I am not a cream lover Style <nsi-PR-ST> I have never been a pink person Sexual <nsi-PR-SEX> I have never been a thong man Table 6.3: Semantic categories of negative self-identifiers and corresponding tags. 22 Each annotated NSI was assigned a tag based on the semantic category of the indefinite NP. In two cases (expert and fan), the categories reflect the presence or absence of the respective lexical item; all other categories were defined based on shared semantic features. The tags consist of a base code (e.g., PRO for non-expert expressions) and optional suffixes indicating subcategories, such as F for profession-related expressions. This tagging structure was designed to allow precise yet flexible querying across conceptual dimensions. 144 6 The Empirical Study: Analytical Approach and Annotation <?page no="145"?> 23 The examples in the table are drawn from various parts of the corpus and are intended to illustrate the tagging logic rather than represent any particular sequence of interaction. In addition to the semantic categorization of NSIs, the corpus was also anno‐ tated for contextual and interactional features to enable a more fine-grained analysis of how NSIs function within discourse. As will be discussed in more detail below, however, assigning forum categories or topics necessarily involves interpretive judgment, given that online forums often span multiple themes, genres, and audience orientations. Moreover, the study did not aim to establish statistically significant correlations between NSIs and thematic contexts. Table 6.4 provides an overview of the annotation categories used to capture dialogic aspects of the forum discussions, as well as both the linguistic and non-linguistic co-text of the NSIs. 23 Meta-textual information about the communicative situation includes: • the forum category in which the NSI occurs (e.g., health, hobby, product-re‐ lated); • the specific topic or subdomain of the forum (e.g., Parkinson’s, car mainte‐ nance); • the title of the discussion thread; • the unique identifier for the document in which the NSI appears; • the participants whose turns were included in the analysis. The co-text surrounding each NSI was further annotated across several struc‐ tural and functional dimensions, as defined by the formal-functional framework introduced earlier: • Functional components of interaction - Turns by various speakers - Speech acts (e.g., greetings) and discourse moves (e.g., advice, experi‐ ence-sharing) • Grammatical units and their functions - Clauses: annotated for ○ pragmatic functions (e.g., directive, declarative) ○ functional relations to other clauses (e.g., contrastive via but) - Phrases: annotated ○ structurally (e.g., NPs) ○ functionally (e.g., discourse markers, stance markers) • Non-linguistic and embedded content - Visual elements (e.g., images) - Quoted or embedded text from other users (non-authorial text) 6.2 Corpus annotation 145 <?page no="146"?> Tag Description Example Metatextual <teiHeader> Encodes document-level meta‐ data including forum type (fo‐ rum), topic (topic), thread title (title), and URL. Each header has a unique identifier (id). <teiHeader id="nsi_002" fo‐ rum="historiccoventry.co.uk" topic="history/ community" ti‐ tle="…" url="…"/ > Turns and functional moves <Sn> Turn by other speaker <S2> Diabetes is diabetes. If we are not careful, we will start differenti‐ ating on this board. <NSIn> Turn by speaker using NSI <NSI2> … <nsi HA>I have never been a <ic HA>smoker</ ic HA> … <smove>, <xsmove> Functional move before/ after NSI <smove> … it’s guesswork IMHO. </ smove> Sentence and clause-level functions <is: type>, <xis: type> Sentence before/ after NSI: s (statement), qu (question), ex (expressive), dir (directive) <xis: qu> Would I be okay to study CIPD intermediate level? <cj: type>, <xcj: type>, <sub: type>, <sup: type> Clause type: conc (concession), + (addition), con (contrast), cau (cause), cq (consequence), rel (relative), cond (conditional) Whilst <cj: conc>I’m not a single traveller … I’m not a programming expert<xcj: cq> so is there any way to fix this? Phrase-level annotation <par>, <xpar> Participial phrase Agreed re John Fahey; I’m not a great fan <np> Noun phrase Descent, 2nd Edition - I have never been a great fan… <pt>, <xpt> Discourse particle Now, I haven’t been a carb nazi since diagnosis Discourse-pragmatic functions <sq>, <xsq> Sequence markers First I’m not a fire door specialist <ev>, <xev> Evaluative adverbs Unfortunately, I’m not a Mac guy <ep>, <xep> Epistemic markers No of course I’m not a pirate <st>, <xst> Stance markers I’ll be honest, I’m not a massive fan of woody cabs 146 6 The Empirical Study: Analytical Approach and Annotation <?page no="147"?> <r>, <xr> Range specifiers … I’ve never been a fan of it being silver <dl>, <xdl> Metadiscourse/ labels DISCLAIMER - I’m not a solicitor <temp>, <xtemp> Temporal specifiers Since Dec I haven’t been a proper mum <per>, <xper> Perspective/ focus adverbs Personally I’ve never been a big fan… <inf> Information status As some of you may know I have not always been a fan… <hl>, <xhl> Highlighting markers I repeat, I’m no expert … and usually wrong <comp>, <xcomp> Comparative adverbials I’m not a native speaker like you <acc>, <xacc> Accuracy modulators Not a particular fan as such Quoted and non-verbal elements <q>, </ q> Quoted forum text <q> Dear Seller, We have charged your credit card … </ q> <a>, </ a> External source (e.g., news link) Embedded article or URL <pic> Non-verbal content <pic> A lovely shark tooth, you must be very excited. Table 6.4: Overview of categories annotated in the corpus. As with the tags for the conceptual categories of identifying NPs, superordinate grammatical categories are represented by a shortcut (e.g., p for phrases), to which suffixes are added to differentiate subcategories. For example, in the sentence I’m not a legal expert,<cj: con> but from that I would suggest that the answer to your question is ‘yes’, the prefix cj indicates that the NSI is followed by a clause introduced by a conjunction, and con specifies the type of conjunction - here, a contrastive one. To distinguish between linguistic elements occurring before or after the NSI, the prefix x is added to the tag for any post-NSI material. For example, if an NSI is preceded by a subordinating concessive conjunction, this is tagged as <cj: conc>, as in: Although <cj: conc><nsi-EX>I am not a <ic-EX>DICOM expert</ ic-EX></ nsi-EX></ cj: conc>, 6.2 Corpus annotation 147 <?page no="148"?> A subordinate concessive clause following an NSI is marked by <xcj: conc>, as in: <nsi-FAN-PE>I’m not a <ic-FAN-PE>big Harry Potter fan,</ ic-FAN-PE></ nsi FAN-PE><xcj: conc>although I’ve seen most of the films.</ xcj: conc> As these examples illustrate, the tags are directional: tags referring to preceding co-text are placed directly before the NSI, while tags referring to subsequent co-text follow it. This convention allows for greater specificity and searchability when analyzing the local environment of NSIs. The advantage of this annotation system is that it enables targeted searches for both conceptual categories of NSIs and their surrounding textual or situational context. For example, using WordSmith (as was done in this study), it becomes possible to retrieve all statements that precede a particular type of NSI, or to examine whether contrastive coordination tends to co-occur with specific semantic categories. One might argue that annotating such formal aspects is redundant, since co-textual elements could theoretically be retrieved using punctuation-based searches or simple pattern matching. However, exploratory analyses showed that in informal web forum data, formal boundaries are often fuzzy: what looks like a sentence might end in a comma, emoji, or dash; what appears to be a statement might be a question. Annotating these structural relationships explicitly proved vital for maintaining consistency and transparency in the categorization process and for avoiding the need to reclassify ambiguous elements during later stages of the analysis. Another benefit of this system is its reusability: for example, a quick search for metadiscursive markers such as note that or I should say immediately reveals that they tend to co-occur with disclaimers of expertise (see Table 6.5). This may suggest that such markers are used when presenting potentially sensitive or epistemically marked information. While their frequent association with expert disclaimers might also reflect the overall prominence of that conceptual category, this possibility is explored in more detail in the analyses that follow. Finally, the development and application of this annotation system - though relatively simple in design - enabled the consistent tagging of 936 instances of informal discourse. It may thus offer a useful model for other researchers working with similar types of CMD. The specific methodological procedures applied to examine the annotated data are described at the beginning of each of the three following analysis chapters, corresponding to the study’s core research questions. The annotated corpus is not publicly available but can be shared upon request for academic purposes. 148 6 The Empirical Study: Analytical Approach and Annotation <?page no="149"?> Concordance example <hl>Like I said</ hl> <nsi-EX>I’m no <ic-EX>programmer</ ic-EX></ nsi-EX> <hl>As I said</ hl>, <nsi-PRO>I am no <ic-PRO>physician</ ic-PRO></ nsi-PRO> <hl>I repeat</ hl>, <nsi-EX>I’m no <ic-EX>expert</ ic-EX></ nsi-EX> <hl>Like I say</ hl> <nsi-EX>I’m no <ic-EX>expert</ ic-EX></ nsi-EX> <hl>again</ hl>, <nsi-EX>I am no <ic-EX>expert</ ic-EX></ nsi-EX> on these type of creatures <hl>Again</ hl>, I must stress here <nsi-EX>I’m no <ic-EX>expert</ ic-EX></ nsi-EX> But <hl>as I say</ hl> <nsi-EX>I’m no <ic-EX>expert</ ic-EX></ nsi-EX> <hl>I should say</ hl> <nsi-EX>I am no <ic-EX>expert</ ic-EX></ nsi-EX> BUT, <hl>like I said</ hl>, <nsi-EX>I am no <ic-EX>expert</ ic-EX></ nsi-EX> on this <hl>again</ hl> <nsi-EX>I’m no <ic-EX>expert</ ic-EX></ nsi-EX> <hl>I said before</ hl>, <nsi-PRO>I’m no <ic-PRO>engineer</ ic-PRO></ nsi-PRO> <hl>as I say</ hl>, <nsi-EX>I’m no <ic-EX>electronics expert</ ic-EX></ nsi-EX> <hl>As I was telling someone in another thread</ hl>, <nsi-EX>I’m not an <ic-EX>ex‐ pert</ ic-EX></ nsi-EX> <hl>I repeat</ hl> that <nsi-EX>I am not an <ic-EX>expert</ ic-EX></ nsi-EX> <hl>Have to say</ hl> <nsi-FAN>I’m not a <ic-FAN>fan</ ic-FAN></ nsi-FAN> of variegated palms <hl>I should add</ hl> that <nsi-FAN>I am not a big <ic-FAN>fan</ ic-FAN></ nsi-FAN> of head phones (this to show <nsi-HA>I am not a <ic-HA>hypochondriac</ ic-HA></ nsi-HA>) Table 6.5: Concordance lines of negative self-identifiers (<nsi-…>) preceded by discourse highlighters (<hl>), with semantic cores (<ic-…>) indicated. To situate the dataset within its broader discursive landscape, I grouped the forums from which NSIs were retrieved into thematic categories and annotated each discussion for its overall topic. This categorization followed a bottom-up logic inspired by Grounded Theory (Hallberg 2006), with categories added iteratively until 18 broad groupings emerged. Some are internally coherent (e.g., forums on specific health conditions), while others are more loosely defined, encompassing diverse platforms with a shared thematic orientation. 6.2 Corpus annotation 149 <?page no="150"?> Forum categorization is rarely straightforward: many address multiple topics, serve hybrid purposes, or target overlapping audiences. Classifications may draw on subject matter (e.g., health, cars), genre (e.g., support board, Q&A), or audience (e.g., professionals, hobbyists), often in combination. For example, www.theiet.org is a professional engineering forum, while www.gramophone. co.uk blends entertainment, expertise, and product focus. The categories follow what Glaser (2002) terms a transcending abstraction - a conceptual structure that privileges analytic relevance over exhaustive description. Since this study focuses on the functions of NSIs rather than on any one thematically defined discourse context, thematic categorization plays only a secondary role in the analysis. It is also difficult to determine whether particular forum types consistently align with specific NSI functions. As the forums examined here are publicly accessible without registration, their domains are reported in full in Appendix 1 to ensure transparency and replicability. The categorization of forums reflects an interpretive grouping for analytical purpo‐ ses and is not intended as a definitive or exhaustive classification, and Table 6.6 presents a condensed thematic profile listing two representative domains per category. Figures refer to discussions rather than NSIs; some discussions contain more than one NSI, which is why the corpus total (936 NSIs) exceeds the number of discussions listed (932). A complete list of all source forums and their categorization is provided in Appendix 1. While these groupings cannot be read as definitive or predictive, they offer a useful point of orientation for understanding the kinds of discursive environments in which NSIs appear. Thematic category No. of discussions Most frequently represented websites Health & illness 201 http: / / www.diabetes.co.uk, https: / / drugsforum.com, https: / / www.parkinsons.org. uk Products 124 http: / / community.evo.co.uk, https: / / www .blackstaramps.com, https: / / www.amazo n.co.uk Games 91 https: / / retropie.org.uk, https: / / forums.fro ntier.co.uk Sports 88 https: / / forums.runnersworld.co.uk, https: / / forum.cyclinguk.org Professional & technical topics 83 https: / / community.tes.com, http: / / www. mig-welding.co.uk 150 6 The Empirical Study: Analytical Approach and Annotation <?page no="151"?> Nature/ animals/ pets 67 http: / / www.nhm.ac.uk, http: / / www.rspb. org.uk Football and other sports clubs (Fans) 66 http: / / www.fm-base.co.uk, http: / / www.t wtd.co.uk Amazon sellers 46 https: / / sellercentral.amazon.co.uk Relationship & family 42 http: / / www.hitched.co.uk, http: / / www.yo uandyourwedding.co.uk IT-related 30 https: / / forums.overclockers.co.uk, https: / / forums.theregister.co.uk Information & debate 25 http: / / forum.iosh.co.uk, http: / / newsforu ms.bbc.co.uk Entertainment, art & travel 24 http: / / www.lse.co.uk, http: / / forum.thefis hy.co.uk Community 21 http: / / theatreboard.co.uk, http: / / www.red dwarf.co.uk Music 19 http: / / bcb-board.co.uk, https: / / www.gra mophone.co.uk TOTAL 932 - Table 6.6: Thematic profile of forums from which negative self-identifiers were retrieved, and number of discussions. As can be seen, the forum category “health & illness” (201 discussions), con‐ sisting of various self-help forums, is the most prominent source from which exchanges containing NSIs were retrieved, with NSIs from a forum on diabetes being most often represented. Another forum assigned to the domain of health is one on drugs, which might not be considered as strictly health-related by everyone. The reason for including this forum, as well as other forums on potentially problematic behaviors rather than health conditions in the category of health - such as https: / / www.gamblersanonymous.org.uk - is that the discussions on these forums tend to problematize, rather than promote these behaviors, thus casting them as conditions rather than mere practices. The second most frequent category of forums is “products” (124 discussions), including some devoted to discussing features of particular products such as (particular brands of) cars, music, knives and food. “Games” (91 discussions) contains forums where people talk about particular computer games or their technical aspects (configurations, installation, software and hardware). Thus, this category - at least regarding the level of technicality of the language 6.2 Corpus annotation 151 <?page no="152"?> used - is similar to “IT-related” (30 discussions), to which forums intended for exchanges on particular software or hardware were assigned. Forums from the category “sports” (88 discussions), in particular one on running, are also very prominent sources of NSIs. Forums from the category “professional & technical topics” account for 83 discussions featuring NSIs. This label includes sites intended for particular professional groups (e.g., teachers [https: / / community.tes.com], welders [https: / / www.mig-welding.co.u k] or pharmacists [https: / / www.pharmacy-forum.co.uk]), but also general job forums (e.g., https: / / www.wikijob.co.uk) and forums dedicated to discussing skilled activities which could, but do not necessarily, constitute professions (e.g., photography, painting, writing), as well as forums dedicated to discussing technicalities of a particular area of expertise usually defined by an object of common interest, e.g., 3D scanning (https: / / laserscanningforum.com) or roads (http: / / www.sabre-roads.org.uk). Among the results, forums grouped under the category “nature/ animals/ pets” (67 discussions) included websites devoted to plants, animals, and pets, ranging from reptiles and amphibians (https: / / www.herpetofauna.co.uk), to barbel (htt ps: / / www.barbel.co.uk), a species of fish, and to turtles (http: / / www.hermann-t ortoise.co.uk), with the Natural History Museum’s website (https: / / www.nhm. ac.uk) and the Royal Society for the Protection of Birds (https: / / www.rspb.org. uk) featuring particularly prominently. “Football and other sports clubs (fans)” (66 discussions) refers to forums for fans of a wide variety of local football (and other sports teams) clubs, referring to themselves e.g., as “Bolton nuts” or “Pride of Nottingham”. A separate category is constituted by “Amazon sellers” (46 discussions). “Relationship & family” (42 discussions) includes forums on various subjects around human relationships and with different purposes, e.g., an advice-giving forum on what could be called manhood or masculinity (http: / / aa ronsleazy.blogspot.co.at), a website aiming to help people find their lost family members (https: / / www.genesreunited.co.uk) or forums to discuss marriage issues (thehappymarriedcouple.blogspot.co.at). The category “information & debate” (25 discussions) covers forums intended to provide information about various subjects, such as taxation and bankruptcy (http: / / bankruptcyhelp.org.uk). “Enter‐ tainment, art & travel” (24 discussions) includes forums on particular films, books and personalities, leisure activities and travel, and “Community” (21 discussions) includes forums for particular local communities (e.g., https: / / www.eastdulwic hforum.co.uk). Finally, “music” (19 discussions) includes forums for discussing artists, albums, and musical genres (e.g., http: / / bcb-board.co.uk/ ). 152 6 The Empirical Study: Analytical Approach and Annotation <?page no="153"?> 6.3 Data presentation and transcription conventions Because the data represent informal, written, asynchronous conversations, I refer to forum contributors as speakers rather than writers. Forum discourse shares many features with spoken interaction (e.g., non-standard orthography, incomplete sentences). These features were preserved in the corpus to maintain stylistic authenticity where they do not impede readability and analysis. In quoted examples, minor spelling corrections are indicated in square brackets; culturally specific, technical, or otherwise obscure terms are briefly glossed in footnotes where feasible. Orthographically idiosyncratic forms (e.g., Dog lover) are reproduced verbatim in quotations, while in my metalinguistic discussion such expressions are normalized (e.g., dog lover). In line with ethical considerations, all potentially identifying information has been removed. To ensure gender inclusivity, I use third-person plural pronouns for speakers; where this is not possible, I default to the female form. All corpus examples are presented according to the notational conventions and abbreviations listed in the Conventions and Abbreviations section (p. i). Chapter-specific symbols are introduced at their first point of use. 6.4 Querying and analyzing the annotated corpus Having described the annotation scheme, this section outlines the tools and procedures used to query and analyze the tagged data. Both qualitative and quantitative analyses were conducted, with the former also feeding into the annotation process itself (see Chapter 7). The primary tool used for quantitative analysis was WordSmith Tools 5.0 (Scott 2008), specifically the Concord function, which allows for the retrieval of linguistic forms or tags and displays them in keyword-in-context (KWIC) format (Hunston 2002: 48; Collins 2019: 9). In this study, the analysis relied not on keywords in context, but on key tags in context - reflecting the conceptual categories defined during semantic annotation. For example, the tag marking instances of product preference disclaimers (as outlined in Table 6.4) was used to retrieve a concordance list of NSIs from that conceptual domain, sorted by the first word to the right of the node (R1). Figure 6.1 illustrates one such list, with tagged structures highlighted in blue and the R1 items in red. Similarly, Concord was used to explore the co-occurrence of NSIs with discourse-structuring elements in the surrounding co-text. Figure 6.2 presents concordance lines showing contrastive conjunctions (tagged as <xcj: con>) following disclaimers of expertise (tagged as <nsi-EX>), providing a 6.3 Data presentation and transcription conventions 153 <?page no="154"?> basis for investigating the interaction between conceptual categories and formal coordination structures. The resulting concordances were examined in greater detail to identify subcategories of identifying NPs and to explore the local discourse functions of NSIs. To supplement these analyses, Wmatrix 4.0 (Rayson 2008) was used to compare the tagged corpus with the BNC Sampler Written Informal. The tool’s built-in semantic tagging and keyness functions allowed for the identification of items and domains that were statistically prominent in my corpus. This offered a useful point of triangulation between researcher-defined conceptual categories and automatically assigned semantic labels, helping to assess the distinctiveness of the corpus in terms of content and thematic focus. Finally, the corpus query processor CQPweb (Hardie 2012) was used to search the Spoken BNC2014 for comparable instances of negative self-identification, employing the simple query syntax provided in the tool’s documentation. This served as a point of external comparison, enabling initial observations about potential differences between written, forum-based usage and informal spoken discourse. Figure 6.1: Concordance list of product preference disclaimers (R1 sort; R1 = first word to the right of the node). 154 6 The Empirical Study: Analytical Approach and Annotation <?page no="155"?> Figure 6.2: Contrastive conjunctions (xcj: con) following expert disclaimers (<nsi-EX>). In some analyses, tagged concordance lines were exported into Excel and structured to cross-reference formal co-textual markers with syntactic roles in the clause. Table 6.7 shows an excerpt where additive conjunctions occurring after NSIs, tagged as <xcj: +>, are aligned with subject-verb-object sequences. Such spreadsheets allowed for systematic identification of verb types and participant roles in NSI-associated clauses, supporting a functional analysis grounded in transitivity theory. NSI Co-text tag Subject Verb Object I have never been a smoker <xcj: +> I be regular swimmer I have never been an em‐ ployee <xcj: +> I be privy to such infor‐ mation I am no expert <xcj: +> I be happy to stand cor‐ rected I’m no expert <xcj: +> I be well aware of that I am not a very good me‐ chanic <xcj: +> I not have Mig/ tig welder I am not a programmer <xcj: +> I have no luck I am no spring chicken <xcj: +> I have history with Meth I am not a heavy guy <xcj: +> I lost 3 kg 6.4 Querying and analyzing the annotated corpus 155 <?page no="156"?> NSI Co-text tag Subject Verb Object I’m not a SKY subscriber <xcj: +> I own SKY box I am no expert at this <xcj: +> I own my Pi I have never been a TGS kind of rider <xcj: +> I own a host of 26 inch wheel street Table 6.7: Excel matrix of negative self-identifiers with additive conjunctions (<xcj: +>). In this chapter, I have presented a framework for analyzing NSIs in relation to their clause-internal, immediate, and more distant clause-external co-text. I argued that to be able to establish how often particular conceptual categories of the structure occur in co-texts with particular meanings and functions, it is necessary to define (a) conceptual categories of identifying NPs, (b) categories of clauses and sentences immediately preceding or following instances of the target structure and formal relations between the structure and these categories and (c) categories of co-text above sentence level which can be relevant for the functions of NSIs. Based on the framework defining these formal and functional categories of co-text, I presented an analytical model for investigating NSIs to answer the research questions addressed in this study. I then showed how I marked said categories of co-text and conceptual categories of identifying NPs in the corpus, and how I annotated the corpus for metatextual information about the data. Finally, I discussed how I used the programs WordSmith 5.0 and Wmatrix 4.0 as well as Excel to quantitatively analyze the corpus of NSIs in their co-texts. Chapter 7 now turns to the conceptual profile of identifying NPs, providing the semantic annotation framework for addressing RQ1. 156 6 The Empirical Study: Analytical Approach and Annotation <?page no="157"?> 7 Conceptually Profiling Negative Self-Identifiers This chapter focuses on the semantics of the identifying NP itself, establishing the conceptual foundation for subsequent functional analyses (Chapters 8 and 9). Whereas Chapter 6 outlined the formal-functional framework, here I develop a conceptual profile of identifying NPs. This addresses RQ1, repeated below: A. With which nouns and NPs do forum users negatively identify? B. To which conceptual categories can these nouns and NPs be assigned, and how prominently (in frequency and lexical variation) are these categories represented? The analysis proceeds in two stages: first, identifying and categorizing the NPs in the corpus; second, using these categories as a basis for examining patterned functional relations in later chapters. Section 7.1 discusses how conceptual profiling is defined in this study, what intricacies of creating such a profile for NPs needed to be considered prior to analysis, and according to which criteria identifying NPs were categorized. Section 7.2 presents the results of the analysis, and Section 7.3 critically reflects on them, formulating expectations about conceptualizations assumed to potentially underlie the functions of NSIs. 7.1 Theoretical considerations and methodological principles The first step of analyzing the functions served by particular NSIs in particular co-texts across interactional situations on discussion forums is to learn about the meanings they express and the forms used to do so. Therefore, the first analysis of the corpus of instances of the structure carried out was to create a conceptual profile of identifying NPs in instances of the structure “I + copula + NOT + identifying NP” (i.e., the full phrase within the NSI, with the lexeme as its lexical core). Recurring pragmatic functions of NSIs from particular conceptual domains, when found in co-texts with specific functions across forum discussions, can offer insights into how speakers routinely represent themselves linguistically in written online interactions. In other words, particular variants of NSIs may serve as strategies for discourse management - and their use in these settings may have implications that extend beyond the immediate situational context. In this way, this study investigates micro-pragmatic functions to learn about <?page no="158"?> conceptualizations about the social world informing the way people interact linguistically on the web, despite not analyzing a particular discourse (in the Faircloughian sense of a particular representation of social life). Regarding the methodological approach toward creating a conceptual profile of identifying NPs, Collins (2019: 9) explains that analyzing corpus data with concordance software “allows us to [not only] read the data both ‘horizontally’, to recover some of the context in which it is used; but also ‘vertically’, to elicit a sense of the patterns of how the term is used across texts”. Creating a conceptual profile means first defining one particular linguistic element or structure (here, every NP contained in the 936 instances of NSIs), the lexical realizations of which are retrieved from the corpus in a vertical format (Marko 2015b: 69). These are then qualitatively examined and assigned to conceptual domains according to particular categorization principles, without (at least initially) taking their co-texts into consideration. In the following, I refer to each unique lexical item or construction as a type (abbreviated as Ty. in tables), and to every instance of its occurrence in the corpus as a token (To.). To explore how particular conceptual categories of NSIs relate to particular meanings expressed by their immediate co-texts, these co-texts were function‐ ally categorized in the analyses presented in Chapters 8 and 9. This means that linguistic elements in the co-text of the structure, retrieved from the corpus in a concordance line format, were analyzed both ‘vertically’ and ‘horizontally’. As for vertical analysis, these co-texts represent the contents of particular structurally defined categories annotated in the data (namely phrases preor postmodifying the structure clause-internally or clauses with particular formal-functional relations to the structure). After being retrieved from the corpus, they were assigned to functional categories (e.g., as processes involving particular participants). As for horizontal analysis, NSIs and their immediate co-texts were analyzed in terms of their functional interaction and their relations with linguistic elements in their wider co-text (Rühlemann & Clancy 2018: 6). To account for instances of the structure in focus as lexicogrammatical choices embedded in their “interactional and situational contexts of occurrence” (Rühlemann & Clancy 2018: 7) more comprehensively, the ‘narrow horizon’ of the analyses of NSIs in their immediate co-texts in concordance line format were supplemented by more detailed, qualitative analyses of examples of the structure. The discussions of these examples thus serve to expand the co-textual ‘horizons’ of the mainly corpus-based and thus vertically oriented study. Additionally, to demonstrate how the gap between microlinguistic analysis and questions on the level of the more macrosocietal context might be bridged, Section 9.3 discusses in detail two instances of NSIs found to be particularly 158 7 Conceptually Profiling Negative Self-Identifiers <?page no="159"?> prominent in the data, referring to their situational contexts of usage and taking aspects of online contexts into account. The practical task of creating a conceptual profile of identifying NPs in NSIs is by no means a straightforward endeavor: firstly, the concepts linguistically represented are fuzzy, and it is thus difficult to draw clear boundaries between them, which means that categorization needs to be principled and explicit about where these boundaries are set. Secondly, meaning is perspectival, which means that precisely which lexemes we choose to refer to a concept, foregrounding different aspects of it, depends on and may reflect the textual, situational and cultural context of using language. And, vice versa, what superordinate concepts we consider particular expressions to belong to is also influenced by our conceptualizations of the world. For example, whether I consider the noun gamer as a profession or as an identity defined by a preference for a leisure activity probably depends on my attitude toward gaming. The rest of the content of the profiled paradigm, too, has an impact on categorization. I assume that the likelihood to assign a particular lexeme to a semantic category is greater the more salient that category is perceived to be in relation to the entire discourse under scrutiny. Take, for instance, the corpus examined here contains numerous references to various professions, which is why the noun doctor was primarily categorized as instantiation of the concept of professions and only in a second step subcategorized as medical profession. In a hypothetical corpus with hardly any references to professions, but many nouns from the semantic field of medicine, we might simply add the noun doctor to this category, foregrounding a different meaning aspect for the noun at hand. To account for the fuzziness of semantic categories while still enabling prin‐ cipled boundaries between them for the sake of systematicity and quantifiability, I developed a framework for classifying NPs based on formal and semantic features that transparently reflects the categorization process. The framework was derived through exploratory data analysis, which revealed that certain meanings were realized by particular lexemes significantly more frequently than others. Taking a prototype-theory-informed approach (Taylor 1995), which posits that conceptual categories have a center - the prototypical instance that best fulfills the category’s defining features - and more peripheral members that share fewer of these features, I devised a classification model to distinguish between relatively clear and fuzzier categories of NPs. Table 7.1 presents the resulting set of classification categories (based on my own interpretive framework), aimed at systematizing the process of identifying NPs. These categories consider the formal structure of the NP and vary in the degree of objectivity they afford. Table 7.2 provides an overview of how the data 7.1 Theoretical considerations and methodological principles 159 <?page no="160"?> were systematically categorized, using attested corpus examples to illustrate the approach. Reason for categoriza‐ tion Definition Example Category-de‐ fining lexeme Nouns (single or head in a nomi‐ nal compound/ phrasal construc‐ tion); instantiated considerably more frequently than other lex‐ emes; designates a superordinate concept frequently referred to by other types. fan: defines category “preference (+ fan)”; frequently instantiated in numerous variants (as head in compounds, with postmodify‐ ing prepositional phrases); desig‐ nates the superordinate concept ‘fandom’/ preference, referred to by many identifying NPs (featur‐ ing fan and similar expressions like lover). Lexeme with categorizable meaning fea‐ ture Nouns (single or head in a nomi‐ nal compound/ phrasal construc‐ tion) with an unambiguously cat‐ egorizable meaning feature. racist: inherently negative → category “characteristics: evalua‐ tive”; lover/ lover of X: inherently positive → category “preferen‐ ces (fan)”. Generic nouns with categorizable specifiers ‘Empty’ or generic nouns preor postmodified by adjectives or other elements with categoriza‐ ble meaning. technological man; person for an HRM strap; bad person. Category-de‐ lineating specifiers Preand postmodifiers which specify a kind of or a more spe‐ cific subdomain of the superordi‐ nate concept to which the noun is assigned. expert at geology → “scientific expertise” as a subcategory of “expertise”. Category-in‐ tensifying specifiers Preand postmodifiers that do not change but intensify the meaning of the classifiable noun; constitute separate types, but not separate conceptual categories. big in a big fan of the IH emblem → same conceptual category as fan of the IH emblem. Analogy/ background knowledge Analogies and background knowledge that support the cat‐ egorization. techie: analogy with foodie → implies a passion for technology; background knowledge: techie implies identification with tech‐ nology and (unprofessional) ex‐ pertise. Table 7.1: Criteria for conceptual classification of nouns and noun phrases. 160 7 Conceptually Profiling Negative Self-Identifiers <?page no="161"?> Exam‐ ple phrase Premo‐ difier/ Speci‐ fier Noun (Head/ Com‐ pound) Post‐ modifier/ Specifier Cate‐ gory Subcate‐ gory Reason for categoriza‐ tion racist — racist — Charac‐ teristics Evaluative inherently negative cow — cow — Charac‐ teristics Evaluative (meta‐ phorical) metaphorical, pejorative member — member — Roles Member‐ ship denotes group relationship smoker — smoker — Habit Substance habitual action user of delay — user of delay Usage — functional use fan of yellow — fan of yellow Prefer‐ ence Visual color preference dog lover dog lover — Prefer‐ ence Nature preference for animals runner — runner — Activity Expertise habitual activ‐ ity techie — techie — Exper‐ tise (expert) tech-related, casual expertise astro‐ nomical type astro‐ nomical type — Exper‐ tise (expert): domainspecific “X type (of per‐ son)” pattern; premodifier signals domain (e.g., scientific, botanical) huge one for com‐ posting huge one for com‐ posting Exper‐ tise (expert): activity/ domain “one for X” pat‐ tern; evaluates person’s com‐ petence/ incli‐ nation in a do‐ main (e.g., composting, painting) legal expert legal expert — Exper‐ tise Legal field-specific expertise lawyer — lawyer — Profes‐ sions — professional role Table 7.2: Overview of categorization process with examples. 7.1 Theoretical considerations and methodological principles 161 <?page no="162"?> Section 7.2 presents the results of this categorization, highlighting the most frequent categories and their internal variation. 7.2 Results The superordinate conceptual categories were established and annotated in the corpus (see Figure 7.1 for an overview with token frequencies). In the following, I discuss each category and the frequencies of NSIs representing them, thereby addressing parts A and B of RQ1. Section 7.3 then presents the overall results, beginning with a conceptual landscape. This landscape illustrates how the categories, discussed separately in this section, are conceptually related. It highlights their fuzziness and identifies ‘peripheral areas’ to show that, in light of the data examined, identifying NPs - and perhaps linguistic identity representation more broadly - are best approached as a continuum, with certain conceptualizations standing out as especially salient rather than as rigidly separable categories. The section concludes with a discussion of the implications of these results. 0 50 100 150 200 250 Conceptual Categories of Identfying NPs Figure 7.1: Overview of conceptual categories of identifying noun phrases. 162 7 Conceptually Profiling Negative Self-Identifiers <?page no="163"?> 7.2.1 Preference The overarching conceptualization of preference is considered in terms of two categories in this profile, viz. constructions with fan and constructions with various head nouns expressing a preference for something. Preference (+ fan) This category contains all identifying NPs featuring the noun fan either on its own, as head in nominal compounds or together with preand postmodifiers. It is internally differentiated according to category-delineating specifiers as described above, that is, according to preor postmodifiers specifying the kind of ‘fandom’. Items were grouped by the head noun’s modifier (pre/ post) to identify subdomains of fandom (products, persons, activities, etc.). Category To. Ty. Items General 15 4 fan (11), huge fan (2), great fan, fanboy Anaphoric reference 4 3 big fan of these, fan of those (2), huge fan of Products/ objects 68 68 amazon fanboy, B/ O fan, big Lambo fan, BMW fan, CS fan, huge 911 fan, O/ A fan, big e21 fan big fan of: cab sims, foreign vehicles, head phones, Hornby decoders, military aircraft, Phase scanners, show different amps, tablets, tele neck pickups, the P20, tiny little satellite speakers, xtc fan of: 154CM, adapters, bulky bezels, butly based inha‐ lants, cradles, crimps, die nuts, diesels, drop bars, ear buds, ebay, ESLA street lights, exercise bikes, Fastpass Plus, gta, Hills, ‘home cinema’, Home Max speaker, lg, Lightroom, linn speakers, mini keys, Model T hot rods, most of the kits being released this season, Panaracer tyres, privacy glass, Run Flats, Serif, sluggish sd cards, synthetics, the AXT-11 template, the Cabrinha bar, the high-heel flare jean, the Kiev class, the old Astra, the ‘relative display’, the videogame, top handle straps; great fan of: Bosch plugs, the first edition of Descent huge fan of: after market lights, RWI, the V10, these elastic straps, Var; massive fan of: woody cabs Visual aspects 24 24 big fan of: blue flowers, magenta, the IH emblem, the opacity, yellow; fan of: busy look, if [it] being silver, modulation effects, rose, that exhaust tip on the black car, the Alu-stars, the bright non black everywhere, the fake-flowers-that-should-look-completely real, the green ones [trays], the kit, the look, the mac-style icon, the Neu‐ roshima theme, the new BMW shape, the white patches; 7.2 Results 163 <?page no="164"?> Category To. Ty. Items huge fan of: the colour; massive fan of: the cream; particular fan of: narrow gauge modelling Persons/ clubs 27 27 big: Harry Potter fan, Heyman fan, Prentiss fan, Chelsea fan huge: Bryan fan, Revolver fan, James Bond fan, Leicester fan big fan of: him fan of: his, Blake, Pitman, Cameron, Chelsea, City, Gel‐ dof, Harnoncourt, Hopkins, Mr Craig, Parelli, PJ’s, Queen, shakira, the tenor’s, theirs, them, Tim Oxbrow’s voice Food/ drink/ substances 14 14 big fan of: creatine, garlic, supplements, whiskey; fan of: brown chocolate, cooked green vegetables, either HP Sauce or Guiness, fruit, Gewurtztraminer, greasy food, lucozade, nicotine, pizza, Walkers SnV Activity 10 9 fan of: going out much (2), air ride, chasing the dragon, exploration, full body workout, going on holidays, not celebrating against old clubs great fan of: skiing, thrashing about in wet snow Activity aspect 7 7 big fan of: poke hacks; fan of: artificial guidance, farming credits in RES sites, having bottles behind the saddle, paying to select a cabin, strikerless formations, the mess of of wires Nature/ animals 7 7 fan of: mice, moths, Slow worms or snakes, swans, the waterlily, variegated palms; great fan of: gulls IT-related 7 7 big fan of: file track; fan of: cgv, clean blend on overdrive, downloading, fixed track, the Instacrew concept huge fan of: the new menus to select vehicles from Ideological 5 5 fan of: ranks, the fashion industry, these ‘5 year plan’ type of things, violence for the sake of violence great fan of: this idea Business-rela‐ ted 3 3 fan of: partnerships, the Scandinavian market great fan of: guaranteed stops Entertainment 3 3 fan of: “So Broken”, Nemesis Sub-Terra, stand-up per‐ formances Body-related practice 3 3 big fan of: fasting fan of: shaving, the steam method Events 2 2 great fan of: my birthday huge fan of: April Fool’s day Total 198 185 - Table 7.3: Noun phrases in the category “preference” (+ fan). 164 7 Conceptually Profiling Negative Self-Identifiers <?page no="165"?> As can be seen from Table 7.3, the superordinate category of “preference” can be internally differentiated into many thematic subcategories, the biggest of which is represented by the subcategory of “products/ objects”. This category includes instances of NSIs postmodified by prepositional phrases specifying a wide range of objects and specific products - from adapters over ear buds of any kind and synthetics to textbooks - as well as visual and other aspects of objects and products - like the mac-style icon or the new BMW shape - as the objects of (non-) fandom. Cases in which the objects of fandom are particular persons/ clubs, like Harry Potter, Geldof or Chelsea, are much rarer, with only 28 tokens as compared to an overall 92 used to refer to aspects of particular objects and products. Regarding the use of NSIs, contrasting speakers with preferences for partic‐ ular products, objects or their visual appearance, the nouns and NPs assigned to these categories mostly refer to highly specific (non-)preferences. An example is 7.1 below, where the NSI serves to evaluate a picture of a kit, construing the speakers’ identity in relation to a category directly - and probably exclusively - relevant in the immediate interactional context. 7.1. A: Here is my effort. Inspired by the Adidas/ Brother design from earlier in the thread. [PICTURE] - B: Great third kit, [NAME] should nick the design. I’m not a fan of the white patches on the shoulders in the home and I’m not sure if I like the white on the away kit but the design is pretty good. There are some absolute crackers on here. I might get another competition going if I can be bothered. At the same time, negatively identifying as a fan of very specific aspects of par‐ ticular things being discussed can serve to represent speakers as knowledgeable in these fields of interest and, thus, index more permanent aspects of their identity. This becomes obvious when scrutinizing such instances of NSIs in more detail, taking more of their co-texts and aspects of the situational context into account (see Chapter 9, where NSIs are functionally analyzed). Consider, for example, 7.2. Here, the NSI locally modifies the interpretation of the speaker’s account of being “impressed” by a certain type of laser scanner, highlighting that the model referred to has exceeded the speaker’s generally low opinion of the scanners in question. Being or not being a fan of Phase scanners is a very specific preference which requires awareness of very detailed aspects of this product category. By negatively identifying as such, before providing a detailed product description, the speaker indexes that they have strong opinions when it comes to scanners. 7.2 Results 165 <?page no="166"?> 7.2. As some of you may know I have not always been a big fan of Phase scanners but things change.Z+F demonstrated the 5010 to us recently and I must say that I was really, really impressed. Its a lot smaller than previous models but has a large on board screen which is easy to use. The data looked a lot cleaner than I have seen before with phase scans and the fact that it can scan at similar ranges to TOF is impressive. And I think I am correct in saying that it now has a level compensator.I look forward to seeing more from this scanner Besides negative identification with particular products or product features, constructions with fan are used to contrast speakers with preferences for particular food, drinks and substances (e.g., cooked green vegetables or nico‐ tine), but also to negatively identify as fans of particular activities like going out much or aspects of activities like having bottles behind the saddle (when riding a bike) or strikerless formations (in football). Speakers also negatively identify via their non-preferences for particular aspects relating to work on the computer; for example, they state not being fans of downloading or file track. The fact that being or not being a fan of something is also used to discuss all kinds of natural phenomena (from the waterlily to gulls), ideological positions (on questions like violence for the sake of violence or the fashion industry) and certain practices related to bodily appearance and well-being (e.g., shaving) suggests that negative identification in this case is routinely used to express viewpoints on a wide range of topics. Judging from the analysis of this data, liking or even admiring - as the noun fan implies - particular products, people, food and activities seems to be an important aspect of discursive self-representation in the examined data. At first glance this may seem mundane - after all, why not talk about what you like and dislike - but the prominence of NSIs of this kind does indeed seem noteworthy when considered in relation to other categories. For instance, only 16 NSIs altogether were used to contrast speakers with ideological categories, which could be taken to imply that mundane, situationally relevant topics play a bigger role for self-representation through negative self-identification on web forums than the ‘big’ questions of our time. This could partly be due to the medium of web forums as platforms for discussing very specific “interests shared by a group of geographically dispersed participants” (Burnett 2000: n.p.). At the same time, self-representation in terms of non-preferences implies awareness of available options, and the discursive manifestation of this awareness in people’s self-representation might, indeed, be seen as reflecting major trajectories of contemporary society, namely those of individualization and consumerism (discussed in Section 2.2). Put bluntly, if speakers did not have the choice between gloss, cream, magenta and yellow things (of a 166 7 Conceptually Profiling Negative Self-Identifiers <?page no="167"?> particular kind) or between the IH emblem, the Mac style icon, the Neuroshima theme and the one with the Alu Stars, and if these choices did not serve as signifiers of identity, negatively identifying with these aspects might not be so relevant. That negative identification with preferences for products, particular looks and activities plays an important role in the data analyzed for this study also shows when considering the conceptual category of preferences containing constructions without fan. Preference (fan) This category features all nouns and NPs that have one meaning feature categorizable as referring to a preference for something or someone, but without containing the noun fan. Table 7.4 illustrates NPs in the category “preference” (fan), grouped by semantic domain. Category To. Ty. Items Product 13 14 avid collector of TP, Dore enthusiast, great believer in tablets, hater of CGI as a whole, linux person, Mac guy, Mac person, ML stooge, person for a HRM strap, petrolhead, slave to period, speaker cables believer, TF evangelist Style 7 7 big ‘dress’ person, dress person, makeup kind of girl, particularly ‘pink’ person, pink person, shoes girl, custom dress shirt kind of person Ideas/ ideology 5 5 advocate of couples separating, believer in the one hat fits all solution, big believer in patterns and stuff, Labour lover, slavish adherent to their politics Sexual 3 3 masturbator, thong man, tit man Body practicerelated 3 3 advocate of high doses, lover of taking laxatives, serial doctors apt person Persons/ clubs 3 3 Hodgson basher, supporter of Jim Price, Radiohead hater Activity aspect 3 3 great lover of positions 2/ 4, lover of the Beagle Point systems, H/ C snob IT-related 2 2 piping guy, database guy Food 2 2 big chocolate lover, cream lover Activity 1 1 UBER audiophile 7.2 Results 167 <?page no="168"?> Category To. Ty. Items Visual aspects 1 1 lover of exactly copied portraits from photos Nature/ animals 1 1 dog lover Total 44 44 - Table 7.4: Noun phrases in the category “preference” (fan). Regarding the objects of non-preference, again, the most prominently represen‐ ted subcategory is that of products (e.g., people negatively identify as having a preference for Mac, Dore and speaker cables). People also negatively identify in relation to stylistic preferences (e.g., as a pink person or as a custom dress kind of person). Another category of preference disclaimers in this category contains identifying nouns referring to preferences for particular ideas and ways of thinking about the social world. The category “preferences” further features nouns and NPs referring to preferences in various spheres of life, from sexuality (being a thong or a tit man) to body-related practices (medication and treatment such as fan of taking laxatives), certain aspects of particular activities (playing certain positions on the guitar or having particular golf or computer game preferences) and preferences relating to food or animals. As can be seen, this category is lexically and semantically less homogenous than the previous one, which is due to the fact that it is not based on one particular lexeme but on meaning relations between the head nouns, which were considered sufficient to assign these indefinite NPs to the same subordinate conceptual category. Looking at this category in terms of the meanings of the head nouns of its members (see Table 7.5 below) reveals that a range of constructions are regularly used to express (non-) preference. For example, speakers routinely negatively identify in terms of affective categories, e.g., as lovers, haters, and enthusiasts. Also frequent are constructions where the object of (non-) preference is combined with a general head noun such as person, guy, girl or man and head nouns expressing belief in or support for a person or idea, such as believer, supporter or even slave. What appears noteworthy about these head nouns is that they are flexibly used to express negative identification with all kinds of things, people and ideas - from brands, to food, and to particular people. This indicates, again, that NSIs perform functions specific to the particular communicative situation in which they are used: depending on the topic discussed, it seems, I can be a chocolate or labour lover, a make-up girl or a database guy. Judging from this - admittedly small - data set, having affection for something or not is an important aspect of identification, foregrounding awareness of choice and individual agency. 168 7 Conceptually Profiling Negative Self-Identifiers <?page no="169"?> Affective To. General To. Belief & support To. Evaluative To. lover - 8 person 9 believer 5 snob 1 hater - 2 guy 3 advocate 2 stooge 1 enthusiast - 1 girl 2 adherent 1 - - -head - 1 man 2 evangelist 1 - - -phile - 1 - - slave 1 - - -or/ -er - 1 - - supporter 1 - - - - - - - basher 1 - - Total - 14 - 16 - 12 - 2 Table 7.5: Semantic categorization of preference (fan) disclaimers. Considering the categories “preference (+ fan)” and “preference (fan)” together, the following conceptual domains are most frequently referred to in negative identification with preferences: Figure 7.2: Conceptual categories of preference disclaimers (all lexical variants). As can be seen, speakers most frequently negatively identify with preferences for particular objects, brands and visual aspects of things they have or buy, 7.2 Results 169 <?page no="170"?> 24 The expression carb nazi is reproduced verbatim from forum data. Its inclusion here reflects its occurrence as an attested self-identifier. The term is derogatory, and its use is not endorsed. followed by particular people, activities and consumption preferences. This suggests that what could be subsumed under the umbrella term of lifestyle preferences - what you buy, who you ‘follow’, what activities you engage in and what you eat and drink - is what speakers in my corpus contrast themselves with most frequently. This indicates that, as already explained earlier, conscious (mainly consumption-related) choice - expressed in terms of non-preferences - is an important aspect of self-representation in the examined data. 7.2.2 Habits A category which is closely related to “preference” is “habits”, the difference between the two categories being that the latter contains nouns describing what the speakers routinely do rather than what they prefer - so for instance, being an evangelical low-carber means that I strictly follow a low-carb diet, which is a routine I have chosen rather than a preference. I could be an evangelical low carber but still a fan of pizza; likewise, I could be a fan of the new Bmw shape (a preference according to my conceptual framework) but still not identify as a brand follower (a habit in my classification). Predominantly, speakers in my corpus negatively identify with consumption (or substance (ab)use) habits, e.g., as a coffee drinker, smoker or pill-popper. Table 7.6 lists noun phrases categorized as “Habits”, grouped by semantic domain. Category To. Ty. Items Food/ drink 15 14 big eater, big meat/ cheese eater, binge drinker, breakfast eater, carb nazi  24 , coffee drinker, evangelical low-carber, guy to take nutrition potions, heavy drinker (2), hot drink person, low-carber, morning eater, normal eater, vegan Substance 15 9 drinker or smoker, drug user, very big drug user, drug user outside of mj, heavy smoker, intravenous drug user, pill-popper, regular smoker, smoker (7) Routine activities 8 6 early morning person, forum person, regular corrie watcher, big gamer, huge bar guy, morning person (3) Product-related 2 2 brand follower, n+1 person Total 40 31 — Table 7.6: Noun phrases in the category “habits”. 170 7 Conceptually Profiling Negative Self-Identifiers <?page no="171"?> 7.2.3 Expertise and Professionalism This section discusses the superordinate category “expertise and professional‐ ism”, which is subdivided into three analytically distinct groups. One is based on the frequently appearing head noun expert; another features lexical elements assignable to the concept of (non-)expertise without containing the noun expert; and the third category contains nouns referring to what would conventionally be considered professions/ job titles. Expertise (+ expert) The second biggest lexeme-based conceptual category is constituted by (variants of) the noun expert, which frequently appears on its own (105 tokens), some‐ times refers back to something mentioned before (18 instances, categorized as anaphoric reference in Table 7.7 below) or occurs together with postmodifying prepositional (and, rarely, other phrases) specifying the field of expertise (40 tokens altogether). Category To. Ty. Items General 105 3 expert (103), full expert, real expert Anaphoric reference 18 12 expert: at these things, at this, in these matters, in this area, in this field (2), on such matters, on the matter, on the subject, on these points, on this (6), on this sort of thing Premodified by adjective Legal 2 1 legal expert (2) Business/ services 2 2 financial expert, postal expert Medical 1 1 medical expert Postmodified by prepositional phrase Nature 11 11 expert: for cetaceans, in fish, in mammals, on bees, on birds, on feline anatomy, on hymenopterans, on ID, on ticks, on their behavior, on these type of creature Technical 7 7 expert: at modelling, on Indian signalling, on kero‐ sene, on metal detecting, on oil, on small horticultural engines, on the various types/ standards of gas cylin‐ ders/ fitting Products 5 5 expert: in Normandy maps, in Nike’s ethics, on Aus‐ trian military uniforms, on shoes, on the dot product 7.2 Results 171 <?page no="172"?> 25 This expression strictly speaking does not fit into any of the categories presented. For the sake of exhaustive categorization of data, though, I am considering expert as head of economist~game theory expert~doctorate here. Category To. Ty. Items IT/ gaming 3 3 expert: at drivers/ optimization, in this verification lark, with Meshlab Arts/ Sports 4 4 expert: in training techniques, on ski jumping, on the 2 step (dancing style), on the popularity of this music Medical 2 2 expert: on HRT, on slimming world Science 1 1 expert: at geology Business/ services 1 1 expert: on house prices Ideological 1 1 expert: on religious matters Leisure 1 1 expert: at this game Linguistic 1 1 expert: at pronouncing things Postmodified by other phrases IT/ Gaming 2 2 expert: when it comes to inserting kits into the game, when it comes to configs Nature 1 1 expert: to tell you what bird it’s from Head in nominal compounds IT/ gaming 8 6 [linux (2), PHP, programming (2), server class CPU, computer game, economist~game theory expert~doctor‐ ate  25 ] expert Technical 6 6 [audio, asbestos, Bluetooth, electronics, vehicle electron‐ ics, wood] expert Product 5 5 [DICOM, Dennis (type C colours), jean, military truck, amplifier] expert Nature 5 4 [conformation, moth, shark (2), wood] expert Total 193 81 - Table 7.7: Noun phrases in the category “expertise” (+ expert). Unspecific expert tokens (no field given) outnumber specific ones. This suggests that web forum users routinely negatively identify as experts without specifying the precise domain in which they lack expertise; in the later functional analysis, I 172 7 Conceptually Profiling Negative Self-Identifiers <?page no="173"?> treat these cases as epistemic disclaimers rather than domain labels. One reason could be that speakers assume their interlocutors would not understand a spec‐ ification of the expertise. Another is that they themselves may lack specialist terms to be more specific. Thus, their use of expertise disclaimers could be seen as reflecting their (potentially implicit) awareness that they are exchanging lay expertise on these forums. In any case, the high number of unspecific disclaimers may suggest that they function more as discourse markers than as truly informative statements. In example 7.3 below, an unspecific disclaimer of expertise is used in an exchange the purpose of which is the identification of the source of “weird [looking]poo”. An NSI is used by speaker B after providing their ‘diagnosis’ upon seeing the picture: 7.3. A: weird poo? ! dogs found this the other day, like poo but maybe not? ! [PICTURE] it was like all hairy, cheers, [NAME]. - B: Hi [NAME]. Looks like fur from a cat! They sometimes bring up what are loosely called ‘fur balls’ and they can look a bit like this. Or some other animal possibly. I’m no expert and may be well off the mark [NAME]! Another example of an NSI epistemically postmodifying assessments and diagnoses is 7.4 below. Just like in the above example, where the NSI is coordinated with a statement commenting on the degree of certainty of the previous utterances (and might well be off the mark), the co-text of the NSI in 7.4 works to reduce the certainty of the speaker’s identification of the beetle in question: 7.4. [PICTURE] Definetly [Definitely] a deathwathch [deathwatch] beetle. if they are you might hear them tapping their heads against the wall. I’m no expert so I might be wrong. Again, in this example, the speaker acknowledges their lay status and the possibility of being mistaken in their identification of the animal in the picture as a “deathwatch beetle” using a modal verb (so I might be wrong); the conditional clause following their assertion also serves to mitigate their claim. Interestingly, though, the speaker at the same time uses the adverb definitely when providing their identification, thus in a way undermining the acknowledgement of their lay status. In the next chapter, I examine whether NSIs of this type are a salient linguistic choice in similar co-texts, which would indicate that they function like 7.2 Results 173 <?page no="174"?> discourse markers, used by speakers to formally index their lay status, allowing them to act like experts while still coming across as modest. In addition to the most frequently used unspecific disclaimers of expertise, there are 41 instances of identifying NPs in the corpus in which the noun expert is postmodified by a prepositional (or other) phrase specifying the field of expertise and 24 nominal compounds with expert as head. A large number of constructions with expert are used to refer to very specific fields of knowledge that probably would not formally or traditionally be considered expert domains, but rather as specific interests people might have - such as Indian signaling, Austrian military uniforms, or metal detecting. Among the 41 postmodified uses of expert, the most frequent associations were with nature (17 tokens), IT/ gaming (16), technical domains (13), and specific products (10). Less common were references to arts/ sports (4), science (1), medicine (2), business/ services (3), and ideological, linguistic, or leisure domains (1 each). These figures point to a notable broadening of the concept of expertise. Speakers in my corpus often negatively identify as experts in the field of nature, mostly on subjects relating to particular species (such as feline anatomy, cetaceans, or sharks). There are also a high number of instances of negative identification with expertise in specific technical and IT subjects - from small horticultural engines, asbestos to programming and Linux- and with particular products (e.g., the dot product) and commodities (e.g., jeans). By choosing to negatively identify with nominal constructions featuring the noun expert to indicate their lack of expertise with certain things, phenom‐ ena, or activities, speakers linguistically represent - and thereby implicitly acknowledge - these as legitimate fields of expertise. This may reflect how the figure of the ever-more specialized expert has become increasingly important in online self-representation. It also suggests an orientation to the demands of the contemporary job market and to forms of self-presentation that align with it. In this context, the concept of employability as an individualized project is worth noting (Moreau & Leathwood 2007). On a blog providing advice on how to succeed professionally, for instance, it is stated that “the best way to make yourself extremely valuable in a team [… is to be] a master of something […] The goal is to become a trusted resource about a certain topic [… and] [t]herefore, a focus area doesn’t mean you can ONLY do that one thing, but simply that you are BEST at doing it” (Ucros 2018). This framing highlights how expertise is not necessarily understood as the result of formal education, but also as something acquired through customer experience. As discussed in Section 2.2.1.2, consumers are increasingly active participants, with online reviews exerting significant influence on buying 174 7 Conceptually Profiling Negative Self-Identifiers <?page no="175"?> 26 Mesembs: horticultural shorthand for Mesembryanthemum, a genus of succulent plants in the ice plant family (Aizoaceae), native mainly to southern Africa and commonly referred to as ice plants or living stones (Encyclopaedia Britannica n.d.). decisions. Similarly, knowledge gained through the use of particular products is constructed as a valid form of expertise - one that is frequently negotiated and displayed in online discourse (Hennig-Thurau et al. 2004). On the other hand, the fact that negative identifications take the form expert + preposition + field suggests that expert is used as a flexible, ad hoc disclaimer - easily adapted to whatever topic is under discussion - and in this sense speakers pragmatically appropriate expert constructions to serve their immediate communicative goals rather than uphold institutional definitions of expertise. Expertise (expert) Finally, beyond the noun expert, a range of other nouns and NPs are also used to refer to individuals with field-specific knowledge. These include 12 tokens referring to general levels of proficiency (from absolute beginner to professional) and 45 tokens - mostly preor postmodified - referring to individuals with specific forms of expertise. The distribution of these forms across structural categories and semantic domains is summarized in Table 7.8. Structural category Domain To. Ty. Items General expertise 12 11 - Noun/ Noun phrase - 12 11 absolute beginner, beginner, champion, consis‐ tent performer, novice, person who can advise you on the matter, pioneer (2), pro, professional, specialist or collector, specialist Specific expertise 45 42 - Noun Technical 2 1 techie (2) Adjective + noun Linguistic 3 3 eloquent wordsmith, particularly lyrical guy, great blogger Medical 3 2 medical person (2), medical professional Nature 2 2 big grower of mesembs,  26 good birdwatcher Legal 1 1 legal eagle Technical 2 2 technological man, expert builder 7.2 Results 175 <?page no="176"?> Structural category Domain To. Ty. Items House‐ work 1 1 very good cook Science 1 1 astronomical type Sports 1 1 expert runner Other 2 2 tactical guru, confident driver Subtotal — 18 15 — Nominal com‐ pounds IT 7 7 advanced IT person, bash guru, computer boff, IT guy, IT person, license guru, Revit master Technical 4 4 electronics guru, fire door specialist, qualified HV switching person, tech geek Business 1 1 VAT specialist Science 1 1 math wiz Subtotal — 13 13 - Noun + preposi‐ tional phrase Health 4 4 professional on OCD, stranger to drugs, stranger to how PD affects people, stranger to injecting IT 2 2 layman when it comes to these things (enable PS one emulator), noob to UAE4ALL House‐ work 2 2 huge one for composting, natural in the kitchen Nature 1 1 great one for birdsong Arts 1 1 authority on paint work Subtotal — 10 10 — Metony‐ mic — 4 4 Armstrong or Cavendish, Mo Farah, Nostrada‐ mus, Aladdin’s genii Total — 57 53 - Table 7.8: Noun phrases in the category “expertise” (expert). The fields of expertise with which speakers most frequently negatively identify are “IT” and “technical”. Other frequently invoked fields include nature and product-related expertise, which likely reflect areas where forum users com‐ monly seek or disclaim practical competence. By contrast, more institutionalized 176 7 Conceptually Profiling Negative Self-Identifiers <?page no="177"?> domains such as medicine, law, or science appeared only occasionally. This may indicate that few people feel qualified enough not to index their lay status in these contexts. It can also be assumed that there is uncertainty in several of the fields of expertise identified in Table 7.8. Turning from the distribution of domains to the lexical forms used, most of the head nouns in such disclaimers of expertise designate persons with a high level of proficiency, such as professional and specialist, or, more informally, boff, geek or wiz. As with NSIs of preference discussed earlier, some disclaimers of expertise are constructed by combining a general head noun such as person with a preor postmodifier providing the ‘specialist’ meaning component, as in advanced IT person or medical person. Conversely, the “specialism” can be carried by a pre-modifying evaluative adjective (e.g., very good), with the head noun specifying the field of expertise (e.g., driver). In one case, eloquent wordsmith, both adjective and head noun express proficiency. This wide variety of (seemingly spontaneous, informal, and even original) nominal constructions, created by speakers to contrast themselves with exper‐ tise in often highly specific fields, shows once again that speakers acknowledge the importance of the concept of expertise. At the same time, they appear to care little about finding ‘proper’ labels for the specialists they claim not to be, freely referring to all kinds of experts to suit their own communicative goals. References to a lack of expertise or experience through nouns like stranger (to something), noob or layman are comparatively rare. This indicates that the notion of the expert - in its different lexical realizations - is the more salient conceptualization in speakers’ linguistic negative self-representation. Perhaps this is because, when interacting on forums, belonging to the group of laypeople is the default and socially preferred option, while explicitly highlighting mem‐ bership in the group of non-experts might be perceived as immodest and thus marked. In this sense, speakers may use these variants of NSIs to mark their discourse as different from, but not inferior to, expert discourse. An example illustrating that acknowledging non-expertise or even poorly evaluating one’s own skills does not stop speakers in my corpus from performing speech acts that contradict the meaning of the NSI is: I am not a very good cook but do recommend experimenting. There are also figurative and metonymic cases, where speakers negatively identify with actual experts (e.g., I’m no Armstrong, I’m no Cavendish) or with figures standing for particular abilities (e.g., I’m no Nostradamus). In the following example, the speaker negatively identifies as Mo Farah, (perhaps self-)ironically acknowledging his lack of professionalism while at the same time emphasizing his own ambitions: Could you give me an 7.2 Results 177 <?page no="178"?> indication of what time the last finishers tend to cross the line in? I am no Mo Farah but I certainly don’t wish to finish last! ! ! Whereas Table 7.8 organizes these forms by structural pattern and semantic domain, Table 7.9 highlights the conceptual categories of head nouns and modifiers used to construct expertise-related self-identifications. Table 7.9 shows that alongside nouns denoting domain-specific expertise (e.g., guru, specialist, professional), the data also include general referents (e.g., person, guy), experience-based expressions (e.g., layman, noob, stranger), and evaluative constructions (e.g., great one, huge one). In addition, occasional figurative labels such as legal eagle illustrate how speakers creatively exploit metaphor to frame (non-)expertise. These forms suggest that negative self-identification with expertise is not limited to straightforward domain terms, but often involves stance-taking and self-positioning through evaluative, experiential, and figura‐ tive resources. Structural category Category To. Ty. Items Head noun - Expertise 15 10 guru (4), specialist (2), professional (2), master, geek, boff, authority, natural, wordsmith, wiz General 8 4 person (4), guy (2), type, man Experience/ inexperience 5 3 stranger (3), noob, layman Evaluative 2 2 huge one, great one Figurative 1 1 eagle Premodifier Evaluative 8 7 big, confident, eloquent, expert (2), good, great, very good Total 39 27 - Table 7.9: Profile of expertise-related head nouns and modifiers by structural and conceptual category. Professions One of the three biggest semantic categories differentiated in this conceptual profile contains single and compound nouns referring to particular professions. Most nouns in this category denote medical professions, with the noun doctor being the second most frequently occurring type after expert in the entire corpus 178 7 Conceptually Profiling Negative Self-Identifiers <?page no="179"?> (103 unmodified instances). The analysis revealed that unspecific disclaimers of expertise are most common in health forums. The fact that the relatively unspecific noun doctor occurs so frequently indicates that NSIs are often used in discussions on health issues, in somewhat ‘delicate’ contexts, where speakers might consider it particularly necessary to epistemically mitigate the impact of their utterances. Speakers in my corpus also frequently negatively identify with professions from the domain of IT (notably being or not being a programmer) and technical professions such as mechanic, engineer or builder. This indicates that medical and technology-related knowledge appears to have an important status for people on web forums (an observation already made when discussing identifying NPs in the conceptual domains of expertise). In contrast, negative identification with academic qualifications, e.g., as geologist or scientist, is comparatively rare. This could reflect that forums tend to be oriented toward discussing problems of everyday life - like building a carport or setting up a computer program - for which practical skills, rather than formal education, play a role. Table 7.10 presents the nouns assigned to the superordinate conceptual domain “professions” in their respective subcategories. Conceptual category To. Ty. Items Medical 40 13 cardiologist, doctor (23), dr., entomologist, geneticist, gp, medic (4), medical professional (2), neurologist, phar‐ macist, pharmacologist, physician (2), psychopharma‐ cologist IT 22 10 beta tester, coder (2), dba, dev, developer (2), faro developer, hard core programmer, html programmer, programmer (10), pt admin or developer, windows pro‐ grammer Technical 21 13 audio engineer, builder (2), car mechanic, chainsaw technician, electrical engineer, electrician, engineer (4), expert tig welder, mechanic (5), pilot, plumber, techni‐ cian, very good mechanic Arts/ sports 15 14 artist (2), cheerleader, designer, dj, motoring journalist, musician, photographer, very good photographer, pro‐ fessional dance instructor, professional footballer, ref, texture artist, university educated writer, writer Science 14 10 chemist, economist, geologist, historian, mathemati‐ cian, nuclear physicist, physicist, rocket scientist, scien‐ tist (5) 7.2 Results 179 <?page no="180"?> Conceptual category To. Ty. Items Business/ finance 11 9 accountant (3), bookseller, experienced investor, finan‐ cial advisor, consultant or pd, postman, salesman, super manager, supervisor Legal 6 2 lawyer (5), solicitor Education 8 6 du student, pshe [personal, social, health and economic education] teacher, qualified teacher, ‘second’, student, ta, teacher (2) Nature 4 4 botanist, gardener, zoologist, hymenopterist Other 2 2 fucking butler, fieldtester Total 142 33 — Table 7.10: Noun phrases in the category “professions”. The overall frequency of nouns from the conceptual domains of “expertise and professionalism” - whether denoting actual professions or constituting constructions with expert and nouns with similar meanings - suggests that knowledge is a key identifying concept in the examined corpus. The conceptual category of professions represents a pole on the continuum of conceptualiza‐ tions I have mentioned, as reference to formal qualifications can be considered the most prototypical - or at least conventional - form of identifying with expertise. The fact that the conceptual category “expertise”, which is based on occurrences of the noun expert with various specifiers, is actually more prominent in the corpus, however, could suggest that in the examined online contexts, (lay) expertise, rather than formal education, is a key concept in people’s self-representation. While contrasting themselves with experts, speak‐ ers also appear to quite confidently engage in the very practice of informally exchanging ‘non-expert’, unprofessional knowledge (as example 7.2 has shown). This could mean that there is a certain tension between what speakers do with language (discussing ‘expert topics’) and what they claim not to be. 7.2.4 Activities This category contains single and compound nouns representing nominaliza‐ tions of verbal processes, that is, nouns used to negatively identify speakers in relation to activities, for example, as a climber, runner, or builder. While these identifying NPs can imply expertise (or lack thereof), they cannot be 180 7 Conceptually Profiling Negative Self-Identifiers <?page no="181"?> straightforwardly assigned to the category of expertise: for example, negatively identifying as a World of Warcraft player primarily signals a choice not to engage in this game, which could reflect either a non-preference or a non-practice (hence non-expertise). This is why “activities” constitutes a fringe category (i.e., overlapping with other categories; see Figure 7.7 for placement in the conceptual landscape). Fringe categories are interesting in that they share features with more clearly definable ones - in this case, expertise, preferences, and habits - and can thus be conceptualized in relation to these reference categories. Together, robust and fringe categories form a continuum of conceptualizations defined by ‘poles’ of unproblematic reference categories - a continuum that coincides with and reveals salient conceptualizations in the corpus. Table 7.11 lists NPs in the fringe category “activities”, grouped by semantic domain, illustrating how such expressions relate to more robust categories like expertise, preferences, and habits. Category To. Ty. Items Sports 13 10 climber, fast runner (2), gym person, “hill walk”, hill‐ walker, runner (4), ultra distance runner, ultra runner, road rider, TGS kind of rider Gaming/ entertainment 6 6 big gambaler [gambler], collector, gamer, hard-core gamer, huge board game player, World of Warcraft player Arts 3 3 fast writer, piano player, fx guy Leisure 2 2 single traveller, happy camper Nature 1 1 “fossil hunter” Total 25 22 — Table 7.11: Noun phrases in the category “activities”. 7.2.5 Characteristics The conceptual category broadly labelled “characteristics” contains identifying NPs used to identify the speaker in evaluative terms, in relation to health and illness, beliefs and ideologies, in demographic or relational terms, and in relation to particular social, physical or other characteristics. As can be seen, speakers most frequently negatively identify with evaluative nouns and NPs, whereby it is possible to distinguish between literally evaluative nouns and NPs (e.g., creep, nasty person) and nouns and NPs used metaphorically (e.g., pig) or 7.2 Results 181 <?page no="182"?> metonymically (e.g., pain) or both (e.g., 12 year old text talking cunt), a distinction commonly discussed in metaphor studies (e.g., Goatly 1997; Barcelona 2003). As these examples show, the evaluative meaning is either provided by the head noun (as in 12 year old text talking cunt) or the premodifying adjective (as in nasty person). Table 7.12 lists NPs in the category “characteristics”, grouped by semantic domain, illustrating the range of evaluative, metaphorical, and metonymical identifications in the corpus. Category To. Ty. Items Evaluative/ derogatory (literal & figura‐ tive) 46 43 12 year old text talking cunt, animal, bad person (2), bad racist, clever man, complete monster, complete scutter monster, creep, danger to my step son, dick, drama queen, fucking product, fussy person, good boy, good girl, good host, hater, hero, hoolie or a fighter, idiot, lunatic overbearing mother, mug, nasty person, ‘overprotective’ mother, pain, pervert, pig, proper mum, prude (2), real stickler about monetary things, robot that someone can program into liking this or that, saint, scare monger, sexist, show off, thief (2), threat to my daughter, twat, violent person, wierdo [weirdo] Metaphorical (non-evaluative) 2 2 spring chicken, tortoise Health/ illness 22 20 addict, alcoholic, bedwetter, compulsive eater, diabetic (3), eczema sufferer, great sufferer, hypochondriac, insu‐ lin user, methamphetamine “addict”, newly diagnosed diabetic, pain patient, very patient patient, parky, pa‐ tient of ARCG [Assisted Reproduction and Gynaecol‐ ogy Centre], pwp [person with Parkinson’s], recover‐ ing addict, severe case, very strong asthmatic, victim of caffeine addiction Ideological/ reli‐ gious 16 14 buddist [Buddhist], catholic (3), communist, ‘defacto leader’, extremist, leaver, libdem, liberal, nationalist, practising christian, purist, rastafarian/ stoner/ hippie, religious person, trumper Relational/ demo‐ graphic 10 7 bridesmaid (2), descendant, descendant of fred archer, parent (3), “real” bridesmaid, schoolboy, youngster Physiological/ physical 10 9 7-12 or a 7-19 or any, easily hypnotised person, flexible person, good sleeper (2), great one for scars healing, great sleeper, heavy guy, sound sleeper, tall or stocky sort of person Social 9 9 “single-person”, face to face group person, ‘I told you so’ type of person, isolated person, networking person, particularly ‘weddingy’ person, ‘people person’, Single‐ ton 182 7 Conceptually Profiling Negative Self-Identifiers <?page no="183"?> 27 What adds to this problem is that the web forums examined here constitute by no means a homogenous discourse type, so they are hard to compare to other discourse types in the first place. Geographic 8 7 Aberdonian, EU resident, Glasto [Glastonbury] newbie, resident, town dweller, UK resident (2), Widnesian res‐ ident Linguistic 5 2 native speaker (3), native English speaker (2) Psychological 4 3 anxious person (2), masochist Temporary/ mood 3 2 happy bunny (2), happy chappie Gender-specific 3 3 ‘girly girl’, one of those girls who have always dreamed about their wedding, sir Sexual orienta‐ tion/ practice 1 1 pedophile Total 137 120 — Table 7.12: Noun phrases in the category “characteristics”. The frequency of references to evaluative expressions suggests that web forum users are careful to position themselves not only regarding the subject matter the respective forum is devoted to, but also as individuals with particular personality traits. The fact that the conceptual category mainly consists of negatively evaluating NPs could indicate that face management plays an important role in the context of web forums. In a discourse context where people who are unlikely ever to meet or have met each other in real life discuss quite personal issues such as health (Rueger et al. 2021), the need to provide information about one’s ‘real-life’ personality and to prevent potential threats to one’s own (positive) face by pre-emptively disclaiming potential negative implications of an utterance might be greater than in other situations (Mackiewicz 2010a; Seargeant & Tagg 2014). Of course, there is no comparative data to test whether the prevalence of evaluative personality disclaimers is higher in the context of the examined web forums than in other discourse contexts. 27 Contrary to what one might expect, identifying NPs defining speakers’ identities in relation to particular worldviews, beliefs or ideologies occur rather infrequently in comparison with nouns and NPs from other conceptual domains. This could have to do with the medium of investigation: web forums might just not be the sites for discussing political issues (in the widest sense), but are rather used for exchanging perspectives and knowledge on very specific fields 7.2 Results 183 <?page no="184"?> of interest. These constitute the raison d’être for the forum in the first place and thus create a sense of shared identity that is ‘disembedded’ from the participants’ normal social and political realities, in which their ideological position might play a more important role. In other words, web forums, as topic-defined sites of knowledge and opinion exchange, might not necessitate ideological positioning, as the shared interest the forum is devoted to is mostly apolitical, setting a discourse context in which being or not being, say, a liberal, does not really matter - after all, liberals and conservatives might be united by their common interest in motorbikes, or by suffering from the same health condition. 7.2.6 Roles Another relatively prominently represented conceptual category of identifying NPs I labelled “roles” because it defines the speaker in relation to particular sites or forums of social interaction. More specifically, this category contains nouns defining speakers via their role, status or behavior in relation to the forum on which they are interacting (e.g., as member or frequent user of a forum), their role in business and trade (e.g., as seller or Prime customer) or their affiliation to organizations and clubs (e.g., as member or active contributor). Table 7.13 lists NPs in the category “roles”, grouped by semantic domain and subcategory, illustrating how speakers define themselves in relation to particular sites or forums of social interaction, business contexts, and organizational affiliations. Category To. Ty. Items Forum-internal 46 38 - Membership 24 17 forum admin, member (8), member of any groups, mem‐ ber of any (forum), member of Seller Central, member of that shithole forum, member of vendor central, member on the forum, member on this site, moderator, OM, prime member (2), pro member, unknown person on Longevity.org, visitor, subscriber Usage 11 10 big poster on this site, forum guy, frequent user, frequent visitor, power user, regular contributor, regular in here, regular poster (2), regular user, very active member Behavior 4 4 big flasher, forum predator, saint on this forum, spam‐ mer Virtual identity 5 5 mobian, part of the elite universe, pirate, quiverfull, rich player Current interaction 2 2 good choice, help 184 7 Conceptually Profiling Negative Self-Identifiers <?page no="185"?> Business roles 31 25 Seller/ provider 25 20 personal seller, “darkroom under the stairs” merchant, approval lending service, ATEX SME, big commercial seller, big seller (2), BMVD seller, branded seller of this product, business owner, business (2), fraudulent seller, landlady, major supplier, massive company, merchant seller, new seller, pro-seller, seller (3), seller with Amazon (2), YouTube partner Buyer/ customer 4 4 big invester, client, Prime customer, victim of online fraud Staff 2 1 employee (2) Affiliation/ membership 6 3 active contributor to FF, member of the previous club, member (4) Total 83 66 - Table 7.13: Noun phrases in the category “roles”. As can be seen, the largest subcategory here is that of “forum-internal” roles. Speakers often negatively identify with their “membership”, for instance as members of particular forums or in relation to particular positions within the forum. Some forums differentiate between regular and prime or pro members, and it apparently makes a difference whether one makes a contribution as a visitor or subscriber. Speakers also negatively identify in relation to their “behavior”, for example as spammers, saints or predators, or with respect to the “current interaction”, as in cases where they identify as help or good choice (for advice). Finally, there are a few instances of negative identification with “virtual identity”, for example as mobian or as part of the elite universe. Negative iden‐ tification with “forum-internal” roles occurs relatively often, which indicates that speakers are aware of and orient themselves toward particular conventions of forum usage. For instance, example 7.5 from my corpus suggests that being a regular poster is associated with greater authority among forum users. 7.5. A: [NAME of C], I see you are fairly new to the site. I can recommend Dwight Van Driver as a completely 100% reliable poster. He is AWAYS [ALWAYS] right. Please take what he has to say as gospel, and he is the only poster here that I would say this about. - B: As a certain tennis player used to say….”You can’t be *serious*” He is not exactly wet behind the ears where this site is concerned : o) - A: OOPS! Just checked his profile, I see he registered 8th Nov alright, but in 2002! My humble apologies! 7.2 Results 185 <?page no="186"?> C: No offence taken - I’ve not been a regular poster since joining, so probably fair to say I’m a little damp behind the ears. Very grateful for all replies. In this exchange, speaker A states that another speaker, C, is new to the site, and goes on to recommend an experienced poster for reference. Upon speaker B’s correction, according to which C is, in fact, not wet behind the ears, C assures A that he is not offended by their wrong accusation, conceding that not being a regular poster probably means being damp behind the ears, that is, inexperienced. The high number of instances of negative identification with forum-specific roles in a corpus of web forum discussions implies, again, a relation between NSIs and discourse context: it seems that NSIs are mainly used to strategically modify the interpretation of speakers’ utterances within the parameters of the given communicative situation. At the same time, negatively identifying with forum-internally relevant categories by using the structure “I + copula + NOT + identifying NP” is, without even taking into consideration the linguistic context of use, a linguistic choice by which forum-internal roles and behavior are conceptualized as NPs and, thus, as entities (i.e., identity) rather than processes (i.e., activity). The fact that being considered a new forum user can be taken as an offense, as the example above suggests, might indicate that forum identities are also relevant for speakers’ ‘forum-external’, that is, more permanent, identity. Besides forum-internal roles, speakers often negatively identify in relation to roles one can assume in business relationships, namely as sellers or providers of particular services and products, as buyers or customers and as staff within a company. Scrutinizing the discourse contexts of some NSIs in this conceptual category suggests that these NSIs are more likely to be used to convey proposi‐ tional content, that is, the ideational function here seems to be in the foreground: the reason why speakers negatively identify, for example, as particular kinds of sellers in the situational context of a sellers’ forum is that they seek to provide more specific information about themselves to facilitate finding a solution to a problem together with the other forum members. The following example is a case in point: 7.6. As the thread title suggests… I’ve been a very long term Amazon seller, for over a decade. I had my selling privileges removed once before, two or three years ago, but they were rescinded after I spoke to someone there. This time it’s been very different. I received no notice or warning, but the privileges were suspended a week or so ago for late shipping. I know this is my own fault. I’m not a big commercial seller but a private individual who sells the odd item, and recently have been so busy with work that I shipped some items a day or two late and shipped others on time 186 7 Conceptually Profiling Negative Self-Identifiers <?page no="187"?> but forgot to click the ‘sent’ button on the site, so my performance notifications for shipping times suffered. In this example, the speaker negatively identifies as a big commercial seller on an Amazon sellers’ forum, pointing out more specifically that she is but a private individual. The function of the NSI here is thus to contrast the speaker not with the category of sellers as a whole, but with other members in the category of sellers, thus specifying what kind of seller she is. The reason why the speaker provides this information is to justify her imperfect shipping performance, that is, specification in this case mitigates the threat to the speaker’s own face represented by her self-criticism. Negatively identifying with big commercial sellers may also index aspects of the speaker’s identity beyond the immediate communicative situation, such as a more permanent ideological opposition to large corporations (the characterization of the items sold as odd, too, seems to undermine the idea that the speaker could be mistaken for a commercial business) - but this is just a speculative interpretation. 7.2.7 Usage/ consumption/ ownership As indicated before, conceptual profiling always involves devising criteria according to which lexical elements can be considered to fall into the same conceptual category. Depending on the chosen criteria, the resulting category is more or less homogenous and robust. Sometimes, as mentioned when discussing the category “activities” in Section 7.2.4, certain linguistic elements seem to occupy a middle ground between other, less problematic categories and do not fit well with any of them. Three small groups of identifying NPs that represent such cases are “usage”, “consumption” and “ownership”, which can be explained by reference to, but do not share sufficient characteristics with, members of the more prominently represented categories. As for the category “usage”, identifying NPs containing the lexeme user are considered to constitute a category in its own right because they do not seem to be assignable to any of the other conceptualizations found to be frequently expressed in the corpus: user does not, per se, imply expertise, because I could be a user of something without being proficient at using it. It is not quite the same as a leisure or sports activity in the sense of climber, either, because in contrast to climb, the nominalized verb use is transitive (and hence, you are a user of something). It can, but does not necessarily, imply a specific preference for something (I could say that I’m a smartphone user simply because this is standard practice today), and it can be interpreted as indicating a relation (i.e., 7.2 Results 187 <?page no="188"?> in the sense of energy suppliers versus energy users) like the nouns assigned to the category of “roles” in my profile. What unites the instances of user in this corpus is that they all specify products or applications as objects of usage, which makes the category conceptually similar to that of product preferences. The nominal compounds assigned to the category of “consumption” imply both preference and habit: being a SKY [TV] subscriber, for example, presupposes habitual use of SKY. Regular consumption can also suggest expertise, since long-term customers may become expert users (e.g., being an O2 customer can imply familiarity with O2 services); the same applies to “ownership”. Summing up, the category “usage/ consumption/ ownership” is deliberately conceptualized as a fringe category (see Figure 7.7 for placement in the conceptual landscape), overlapping with preferences, habitual practices, and expertise. Such fringe categories are analytically valuable, as they appear to reflect potentially shifting conceptualizations of identity. Table 7.14 lists NPs in this fringe category, grouped by semantic domain, to illustrate how such identifiers intersect with preferences, habits, and expertise. Category To. Ty. Items Usage 8 8 big user of Talking Point, big user of this, big user of wire, Maya user, serious user of the later versions of FTM, user of delay, user of Facebook or Twitter, Vibe user Consumption 7 7 LTSB customer, o2 [O2] customer, perfume buyer, Sky customer, SKY subscriber, TalkTalk customer, TT cus‐ tomer Ownership 2 2 owner of a BTS, owner of a house Total 17 17 — Table 7.14: Noun phrases in the category “usage/ consumption/ ownership”. 7.3 Summary and implications of results The aim of this chapter was to establish a conceptual profile of the nouns and NPs with which speakers in my corpus negatively identify. I described, classified and quantified identifying NPs occurring in my dataset of 936 NSIs used in web forums, differentiating them according to their ideational meanings and taking into consideration their formal appearance (i.e., whether they are represented by (potentially preor postmodified) single nouns or compounds). 188 7 Conceptually Profiling Negative Self-Identifiers <?page no="189"?> The categories in this profile serve to explicate and quantify one - crucial - variable of the NSIs in my corpus, viz. the possible forms and meanings of the identifying NP, serving as the departure point for the analysis of the interaction between NSIs and their linguistic context. To approach the categorization process systematically and ensure maximal transparency, I set up an initial framework according to which nouns can be assigned to superand subordinate categories. Borrowing the concept of prototypes, I argued that establishing a conceptual profile, at least in the case of this analysis, which exclusively examined nouns and NPs, might best be approached in terms of robust categories (with definable characteristics and a prototypical representative) emerging from iterative data analysis and constituting the departure point for establishing other categories in relation to them. Approaching nouns and NPs this way was intended to handle the problem of category fuzziness: being clear about the prototypical features of an element assigned to a category, and acknowledging that this ‘focal semantic point’ is an interpretative decision and thus not entirely objective, is, I think, preferable to postulating clear-cut boundaries between rigidly constructed categories of authentic language in use. In other words, I assume that salience is not something that can best be accounted for by presenting tables with category frequencies, as claiming that a category with 101 members is conceptually more important than one with 90 members probably does not say much about the relations between language, the mind and the social world. Based on these considerations, the final representation of the results of this analysis of identifying NPs here takes the form of a lexicosemantic landscape, representing the different conceptualizations found to be distinguishable and more or less salient in the corpus. 7.3.1 Conceptual landscape of identifying noun phrases The following emerged as ‘focal concepts’, that is, as semantically differentiable in my data. As will be seen, they largely correspond to the superordinate tags introduced earlier; the reason for this is that - as mentioned - the data were not annotated prior to, but as part of, the analysis. • Expertise and professionalism: knowledge/ skill in a domain (e.g., expert, physician, math wiz) • Preference: choices/ likes (e.g., dog lover, fan of air ride) • Activity: regular practices with potential proficiency (e.g., climber, winter camper) • Personal characteristics: traits/ beliefs (e.g., flexible person, masochist) 7.3 Summary and implications of results 189 <?page no="190"?> 28 Some people might, however, object to this claim, arguing that it is indeed possible to become an expert smoker or a very experienced early morning person. • Role: social/ interactional positions (e.g., employee, member) • Habit: routines without expertise (e.g., smoker, early morning person) 28 To check, at least informally, if a semantic tagger would also identify the concepts of expertise and professionalism (i.e., “knowing”) and preference (i.e., “liking”) as salient upon being ‘fed’ a (derived) corpus of identifying NPs, I had the tagging software Wmatrix 4.0 (Rayson 2008) create a key word cloud (Figure 7.5). This cloud represents particular words which are significantly more often represented in the NSIs in my data than in a reference corpus (in this case the informal written component of the BNC). I also had Wmatrix produce a semantic tag cloud (Figure 7.6), which compares automatically tagged semantic domains of the examined corpus to semantic domains identified in a reference corpus, again, the BNC Sampler Written Informal Corpus. Of course, this has to be taken with a grain of salt, because comparing a list of NPs to a corpus of texts is not technically a valid comparison: after all, a list of NPs will significantly differ from a normal text corpus by virtue of having a significantly larger (namely 100%) proportion of nouns. Still, what such a key word and semantic cloud can, indeed, reveal is the prominence of particular nouns or semantic classes of nouns in relation to other nouns and semantic classes of nouns in the examined corpus as opposed to the meanings prominently featuring in another corpus. Figure 7.5: Word cloud of key identifying noun phrases. The most frequently occurring words in my set of identifying NPs as compared to the reference corpus are, firstly, the adjectives big and huge, as well as the preposition of, reflecting that the analyzed nouns and NPs are often premodified by these adjectives and postmodified by prepositional phrases. Regarding the word class of interest, nouns, the most prominently represented nouns in this cloud are doctor, expert, programmer and mechanic (belonging to the conceptual category “expertise” in my classification), eater and smoker (which I classified 190 7 Conceptually Profiling Negative Self-Identifiers <?page no="191"?> as habits), runner (an activity in my framework), seller (i.e., a role) as well as fan (preferences). The prominence of the nouns guy and person is due to the fact that, as mentioned, many identifying NPs are constructions where a category-changing specifier (usually a premodifying noun or adjective) is combined with a semantically neutral head noun (as in Mac guy, guy for a HRM strap, medical person). The semantic cloud (Figure 7.6) also identifies the semantic fields “knowl‐ edgeable”, “like” and “medicines and medical treatment” as key domains. The keyness of the first domain is due to the frequent occurrence of the noun expert, the category “like” is based on the numerous instances of fan, and the semantic field “medicines and medical treatment” in Wmatrix is mainly made up of references to doctor and other medical professions. As mentioned before, most NSIs were found on health forums. This could mean that health is the context in which the structure of interest is used most frequently or that web forums for this topic are particularly frequent. Regarding the prominence of the semantic field “size: big”, this is due to the fact that the analyzed nouns and NPs are often premodified by the adjectives big and huge; the category “unmatched” (a default Wmatrix category) stands out in this cloud because the examined data is full of references to brand names uncategorisable for Wmatrix, such as Longevity.org and meshlab, as well as informal expressions unknown to the program, such as weirdo, twat or Lib Dem. Figure 7.6: Word cloud of key conceptual domains of identifying noun phrases. Figure 7.7 below represents the conceptual categories I have identified in the form of one prototypical member each, with larger fonts representing relatively more frequent occurrences. The figure is framed by the superordinate, or focal semantic categories introduced above, which serve as points of orientation and which are arranged to indicate their (postulated) relations and fuzzy boundaries/ overlaps. Thus, for instance, I’m not a dress person, representing the conceptual category “preference (fan)”, occupies a middle ground between the concepts of “preference” and “personal characteristics”, and similarly I’m not a member is negative self-identification with a role, according to the postulated 7.3 Summary and implications of results 191 <?page no="192"?> focal concepts, but it can also imply a conscious choice against membership and hence be considered to index a (non-) preference. Of course, it could also have implications for a person’s level of expertise (which shows in the differentiation between prime members and regular members etc.), which is why it is positioned somewhat halfway between the semantic poles “expertise” and “preference” here. The reason why “expertise” and “preference” are considered as ‘poles’, that is, as main orientation points standing out in this conceptual landscape, is their prominence in terms of two aspects: • Frequent occurrence of a lexeme: a particular lexeme (type) reoccurs in variants and thus constitutes and prototypically represents a conceptual category in its own right (category-defining lexeme, e.g., fan). • Frequent representation of a conceptual category: a definable concept (e.g., “preference”) is frequently represented (token frequency) by a variety of semantically related types. • Summing up, the analysis revealed that the nouns expert and fan not only appear prominently in the corpus overall but also occur in many compounds and together with prepositional phrases referring to specific kinds of ‘fandom’ and expertise. Another category which emerged as relatively prominent is that of personal characteristics, which indicates that NSIs are used by people engaging in web forum discussions to carefully position themselves not only regarding their knowledge and preferences, but also in terms of very personal aspects, such as their personalities, bodies and behaviors. Figure 7.7: Conceptual landscape of negative self-identifiers. 192 7 Conceptually Profiling Negative Self-Identifiers <?page no="193"?> To learn which conceptual categories of NSIs feature prominently in a different set of data, I applied the same categorization system to conceptually profile the set of 264 instances of the variant “I’m not a/ n + identifying NP” produced by searching the Spoken BNC2014 (see Section 3.4.2, which explains the details of the queries performed). In contrast to my self-compiled corpus of a pre-defined set of formal variants of NSIs in their co-texts as used in UK web forum discus‐ sions, the Spoken BNC2014 is a representative sample of spoken face-to-face interaction in British English. This means that a comparison between these two vastly different corpora cannot be deemed statistically significant. Still, such a comparison can serve to identify tendencies of certain conceptual categories to be prominently represented in NSIs used in different contexts. The most frequently represented conceptual categories of nouns and NPs with which speakers in this data contrasted themselves are presented in Table 7.15 below. Category To. Items Preference 96 - Preference (+ fan) 61 - General 14 fan (4) [big (6), huge (2), massive (2)] fan Food/ drink 20 big carrot fan, big McDonald’s fan, lentil fan big fan of: artichokes, beer, crumble, French beers, pizza, porridge, pumpkin pie, sparkling wines, those wraps, churros fan of: ginger, pizza, waffles, wine huge fan of: Chinese [food], Skittles; massive like fan of it [Chinese food] Persons 8 big Ridley Scott fan, Larry David fan, big fan of them fan of: anyone, his, like Ameri-, the whiners, them Arts & entertain‐ ment 6 big Alice in Wonderland fan, massive Shakespeare fan, mas‐ sive Queen fan fan of: sequels (Disney), the song big fan of R and B Nature 3 big fan of leaves, great fan of the trotter, massive insect fan Products/ objects 3 fan of these TTs, massive fan of ankle length things, big fan of that [ Joy Division poster] Specific but not specified 3 big fan of the (.), fan of that kinda stuff, big fan of it Ideological/ reli‐ gious 2 fan of the stated aims, big fan of change 7.3 Summary and implications of results 193 <?page no="194"?> 29 Resty: informal, nonstandard adjective meaning ‘inclined to rest or stay still’, used here as the opposite of ‘fidgety’. Category To. Items Sports 1 big rugby fan Metalinguistic 1 fan of “hon” Preference (fan) 35 - Food/ drink 14 big drinker (2), big drinker of coffee and tea, condiment per‐ son, dunker, dunker of croissants, great lover of aubergines, great one for a fry up, lover of Marmite, mangetout person, Rice Krispies fiend, vegan, vegetarian, very fishy person Activities 14 bath man, beachy surfy kinda person, big film lover, big shopper, boat person, cave person, foodie, gym person, Mas‐ sive Formula One person, music person, phone person, pub person, sand person, sports person Activity aspectrelated 2 massive gig person, speeder Sexual practice 2 one night stand kind of guy, threesome girl General 1 slave to it Products/ objects 1 sort of massive opponent of ebooks Style 1 leggings wearer Characteristics 87 - Evaluative (literal) 25 ageist, amazing person, bad person, bad sort of sea traveler, cheater, fool, freak, fucking idiot, fucking imbecile, gayist, good parent, idiot (3), lair [liar], lazy person, racist (3), snob, twit (3), violent person, winer [whiner] Evaluative (meta‐ phorical/ metony‐ mic) 16 asshole (2), bastard, bitch, bum, cow, dick, dickhead (2), fanny, friendly voice, heartless monster, little wife, pretty face, princess, sponger General 11 adult, child (3), lad, little girl, man (2), parent, teenager, young man Psychological 8 deep thinker, dweller, impatient person, perfectionist, roman‐ tic, tough hard guy, quitter, resty person  29 Ideological/ reli‐ gious 6 anti-Semite, granddad about sex on the TV, Muslim, PC per‐ son, practising Christian, traditional by the book Christian type 194 7 Conceptually Profiling Negative Self-Identifiers <?page no="195"?> Category To. Items Social 6 celebrity, couple, loud party-throwing person, people person, team player, widow Geographic 4 Brummie, fucking northerner, Geordie, southerner Physical/ physio‐ logical 4 big eater, light sleeper, size sixteen, small person Metaphorical 2 animal, spring chicken Sexual orientation 1 Lesbian Gender-specific 1 girly girl Metonymic 1 a hundred percent Health/ Ill-health 1 alcoholic Other 1 exam-taking person Expertise & pro‐ fessionalism 54 - Professions 33 academic, acoustic engineer, Chinese teacher, counsellor, en‐ gineer, good cook, good enough hairdresser, good musician, graduate, great painter, great sewer, hair- (dresser), engineer, legal guardian, meteorologist, million pound businessman, model, natural teacher, nurse or a doctor, professional baker, professional skier, qualified counsellor, researcher, scientist, sheriff, Soldier, soldier, spy, student (2), surgeon, uni graduate, writer cum author Specific expertise (expert) 14 bad player, good judge of character, good reader, good talker, good walker (2), great reader, natural driver, performing cat, quick learner, terrible driver, very good critique, very good photographer, very good swimmer Specific expertise (+ expert) 3 expert on Chinatowns, expert in politics, DIY expert General expertise (+ expert) 2 expert General expertise (expert) 2 authoritative person, specialist 7.3 Summary and implications of results 195 <?page no="196"?> Category To. Items Roles 8 - Role: trade/ profes‐ sional 4 charity (2), customer, regular customer Role: crime 2 rapist, victim Role: affiliation 1 part of it (their team) Role: virtual 1 Sim Usage/ consump‐ tion/ ownership 1 one of your big users Others 18 - Total 264 - Table 7.15: Conceptual categories of negative self-identifiers in the Spoken BNC2014. As can be seen, the three conceptual categories of identifying NPs found to be key in the data examined for this study - “Preference”, “Expertise & Profession‐ alism”, and “Characteristics” - also emerge as key in the Spoken BNC2014. While NSIs from the conceptual domain of “Expertise & Professionalism” represent the largest category in the corpus examined for this study, they account for only 20% of NSIs in the Spoken BNC2014. In that corpus, negative self-identification with particular “Characteristics” accounts for 33% of instances, compared to just 15% in the present data. This suggests that “Expertise” may play a greater role in online forum contexts than in spoken offline conversation, where evaluative disclaimers appear more frequent. While this comparison is necessarily small in scale, it may indicate that “Expertise” is a site of conceptual struggle on online forums - a point discussed in more detail in the next section. The comparison is indicative only: it contrasts the set of NSIs with a derived collection of NSIs taken from a representative spoken corpus, and differences are not tested for statistical significance but should be read as tendencies. Table 7.16 provides an overview of the relative frequency of the three most prominent conceptual categories of NSIs in both corpora. 196 7 Conceptually Profiling Negative Self-Identifiers <?page no="197"?> 30 Type counts are not provided in this table, as the focus is on category-level distribution rather than lexical diversity. Percentages do not sum to 100% because infrequent categories (“Other”) are excluded. Corpus Total no. of NSIs Prefer‐ ence (To.) Prefer‐ ence (%) Exper‐ tise & profes‐ sional‐ ism (To.) Exper‐ tise & profes‐ sional‐ ism (%) Char‐ acteris‐ tics (To.) Char‐ acteris‐ tics (%) Corpus of NSIs 936 246 26% 393 42% 136 15% Spoken BNC 2014 264 96 37% 54 20% 87 33% Table 7.16: Conceptual categories of negative self-identifiers: Corpus vs. the Spoken BNC2014. 30 7.3.2 Implications: “I might not be an expert, but I know what I do not want” Regarding the question of potential implications of these results in light of the social research interests of this study, the way of positioning the conceptual categories emerging from my analysis in this figure is intended to visualize not only their respective prominence in terms of tokens assigned to them, but also my evaluation of these results. I do not think that it is a coincidence that the domains of “expertise” and “preference” are so prominently represented, neither do I think that (not) being an expert or a fan are entirely unrelated conceptualizations. Just as the concept of expertise is so often made relevant, thereby implicitly acknowledging the hierarchical difference between experts and laypersons and indexing membership with the social group of non-experts, the concept of choice by non-preference is also frequently foregrounded, constructing speakers as ‘wise choosers’ - an identity category which might become just as important as (or even more important than) that of the expert when strategically managing one’s utterances in web forum interaction. My analysis revealed that the structure in focus is highly context-dependent, being used to contrast speakers with categories deemed relevant for their self-representation in particular communicative situations. This is precisely why NSIs say something about these contexts of usage: if speakers make their non-affiliation with the category of 7.3 Summary and implications of results 197 <?page no="198"?> experts relevant in the context of discussing subjects traditionally associated with a high degree of technicality and complexity, they in fact undermine the importance of formal education in these subjects. At the same time, other strategies of construing expertise might become more pivotal in these contexts. According to Sillence (2010), trustworthiness is mainly established through showing (and thus, per implication, not necessarily mentioning) expertise. The analysis suggests that one of these strategies could be to identify via (very specific) non-preferences on forums defined by a common subject of interest, which serves to construe a highly specific position in an already specialized discourse. For example, on a forum on the football management simulation game Football Manager (which, I shall argue, could be considered a niche interest in the first place), negatively identifying as fan of the AXT-11 template in example 7.7 below construes the speaker as very involved with the subject of the forum. 7.7. A: scrapped my old template because it was shit, now started on a new one [PICTURE] - B: i like it GGG75 the design is good but I am not a fan of the AXT-11 template just looks a bit dirty Using epistemic disclaimers to index one’s lay status while discursively per‐ forming expertise could represent an effective strategy of coming across as trustworthy, but careful and self-aware regarding the epistemic status of one’s utterances, as well as socially sensitive, using language to create a modest, rather than authoritative, impression. In other words, speakers may construe themselves as experts while negatively identifying as such, drawing on a range of linguistic strategies to do so. Thus, liking and knowing appear as related concepts in my corpus insofar as negatively identifying with very specific preferences constructs the speaker as opinionated in the sense of being well-aware of the plurality of available options and, thus, knowledgeable - albeit not in the traditional sense of having formal qualifications or expertise. I therefore assume, based on the results of this analysis, that the following two related conceptual trends (defined by Marko 2015a: 130 as “general beliefs and attitudes organising more specific ones”, which “come to the fore as trends over larger stretches of text”) manifest themselves in people’s use of NSIs on web forums: • De-expertization: using NSIs to disclaim epistemic authority, and, thus mark an assertion as opinion. This can serve various interactional functions, for example, it can serve to open the floor to debate, index speaker alignment 198 7 Conceptually Profiling Negative Self-Identifiers <?page no="199"?> with a community of non-experts and thus highlight common ground, or even ‘mock’ people claiming to be or perceived as experts. • Re-expertization by informed choice: identifying via non-preferences in a discourse community defined by a common preference, thereby positioning oneself as making an informed choice. The foregrounding of individual choice could be considered an emancipatory tendency, as the opposition between fan and non-fan involves less of a power differential than that between experts and non-experts by construing the speaker as the agent of choice rather than as someone lacking expertise. The significance of these results with a view to potentially changing conceptu‐ alizations of self-identity can be illustrated quite well when considering them in reference to a simple example, namely an extract from a blog on the topic of professional specialization. The following statements taken from a blog post called “How to choose your specialization” (Grassilli 2013) suggest that these conceptual trends - examined only in relation to NSIs here - might, indeed, play a role for people’s self-identity more generally. More specifically, they can be considered to exemplify a shifting conceptualization of expertise insofar as it seems to be increasingly related to, and seen as resulting from, choice and preference rather than, say, formal education: 7.8. A - focus on what the market needs - first analyze the market, then choose the subjects to specialize in B - focus on what you like - choose the areas you are passionate about, then offer your expertise to the market. What if you could be the specialist in gluten-free or vegan products, for example, because you’re passionate about healthy food? In example 7.8 above, the author discursively merges the notions of specialist knowledge (using the verbs focus, analyze, and specialize) and personal prefer‐ ence and choice (choose, like, passionate). The rhetorical “What if ” question seems to summarize the possible implications of results of this analysis: the simultaneous prominence of the concepts of expertise and preference in the examined forums might suggest that the boundaries between these two concepts are becoming increasingly blurred, to the effect that being an experienced consumer might be perceived as expertise (this is supported, e.g., by Mackiewicz (2010a: 4), who shows how asserting product familiarity may serve as a credibility marker). If such conceptualizations can be shown to manifest themselves in the way people position themselves toward their own utterances in everyday, informal conversations, this would imply that they implicitly structure identity 7.3 Summary and implications of results 199 <?page no="200"?> management in discourse and, thus, are becoming naturalized - which, as will be discussed in more detail below, can have sociopolitical implications beyond the local communicative situations in which people use the structure “I + copula + NOT + indefinite NP”. In other words, of particular interest for this study are NSIs which are routinely used to serve particular pragmatic functions, thus constituting ‘commonsensical’ ways of self-representation in online interaction. Judging from the semantic analysis of 246 NSIs as used in the Spoken BNC2014, negative self-identification with preference and expertise/ professionalism ap‐ pear to represent quite routinised ways of speaking: together with NSIs used to contrast speakers with particular (mainly evaluative) personal characteristics, they are the conceptual categories most frequently represented in the sample of spoken British English I have analyzed. To conclude, in this chapter I have answered RQ1 of this study. I analyzed which NPs forum users negatively identify with and created a conceptual profile of identifying NPs. By determining the frequency of occurrence and lexical variation of the nouns and nominal structures assigned to various conceptual categories, I established how prominently the respective categories, and particular lexemes within them, feature in the examined data. The analysis revealed that negative identification with nouns from the fields of preference and expertise is particularly prominent in the corpus. In Section 2.2.1, I explained how expertise relates to notions of both risk and trust. I suggested that online forum communities represent sites where novel forms of expertise, based on lay experience, may be created, distributed, and negotiated, and that they are trusted by peers precisely because of their subjective nature (Vermeulen & Seegers 2009; Vásquez 2014; Rueger et al. 2021). It follows that foregrounding preference through negative self-identification (e.g., as a fan) and disclaiming expertise in order to align with a lay community may figure as preferred discourse management strategies in these contexts. To study NSIs in their co-texts, I turn to RQ2 in Chapter 8 and functionally analyze their use. More precisely, I examine the formal-functional relations between NSIs and their co-texts, and whether co-texts with specific experiential functions co-occur with particular conceptual categories of NSIs. 200 7 Conceptually Profiling Negative Self-Identifiers <?page no="201"?> 8 Negative Self-identifiers and Their Immediate Co-Texts While the previous chapter examined the meanings of identifying NPs with which speakers contrast themselves, this chapter and the next address RQ2: A. What are the formal-functional relations of NSIs in my corpus and their clause-internal and clause-external co-texts? B. What are the meanings and functions of co-texts with certain formal links to the NSI structure, and how frequently are they represented in the corpus? C. Do co-texts with specific meanings and functions occur together with particular conceptual categories of NSIs in patterned ways? More specifically, the analyses presented in this chapter examine, first, the clause-internal co-texts of NSIs to establish whether, and how, instances of the structure in my corpus are preand/ or postmodified. Second, they analyze the formal-functional relations between instances of the NSI structure in my corpus and their clause-external co-texts, that is, sentences and clauses preceding and following them (RQ2A; Section 8.1). Third, they present an experiential profile of co-texts with formal-functional links to the structure (RQ2B) and identify potential co-occurrence patterns between co-texts with particular functions and particular conceptual categories of NSIs (RQ2C; Section 8.2). As will be explained, establishing categories of co-text for the structure in focus raised several theoretical and methodological issues concerning the criteria by which such categories can or should be set up. These analyses yield a contextual profile of NSIs. This profile not only sheds light on the usage of NSIs in general, but also helps assess whether the expectations about deand re-expertization formulated in the previous chapter play a role in their functions. The interaction of NSIs with their co-text can be examined from two perspectives. The first would be to start with the conceptual category of the identifying NP and examine the relations of instances of that category with their co-text. Alternatively, one can start by looking at the co-text of all NSIs, initially ignoring the conceptual category of the identifying NP, in order to identify categories of context which occur particularly often. These prominently represented categories can then be examined to determine whether they co-occur with particular types of NSIs. <?page no="202"?> In the analysis to be presented below, I took the second approach, that is, I began by examining formal-functional categories of context preceding and following all NSIs in the corpus, regardless of the conceptual categories of identifying NPs, to establish whether particular contexts occur particularly frequently before or after instances of the structure in focus. Thus, the analysis sheds light on the usage of the structure “I + copula + NOT + identifying NP” in this corpus from a more general, structure-oriented perspective to find out whether there are (a) particular positions in a conversation and in a sentence where NSIs occur relatively frequently, and (b) particular forms and meanings which constitute the co-texts of NSIs particularly often. These conceptually prominent contexts are then examined in more detail, taking into account the conceptual values of the NSIs they interact with. 8.1 General co-textual profile The internal and external co-texts of NSIs can initially be differentiated as follows: • Turn-internal co-text - Clause-internal co-text: This refers to textual elements preand postmodifying the structure “I + copula + NOT + indefinite NP” (see Table 8.1 below for examples). - Clause-external co-text: This refers to the clauses and sentences immediately preceding and following the NSI, that is, independent sentences, coordinated or super-/ subordinated clauses. Assuming that there is a tighter conceptual link between sentences and clauses that are formally linked, I prioritize the conjoined clause or sentence in the contextual analysis. For example, if an NSI is preceded by an independent sentence not related to it by a connecting adverb or a conjunction, and followed by an independent sentence introduced by however, I analyze the second sentence first. Table 8.1 provides examples of clause-internal context (shaded in grey) and clause-external context. It also indicates the order in which elements of the former are considered in the analysis: 202 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="203"?> Order of analysis 3. 1. 2. (4.) Clause-inter‐ nal co-text - I’m not a tech‐ nological man at all - - No of course I’m not a pirate Clause-exter‐ nal co-text No offence taken - I’ve not been a regular poster since joining, so probably fair to say I’m a little damp behind the ears. - Perhaps your doc‐ tor can send you to someone who can help you sort it out and reduce the lkelyhood [likelihood] of infection? I have never been a fan of shaving as I was al‐ ways told hair is there for a reason, although I know lots of ladies de hair themselves in a number of ways Table 8.1: Internal and external co-texts of negative self-identifiers analyzed. • Turn-external co-text This refers to textual elements pertaining to turns of other speakers, for example, A’s turn in example 8.1 from my corpus. In this example, the NSI fulfills a different function than for example, disclaimers of expertise, merely conveying factual information. 8.1. A: If you’re an individual seller you are bound by Amazon’s default shipping prices---you can only create your own postage prices if you are a pro seller (subscription only) - B: No I am not a personal seller and I have set all my shipping prices for my listings. All three kinds of co-text are relevant when analyzing NSIs, as they might occur as independent sentences or as part of complex clauses, turn-initially and turn-finally and as they may or may not be uttered in response to another speaker’s turn. In addition, of course, NSIs might be pre-or postmodified by adverbs, phrases and discourse particles with various functions, which should also be taken into consideration. 8.1 General co-textual profile 203 <?page no="204"?> To provide a general co-textual profile of NSIs, I differentiated between them according to their position in the turn and sentence, and their formal relations with different kinds of co-text for a general overview of the use of the structure. I then analyzed in more detail particular co-texts with particular relationships with the matrix clause which emerged as frequent from the first contextual analysis. The general co-textual profile presented further below takes into considera‐ tion where in the turn the NSI appears, considers textual elements modifying the structure “I + copula + NOT + indefinite NP”, and distinguishes between NSIs based on whether they are followed by a coordinated, superor subordinated or independent clause. A distinction between independent sentences preceding and following NSIs is made according to sentence type (declarative, interroga‐ tive, directive, exclamative) and different functional relations between clauses and sentences established by adverbs and conjunctions (cause, contrast, conces‐ sion, consequence). Of course, not all functional relationships are established by adverbs or conjunctions; two independent sentences without any explicit connective may still express contrasting meanings. Table 8.2 summarizes the criteria for creating the general co-textual profile: Preceding co-text Negative self-identifier Following co-text • 0 = turn-initial • Independent sentence (declar‐ ative, interroga‐ tive, directive, or exclamative) • Coordinated or subordinated clause • Other element • A separate sentence, possibly in‐ troduced by a conjunction or a conjunctive adverb, which estab‐ lishes a functional relationship with the previous sentence • Part of a complex clause (i.e., as subordinate, superordinate, or co‐ ordinate clause) with a particular functional relationship with its syntactically related co-text • Parenthetical • Preand postmodification • 0 = turn-final • Independent sen‐ tence (declarative, interrogative, direc‐ tive, or exclamative) • Coordinated or sub‐ ordinated clause in‐ troduced by a con‐ junction • Independent sen‐ tence introduced by a conjunctive adverb • Other element Table 8.2: Criteria for a general co-textual profile, including positional, structural, and functional criteria used to classify the preceding and following co-texts of negative self-identifiers. The general co-textual profile presented in Table 8.3 below gives an initial over‐ view of the relations between NSIs and their co-texts, that is, considers which linguistic elements precede and follow them. Parts of speech directly preceding and thus occupying the first position to the left of the NSI are categorized as 204 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="205"?> L1 co-texts, while those directly following the structure, occupying the first position to the right, are referred to as R1 co-texts. Focusing on these positions offers a systematic approach to functional analysis, as formal adjacency is typically indicative of functional proximity - reflecting the principle that grammatical structure is often iconic of meaning (Givón 1979; Glynn 2007). High-frequency categories are highlighted by grey shading. Tag Category of clause-external co-text L1 R1 <cj: cau> Conjunction/ adverb: cause (as, because, since, when, in case) 41 20 <cj: con> Conjunction/ adverb: contrast (but, however, though, yet) 86 279 <cj: conc> Conjunction/ adverb: concession (although, while, even if) 27 5 <cj: cq> Conjunction/ adverb: consequence (so, thus) 4 94 <ina> Turn-initial: response 120 0 <ini> Turn-initial: not responding 52 0 <hello_r> Turn-initial, following greeting: response 43 0 <hello_nr> Turn-initial, following greeting: not responding 15 0 <is: s> Independent sentence: declarative 376 196 <is: q> Independent sentence: interrogative 47 23 <is: imp> Independent sentence: directive 4 5 <is: exp> Independent sentence: exclamative 31 10 <corr> Correction 0 40 <end> Turn-final 0 87 Others Rare categories (≤20 instances; e.g., relative pronouns, pa‐ rentheses, noun phrases, non-verbal elements, quotes, lists, infinitive phrases, superordinate clauses) 59 34 Total — 936 936 Table 8.3: Clause-external co-text categories and their frequencies, with formal category tags and position frequencies for left (L1) and right (R1) contexts (L1 = first token to the left; R1 = first token to the right). As can be seen, the examined NSIs are most often preceded by declarative sen‐ tences (376 cases). They appear turn-finally without further postmodification in only 87 cases overall and turn-initially in 228 cases overall, framing what 8.1 General co-textual profile 205 <?page no="206"?> speakers are about to say; 163 of these ‘framing’ instances are uttered in response to a previous turn by another speaker, sometimes preceded by a greeting. (An example would be Hi [NAME], welcome to the forum, I’m no expert on Slimming World, but have read on here that some plans are more suitable than others for people with diabetes.) The high number of NSIs occurring at the beginning of responding turns might indicate they are salient linguistic choices in the context of giving advice, being used as turn-initiating disclaimers before presenting one’s opinion on a particular subject. Responding turn-initiating NSIs are also the ones most likely to take up an identification category mentioned or implied by the previous speaker’s turn. A case in point is example 8.2 below, where B’s negative identification as an expert on oil clearly relates to A’s request for advice on which oil to use: 8.2. A: Hi everyone, can you help with this question please.We have just bought a Mc Culloch Petrol Rotary Mower 46s 500 series. It has a Briggs and Stratton Engine, and recommends using SAE30 oil, which we have bought.As you know this is quite expensive to buy, can we use an alternative, like say 5w30, 10w30 or 20w50, without there being any risk of damaging the engine.Our rotovater [rotavator] engine (Tecumseh 3.5) and our Qualcast Classic 45s petrol mower, both advise SAE 30 oil for the engines, but also says a multigrade oil of 15w40 or 20w50 is acceptable, we use the 20w50 in both, with no problems. Your advice would be most appreciated. - B: Hello [NAME], I am not an expert on oil, […]. NSIs are often introduced by conjunctions and conjunctive adverbs coordinating them with preceding textual material (overall 182 cases, an example would be I took a wild guess at the 20’s/ 30’s but I’m no historian), and overall 536 NSIs, that is, almost 60% of all examined instances, are followed by a conjunction or conjunctive adverb linking them to following textual material (e.g., I’m not an owner of a BTS yet. But I already have myself a 4x2x2 vivarium). Occasionally, NSIs occur in parentheses, either themselves or followed by one, but these cases are rare. Only in a small minority of examined cases does a turn-medial NSI appear on its own, that is, as a separate sentence or without any preor postmodifier. This means that the instances of the structure examined are mostly tightly interwoven with the rest of the text, which points toward their high functional interactivity. In 40 cases, NSIs are followed by what I named “correction”, that is, affirmative identifiers such as: I’m by no means an expert - I’m a complete amateur! These corrections either appear as declarative sentences as in this example, or as elliptical constructions, as in I am no pain patient, [I am] just a substituted ex-heroin user. 206 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="207"?> In the following, I first discuss clause-internal preand postmodification of NSIs (e.g., as in I am no expert (at ALL)), and then move on to consider clause-external textual elements with different formal and functional relations with NSIs. I will then present the results of the analyses of NSIs formally related to their co-text by conjunctions and adverbs, as the most frequent type of appearance of the structure in my corpus, and of instances preceded by declarative sentences, which are the second most frequent type. 8.1.1 Clause-internal modification The following table gives an overview of linguistic elements which with NSIs are preand postmodified. In total, 170 NSIs come with modifiers. Category Preceding text To. Following text To. Temporal adjunct in this time, throughout my career, since…, so far, usu‐ ally 5 12 hours yet, after [+ clause], as long as [+ clause], before (3), even before [+ clause], for [+ NP denoting a time pe‐ riod] (14), lately, long, long enough to (2), not even as a kid, over the last month, since… (7), so far, these last 20 months, until (2), up to now, yet (3) 41 Information status/ impor‐ tance high‐ lighter as/ like I say/ said (10), note (3), I should/ have to say (3), again (2), I repeat (that) (2), I must stress here, I want to be clear that, please note, once again, just to be clear, this is to show, I should add that, mind, you will be aware that, as you see, as many on this forum will know, as I was telling sb. in another thread, as you can see from my previous post in this thread, you know, as my subject states, I said before 36 mind you 1 Discourse markers well (5), now (3), anyway (3), ok (2), oh, hesitation (hmm), stance (I must ad‐ mit) 16 — — 8.1 General co-textual profile 207 <?page no="208"?> Category Preceding text To. Following text To. Information label PS (4), disclaimer (2), memo, edit, as an aside, FYI, just a thought, warning 12 — 0 Range of ap‐ plication speaking of…, with regard to…, for…, except for, as for…, basically 6 when it comes to (2), if it comes to, except for, espe‐ cially when… 5 Sequence finally (3), first, first off, furthermore, let me start off by saying that, the second thing is that, other than that 9 by the way 1 Epistemic obviously (2), clearly (2), contrary to how it may seem, granted, of course 7 obviously 1 Intensifica‐ tion — — at all (3), by any means (3), by any stretch of the imagination, in the slight‐ est 8 Comparative like your friend, like you, like…you mention, likewise 4 compared to, as much as, like you, like some of the great people around here 4 No-intensifi‐ cation no, I’m not a low carber; no, I’m absolutely not an ML stooge; no, I’m not a high or even particularly lyrical guy; no, seriously, I’m not a spammer 4 — — Specification — — as such, really 2 Locative ad‐ junct — — in…, outside of…, in a … environment, on here 4 Evaluative sadly, unfortunately 2 sadly, unfortunately 2 Total - 101 - 69 Table 8.4: Preand postmodifiers of negative self-identifiers. As can be seen, in 46 cases NSIs in my corpus occur together with temporal adjuncts, which means these are the most frequent type of clause-internal preand postmodification of the structure in focus and, thus, that the temporal dimension of not belonging to a particular identity category is most often fore‐ 208 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="209"?> grounded. Table 8.5 below shows to which conceptual categories (as discussed in the previous chapter) temporally modified NSIs can be assigned. Category To. Role: forum-related 9 Characteristics: health/ ill-health 5 Professions 5 Role: membership 4 Activity/ expertise 3 Characteristics: social 3 Preference (+ fan): product 2 Preference (+ fan): person 2 Role: trade 2 Usage/ consumption/ ownership 2 Characteristics: physical/ physiological 1 Preference (+ fan): food/ drink 1 Specific expertise (+ expert) 1 Expertise (expert) 1 Preference (fan): product 2 Habits: routine 1 Habits: substance 1 Role: virtual 1 Total 46 Table 8.5: Conceptual categories of temporally modified negative self-identifiers. Speakers in my corpus most frequently specify the temporal validity of their negative self-identification when talking about membership of particular groups and forums (e.g., member of the previous club, active contributor to [FORUM NAME]). This indicates that duration of membership and frequency of contribution on a particular forum plays a role for the discursive self-repre‐ sentation of people interacting on the examined forums. This might have to do 8.1 General co-textual profile 209 <?page no="210"?> with perceived and actual authority in this discourse, with longer membership potentially being associated with greater prestige and practically with more user rights in comparison to regular/ short-time users (an observation also made when discussing example 7.3 in the previous chapter). Example 8.3 features an instance of an NSI serving as a factual self-description by the speaker. It is a case of long-term membership and regular forum activity being represented as linked to a status of greater respect among forum users. The topic of the thread is the question of who should be nominated as “most intelligent/ intellectual poster”, and for the speaker, only long-term posters deserve this award: 8.3. Re: Most intelligent/ intellectual poster 2015 Got to be [NAME], [NAME] or [NAME] for me. Always enjoy debating them even when we disagree. Have decided to play rock paper scissors with my lovely wife and eldest son in order to pick which of the three will get my final vote. I’m surprised (albeit pleased) to see my name on the list given how little I have posted this year and would like people to know I don’t think they should vote for me because I don’t deserve to win it when I haven’t been a regular contributor. The second largest group of modifiers are what I classify as “information status/ importance highlighters”. These include (a) expressions that metadis‐ cursively comment on the status of the information provided (i.e., speakers often highlight that they have already negatively identified with a particular conceptual category, e.g., by using the premodifier as I have said), and (b) expressions that indicate the speaker considers the information conveyed by the NSI - their non-identity claim (i.e., their act of negative self-identification) - as important for the recipient (e.g., please note). This suggests that, at least in these cases, speakers use negatives not en passant but deliberately, foregrounding them metadiscursively for other participants. Regarding whether information status/ importance highlighters occur more frequently with particular conceptual categories of NPs, it is note‐ worthy that speakers are particularly likely to emphasize their negative self-identification when it comes to being or not being an expert. As shown in Table 8.6, information status/ importance highlighters most frequently (in 26 of 37 cases) modify NSIs from the conceptual category of “expertise and professionalism”. 210 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="211"?> Category To. Expertise & professionalism 26 Preference 5 Characteristics 4 Habits 1 Ideological 1 Total 37 Table 8.6: Conceptual categories of negative self-identifiers adjacent to highlighters. The classification system for preand postmodifiers of NSIs, presented in Table 8.4 above, also distinguishes a category labelled “information”. This category includes modifiers which metadiscursively name the information provided by the NSI, for example, PS (postscript), disclaimer, or even warning. These labels in some cases comment on the NSI in terms of information structure, that is, by setting the NSI apart from the rest of the utterance by classifying it, for example, as a PS or as an aside. They may also index the type of information being provided, for example, a warning or just a thought. Since labels of this kind occur together with NSIs in only 12 cases, though, this co-occurrence cannot be claimed to constitute a pattern of language use. 8.1.2 Negative self-identifiers in their clause-external co-text As shown in the general overview of the co-text of NSIs (Table 8.3), the structure is often preceded, and mostly followed, by contrasting conjunctions (notably but, which is the conjunction contrasting an NSI with another clause in 68 of 86 cases). This indicates, regardless of the content of the examined utterances, that NSIs are most frequently used to highlight a contrast between what speakers claim not to be and what they say immediately before or after using the NSI. This could point toward a metadiscursive use of the structure in focus, which will be explored in more detail in Chapter 9. There are 138 NSIs followed by the coordinating conjunction and, which could indicate that many NSIs occur together with linguistic elements more specifically describing what not being an X means or entails (e.g., I have never been a pill-popper and just hate taking medication). (The experiential meanings of the clauses coordinated with NSIs are examined in more detail in Section 8.2.) In 94 cases, an NSI is followed by coordinated clauses introduced by so (and other connectors with 8.1 General co-textual profile 211 <?page no="212"?> consequential meaning, e.g., thus); complementarily, 41 NSIs are preceded by a causal conjunction, mainly as, which means that speakers often state the consequences of their non-identification (or explain particular states of affairs by NSIs). In addition, 48 NSIs are preceded, and 22 followed, by interrogatives, which means that NSIs are also used in the context of asking questions. However, given that in 571 cases they are preceded or followed by a declarative sentence, making statements is the most popular linguistic choice in the immediate context of NSIs. Table 8.7 below provides an overview of the most frequent formal-functional relations between NSIs and their immediately adjacent co-text, to which they are formally linked by adverbs and conjunctions. As can be seen, a total of 717 NSIs, corresponding to 77% of the instances of the structure in this corpus, are related to preceding or subsequent textual material. Of course, as mentioned before, there may be relationships (e.g., of cause) between sentences unrelated by adverbs marking that relation. One example is I’m not a big fan of the opacity, it makes the whole screen look washed out when only one item is bright and the rest are dull. Due to the dominance of NSIs formally related to their co-text in the corpus and the conceptually tighter association between sentences whose functional relation is formally marked, the analysis of relations between NSIs and formally coordinated and subordinated sentences was given priority in this study. The results of this analysis are presented in Section 8.2. Relationship Preceding co-text Following co-text Number of instances Contrast and concession NSI as contrastive clause 86 279 365 NSI as concessive clause 6 16 22 NSI with concessive co-text 4 5 9 Subtotal - - 396 % of all NSIs in the corpus - - 42% Cause and consequence NSI as cause 41 93 134 NSI as consequence 4 20 24 Subtotal - - 158 % of all NSIs in the corpus - - 17% 212 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="213"?> Addition NSI as additive clause 24 138 162 Subtotal - - 162 % of all NSIs in the corpus - - 17% Total 165 551 716 Table 8.7: Formal-functional relations between negative self-identifiers and adjacent co-texts, based on conjunctions and conjunctive adverbs. 8.1.3 Summary Summing up, this section has presented the results of the analysis of the formal relations between NSIs and their clause-internal as well as clause-external co-text. I examined the preand postmodifiers of NSIs, which revealed that 170 instances of the structure in the corpus are modified clause-internally, with temporal modifiers occurring most frequently. The analysis of clause-external relations showed that a large majority - 717 instances - of NSIs are formally linked to their co-text via conjunctions and conjunctive adverbs. Among these, contrastive relations are by far the most frequent, with contrastive adverbs and conjunctions linking NSIs to following textual material in 279 cases. This analysis has shown which formal relations between NSIs and their co-texts are most prominent in the data examined and provides the empirical basis for the more detailed analysis of these frequent relations (as summarized in Table 8.7) in Section 8.2, thereby contributing to answering RQ2. 8.2 Negative self-identifiers and their formally related co-texts The analysis in this section draws on the transitivity framework (Halliday & Matthiessen 2014) to categorize sentences and clauses formally linked to NSIs by conjunctions or adverbs. This categorization reveals patterns in the experiential structure of textual material in the proximate co-text of the NSI structure. In addition, the analysis explores whether particular types of experiential co-texts and conceptual categories of NSIs tend to co-occur. In this way, the chapter contributes to answering RQ2. Using transitivity lets us test whether NSI categories pattern with particular participant-process configurations, directly addressing RQA. 8.2 Negative self-identifiers and their formally related co-texts 213 <?page no="214"?> The transitivity system is theorized to enable us to represent and construe our experience of the world as “a quantum of change in the flow of events as a figure” (Halliday & Matthiessen 2014: 213), that is, it allows us to represent the constant flow of our experiences and thus classify and make sense of them. The reason for drawing on the transitivity framework to analyze the proximate co-texts of NSIs is that this makes it possible to approach the data more systematically than analyses of clauses and sentences according to their overall meaning/ function, making it possible to divide up the authentic language material under scrutiny into relatively clearly differentiable and, importantly, quantifiable categories. Another big advantage of using the transitivity framework instead of a ‘func‐ tionally holistic’ categorization scheme to create a general profile of the co-texts of a relatively large sample of NSIs is that this framework takes into account that textual material’s information structure, which is relevant as it makes a functional difference. To illustrate what I mean, consider the following examples from my corpus: 8.4. I’m no expert, but surely a jump from 19k to 32 in a week is far too much 8.5. I am no expert, but I believe any GAD test over 50 indicates an autoimmune condition 8.6. I am no expert but you may want to check Plate 1, item 4 of the book British Caenozoic fossils published by the natural history museum. 8.7. I am no expert on such matters, but 150/ 90 or 150/ 80 isn’t to [too] far above normal bp, so it may be possible to reduce this by keeping to your diet plan and starting to exercise daily. In all four examples, the speaker negatively identifies as an expert, contrasting the NSI with a sentence introduced by but. Now, if we ignored the different perspectives the sentences represented in 8.4 to 8.7 take, and just considered their overall functions, 8.4 and 8.5 and, respectively, 8.6 and 8.7 would be deemed the same: 8.4 and 8.5 represent claims that could be verified or falsified (by actual experts, for example), while 8.6 and 8.7 can be classified as advice. However, assuming that lexicogrammatical choices reflect and construe our conceptualization of the world, the order in which information is presented and the meanings of the very words chosen, do matter. Indeed, upon closer inspection, the four sentences are quite different: the statement in 8.4 is introduced by the adverb surely, which serves to increase the certainty of the speaker’s assessment of the increase of running distance. By contrast, the introductory I believe explicitly marks the statement in 8.5 as the speaker’s perspective. The second clause in 8.6 has the addressee - you - as its subject and 214 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="215"?> theme and serves as a piece of advice. Example 8.7, too, is intended to provide information of use to the addressee. Here, the recipient is not directly addressed though; instead, the NSI is formally coordinated with - and functionally relates to - an assessment of blood pressure values. This means that the sentence is not primarily about the addressee, but about blood pressure norms, which renders the advice more indirect. Together, examples 8.4-8.7 illustrate how adverbials (surely), stance markers (I believe), addressee-orientation (you), and indirect advice shift the experiential profile of the co-text. To know precisely which participants and processes are formally coordinated with NSIs makes it possible to categorize and, ultimately, quantify the textual material in the immediate context of NSIs. If representations of certain experi‐ ences - that is, certain “factual-notional” clause structures (Halliday 1970/ 1976: 159) - can be found to occur together with particular categories of NSIs in patterned ways, this would imply that speakers routinely (linguistically and thus conceptually) associate particular identification concepts with particular representations of their experiences of the world. To refer back to my expecta‐ tions about de-expertization and re-expertization based on the results of the conceptual profile in the previous section, the prominence of identifying NPs from the conceptual domains of “expertise” and “preference” raises the question of whether these concepts can be found to be frequently made relevant in the context of representing particular processes involving particular participants. Considering the proximate co-texts of NSIs solely in terms of participants and processes does not, of course, suffice to fully capture the functions of NSIs in relation to these contexts. Example 8.8 from my corpus illustrates that examining co-texts of NSIs in terms of the experiential metafunction can only be a starting point for further analyses: 8.8. I’m not a road rider so can’t really comment on whether this actually happens or not. What my analysis of participants and processes in the context of NSIs presented below can capture is that a co-textual clause formally marked as consequence of the negative identification as road rider by so is a process with I in the sayer role (comment constituting a verbal process). If verbal processes frequently occur as co-texts of NSIs formally marked as consequence, this could be taken to indicate that negatively identifying with a particular category is a linguistic choice that speakers tend to make when metadiscursively referring to what they say. To really reveal what function this linguistic choice serves, however, one would have to consider that the finite verbal group in the example above is modified 8.2 Negative self-identifiers and their formally related co-texts 215 <?page no="216"?> by a negated modal verb, can, which implies that being or not being a road rider is associated with the speaker’s ability to make a comment. This means that aspects such as mood and polarity would also need to be taken into consideration to get a fuller picture of the functional relationship between NSIs and their co-texts. Therefore, the experiential co-textual profile discussed and presented in Sections 8.2.1 to 8.2.4 below should be considered as the first of two analytical steps. This first analysis examined the co-texts of NSIs in terms of one strand of language as differentiated by SFL, namely in terms of their experiential functions (Gwilliams & Fontaine 2015: 1). It aimed to find what - if any - experientially describable co-texts are frequently formally (and thus functionally) related to NSIs. The second step of this analysis - presented in Chapter 9 - was to consider co-texts representing particular participant-process configurations, identified as frequent in the first analysis, in terms of their communicative functions. This second step therefore contributes to answering RQB, which asks about the meanings and functions of co-texts with certain formal links to the NSI structure, and their frequency in the corpus. In short, the analysis proceeds in two steps: (i) an experiential profiling of NSI co-texts, and (ii) a functional analysis of the most frequent participant-process configurations, presented in Chapter 9. 8.2.1 Profiling co-textual elements: method and categorization principles This analysis aimed to find which process types, involving particular partici‐ pants, are represented by clauses with certain formal links to NSIs. It considered the immediate and functionally most tightly related co-text of NSIs, that is, sentences and clauses formally related to the structure by means of a conjunc‐ tion (e.g., but) or an adverb (e.g., therefore). In corpus analytic terms, this co-text is, again, defined as L1 and R1, whereby L1 and R1 here constitute sentences and clauses rather than individual lexical items (e.g., I’m not a heavy smoker, but I have been smoking for about 3-4 years now; I am not a forum person myself although I see the benefit of it). In this analysis, I operationalize ‘proximate co-text’ as these L1 and R1 sentences/ clauses, a definition that ensures the procedure is systematic and reproducible. These co-texts were categorized by drawing on the framework of transitivity as proposed by Halliday and Matthiessen (2014) (also cf. Koller 2020 for an introduction). The decision to use the transitivity framework to categorize sentences and clauses formally related to NSIs was made to lend systematicity to the analysis and help establish the frequency of particular lexicogrammatical choices in the co-text of the structure in focus. However, as will be discussed in 216 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="217"?> 31 According to Halliday and Matthiessen (2014: 148), the notion of ‘subject’ is understood in terms of “modal responsibility”, that is, both as a semantic and as a grammatical category. more detail, assigning the textual material at hand to the categories provided by the transitivity framework often means drawing quite rigid boundaries between instances of language in use, suggesting clear differences (or similarities) between utterances intuitively serving quite similar (or different) functions. This is also acknowledged by Halliday and Matthiessen (2014: 216), who explain that their “model of experience, as interpreted through the grammatical system of transitivity, is one of regions within a continuous space”. To account for the fuzziness of process type categories and shed light on the co-texts of NSIs from more than just one perspective, process types found to be frequently instantiated by the textual material examined in this analysis were approached from a functionally more holistic perspective in the analyses presented in Chapter 9. Below, I discuss how I categorized the co-texts of NSIs in terms of participants and processes and discuss some of the theoretical and methodological intricacies involved in this process. My analytical framework categorizes clauses according to process types they realize. It subsumes under the label “Role-1” participants functioning as: actor in material processes, sayer in verbal processes, carrier/ token in relational processes, existent in existential processes, senser in mental processes and behaver in behavioral processes. This, in most cases, also means that these participants constitute the subjects 31 of the clause. An exception is represented by passive clauses, which are marked by a subject-object-swap. While this swap does not affect the functional roles taken by participants (Koller 2020: 59), passive clauses often do not explicitly feature the actor, constituting agentless passives, as in my videos are monetized. In cases such as this one, the subjects of the examined clauses were included as participants in the tables summarizing my findings despite not taking Role 1 according to my framework. In pre-projected mental clauses serving modal assessments, such as it seems to me [that…] or it would surprise me [if…], too, the speaker is listed as Role-1 participant despite not representing the subject of the clause (Halliday & Matthiessen 2014: 257). As a practical rule, if the senser is lexically present, the clause is treated as mental regardless of its grammatical function. In contrast, in relational processes such as it seems unusual, it serves as carrier and thus as Role-1 participant. In existential processes in clauses introduced by a non-personal pronoun there, the existent is listed as Role-1 participant (ibid.: 268). While I include attributes/ values related to Role-1 participants in relational processes in my presentation of results (differentiating them according to the 8.2 Negative self-identifiers and their formally related co-texts 217 <?page no="218"?> 32 Except, as has been pointed out above, in agentless passives. verb relating them to the subject), I did not consider the meanings of participants taking Role 2 according to the framework presented in Table 8.8 below. 32 The reason for the decision to exclude these participants from this analysis is that it is intended as a first experiential profile, the purpose of which is to reveal frequently occurring patterns of participant-process co-occurrence. Role-1 participants (i.e., subject-like participants: Actor, Sayer, Carrier, etc.) were further distinguished in this analysis according to formal criteria (e.g., whether the participant is realized by a pronoun or by an indefinite NP) and semantic criteria (e.g., whether the participant denotes a thing or a person), as follows (an overview with attested corpus examples is given in Table 8.8): • First-person singular pronouns (I → speaker) • Second-person singular pronouns (you → addressee) • Third-person singular or plural pronouns (e.g., they) or a NP (e.g., the juddering) These participants were also differentiated semantically depending on whether they refer to humans, animals, inanimate objects, or abstract ideas (Halliday & Matthiessen 2014: 385-6). The category of non-human third-person subjects also includes pro-forms referring back to preceding co-text, either in the same turn or in a different speaker’s turn (e.g., I’m no doctor, but it doesn’t sound like a great idea to me, where it refers to an idea expressed by another speaker). • Clauses (e.g., getting your diabetes under control) • Co-texts that do not represent processes - for example, single NPs func‐ tioning as substitutes for full clauses (e.g., Hi Gareth, I’m no expert, so no advice, where no advice appears to substitute for “I give you no advice”), or as expressives (e.g., I’m no photographer… but jeeeeeeesus! ). While these do not constitute processes, they are included in the result tables for completeness; they count toward the grand totals but - being non-clausal - are not assigned a process type and are excluded from the process subtotals (listed as “Others (NP)”). As discussed above, categorizing clauses representing the co-texts of NSIs within the transitivity framework can be complex. One difficulty lies in distin‐ guishing between mental and relational clauses, which often appears somewhat artificial. As Halliday and Matthiessen (2014: 542) note, “‘personal’ ‘attributive’ clauses are closely agnate with projecting ‘mental’ clauses” in that both can construe “inner experiences” and serve as modal assessments of a hypotactically 218 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="219"?> 33 In fact, they can also construe “outer experiences”, just like material processes. As an example, they cite cases of present progressive predicates as in She’s walking into the dining room. In my analysis, I treated such instances as material processes. projected clause. 33 For instance, being confident that something is the case (a relational process type) is very similar to thinking that something is the case (a mental process type). 8.9. I’m no expert but I don’t think the juddering is coming through the clutch peddle [pedal]. 8.10. I am no pro but I am confidant [confident] the first photo is a common and harmless hover fly. The difference between mental and relational clauses being used to provide modal commentary on acts or facts lies in the type of relationship they construe. Mental clauses construe a relationship between a senser (e.g., I) and a phenomenon, which may be formally expressed as a separate clause (e.g., the juddering is coming through the clutch peddle [pedal]), whereas relational clauses construe a relationship between a carrier (e.g., I) and an attribute (e.g., confident) (ibid: 261). Further instances that may appear very similar, but actually represent differ‐ ent process types, include clauses in which a preparatory it serves as subject of a modal assessment. Compare, for example, it would be my guess - which, despite being marked as the speaker’s perspective through the possessive my, constitutes a relational process with it as Role-1 participant) - and it would surprise me, which, although the senser appears in the structural position of direct object, is a mental process type, with the pre-projected clause serving as the inducer). 8.11. I’m no expert, but it would be my guess that since the lining of your lungs has a lot of mucous [mucus] to keep dust and the like from building up in there, it would serve the same function to keep resin from accumulating in the larynx. 8.12. It would surprise me if they couldn’t vomit. However, I’m no zoologist. Another issue that needs to be addressed when categorizing the data according to the transitivity framework is how to handle metaphor (Koller 2020: 68). Two types of metaphor are relevant here: • Grammatical metaphors: These involve processes and participants being realized by grammatical units other than those typically associated with 8.2 Negative self-identifiers and their formally related co-texts 219 <?page no="220"?> them -that is, there is an incongruent relationship between semantics and lexicogrammar (Halliday & Matthiessen 2014: 665). For example, while talking to someone is a process typically realized by a verbal group, it is nominalized in have a talk. Depending on whether this nominalized process is considered a (non-congruent) verbal process or whether have is interpreted as construing a possessive relationship between a carrier and the attribute talk, different analyses result. In the tables below, have a talk is classified as a relational process, just like have an idea is counted as relational rather than mental. This form-based approach is intended to provide an overview of the lexicogrammatical choices found in the co-texts of NSIs. In the analyses presented in Chapter 9, frequently instantiated process types - whether used congruently or via grammatical metaphor - are further examined in terms of their discourse-pragmatic functions, with potential metaphorical readings noted. • Conceptual metaphors: These involve “mapping[s] of semantic features from a source domain to a target domain” (Koller 2020: 78, italics in original) and serve to express and even “comprehend one aspect of a concept in terms of another” (Lakoff & Johnson 1980: 458). An example from my corpus is fall in love, which realizes the conceptual metaphors L O V E I S A C O N TAIN E R and C O N T R O L I S U P (Lakoff & Johnson 1980). In transitivity terms, fall in love involves the material process fall being used to express a mental-emotive process. Other examples include mappings from abstract to concrete domains, such as I D E A S A R E O B J E C T S : get it (material process for a mental-cognitive one), take blame (material process for mentally or verbally accepting responsibility), or add thoughts (material process used to metaphorically represent verbal expression - adding to a figurative pile of thoughts). For the sake of a consistent, form-based systematization, such expressions are categorized according to the source domain process type. Thus, fall, get, and take are classified as metaphorical material processes. I am aware that this approach renders the analysis somewhat artificial - counting fall (in love) as a material process obscures that the speaker is linguistically representing a mental-emotive experience. However, as Koller (2020: 69) explains, from an SFL perspective, “all language use represents a motivated choice from the resources that is the language system”, meaning that “the use of a metaphor by a speaker or writer is functional”. Figurative expressions are therefore briefly commented on in footnotes in the presentation of data. Finally, some verbs metaphorically instantiate different subtypes within the same process type (Koller 2020: 68). For instance, find may express a 220 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="221"?> mental-perceptive process literally but a mental-cognitive one metaphorically (as in I find this difficult). In this study, such metaphorical extensions within the same process type are not distinguished, since the conceptual ‘stretch’ is relatively minor. Accordingly, find is always assigned to the mental process category, regardless of whether it is used in its literal (perceptive) or metaphor‐ ical (cognitive) sense. In fact, my framework does not differentiate between subtypes of process types at all. This form-based approach was chosen to ensure consistency, even though it reduces fine-grained semantic distinctions. The following examples of informal language use in online forums further illustrate the challenges of categorizing the data according to a maximally ‘objective’ - that is, formal and framework-based - system. 8.13. Looks like a hoverfly larva, but I am no expert at all and well used to being blown up about my erroneous guesses. 8.14. I am no expert and very happy to stand corrected on the matter ; -). 8.15. I’ve never been a fan of going out much anyway, but like everybody, I’m more than capable of having a few [beers] too many. In all three cases, attributes are ascribed to the speaker: being used to being ‘blown up’ [criticized), being happy to do something, and being capable of something. Functionally, however, these utterances differ substantially. While in 8.13 and 8.14 the self-characterizations mitigate the speaker’s claims, making them appear more tentative or modest, the utterance in 8.15 frames an ability (“being capable of doing something”) as a reference to a tendency toward binge drinking. This illustrates that, while categorizing co-texts by participants and processes provides a useful overview of patterned linguistic structures surrounding NSIs, closer contextualized analysis reveals a wide functional range. Table 8.8 summarizes the process-participant configurations identified in co-texts preceding (L1) and following (R1) the NSI. Process role Corpus example (Role-1 partici‐ pant & verb) Role-2 participant Process category Meta‐ phorical? First person (Actor) I change the way I run Material No First person (Actor) I fall (in love) — Material Yes First person (Senser) I don’t think the juddering [is coming Mental No 8.2 Negative self-identifiers and their formally related co-texts 221 <?page no="222"?> 34 Come down, here, is used to mean “to stop being high on (a drug)” (Merriam Webster: n.d.). Since this experience of coming down occupies a middle ground between a mental and a material process, constituting an involuntary sensory experience but also a sort of ‘activity’, it is classified as a behavioral process here (Halliday & Matthiessen 1999: 136). Process role Corpus example (Role-1 partici‐ pant & verb) Role-2 participant Process category Meta‐ phorical? through the clutch pedal] First person (Senser) It would surprise me if they couldn’t vomit Mental No First person (Sayer) I suggest Lantus alone might not be the best choice for an LADA type Verbal No First person (Carrier) I have a vivarium Relational: attributive No First person (Carrier) I have a conversation Relational: attributive Yes First person (Carrier) I am confident Relational: attributive No First person (Carrier) I keep an eye (out) Relational: attributive Yes First person (Behaver) I came down  34 — Behavioral Yes First person (Existent) I am — Existential No Second per‐ son Sg. 6 (Actor) I am no scientist but from some research, you initially have to melt all the contents in the bowl Material No Clause (Actor) I’m no expert but as already been suggested, getting your diabetes un‐ der control will help — Material (clause) No Third person (Token) I am not a DU stu‐ dent, but my part‐ ner is one Relational No 222 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="223"?> 35 Since table entries are often read in isolation, some examples may initially appear ambiguous or misleading. For instance, I’m no expert, but cocaine is… might be misread as a sarcastic comment implying that cocaine is the “expert”. In context, however, the coordinated clause identifies cocaine as “the most harmful drug”, indicating a judgment about the substance. The functions of such relational processes are further discussed below (Table 8.15). 36 While the clause introduced by but is an existential process, the identifying relative clause following it is actually important for the function of the utterance as a whole: that the people referred to have helped the speaker is represented as relevant in relation to their status as a layperson. This, again, shows that transitivity analysis of the clauses immediately adjacent and formally linked to NSIs just serves as a starting point for finer-grained analysis. As for existential processes, in particular, these were found to be very rare, which is why they are not discussed in more detail. Process role Corpus example (Role-1 partici‐ pant & verb) Role-2 participant Process category Meta‐ phorical? Third person (Carrier) I’m no expert, but co‐ caine is the second most addictive and most harmful drug out there  35 Relational: attributive No Third person (Actor) I am no dog lover but the whole thing has royally pissed [me] off me Material Yes Third person (Token) I’m no doctor but it doesn’t sound like a great idea Relational: identifying No Third person (Carrier) I’m no expert, but it would be my guess [that the lining of your lungs…] Relational: attributive Yes (Actor) I am not a YouTube partner but [my vid‐ eos] are monetized my videos Material (pas‐ sive) No (Existent) I’m not an expert but there are many peo‐ ple on here who have been very helpful to me  36 many people Existential No Table 8.8: Process-participant configurations in the immediate co-texts of negative self-identifiers, identified within the transitivity framework (Halliday & Matthiessen, 2014). The table lists Role-1 participants (Actor, Senser, Sayer, Carrier/ Token, Behaver, Existent), their associated processes, Role-2 participants where applicable, and notes on metaphorical realization. 8.2 Negative self-identifiers and their formally related co-texts 223 <?page no="224"?> 37 All corpus examples in this chapter are formatted according to the notational conven‐ tions and abbreviations listed in the Conventions and Abbreviations section (p. i). 38 Cf. Halliday and Matthiessen (2014: 680 f). 8.2.2 Contrast and concession Firstly, co-texts logically linked to NSIs by contrasting conjunctions and adverbs were examined in accordance with my framework for experiential context analysis. As has been shown, it is this type of formally marked functional relation between NSIs and their immediate co-text that occurs most frequently in the examined data. Table 8.9 below shows the identified process types, the verbs denoting these processes and their frequency of occurrence. 37 Contrastive links overwhelmingly put I in Role-1 mental processes (mostly cognition), while non-human third-person relational processes represent the second most frequent configuration. CONTRAST AND CONCESSION CONTRAST C O - T E XT F O L L O WIN G NSI S Process Role-1 Referents - - First person s. g. (I) To. Mental agree, appreciate (2), assume (2), believe (5), decide (2), discover, doubt, enjoy (3), experience, find (2), guess (3), hear (3), hope, imagine, know (4), learn, like (3), look at, love (3), notice, prefer, recall, research, see [understand] (2), see (4), suspect (2), think (17), understand (2), want to (5), wish (2), wonder (2), worry, like to 82 Modal assessments in mental ‘please’ clauses 38 - it does seem to me, it looks to me 2 Relational be (16): comfortable, confident, certain, diabetic, a fan of this little knife, impressed, married, more than capable of sorting it, on the periphery of the app, a private individual, shocked, a sole trader, a spy, sure (3) find myself (2): oddly refreshed, drawn to these shades feel: humiliated get [become] (2): stuck, unlucky 34 have (9): a conversation, a few years of experience, a tinker, a vivarium, experience, high cholesterol, impression, memory, lower back pain - own (2): a diablo red 106s, a Dennis - 224 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="225"?> 39 The instances of get referred to here are get stitches, get sleep and get it. The verb get is thus interpreted as being used in a sense similar to that of ‘acquire’, which Halliday and Matthiessen (2014: 236) consider a material clause type. Get it is used to mean ‘understand it’ by the speaker, that is, it is an instance of a metaphor through which an idea is conceptualized as an object. 40 [I] manage to get up running is categorized as material process here because the first verbal group manage to serve as an extension of the second verbal group get up, a semantic relation which is referred to as conation in SFL (Halliday & Matthiessen 2014: 572). CONTRAST AND CONCESSION need: the DotProduct - keep: an eye - Material change, create, cut in 1/ 2, cut out, deal with, go back, get (3),  39 go fossiling, go with the flow, have fitted, hesitate to come back, make up for sth [something], manage to get up running,  40 play, raise, smoke, start clearing, take [photo], use (3), visit, water, work (3) 28 Verbal offer, recommend, say (5), share, suggest (2), tell (2), thank 13 Behavioral sneeze 1 Existential be 1 - Subtotal 161 - Third person (non-human) - - [QUOTE], 150/ 90 or 150/ 80, 4-aco-dmt, a few thoughts, a jump from 19k to 32 in a week, a lot of ibuprofen, all the photos, an interface, an Italian Restaurant, cocaine, half of my plate, high gain amps, it [anaphoric ref] (2), it [if it did re-open], it [impersonal] (2), it [never having been a professional footballer], it [O/ A], it [to remove the fizz], Kermode’s love of the best films in the genre, magenta, mine [my photo], my brain, my guess, my understanding (2), North, Soho, some other tips, something weird about a person who can take 200 mg of trazadone without a tolerance within an hour and say they felt a “weed like high” and then not be knocked the fuck out, swear words and obscenity, that the XK engine is one of the best engines ever made, that thing, the 6 week outage, the active ingredients, the alternative, the Bosch FR, the final hurdle, the whole thing, the list of bits and pieces that I like to carry, the only way to deal with OCD, the palladium, the story, these [caves], these two [nutrition potions], they [e21], they [mac style icons], they [tits], things, this [advice], this [anaphoric ref] (3), this [behavior], this [experience], this [pic] (5), this [poke hack], this [woody cab], this stuff, those pretty flowers, UTR numbers, whippets, your offshore installation, your raison d’être 70 Goal as subject 2 my videos, sperm - 8.2 Negative self-identifiers and their formally related co-texts 225 <?page no="226"?> CONTRAST AND CONCESSION Relational be (34): 0 degrees, a Austro-Hungarian Uniform, a few thoughts, a generator isnt [isn’t] a portable/ moveable or stationary appliance, a good move, far above normal, far too much, good, helpful, mar‐ vellous, much more comprehensive programme, my hair trial, my MIL2B’s, my shaky understanding of the stiffness of tubes, nice, nifty, non starchy veg, not quite as the mail reports, odourless, one of outsmacking, opiates, outstanding, roughly right, scary, something special, that endurance training can itself cause Wenckeback phe‐ nomenon to occur, that my unit has some problem with the input sensitivity, the Bosch FR, the middle ground between white gold and platinum, the only thing from your report that may suggest MS, the place, the second most addictive substance, to correct the first and last name one to my full passport name, to learn how to change the way that we react to our thoughts 53 look (6): better and better, damn good, good, different, phenomenal, unsafe - have (4): live conductor, a letter at the end, adverse reaction, uses - look like (2): monster, salvia-divinorum - - seem (3): unusual, a summary of advice I have had before, popular - need, require (2): solid state pre amplification on top, the presence of water - sound like (2): PD, a great idea - take: 20 mins - Material bypass, cause, change, expand, hurt, lead, move, monetize, make, pick up, piss off, stop, supply 13 Existential an Italian restaurant, a few thoughts, something weird about a person who can take 200 mg of trazadone without a tolerance within an hour and say they felt a “weed like high” and then not be knocked the fuck out, swear words and obscenity, the 6 week outage 5 Verbal define 1 - Subtotal 72 - Third person (human/ animal) - - any dementia sufferer, anyone (2), Blake, Fred, GT towing, he, he [Captain of Liverpool FC], he [the tenor], many people who have been very helpful, MC fans, my partner, Sally Ann, several musicians, she, the bird, they [anaphoric ref] (2), they [Queen], who - Relational be (7): fine, linked to Amy’s life, DU student, expert, one of those bands that you occasionally realize wrote great song, son of William Archer, about the size of a seagull 8 226 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="227"?> 41 This is an instance of a metaphorical material process representing a verbal process (in the sense of confessing to guilt). 42 Verbs such as try and start (to do something) are treated as expansions in SFL. This means that they imply that the process they expand is actually being carried out (so try fixing here implies doing the fixing - whether this is successful or not) (cf. Halliday and Matthiessen 2014: 584). CONTRAST AND CONCESSION sound: good - Material animate, give stick, service, take [blame],  41 try fixing,  42 use, receive, win 8 Verbal encourage 2 Mental think 1 Existential many people who have been very helpful 1 - Subtotal 20 - Second person (You) - Material check, fault, get, go wrong, increase, invalidate, list, melt, pique, protect yourself, search, use (2) 13 Relational be: aenemic [anemic] have: POA for welfare need: more protein 3 Mental believe 1 - Subtotal 17 - First person pl. (We) - Relational get: lucky 1 - Subtotal 1 - Clause - - doing stats, getting your diabetes under control, to see new projects around, to despise the scum - Relational be (3): not that complicated, interesting, in the genes 3 Material help 1 - Subtotal 4 8.2 Negative self-identifiers and their formally related co-texts 227 <?page no="228"?> 43 Since proving something to others necessarily involves verbalizing what one considers evidence of a particular claim, I see prove as a verbal process. CONTRAST AND CONCESSION - Others - NP just an idea, maybe something like this 2 Exclama‐ tive Jeeeeeesus, thanks for the suggestions 2 - Subtotal 4 Total 279 C O -T E XT P R E C E DIN G NSI S Process Role-1 Referents - - First person s. g. (I) - Mental believe, consider, guess, hope (2), know (2), like (3), mean, prefer, read (2), see [watch], taste, think (7), understand (2), wonder 27 Modal assessments in mental ‘please’ clauses - it would surprise me 1 Material do exercise, edit, follow, put up image, show, sort, take (2), try to drink, use 10 Relational be (7): allowed, Christian, happy, hooked, new to the forum, sure, vegetarian become: healthier have: experience 9 Verbal say (3), need to consult, prove  43 5 - Subtotal 52 - Third person (non-human) - - better ways, heavy industry and pollution, it [advice], it [eclecti‐ cism], it [manual], it [PICTURE] (4), it [anaphoric ref] (4), it [truck], it [VW Golf Mk4], my room, my username, nothing I can say here, nothing much to do, printing, resources, some of them [photos of A rezia], the design, the pace, the term, these [PICTURE], this [bed], this [QUOTE] - Relational be (16): a user rights/ permissions thing in win 10, arguable, as strong as, a case of slipping into ‘gravity’ mode, ChamSys.be, complete rip-off, confusing, false widow spiders, good, King Alfred’s Cakes, 20 228 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="229"?> CONTRAST AND CONCESSION limited, meadow tick, my hobby, small, strikingly dissimilar, way to run programs look (like) (3): British, hoverfly larva, lime - seem: strange - Material change, disappear, handle, kill, start 5 Existential better ways, heavy industry and pollution, nothing much 3 - Subtotal 28 - Third person (human/ animal) - - Buzz, my HV, my mum, the pope, they [bees] - Verbal recommend, say, suggest, tell 4 Relational be: gentle little creatures 1 - Subtotal 5 - Second person (You) - Material lose 1 - Subtotal 1 - Total 86 - TOTAL CONTRAST 365 CONCESSION NSI A S C O N C E S S IV E C LAU S E : F O L L O WIN G C O -T E XT Process Role-1 Referents - - First person s. g. (I) - Mental enjoy, imagine, see, think 4 Verbal admit, answer, point out 3 Material cruise, visit 2 Relational deserve: one of those frames 1 - Subtotal 10 8.2 Negative self-identifiers and their formally related co-texts 229 <?page no="230"?> CONTRAST AND CONCESSION - Third person (non-human) - - it [privacy glass], the kindle - Relational fit into: elastic retainers suit: the car 2 - Subtotal 2 - Third person (human/ animal) - - he, schools, someone - Relational be (2): to Graceland, a recovering addict 2 Material hire 1 - Subtotal 3 - Second person (You) - Relational keep: friend 1 - Subtotal 1 - Total 16 NSI A S C O N C E S S IV E C LAU S E : P R E C E DIN G C O -T E XT Process Role-1 Referents To. - First person s. g. (I) - Material get, buy 2 Behavioral wait 1 Relational deserve: a Christmas 1 - Subtotal 4 - Third person (non-human) - - this dosage, the TVs - Material help 1 Relational have: a nice design 1 - Subtotal 2 - Total 6 230 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="231"?> 44 According to Koller (2020: 60), the sentence “The picture is on the wall” is an existential process, which means that on the wall, according to this interpretation, represents a circumstance. This, I think, is reasonable if we consider the wall to represent the place where the picture ‘resides’, that is, exists more permanently. In Halliday and Matthiessen (2014: 301), relations of being in, or to, particular places, are seen as outer experiences construed by relational process types. As an example, they cite “She’s in the dining room”. Accordingly, be [to Gracelands], representing a process of visiting, rather than existing in, a place, is listed as a relational process here. CONTRAST AND CONCESSION C O N C E S S IV E C LAU S E = F O L L O WIN G C O -T E XT Process Role-1 Referents - - First person s. g. (I) - Material play, register 2 Mental see, love 2 Relational be [to Gracelands] 44 1 - Total 5 C O N C E S S IV E C LAU S E = P R E C E DIN G C O -T E XT Process Role-1 Referents - - First person s. g. (I) - Mental hope, know, see 3 Relational have: PD 1 - Total 4 TOTAL CONCESSION 31 T O TAL C O N T R A S T A N D C O N C E S S I O N 396 Table 8.9: Transitivity analysis of contrastive and concessive co-texts of negative self-identifiers, listing referents, process types, verbs, and their frequencies. The analysis revealed that, in most cases, an NSI is preceded or - much more frequently - followed by a clause with I, that is, the speaker, in Role 1. The second most frequently occurring Role-1 participants are inanimate third-person NPs, that is, things and ideas. The table below provides an overview of the relative frequency of the Role-1 participants in clauses presenting a contrast or a concession to NSIs. 8.2 Negative self-identifiers and their formally related co-texts 231 <?page no="232"?> Role-1-participant type Frequency First person s. g. (I) 236 Third person (non-human) 104 Third person (human/ animal) 28 Second person s. g. (You) 19 Others (We, clausal subjects, NPs, exclamatives) 9 Total 396 Table 8.10: Role-1 participant types in contrastive and concessive co-texts of negative self-identifiers. Overall, of process types in clauses of contrast and concession preceding or following NSIs (396 in total), mental processes with I in the senser role occur by far the most frequently (121 instances, i.e., 30 per cent of all clauses and sentences linked to NSIs by means of contrasting and concessive conjunctions). The second most frequent participant-process configuration is non-human third-person NPs in relational processes: 77 cases, or 19 per cent of the contexts examined, are relational processes with inanimate third-person participants in Role 1. The third most frequent participant-process relation is first-person participants in relational (47) and material (44) processes. Categorized as “others” in this table are the few cases where an NSI is not followed by a process but, for example, by an NP, as in I’m no photographer … but jeeeeeeesus! Table 8.11 summarizes the distribution of Role-1 participant types in relation to process types occurring in contrastive and concessive co-texts of NSIs, distinguishing between preceding and following clause contexts. This overview allows us to see which participant-process configurations are most frequent in such environments and whether these tendencies differ depending on whether the co-text precedes or follows the NSI. 232 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="233"?> 45 Read across rows to see whether a given Role-1/ process configuration occurs more often in Pre vs. Fol position. Process Mental Mate‐ rial Verbal Exis‐ tential Rela‐ tional Behav‐ ioral To‐ tal Role-1 part. - Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. - First person s.g. (I) - 31 90 12 32 5 16 — 1 11 36 1 1 236 Total 121 44 21 1 47 2 Third person (nonhuman) - — — 6 12 — 1 3 5 21 56 — — 104 Total — 18 1 8 77 — Third person (human) - — 1 — 9 4 2 — 1 1 10 — — 28 Total 1 10 6 1 11 — Second person (You) - — 1 1 13 — — — — — 4 — — 19 Total 1 14 — — 4 — Total - 123 86 28 10 139 2 387 Others - 9 TOTAL 396 Table 8.11: Role-1 configurations in contrastive/ concessive co-texts of negative self-iden‐ tifiers, with frequencies in preceding (Pre.) and following (Fol.) positions. 45 Mental processes can commonly be differentiated according to whether they designate cognitive, emotive, perceptive or desiderative processes. As Table 8.12 shows, mental processes with I in the senser role in clauses and sentences contrasted with NSIs are predominantly processes of cognition: 8.2 Negative self-identifiers and their formally related co-texts 233 <?page no="234"?> Verbs in mental processes with I as senser Subtype Frequency Verbs Cognitive 75 agree, assume (2), believe (6), consider (2), decide (2), discover, doubt, experience, guess (4), imagine (2), it does seem to me, it looks to me, it would surprise me, know (7), learn, look at, mean, notice, read (2), realize, recall, research, think (27), understand (4), wonder (3) Emotive 19 appreciate (2), enjoy (4), like (6), love (4), prefer (2), worry Perceptive 15 find (2), hear (3), see (9), taste Desiderative 12 hope (4), want to (5), wish (2), would like to see Total 121 - Table 8.12: Mental process types in contrasting co-texts of negative self-identifiers (I as senser). Considering the conceptual categories of the NSIs (as discussed in Chapter 7), with which these clauses and sentences are related by means of a contrasting conjunction or adverb, yields the following results: Conceptual categories To. Expertise (general) 25 Expertise (specific) 23 Professional 20 Expertise (expert) 9 Characteristics: evaluative 5 Preference (+ fan): person 5 Preference (+ fan): general 4 Roles: forum-related 3 Characteristics: general 3 Preference (+ fan): activity 3 Preference (+ fan): visual aspects 3 Habit: routine 3 234 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="235"?> Preference (+ fan): product 2 Characteristics: health/ ill health 2 Preference (fan): product 2 Preference (fan): style 2 Habit: substance 2 Characteristics: social 1 Preference (+ fan): nature 1 Ideological 1 Usage/ consumption/ ownership 1 Habit: food/ drink 1 Total 121 Table 8.13: Categories of negative self-identifiers contrasted with mental processes (I as senser). Overall, 77 of 121 (i.e., 63%) NSIs contrasted with mental (as we have seen, mostly cognitive) processes with the speaker in the senser role are from the conceptual domain of “expertise and professionalism”. This is not surprising, given that these categories were generally the most frequent. In light of what I have said in the previous chapter about deand re-expertization, though, this appears interesting insofar as stating one’s view immediately before or after using a disclaimer of expertise can be interpreted as acknowledgement of experts and, at the same time, as subversion of potential implications of the speakers’ lay status in the sense of not being able or feeling authorized to comment on a given subject. As stated earlier, the second most frequent type of process formally contrasted with NSIs are relational processes with a non-human third-person subject Role 1 (77 of 396, that is, 19% of the sentences formally contrasted with NSIs). Relational processes can be classified into identifying processes (where a is the identity of x; a corpus example would be mine is my MIL2B’s) and attributive processes (where a is an attribute of x, as in this story is not quite as the Mail reports it). Both types of relational process can be further subcategorized into intensive, possessive and circumstantial processes, which are exemplified in Table 8.14 below (examples from my corpus are represented in italics). 8.2 Negative self-identifiers and their formally related co-texts 235 <?page no="236"?> 46 Example constructed for illustration; all other examples from the corpus. 47 Table adapted from Halliday and Matthiessen (2014: 265). 48 It should be added that the transitivity framework as proposed by Halliday and Matthiessen (2014), is in fact much more intricate than what I have presented here, postulating much finer-grained subcategorizations of attributive and identifying processes. For the purposes of this study, though, I am not taking SFL to its true ‘depths’, complementing categorizations according to functional grammar with less abstract, meaning-oriented, conceptual analysis. Attributive Identifying Intensive (x is a) they’re looking better and bet‐ ter half of my plate is non starchy veg Possessive (x has a) UTR numbers don’t have a let‐ ter at the end mine is my MIL2B’s Circumstantial (x is [preposition] a) a Kindle fits into elastic retain‐ ers the meeting is at 12  46 Table 8.14: Relational processes in contrastive co-texts of negative self-identifiers, sub‐ divided into intensive, possessive, and circumstantial types. 47 I further categorized the data semantically to find if particular meanings frequently occur in data representing this participant-process configuration. This was done by initially differentiating among relational processes with non-human third person participants drawing on (but departing from) this framework. 48 Attributive Identifying Evaluation 30 Identification 28 arguable, better and better, as strong as “Whatever you do don’t drink alco‐ hol! ”, confusing, damn good, different, far too much, good (3), (have) nice design, helpful, limited, marvellous, nice, nifty, not quite as the mail reports, outstand‐ ing, phenomenal, popular, roughly right, scary, small, so far above normal, some‐ thing special, strange, strikingly dissim‐ ilar, (suit) the car, unsafe, unusual a Austro-hungarian uniform, a case of slipping into ‘gravity’ mode, a complete rip-off, a few thoughts, a good move, a great idea, a monster, a much more comprehensive programme, a summary of advice I have had before, a user rights/ permissions thing in win 10, a way to run programs, a winchat, British, false widow spiders, hoverfly larva, King Alfred’s Cakes, lime, meadow tick, my shaky understanding of the stiffness of tubes, one of outsmacking, opiates, place where we all need open and honest opinions, salvia-divinorum, the most addictive and must harmfull [harmful] drug, the only thing from your report that may suggest MS, the only VW listing, to correct the first and last name one to my full passport name, to learn how to change the way that we react to our thoughts 236 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="237"?> Description 9 Informative reference 5 a letter at the beginning, adverse reac‐ tion, odourless, PD, the presence of water, uses, fit into: elastic retainers, (have) live conductor, (have) solid state pre amplifi‐ cation on top [NAME].be, my hair trial, my hobby, my MIL2B’s, non starchy veg Spatial/ temporal specification 2 Claims 3 0 degrees, 20 mins that a generator isnt a portable/ moveable or stationary appliance the middle ground be‐ tween silver and platinum; that endurance training can itself cause Wenckeback phenom‐ enon to occur; that my unit has some problem with the input sensitivity Total 41 Total 36 Table 8.15: Values and attributes in relational processes with third-person non-human Role-1 participants. As can be seen, relational processes with non-human third-person participants largely fall into the categories of evaluation (30 instances) and identification (28 instances), which means that co-texts formally contrasted with NSIs frequently represent speakers’ assessments of particular objects or ideas, the potential interpretations of which are modified by NSIs. Take, for instance, the sentence I’m no fan of gta (except vice city) but this does look phenomenal, where the NSI is used to relativize the speaker’s evaluation of this (anaphorically referring to the computer game GTA 5) by fending off its potential implications, namely, that they are, indeed, a fan of this game. Interesting examples are those where speakers use NSIs to epistemically mitigate their - sometimes very precise and seemingly informed - categorizations of things. For instance, in I’m no expert but those pretty flowers look just like salvia-divinorum-flowers, the speaker assigns the referent to the class of Salvia divinorum flowers, using a technical term and thus - albeit tentatively, using the comparative look just like instead of the epistemically more certain are - providing expert information. There are also cases where negative self-identification with expertise reduces the certainty of already hedged opinions, such as I am no expert but it seams [seems] like a summary of advice I have had before. Here, the speaker classifies textual material presented in the anaphoric co-text as repeating things they already know, thus criticizing information which claims to be in their best interest. However, they do so very tentatively. Such cases of NSIs being contrasted with contexts representing personal evaluations and 8.2 Negative self-identifiers and their formally related co-texts 237 <?page no="238"?> assessments suggest that speakers carefully position themselves in relation to both subjective opinion and factual knowledge. Not being an expert does not stop speakers from evaluating advice, challenging the hierarchy traditionally involved in advice-giving (someone tells you what is best for you, thus speaking from a position of greater authority), or from using technical terms in ‘diagnoses’ which are, simultaneously, perspectivized by using expertise disclaimers. The motto, it seems, is “what I believe or know depends on what I am (not)” - and the effect of using NSIs to relativize both what is presented as knowledge and what is framed as opinion is that the conceptual boundaries between the two appear to conflate: if everything is relative, nothing can be accepted as certain. A minor category I distinguished here is “claims”, which contains clauses occurring as values in relational processes with participants such as my under‐ standing, thus representing - modalized - claims contrasted with NSIs. The effect of NSIs contrasted with such sentences, too, is to epistemically mitigate them. Negative self-identifiers are also used in contexts representing what I classified as “descriptions”, that is, attributive processes which relate the carrier to particular features, such as having a nice design or being odourless, as in the sentence I’m not a big fan of garlic but this stuff is meant to be odourless. Here, the NSI is used in the context of discussing the pros and cons of Kyolic Aged Garlic Supplements, modifying the speaker’s positive assessment of the supplements’ lack of garlic smell and thus positioning the speaker in relation to a consumption preference/ recommendation. Finally, there are some cases where identifying relational processes are used to provide information about or to show something, as in My username is [NAME], but I’m not a developper [developer], or I’m not an OM but this was my hair trial. In these two examples, the NSI is used to contrast the speaker with a category that is considered relevant for or implied by the textual material with which it is coordinated. Example 8.16 below represents the exchange within which the NSI I’m not a developer occurs and shows the source of this identity claim, namely a false identity ascription implied by speaker A’s request. Upon A’s request to add two functions (to a programming forum where features and information are exchanged), explicitly addressed to B, the latter positively identifies with their username, but contrasts this affirmation with what A’s request implicitly presupposes about B: namely, that they can fulfill the request. Pragmatically speaking, the NSI here serves to alert A to the non-fulfillment of the preparatory condition required for the request to be performable (namely, that B is a developer and can thus add the features). 238 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="239"?> 8.16. A: Probably my two most used functions setting up a busking show on other consoles.. Rate Masters and Speed Masters. Please add this soon, [NAME]! - B: Feature request should be posted in the bug tracker! My username is [NAME], but I’m not a developer In the second example (8.17), the speaker negatively self-identifies as OM, that is, already married. She contrasts this with her own status (‘on the market’), clarifying that the picture she shares is from a hair trial rather than her actual wedding hairstyle. The NSI thus functions to acknowledge her lack of eligibility for full participation while still claiming the right to “join in” the discussion. 8.17. A: I thought I knew what I wanted for my hair but I’m having second thoughts and wanted to see some other styles. Can you show me what you did? (you can pixelate your faces if you want! ) - B: I’m not an OM but this was my hair trial and how I’ll be having my hair, hope you don’t mind me joining in! In summary, the analysis presented here showed that the co-texts contrasted with NSIs mostly represent (cognitive) mental processes with the speaker in the senser role, and that these are most often contrasted with NSIs from the conceptual domain “expertise and professionalism”. This indicates that NSIs examined here are most frequently used as epistemic disclaimers contrasted with the speaker’s opinions and claims. The second analysis looked at the second most frequent experiential type of sentences contrasted with instances of the structure, namely inanimate third-person Role-1 participants in relational pro‐ cesses. It showed that these co-texts most often express speakers’ assessments and ‘diagnoses’, thus fulfilling similar functions as mental process contexts contrasted with the structure. Hence, these results support the finding that modification of opinions and claims is the most prominent function of NSIs. 8.2.3 Cause and consequence In this analysis, I approached NSIs followed or preceded by cause-consequence conjunctions similarly to those examined in contrasting or concession relations in the previous section. While NSIs in contrastive contexts primarily serve as hedging devices, those in cause-consequence contexts likely fulfill different functions. Specifically, this section focused on various types of relations between NSIs and their immediate co-texts, categorized using the transitivity framework. 8.2 Negative self-identifiers and their formally related co-texts 239 <?page no="240"?> 49 Find can be seen as a mental-perceptive process being used to metaphorically represent a mental-cognitive one (cf. Koller 2020: 68). • NSI as a cause, as in a. I can’t really speak for FF as I haven’t been a member for years and b. I’ve never been a smoker so don’t know how difficult that would be c. I’ve never been a smoker so don’t know how difficult that would be • NSI as a consequence, as in a. I’ve been involved with electric drive systems since the 1970s so I am no novice and b. I’ve not been a frequent user lately as have been busy The most frequently occurring cases are NSIs serving as cause, preceded or followed by co-texts representing their consequences, that is, cases (a) (41 sen‐ tences) and (b) (93 sentences). Speakers in my corpus thus often explain what not belonging to a particular identity category means and use NSIs as explanations for particular states of affairs. Table 8.16 presents the distribution of process types and associated verbs in the immediate co-text of NSIs occurring in causal or consequential relations, classified according to the transitivity framework. As with the contrastive/ concessive analysis above, frequencies are given separately for co-texts preceding (as cause) and following (as consequence) the NSI, and are further broken down by Role-1 referents. C AU S E A N D C O N S E Q U E N C E F O L L O WIN G C O -T E XT A S C O N S E Q U E N C E Process Role-1 Referents To. - First person s. g. (I) - Mental decide, find,  49 guess, know (5), like, listen, miss, opt for, prefer, think (4), view 18 Relational be: authoritative, concerned, damp behind the ears, interested, sure, surprised, unable, unsure, worried, wrong (2) pretend to be: expert got: feeling do without: chorus have: enough food, feeling, great girl, gross mistakes, idea (3) lack: knowledge own: a harness 23 Verbal admit, answer (2), comment (2), explain, guarantee, name, talk, vouch for 10 240 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="241"?> 50 Adding thoughts is, of course, used metaphorically to describe the process of verbally expressing one’s thoughts. 51 Base on is a grammatical metaphor in which a material process of physically putting something on a basis serves to figuratively refer to the cognitive-mental process of basing one’s arguments, claims etc. on solid ground. 52 This is a conceptual metaphor, where the material process of falling stands in for the mental-emotive one of (involuntarily) developing feelings for someone. 53 These, again, are instances of get being used metaphorically to stand in for a men‐ tal-cognitive process. 54 Used metaphorically to refer to the process of interpreting something in a certain way. C AU S E A N D C O N S E Q U E N C E Material add [thoughts],  50 base on,  51 fall [in love],  52 get [it] (2),  53 give, go in, have sth. made, wander, send 10 - Subtotal 61 - Third person (non-human) - - all repairs, any advice, any pointers in that direction, any way to fix this, HF start, it (8), my question, some errors within the hint text, that, the focus, these, what - Relational be: the best I could do, an essential feature, a huge board, one of those beasts, on the practical application, outside my experience, rubbish, the only types of things I know to try, the optimal way to fix this, title, useful, your choice 12 Existential all repairs, any way to fix this, it, some errors within the hint text (be) 4 Material go off 3 Subject as goal in agentless passives - - be set up, be received - - Subtotal 19 - Second person (You) - Mental bear with, forgive 2 Material take, take sth as (3),  54 attempt (sth.) 5 Relational have: conflicts need: additional advice 2 - Subtotal 9 - Clause - - to crave it, to say this - 8.2 Negative self-identifiers and their formally related co-texts 241 <?page no="242"?> 55 Shed light is a metaphorical material process, representing the act of facilitating understanding. 56 While instantiating a material process type, enlighten metaphorically encodes a change in awareness or understanding. 57 Both the discouragement (e.g., That put me off) and postponement (e.g., We put off the meeting) senses are metaphorical extensions of material action. The former conceptualizes affective distancing as spatial displacement; the latter treats time as space, allowing abstract events to be metaphorically ‘moved’. 58 The material process of putting somebody in their place is also metaphorical, expressing a social-evaluative act rather than literal physical movement. 59 Take something as metaphorically realizes a process of interpretation or stance-taking through a material clause structure. C AU S E A N D C O N S E Q U E N C E Relational be: unusual, easy 2 - Subtotal 2 - Third person (human) - - somebody - Material shed [light]  55 1 - Subtotal 1 - Others NP no advice 1 - Subtotal 1 Total 93 P R E C E DIN G C O -T E XT A S C O N S E Q U E N C E Process Role-1 Referents - - First person s. g. (I) - Mental appreciate (2), choose, hope, know (3), learn, like, relate, think, understand 12 Relational be: impressed, sure (2), unable to get more stock to sell, unclear, undecided have: idea 7 Verbal ask (2), answer, prove, say, speak for 6 Material do, do poorly, enlighten sb.,  56 post, put off,  57 put sb. in their place,  58 register, take sth. as  59 8 Subtotal 33 242 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="243"?> 60 The expression get to grips with functions as a metaphor for coping with or gaining an understanding of how to approach a task or situation. It is categorized as a relational process here, based on an interpretation of get to as meaning ‘coming to have’ (a grip). 61 As already explained above, come down is used metaphorically here to describe the process of feeling the effect of drugs wearing off. C AU S E A N D C O N S E Q U E N C E - Third person (non-human) - - it (decision), this (advice), nothing much to add, that (the forum is shit), the same - Relational be: no-brainer, the best I can do apply to: Heinz’s HP sauce 3 Behavioral stand 1 Existential be (nothing much to add) 1 Subtotal 5 - Third person (human) - - they - Verbal offer 1 Subtotal 1 - Clause - - to ask me, to go see him - Relational be: difficult, pointless 2 - Subtotal 2 Total 41 F O L L O WIN G C O -T E XT A S R E A S O N - First person s. g. (I) - Relational be: busy, too little, used to old ones get: to grips  60 4 Mental feel, think (2) 3 Material pay for, take up 2 Behavioral come down  61 1 - Subtotal 10 8.2 Negative self-identifiers and their formally related co-texts 243 <?page no="244"?> C AU S E A N D C O N S E Q U E N C E - Third person (non-human) - - contract of employment, ear buds, much room for misunderstanding, they (bottles), they (hydration packs) - Existential be (2) (contract of employment, much room for misunderstanding) 2 Material fall out, bounce out 2 Relational look: clumsy 1 - Subtotal 5 - Third person (human) - - they (the local branch of the Alzheimer’s Society) - Relational be: on doorstep 1 - Subtotal 1 - Other: NP - - (because of) their ultra-thin sidewalls, the disastrous start 2 - Subtotal 2 - Second person (You) - Mental know 1 - Subtotal 1 - Clause - - eating just 800 cals a day - Relational be: hard 1 - Subtotal 1 Total 20 244 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="245"?> 62 In get popped out, although I is the grammatical subject, it is the goal of the material process, not the actor. 63 Read across rows to see whether a given Role-1/ process configuration tends to occur more often in Pre vs. Fol position C AU S E A N D C O N S E Q U E N C E P R E C E DIN G C O -T E XT A S R E A S O N - Process Role-1 Referent - - First person s. g. (I) - Mental focus on 1 Relational be: involved with electric drive systems 1 Material get: popped out  62 1 - Subtotal 3 - Third person (non-human) - - my cruise pace - Relational be: about 9.20 minutes miles 1 - Subtotal 1 Total 4 T O TAL C AU S E A N D C O N S E Q U E N C E 158 Table 8.16: Transitivity analysis of causeand consequence-related co-texts of negative self-identifiers, listing referents, process types, verbs, and their frequencies. For ease of reference, the results of the analysis presented above are summarized again in Table 8.17. 63 8.2 Negative self-identifiers and their formally related co-texts 245 <?page no="246"?> NSIs as Cause Process - Men‐ tal Mate‐ rial Verbal Exis‐ tential Rela‐ tional Behav‐ ioral Total Role-1 part. - Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. - First Person s. g. (I) -- 12 18 4 10 6 10 — — 7 23 — — 90 Total 30 14 16 — 30 — - Third person (nonhuman) - — — — 3 — — 1 4 3 12 1 — 24 Total — 3 — 5 15 1 - Second person (You) -- — 2 4 5 — — — — — 2 — — 13 Total 2 9 — — 2 — - Clause - — — — — — — — — 2 2 — — 4 Total — — — — 4 — - Third person (hu‐ man) -- — — — 1 1 — — — — — — — 2 Total — 1 1 — — — - Other - — — — — — — — — — — — — 1 Total — — — — — — 1 Total - 32 27 17 5 51 1 134 - NSIs as Consequence Process - Men‐ tal Mate‐ rial Verbal Exis‐ tential Rela‐ tional Behav‐ ioral Total Role-1 part. - Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. Pre. Fol. - First person (I) - 1 3 1 2 — — — — 1 4 — 1 13 Total 4 3 — — 5 1 - Third person (nonhuman) - — — — 2 — — — 2 1 1 — — 6 Total — 2 — 2 2 — - 246 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="247"?> Second person (you) — 1 — — — — — — — — — — 1 Total 1 — — — — — - Clause - — — — — — — — — — 1 — - 1 Total — — — — 1 — - Third person (hu‐ man) - — — — — — — — — — 1 — — 1 Total — — — — 1 — - Other - — — — — — — — — — 2 — — 2 Total — — — — — — 2 Total - 5 5 — 2 11 1 24 Table 8.17: Role-1 participant and process types in causally related co-texts of negative self-identifiers, with frequencies in preceding (Pre.) and following (Fol.) positions. Overall, as shown in Table 8.18, processes with the speaker as Role-1 participant are most frequent in the examined co-texts again, followed by non-human third-person subjects. Role-1-Participant Type Frequency First person (I) 103 Third person (non-human) 30 Third person (human) 3 Second person (you) 14 Clause 5 Other 3 Total 158 Table 8.18: Role-1 participant types in causal co-texts of NSIs. As Table 8.18 shows, Role-1 participants in causal co-texts are predominantly first person (I: 65%), followed by third-person non-human subjects (19%); third-person human subjects are rare (2%). This pattern suggests that speakers most often construe negative self-identification as a cause for their own subsequent stance or assessment. The resulting consequence clauses are chiefly 8.2 Negative self-identifiers and their formally related co-texts 247 <?page no="248"?> mental and relational; notably, many clauses that are formally relational func‐ tion to report mental states (e.g., modal assessments such as be sure/ unsure or nominalizations such as have an idea). Among third-person subjects, non-hu‐ man referents (things/ ideas) overwhelmingly dominate, and in these cases the consequence is typically a relational clause assigning an attribute, e.g., I’m no writer, so it’s probably rubbish anyway. In short, negative identification is frequently framed as a reason for (not) having particular assumptions (mental processes) and for ascribing attributes to things (relational processes); for example: I have never been a bridesmaid so have no idea! ; I’m no writer, so it’s probably rubbish anyway. Regarding the question of which conceptual categories of identifying NPs occur in mental processes with I as the senser, most come from the superordinate domain “expertise and professionalism” (13 instances overall), followed by those from “roles” (6 instances) and “preference” (6 instances), with the remainder spread across other conceptual domains. Table 8.19 shows the full breakdown by category and subcategory for NSIs occurring in causal relations. Category Subcategory Frequency Expertise & professionalism - 13 - Professional 6 - (+ expert): specific 3 - (+ expert): general 2 - (expert): specific 2 Roles - 6 - Forum 3 - Business 3 Preference - 6 - (+ fan): product 3 - (+ fan): activity 1 - (+ fan): person 1 - (fan): product 1 248 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="249"?> Characteristics 5 - Health/ ill-health 3 - Evaluative 1 - General 1 Habits - 3 - Food/ drink - - Routine 1 - Substance 1 Ideological — 1 Total - 34 Table 8.19: Categories of negative self-identifiers causally related to mental processes (I as senser). 8.2.4 Addition Regarding clause-external co-texts related to NSIs by additive conjunctions, the following table demonstrates that additively connected co-texts in most cases follow the matrix clause. The process type configurations involving Role-1 participants that occur most often are I as the senser in mental processes and as the carrier/ token in relational processes (again, relational processes here should be taken with a grain of salt, because many of them are used to express processes of thinking, such as be aware/ sure, or emotive processes, like be disappointed, unimpressed). In a total of 22 cases, inanimate third-person subjects take Role 1 in relational processes and again, it is mainly judgments which are passed (things are evaluated, e.g., as good, repulsive or average). Table 8.20 summarizes the distribution of process types and associated Role-1 referents in clause-external co-texts related to NSIs by additive conjunctions. As with previous tables, frequencies are given separately for co-texts following and preceding the NSI, with the majority occurring in the following position. 8.2 Negative self-identifiers and their formally related co-texts 249 <?page no="250"?> 64 Lean toward, while constituting a material process type, is used metaphorically to represent the mental-cognitive process of favoring one thing or idea over another. 65 If you pass something on, that is technically a material process; here, it is used to metaphorically represent the process of verbally relating information. ADDITION C O -T E XT F O L L O WIN G N E G ATIV E S E L F - ID E N TI F I E R S Process Role-1 referent To. - First person s. g. (I) - Rela‐ tional be (21): aware, Aladdin’s genii, disappointed, fan of TVRs, happy to stand corrected, in the chair at FFD, more than capable of wiring a house up, off the mark, one for writing reviews, privy to such information, regular swimmer, sure (5), that old, type 2, unimpressed, used to being blown up about sth., wrong feel (2): guilty, disinterested have (5): account, chance, history with Meth, luck, Mig/ tig welder own (3): a host of 26 inch wheel street, my Pi, SKY box lose: 3 kg gain: satisfaction stand: ready 34 Mental enjoy, experience, find (2), hate, intend, know (6), look for, prefer (2), read, regret, relate, see (2), seek, suspect, think (5), understand (2), want [to], wish, wonder (2), would like 34 Material battle my way, complete, do (2), eat, get, go out, install, label, lean toward,  64 make, open, pass on,  65 play (2), sell, set, sign up, walk 19 Verbal advise, answer, ask, speak, welcome 5 Subtotal 92 - Third person (human/ animal) - 5-stud to 4-stud converters, a lot of detail, all the shoes I ever bought, Corrie, everyone’s stroke, healthier version, historical reasons, it, it [if you eat a diet that is almost totally carby], my name, my size 10 feet, Neuroshima theme, no two strokes, skis, storylines like this, it [that my fiance said he didn’t like makeup], it [that so many called high end production folders come out with this steel], it [the c word], the emails, the information above, the last step, the thoughts and delusions, the underactive thyroid discovered at my pre-op, this [anaphoric ref] (2), this issue, this track, times (2), to upset my pattern, what you do for your children, yogurts I eat - Rela‐ tional be (18): a litgant [litigant] in person, anyone one this thread is either, average, awful, beyond my capabilities, Bez, cobbled together, different, disappointing, easy to describe, full or low fat, good, one of the most dangerous, promising anything, repulsive, same, your call 22 250 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="251"?> ADDITION cost: less lose: appeal take: much have: adverse effect Material change (4), come, help, seal, slaps, sort 9 Existen‐ tial a lot of detail 1 Subtotal 32 - Third person (human) - - Amazon, Amazon Prime, anyone on this thread, Chelsea, most people in our family, my husband, someone, that boy, the captain of the football team, they - Rela‐ tional be (3): great sleepers, one of my favorite clubs, very bad deserve: hero have got: money 5 Material charge, give, help, invite 4 Mental consider 1 Subtotal 10 - Second person (You) Mental rely on 1 Rela‐ tional need 1 Verbal contact 1 Subtotal 3 - Impersonal (agentless passive) Material sort 1 Subtotal 1 TOTAL 138 C O -T E XT P R E C E DIN G N E G ATIV E S E L F - ID E N TI F I E R S Process Role-1 referents - First person s. g. (I) 8.2 Negative self-identifiers and their formally related co-texts 251 <?page no="252"?> 66 Get is used metaphorically in the sense of ‘understanding’ here, but following my approach toward categorizing metaphorical processes outlined in Section 8.2.1, it is listed as material process type. ADDITION Rela‐ tional be (9): a fan, a fan of wago type connectors, convinced, deputy, doctor, fed up, sorry (2), without the amp have: kids 10 Material Get,  66 register, stop, live 4 Mental hate, love, want [to] 3 Verbal predict 1 Subtotal 18 - Third person (non-human) - - control, new bond film, programming experience, that logic - Rela‐ tional be: part of the application questions look: good 2 Existen‐ tial control 1 Material work 1 Subtotal 4 - Clause - - waking up covered in urine - Rela‐ tional be: my case 1 Subtotal 1 - NPs - - good luck 1 Subtotal 1 Total 24 TOTAL ADDITION 162 Table 8.20: Transitivity analysis of additively related co-texts of negative self-identifiers, listing referents, process types, verbs, and their frequencies. 252 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="253"?> 67 Read across rows to see whether a given Role-1/ process configuration tends to occur more often in Pre vs. Fol position. Considering, again, the overall instantiations of the different process types with first-person Role-1 participants reveals that relational and mental pro‐ cesses dominate; for inanimate third-person participants, it is, again, relational processes which are clearly most prominent. Table 8.21 summarizes these distributions. Role-1 referent Process type Preceding Following Total First person s. g. (I) Relational 10 34 44 - Mental 3 34 37 - Material 4 19 23 - Verbal 1 5 6 Total - - - 110 Third-person NP (non-human) Relational 2 22 24 - Material 1 9 10 - Existential 1 1 2 Total - - - 36 Table 8.21: Role-1 participant and process types in additively related co-texts of negative self-identifiers, with frequencies in preceding (Pre.) and following (Fol.) positions. 67 Table 8.22 below reports the conceptual categories of identifying NPs additively related to mental processes with I as senser. As can be seen, a variety of conceptual categories appear in this context, though it is again NSIs from the domain “expertise and professionalism” which dominate (14 overall), followed by preference disclaimers (11 overall). 8.2 Negative self-identifiers and their formally related co-texts 253 <?page no="254"?> Category Frequency Professional 6 Preference (+ fan): product 6 Expertise (expert) 5 Characteristics: evaluative 3 Expertise: general 2 Preference (+ fan): food 2 Preference (fan): product 2 Role: forum 2 Usage/ ownership/ consumption 2 Role: trade 1 Characteristics: gender-specific 1 Characteristics: physical/ physiological 1 Activity/ expertise 1 Habit: substance 1 Preference (+ fan): visual 1 Total 37 Table 8.22: Categories of negative self-identifiers additively related to mental processes (I as senser). 8.3 Summary The analyses presented in this chapter address RQ2A by determining the experiential (ideational) meanings of co-texts with formal relations to the NSIs in my corpus. Table 8.23 summarizes the distribution of process types in 443 co-texts where I serves as the Role-1 participant, linked to NSIs by contrast/ concession, cause/ consequence, or addition. Overall, mental processes account for almost half of the examined cases (192 of 443; 43%) and are the most prominent process type in co-texts connected by contrast or concession. In co-texts related additively or causally to NSIs, relational processes constitute the largest category. This suggests that NSIs are most often formally - and thus conceptually - connected to processes of thinking and feeling. 254 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="255"?> Process Type Addition Cause/ consequence Contrast/ concession Total Mental 37 34 121 192 Relational 43 35 47 125 Material 23 17 44 84 Verbal 6 16 20 42 Total 109 102 232 443 Table 8.23: Process types in co-texts with I as Role-1 participant, formally links to negative self-identifiers by different conjunction types (n = 443). Across relation types, mental processes dominate in contrastive/ concessive contexts, while relational processes slightly outweigh mental ones in causal/ consequential and additive contexts; all other process types occur only at much lower frequencies. Regarding the overall proportions of conceptual categories of identifying NPs in co-texts with I as a participant in a mental process, NPs relating to “expertise and professionalism” occur most often. Of the 192 NSIs formally related to mental-process contexts, 103 (53%) contrast the speaker with a noun denoting a profession or a person with expertise. The second most prominent category is “preference” (38, i.e., 20%). Although disclaimers of expertise are somewhat more frequent in mental-process contexts (53%) than in the corpus overall (41%), this difference does not provide strong evidence of a closer association. Still, the results point toward a tendency for disclaimers of expertise and mental processes to co-occur in the data examined. It is not surprising at first sight that co-texts formally related to NSIs are predominantly mental processes with the speaker in the role of senser: after all, web forums are spaces for exchanging views and knowledge on subjects of shared interest, so talking about what one thinks and knows is hardly unusual. At the same time, it is striking that speakers emphasize their lack of expertise or authority immediately before or after expressing their views, since one might assume that forums, by their very nature, are not sites intended for expert discourse. As argued before, against this background, routinely highlighting non-membership in the group of experts appears significant. This suggests that even in lay arenas, participants remain oriented to expert authority and manage that orientation through the use of NSIs. Table 8.24 shows the distribution of conceptual categories of NSIs occurring in mental-process co-texts where I functions as the senser, with a clear concen‐ 8.3 Summary 255 <?page no="256"?> tration in “expertise & professionalism” and smaller but notable clusters in “preference” and “characteristics”. Category Subcategory Frequency Expertise & professionalism 103 - Professional 32 - (+ expert): general 29 - (+ expert): specific 26 - (expert) 16 Preference 38 - (+ fan): product 11 - (+ fan): person 6 - (fan): product 5 - (+ fan): activity 4 - (+ fan): general 4 - (+ fan): visual 4 - (fan): style 2 - (+ fan): food 2 Characteristics 18 - Evaluative 9 - Health/ ill health 5 - General 4 Roles 13 - Forum 8 - Business 5 Habits 10 - Routine 4 - Substance 4 - Food/ drink 2 256 8 Negative Self-identifiers and Their Immediate Co-Texts <?page no="257"?> Usage/ consumption/ ownership 3 Ideological — 2 Other — 5 Total - 192 Table 8.24: Categories of negative self-identifiers in mental process co-texts (I as … senser). Mental processes with I in the senser role were the most frequent partici‐ pant-process configuration across all three examined context types: contras‐ tive/ concessive, causal/ consequential, and additive. This subset of 192 instances of negative self-identification - together with formally related co-texts repre‐ senting relational and material processes with the speaker as Role-1 participant (125 and 84 instances, respectively) - was selected for a more detailed functional analysis, the results of which are presented in the next chapter. 8.3 Summary 257 <?page no="259"?> 9 The Discourse Functions of Negative Self-identifiers The previous chapter presented a general co-textual profile of NSIs, explor‐ ing how they are modified clause-internally and drawing on the transitivity framework to create an experiential profile of their clause-external co-texts. This chapter examines the discourse-pragmatic functions of NSI co-texts and asks whether these relate to particular conceptual categories of the structure. Together, the analyses address RQ2 by linking NSI types to patterned discourse functions in their local co-texts. The first analysis focused on NSI co-texts with specific formal relations identified as prominent in Chapter 8. The goal was to establish whether particular meanings with particular discourse functions occur together with specific conceptual categories of NSIs in patterned ways. More specifically, the analysis examined sentences and clauses that: • immediately precede or follow NSIs; • are linked to NSIs by a contrasting, causal/ consequential, or additive conjunction or adverb (the most frequent formal ties in the corpus, see Chapter 8); • realize mental (Section 9.1.1), relational (Section 9.1.2), or material processes (Section 9.1.3), the three most frequent types of experientially differentiated co-text. The decision to analyze these subsets was based on the assumption that formal and semantic similarities across these co-texts provide a good basis for finer-grained comparison. For each process type, a functional framework was devised prior to analysis; each subsection introduces this framework and then presents results. The second part of the chapter (Section 9.2) examines textual material preceding NSIs that is not formally linked to the structure. It functionally analyzes sentences and higher-level units to determine, first, whether the findings from formally related co-texts also apply to this different subset of data, and second, how NSIs interact with categories of co-text identified on functional rather than purely formal grounds. Section 9.3 then relates the analysis of NSIs as microlinguistic choices to broader issues of social relevance by presenting a qualitative analysis of two instances in their local discourse <?page no="260"?> contexts. Finally, Section 9.4 reviews the overall results of the chapter and concludes by addressing RQ2. 9.1 Functional analysis of formally related co-texts 9.1.1 Functional categorization of mental processes with I as the senser The finding that NSIs often co-occur with first-person mental clauses suggests a patterned use in epistemic stance contexts, constituting a linguistic choice in local discourse contexts with shared formal and semantic features. The next step was to find out what overall communicative functions sentences constituting mental processes with the speaker as the senser fulfill and how they functionally interact with the NSIs with which they are formally related (by contrasting, causal/ consequential or additive conjunctions or adverbs). To give an example of how NSIs - or rather their co-texts - are approached in the analysis presented here, two sentences that featured in the previous stage of analysis are: 9.1. I can’t relate to that because I’ve never been a schoolboy since Viz has been around 9.2. I’m no expert in these matters but i do know that digestion effectively starts in the mouth with your saliva For the analysis presented in the previous chapter, the underlined sentences in both examples were categorized as mental processes (based on the meanings of the verbs relate to and know). Example 9.1 features the speaker as the senser causally related to an NSI and 9.2 also features the speaker as the senser, but here contrasted with the NSI. Functionally, however, 9.1 and 9.2 are very different: while 9.1 could be described as indexing the speaker’s lack of understanding and experience and thus, perhaps, their lack of authority on a particular subject and/ or empathy with the addressee, 9.2 is a representation of the speaker’s opinion, a claim to knowledge in a particular field, so to speak. And then there are cases like I’m really not a fan of the above as I think its sub-optimal, where the clause projected by the cognitive verb think expresses a judgment of a previously mentioned referent or proposition (the above). To capture such differences, and thus get a fuller picture of the co-texts in which NSIs are used, this section presents a framework for functionally categorizing the data subset defined above. 260 9 The Discourse Functions of Negative Self-identifiers <?page no="261"?> Establishing and eventually quantifying functional categories is intricate insofar as one and the same utterance - as I argued earlier - may serve to index more than just one aspect of the communicative situation and may simultaneously fulfill various textual and interpersonal functions. A particular function, in turn, may be realized by more than just one form or structure. Besides the possibility of many-to-many form-function mappings, another reason why it is practically impossible to set up solid formal criteria for assigning sentences and clauses to functional categories is that function always takes the communicative situation and the recipients into account, which means that pragmatic phenomena can never be explicated by reference to linguistic forms only. This is why speech act theory usually sets up conditions rooted in the lifeworlds of language users rather than in language alone, vaguely defining, for example, advice as “telling you what is best for you” (Searle 1969: 67) and matching this overall function with linguistic forms prototypically used to realize it. To close the gap between entirely form-based, bottom-up functional analysis (which would be too rigid, ignoring cases where a function is realized other than by the predefined forms) and entirely top-down functional analysis (which runs the danger of being too unsystematic), my analysis, based on iterative qualitative analyses of the data, sets up formal features to make the categorization process maximally transparent, while acknowledging that such a framework is to a certain extent constructed, representing an attempt to bring to light tendencies observable in the data, rather than a definite, deterministic ascription of functions. Methodologically, I proceeded by broadly differentiating between intuitively different functions and then analyzing these broad categories in more detail to identify linguistic features shared by items in the respective categories. I then once again compared the category members against the postulated sets of features, weeding out utterances deviating from the detected patterns or changing the criteria for category inclusion if the resulting category turned out to be too heterogeneous and thus probably too broad. I repeated this process until I arrived at a set of features operationally defined as necessary and sufficient for category inclusion. Below, I briefly discuss the formal features that can be considered when functionally categorizing the material at hand. 9.1.1.1 Aspects to consider when categorizing mental processes Processes and participants All sentences considered in this analysis constitute mental processes, but - as already mentioned in the previous chapter - they can be distinguished 9.1 Functional analysis of formally related co-texts 261 <?page no="262"?> semantically according to whether they denote processes of perception, cogni‐ tion, volition or emotion, as well as grammatically according to whether their complement is an object NP (as in I love you), a clause (as in I think that I love you) or a verb (as in the ironic example I love revising PhD chapters). This alone makes it possible to distinguish examples from my corpus such as the ones below: 9.3. I can see small teeth at the front of the lower jaw. 9.4. [I] just hate taking medication. 9.5. [I] would like to reduce this dosage to a minimum when possible. Example 9.3 represents the speaker’s perception of a phenomenon, while example 9.4 expresses the speaker’s emotive attitude toward an activity. In 9.5, a desiderative verb (like) is complemented by an infinitival clause realizing a material process (reduce) with the speaker as the actor, which licenses classifying this sentence as “intention.” The infinitival complement can also construe the speaker as a non-agentive senser of a phenomenon outside their control; for example, [I] would like to see [it] given a go now expresses what the speaker desires, but they are not the actor of the desired event (to give [something] a go), merely the senser of the perceptual process (see), and thus not capable of bringing about the event. This means the sentence is better categorized as “wish” than as “intention”. Mental processes with I as the senser do not only take verb complements: they can also have NP complements, as in example 9.6, to the effect that the process the sentence is about is the speaker’s cognitive experience. Alternatively, they may project clauses, as in example 9.7: 9.6. I don’t understand all the problems. 9.7. [I] thought Boxing Day was Christmas day. Here, the verb think projects a clause representing the speaker’s thought, so the sentence is also - or even mainly - about the process represented by the clause complement. This makes an important difference for the function of the sentence as a whole. While the sentence in 9.6 serves to say something about the speaker’s understanding/ grasp of something, [I] thought in example 9.7 epistemically frames the proposition “Boxing Day is Christmas day”. In cases such as this one, where the mental process projects a clause, the overall function of the examined sentence crucially depends on the function of that 262 9 The Discourse Functions of Negative Self-identifiers <?page no="263"?> 68 Formation here refers to the in-game arrangement or layout of players or units, often shared by users to illustrate preferred strategies or configurations within the game environment. clause which in turn depends on the participants and processes it represents (as well as modality and polarity, but see further below). For instance, [I] thought Boxing Day was Christmas day can be considered to represent what the speaker believes because “Boxing Day is Christmas day” is an identifying process by which a non-human third person subject is identified as another one, that is, a claim about a third party is made. Similarly, attributive processes projected by cognitive mental process verbs, such as I think any of those scenarios sounds feasible, represent the speaker’s opinion on a particular subject. When the addressee - you - features as participant in a clause projected by a mental process verb, this can indicate that the function of the overall sentence is addressee-oriented, that is, that the sentence serves as advice, expression of empathy, (dis-)agreement and the like. For instance, example 9.8 has an interactive character by virtue of making a claim about the response of the addressee of the message, you, to content provided before or after the utterance, referred to by this. 9.8. I thought you might find this useful. The function of I thought, here, is to epistemically frame the utterance, metadis‐ cursively perspectivizing it (i.e., marking it as the speaker’s thought) and serving as a mitigating/ modesty device. Another case in point is I do understand where you are, where the noun clause complement refers to the (metaphorical) position of the addressee, expressing empathy. There might, however, also be cases where the addressee is the subject of the embedded clause and the sentence is not interactive in the sense just described. For example, imagine I told someone “I think you might suffer from stage fright”. Here, my statement, despite featuring the addressee as the senser in the embedded clause, serves to represent my knowledge/ beliefs as in example 9.7 discussed above. When the projected clause self-referentially describes the speaker’s commu‐ nicative behavior in the forum, as in example 9.9, the sentence as a whole can be considered to index aspects of the current interaction. 9.9. I thought I’d share my formation. 68 9.1 Functional analysis of formally related co-texts 263 <?page no="264"?> 69 See Halliday and Matthiessen (2014: 584) for a discussion of projecting clause nexuses with mental processes of desideration, whose subjects may or may not be identical to those of the clauses they project. In this example, the framing I think serves to modify the meaning of the process represented by the clause complement (I’d share my formation), meta‐ discursively signaling awareness and thus highlighting to other participants the process of sharing something. Tense and aspect Besides the meaning of the verb and its complements, another element that may be considered as a criterion for functional categorization of sentences coordinated with NSIs is their tense and aspect. Tense and aspect are elements of the verbal group system network (as discussed by Halliday & Matthiessen 2014: 410). For example, a desiderative verb in the present progressive tense with an infinitive verb complement, such as [I] am wanting to limit the function that the Vodafone hardware plays, describes a situation currently faced by the speaker and, thus, a context in which the NSI - in this case I’m not an advanced IT person when it comes to setting up networks - is relevant. By contrast, if the same sentence occurred in the past tense - [I] wanted to limit the function that the Vodafone hardware plays - it would rather be interpreted as an account of what the speaker did prior to posting to the forum. Likewise, the use of the present perfect tense as in I have seen a lot of things said and written about “bioidentical/ compounded” hormone indicates that the speaker is talking about their experiences, while I can’t view them clearly refers to a problem relevant at the moment of posting to the forum. The only functional differentiation that is possible based on tense and aspect is to distinguish between textual material depending on whether it refers to what is currently the case, what has been the case up to now and what happened prior to the moment of writing, so tense and aspect - like the other criteria mentioned here - can only be two among several distinguishing features. Scope of subject influence Another feature that makes a difference for the meanings of the sentences examined here is what I refer to as scope of influence, differentiating wishes from intentions: what distinguishes the sentences I hope he suffers for what he has done and I want to remove the facepacks I added is that the mental process verb hope in the first one projects a clause with another participant - he - as subject and senser, representing a state of affairs not under the speaker’s influence, that is, a wish. By contrast, wanting to do something means volition and thus, typically, intentionality. 69 264 9 The Discourse Functions of Negative Self-identifiers <?page no="265"?> Polarity and modality Finally, the functions of the examined sentences depend on their polarity and modality: for instance, one of the reasons why example 9.10 is classified as “problem” and example 9.11 as “perception” in the framework I devised is the polarity of the clause (not being able to view something is taken not to be in the speaker’s interest). 9.10. I can’t view them clearly. 9.11. I can see small teeth at the front of the lower jaw. However, it is not only clause polarity that makes a categorial difference but also the object (thus, if negated - I can see no small teeth, for example - 9.11 would still be interpreted as an analytical description of a picture the speaker is looking at). This shows once again that language function is the result of a variety of factors and, thus, that categorizing authentic language in use needs to take multiple factors into account, the weighting of which may differ from case to case. In this way, I think, linguistic categorization is a matter of identifying family resemblance (Geeraerts 1989) rather than of comparing data against a fixed set of features. In other words, as stated earlier, I believe that justifying functional categori‐ zation by referring to the presence of particular forms can make an analysis more transparent, but given that one form may fulfill many different functions depending on contextual parameters, and that one function can be realized through many different forms, it is unrealistic to account for function by means of a fixed combination of forms. As Kuzar (2012: 14) puts it, “a construction is not an entity pre-existing its assessment”, but “a form that is extracted by way of categorization from the substance of language for particular descriptive purposes, using relevant conceptual tools”. The advantage of the form-based cat‐ egorization applied here is that it draws on the similarity of co-texts previously grouped according to their experiential functions. This provides a systematic starting point for exploring finer functional distinctions - for instance, clauses beginning with I think are structurally constrained in ways that make it easier to anticipate what may follow and what communicative functions such patterns can serve. 9.1 Functional analysis of formally related co-texts 265 <?page no="266"?> 9.1.1.2 Functional categories of mental processes The following table presents the functional categories used to differentiate between co-texts coordinated with NSIs and representing mental processes. It shows how these categories are distinguished in terms of the meaning of the mental process verb, the possible verb complements, as well as the other aspects just discussed, providing concrete examples from the corpus. Category Mental process meaning Verb comple‐ ment Other aspects considered Examples Knowledge/ opinion Cognitive, perceptive Sen‐ tence/ NP Sentence enco‐ des internal state or claim to knowledge I think, I have experienced, I see the benefit of it My red palms must be as‐ sociated with my cancer That different parts of the world may have different properties… Knowledge/ understand‐ ing refer‐ ence Cognitive Intransi‐ tive/ NP/ sentence Includes episte‐ mic disclaim‐ ers, questions of grasp or ability Just guessing really, I don’t know the pros and cons of this item, I believe, I have a good grasp… Addresseeoriented Cognitive, percep‐ tive, emo‐ tive NP/ sentence Complement includes or im‐ plies secondperson refer‐ ence (e.g., you, share) I highly appreciate your re‐ ply, I know what you’re saying, I thought you might find this useful, I’d share my formation Perception Perceptive NP Present simple or progressive tense I can see small teeth at the front of the lower jaw Preferences/ habits Emotive NP Present simple, progressive or perfect tense I do like, I prefer 508s, the 52 neck pickup Wish Desidera‐ tive NP/ sentence Wish or hope directed toward things outside speaker’s con‐ trol I wish [them] the best, I hope he suffers for what he has done Intention Desidera‐ tive, emo‐ tive Infini‐ tive/ gerund clause Desire or dislike for an action within speaker’s control I want to remove the face‐ packs, I don’t like writing all the NH people off 266 9 The Discourse Functions of Negative Self-identifiers <?page no="267"?> Problem Cognitive, emotive, perceptive Intransi‐ tive/ NP/ PP/ Clause Inherently neg‐ ative stance; in‐ cludes evalua‐ tive/ attributive structures I worry about silly things, I can’t view them, I am find‐ ing it tricky… Background Cognitive Sentence Mental verb functions pri‐ marily as dis‐ course marker for stance I know I have two small children Unspecified — — Past-tense mental action verbs express‐ ing a state or default position I chose, I opted Decisionmaking Cognitive PP + NP/ gerund Reflects plan‐ ning or indeci‐ sion I might decide, I am think‐ ing of selling all three cop‐ ies Experience Cognitive, percep‐ tive, emo‐ tive, de‐ siderative NP/ sentence Past or perfect aspect; mental action verbs used to reflect internal states retrospectively I have not experienced many changes in my med‐ ication Table 9.1: Framework for functionally categorizing mental process co-texts (I as senser). The first category, “knowledge representation/ opinion”, comprises sentences with a mental process verb and a projected clause indicating an assumption (example 9.12) or a mental process verb with an NP complement expressing an opinion (example 9.13). 9.12. I think what is happening is this: […] 9.13. I see the benefit of it 9.14. I thought you might find this useful 9.15. I know how hard it is to pull yourself away from *anything* Excluded from this category are mental process verbs projecting “addressee-ori‐ ented” clauses, such as examples 9.14 and 9.15. While the former addressee-ori‐ ented clause is classified as such by having a second-person subject, the latter is identified as such by implicature: the illocutionary force of claiming to know 9.1 Functional analysis of formally related co-texts 267 <?page no="268"?> what the addressee is experiencing is to show understanding, not (just) to make a claim to knowledge. The category “knowledge/ understanding reference” contains sentences that are used to say something about the speaker’s own knowledge or ability. Linguistically, this either means using mental process verbs intransitively (9.16), with an NP complement (9.17), or using a mental process verb projecting a claim about the speaker’s ability or knowledge (9.18). I have added not-negations in square brackets in these examples from my corpus to indicate that in all three cases, polarity changes what is said about the (scope, kind or existence of the) speaker’s knowledge or ability. 9.16. [not] just guessing 9.17. I [don’t] know the pros and cons of this item 9.18. I [don’t] believe I have a good grasp of the laws of the game This means that this category includes mental processes which constitute the proposition rather than projecting one in an embedded clause. Co-texts classified as “knowledge/ understanding reference” thus function to explicate the speaker’s position toward a particular subject or to provide a self-assessment of their ability/ knowledge, sometimes referring back and forward in the ongoing discourse, for example to preceding text as in now I know why [x happened]. This category, too, excludes cases where the complement is addressee-oriented, as in 9.19, the function of which is not (only) to make a claim about the speaker’s knowledge of what is being said, but to signal understanding for the addressee. 9.19. I know what you are saying The category “perception” is relatively straightforward; the criteria for inclusion are the presence of a mental process verb of perception in the present tense and an NP complement. “Preference” and “habits” are defined as mental processes where an emotive verb in the present or present perfect tense expresses the speaker’s attitude toward a nominative phenomenon (e.g., I do like 508s a lot), or a phenomenon expressed by a verb complement (i.e., usually activities, I just love beachcombing). The category “background” comprises sentences with projected clauses describing the speaker’s current situation, as in I know I have two small children (where I know serves like a discourse marker, framing the projected clause in terms of stance) and sentences with mental action verbs in the past tense de‐ 268 9 The Discourse Functions of Negative Self-identifiers <?page no="269"?> scribing what has happened so far, that is, a status quo rather than an experience, for example, I chose ‘Other infringemet’[infringement]. The categories “wish” and “intention” include sentences describing a ‘desired situation’, usually by featuring desiderative verbs (9.20). Sentences assigned to this category also tend to be in the present progressive tense, as in 9.21. 9.20. I hope I won’t need both 9.21. I am wanting to limit the function that the Vodafone hardware plays Then, there are sentences categorized as “problem”, either because they contain an inherently negative mental process verb (e.g., worry), because they come with a negated modal (I can’t view them clearly), or because the speaker attributes a ‘problem’-adjective to a carrier (as in I’m finding it tricky to research…). The category “decision-making” includes cognitive verbs referring to decision-mak‐ ing (e.g., decide) or describing situations of decision-making, being used in the present progressive tense together with a gerund (e.g., I’m thinking of selling all three copies). There are a few borderline cases of sentences classified in the previous analysis as mental processes and thus featuring in the data set examined here which seem to occupy a middle ground between mental processes and material processes, like decide, research or discover. These, despite being processes primarily involving the mind, also display characteristics of material processes in that, unlike more prototypical mental processes, they could be answers to the question “What happened yesterday? ” (thus, the invented example I decided that I will move to Germany would work as an answer, whereas I thought that I would move to Germany can maximally serve as ‘preface’ to an event - e.g., but then, I changed my mind). Mental action verbs of this kind, occurring in the examined data in the past and present perfect tense (e.g., i researched and talked to a whole bunch of people), are categorized as “experience” to capture this difference in relation to ‘proposition-projecting’ verbs. Also included in this category are mental processes realized by verbs of perception in the past and present perfect tense (e.g., I’ve seen most of the films), as these also constitute experiences of the mind. 9.1.1.3 Results The following table shows how many of the examined co-texts (differentiated by their relations with the NSI - contrast, cause/ consequence, and addition) 9.1 Functional analysis of formally related co-texts 269 <?page no="270"?> were assigned to the functional categories just described (a full overview of the data assigned to these categories is, for reasons of space, provided in Appendix 2A). Functional category Contrast/ concession Cause/ consequence Addition Total Knowledge representa‐ tion/ opinion 55 5 12 72 Knowledge/ understanding reference 9 15 7 31 Preferences/ habits 13 5 6 24 Experience 17 0 1 18 Addresseeoriented 12 4 1 17 Intention 5 0 5 10 Background 1 2 3 6 Decision-mak‐ ing 2 2 0 4 Wish 4 0 0 4 Perception 2 0 1 3 Problem 1 1 1 3 Total 121 34 37 192 Table 9.2: Functional categories of mental process co-texts (n = 192). Totals reflect clauses, not tokens of NSIs; a clause is counted once under its formal link (contrast/ cause/ addition). An overview of the results of the functional analysis of contexts constituting mental processes with I as the senser - which are most often formally linked to NSIs - shows that 72 of 192 (37%) fall into the category “knowledge representation/ opinion”, serving to express speakers’ beliefs and opinions, as in I’d guess a telemark is something to do with the binding. This context type stands out as particularly frequent when formally contrasted with NSIs: 55 of 121, that is, 45% of contrastive co-texts constituting mental processes with I as senser fall 270 9 The Discourse Functions of Negative Self-identifiers <?page no="271"?> into this category, and 50 of these 55 contexts are contrasted with disclaimers of expertise and professionalism. This indicates that NSIs are characteristically used to mitigate the epistemic status of information provided by speakers and the beliefs they hold. The second most frequently represented category of formally related co-text is “knowledge/ understanding reference”. In 31 of 192 cases examined here, NSIs are used in co-texts where speakers reflect on their own understanding or ability regarding a particular subject or the ongoing discourse. More than half of the NSIs used in these contexts can be categorized as disclaimers of expertise and professionalism: In 16 of 31 contexts in this category, NSIs such as zoologist, hard-core programmer or simply expert are used. This indicates that speakers explicitly position the knowledge and expertise they refer to in relation to (various forms of) expertise. Interestingly, it is mainly co-texts formally marked as cause or consequence of NSIs which can be assigned to this functional category, that is, speakers often causally associate their knowledge and abilities with NSIs in their self-representation on forums. Rhetorically, this could work to justify potential limitations of the expertise speakers share on web forums, but it could also index that speakers, despite not being formally accredited experts, are aware of their knowledge and skills and thus project epistemic self-confidence. For example, although the NSI in I am not an expert in Normandy maps, so I based [mine] on GJS Close Combat Maps and books I’ve read about the battle positions the speaker as a layperson pointing out potential flaws in the map they have created, it allows them to describe how they autodidactically acquired the skills necessary to simulate the map in question. Indeed, the rest of their posting suggests that they are proud of what they have created: To this I add that the battle in BA be attractive and entertaining for both sides. I tried to simulate the map as I know… and can. Conceptual category of identifying NP To. Professionalism 6 Expertise (expert) 5 Characteristics: health/ ill health 3 Expertise (+ expert): specific 3 Role: forum-related 3 Characteristics: evaluative 3 Characteristics: general 2 9.1 Functional analysis of formally related co-texts 271 <?page no="272"?> Conceptual category of identifying NP To. Role: trade 1 Expertise (+ expert): general 1 Activity/ expertise 1 Usage/ consumption/ ownership 1 Characteristics: physical/ physiological 1 Habit: substance 1 Total 31 Table 9.3: Categories of negative self-identifiers in “knowledge/ understanding” co-texts (n = 31). The corpus example below is an example of this kind of knowledge negotiation. The speaker first relates their experience with a particular item, just to add they do not know enough to judge its quality. The speaker goes on to cite their bike mechanic using, again, technical terms like octolink and bb shell in doing so. Thus, professional authority is simultaneously acknowledged - the real mechanic is cited - and appropriated, as the speaker not only passes on the information provided by the expert, but also adds their own experience with the product in the sentence preceding the NSI. 9.22. I was pleased to be able to find the right part very easily, and it arrived speedily. I am not a mechanic so I don’t know the pros and cons of this item except that my bike mechanic says that the octolink is a bit outdated, and is always whingeing about the eccentric bb shell. The next category of co-texts formally coordinated with NSIs concerns “prefer‐ ences/ habits”. These contexts typically involve speakers distancing themselves from particular likes, dislikes, or habitual behaviors, often through explicit preference disclaimers. In 16 of the 24 instances identified here, the NSI specifies a preference (e.g., fan of cradles, Petrolhead, fan of narrow gauge modelling), with the remaining examples referring to personal characteristics, expertise, or habitual activities. 272 9 The Discourse Functions of Negative Self-identifiers <?page no="273"?> Conceptual category of identifying NP To. Preference (+ fan): product 7 Preference (+ fan): activity 3 Characteristics: health/ ill health 2 Expertise (+ expert): specific 2 Preference (+ fan): general 2 Ideological 2 Characteristics: general 1 Characteristics: gender-specific 1 Expertise (+ expert): general 1 Preference (+ fan): person 1 Preference (+ fan): food/ drink 1 Preference (+ fan): visual aspects 1 Habit: food/ drink 1 Habit: substance 1 Preference (fan): product 1 Total 24 Table 9.4: Categories of negative self-identifiers in “preferences/ habits” co-texts (n = 24). The use of negative identification with particular preferences in formally related co-texts describing the speakers’ preferences and habits suggests that one function of NSIs examined here is preference-specification in the negative. The following corpus example illustrates this well: 9.23. Folks, I’m looking for a holder or cradle to hold my iPhone 6S. Needs to be handy reached, but nothing too obtrusive. I’ve never been a fan of cradles really and usually prefer the OEM option but the specdock is the closest thing to factory and it’s a little low. Here, the speaker negatively identifies as a fan of cradles, only to talk about their preference for other options. Examples such as this are interesting because while speakers in my corpus frequently contrast themselves with expertise to indicate a lack of (epistemic) authority, when they negatively identify with preference 9.1 Functional analysis of formally related co-texts 273 <?page no="274"?> categories, this often has the opposite effect. To stay with the example, the speaker, by negatively identifying as a fan of cradles, presupposes knowledge of this specific sense of a (phone) cradle, when describing their preference. By using the technical terms (OEM option, specdock), they come across as an expert on the subject. Co-texts describing the speakers’ experience occur 18 times in the examined data; 17 of them are formally contrasted with NSIs, as in 9.24. 9.24. I am not a piping guy but have heard nothing but positives about the Plant module with regards to valvels [valves] and folanges [flanges] and such. Nine of these contexts relate speakers’ experience gained from reading (9.25), researching and talking to others (9.26). 9.25. I’m certainly no expert on HRT, but over the last 5 years or so I have seen a lot of things said and written about “bioidentical/ compounded” hormones. 9.26. I really am no ‘jean expert’, but I researched and talked to a whole bunch of people on whether to size up or stay true to fit. Regarding the relation between contexts of this kind and the conceptual categories of NSIs used, the sample of 18 instances of negative identification here is probably too small to identify patterns. Another 17 of the examined contexts could be classified as “addressee-ori‐ ented”, that is, they serve to signal aspects of the communicative situation. More precisely, these can be differentiated according to whether they metadis‐ cursively comment on the speaker’s own utterance, on other users’ utterances (in this case mostly to express empathy with what they say), or serve as speech acts such as requests and expressives. That speakers represent NSIs as a contrast to sharing something they have created (i.e., before or after co-texts categorized as “metadiscursive comment: self ” in Table 9.5 below) suggests that they discursively perform modesty when it comes to presenting what they know or have done, emphasizing their lay status and making their contributions highly tentative. For example, 9.27 is a case in point, and incidentally also shows that speakers in my corpus acknowledge the authority of experts in all kinds of fields, including postal matters. 274 9 The Discourse Functions of Negative Self-identifiers <?page no="275"?> Addressee-oriented co-texts To. Metadiscursive comment (self) 8 I thought I’d ask here first - [I] just wish to share my experiences when possible I hope my post helps you even just a little Hope this helps [I] thought I would point this product out I thought I’d share my formation I thought you might find this useful I didn’t want to read and run Metadiscursive comment (other) 5 I know what you’re saying - I can understand your worries I do understand where you are I know how hard it is to pull yourself away from anything I can see how hard it is Speech act: request 2 I would really appreciate some feedback on this one - Hope others can provide help Speech act: expressive 2 I highly appreciate your reply - [I] just wanted to thank you all Total 17 Table 9.5: Categories of addressee-oriented co-texts of negative self-identifiers (n = 17). As for the other categories and their frequencies, the analysis revealed that NSIs are formally related to co-texts which express the speaker’s intention or wish (overall 14 sentences, e.g., [I] am seeking advice and information about it), in which speakers describe a situation of decision-making (4 sentences, e.g., I might decide on a couple of nights to run during the week) or a problem they are facing 9.1 Functional analysis of formally related co-texts 275 <?page no="276"?> (3 sentences, e.g., [I] am finding it tricky to research equivalent posts abroad). This can - tentatively, given the small sample - be interpreted as pointing to a tendency for the structure “I + copula + NOT + indefinite NP” to be used when it is practically relevant, that is, when belonging or not belonging to a particular category appears to make a difference with respect to a situation representing some kind of ‘turning point’, either because of an impending event or because of desired change. For example, in the sentence I am no Mo Farah but I certainly don’t wish to finish last, a relation is established between not identifying as Mo Farah and, thus, as a real athlete, and the desired outcome of a situation, viz. the speaker’s participation in a marathon. Looking at the whole thread reveals that the post is about the speaker’s participation in 5000 or 10000 road races. Examining the conceptual categories of identifying NPs in NSIs related to co-texts classified as intentions and wishes shows that in eight cases, speakers negatively identify as an expert or professional, in four cases with particular preferences and disclaimers and in two cases with evaluative categories. This might indicate that reaching particular goals tends to be linguistically associated with expertise, but a larger sample of data would have to be examined to provide stronger support for this assumption. Finally, in six cases, NSIs are formally related to sentences that can be classified as providing background knowledge about the speaker. A (sad) example is 9.28. Here, the mental process phrase I know projects a clause describing the speaker’s general situation (she has two children), which is contrasted with negative identification as a proper mum. Thus, the speaker implies that having two children alone is not enough to qualify as a proper mother, but that particular behaviors or actions are required which she cannot do because of her illness. 9.28. I know I have two small children aged 3 and 7 but since dec I haven’t been a proper mum I’ve been ill […] In summary, the present section presented an analysis of sentences formally related to NSIs and constituting, or being framed by, a mental process with the speaker in the role of the senser. These were categorized according to a set of formal and semantic criteria, and it was examined whether particular functional categories of co-text tend to co-occur with particular conceptual categories of NSIs. The analysis showed, firstly, that speakers most frequently contrast NSIs from the conceptual domain “expertise and professionalism” with sentences representing their knowledge and opinion, mitigating the epistemic status of their claims. It was also found that speakers frequently causally relate NSIs 276 9 The Discourse Functions of Negative Self-identifiers <?page no="277"?> to sentences in which they refer to their own understanding and expertise, thus showing a high sensitivity toward the status of the information shared and negotiated. When talking about what they like and dislike, speakers in my corpus often negatively identify with preference disclaimers; one function of this usage of the structure is to specify preferences in the negative, that is, by saying what they do not like, speakers often in fact provide more specific information about what precisely they want. Negative self-identifiers were found to be formally coordinated and thus conceptually linked to sentences metadiscursively commenting on the ongoing interaction between the speaker and other people posting to the forum, serving mitigating purposes. That speakers use the structure in focus so often to mitigate their claims, but also to strategically manage the interpretation of their own interactive behavior on the forum, could indicate they perceive the information they share and negotiate online as epistemically as well as socially delicate. In other words, it appears that the use of NSIs in the examined online contexts does not only point toward struggles around (various levels and forms of) expertise, but also toward speakers’ attempts to discursively represent themselves as likeable and, thus, trustworthy, online personas. NSIs were also found to be formally related to co-texts describing speakers’ experiences, that is, accounts of what they have read or watched, and how they felt about it. This could be interpreted as pointing to a certain discursive tension between the static category of being and the more dynamic conception of experiencing, lending weight to the concept of (lay) experience. More data has to be scrutinized, though, to find if this observation holds true beyond the small sample examined here. Finally, the analysis revealed that NSIs occur in co-texts describing speakers’ wishes, intentions, problems and current situations. This might indicate that negative self-identification has less to do with speakers’ permanent self-concept than with what they deem relevant in relation to the situations they find themselves in at the moment of speaking, fulfilling rhetorical rather than merely assertive functions. All in all, the analysis showed that NSIs are frequently used in contexts where speakers reflect on their knowledge and abilities, which implies that the instances of negative self-identification examined here are a salient linguistic choice in the context of speakers talking about what they think they know and can do. The tendency for speakers in my data to represent themselves in terms of their authority could reflect their orientation toward notions of expertise based on a hierarchy between experts and non-experts (Williams 2014). These notions may be relevant in the immediate situational context: for example, in a discussion on a very technical topic, a speaker may index epistemic 9.1 Functional analysis of formally related co-texts 277 <?page no="278"?> uncertainty of their utterance using an NSI, thus at the same time signaling awareness that expertise, as a feature of the self-identity they enact online, is a contractual achievement (Bigi 2011: 69; Leppänen et al. 2014: 112) - whether or not what they say is deemed trustworthy depends on how other forum users perceive their credibility, and not just on whether or not they are, or pretend to be, accredited experts on something in the offline world. On the contrary, in the context of web forums, speakers’ own experiences and preferences might even be more important (Rudolf von Rohr et al. 2019), reflecting a tendency for education and knowledge to become redistributed and for traditional notions of expertise to be challenged (Beck 1992: 57). Thus, beyond the immediate situational context, the routine use of expertise disclaimers could be seen as reflecting speakers’ underlying conceptualizations about the social world more generally, in which risks, competing systems of expertise and struggles around trust play a key role. 9.1.2 Functional categorization of relational processes with I as the carrier/ token This section presents my analysis of relational processes with I in the role of the token/ carrier, which were revealed to be the second most frequent participant-process type configuration in sentences formally related to NSIs (Chapter 8). The goal of this analysis was to establish whether the tendencies observed in the data examined in the previous section also manifest themselves in the rest of the data, or if other uses of NSIs emerge as more frequent. As for the method used to analyze this data segment, I adapted the framework used for the analysis of mental processes presented in 9.1.1 and added new categories that could be identified as relevant in the process of qualitatively analyzing the data. Like the framework for categorizing mental processes, the one applied in the analysis presented here refers to the formal appearance, grammatical and semantic features of the linguistic elements examined. Thus, for instance, tense can be the decisive factor when differentiating between the corpus example I have been a spy, in the house of love (which would be categorized as “experience”) and the invented example I am a spy, in the house of love (which, because it is uttered in the present tense, would be classified as “personal characteristic”). Then, there are figurative idioms in the data, such as being in the same boat. The functions of these multi-word units are based on an interaction of formal and lexical aspects and hence cannot be pinned down in purely syntactic/ grammatical terms or reduced to the figurative interpretation of one particular word. This, however, does not pose a problem insofar as this 278 9 The Discourse Functions of Negative Self-identifiers <?page no="279"?> framework, like the previous one, is data-based. This means that rather than introspectively creating and listing all possible formal realizations of functional categories, I specified those formal and semantic criteria in this framework which actually occur and can thus be used for classification of the data, whilst acknowledging that not all expressions can be described this way. The resulting framework, shown in Table 9.6, outlines the functional categories identified for relational processes with I as the carrier/ token, along with their formal features and representative examples. Category Description Formal appearance Example Personal characteristics General info about person (age, physi‐ cal or psychological traits, relations) Copula + adjective I’m not that old Copula + indefinite NP I am a Christian Copula + adjective [human relation] I am married to a T1 Have + indefinite NP I have a great girl Emotional states States of feeling Copula + adjective [‘intransitive’ emo‐ tion] couldn’t be happier Response - Reactions to exter‐ nal stimuli Copula + adjective [response] I am shocked to hear that… Present or past simple I was impressed with the documentary Present perfect I have been unim‐ pressed by Aimard Knowledge/ understanding - Referring to own knowledge or un‐ derstanding Copula + adjective [ability/ knowledge] I am comfortable with computers Copula + NP [exper‐ tise/ professionalism] I am not one for writing reviews Verb of possession + NP I probably had suffi‐ cient experience Knowledge representa‐ tion/ opinion - Stating what one believes Copula + adjective [certainty] + S/ PP I’m pretty sure you can get… Verb of possession + NP I have got the im‐ pression that… Copula + other adjec‐ tives I probably wouldn’t be allowed 9.1 Functional analysis of formally related co-texts 279 <?page no="280"?> Category Description Formal appearance Example Possession and entitlement - Stating what one possesses or de‐ serves Verb of possession (e.g., have, own) + NP I already have my‐ self a vivarium Deserve + NP I deserve one of those frames Preferences/ habits - Preferences or ha‐ bitual choices Copula + adjective [‘transitive’ emotion, preference] I keep finding myself drawn to… Copula + NP [prefer‐ ence] I am a fan of this lit‐ tle knife Problem - Situations per‐ ceived negatively Have + NP [negative or negated positive] I have had no luck… Copula + adjective [inability] I am unable to source products… Copula + not + adjec‐ tive [inability] I am not able to get more stock… Experiences - Present or past ex‐ periences Copula + adjective [experience] I am becoming healthier Copula + locative [metaphor] I have been on the periphery of the app Verb of possession + noun I have had lower back pain Addresseeoriented/ discourseinternal - Empathy or expres‐ sive functions di‐ rected at the ad‐ dressee Adjectives of compar‐ ison (same, similar…) I’m in the same boat as you Copula + adjective [addressee-directed emotion] I’m sorry Decision-mak‐ ing Describing uncer‐ tainty about up‐ coming situations Copula + adjective [certainty] + PP/ VP I’m really not sure what activity to do Forum-related characteristics/ attitudes Describing oneself in relation to the fo‐ rum Reference to forum (explicit or via ‘here’) I’m new to the forum Table 9.6: Framework for functionally categorizing relational process co-texts (I as carrier/ token). 280 9 The Discourse Functions of Negative Self-identifiers <?page no="281"?> 9.1.2.1 Results Table 9.7 below presents an overview of the numbers of sentences assigned to the functional categories as specified by the above framework. A full overview of the data assigned to these categories is, for reasons of space, provided in Appendix 2B. As can be seen, the most frequently represented co-texts are “knowledge/ understanding reference”, that is, co-texts in which speakers refer to their knowledge or understanding of particular subjects (26 instances), co-texts classified as “knowledge representation/ opinion”, where a relational process with the speaker as the carrier represents their position toward a projected proposition (as in [I] have got the impression that they don’t flex their immune protocols much) (20 instances) and sentences describing speakers’ experience (22 instances). These results indicate that, irrespective of whether sentences formally related to NSIs constitute mental or relational processes, they often represent what speakers think they know or do not know, on the one hand, and what speakers have experienced, on the other. Functional category Contrast/ concession Cause/ consequence Addition Total Knowledge/ under‐ standing reference 3 15 8 26 Experience 14 4 4 22 Knowledge repre‐ sentation/ opinion 10 6 4 20 Personal charac‐ teristics 4 1 8 13 Preferences/ habits 5 2 3 10 Possession and entitlement 4 1 5 10 Response 2 2 5 9 Problem 1 3 2 6 Emotional states 1 1 1 3 Addressee-orien‐ ted 1 0 2 3 9.1 Functional analysis of formally related co-texts 281 <?page no="282"?> Functional category Contrast/ concession Cause/ consequence Addition Total Forum-related characteristics/ attitudes 2 0 0 2 Decision-making 0 0 1 1 Total 47 35 43 125 Table 9.7: Functional profile of relational processes with different formal links to negative self-identifiers (n = 125). As summarized in Table 9.8, the distribution of NSI conceptual categories in “knowledge representation/ opinion” and “knowledge/ understanding reference” co-texts shows that 29 of 46 instances fall within expertiseand profession‐ alism-related domains. A similar pattern appears in “experience” co-texts: expertise/ professionalism NSIs account for about half of the 22 instances (10 of 22; see Table 9.9). This suggests that identifying as an expert or professional also matters when speakers talk about their own experiences. Given the small sample, however, firmer generalizations should await the larger dataset reported in the next section. NSI conceptual category Frequency Professional 11 Expertise (+ expert): general 11 Expertise (+ expert): specific 6 Roles: forum-related 4 Characteristics: evaluative 3 Characteristics: health/ ill health 2 Activities/ expertise 1 Characteristics: general 1 Characteristics: linguistic 1 Usage/ consumption/ ownership 1 Expertise (expert): general 1 Habits: food/ drink 1 282 9 The Discourse Functions of Negative Self-identifiers <?page no="283"?> Habits: substance 1 Roles: trade 1 Preference (+ fan): general 1 Total 46 Table 9.8: Categories of negative self-identifiers in “knowledge” co-texts (n = 46). NSI conceptual category Frequency Expertise (+ expert): specific 4 Expertise (+ expert): general 3 Professional 2 Activity/ expertise 1 Characteristics: physical/ physiological 1 Characteristics: general 1 Characteristics: social 1 Expertise (expert): specific 1 Preference (+ fan): visual aspects 1 Preference (+ fan): product 1 Preference (+ fan): general 1 Habits: food/ drink 1 Habits: substance 1 Ideological 1 Characteristics: evaluative 1 Role: forum-related 1 Total 22 Table 9.9: Categories of negative self-identifiers in “experience” co-texts (n = 22). Sentences constituting relational processes of possession, that is, those describ‐ ing speakers’ relations - and thus experience - with particular products might also be seen as pertaining to the superordinate category of “experience”. An 9.1 Functional analysis of formally related co-texts 283 <?page no="284"?> example of such a ‘product experience’ sentence is 9.29, where having owned something for a particular period of time is formally and thus conceptually related to negative self-identification as an expert. 9.29. Before I detail what I did, please note I am no expert at this, and have only owned my Pi just over a week. Further examples of proficiency or knowledge in particular areas being related to ownership are the following: 9.30. Now, I’m not a very good mechanic, and I dont have a lathe or a Mig/ tig welder. 9.31. I’m no Dennis expert but I do own one bought new by my Dad. 9.32. I’m not a SKY subscriber and don’t own a SKY box so I cannot comment on the details in relation to the settings on the box. Cases such as these are interesting insofar as they suggest that authority, in the examined data, while often explicitly referred to, is not construed as something ‘out of reach’ for laypeople discussing on forums. In contrast, in these examples, ownership alone qualifies forum users to be authoritative on the subjects of importance in these situations (an observation I will come back to in Section 9.2). In 19 cases, speakers talk about their preferences and habits (e.g., 9.33) or their emotive responses to persons, things and events (e.g., 9.34). 9.33. I am not a fan of modulation effects so do without chorus, tremelo and phaser. 9.34. I’m certainly not a linux expert so I’m constantly being surprised at the small nuances here and there. Interestingly, in these contexts, expertise disclaimers do not dominate; instead, a wide variety of preference disclaimers are used in 15 of 19 cases. This indicates that when speakers talk about what they like and usually do, negative self-identification with preference disclaimers (frequently constructions with fan) are preferred - a trend already revealed by the analysis of mental process sentences. The distribution of these semantic categories in “preferences/ habits” and “response” co-texts is shown in Table 9.10. 284 9 The Discourse Functions of Negative Self-identifiers <?page no="285"?> NSI conceptual category Frequency Preference (+ fan): product 3 Preference (+ fan): activity 2 Preference (+ fan): food/ drink/ substance 2 Preference (+ fan): person 2 Expertise (+ expert): specific 1 Preference (+ fan): it 1 Preference (+ fan): nature 1 Preference (+ fan): visual aspects 1 Habit: food/ drink 1 Preference (fan): activity 1 Preference (fan): product 1 Professional 1 Role: virtual 1 Characteristics: evaluative 1 Total 19 Table 9.10: Categories of negative self-identifiers in “preference/ habit” and “response” co-texts (n = 19). 9.1.3 Functional categorization of material processes with I as the actor In addition to considering mental and relational processes with I as the senser which altogether account for 316 instances of negative identification, that is, 34% of the examined data, I also examined the smaller category of material processes with the speaker in the actor role formally related to NSIs (84 instances) in more detail. The purpose of this analysis was to find if representations of what speakers do (rather than what they think or consider themselves to be) can be found to co-occur with particular NSIs with particular functions - possibly different from those revealed by the analyses presented so far. 9.1 Functional analysis of formally related co-texts 285 <?page no="286"?> 9.1.3.1 Aspects to consider when categorizing material processes Categorizing material processes with a first-person subject in the actor role is relatively unproblematic in comparison to analyzing mental and relational processes, because the former are less abstract. Compare, for instance, the straightforward meaning of the verb “use” in I am using a little more butter than I did and the attribute “used” and its relation to the overall function of the sentence in [I am] well used to being blown up about my erroneous guesses, which serves as a self-ironic comment by the speaker on their lack of knowledge, which, apparently, is often identified as such by others. Material processes as metaphors Of course, also with material processes, there are cases where categorization is difficult; what often constitutes a problem for categorization here are concep‐ tual metaphors, that is, meaning extensions of verbs technically representing material processes, like in the following examples: 9.35. I follow Daveo/ Boycie, but I’m not a sexist. 9.36. I’m no expert. But I’ll still go with the flow as long as I don’t have any tasty cargo or somewhere to be. 9.37. I was going to put up the image of people injecting their naughty bits with saline, but I’m not a complete monster. In example 9.35, the meaning of follow is constrained by the implied ‘virtual’ context, that is, what is meant is not the physical process of walking behind them, but probably rather the habit or preference of staying informed about these footballers’ lives and views, or the moral support for their beliefs. Similarly, to go with the flow in example 9.36 does, of course, not refer to an actual process of moving, but in this case appears to refer to spontaneous, flexible behavior of the speaker in situations where they do not have any other duties (here, in a computer game). In example 9.37, the literal meaning of putting up an image (on a wall) is extended so that it serves as metadiscursive comment on what the speaker does not want to describe (but does, nevertheless). Such cases were categorized by functional comparison and assignment to more fre‐ quently represented, clear-cut categories of sentences formally related to NSIs constituting material processes. Finally, there are cases of material processes in the examined data which are used figuratively to function like mental and relational processes categorized as “knowledge representation/ opinion” and “knowledge/ understanding reference” in the previous analyses, and which 286 9 The Discourse Functions of Negative Self-identifiers <?page no="287"?> were consequently assigned to these categories. For example, taking a guess at something serves to express one’s knowledge/ opinion on a particular subject, and getting an argument means understanding it. Meaning of verb and goal - but where to draw the lines? In many cases, the meanings of the verb and its object complement, that is, the goal, alone suggest a conceptual difference. As with all conceptual categorization, though, drawing the line between different conceptual categories is a matter of focus on particular meaning aspects at the expense of others, where the decision to focus on particular ones probably results from their prominence in the data - that is, recurrent meaning aspects are probably more likely to be selected as category-defining rather than ones which only appear in individual cases. Compare, for example, “use + butter” in 9.38, “use + it” in 9.39 and “use + accountant” in 9.40. 9.38. I am using a little more butter than I did. 9.39. [I] have never used it [an app called Fastpass Plus] till this morning. 9.40. i use an accountant who I pay a lot to. Even though all three belong to the same category by virtue of describing what the speakers use, one could argue that while the first and the third example both describe habits (employing present tense progressive and simple, respectively), the second one relates the speaker’s experience with using something (present perfect tense) (this is how these co-texts were categorized in my analysis); however, using butter is very different from using an accountant. Tense and aspect As previously discussed, tense and aspect play an important role for categori‐ zation in this analysis, because there arguably is a conceptual difference between [I] am battling my way through the install (the install has not been completed) and (the invented example) [I] have battled my way through the install (the battle has been won). Modality and polarity For the functional categorization of material processes in particular, modality and polarity play a role as well. To give an example in which modality makes a categorical difference in my profile, 9.41, by virtue of containing the modal would, is categorized as an addressee-oriented sentence (serving as a piece of advice by providing information about what the speaker would do, were they in 9.1 Functional analysis of formally related co-texts 287 <?page no="288"?> the addressee’s situation); another example is 9.42, where the modal can makes the sentence function as a statement about the speaker’s ability, rather than, say, a mere habit of walking long distances. An instance where polarity played a role for categorization is 9.43, where the inability to edit shipping prices (despite having been a merchant seller) indicates a problem faced by the speaker. 9.41. I’m no expert by any means, but i would definitely go back to the doctors. 9.42. I am not a severe case (yet) and can still walk a long way. 9.43. Still cannot edit shipping prices. BUT i have not been a merchant seller for most of this year. 9.1.3.2 Functional categories of material processes Table 9.11 below presents the functional categories of clauses and sentences representing material processes with the speaker in the actor role. Category name Description Tense and other for‐ mal specifications Examples Situative anchor Factual background information about the speaker’s current sit‐ uation, setting the scene for the rest of the posting - As I’ve already paid for a year’s subscription to the Telegraph; I as usual did not get the e-mail; I’m leaning to‐ ward Z1 and Cat 2 - Descriptions of gen‐ eral facts that relate to the speaker, but are outside their scope of influence - However I’m not a big seller and I don’t get many customer emails anyway Experi‐ ence References to experi‐ ences stretching over or repeatedly experi‐ enced during time spans from a past mo‐ ment up to the mo‐ ment of speaking Present or present perfect I am not a vat specialist or anything but have worked with it a lot on accounting systems; I usually get stitches when I bounce up and down too much - References to actions just completed and immediately relevant to the subject of the ongoing conversation Present perfect I’ve had a tinker with the levels on that photo 288 9 The Discourse Functions of Negative Self-identifiers <?page no="289"?> Measures Actions just taken by speakers to address a specific problem Present perfect, past I have cut the bread consumption in ½ Habits/ principles Activities done rou‐ tinely or out of princi‐ ple Present simple, willfuture; often marked by adjuncts of time (e.g., whenever I get injured, when some‐ thing hurts, regularly) or conditionals (e.g., if there’s a problem) If there’s a problem I’ll sort it then and there Ability/ inability Sentences describing the speaker’s ability or inability to carry out particular actions (lit‐ eral or metaphorical) May be marked by the semi-modal “can” Still cannot edit ship‐ ping prices; I don’t get this argument Addres‐ seeoriented/ discourse -internal Sentences metadiscur‐ sively referring to past, present, or future communicative (and thus addressee-orien‐ ted) actions by the speaker in the dis‐ course context of the forum as a site of so‐ cial interaction - I haven’t been a big flasher these last 20 months, but I will more than make up for it with my report (by negatively identifying as flasher, the speaker means that she has not posted many pictures recently); I only pass on advice Hypothetical actions by the speaker serving as advice Use of the modals “would” or “could” I would definitely go back to the doctors with your symptoms Hypothet‐ ical ac‐ tions Hypothetical actions by the speaker not serving as advice - “I would buy one even though I’m not a Chel‐ sea fan” Knowl‐ edge rep‐ resenta‐ tion/ opinion Material processes used metaphorically to refer to processes of thinking in utterances where speakers state what they believe - I don’t get it Knowl‐ edge/ under‐ standing reference Material processes used metaphorically to refer to the speak‐ er’s own knowledge or understanding - I took a wild guess at the 20’s/ 30’s Table 9.11: Framework for functionally categorizing material processes (I as actor). 9.1 Functional analysis of formally related co-texts 289 <?page no="290"?> 9.1.3.3 Results Table 9.12 provides an overview of the numbers of sentences assigned to each of these categories. Again, a table with all the data assigned to these categories is provided in Appendix 2C. Category Contrast/ concession Cause/ consequence Addition Total Experience 18 1 3 22 Habit/ principle 11 2 6 19 Situative anchor 3 4 9 16 (In)ability 1 3 2 6 Measures 3 2 0 5 Addressee-orien‐ ted/ discourse-internal 3 1 1 5 Hypothetical actions 3 1 1 5 Knowledge represen‐ tation/ opinion 1 0 0 1 Knowledge/ under‐ standing reference 1 3 1 5 Total 44 17 23 84 Table 9.12: Distribution of material process functions across formal links to negative self-identifiers (contrast/ concession, cause/ consequence, and addition; n = 84). Experiences as co-texts The analysis revealed that the most frequent functional category of sentences formally related to NSIs and representing material processes with the speaker in the actor role is “experience” (22 sentences overall). This category mainly includes sentences referring to activities and accomplishments by the speaker, such as cruise the Voyager, raise tadpoles, design frames etc. (12 instances). In eight cases, the sentence formally related to the NSI expresses a material process of buying, using, or consuming something, for example, take Senna, do E/ mdma, play with it. One sentence represents an involuntary experience, namely get stitches. Of the 12 sentences categorized as “activities and accomplishments” here, nine are formally contrasted with disclaimers of expertise, which is 290 9 The Discourse Functions of Negative Self-identifiers <?page no="291"?> interesting insofar as not being an expert of some kind is juxtaposed with having, in fact, done and accomplished things. The functional subcategorization of “experience” co-texts, with items assigned to each subcategory, is shown in Table 9.13. Category Items Frequency Activities and accomplish‐ ments I did create a profile; [I] have raised many of my tadpoles in a tank; [I] have only completed one marathon; I’ve quite regularly visited the forum and read peoples’ experiences with OCD; I’ve not even done a marathon; I’ve managed to get up to running 10 minutes; in the past have always worked; I never watered Argyrodermas; I have worked in a job where I was on my feet 12 hours a day; I have fallen in love with this color; I have cruised 4 times on Voyager; I have been “designing” Spa’s frames for a while now 12 Consump‐ tion/ usage I have taken senna before; I get better quality sleep without opiates; [I] have had orthotics fitted profes‐ sionally; [I] have been known to do E/ mdma when out clubbing; have worked with it a lot on accounting systems; have never used it till this morning; I’ve played Football Manager for many years; I have “played” with it a few times 8 Involuntary experience I usually get stitches when I bounce up and down too much; had to wait over 2 weeks for the money to be available 2 Total - 22 Table 9.13: Functional categorization of “experience” co-texts (n = 22). Habits/ principles as co-texts The second most frequently occurring type of contexts representing material processes is those subsumed under the category “habits/ principles”. On closer inspection, these can be differentiated according to whether they say something about practices the speaker generally engages or does not engage in or principles the speaker generally follows (e.g., I am not a bookseller and i actually do not label my own items), activities the speaker regularly does (e.g., I’ve never been a fast runner and quite often go out with the Durham City Harriers) or about the speaker’s consumption habits (e.g., I’m not a heavy smoker, but i have been smoking for about 3-4 years now). One sentence seems to fall in between constituting a habit and a general practice, describing a preference more than a certain habitual behavior, namely, I follow Daveo/ Boycie. Table 9.14 shows the 9.1 Functional analysis of formally related co-texts 291 <?page no="292"?> functional categorization of “habit/ principles” co-texts, with items assigned to each subcategory. Category Items Frequency Activities I still use it [the uni athletic track] whenever I get injured; When something hurts, I change the way that I run; I rarely play without the S1 on; I usually play solo; I’ll still go with the flow as long as I don’t have; Only ever go in when I clock up a few complaints; I like to wander around the huge expanse of heather moorland between St. Fillans and Loch Tay; Quite often go out with the Durham City Harriers; I deal with asbestos on a regular basis; I go fossiling a bit; I do some kind of exercise every day 11 Consump‐ tion habits I am using a little more butter than I did; I have been smoking for about 3-4 years now; I eat, sometimes, 2 boiled eggs daily 3 General prac‐ tice I actually do not label my own items; I use an accountant who I pay a lot to; I only sell on Amazon; If there’s a problem, I’ll sort it here and there 4 Preference I follow Daveo/ Boycie 1 Total - 19 Table 9.14: Functional categorization of “habit/ principles” co-texts (n = 19). As with the other process types discussed, the kind of interaction between NSI and formally related co-texts describing activities speakers routinely engage in depends on the formal link between the structure of interest and its co-text: if the coordination and is used, the activity intensifies, that is, describes in more detail what the NSI entails, or simply adds to its meaning (example 9.44). If the formal link is one of cause or consequence, the activity described causally relates to what is indicated by the NSI (example 9.45), and in case of a contrasting relation between NSI and its co-text, the activity contradicts implications created by the NSI (example 9.46). 9.44. I’m not a great lover of positions 2/ 4 and I rarely play without the S1 on. (Intensification) 9.45. I’m not a serial doctors apt person so only ever go in when I clock up a few complaints. (Consequence) 9.46. I am not an asbestos expert however I deal with asbestos on a regular basis. (Contradiction) 292 9 The Discourse Functions of Negative Self-identifiers <?page no="293"?> 70 Considering more of the co-text surrounding this NSI reveals that it represents a contrast to the co-text following it. The speaker goes on to explain that “I’m always at the back of the pack, but nobody minds waiting for a few minutes every now and then for me to catch up”. In five out of six cases where a habitual activity is contrasted with an NSI, this NSI can be classified as a disclaimer of expertise. Despite the non-generalizability of the small sample here, this seems to support previous analyses presented in this chapter which revealed that implicit references to tacit knowledge - gained, these co-texts suggest, from having experienced something or, as in this case, from regularly doing something - tend to be contrasted with expert identities. In the three cases where an NSI is added to a co-text representing an activity, the NSIs, too, come from the conceptual domains of “activity” and “preference (fan): activities”. This can mean that the sentence as a whole characterizes the speaker’s preferences. Or, the textual material additively related to the NSI can contradict or be interpreted as an entailment of what is implied by the instance of the structure, as in example 9.47 below. Here, the speaker first negatively identifies with the category of “fast runners”, just to add that they often join a running club named Durham City Harriers. Depending on whether or not it is evident from the co-text, or implicitly assumed to be known, that this is a club for slow runners, this could either be interpreted as a consequence of the negative identifier (if we assume that the Durham Harriers are slow, too) or as a contradiction to the speaker’s preceding self-identification as slow (if we assume that the Durham Harriers are regular or fast runners). 70 9.47. I’ve never been a fast runner and quite often go out with the Durham City Harriers on their runs. In the two cases where co-texts representing activities are causally related to NSIs, speakers negatively identify with preferences (e.g., fan of thrashing about in wet snow), the causes or consequences of which are then described in more detail, to the effect that the whole sentence describes what the speaker likes or dislikes. The three instances of speakers referring to their consumption habits are formally related to NSIs from the conceptual domain “habits/ principles”, specifically in relation to food and substances. 9.1 Functional analysis of formally related co-texts 293 <?page no="294"?> Situative anchor Finally, another category of sentences formally related to NSIs and constituting material processes which occurs often enough to be mentioned here is what I referred to as “situative anchors” above; these are sentences which describe a ‘status quo’. This status quo, that is, a description of the situation as it is at the moment of speaking, can result from a past happening or action causing a result state, (e.g., I’ve already paid for a year’s subscription to the Telegraph), from an activity that has been going on up to the moment of speaking (e.g., I’ve been putting it off) or be constituted by a general fact or state (e.g., I don’t get many customer emails). In co-texts of this type, it appears that NSIs more often serve informative purposes than in some of the previously examined ones. Referring back to the analysis of mental process contexts, it has been shown that NSIs, notably those from the conceptual domain of expertise, often merely function to epistemically mitigate information provided by the speaker, serving rhetorical rather than assertive functions. An example is I am no expert, but I believe any GAD [generalized anxiety disorder] test over 50 indicates an autoimmune condition, where the NSI hedges the speaker’s statement, namely their interpretation of a GAD test. By contrast, in the following extract from my corpus, where a speaker uses an NSI in a co-text classified as “situative anchor” (underlined), the function of negatively identifying as an EU resident is not that of a stance marker, but to negatively assert information that is intended to help the addressee better understand the speaker’s situation - due to the fact of not being an EU resident but having a UK LTD (company), they might be liable to pay VAT in other countries. 9.48. I am not a EU resident though Ive registered a UK LTD on my name and then successfully registered as a business seller on AMZ UK. During the signup process I didnt pay much attention to the ‘ SELL ON OTHER EUROMARKETS’ (something like that) and ticked the box. I am now in concern that I have to obtain not only UK VAT # but German, French, Spain and Italian VAT # when I reach the limit of 15k. Another instance of an NSI being used with seemingly more ‘mundane’, assertive functions in its formally related co-text is I have not been a PWP as I took up an assistant psychologist post instead, where taking up one job is represented as excluding the possibility of identifying with another position (that of a PWP). The functional subcategorization of “situative anchor” co-texts, with items assigned to each subcategory, is shown in Table 9.15. 294 9 The Discourse Functions of Negative Self-identifiers <?page no="295"?> Category Items Frequency Completed actions and happenings - Perfect Ive registered a UK LTD on my name; I’ve already paid for a year’s subscription to the Telegraph; I couldn’t miss taking a photo of this male Slow Worm; I took up an assistant psychologist post instead; I never signed up for this; [I have] not opened it for a while; I have set all my shipping prices for my listings; I have just registered for this purpose; I stopped then as defeat for the Federation looked certain; I as usual did not get the e-mail 12 Ongoing - Progressive [I] am battling my way through the install; I’ve been putting it off; [I] was just installing it through the config menu 8 State - Non-progres‐ sive [I] don’t get many customer emails; I get popped or run more often than not when I unwisely try PVP; I live a good 2 hours away 1 Total - 16 Table 9.15: Functional categorization of “situative anchor” co-texts (n = 16). All this indicates, once again, that NSIs and the co-texts with which they are formally related are conceptually related as well, and judging from the data examined here, patterns of recurring conceptual relations can be observed. In any case, the analysis shows that NSIs tend to interact strongly with the rest of what is being said and have discourse-modifying purposes: thus, referring to an example from Table 9.14 above, eating two boiled eggs daily (a consumption habit) formally and functionally interacts with not being a vegan (categorized here as food-related habit), just like not being a cream lover (a food preference) relates to the quantities of butter used (a consumption-related activity). The analysis of material processes formally related to NSIs shows that speakers often use the structure “I + copula + NOT + identifying NP” in the context of talking about what they have done (“experiences”) and in the context of what they routinely do (“habits”). Experiences with particular activities and accomplishments implying the acquisition of skills and knowledge were shown to frequently occur together with disclaimers of expertise. Overall, 12 out of 22 sentences classified as “experience” are formally related to NSIs with identifying NPs from the conceptual domain of professionalism and expertise, and all but one of them are formally contrasted with experience contexts. Regarding co-texts classified as “habits/ principles”, these sentences are rela‐ ted to an expertise disclaimer in nine out of 19 cases. The other NSIs in these co-texts come from the domains “preference” (five instances), “activity” (two 9.1 Functional analysis of formally related co-texts 295 <?page no="296"?> instances), “habits: food/ substances” (two instances) and various others (three instances). So, it could be that talking about one’s experience is more closely associated with being or not being an expert, while talking about one’s habits and principles is associated with one’s preferences. Finally, the analysis also showed that NSIs formally related to material processes can serve to provide more specific information about the speaker against a situational background (“situative anchor”) described by the sentence with which the NSI is formally linked. This indicates that the structure also has predominantly descriptive uses, where it serves to characterize the speaker in relation to a particular situation. 9.1.4 Summary Section 9.1 has presented functional analyses of sentences constituting mental, relational and material processes with I in the role of the senser, the carrier/ token and the actor, as these were the most frequent configurations of participants and processes represented in sentences formally related to instances of the structure in focus (see Chapter 8). Analyzing each process type category separately, comparing formally and functionally similar textual material and accounting for the formal and semantic properties of sentences representing mental, relational and material processes was intended to render the analysis as criteria-based and as fine-grained as possible. Overall, 401 sentences formally linked to NSIs were analyzed in depth to learn with which co-texts NSIs are formally and thus conceptually associated. The result of this is a functional profile of ideationally and formally defined co-text types preceding and following NSIs in this corpus. It is represented in Table 9.16 below. The table also shows the respective conceptual categories of identifying NPs for the four most frequently represented, that is, dominant categories of co-text. As mentioned already, these numbers have to be taken with a grain of salt because of the difficulties with conducting definite categorizations of authentic language in use. Functional category Mental Re‐ lat‐ ional Ma‐ te‐ rial Total Dominant con‐ ceptual categories of NSIs (%) Knowledge representa‐ tion/ opinion 72 20 1 93 66 Expertise/ prof. Knowledge/ understand‐ ing reference 31 26 5 62 47 Expertise/ prof. Experience 18 22 22 61 54 Expertise/ prof. 296 9 The Discourse Functions of Negative Self-identifiers <?page no="297"?> Preferences/ habits/ principles 24 10 19 53 59 Preference Background/ situative anchor (incl. possession & entitlement) 6 10 16 32 - Addressee-oriented/ discourse-internal 17 3 5 25 - Personal characteristics - 13 - 13 - Intention 10 - - 10 - Problem 3 6 - 9 - Response - 9 - 9 - (In)ability - - 6 6 - Decision-making 4 1 - 5 - Measure - - 5 5 - Hypothetical actions - - 5 5 - Wish 4 - - 4 - Perception 3 - - 3 - Emotional states - 3 - 3 - Forum-related charac‐ teristics/ attitudes - 2 - 2 - Total 192 125 84 401 - Table 9.16: Functional profile of clauses with formal links to negative self-identifiers (n = 401). Considering the results of the analyses of mental, relational and material processes together reveals that, whereas some categories are only relevant for a particular process type (e.g., the category of (in)ability only includes material processes), others recur independent of the process type the sentences assigned to that category represent. For example, the category of “experience” is relevant for all three process types examined here: 9.49 (relational), 9.50 (material) and 9.51 (mental) all represent experiences according to my framework, despite constituting different processes from a transitivity perspective. 9.1 Functional analysis of formally related co-texts 297 <?page no="298"?> 9.49. I lost 3 kg 9.50. [I] have only completed one marathon 9.51. I have seen a lot of things said and written about “bioidentical/ compounded” hormones This might, of course, be interpreted differently: the category could either be weak, allowing too much textual material in, the process type distinction might simply not be as relevant for analyzing what speakers actually do with language or the category might be considered particularly relevant. I believe there are arguments for all three interpretations, the third is not plausible, based on the assumption that even though categorization and quantification by just one person is somewhat problematic for said reasons, reaching conclusions requires accepting - rather than deconstructing - the constructedness of linguistic analysis. Disclaimers aside, these numbers still serve to indicate certain trends that can be observed in the data: NSIs are most often, namely in 93 of 401 sentences examined here, formally coordinated with sentences representing the speaker’s knowledge or opinion, and 66% of these knowledgeand opinion-representing sentences are formally related to NSIs from the conceptual category “expertise and professionalism”. Considering that the category “knowledge/ understanding reference” is very similar to “knowledge representation/ opinion” in that it includes sentences in which speakers comment on what they think they know, have learned etc., the overall number of NSIs formally coordinated with knowledge-representing or knowledge-negotiating material is even larger. Negative self-identifiers classified as disclaimers of expertise correlate most strongly with all but one category of co-text, namely, what I referred to as “preferences/ habits/ principles” here, that is, speaker’s descriptions of what they routinely, normally or principally do or like. In these co-texts, it is preference disclaimers which dominate (in 59% of the cases examined here, a preference disclaimer occurs in such a context). This corroborates previous observations of preference disclaimers occurring in co-texts describing speakers’ tastes and routines, indicating that negatively identifying with particular preferences often serves to specify personal choices and preferences in the negative. Functionally categorizing the immediate co-texts of NSIs formally linked to the structure, making up 43% of the NSIs in my corpus, provided an overview of the types of discourse contexts in which variants of the structure “I + copula + NOT + indefinite NP”, as used by people interacting on UK web forums, appear. Since structures like NSIs are multifunctional, indexing and reflecting various aspects of the textual and situational context in which they are used, determining precisely 298 9 The Discourse Functions of Negative Self-identifiers <?page no="299"?> which functions are served by individual instances of particular types of NSIs in particular co-texts is only possible on a case-by-case basis. As the functional profile in Table 9.16 above shows, it is still possible to identify a set of key functions that the structure serves. In short, while NSIs are multifunctional, the analysis shows that they most often serve as epistemic stance markers in knowledgeand opinion-related talk, and as negative specifiers in preference-related talk. These functions are summarized with corpus examples in Table 9.17 below: Category Subcategory Definition Corpus example Descrip‐ tive posi‐ tioning General Negative self-identification to provide details about sit‐ uations speakers find them‐ selves in, represented by co-texts classified as “situa‐ tive anchor” and “possession & entitlement” I know I have two small chil‐ dren aged 3 and 7 but since Dec I haven’t been a proper mum Intention/ wish Negative self-identification to position speakers in rela‐ tion to things they intend to do or want to happen I want to go back to what I had before Google Chrome and I am not a computer boff, please help. Decisionmaking/ problemsolving Negative self-identification to position speakers in rela‐ tion to decisions they are facing and problems they are confronted with I’m pretty much not an early morning person so I might decide on a couple of nights to run during the week and go home early those nights Discur‐ sive posi‐ tioning Attitudi‐ nal/ pref‐ erence Negative self-identification with preferences to position speakers more specifically in terms of personal choices I am not a vegetarian but I have never been a big meat/ cheese eater, I would normally prefer grains, vegetables, fruits, etc. Epistemic Negative self-identification with expertise to epistemi‐ cally modify utterances de‐ scribing speakers’ knowl‐ edge and experience I’m no expert but I’m pretty sure you can. Interper‐ sonal/ metadis‐ cursive Negative self-identification to modify the interpretation of co-texts explicitly addressing the interlocutor or metadis‐ cursively commenting on the ongoing social interaction I am no expert, but I am in the same boat as you. I should probably post this on a triath‐ lon forum but as I’m not a member of any, I thought I’d ask here first. Table 9.17: Functional categories of negative self-identifiers, with descriptions and illustrative corpus examples. 9.1 Functional analysis of formally related co-texts 299 <?page no="300"?> RQ2B and RQ2C asked what the functions of co-texts with particular experien‐ tial meanings and formal links to the target structure were, and how frequently these categories were represented in the corpus. The analyses in this section examined co-texts that were formally coordinated with, or superordinate or subordinate to, NSIs. The next section turns to a different type of co-text: textual material preceding the focus structure without being formally related to it. Unlike the analyses presented so far, which considered only clauses and sentences, the following analysis also includes textual units above the sentence level, where formal linkage is absent but functional patterning persists. 9.2 Functional analysis of co-texts preceding negative self-identifiers To answer RQ2A-C with analyses extending beyond a single formally defined linguistic paradigm (clauses and sentences formally linked to the NSI structure), this section reports the results of a functional analysis of textual material preceding NSIs. In my corpus, declarative sentences are the most frequent element in this position, occurring 376 times. The analysis considers not only the declarative sentence immediately preceding the matrix clause containing the NSI, but also the communicative function of higher-level textual units in which these sentences are embedded (see Section 6.1 for the analytical model and criteria used to define such units). In the following, I first explain in more detail why it is important to consider units above sentence level to account for the functions of NSIs and what intricacies are involved in qualitatively distinguishing such categories of co-text. I then present and describe the superordinate communicative functions which can be identified in the examined data and go on to discuss the individual cate‐ gories in greater detail, providing information about their respective frequencies of occurrence and their relations with the conceptual category of NSIs. The following extract from my corpus illustrates how NSIs may interact not only with the sentences immediately preceding them, but with entire functional units of text: 9.52. Hi all I have had a couple of Blackstar amps for gigging but this is my first practice amp from them As standard pretty damn good Played it for a couple of weeks and then thought this could be better Contacted Watford valves in the UK and bought a Celestion Super 8 and their recommended valve kit,a Harma ECC 83 retro and a Phillips 12ax7 300 9 The Discourse Functions of Negative Self-identifiers <?page no="301"?> What a difference Cleaner at much higher volume and the drive channel is classic Brit Rock. The fizz is gone I am no engineer but it only took me 20 mins to do. This is an instance of an NSI preceded by a declarative sentence stating the result of an entire story of what the speaker did to repair an amplifier: the fizz is gone (shaded in grey). While explaining the function of the negative identifier just in relation to this immediately preceding sentence would work (referring back to the functional framework introduced in Section 9.3, this sentence could also be classified as “situative anchor”), the use of the NSI is much more interesting if one considers that it can also be interpreted as referring back to the entire short narrative, the result of which is stated by the sentence immediately preceding it. By using an expertise disclaimer right after relating this experience story, the speaker undermines the expertise implied by their account (a conceptual strategy I refer to as de-expertization, see Sections 7.3.2 and 9.3.1), only to add that it only took 20 minutes to do, thus again highlighting the simplicity of the procedure and, thus, their skills (I have called this re-expertization). To learn more about the functions of NSIs in interaction with their wider discourse context, the analysis presented in this section considers as co-text not only the immediately preceding sentence, but the entire functional textual unit preceding the NSI. Through iterative data analysis, I established a set of functional categories appearing in the (formally unrelated) context preceding NSIs. It should be mentioned that, again, the established categories are fuzzy, which means determining their respective frequencies of occurrence in the examined data reveals trends in people’s language use rather than definite numbers for clear-cut categories. For instance, both textual materials classified as “knowledge representation (self)” and “answer” in this analysis effectively serve to assert propositional content that the speaker believes to be true - more or less, since epistemic stance may vary. The only difference is that I counted as answers those utterances which contain linguistic elements whose referents have clearly been mentioned in another speaker’s utterance preceding it, such as personal pronouns as in It has bluetooth built in, the definite article ‘the’ as in The sheep are safe, Geordie or, simply, yes or no marking utterances like yes you can definitely print FBA labels on the RM DMO compatible thermal printers. 9.2 Functional analysis of co-texts preceding negative self-identifiers 301 <?page no="302"?> 9.2.1 Functional categories of preceding co-texts Table 9.18 below presents a functional profile of textual material preceding NSIs. It defines superordinate and subordinate functional categories and provides attested examples from the corpus for each of them. That some categories can be distinguished in the first place, and that some differentiable categories are related, pointing toward higher-order conceptual trends, gives an indication of prominent conceptualizations in the examined data, without even looking at frequencies. Category Subcategory Definition Example Situative anchoring back‐ ground Describing aspects of a situation currently faced by the speaker (present orientation) or anticipated by the speaker (future orienta‐ tion) This was the first time I had left him since being diagnosed on Friday problem Describing a problem‐ atic situation faced by the speaker at present or in the past I read and watched many tuto‐ rials but nothing helped. decision -making Describing a situation in which the speaker is confronted with a deci‐ sion I notice some dealers now have 5243 (R3126) in stock and I am thinking of getting one. intention/ desire/ seeking Expressing an intention, wish, or search for something (e.g., want‐ ing, planning, or seek‐ ing advice) Does anyone have any sugges‐ tions about posting books not using the Royal Mail? Their prices are going up again in April, unbelievable after their crappy performance! Attitudinal positioning opinion Stating one’s opinion on a subject (+ presence of evaluative elements) That, in itself, is quite an im‐ pressive feat habit/ preference Reference to habits or habitual preferences (indirect stance through lifestyle or practice) I eat no red meat 302 9 The Discourse Functions of Negative Self-identifiers <?page no="303"?> Experience sharing event story Relating a sequence of ‘tellable’ events not in‐ volving the speaker Here’s one more - Pete lost in second round on 26th June 2002 and Roger on 26th June 2013. experience story Relating a personal ex‐ perience (rather than an ‘external’ event) Needless to say, I removed all the embed code from the web‐ site and apologized. itemsharing Referring to something one is about to show to other forum users Hello […] In this thread I will show you my Mourinho-esque tactics I have tried to create. product experience story Relating one’s experi‐ ence with a particular product, service, or en‐ tertainment (constitut‐ ing the macro-theme of the account); usually in‐ volves, and sometimes only consists of, some kind of assessment Missed it first time round but watched *The Road to Corona‐ tion Street* last night on BBC 4 and thoroughly enjoyed it - the woman playing Pat Phoe‐ nix was a brilliant likeness. shared experience Explicitly establishing a relation between one’s own and another partic‐ ipant’s experience I have noticed the same thing over the last 6 months or so. shared experience (advice) Drawing on one’s own experience to offer ad‐ vice indirectly by re‐ counting a comparable situation rather than is‐ suing a direct recom‐ mendation If I had decided to repair using MIG, I would have toiled for sure; my preference for O/ A on these types of occasions made repairs far easier - one can dance the flame around, giving the welder far more control on heat input; where with MIG, it’s a quick burst and pray you don’t burn through experience/ advice Providing advice through a personal ex‐ perience, identified as such on the basis of linguistic elements like directives, verbs (recom‐ mend, suggest), modals (should), or hypothetical constructions (if I were you) One thing to mention is that when you have an issue with a command, aren’t sure the method to something, or have a suggestion, they are only a phone call or an email away. They are always looking to improve the software and sug‐ gestions are discussed with the users. I highly recommend this software. 9.2 Functional analysis of co-texts preceding negative self-identifiers 303 <?page no="304"?> Knowledge negotia‐ tion deictic assessment Making an (evaluative) assessment of a tex‐ tual or visual referent present for all partic‐ ipants in the ongo‐ ing communicative sit‐ uation Water’s really nice, the trans‐ lucent blue effect looks almost jelly-like. Transparent is okay, but the teeny bit of tongue/ mouth showing looks out of place on this one. diagnosis Diagnosing/ identifying a textual or visual ref‐ erent present for all participants in the on‐ going communicative situation I wonder if the smell is actually sulphurous, suggesting a sul‐ phide mineral. But the density is too low for pyrite/ iron/ many other metals or metallic min‐ erals knowledge represen‐ tation (self) Presenting information on a particular subject as one’s own knowledge I’ve read on the internet that it is not wise to take if you are asthmatic. knowledge represen‐ tation (other) Presenting information on a particular subject as knowledge obtained from somebody else I had schema therapy and that helped me a lot and my schema therapist really pushed the so‐ cial side of things, and I think she has a point that there’s something social. knowl‐ edge/ under‐ standing reference Referring to knowledge and information on a meta-level (i.e., talking about, rather than pre‐ senting, knowledge) Making games though, so I am pretty neck deep in related fields. There is really only one person you can trust with your personal information and that is yourself. 304 9 The Discourse Functions of Negative Self-identifiers <?page no="305"?> 71 The distinction between metadiscursive and metapragmatic elements made in this analysis rests on an analytical distinction between explicit (metadiscursive) and largely implicit (indexical) metapragmatic devices. Accordingly, “I’m going to avoid your game” constitutes an explicit comment on the ongoing interaction, whereas “why I want this game: ” frames the following co-text as justification or explanation without making this function explicit; the recipient is left to infer the connection between why, the colon, and the subsequent text. For a comprehensive discussion of distinctions between metacommunication, metapragmatics, and metadiscourse, see Dynel (2023). Interaction advice Providing advice, which is identified as such on the basis of the presence of certain linguistic ele‐ ments (e.g., directives, verbs like recommend, suggest; modals like should; structures like if I were you etc.) You might as well do it with‐ out spiralling into too many addictions. metadiscursive (various) 71 Commenting on the on‐ going communicative situation I’m sorry, I’m going to avoid your game (or at least what it seems to me) request Directly or indirectly asking other speakers to do something Any assistance would be greatly appreciated in combining these functions into 1 profile. question Request for information I’m not really sure where to post this as it just concerns injecting in general but seeing as the drug injected was heroin this seemed appropriate. Can a mod move it to the appropriate place if need be? My question concerns the colour of arterial blood. answer Answering a question by another speaker If by numbers you mean the 3 digit tactical marking you often (but not always) see on the tanks: No. agreement/ disagree‐ ment Expressing alignment or non-alignment with something another speaker has said I see your point (not literally, you understand), or whoever made it originally. metaprag‐ matic in‐ dex Indexing the pragmatic function of the follow‐ ing text Why do I want this game: Table 9.18: Functional profile of co-texts preceding negative self-identifiers, with defini‐ tions and illustrative corpus examples. 9.2 Functional analysis of co-texts preceding negative self-identifiers 305 <?page no="306"?> 9.2.2 Results Determining how frequently these functional categories are represented by co-texts preceding NSIs (criterion for inclusion: L1 = declarative sentence), yielded the results presented in Table 9.19 below. In the following, I discuss these results in more detail, presenting and qualitatively analyzing concrete examples from my corpus. Functional category Frequency Product experience 67 Advice 34 Experience story 29 Metadiscursive (various) 26 Situation: problem 22 Situation: background 21 Knowledge/ understanding reference 19 Knowledge representation (self) 19 Opinion 15 Diagnosis 15 Experience/ advice 15 Situation: seeking 13 Question 11 Answer 10 Deictic assessment 10 Shared experience 10 Assessment 7 Knowledge representation (other) 6 Situation: decision-making 5 Habit/ preference 5 Item sharing 5 Request 5 306 9 The Discourse Functions of Negative Self-identifiers <?page no="307"?> Agreeing/ disagreeing 2 Metapragmatic index 2 Event story 2 Self-assessment 1 Total 376 Table 9.19: Frequencies of functional categories in co-texts preceding negative self-iden‐ tifiers (n = 376). As can be seen from this functional profile, the co-texts examined can most often be categorized as experience accounts”. The most prominent representatives of that category by far are product experience stories (67 instances, 18%). This means that NSIs are a linguistic choice that speakers frequently make after sharing their experience with, and evaluating, products and services (of course, this could also be because reviewing products is a prominent practice on forums and many discussions examined here were taken from product forums). The second most frequently appearing type of preceding co-text identified in the analysis is advice (34 instances, 9%), followed by what I labelled “experience stories” (29 instances, 8%). Another type of co-text frequently appearing before NSIs are situative anchorings of the types “problem” and “background”, that is, speakers often use NSIs after describing the general situation in front of which they position themselves. Co-texts of the type “knowledge/ understanding reference” and “knowledge representation: self ” also appear frequently: in a total of 38 cases (10%), NSIs are used after stating what one thinks and reflecting upon what one believes. Regarding relations of co-occurrence between particular conceptual catego‐ ries of NSIs and co-text types, Table 9.20 below shows the conceptual categories of NSIs following the most frequently identified types of preceding co-text, namely “product experience”, “knowledge representation (self)” and “knowl‐ edge/ understanding reference”, “advice”, “experience story”, “metadiscursive”, “situation: problem” and “experience/ advice”. The most frequently occurring conceptual categories of NSIs following these co-texts are highlighted. 9.2 Functional analysis of co-texts preceding negative self-identifiers 307 <?page no="308"?> NSI conceptual category Prod‐ uct ex‐ peri‐ ence Knowl‐ edge ref‐ erence/ repre‐ senta‐ tion Ad‐ vice Expe‐ rience story Meta‐ dis‐ cur‐ sive Situa‐ tion: prob‐ lem Expe‐ rience & ad‐ vice Characteristics (various) 9 4 2 11 5 2 - Expertise & pro‐ fessionalism 13 27 23 7 10 8 8 Habits (various) 3 1 - 2 3 1 1 Preference (fan) (various) 8 1 1 1 1 2 - Preference (+ fan) (various) 26 2 6 5 1 3 5 Roles (various) 3 3 1 3 5 4 - Usage/ consumption/ ownership 5 - 1 - 1 1 - Total 67 38 34 29 26 21 14 Table 9.20: Negative self-identifiers by functional category of preceding co-texts. As can be seen, contexts in which speakers describe their experiences with particular products and services are most often followed by preference disclaim‐ ers featuring the lexeme fan (26 instances). An example of an NSI from this conceptual category being used after a product experience account is 9.53 below. Here, an NSI is used to specify the speaker’s desired appearance of a car just after extensively relating their experience with it. 9.53. Anyway, this is my white 180k 106 from the start […] It looks and sounds awesome! And in typical teenage/ silly child fashion, I found out it pops, so I’ve been driving past people and making it pop to scare them. Easily the best way to get myself pulled overAnyway, more photos.I haven’t done TOO much, mostly just cleaned the interior and fitted a parcel shelf etc. Tomorrow night I’m going to do a deep clean on the carpet (Not fitted) and try to get that fitted nicely, possibly stick the B and C pillar trims on and generally just get it washed etc for FCS, and try to figure out why the bloody demist doesn’t work. And I really need to lose the Saxo mirrors, I’m not a fan. 308 9 The Discourse Functions of Negative Self-identifiers <?page no="309"?> That preference disclaimers tend to occur after such product experience accounts suggests that, while speakers are hesitant to represent what they know as unproblematic, formally coordinating epistemic disclaimers with what they say is true, the NSIs they use in product experience contexts are more self-confident: negating having a preference for something is a powerful way of indexing awareness and liberty of choice. Examining other instances of product experience stories followed by preference disclaimers suggests that being or not being a fan of something is also associated with expert knowledge. Consider, for example, the following product experience story, which incidentally is an instance of two NSIs occurring in one posting - the first one positions the speaker as a layperson (they negatively identify as a motoring journalist and, thus, as someone possessing the expertise to write about engines), and the second one serves to specify their engine preferences: 9.54. Big thanks again to elmsDirect for the loan of the big 7 over the Gaydon weekend. I’m no motoring journalist, but here’s a few thoughts from a couple of hundred mixed-use miles….I’ve never driven the logical competition (Merc S class, Lexus LS, Jag XJ etc) and assuming this niche of car is aimed at big mileage, (mainly) motorway use, the 7 hits the mark. I’m no fan of diesels, but can’t fault over 300 bhp, loads of torque and still an average of 28 mpg overall and approaching 40 on the motorway This product experience story - classified as such because of the metadiscursive comment by the speaker, who frames their story as “just a few thoughts from a couple of hundred mixed-use miles” - constructs the speaker as very knowledgeable about cars: they refer to a variety of specific car brands and identify them as “logical competition” of the specimen they are talking about, but the NSI they use after sharing this experience - fan of diesels - is also followed by a recommendation constructing the implied addressees as equally well-versed when it comes to cars and thus capable of interpreting the significance of the given values (e.g., “28 mpg overall” - despite “torque”). This means that negatively self-identifying as a fan of diesels here positions the speaker as an opinionated car expert, albeit not a professional one, for example, a motoring journalist. As indicated in Table 9.20 above, presenting or metadiscursively referring to one’s knowledge is most often followed by expertise disclaimers; the same applies to co-texts classified as advice in my functional profile. The following example is a case in point. Here, the NSI is used to mitigate the speaker’s recommendation and is followed by metadiscursive comment on the advice just 9.2 Functional analysis of co-texts preceding negative self-identifiers 309 <?page no="310"?> given, labelling it as opinion and inviting other speakers to express their views on the subject, too. 9.55. Whatever oil you use change it at the recommended times and keep the air filter clean. I repeat that I am not an expert and welcome other opinions. Negative self-identification with particular characteristics (such as hero, prude or bad person) is, in 11 cases, preceded by co-texts classified as “experience”. This could indicate that when talking about personal experiences, self-description and thus characterization in the negative is more relevant than negative self-identification with, and thus reference to, expertise and professionalism. Take, for instance, the following examples: 9.56. Needless to say, I removed all the embed code from the website and apologized. I’m no prude (65 year old hippie and ex Navy) and I’m sure some of my friends would find humor in the ‘suggestions’ but this particular friend was not amused. 9.57. Then the final nail in the coffin was when my son became old enough to [do? ] things that i liked too, he became my best friend and the love of my life, different than my wife, i became someone he thought the world of and wanted to be just like me, i was his hero. I have never been a hero to anyone […] In example 9.56, the speaker’s account of what they have done is followed by a character disclaimer, serving to fend off possible implications of their story. In example 9.57, the negative identification with the evaluative concept of hero is used with a slightly different function, negating not the speaker’s self-identification as a hero in its entirety, but limiting their hero status to their relationship with their son as previously related, in which he thinks he served as a role model. Thus, the NSI, on the one hand, marks the related story as unique and, on the other hand, serves to make the experience account more modest. The comparatively higher frequency of negative identification with particular personal characteristics in experience contexts suggests that, when relating their own experiences, speakers are more concerned with discursively managing the interpretation of their personalities and personal features than in other contexts. When describing a problematic situation, speakers negatively identify as experts or professionals more often than they do with other conceptual cate‐ gories, but since the sample of reference here is quite small, this relation of co-occurrence cannot safely be claimed to constitute a pattern. However, pre‐ vious analyses presented in Section 9.1 have shown that NSIs are often used in co-texts describing situations in which speakers are confronted with decisions, 310 9 The Discourse Functions of Negative Self-identifiers <?page no="311"?> problems, intentions and wishes, which indicates that cases such as example 9.58 below are, indeed, a characteristic occurrence of the structure in focus. Here, an expertise disclaimer is used after presenting a problematic situation and is potentially relevant for its solution: if the speaker was a programmer, it would probably be easier for them to tackle the problem described. 9.58. Hy. I got this generic TFT screen and I’ve been trying forever to get it working. I have no clue what to do anymore. I’ve followed this tutorial and I couldn’t get it working because the X file isn’t there to edit : P. I’m no programmer … I would really appreciate some help Another potentially interesting - but, overall, not very frequent - co-occurrence of particular functional categories of co-text and conceptual categories of NSIs is between metadiscursive comments and negative self-identification with particular roles relevant in the situative discourse context. An example of such a situationally relevant role is saint on this forum in the NSI represented in 9.59 below. The reason why I consider this NSI situationally relevant is that being or not being a saint on a particular forum is a self-description that only matters on that forum (though, of course, it might be argued that if I identify as a well-behaving or ill-behaving person in the context of a forum, this has implications for my self-identity and representation beyond the immediate discourse situation). The NSI in this example is used to modify the interpretation of a previously uttered critical comment about the addressee’s behavior on the forum. By negatively identifying with the category of forum saints, the speaker mitigates the face threat of their criticism (just to amplify it again in the subsequent sentence introduced by but). 9.59. DF, you habitually go after people and habitually get disproved. I’m no saint on this forum, but your raison d’être here appears to be one of outsmarting, one which perpetually fails. Finally, another relation of co-occurrence between a functional category of preceding co-text and a particular type of NSI is worth commenting on, namely that between contexts classified as metadiscursive and disclaimers of expertise. Regarding metadiscursive co-texts, it is possible to differentiate between intra-textual metadiscourse and extra-textual metadiscourse, that is, metadiscourse commenting on the communication taking place on the forum, and metadiscourse referring to communication that took place prior to the forum discussion and between other participants than the speaker and addressee 9.2 Functional analysis of co-texts preceding negative self-identifiers 311 <?page no="312"?> whose conversation I am studying. The following extract from my corpus is an example of intra-textual metadiscourse followed by an NSI: 9.60. I will be very happy to share with everyone on this thread, I am not a doctor or anything but doing stats is not that that [sic] complicated. Here, a disclaimer of expertise is used after an utterance metadiscursively commenting on what the speaker is aiming or willing to do on the forum, namely, to share something, thus pre-emptively marking the content to be shared as lay content. This, once again, indicates that a key function of NSIs in the examined corpus is to epistemically mitigate what speakers believe, claim to know and exchange. Another example is 9.61, an instance of extra-textual metadiscourse which illustrates the tension between lay and expert identities in the data. Here, an NSI, used turn-medially by speaker B just before providing information they have researched, not only contrasts the speaker with the group of doctors in general, but also mockingly comments on the “helluva mouthful” of medical jargon which, it is implied, can only be understood by doctors (informally referred to as “discharge docs” by speaker A): 9.61 A: The discharge docs say “ischemic stroke, mri showed right sided deep white matter and small cortical infarcts in a watershed distribution CT angi [angiogram]and cartaroid [carotid] artery showed no signs signifi‐ cant stenosis I understand there are layers of brain that do different things, but my googlefu must be broken too. If anyone can translate medical into english I would appreciate it. Thanks - B: Hi That’s one helluva mouthful! The jargon must be useful to a medical profes‐ sional, but it’s just not English is it! I’m no doctor, so can’t be authoritative I’m afraid. But here’s a few definitions from the internet. Interestingly, both speakers agree that the cited diagnosis is “not English” and therefore needs to be translated. To help with the translation, speaker B, after using the NSI, presents speaker A with a list of definitions. When discussing the role of experts in affluent late modern societies initially, I referred to theories which see experts as mediators of knowledge, that is, as persons who filter the abundance of available information, helping others assess the value and use of circulating knowledge (Bauman 1987). In this exchange, speaker B seems to be doing precisely that: they mediate between users of “medical” and users of “English”, drawing on information they have researched online. Hence, speaker 312 9 The Discourse Functions of Negative Self-identifiers <?page no="313"?> B, while negatively identifying as a doctor and thus as someone possessing formal medical education necessary to understand the cited jargon, acts, de facto, as an expert. 9.2.3 Summary The analysis in Section 9.2 examined preceding co-texts of NSIs without formal links to the structure. Functional categories were established through qualitative analysis and their frequency determined. As in Section 9.1, the aim was to test whether certain co-text categories tend to co-occur with particular conceptual categories of NSIs. Of 376 instances, 67 (18%) were preceded by product experience stories, 26 of which (39%) were followed by preference disclaimers. Thirty-eight (10%) represented or commented on knowledge and information, most of which (27; 71%) were followed by expertise disclaimers. In 34 cases (9%), advice served as preceding context, and 23 of these (68%) were followed by expertise disclaimers. These results support the broader tendency (see Section 9.4) for NSIs in the domain of “expertise and professionalism” to occur in co-texts representing or commenting on knowledge and information, and for preference disclaimers to be used in contexts of consumption-related preferences. Section 9.3 zooms in on two illustrative cases to show how these patterns unfold in context. 9.3 Qualitatively studying negative self-identifiers in context 9.3.1 Linking micro-level functions and macro-conceptualizations The present study has, so far, been concerned with studying fine-grained meaning differences between NSIs and exploring their relations with, and the functions of, their immediate textual surroundings in detail and across situations of usage. The analysis in Chapter 7 has shown that NSIs from two conceptual domains, viz. expertise and preference, feature particularly prominently in the examined corpus; the analyses in Chapter 8 and Sections 9.1-9.2 of this chapter revealed that expertise disclaimers occur particularly often in co-texts representing mental and relational process types, expressing what speakers (believe to) know, and that preference disclaimers occur particularly often in co-texts describing speakers’ experiences with products. This, I argued, indicates 9.3 Qualitatively studying negative self-identifiers in context 313 <?page no="314"?> that NSIs reflect speakers’ orientation to two related conceptual trends, which I referred to as deand re-expertization by informed choice. De-expertization is understood as the process by which the notion of formally accredited, ‘top-down’ expertise is simultaneously acknowledged (because it is made relevant) and undermined (as speakers negatively identify as experts and professionals) in online settings such as forums. Epistemic disclaimers are a microlinguistic identification strategy by which speakers explicitly contrast their utterances or contributions with (what they consider to be perceived as) expertise in the situational context. This means that their use reflects speakers’ orientation to conceptualizations of (various forms and levels of) expertise in their linguistic self-representation on forums. The kinds of expertise circulating on online platforms such as forums, which connect people sharing particular interests, but potentially pertaining to superdiverse societies in a globalized world (Tagg & Lyons 2018), have been the subject of recent scholarly debate. As discussed in Section 2.2.1, the notion of expertise has been critically explored in terms of its role in interactions among experts and non-experts (Bigi 2011) and its reconceptualization toward lay expertise (Williams 2014, Rueger et al. 2021) in the context of a more democratic, participatory knowledge negotiation in digital discourse (KhosraviNik & Unger 2016). More specifically, studies have also looked at online communities creating collective expertise, for example in the field of IT (Coleman 2017). Research has also taken note of an ideological struggle around whose expertise is to be trusted in a risk society (Beck & Beck-Gernsheim 2001), especially in a globally relevant political climate that has been discussed as one of anti-intellectualism (Merkley 2020) and with reference to antiauthoritarian online movements. In online contexts, especially, which are marked by the absence of cues available in face-to-face interaction, Seargeant and Tagg (2014) argue that trust is established through linguistically projecting particular identities. A crucial aspect of construing a credible identity online is authenticity - understood as a connection between the online and offline personae - which they describe as foundational for establishing trust in interaction with ultimately unpredictable, diverse, and theoretically almost infinite audiences (7-8). On forums, where expertise based on various sources of knowledge (from formal education to personal experience) is being exchanged and re-negotiated in a general political climate of distrust, the importance of being honest and explicit about the epistemic status of one’s utterances is likely to be perceived as important. As my analysis has shown, epistemic management is one of the key functions of disclaimers of expertise. Beyond that, negatively identifying as an expert of some sort may also fulfill other functions on forums: not being an expert can 314 9 The Discourse Functions of Negative Self-identifiers <?page no="315"?> serve to implicitly align with particular groups of non-experts, for whom other, more experiential and subjective forms of expertise may be of greater relevance. Re-expertization by informed choice refers to the conceptual strategy by which speakers establish themselves as authentic and, thus, credible, by drawing on other notions than that of formal expertise. As I have argued, lay expertise has become equally or even more important than formal expertise for adviceand support-seekers in online communicative settings, for example in the context of health (Taylor & Bury 2007; Barker & Galardi 2011), but also as consumers con‐ sult online reviews of products when faced with a buying decision (Vermeulen & Seegers 2009; Mackiewicz 2010a, b; Vásquez 2014). Against this background, negative self-identification as an expert can show speakers’ awareness of and affiliation with the lay audience they are addressing and create a sense of belonging to a group (of product reviewers, for example). Alternatively, the notion of expertise may not be explicitly referred to at all, as other attributes of identity could be increasingly important in the context of web forums. The discursive practices on web forums may follow their own interactional norms implicitly agreed on by the forum community and/ or explicitly manifested by affordances of the site. For example, more experienced forum members might be accredited with more expertise than new posters, and the importance of individual posters’ status on a particular forum may be reflected by the very design of the site (for instance, forum members mostly have profiles indicating their status in terms of experience, popularity, etc.). Given that authenticity and, thus, trustworthiness, is seen as coherence between online and offline persona, the personal tastes and preferences of individuals interacting online may be more important for forum members’ authority. As the analyses of NSIs presented in the previous chapters suggest, NSIs are most often used to contrast speakers with particular preferences and, thus, manifest their awareness of the choices available in a particular situation (e.g., when making a purchasing decision). This means that making explicit what one does not like could serve to position speakers as authentic in the sense of having their own opinion that is based on personal experience. They could also interact in interesting ways with other linguistic strategies of asserting expertise (as suggested, e.g., by Richardson 2003, Mackiewicz 2010a, Rudolf von Rohr et al. 2019). As the discussion below indicates, for example, self-representation as authentic and, thus, credible, can also involve using a playful, rhetorically self-conscious language style (de Lange et al. 2015), and thus entirely transcend the paradigm of traditional expertise as an indicator of trustworthiness. In the following, I discuss two instances of negative self-identification in detail, one from the domain of expertise and one from the domain of preferen‐ 9.3 Qualitatively studying negative self-identifiers in context 315 <?page no="316"?> ces, to illustrate how these conceptual strategies can be accounted for when scrutinizing language used in interactions on forums. This section moves from the microlinguistic analysis of NSIs to a broader discussion of their role in online discourse communities. Two examples are examined in detail, with attention both to their linguistic form and to the forum contexts in which they appear. The analyses provide a starting point for engaging with work on identity performance in online forums (see Sections 2.3.1 and 2.3.2) and with debates on (lay) expertise and contemporary struggles over knowledge and authority (see Sections 2.2.1 and 2.2.2). 9.3.2 Disclaiming expertise In this section, I discuss an instance of an NSI being used by a speaker to epistemically frame the interpretation of their utterance, managing their identity in interaction with other forum users and orienting to particular con‐ ceptualizations about expertise relevant in this context. This exchange, dating from 2017, comes from the forum hosted by Retropie, a website that presents and offers a (donation-based) download of the software RetroPie, which “allows you to turn your Raspberry Pi, ODroid C1/ C2, or PC into a retro-gaming machine” (retropie.org.uk). On this forum, registered users can have general discussions about gaming, “show off ” their “custom builds, themes and splashscreens” (ibid.) and get help and support. This is thus a forum for people who engage in building and enhancing their own hardware and co-create, rather than merely consume, the games they play. RetroPie hence represents a highly specialized field of interest, and a quite advanced level of expertise in hardand software is required to be able to even participate in the discourse on this forum. This impression is supported by the fact that the help and support forum asks users to refrain from posting any requests until after they have read and followed the advice on the website. At the point of data collection and analysis (December 2020), the forum was very lively, with new postings appearing regularly. Users need to be registered to post, usernames do not generally reveal much - if anything - about users’ offline identities (e.g., gt700, DonkeyKong17, tsinapah) and forum members use avatars, many of which are retro-looking, pixelated characters from games. Thus, the identities that users of this forum represent online show very little about (other aspects of) their offline identities. Rather, they linguistically and visually foreground those aspects that are relevant and community-building on RetroPie: interest in gaming and programming (as well as a ‘retro’ nostalgia for the latter). There is a contrast here with social networks such as Facebook, 316 9 The Discourse Functions of Negative Self-identifiers <?page no="317"?> where, according to Page (2014: 48), it is likely that “interactions online influence the identities and interactions that spill over into offline contexts”, which is connected to a certain risk (e.g., of losing one’s reputation through a posting judged as problematic by one’s audience). However, in the context of forums such as the one provided by RetroPie, the focus is not on individuals and their real-life identities, but solely on gaming as an interest shared by forum participants (as I have argued in Section 2.2.1.2, the notion of collective expertise is an important feature in ‘nerd’ discourses). At the same time, though, it should be mentioned that the shared interest of these users is one that requires a high level of expertise about gaming and programming gained through real-life practice and design skills (cf. Coleman 2017: 92) - as the very fact that these avatars are mostly self-created indicates. Thus, while the risk of threatening one’s real-life reputation can be deemed low on a forum such as this one, the risk of losing face by coming across as an amateur in the forum community appears particularly high. This may be amplified by users’ awareness of the design of the forum: there are forum admins and moderators, who are likely to enjoy authority status, and the number of views is displayed for each posting. Against this background, B’s use of two NSIs in example 9.62 below, at the beginning and again at the end of their response to A’s question posting, can be seen as disclaiming expertise at a very high level: 9.62. A: Im running a raspberry pi 3 with Windows 10. I have retropie 4. I was trying to fumble my way around getting mames to work and in the heat of frustration I deleted my mame4all folder. How do i reinstall it? - B: I am no expert but I didn’t think the Windows 10 IoT core even had access to the GPU, so you cannot drive a display properly? I thought it just shared it’s name and is not actually Windows at all. What are the benefits of trying to doing this anyway? I really don’t understand why you would want this, with closed source software the learning stops at some point, you can’t learn stuff about your Windows-Pi even though it was created for learning! Even the example code (read a temp sensor, display using webserver) they give on the home page starts with: “Copyright (c) Microsoft. All rights reserved.” Intrnet [Internet] of things on a Pi was sold to me once as a way to make it easier to connect things that are attached to the Pi (like sensors, motors, lights, etc) interact with a Windows 10 PC but not a way to run programs or apps. BUT, like I said, I am no expert on this and times may have changed, just thought this might save you time. Also, you really need to read this: https: / / retropie.org.uk/ forum/ topic/ 3/ read-this-first before asking for help, especially with something odd like this. - A: raspberry pi 3 has wifi and can be discovered on the network. I was moving roms to the rom folder on the SD card from my PC. I went to delete a rom and accidentally deleted the mame4 all entire folder. 9.3 Qualitatively studying negative self-identifiers in context 317 <?page no="318"?> Could i just recreate a folder, because I dont think it would have deleted the entire emulator. - B: Yes, I just don’t get why you are using this IoT windows? It makes no sense to me and I would like to learn why you wouldn’t just use the image provided. Do you think that without Windows, it can’t be discovered on the network? - C: @B no offence to OP but I’d be willing to bet that most people who are running Windows 10 iot on the pi would probably know the simple steps of recreating a folder if the got to the point of getting retropie functioning on iot ; ) On the micro level, the first NSI epistemically mitigates the coordinated clause introduced by but, which represents a mental process type and instantiates what I have referred to as knowledge representation/ opinion in the analyses presented in this chapter so far. The speaker’s claim is further marked as tentative by their use of a question mark at the end of the sentence. The second NSI at the end of their posting - introduced by an emphatically capitalized BUT - seems to serve as a final reminder to the audience of the low certainty of the claims the speaker is making in their relatively lengthy posting. B’s second posting, too, represents this user as careful and modest toward their interlocutors. The user stresses that they “would like to learn” and comments on their lack of understanding (“it makes no sense to me”). The speaker is clearly aware of the epistemic delicacy of the question asked and their response, as they refer A to the manual intended, simply put, to avoid stupid questions being asked in the help thread - in a sort of post-script introduced as such by also after the epistemic disclaimer framing their main utterance. The epistemic mitigation strategies used by B appear justified if we assume that B reflexively designed their utterance with the audience of other IT ‘nerds’, but also with the affordances of the forum per se in mind. As stated above, the fact that users need to be registered to contribute and have individualized profiles certainly puts a certain pressure on them in terms of their reputation among peers. Additionally, the user is probably aware of the relatively high number of views postings may get (e.g., this discussion was viewed 1,600 times; but there are also postings with 84,000 views). Looking at the language used in this discussion in more detail reveals that B’s interlocutors linguistically represent themselves as experts by using a variety of strategies of asserting expertise (see Section 2.2.1.3): the discourse is marked by the use of technical terms known to the community of Raspberry Pi users (e.g., raspberry pi 3, retropie 4, mame4all folder, rom folder). But there are also more self-conscious assertions of expertise in this discussion, such as C, a forum admin who apparently followed A and B’s exchange, ridicules their assumption that a 318 9 The Discourse Functions of Negative Self-identifiers <?page no="319"?> 72 At least from a lay perspective - I am not an expert in retropie4 matters. person capable of running Windows iOT on a Raspberry Pi might not be able to recreate a deleted folder. However, C linguistically expresses awareness of their utterance’s potential to threaten the face of the original poster (“OP”) and B’s advice, marking their utterance with the words “no offence”. At the same time, speaker C positions themself as an expert by referring to the steps of recreating a folder as ‘simple’, and, thus, judging A’s question and B’s answer as interaction among people with a lower level of expertise. Referring to the discussion of identity as contractual achievement (Leppänen et al. 2014: 112), this interaction shows that B’s attempt of being helpful and, thus, representing themselves as more knowledgeable than A by referring them to the must-read-manual, is sanctioned by C, who does not accept this speaker as authoritative on the subject and implicitly confirms their non-alignment with the group of experts. The use of the NSI in this discussion thus suggests that all three interlocutors are aware of, and, in their self-representation, orient to, their perception of the generally high level of expertise marking the discourse on the forum. As B’s careful epistemic framing of their posting and C’s response suggests, asking the right questions and giving responses based on reasonable background assumptions about what levels of expertise can be expected of peers is important when performing identity on this forum. Theorizing media ideologies, Gershon (2010: 284) argues that “[p]eople’s understandings of both language and media will shape, although not determine, their communicative practices”. Citing Silverstein (2001), she highlights that ideology underlies people’s beliefs about what is effective communication. Against the background of a general sociopolitical climate marked by a heightened awareness of, and ideological struggle around, different forms and levels of expertise circulating online, it is particularly interesting that in online communities sharing a common interest and concomitant expertise in a specific field, even minor knowledge differences 72 are linguistically recognized as such: it shows that these forums, representing communities of like-minded people seeking to jointly and democratically advance knowledge in a particular field, have their own standards of what represents expertise and authority of knowledge. These, at least in the case of the forum examined here, exert a key influence on what is perceived as appropriate communicative behavior, with the must-read manual mentioned above formulating explicit guidelines on what questions to ask (and, in fact, even how to format and tag them). The use of the epistemic disclaimer discussed here can be considered a micro-level pragmatic manifestation of users’ orientation to these standards. As a patterned 9.3 Qualitatively studying negative self-identifiers in context 319 <?page no="320"?> linguistic choice in similar discourse contexts, NSIs of this kind can thus be seen as highlighting developments in the social world at large, with supra-local, superdiverse communities of lay experts coming to represent key spaces for creating knowledge in ever more specific fields. 9.3.3 Disclaiming preference The following exchange from AVforums features an instance of an NSI from the second main conceptual category the analyses have identified, namely preference disclaimers. AVforums is a platform self-describing as “The No. 1 Home Entertainment Tech Community & Resource”, which means that it is a forum devoted to reviewing products online. As is the case on Retropie, forum users have to be registered to participate in discussions and have publicly visible user profiles, featuring their usernames (mostly aliases), providing information about when they joined the forum, how many postings they have contributed and how many reactions these obtained, as well as defining them according to their experience and authority as a poster by ranking them on a scale between “novice” and “distinguished member”. This already indicates that consumers’ expertise is a defining feature of their identity as discursively construed and negotiated on this forum. At the same time, users also make use of signature lines to personalize their profiles, which indicates that they also exploit the affordances of the medium to represent themselves as individuals with their own preferences and beliefs. This, in turn, suggests that (at least longer-term) users care about their reputation among peers, and design their utterances in awareness of the rest of the forum community. In exchange 9.63, speaker A negatively identifies as “fan of the Home Max Speaker”, a product which they report to have been looking forward to. 9.63. A: I’m very much into Google and Smart home stuff, have a LOT of various Google/ Next Smart speakers, Hubs, Smoke detectors and doorbells. Was looking forward to ‘finally’ a redesign of the standard Google Home speaker, and looks like we’ve just seen it. [personal opinion] I’d [I’m] sad, as I don’t like it. Looks like something with zero design, zero character, bland bland bland. A band across it, a stripe, a groove, something, ANYTHING might have helped, but this just looks like nothing. Perhaps a bar of soap or the box something comes in? : ( I’m 100% sure it will sound better, but am sad if this is actually it. : ( Is this just me feeling this way? - B: I personally think it looks better than the previous Google home. It’s not supposed to stand out and that’s the point. It’s supposed to be subtle and blend in rather than look like a “Smart Speaker”. It’s all personal opinion though. 320 9 The Discourse Functions of Negative Self-identifiers <?page no="321"?> A: Yes, I can see that, it certainly does not stand out. In that photo, it almost looks like it’s a box which might have some headphones inside. ; ) Honestly I’m not after much. Perhaps just a little metallic band across it, in a similar tone to the fabric. Just some small element of design that takes the edge of the fabric box look. To be totally honest, I’m not a fan of the Home Max Speaker for the same reason. It’s just lacking something, just an element to stop it looking like a fat grey lump : ) All personal opinion of course : ) Oddly enough that rippled dark matt blue wall behind it, looks almost exactly the same as the colour scheme in my back bedroom! Speaker A is a “distinguished member” on AVforums, has published several thousand posts over more than a decade, and received thousands of reactions from other users. They also seem to place importance on their image on the forum, using a profile picture and a humorous signature line to convey personality. Providing cues about one’s social identity has, indeed, been found to be an important factor in creating an authentic and hence credible reviewer persona on online forums (Vásquez & China 2019: 193). Furthermore, A’s use of the preference disclaimer is hedged by an introductory “to be totally honest”, and generally, both A and B use several linguistic devices of marking their reviews as personal opinion: explicitly, through using the metadiscursive labels [personal opinion] and all personal opinion of course : ), in which the smiling emoji appears to serve as an additional hedge, but also more implicitly through a number of verbs and adjectives referring to emotions, for example, look forward, feel, be sad. Besides framing their own evaluation as subjective, speaker A also addresses the forum community, asking for their feelings about the product in question, designing a context in which reviewing products is a matter of opinion, rather than expertise. This is also supported by the epistemic mitigation expressed by “I’m not a 100% sure”. Mackiewicz (2010a), drawing on Beason (1991), argues that construing expertise is based not only on displaying knowledge, but also on self-representation as ethical in the sense of having “good intentions”. By rendering their evaluation of the Google home speaker as very tentative and emotional and showing interest in the feelings of forum peers, speaker A projects a modest, non-imposing image of themselves. Interestingly, though, despite being explicit about the subjective character of their evaluation - which microlinguistically crystallizes in their use of the NSI “I’m not a fan” - this speaker rhetorically stylizes their utterance in a way that makes them appear as very self-assured about their subjective product assessment. The speaker, for example, emphasizes their dislike by repeating the 9.3 Qualitatively studying negative self-identifiers in context 321 <?page no="322"?> adjective bland three times for rhetorical effect, a technique that dramatizes the perceived design failure. They also employ a longer list with an end-climax (A band across it, a stripe, a groove, something, ANYTHING), which enumerates suggestions for improving the design of the product and is thus implicitly addressed to Google, as the producer of the speakers being discussed. Through stating that “anything might have helped” to improve the design, the speaker construes the task of product design as an easy one. Thus, they imply, on the one hand, that Google failed at even a basic job and, on the other, that they know better. Speaker A also makes reference to their familiarity with products from the same brand (“I’m very much into Google and Smart home stuff, have a LOT of various Google/ Next Smart speakers”), which, according to Mackiewicz (2010a: 12), is another strategy of linguistically construing expertise in online product reviews. This suggests that in the context of reviewing products, negative self-identification with preferences is used to linguistically index the subjectivity of product assessments. These assessments, however, construe speakers as experts. Judging from the exchange discussed here, not identifying as a fan can therefore be seen to interact in two ways: (a) with speakers’ discursive construal as credible and likeable peers, who present themselves as ordinary consumers merely report‐ ing their personal (non-)preferences while managing face within the forum in-group; and (b) with speakers’ self-representation as expert consumers, acting as a self-assured ‘product jury’ that positions an out-group of sellers under scrutiny. Negative self-identification can thus index participants’ orientation to the forum community, as the analyses in this section illustrate, while also signalling awareness of wider digitally mediated contexts, where disalignment with product features positions a poster as an opinionated expert customer, linking to broader debates on lay expertise (see Section 2.2). 322 9 The Discourse Functions of Negative Self-identifiers <?page no="323"?> 10 Discussion and Conclusion: I Am What I’m Not The goal of this study was to understand how speakers position themselves in web forum discussions using variants of the phrase “I + copula + NOT + indefinite NP”. To pursue this goal, I began by exploring the philosophical and discursive implications of saying what one is not. As discussed in Chapter 2, such statements raise fundamental questions about what it means to ‘be’ someone, highlighting that identity is not a fixed essence but a socioculturally variable and historically contingent construct. Chapter 3 argued that positioning oneself negatively in discourse is linguis‐ tically significant because negations presuppose affirmatives - which may never be explicitly stated. As such, negative self-identifications often reveal speakers’ situational perceptions more than any stable ‘identity’ - if identity is even best conceived as a nominal concept at all. Building on this, Chapter 4 drew on functionalist approaches to language to argue that identity is perhaps better understood as a process. Just as language itself is increasingly seen as dynamic and performative, particularly in digitally mediated, disembodied contexts where diverse social settings and lifeworlds collapse into a single communicative space, so too might identity be better conceptualized as languaging (Page et al. 2014). Chapters 5 and 6 then detailed the empirical design of the study, addressing the methodological and ethical considerations involved in sampling, processing, and analyzing forum discourse. The analyses presented in Chapters 7, 8, and 9 revealed how NSIs function at the microlinguistic level. Key findings include: • Certain conceptual categories of self-identification - especially preferences and expertise - are particularly prominent. • NSIs tend to co-occur with specific discourse contexts, indicating shared pragmatic functions and discourse-level expectations. • These patterned uses shed light on how speakers discursively manage authority and authenticity, often in relation to topic-specific norms and perceived audience expectations. The remainder of this chapter interprets these findings in depth. It first revisits the key usage patterns and functional tendencies identified in the corpus, then considers their implications for the discursive construction of self and expertise <?page no="324"?> online. Finally, it reflects critically on the ideological issues raised by these patterns and considers the contribution and limitations of the study. 10.1 From patterns to profiles: A functional overview The analyses revealed several consistent usage patterns and pragmatic functions of the structure, which are summarized below. A central finding is that NSIs in the corpus cluster around two particularly salient conceptual domains: “preference” and “expertise and professionalism”. These two categories dominate both in terms of frequency and in the variety of forms they take, with the nouns fan and expert accounting for a striking proportion of the data (198 and 193 instances respectively - together 391 of 936 NSIs, i.e., 42%). Their prominence suggests that negative identification in these areas plays a central role in how speakers discursively navigate taste, authority, and credibility online. The linguistic realization of these categories is varied: both fan and expert appear as heads of complex nominal phrases, often preor postmodified, and embedded in diverse discourse functions. Their use points to two recurring conceptualizations that underlie much of the structure’s pragmatic work: de-expertization (the explicit rejection of expert status) and re-expertization by informed choice (the indexing of credibility through knowledge and discern‐ ment rather than formal authority). There also appears to be some thematic alignment between forum topics and the conceptual categories of NSIs. For example, there is a tendency for health-re‐ lated disclaimers to occur in health forums and for preference disclaimers to cluster in product or leisure discussions. While, as previously discussed, these patterns are based on a broad and approximate categorization of forums, they nonetheless support the broader observation that negative self-identification is a highly context-sensitive practice, shaped by speakers’ perception of what kinds of identity categories are interactionally relevant. Beyond the categories themselves, the co-textual positioning of these struc‐ tures also reveals patterned usage. As shown in Chapter 8, the most frequent relation is contrast (396 instances), followed by addition (162) and cause/ conse‐ quence (158). These patterns suggest that the structure often serves to preface or frame subsequent contributions and functions as part of a broader strategy for managing stance and mitigating epistemic authority. The transitivity analysis of co-texts confirmed this pattern: across all three relation types, mental process clauses with the speaker as senser are the most frequent (192 cases). In over half of these, the matrix clause is a disclaimer of expertise or professionalism. Other 324 10 Discussion and Conclusion: I Am What I’m Not <?page no="325"?> frequent configurations include relational (125 instances) and material processes with the speaker in the actor role (84 instances). In other words, NSIs in this corpus are often embedded in contexts that foreground thought, experience, and subjectivity, rather than objective knowledge claims. Chapter 9 extended this analysis by exploring how these co-textual con‐ figurations contribute to the broader discourse functions of the structure. The most common types of co-text - those metadiscursively commenting on knowledge (155 instances) and those representing experience (61 instances) - again show the dominance of expertise disclaimers, reinforcing the idea that epistemic positioning is a central concern for speakers. When the focus shifts to tastes, habits, or emotional responses, preference disclaimers take precedence, especially in evaluative and opinion-laden discourse. A complementary analysis of preceding co-texts further supported this pattern. Of the 376 instances examined, 67 were preceded by product experience stories, 26 of which were followed by preference disclaimers. Statements pre‐ senting or commenting on information (38 instances) were frequently followed by disclaimers of expertise (27), while advice-giving contexts (34) mainly preceded instances of NSIs from the domain “expertise and professionalism” (23 instances). These findings indicate that NSIs are shaped by, and help to shape, the surrounding discourse - whether aligning with personal experience, hedging informational claims, or softening directive speech acts. Finally, Chapter 9 concluded with two case studies that illustrated how these patterns unfold in specific discursive and interactional contexts. In a programming forum, a speaker rejected the label expert, implicitly acknowledg‐ ing the high level of competence among forum members. In a product forum, disidentification as a fan emerged as a key component of a speaker’s effort to construct a persona of critical and authentic consumer expertise. These examples show how self-representation via negative identification is not merely a reflex of politeness or modesty, but a strategic form of positioning within local epistemic and social hierarchies. Together, these findings demonstrate that the matrix structure under inves‐ tigation is not simply a grammatical curiosity. Rather, it is a pragmatically loaded form that indexes socially meaningful categories, reflects situational expectations, and helps speakers manage the tensions of expertise, preference, and authority in everyday online interaction. 10.1 From patterns to profiles: A functional overview 325 <?page no="326"?> 10.2 Self-representation as epistemic positioning From a pragmatic viewpoint, it can be concluded that certain types of NSIs are used like stance markers, routinely serving the same or very similar functions as they interact with textual and situational contexts. As the analyses have shown, there is a tendency for certain thematically defined forum types to co-occur with certain conceptual categories of NSIs. For example, negative identification with lifestyle categories (e.g., as a smoker, morning person etc.) frequently occurs in the context of health forums, which indicates that speakers deem these categories relevant when talking about health issues and, thus, that they share background assumptions about the topics discussed on these forums. Self-representation in forum discourse is highly situation-dependent: what speakers say about themselves hinges on topic and audience, and different discourse contexts prompt them to linguistically index different aspects of their identity - whether situationally and discursively relevant or more permanent. As argued before, according to Vásquez (2014), this can be considered to manifest speakers’ underlying assumptions about how to best represent themselves in the context of online forums. Some categories people contrast themselves with, however, appear to stand out transtextually, serving the same functions irrespective of the kind of forum or topic of the discussion, notably NSIs from the conceptual domains “expertise and professionalism” and “preference”. From a discourse-analytical perspective, it is these routinely used, pragmaticalized uses of NSIs which appear most interesting, as they point to socially shared conceptualizations which may have become so naturalized that they inform how people talk on a day-to-day basis, on forums and possibly in other discourse contexts as well. As for disclaimers of expertise, they were found to be frequently used across forums to serve essentially the same function in similar co-texts. Notably, they are frequently used by speakers to epistemically mitigate information they provide and comment on to reduce the hierarchy of authority to knowledge implied by advice-giving and to epistemically modify the interpretation of experience stories they relate on forums. Thus, they seem to function to create common ground among forum participants, framing personal opinions and experiences as non-authoritative and thus helping to represent speakers as modest and polite. By routinely contrasting themselves with experts and professionals while discussing topics that would traditionally be associated with high levels of expertise, forum participants - if their utterances across texts are considered as a whole - represent themselves in contrast to what they perceive as ‘experts’. Thus, they create a sense of pertaining to a group of lay experts, discursively positioning personal and experiential knowledge as expertise (Williams 2014; Rueger et al. 2021). If statements can be rendered more modest by negatively self-identifying as an expert, this suggests that identifying as 326 10 Discussion and Conclusion: I Am What I’m Not <?page no="327"?> one could potentially be considered immodest and, thus, that the group of experts and professionals is discursively construed as superior (Beck & Beck-Gernsheim, 1992: 57). This superiority, however, is undermined at the same time as NSIs are used in co-texts presenting knowledge, opinions and advice. Such uses probably fulfill local discourse management functions - in-group solidarity is possibly felt to be stronger against the background of an implicit ‘other’ perceived as somewhat ‘out of reach’. This resonates with research on lay expertise, which shows that forum users often value peer advice highly, considering it to facilitate interaction with professionals or even to be more beneficial than counselling by accredited experts (e.g., Mattson & Hall 2011; Rupert et al. 2014). My findings on the use of NSIs may therefore also have implications beyond the immediate communicative situations in which they occur - a point developed further in Section 10.3, where I connect local pragmatic functions to broader negotiations of trust and expertise. The second conceptual category of NSIs, preferences, stood out in all analyses not only because it is frequently instantiated by the same lexeme, viz. fan, but also because it is used in similar co-texts across texts to serve similar functions. While the functions of preference disclaimers of the type I’m not a fan of Diesels are not as easy to pinpoint as those of disclaimers of expertise, they were found to perform interesting functions across forum discussions examined: not being a fan - notably of particular products - appears to be a prominent strategy of conveying taste and the ability to make informed choices. It was found that preference disclaimers sometimes appear in highly technical co-texts, where they seem to act as indices of authority. I have argued that this, viewed in relation to the prominence of expertise disclaimers in the examined data, could indicate that in the context of web forums, self-representation in terms of non-preferences is a popular strategy of establishing credibility. In other words, while speakers apparently feel the need to epistemically mitigate their utterances (which, as we have seen, are often already framed as their opinion by virtue of being projected by mental processes with the speaker in the senser role), they do not hesitate to say what they do not like. This, on the one hand, can be interpreted as reflecting the nature and purpose of forums, which prototypically constitute sites for discussing opinions among like-minded laypeople rather than sharing information from a position of authority (cf. Vásquez 2014: 66 on forms of expertise in online contexts). On the other hand, the analysis has shown that speakers in my corpus often negatively self-identify as experts or professionals in contexts of linguistically acting like experts and use preference disclaimers to establish authority of a different kind. This points to a more permanent, culturally conditioned, underlying conceptual struggle around knowledge, opinion and choice, and authority. More specifically, I think that the 10.2 Self-representation as epistemic positioning 327 <?page no="328"?> 73 Figures in this chapter reproduce memes for scholarly commentary under fair use. 74 MEME. (n.d.). I’m not an expert myself, but the experts are wrong [Meme]. Retrieved January, 2020, from https: / / me.me/ i/ im-not-an-expert-myself-but-the-experts-are-wro ng-e9914076fa1647a696703062656d05e8 75 Make a Meme.org (n.d.). Not an expert, still an authority figure [Meme]. Retrieved January 2020, from https: / / makeameme.org/ meme/ not-an-expert use of NSIs examined here can be considered to indicate a reconceptualization of what constitutes reliable knowledge, with the boundaries between personal experience and choice and authorized information becoming simultaneously more pronounced (non-experts vs. experts) and increasingly blurred (if nobody is an expert, what remains are experience and opinion) (cf. Seargeant & Tagg 2014: 3 on previous dichotomies of amateur and professional potentially becoming obsolete). Next, I will elaborate on what I think is the key ideological issue at stake here. 10.3 Between expertise and opinion: Authority and identity in online forums The results of this study indicate that being or not being an authority on a particular subject appears to be a question implicitly present on numerous occasions, informing our conversational style insofar as we deem necessary to explicitly address it. This can be illustrated easily by informally consulting the web and googling the phrase “I’m not an expert”. The results point toward an awareness among the web community of the frequent usage of this phrase, mocking it in countless memes, some of which are shown below 73 : Figure 10.1: Expert meme 1. 74 Figure 10.2: Expert meme 2. 75 328 10 Discussion and Conclusion: I Am What I’m Not <?page no="329"?> 76 Make a Meme.org (n.d.). I’m the expert, I know everything about everything! I am an expert in everything, even things I can’t pronounce… [Meme]. Retrieved August 2025, from https: / / media.makeameme.org/ created/ im-the-expert-40a906390e.jpgSource: htt ps: / / media.makeameme.org/ created/ im-the-expert-40a906390e.jpg Figure 10.3: Expert meme 3. 76 All three memes ridicule people who, despite admitting to not being experts, reject expert opinions and claim authority. This appears to reflect a conceptual struggle around authority and the question of what aspects of people’s identities can be regarded to substantiate views they share online and contribute to public discourse which, especially in the case of political decision-makers, have far-reaching ramifications: the issue, it seems, is a perceived conceptual contrast between authority based on opinion and authority based on expertise. However, the results of this study suggest that the line between these two notions is becoming increasingly difficult to draw in the context of online forums. As Carr (2010) argues, what constitutes expertise is not merely a matter of formal qualifications but hinges on the speaker’s ability to enact appropriate register choices that mediate between cultural knowledge and a presumed non-expert audience. This view supports the idea that discursive displays of knowledge - even when framed as non-expert - may still function as enactments of expert identity. 10.3 Between expertise and opinion: Authority and identity in online forums 329 <?page no="330"?> It was found that while NSIs are mainly used as epistemic mitigators, indi‐ cating that speakers’ perceive a difference between expert and lay knowledge, negative self-identification as an expert or professional does not necessarily preclude acting like one: the study revealed, firstly, that epistemic disclaimers are used in co-texts representing speakers’ knowledge, hedging information nevertheless shared, and, secondly, that preference disclaimers are used by speakers to index awareness of options and thus expertise, especially when it comes to making consumption choices. In this sense, disavowing expertise may not signal a lack of authority, but rather a strategic move to avoid discursive closure and maintain an open, participatory stance. Au and Eyal (2022) highlight how expert identities are often contested and negotiated, and suggest that rejecting the label of ‘expert’ can be a means of continuing a discourse without invoking the legitimizing (and potentially excluding) power of expertise. Beyond the data examined here, these findings can be related to a controversy about the notion of experts in the wider social context: on the one hand, even very small-scale decisions have become highly significant for people’s reflex‐ ively constituted self-identity and the amount and accessibility of information has exponentially increased, which is why seeking help from information mediators has become more and more important. On the other hand, in the words of Jensen et al. (2012: 2), “expert knowledge is generally contested and branded with uncertainty” because to be an expert does not necessarily mean to have formal qualifications - on the contrary, the emphasis on the practical relevance of knowledge for decision-making, the focus on action rather than contemplation, is what characterizes the notion of expert in the first place. This, according to Grundmann (2017), has led to a paradox of reliance on and skepticism toward experts: The emergence of the knowledge society has led to a proliferation of, and dependence on expertise. It has led to a loss of trust in scientific experts while at the same time generating forms of expertise that are not based on professional accreditation or scientific reputation. Put more simply, given that there is so much to know and so much information that needs to be consolidated to create knowledge, being an expert in more than a tiny fragment of a subject is hardly possible, which could explain that people interacting online deem it necessary to highlight that they are not experts, but sharing informed opinions. At the same time, becoming an expert in a particular field does not necessarily require formal education, which could explain why negative self-identification as a fan features often in discussions on particular 330 10 Discussion and Conclusion: I Am What I’m Not <?page no="331"?> products, serving to highlight speakers’ knowledge of, and experience with the items they are reviewing. It also appears worth mentioning that this reconceptualization of expertise in online communicative contexts seems to be prone to creating unease and controversy. For instance, the following extract from a digital agency’s blog indicates struggle around the criteria for knowledge to be considered valid and trustworthy: My problem with user-generated content is that it gives the impression that every‐ one’s opinion is equal, and that’s simply not true. My wife is a highly trained ICU nurse, meaning that while I may think that the bone sticking out of my leg is nothing to worry about; I need to listen to her and go to the hospital. After all, she’s the expert. Our digital culture has given us all a platform where we can express our opinions to the entire world, but that doesn’t necessarily mean we should listen. In a world where everyone is an expert, nobody is an expert. (wedü 2019) As van Dijk (2002) explains, societies are based on shared sociocultural knowl‐ edge, that is, common ground, which includes not only knowledge, but also eval‐ uative elements as well as beliefs about knowledge criteria. This presupposed knowledge becomes empirically observable in the form of presuppositions in discourse, that is, in the things we need to know to understand a text. Language users verbally indicate the status of their knowledge. For example, as we have seen, NSIs in this study were found to often occur before sentences starting with such epistemic markers, for example, I know. Thus, the most frequently observable function of NSIs in the examined data is to epistemically modify an epistemic index - people not only mark their opinions as such, but additionally index the authority status of their opinion. The kind of knowledge normally of interest in CDS is group knowledge, with a focus on group ideologies. This study examined a phrase used by people pertaining to various online groups, so the results can provide insights into conceptualizations structuring people’s self-representation across forum communities. That being or not being an expert is presupposed as a quality criterion for people’s opinions irrespective of their group membership or the topic discussed indicates that this structure has become normal to use, functioning like a pragmatic marker. This is interesting insofar as it suggests that expertise is implicitly acknowledged as relevant notion in various online groups, which could be seen as reflecting developments in the social world at large. If we believe theories of populism (e.g., Mudde & Kaltwasser 2012; Montgom‐ ery 2017) society today is experiencing more than a mere opposition between groups with different views - we appear to be facing a more fundamental issue 10.3 Between expertise and opinion: Authority and identity in online forums 331 <?page no="332"?> of distrust, blurring the boundaries between what is considered knowledge and what is considered mere opinion, the effect of which might be an increasing deconstruction of groups jointly agreeing on and thus legitimizing knowledge and a tendency toward individuals becoming their own source of knowledge and opinion. This can have both democratizing and anti-democratic effects (Riedel 2017). In online contexts such as forums, individuals can unite to freely share and negotiate expertise that is “geographically dispersed, and interpersonally removed” (Vásquez 2014: 66). This knowledge can represent a contrast to traditional expertise (a notion which, as has been discussed, has been referred to as lay expertise, by, e.g., Williams 2014), which may contribute to the wellbeing of people interacting online, as shown by the example of lay expertise in the field of health (Barker & Galardi 2011). The views shared within these communities may also undermine those held by what is considered the ‘elite’, and this may have problematic implications, as shown for instance by what has been referred to as right-wing populist “echo chambers” (KhosravNik 2017). The discursive foregrounding of non-expertise construes a difference not only between the online community, exchanging lay expertise, and the group of experts, but also fulfills face management functions within the forum community. Thus, using expertise disclaimers can serve to construe forum participants as modest lay experts, who carefully mitigate their views, and ignorant non-experts, who - as one of the memes above puts it - have no idea, but strong opinions. Regardless of which actual ideological backgrounds these language users may come from, the discursive perpetuation of a mélange of skepticism, modesty, and claims to truth can be conducive to a friendly and constructive atmosphere on a web forum. This supports Au and Eyal’s (2022) view that distancing oneself from the role of expert can enhance discursive participation by reducing the social risks associated with epistemic authority. However, other corners of the online world not explored here, as well as recent political events, marked by distrust toward and instability of even the traditionally most authoritative political and legal institutions, suggest that a gradual loss of trust among and between members of society, manifesting itself in even the most mundane discourse becoming an epistemological tightrope walk, might not be supportive of a peaceful, democratic and constructive societal discourse (but see more on this question below). Yet language is never static: users continually renegotiate and reshape the norms of social interaction. The results of this study suggest not only an ongoing struggle around expertise, but also that other concepts may be emerging as less problematic and more effective strategies of face management in discourse. Representing oneself as an authority can increasingly appear to rest on fragile 332 10 Discussion and Conclusion: I Am What I’m Not <?page no="333"?> 77 Napster was a peer-to-peer file-sharing service launched in 1999 that allowed users to exchange music files directly, challenging traditional music industry distribution models (see Giesler & Pohlmann 2003a; 2003b). epistemic ground and may invite criticism from others - a risk that many web users are keen to avoid in times of heightened online controversy. Alternative ways of establishing trust, particularly those that foreground ethical stance, can therefore serve as more acceptable and persuasive options. This view is echoed by Mackiewicz (2010a), who argues that ethical soundness, in the sense of “meaning well”, is a key strategy for constructing a credible online persona. Indeed, as the analysis has shown, negatively identifying as a fan in often highly technical discourse contexts can function as a powerful means of communicating authority without requiring the risks of explicit epistemic commitment. Anyone can be a fan, and anyone can be a follower - legitimacy here derives not from institutional credentials but from the consumption choices that participants display and discuss. As Li (2021: 163) observes, fandom is best understood as a performative space in which individuals “share feelings and thought, interact […] and join and create communities”. She identifies active, rather than passive, consumer engagement and communal participation as central pillars of fan culture. Both are evident in the corpus: not only is non-preference a recurring identification strategy, but NSIs also frequently appear in metadiscursive comments, enabling participants to signal communicative self-awareness and to frame their identities as reflex‐ ively constructed in interaction with others. The empowerment of fans as ‘enlightened consumers’ can be viewed from at least two perspectives. On the one hand, the unprecedented ease of ac‐ cessing, producing, circulating, and contesting information carries significant emancipatory potential and is often rightly linked to the democratization of knowledge and processes of subpoliticization. On the other hand, studies of consumption communities, such as those surrounding Napster, 77 show how consumer practices - framed as gifting rather than monetary exchange - can inadvertently disrupt traditional market structures (Kim et al. 2016; Giesler & Pohlmann 2003a). At the same time, however, what appears as emancipation eventually oper‐ ates under the same socio-political conditions that confine, namely, consumer‐ ism: marketers have long recognized the potential of exploiting the content generated by online communities for strategic marketing measures (e.g., Wag‐ ner et al. 2017). According to Giesler and Pohlmann’s (2003b) discussion of online communities like Napster, their emancipatory potential is paradoxical because what makes it possible - communication about consumption - is also 10.3 Between expertise and opinion: Authority and identity in online forums 333 <?page no="334"?> an act of “autopoiesis of the social form of emancipation”. As they put it, “you have to chase the king to ensure your status; likewise, chasing the king, you ensure his status” (5). As indicated above, undermining expertise and formal knowledge by con‐ trasting them with personal opinion and experience could also be seen as problematic - insofar as the mere reliance on emotionalized ‘fandom’ can have very real consequences. Indeed, 2016 became known as the year of “post-truth politics” (Montgomery 2017) - a climate that, as Crouch (2017) observes, rests on the assumption that “emotional truth does not need evidence of a scientific kind; a feeling is true, if someone truly feels it”. Or, as Suiter (2016: 25) argues, in post-truth discourse, appeals to emotion are dominant and factual rebuttals or fact checks are ignored on the basis that they are mere assertions. This combination arguably results in the emergence of swathes of expressive voters moved by dangerous rhetoric and nativism. The swirling impact of these variables has helped undermine the legitimacy of the liberal order, opening the door to illiberal forces and increasing the potency of populist and nationalist appeals. The frequency of negative identification with expertise and professionalism, on the one hand, and with preferences, on the other, and the functions thereof, is of course only one microlinguistic pragmatic phenomenon which can be interpreted in relation to these larger societal trends. However, because of the microlinguistic focus of this study, the evaluation of its results in terms of their socio-political significance must remain tentative. Still, I hope to have demonstrated that examining how people use NSIs in routine, informal online interaction can serve as a productive starting point - or a meaningful contribution - to CDS. What people say they are not, it has been shown, can indeed say something about how they see the world, and exploring what many people, in similar discourse contexts and situations, claim not to be can contribute valuable insights into how discursive representation of self-identity relates to conceptualizations structuring socially shared beliefs about what can be said, by whom, in which situations. In the next, final section, I will critically reflect on this study as a whole, pointing out its contributions and limitations as well as potential for further research yet unexploited. 10.4 Contributions: Method, theory, and social insight This study advances research on negatives by examining them in their actual textual surroundings, as called for by Jordan (1998: 706) and Roitman (2017), 334 10 Discussion and Conclusion: I Am What I’m Not <?page no="335"?> and by offering detailed insights into the pragmatic functions of NSIs in digital discourse. Framing this inquiry at the intersection of interactionally oriented pragmatics and CDS, it demonstrates how meticulous linguistic analysis can contribute to our understanding of how contexts are dynamically construed in interaction (Fetzer 2007: 4). It affirms that discourse is not external to pragmatics but integral to it, and that discourse-pragmatic analysis can serve as a productive entry point into larger discursive formations. In doing so, the study supports Barron and Schneider’s (2014: 1) argument that the pragmatics of discourse and the pragmatics of utterances are complementary levels of analysis, and it illustrates what it can look like in practice when “micro meets macro” (Fetzer 2007). By focusing closely on the local functions of NSIs, the study follows a linguistic tradition concerned with mapping grammatical form to discourse function (Sinclair & Coulthard 1975; Edmondson 2014). It responds to calls for micro-analytical approaches that yield empirically grounded insights into online communication practices, in contrast to more speculative or anecdotal accounts (Herring 2004b: 338). The corpus-based methodology adopted here enables pattern identification through a replicable analytic process (Adolphs 2006: 7-8), reinforcing the value of corpus approaches for analyzing digital language use (Knight 2015: 20). Empirically, the study sheds light on how NSIs function as strategies of utterance design and identity positioning, contributing to research on identity work in digital discourse (Page 2012; Androutsopoulos 2014; Deumert 2014; Leppänen et al. 2014; Tagg et al. 2017; Bou-Franch & Blitvich 2018). The findings also illuminate broader discursive negotiations of epistemic authority and trust - a key concern in studies of online expertise construction (Mackiewicz 2010a, b; Vásquez 2014; Rudolf von Rohr et al. 2019; Rueger et al. 2021). These findings also exemplify the broader value of combining fine-grained pragmatic analysis with corpus methods to explore identity performance in context. While the study takes a specific linguistic structure as its starting point, its contribution lies in showing how local identity work, pragmatic positioning, and utterance design are tied to wider discursive negotiations of trust, legitimacy, and expertise. In this way, it speaks to broader questions about how everyday interactional choices are embedded in - and help reproduce - social meanings and norms. This highlights the potential of corpus-informed, discourse-pragmatic approaches to connect micro-level communicative strat‐ egies with macro-level concerns across digital contexts. More broadly, it shows how local stance work is tied to wider negotiations of trust and authority in digital contexts. 10.4 Contributions: Method, theory, and social insight 335 <?page no="336"?> 10.5 Looking forward: Reflections and research horizons This study has provided insights into the use of NSIs in informal UK web forum interactions, but it has also raised further questions and revealed methodological intricacies worth noting. At first glance, studying the functions of variants of just one sentence type seemed almost too narrow to sustain a full-scale project. After years of working with instances of “I + copula + NOT + identifying NPs”, I can say those doubts were unfounded: negative self-identification opens enough theoretical, methodological, and social questions for several projects of this scale. One persistent challenge was explaining a context-dependent linguistic struc‐ ture in relation to both its immediate and wider textual and non-textual contexts, while also grappling with the question of how language relates to identity - and whether identity is discursively constructed or pre-exists language use. As noted in Chapter 4, approaches in linguistics, discourse analysis, and social sciences answer this differently. For my purposes, deciding how much co-text and context to include, and on what basis, was crucial. The analyses here draw only on a small proportion of possible co-texts and contextual aspects relevant to NSI functions. Even if not every angle could be explored, these questions - however complex and open-ended - should not be ignored. Because NSIs are context-dependent and can point in several directions at once, it is impossible to determine precisely which aspects of co-text and context they interact with, except perhaps in case-by-case analysis. In a corpus-prag‐ matic approach, an exhaustive account of all functions is not feasible. Instead, I categorized NSIs and their co-texts, then cross-tabulated them. While this offers a functional profile of conceptualizations that co-occur, it does not claim to provide an exhaustive pragmatic analysis of all 936 cases. Categories are inherently fuzzy, and assigning naturally occurring language to them involves subjective decisions - especially in a single-researcher project. I sought to reduce bias by creating detailed categorization frameworks with explicit criteria and by discussing these at length, even if this left less space to explore other aspects of the corpus. This study does not replicate any single previous project, nor does it sit wholly within one theoretical or methodological tradition. Instead, it follows a problem-oriented approach - “sitting down with a piece of data” in the sense of Sacks’ CA (1984a: 27) - and finding systematic ways to analyze it. This pluralism risks sacrificing some of the internal coherence of more established frameworks. For example, SFL could have been applied more comprehensively; CDS methods usually prescribe step-by-step analytical categories, which I have 336 10 Discussion and Conclusion: I Am What I’m Not <?page no="337"?> not consistently followed; corpus linguists might question the small, manually annotated corpus and limited statistical analysis; and CMC scholars might find the characterization of forum interaction less comprehensive than they expect. At the same time, this pluralism opens up new ways of combining insights across traditions, offering a flexible, problem-oriented perspective that more rigid frameworks might not have allowed. The corpus, now six years old, reflects a medium - online discussion forums - that many consider to be in decline. If starting today, I might choose a more current platform or an existing CMC corpus for reasons of convenience and contemporaneity. Yet the self-compiled, self-annotated nature of the corpus is also a strength: it allowed me to address, in a concrete way, the question of what data and how much of it is needed to study a context-dependent structure. The annotation of co-text, too, addresses a larger debate in corpus pragmatics: the tension between corpus linguistics’ reliance on form and pragmatics’ emphasis on function. I aimed for systematicity by drawing on formal aspects, while also considering larger textual units and detailed individual examples. A further methodological question was how to establish consistent, trans‐ parent categories for clauses and larger discourse units. The SFL transitivity framework offered a high level of systematicity but sometimes foregrounded fine-grained formal distinctions at the expense of holistic functional interpre‐ tation. To balance this, I also examined larger functional units and discourse patterns to situate clause-level analyses within the broader organization of the texts. Many potentially fruitful lines of inquiry remain. For instance, NSIs often appear turn-initially, responding to previous speakers; future work could explore whether they take up identifying categories from prior turns, and with what patterns. Similarly, my detailed annotation scheme (Chapter 6) was more extensive than necessary for the current research questions and could support other analyses. One promising avenue is examining how speakers present themselves as knowledgeable immediately after disclaiming expertise, particularly in technical co-texts. Others could explore NSIs within specific forums - for example, the many postings in this corpus drawn from health contexts. The socio-political implications of these findings have been kept tentative, but the study raises further-reaching questions. The use of NSIs may reflect tensions between formal authority and consumer authority, and between traditional politics and populist or sub-political movements in which online interaction plays a central role. Understanding how individuals represent themselves in everyday language, and how this connects to larger societal trends, offers a 10.5 Looking forward: Reflections and research horizons 337 <?page no="338"?> valuable complement to studies focusing on explicitly political discourse. It may not always be enough to look to politics to understand societal change; the mundane, banal exchanges of daily life can also be revealing. Fittingly, I will end with a disclaimer - consciously echoing the form at the heart of this book: I am not claiming to have said the last word on negative self-identification; in that sense, I wouldn’t call myself an expert. But as the title says: I am what I’m not. This, ultimately, is the paradoxical value of negative self-identification: what people deny themselves to be can be as socially and discursively meaningful as what they claim to be. 338 10 Discussion and Conclusion: I Am What I’m Not <?page no="339"?> 11 References Adolphs, S. (2006). Introducing electronic text analysis: A practical guide for language and literary studies (1st ed.). Routledge. https: / / doi.org/ 10.4324/ 9780203087701 Agar, M., Glaser, B., Strauss, A., Hammersley, M., Kirk, J., Miller, M., Miles, M., Huberman, M. A., Seale, C., & Silverman, D. (2004). Quality and credibility. In C. Seale, G. Gobo, J. F. Gubrium, & D. Silverman (Eds.), Qualitative research practice (pp. 378-392). SAGE. https: / / doi.org/ 10.4135/ 9781848608191 Ahuja, M. K., & Galvin, J. E. (2003). Socialization in virtual groups. Journal of Management, 29(2), 161-185. https: / / doi.org/ 10.1177/ 014920630302900203 Aijmer, K. (2013). Understanding pragmatic markers: A variational pragmatic approach. Edinburgh University Press. https: / / doi.org/ 10.1017/ S1360674315000350 Aijmer, K. (2018). Corpus pragmatics: From form to function. In A. H. Jucker, K. P. Schneider, & W. Bublitz (Eds.), Methods in pragmatics (Handbooks of Pragmatics, Vol. 10, pp. 555-585). De Gruyter. https: / / doi-org. / 10.1515/ 9783110424928-022 Aijmer, K., & Rühlemann, C. (Eds.). (2014). Corpus pragmatics: A handbook. Cambridge University Press. https: / / doi.org/ 10.1017/ CBO9781139057493 Akman, V., & Bazzanella, C. (2003). The complexity of context: Guest editors’ introduction. Journal of Pragmatics, 35(3), 321-329. https: / / doi.org/ 10.1016/ S0378-2166(02)00138-8 All European Academies (ALLEA). (2019). Democracy in a digital society: Trust, evidence and public discourse in a changing media environment. https: / / allea.org/ democracy-in -a-digital-society-trust-evidence-and-public-discourse-in-a-changing-media-environ ment/ Allwood, J. (2003). Meaning potentials and context: Some consequences for the analysis of variation in meaning. Cognitive Approaches to Lexical Semantics, 10, 1-37. https: / / doi.org/ 10.1515/ 9783110219074.29 Althusser, L. (1971). Ideology and ideological state apparatuses. In Lenin and philosophy and other essays (pp. 2-33). New Left Books. Anderwald, L. (2002). Negation in non-standard British English: Gaps, regularizations and asymmetries. Routledge. Andresen, M. A. (2009). Asynchronous discussion forums: Success factors, outcomes, assessments, and limitations. Educational Technology & Society, 12(1), 249-257. Androutsopoulos, J. (2006). Introduction: Sociolinguistics and computer-mediated com‐ munication. Journal of Sociolinguistics, 10(4), 419-438. https: / / doi.org/ 10.1111/ j.1467-9 841.2006.00286.x Androutsopoulos, J. (2008). Potentials and limitations of discourse-centred online eth‐ nography. Language@Internet, 5, Article 8. https: / / scholarworks.iu.edu/ journals/ inde x.php/ li/ article/ view/ 37571 <?page no="340"?> Androutsopoulos, J. (2014). Languaging when contexts collapse: Audience design in social networking. Discourse, Context & Media, 4-5, 62-73. https: / / doi.org/ 10.1016/ j.d cm.2014.08.006 Androutsopoulos, J., & Juffermans, K. (2014). Digital language practices in superdiversity: Introduction. Discourse, Context & Media, 4-5, 1-6. https: / / doi.org/ 10.1016/ j.dcm.201 4.08.002 Androutsopoulos, J., & Busch, F. (2021). Digital punctuation as an interactional resource: The message-final period among German adolescents. Linguistics and Education, 62, 100871. https: / / doi.org/ 10.1016/ j.linged.2020.100871 Aranda, A. M., Sele, K., Etchanchu, H., Guyt, J. Y., & Vaara, E. (2021). From big data to rich theory: Integrating critical discourse analysis with structural topic modeling. European Management Review, 18(3), 197-214. https: / / doi.org/ 10.1111/ emre.12452 Archer, D. (2012). Corpus annotation: A welcome addition or an interpretation too far? Studies in Variation, Contacts and Change in English, 10. https: / / varieng.helsinki.fi/ ser ies/ volumes/ 10/ archer/ Archer, D., & Culpeper, J. (2018). Corpus annotation. In A. H. Jucker, K. P. Schneider, & W. Bublitz (Eds.), Methods in pragmatics (Handbooks of Pragmatics, Vol. 10, pp. 495-525). De Gruyter. https: / / doi-org/ 10.1515/ 9783110424928-020 Armstrong, N., Koteyko, N., & Powell, J. (2012). “Oh dear, should I really be saying that on here? ”: Issues of identity and authority in an online diabetes community. Health, 16(4), 347-365. https: / / doi.org/ 10.1177/ 1363459311425514 Arnaut, K. (2012). Super-diversity: Elements of an emerging perspective. Diversities, 14(2), 1-16. Atkinson, M., DeWitt, D., & Joseph, E. (2017). Conspiracy theories in the 2016 election: Costly signals or conventional wisdom. In J. C. Lucas, C. J. Galdieri, & T. S. Sisco (Eds.), Conventional wisdom, parties, and broken barriers in the 2016 election (pp. 163-179). Lexington Books. Atlas, J. D. (1974). Presupposition, ambiguity and generality: A coda to the Russell-Straw‐ son debate on referring [Unpublished manuscript]. Au, L., & Eyal, G. (2022). Whose advice is credible? Claiming lay expertise in a COVID-19 online community. Qualitative Sociology, 45(1), 31-61. https: / / doi.org/ 10.1007/ s11133 -021-09492-1 Auer, P., & Roberts, C. (2011). Introduction: John Gumperz and the indexicality of language. Text & Talk, 31(4), 381-393. https: / / doi-org./ 10.1515/ text.2011.018 Auer, P., Heller, M., & Roberts, C. (2014). John J. Gumperz. In J.-O. Östman & J. Verschueren (Eds.), Handbook of pragmatics online. John Benjamins. https: / / doi.org/ 1 0.1075/ hop.18.gum1 Auinger, A., & Fischer, M. (2008). Mining consumers’ opinions on the web [Conference paper]. In Proceedings of FH Science Day 2008 (pp. 410-419). Linz, Austria. Retrieved 340 11 References <?page no="341"?> from https: / / www.researchgate.net/ publication/ 259502334_Mining_consumers%27_ opinions_on_the_web Austin, J. L. (1975). How to do things with words (2nd ed.). Oxford University Press. (Original work published 1962) Bach, K., & Harnish, R. M. (1979). Linguistic communication and speech acts. MIT Press. Bahri, L., Carminati, B., & Ferrari, E. (2018). Knowledge-based approaches for identity management in online social networks. WIREs Data Mining and Knowledge Discovery, 8(5), e1262. https: / / doi.org/ 10.1002/ widm.1260 Baker, P. (2006). Using corpora in discourse analysis. Continuum. Baker, P. (2013). Introduction: Virtual special issue of Gender and Language on corpus approaches. Gender and Language, 1. https: / / doi.org/ 10.1558/ 8psxqda5wh3d Baker, P. (2014). Using corpora to analyse gender. Bloomsbury Academic. Baker, P., & Levon, E. (2015). Picking the right cherries? A comparison of corpus-based and qualitative analyses of news articles of masculinity. Discourse & Communication, 9(2), 221-236. https: / / doi.org/ 10.1177/ 1750481314568542 Baker, P., Gabrielatos, C., KhosraviNik, M., Krzyżanowski, M., McEnery, T., & Wodak, R. (2008). A useful methodological synergy? Combining critical discourse analysis and corpus linguistics to examine discourses of refugees and asylum seekers in the UK press. Discourse & Society, 19(3), 273-306. https: / / doi.org/ 10.1177/ 0957926508088962 Bamman, D., & Smith, N. A. (2015). Contextualized sarcasm detection on Twitter. In Proceedings of the Ninth International AAAI Conference on Web and Social Media (ICWSM 2015) (pp. 574-577). AAAI Press. https: / / ojs.aaai.org/ index.php/ ICWSM/ arti cle/ view/ 14612 Barcelona, A. (2003). Metaphor and metonymy at the crossroads: A cognitive perspective. De Gruyter Mouton. https: / / doi.org/ 10.1515/ 9783110894677 Barker, K., & Galardi, T. R. (2011). Dead by 50: Lay expertise and breast cancer screening. Social Science & Medicine, 72(8), 1351-1358. https: / / doi.org/ 10.1016/ j.socscimed.2011. 02.024 Bar-Lev, S. (2008). “We are here to give you emotional support”: Performing emotions in an online HIV/ AIDS support group. Qualitative Health Research, 18(4), 509-521. http s: / / doi.org/ 10.1177/ 1049732307311680 Barron, A., & Schneider, K. P. (2014). Discourse pragmatics: Signposting a vast field. In K. P. Schneider & A. Barron (Eds.), Pragmatics of discourse (pp. 1-34). De Gruyter Mouton. https: / / doi.org/ 10.1515/ 9783110214406.1 Bartlett, T. (2017). Introduction: Reading systemic functional linguistics. In T. Bartlett & G. O’Grady (Eds.), The Routledge handbook of systemic functional linguistics (pp. 1-8). Routledge. Barton, D., & Lee, C. (2013). Language online: Investigating digital texts and practices. Routledge. https: / / doi.org/ 10.4324/ 9780203552308 11 References 341 <?page no="342"?> Bateman, P. J., Gray, P. H., & Butler, B. S. (2011). The impact of community commitment on participation in online communities. Information Systems Research, 22(4), 841-854. https: / / doi.org/ 10.1287/ isre.1090.0265 Bauman, Z. (1987). Legislators and interpreters: On modernity, post-modernity and intel‐ lectuals. Polity Press. Bauman, Z. (2012). Liquid modernity. Polity Press. Baym, N. K. (1996). Agreements and disagreements in a computer-mediated discussion. Research on Language and Social Interaction, 29(4), 315-345. https: / / doi.org/ 10.1207/ s 15327973rlsi2904_2 Beason, L. (1991). Strategies for establishing an effective persona: An analysis of appeals to ethos in business speeches. The Journal of Business Communication, 28(4), 326-346. https: / / doi.org/ 10.1177/ 002194369102800403 Beck, U. (1992). Risk society: toward a new modernity. SAGE. Beck, U., & Beck-Gernsheim, E. (2001). Individualization: Institutionalized individualism and its social and political consequences. SAGE. Beißwenger, M., & Storrer, A. (2008). Corpora of computer-mediated communication. In A. Lüdeling & M. Kytö (Eds.), Corpus linguistics: An international handbook (Vol. 1, pp. 292-308). De Gruyter. Belcher, D. D. (2023). Digital genres: What they are, what they do, and why we need to better understand them. English for Specific Purposes, 70, 33-43. https: / / doi.org/ 10.10 16/ j.esp.2022.11.003 Bennett, P. (1999). Understanding responses to risk: Some basic findings. In P. Bennett & K. Calman (Eds.), Risk communication and public health (pp. 3-32). Oxford University Press. Benson, P. (2015). YouTube as text: Spoken interaction analysis and digital discourse. In R. H. Jones, A. Chik, & C. A. Hafner (Eds.), Discourse and digital practices: Doing discourse analysis in the digital age (pp. 81-96). Routledge. Benwell, B., & Stokoe, E. (2006). Discourse and identity. Edinburgh University Press. htt ps: / / doi.org/ 10.1515/ 9780748626533 Bhatia, V. K. (2002). Applied genre analysis: A multi-perspective mode. Ibérica, 4, 3-19. Biber, D. (1993). Representativeness in corpus design. Literary and Linguistic Computing, 8(4), 243-257. https: / / doi.org/ 10.1093/ llc/ 8.4.243 Biber, D., Conrad, S., & Cortes, V. (2003). toward a taxonomy of lexical bundles in speech and writing. In A. Wilson, P. Rayson, & T. McEnery (Eds.), Corpus linguistics by the Lune: A festschrift for Geoffrey Leech (pp. 71-92). Peter Lang. Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman grammar of spoken and written English. Pearson Education. Bigi, S. (2011). The persuasive role of ethos in doctor-patient interactions. Communica‐ tion & Medicine, 8(1), 67-76. 342 11 References <?page no="343"?> Billig, M. (1999). Conversation analysis and the claims of naivety. Discourse & Society, 10(4), 572-576. Biriyai, H., & Emmah, V. (2014). Online discussion forum: A tool for effective stu‐ dent-teacher interaction. SSRN Electronic Journal. https: / / doi.org/ 10.2139/ ssrn.25250 47 Blommaert, J. (2005). Discourse: A critical introduction. Cambridge University Press. Blommaert, J., & Rampton, B. (2011). Language and superdiversity. Diversities, 13(2), 1-21. Blommaert, J., Smits, L., & Yacoubi, N. (2018). Context and its complications. Tilburg Papers in Culture Studies, 208. https: / / research.tilburguniversity.edu/ en/ publications/ context-and-its-complications Bloor, M. (2016). The construal of terminal illness in online medical texts: Social distance and semantic space. In S. Gardner & S. Alsop (Eds.), Systemic functional linguistics in the digital age (pp. 120-133). Equinox. Blum-Kulka, S. (1982). Learning how to say what you mean in a second language: A study of speech act performance of learners of Hebrew as a second language. Applied Linguistics, 3(1), 29-59. https: / / doi.org/ 10.1093/ applin/ III.1.29 Boas, G. (1964). The romantic self: An historical sketch. Studies in Romanticism, 4(1), 1-16. Bond, O. (2011). Negation in clause linkages. University of Salford. Bou-Franch, P., & Garcés-Conejos Blitvich, P. (Eds.) (2019). Analyzing digital discourse: New insights and future directions. Palgrave Macmillan. Bourdieu, P. (1998, March 13). Eine Utopie grenzenloser Ausbeutung wird Realität: Die Sachzwänge des Neoliberalismus. Le Monde Diplomatique: Beilage zur taz. boyd, d. (2014). It’s complicated: The social lives of networked teens. Yale University Press. Boyd, M. (2018). Critical discourse analysis and the editorial 2.0: News reception and user-generated comments in discourses about (im)migration. Altre Modernità, October, 1-22. https: / / doi.org/ 10.13130/ 2035-7680/ 10748 Bradford, B., Jackson, J., & Taylor, E. (2024). Introduction: Policing the permacrisis. The Political Quarterly, 95(3), 392-398. https: / / doi.org/ 10.1111/ 1467-923X.13441 Bromme, R., & Jucks, R. (2018). Discourse and expertise: The challenge of mutual understanding between experts and laypeople. In M. F. Schober, D. N. Rapp, & M. A. Britt (Eds.), The Routledge handbook of discourse processes (2nd ed., pp. 222-246). Routledge/ Taylor & Francis Group. https: / / doi.org/ 10.4324/ 9781315687384-13 Brookes, G., & McEnery, T. (2019). The utility of topic modelling for discourse studies: A critical evaluation. Discourse Studies, 21(1), 3-21. https: / / doi.org/ 10.1177/ 14614456188 14032 Brown, P., & Levinson, S. C. (1987). Politeness: Some universals in language usage. Cambridge University Press. Brubaker, R. (2004). Ethnicity without groups. Harvard University Press. 11 References 343 <?page no="344"?> Bruckman, A. (2002). Studying the amateur artist: A perspective on disguising data col‐ lected in human subjects research on the Internet. Ethics and Information Technology, 4(3), 217-231. https: / / doi.org/ 10.1023/ A: 1021316409277 Bubenhofer, N. (2008). “Es liegt in der Natur der Sache…” Korpuslinguistische Untersu‐ chungen zu Kollokationen in Argumentationsfiguren. In C. Mellado Blanco (Ed.), Studien zur Phraseologie aus textueller Sicht (pp. 53-72). Kovač. Burnett, G. (2000). Information exchange in virtual communities: A typology. Information Research, 5(4). http: / / informationr.net/ ir/ 5-4/ paper82.html Cameron, D. (1990). Demythologizing sociolinguistics: Why language does not reflect society. In J. E. Joseph & T. J. Taylor (Eds.), Ideologies of language (pp. 79-93). Routledge. Cameron, D. (2001). Working with spoken discourse. Cambridge University Press. Canca, C. (2019). AI & Global Governance: Human rights and AI ethics: Why ethics cannot be replaced by the Universal Declaration of Human Rights. In AI & global governance. United Nations University Centre for Policy Research. https: / / cpr.unu. edu/ publications/ articles/ ai-global-governance-human-rights-and-ai-ethics-why-eth ics-cannot-be-replaced-by-the-udhr.html Canca, C. (2020). Operationalizing AI ethics principles. Communications of the ACM, 63(12), 18-21. https: / / doi.org/ 10.1145/ 3430368 Cano Macias, R., & Ruiz Vera, J. M. (2024). Dynamics of opinion polarization in a population. Mathematical Social Sciences, 128, 31-40. https: / / doi.org/ 10.1016/ j.mathso csci.2024.01.009 Cappelen, H., & Lepore, E. (2005). Insensitive semantics. Blackwell. Carr, E. S. (2010). Enactments of expertise. Annual Review of Anthropology, 39(1), 17-32. https: / / doi.org/ 10.1146/ annurev.anthro.012809.104948 Carston, R. (1988). Negation, “presupposition”, and the semantics-pragmatics distinction. Journal of Linguistics, 34(2), 309-350. http: / / www.jstor.org/ stable/ 4176477 Carston, R. (1994). Truth-conditional semantics. In J.-O. Östman & J. Verschueren (Eds.), Handbook of pragmatics online. John Benjamins. https: / / doi.org/ 10.1075/ hop.m2.tru1 Chen, K., Luo, Y., Hu, A., Zhao, J., & Zhang, L. (2021). Characteristics of misinformation spreading on social media during the COVID-19 outbreak in China: A descriptive analysis. Risk Management and Healthcare Policy, 14, 1869-1879. https: / / doi.org/ 10.21 47/ RMHP.S312327 Chernev, A., Hamilton, R., & Gal, D. (2011). Competing for consumer identity: Limits to self-expression and the perils of lifestyle branding. Journal of Marketing, 75(3), 66-82. https: / / doi.org/ 10.1509/ jmkg.75.3.66 Cherny, L. (1999). Conversation and community: Chat in a virtual world. CSLI Publications. Chomsky, N. (1986). Knowledge of language: Its nature, origin, and use. Praeger. Coffin, C., Lillis, T., & O’Halloran, K. (Eds.). (2010). Applied linguistics methods: A reader. Routledge. 344 11 References <?page no="345"?> Coleman, G. (2017). From internet farming to weapons of the geek. Current Anthropology, 58(S15), S91-S102. https: / / doi.org/ 10.1086/ 688697 Collins COBUILD Advanced English Dictionary. (n.d.). Twat. In Collins English dictionary. https: / / www.collinsdictionary.com/ Collins, L. (2019). Corpus linguistics for online communication: A guide for research. Routledge. Condoravdi, C., & Lauer, S. (2011). Performative verbs and performative acts. In I. Reich, E. Horch, & D. Pauly (Eds.), Proceedings of Sinn & Bedeutung 15 (pp. 1-15). Universaar - Saarland University Press. Conoscenti, M. (2018). NATO’s social media strategic communication in the making. Altre Modernità, October, 23-47. https: / / doi.org/ 10.13130/ 2035-7680/ 10750 Cook, V. (2002). The functions of invented sentences: A reply to Guy Cook. Applied Linguistics, 23(2), 262-269. https: / / doi.org/ 10.1093/ applin/ 23.2.262 Cotgrove, L. (2025). Novel methods of intensification in young people’s digitally mediated communication. In L. Cotgrove, L. Herzberg, & H. Lüngen (Eds.), Exploring digitally mediated communication with corpora: Methods, analyses, and corpus construction (pp. 137-162). De Gruyter. Coupland, N. (2014). Social context, style and identity in sociolinguistics. In J. Holmes & K. Hazen (Eds.), Research methods in sociolinguistics: A practical guide (pp. 290-303). Wiley-Blackwell. CQPweb. (n.d.). CQPweb simple syntax help. https: / / cqpweb.lancs.ac.uk/ doc/ cqpweb-sim ple-syntax-help.pdf Creese, A., & Blackledge, A. (Eds.). (2018). The Routledge handbook of language and superdiversity. Routledge. https: / / doi.org/ 10.4324/ 9781315696010 Crooks, V. (2006). I go on the internet; I always, you know, check to see what’s new: Chronically ill women’s use of online health information to shape and inform doctor-patient interactions in the space of care provision. ACME: An International Journal for Critical Geographies, 5(1), 50-69. Crouch, C. (2017, February 23). Balancing reason and emotion in democracy. Politics: British Academy Blog. https: / / www.thebritishacademy.ac.uk/ blog/ balancing-reason-a nd-emotion-democracy Crystal, D. (2001). Language and the Internet. Cambridge University Press. Culpeper, J. (2009). The metalanguage of impoliteness: Explorations in the Oxford English Corpus. In P. Baker (Ed.), Contemporary corpus linguistics (pp. 64-86). Continuum. Culpeper, J., & Archer, D. (2008). Requests and directness in Early Modern English trial proceedings and play-texts, 1640-1760. In A. H. Jucker & I. Taavitsainen (Eds.), Speech acts in the history of English (Pragmatics & Beyond New Series, Vol. 176, pp. 45-84). John Benjamins. 11 References 345 <?page no="346"?> Danesi, M. (2016). The semiotics of emoji: The rise of visual language in the age of the internet. Bloomsbury Academic. Danet, B., & Herring, S. C. (Eds.). (2007). The multilingual internet: Language, culture, and communication online. Oxford University Press. Darics, E., & Koller, V. (2018). Language in business, language at work. Palgrave Macmillan. https: / / doi.org/ 10.1057/ 978-1-137-60117-6 Davies, B., & Harré, R. (1990). Positioning: The discursive production of selves. Journal for the Theory of Social Behaviour, 20(1), 43-63. https: / / doi.org/ 10.1111/ j.1468-5914.19 90.tb00174.x Dawson, M. (2013). Late modernity, individualization and socialism: An associational critique of neoliberalism. Palgrave Macmillan. https: / / doi.org/ 10.1057/ 9781137003423 De Beaugrande, R. (2008). How ‘systemic’ is a large corpus of English? In A. Gerbig & O. Mason (Eds.), Language, people, numbers: Corpus linguistics and society (pp. 43-60). Rodopi. De Kok, B. C. (2008). The role of context in conversation analysis: Reviving an interest in ethno-methods. Journal of Pragmatics, 40(5), 886-903. https: / / doi.org/ 10.1016/ j.pra gma.2007.06.006 De Lange, M., Raessens, J., Frissen, V., Lammes, S., & de Mul, J. (Eds.). (2015). Playful identities: The ludification of digital media cultures. Amsterdam University Press. http s: / / doi.org/ 10.5117/ 9789089647723 De Oliveira, S. M. (2013). Address in computer-mediated communication. In S. C. Herring, D. Stein, & T. Virtanen (Eds.), Pragmatics of computer-mediated communication (pp. 291-314). De Gruyter Mouton. Dear, P. (2004). Mysteries of state, mysteries of nature: Authority, knowledge and expertise in the seventeenth century. In S. Jasanoff (Ed.), States of knowledge: The co-production of science and the social order (pp. 206-224). De Gruyter. Dee, J. (2017, April 21). Internet forums are, generally speaking, long dead; don’t do them. Medium. https: / / medium.com/ web-design-web-developer-magazine/ internet-forums -are-generally-speaking-long-dead-dont-do-them-85031404fd6d Delahunty, J. (2012). ‘Who am I? ’: Exploring identity in online discussion forums. International Journal of Educational Research, 53, 407-420. https: / / doi.org/ 10.1016/ j.ij er.2012.05.004 Demata, M., Heaney, D. B., & Herring, S. C. (Eds.). (2018). Language and discourse of social media: New challenges, new approaches. Special issue of Altre Modernità, 1-168. Denzin, N. K. (2017). The research act: A theoretical introduction to sociological methods. Routledge. (Original work published 1970) Derewianka, B. (2016). Trends and issues in genre-based approaches. RELC Journal, 34(2), 133-154. https: / / doi.org/ 10.1177/ 003368820403400202 346 11 References <?page no="347"?> Deschrijver, C. (2021). On the metapragmatics of ‘conspiracy theory’: Scepticism and epistemological debates in online conspiracy comments. Journal of Pragmatics, 182, 310-321. https: / / doi.org/ 10.1016/ j.pragma.2021.02.010 Deumert, A. (2014). The performance of a ludic self on social network(ing) sites. In P. Seargeant & C. Tagg (Eds.), The language of social media: Identity and community on the internet (pp. 23-45). Palgrave Macmillan. https: / / doi.org/ 10.1057/ 9781137029317_ 2 Diabah, G. (2023). Gendered discourses and pejorative language use: An analysis of YouTube comments on We should all be feminists. Discourse, Context & Media, 51, 100667. https: / / doi.org/ 10.1016/ j.dcm.2022.100667 Dobson, A. (2015). Postfeminist digital cultures: Femininity, social media, and self-repre‐ sentation. Palgrave Macmillan. https: / / doi.org/ 10.1057/ 9781137404206 Douglas, K. M., Uscinski, J. E., Sutton, R. M., Cichocka, A., Nefes, T., Ang, C. S., & Deravi, F. (2019). Understanding conspiracy theories. Political Psychology, 40(S1), 3-35. https: / / doi.org/ 10.1111/ pops.12568 Dringus, L. P. (2010). Temporal transitions in participation flow in an asynchronous discussion forum. Computers & Education, 54(2), 340-352. https: / / doi.org/ 10.1016/ j.co mpedu.2009.08.011 Dubrofsky, R. E., & Wood, M. M. (2014). Posting racism and sexism: Authenticity, agency and self-reflexivity in social media. Communication and Critical/ Cultural Studies, 11(3), 282-287. https: / / doi.org/ 10.1080/ 14791420.2014.926245 Duranti, A. (1997). Linguistic anthropology. Cambridge University Press. Dynel, M. (2023). Lessons in linguistics with ChatGPT: Metapragmatics, metacommu‐ nication, metadiscourse and metalanguage in human-AI interactions. Language & Communication, 93, 107-124. https: / / doi.org/ 10.1016/ j.langcom.2023.09.002 Dynel, M., & Zappavigna, M. (2023). Enacting polyvocal scorn in #CovidConspiracy tweets: The orchestration of voices in humorous responses to COVID-19 conspiracy theories. Discourse, Context & Media, 52, 100670. https: / / doi.org/ 10.1016/ j.dcm.2023.1 00670 Eckert, P. (2012). Three waves of variation study: The emergence of meaning in the study of sociolinguistic variation. Annual Review of Anthropology, 41, 87-100. https: / / doi.or g/ 10.1146/ annurev-anthro-092611-145828 Edmiston, D., Hyde, E., & Adnan-Smith, T. (2025). Submerged: Surfacing deep poverty during permacrisis. Sociology. https: / / doi.org/ 10.1177/ 00380385251320564 Edmondson, W. (2014). The emergence of discourse analysis as a disciplinary field: Philosophical, pedagogic and linguistic approaches. In K. P. Schneider & A. Barron (Eds.), Pragmatics of discourse (pp. 65-96). De Gruyter Mouton. https: / / doi.org/ 10.151 5/ 9783110214406-004 11 References 347 <?page no="348"?> Ekström, M., & Stevanovic, M. (2023). Conversation analysis and power: Examining the descendants and antecedents of social action. Frontiers in Sociology, 8, 1196672. https: / / doi.org/ 10.3389/ fsoc.2023.1196672 Ekström, M., Patrona, M., & Thornborrow, J. (2018). Right-wing populism and the dy‐ namics of style: A discourse-analytic perspective on mediated political performances. Palgrave Communications, 4, Article 83. https: / / doi.org/ 10.1057/ s41599-018-0132-6 Encyclopaedia Britannica. (n.d.). Ice plant. In Britannica. Retrieved August 2025, from h ttps: / / www.britannica.com/ plant/ ice-plant Evans, S. K., Pearce, K. E., Vitak, J., & Treem, J. W. (2017). Explicating affordances: A conceptual framework for understanding affordances in communication research. Journal of Computer-Mediated Communication, 22(1), 35-52. https: / / doi.org/ 10.1111/ j cc4.12180 Fairclough, N. (1992). Discourse and social change. Polity Press. Fairclough, N. (2003). Analysing discourse: Textual analysis for social research. Routledge. Fairclough, N. (2010). Critical discourse analysis: The critical study of language (2nd ed.). Routledge. https: / / doi.org/ 10.4324/ 9781315834368 Farkas, J., & Schou, J. (2023). Post-truth, fake news and democracy: Mapping the politics of falsehood (2nd ed.). Routledge. https: / / doi.org/ 10.4324/ 9781003434870 Ferrara, K., Brunner, H., & Whittemore, G. (1991). Interactive written discourse as an emergent register. Written Communication, 8(1), 8-34. https: / / doi.org/ 10.1177/ 074108 8391008001002 Fetzer, A. (2007). Context and appropriateness: Micro meets macro. John Benjamins. https : / / doi.org/ 10.1075/ pbns.162 Fillmore, C. (1963). The position of embedding transformations in a grammar. Word, 19(2), 208-231. Foner, N., Duyvendak, J. W., & Kasinitz, P. (2017). Introduction: super-diversity in everyday life. Ethnic and Racial Studies, 42(1), 1-16. https: / / doi.org/ 10.1080/ 01419870 .2017.1406969 Forey, G., & Sampson, N. (2017). Textual metafunction and theme: What’s “it” about? In T. Bartlett & G. O’Grady (Eds.), The Routledge handbook of systemic functional linguistics (pp. 131-145). Routledge. Foucault, M. (1972). The archaeology of knowledge (A. M. Sheridan Smith, Trans.). Pantheon Books. (Original work published 1969). Fox, S. (2023). Linguistic turn. In M. Huber & M. Döll (Eds.), Bildungswissenschaft in Begriffen, Theorien und Diskursen (pp. xxx-xxx). Springer VS. https: / / doi.org/ 10.1007 / 978-3-658-37858-5_45 Freud, S. (1961). The ego and the id ( J. Riviere, Trans.). In J. Strachey (Ed. & Trans.), The standard edition of the complete psychological works of Sigmund Freud (Vol. 19, pp. 12-66). London: Hogarth Press. (Original work published 1923) 348 11 References <?page no="349"?> Gagné, C. L., & Spalding, T. (2010). Relational competition during compound interpre‐ tation. In S. Scalise & I. Vogel (Eds.), Cross-disciplinary issues in compounding (pp. 287-300). John Benjamins. https: / / doi.org/ 10.1075/ cilt.311.21gag Gajewski, J. R. (2007). Neg-raising and polarity. Linguistics and Philosophy, 30(3), 289-328. https: / / doi.org/ 10.1007/ s10988-007-9020-z García-Gómez, A. (2017). Teen girls and sexual agency: exploring the intrapersonal and intergroup dimensions of sexting. Media, Culture & Society, 39(3), 391-407. https: / / do i.org/ 10.1177/ 0163443716683789 Garzone, G. E. (2018). Sharing knowledge on LinkedIn groups: Focus on legal practition‐ ers. Altre Modernità, October, 64-86. https: / / doi.org/ 10.13130/ 2035-7680/ 10752 Gee, J. P. (2004). Situated language and learning: A critique of traditional schooling. Routledge. Geeraerts, D. (1989). Introduction: Prospects and problems of prototype theory. Linguis‐ tics, 27(4), 587-612. https: / / doi.org/ 10.1515/ ling.1989.27.4.587 Georgakopoulou, A. (2007). Small stories, interaction and identities. John Benjamins. Georgakopoulou, A. (2013). Narrative analysis and computer-mediated communication. In S. Herring, D. Stein, & T. Virtanen (Eds.), Pragmatics of computer-mediated commu‐ nication (pp. 695-716). De Gruyter. https: / / doi.org/ 10.1515/ 9783110214468.695 Georgakopoulou, A. (2014). Small stories transposition and social media: A microper‐ spective on the ‘Greek crisis.’ Discourse & Society, 25(4), 519-539. Georgakopoulou, A. (2015). Small stories research. In A. De Fina & A. Georgakopoulou (Eds.), The handbook of narrative analysis (pp. 255-271). Wiley. https: / / doi.org/ 10.100 2/ 9781118458204.ch13 Georgakopoulou, A. (2022). Co-opting small stories on social media: A narrative analysis of the directive of authenticity. Poetics Today, 43(2), 265-286. https: / / doi.org/ 10.1215/ 03335372-9642609 Georgakopoulou, A., & Spiliotti, T. (2016). Introduction. In A. Georgakopoulou & T. Spiliotti (Eds.), The Routledge handbook of language and digital communication (pp. 1-16). Routledge. Germain, T. (2024a, October 31). The ‘bias machine’: How Google tells you what you want to hear. BBC Future. https: / / www.bbc.com/ future/ article/ 20241031-how-google-tells -you-what-you-want-to-hear Germain, T. (2024b, May 24). Google just updated its algorithm. The Internet will never be the same. BBC Future. https: / / www.bbc.com/ future/ article/ 20240524-how-googles-ne w-algorithm-will-shape-your-internet Gershon, I. (2010). Media ideologies: An introduction. Journal of Linguistic Anthropology, 20(2), 283-293. https: / / doi.org/ 10.1111/ j.1548-1395.2010.01070.x 11 References 349 <?page no="350"?> Ghaffari, S. (2020). Discourses of celebrities on Instagram: Digital femininity, self-repre‐ sentation and hate speech. Critical Discourse Studies, 19(2), 161-178. https: / / doi.org/ 1 0.1080/ 17405904.2020.1839923 Ghose, A., & Ipeirotis, P. G. (2011). Estimating the helpfulness and economic impact of product reviews: Mining text and reviewer characteristics. IEEE Transactions on Knowledge and Data Engineering, 23(10), 1498-1512. https: / / doi.org/ 10.1109/ TKDE.20 10.188 Gibson, W. (2008). Intercultural communication online: Conversation analysis and the investigation of asynchronous written discourse. Forum Qualitative Sozialfor‐ schung / Forum: Qualitative Social Research, 10(1). https: / / doi.org/ 10.17169/ fqs-10.1.12 53 Giddens, A. (1991). Modernity and self-identity: Self and society in the late modern age. Polity Press. Giddens, A. (1993). Sociology (2nd ed., fully rev. and updated). Polity Press. Giesler, M., & Pohlmann, M. (2003a). The anthropology of file sharing: Consuming Napster as a gift. Advances in Consumer Research, 30, 273-279. Giesler, M., & Pohlmann, M. (2003b). The social form of Napster: Cultivating the paradox of consumer emancipation [Unpublished manuscript]. University of Konstanz. https: / / www.researchgate.net/ publication/ 259502265_The_Social_Form_of_Napster_Cultiv ating_the_Paradox_of_Consumer_Emancipation Ging, D. (2019). Alphas, betas, and incels: Theorizing the masculinities of the manosphere. Men and Masculinities, 22(4), 638-657. https: / / doi.org/ 10.1177/ 1097184X17706401 Givón, T. (1978). Negation in language: Pragmatics, function and ontology. In P. Cole (Ed.), Syntax and semantics, Vol. 9: Pragmatics (pp. 69-112). Academic Press. Givón, T. (1979). Syntax and semantics, Vol. 12: Discourse and syntax. Academic Press. Givón, T. (1993). English grammar: A function-based introduction. John Benjamins. Givón, T. (2018). On understanding grammar (Rev. ed.). John Benjamins. Glaser, B. G. (2002). Constructivist grounded theory? Forum Qualitative Sozialfor‐ schung / Forum: Qualitative Social Research, 3(3), Article 12. Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory: Strategies for qualitative research. Aldine. Glynn, D. (2007). Iconicity and the grammar-lexis interface. In O. Fischer, J. Ljungberg, & E. Tabakowska (Eds.), Insistent images (pp. 269-288). John Benjamins. Goatly, A. (1997). The language of metaphors. Routledge. https: / / doi.org/ 10.4324/ 9780203 210000 Godden, D. M., & Walton, D. (2006). Argument from expert opinion as legal evidence: Critical questions and admissibility criteria of expert testimony in the American legal system. Ratio Juris, 19(3), 261-286. https: / / doi.org/ 10.1111/ j.1467-9337.2006.00331.x 350 11 References <?page no="351"?> Goffman, E. (1974). Frame analysis: An essay on the organization of experience. Harvard University Press. Goffman, E. (1981). Forms of talk. University of Pennsylvania Press. Goldman, A., & O’Connor, C. (2019). Social epistemology. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Fall 2019 ed.). Metaphysics Research Lab, Stanford University. Retrieved August 2025, from https: / / plato.stanford.edu/ archives/ fall2019/ entries/ epistemology-social/ Gordon, C. (2011). Gumperz and interactional sociolinguistics. In R. Wodak, B. Johnstone, & P. Kerswill (Eds.), The SAGE handbook of sociolinguistics (pp. 67-84). SAGE Publi‐ cations. https: / / doi.org/ 10.4135/ 9781446200957.n6 Graham, P. W. (2000). Hypercapitalism: A political economy of informational idealism. New Media & Society, 2(2), 131-156. https: / / doi.org/ 10.1177/ 14614440022225742 Grassilli, C. (2013, April). How to choose your specialization. Translator Thoughts. Grice, H. P. (1975). Logic and conversation. In P. Cole & J. L. Morgan (Eds.), Syntax and semantics, Vol. 3: Speech acts (pp. 41-58). Academic Press. Grundmann, R. (2017). The problem of expertise in knowledge societies. Minerva, 55(1), 25-48. https: / / doi.org/ 10.1007/ s11024-016-9308-7 Gumperz, J. J. (1982). Discourse strategies. Cambridge University Press. Gumperz, J. J. (1996). The linguistic and cultural relativity of inference. In J. J. Gumperz & S. Levinson (Eds.), Rethinking linguistic relativity (pp. 374-407). Cambridge University Press. Gumperz, J. J., & Cook-Gumperz, J. (2008). Studying language, culture, and society: Sociolinguistics or linguistic anthropology? Journal of Sociolinguistics, 12(4), 532-545. Gwilliams, L., & Fontaine, L. (2015). Indeterminacy in process type classification. Func‐ tional Linguistics, 2, Article 8. https: / / doi.org/ 10.1186/ s40554-015-0021-x Hafner, C. A., & Pun, J. (2020). Editorial: Introduction to this special issue: English for academic and professional purposes in the digital era. RELC Journal, 51(1), 3-13. http s: / / doi.org/ 10.1177/ 0033688220917472 Hall, S. (2011). Introduction: Who needs ‘identity’? In S. Hall & P. du Gay (Eds.), Questions of cultural identity (pp. 1-17). SAGE Publications Ltd. https: / / doi.org/ 10.4135/ 978144 6221907.n1 (Original work published 1996) Hallberg, L. R.-M. (2006). The “core category” of grounded theory: Making constant comparisons. International Journal of Qualitative Studies on Health and Well-Being, 1(3), 141-148. https: / / doi.org/ 10.1080/ 17482620600858399 Halliday, M. A. K. (1970). Functional diversity in language as seen from a consideration of modality and mood in English. Foundations of Language, 6(3), 322-361. Halliday, M. A. K. (1973). Explorations in the functions of language. Edward Arnold. Halliday, M. A. K. (1978). Language as social semiotic: The social interpretation of language and meaning. Edward Arnold. 11 References 351 <?page no="352"?> Halliday, M. A. K. (1991). Corpus studies and probabilistic grammar. In K. Aijmer & B. Altenberg (Eds.), English corpus linguistics (pp. 30-33). Longman. Halliday, M. A. K. (2003). Introduction: On the “architecture” of human language. In J. J. Webster (Ed.), On language and linguistics (Vol. 3, Collected works of M. A. K. Halliday, pp. 1-49). Continuum. Halliday, M. A. K., & Hasan, R. (1989). Language, context, and text: Aspects of language in a social-semiotic perspective (2nd ed.). Oxford University Press. Halliday, M. A. K., & James, Z. L. (2005). A quantitative study of polarity and primary tense in the English finite clause. In G. Fox, M. Hoey, & J. M. Sinclair (Eds.), Techniques of description: Spoken and written discourse. A festschrift for Malcolm Coulthard (pp. 32-66). Routledge. Halliday, M. A. K., & Matthiessen, C. (1999). Construing experience through meaning: A language-based approach to cognition. Continuum. Halliday, M. A. K., & Matthiessen, C. (2014). Halliday’s introduction to functional grammar (4th ed., revised by C. Matthiessen). Routledge. Hardie, A. (2012). CQPweb—Combining power, flexibility and usability in a corpus analysis tool. International Journal of Corpus Linguistics, 17(3), 380-409. https: / / doi.o rg/ 10.1075/ ijcl.17.3.04har Hardt-Mautner, G. (1995). Only connect: Critical discourse analysis and corpus linguistics (UCREL Technical Papers, 6). Lancaster University. Harvey, K., & Koteyko, N. (2013). Exploring health communication: Language in action. Routledge. Hasan, R. (1995). The conception of context in text. In P. H. Fries & M. Gregory (Eds.), Discourse in society: Systemic functional perspectives. Meaning and choice in language: Studies for Michael Halliday (pp. 183-283). Ablex Publishing. Hasan, R. (2005). Semiotic mediation and three exotropic theories: Vygotsky, Halliday and Bernstein. In J. Webster (Ed.), Language, society and consciousness (Vol. 1, pp. 130-156). Equinox. Hasan, R. (2009). A view of pragmatics in a social semiotic perspective. Linguistics and the Human Sciences, 5(3), 251-279. https: / / doi.org/ 10.1558/ lhs.v5i3.251 Haugh, M. (2018). Corpus-based metapragmatics. In A. H. Jucker, K. P. Schneider, & W. Bublitz (Eds.), Methods in pragmatics (pp. 619-644). De Gruyter. https: / / doi.org/ 10.15 15/ 9783110424928-024 Hennig-Thurau, T., Gwinner, K. P., Walsh, G., & Gremler, D. D. (2004). Electronic word-of-mouth via consumer opinion platforms: What motivates consumers to articulate themselves on the Internet? Journal of Interactive Marketing, 18, 38-52. htt ps: / / doi.org/ 10.1002/ dir.10073 352 11 References <?page no="353"?> Henriksen, B. E. (2011). Language and politics. In A. Mooney, J. S. Peccei, S. LaBelle, B. E. Henriksen, E. Eppler, A. Irwin, P. Pichler, S. Preece, & S. Soden (Eds.), Language, society and power: An introduction (3rd ed., pp. 46-68). Routledge. Herring, S. C. (1998). Le style du courrier électronique: Variabilité et changement. Terminogramme, 84-85, 9-16. Herring, S. C. (2001). Computer-mediated discourse. In D. Schiffrin, D. Tannen, & H. E. Hamilton (Eds.), The handbook of discourse analysis (pp. 612-634). Blackwell. Herring, S. C. (2004a). Slouching toward the ordinary: Current trends in computer-medi‐ ated communication. New Media & Society, 6(1), 26-36. https: / / doi.org/ 10.1177/ 14614 44804039906 Herring, S. C. (2004b). Computer-mediated discourse analysis: An approach to research‐ ing online behavior. In S. A. Barab, R. Kling, & J. H. Gray (Eds.), Designing for virtual communities in the service of learning (pp. 338-376). Cambridge University Press. Herring, S. C. (2007). A faceted classification scheme for computer-mediated discourse. Language@Internet, 4(1). Herring, S. C. (2013). Relevance in computer-mediated conversation. In S. C. Herring, D. Stein, & T. Virtanen (Eds.), Pragmatics of computer-mediated communication (pp. 245-268). De Gruyter. https: / / doi.org/ 10.1515/ 9783110214468.245 Herring, S. C. (2014). Language and the Internet. In W. Donsbach (Ed.), The concise encyclopedia of communication. Wiley-Blackwell. (Pre-print version retrieved January 2, 2021, from https: / / ella.sice.indiana.edu/ ~herring/ concise.pdf Herring, S. C. (2019). The co-evolution of computer-mediated communication and computer-mediated discourse analysis. In P. Bou-Franch & P. Garcés-Conejos Blitvich (Eds.), Analyzing digital discourse: New insights and future directions (pp. 25-67). Palgrave Macmillan. Herring, S. C. (Ed.). (1996). Computer-mediated communication: Linguistic, social and cross-cultural perspectives. John Benjamins. Herring, S. C., Stein, D., & Virtanen, T. (2013). Introduction to the pragmatics of computer-mediated communication. In S. C. Herring, D. Stein, & T. Virtanen (Eds.), Pragmatics of computer-mediated communication (pp. 3-34). De Gruyter. Herring, S., & Dainas, A. (2017, January). “Nice picture comment! ” Graphicons in Facebook comment threads. In Proceedings of the Hawaii International Conference on System Sciences (HICSS). https: / / doi.org/ 10.24251/ HICSS.2017.264 Herring, S., & Kapidzic, S. (2015). Teens, gender, and self-presentation in social media. In International encyclopedia of the social & behavioral sciences (pp. 146-152). https: / / do i.org/ 10.1016/ B978-0-08-097086-8.64108-9 Heyd, T. (2014). Doing race and ethnicity in a digital community: Lexical labels and narratives of belonging in a Nigerian web forum. Discourse, Context & Media, 26(4), 38-47. https: / / doi.org/ 10.1016/ j.dcm.2013.11.002 11 References 353 <?page no="354"?> Hidalgo-Downing, L. (2000). Negation, text worlds, and discourse: The pragmatics of fiction. Ablex Publishing. Hodkinson, P. (2007). Interactive online journals and individualization. New Media & Society, 9(4), 625-650. https: / / doi.org/ 10.1177/ 1461444807076972 Horn, L. R. (1978). Remarks on neg-raising. In P. Cole (Ed.), Syntax and semantics: Vol. 9. Pragmatics (pp. 129-220). Academic Press. Horn, L. R. (1985). Metalinguistic negation and pragmatic ambiguity. Language, 61(1), 121-174. Horn, L. R. (2001). A natural history of negation (2nd ed.). University of Chicago Press. Horn, L. R., & Wansing, H. (2020). Negation. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Spring 2020 ed.). Stanford University. https: / / plato.stanford.edu/ archiv es/ spr2020/ entries/ negation/ Huang, Y. (2015). Pragmatics (2nd ed.). Oxford University Press. Huddleston, R. D., Hudson, R. A., Winter, E. O. O., & Henrici, A. (1968). Sentence and clause in scientific English (OSTI Report No. 5030). University College, London. Hunston, S. (2002). Corpora in applied linguistics. Cambridge University Press. Hunt, D. (2015). The many faces of diabetes: A critical multimodal analysis of diabetes pages on Facebook. Language & Communication, 43, 72-86. https: / / doi.org/ 10.1016/ j. langcom.2015.05.003 Hutchby, I. (2001). Conversation and technology. Polity Press. Hutchby, I., & Wooffitt, R. (1998). Conversation analysis: Principles, practices and applica‐ tions. Polity Press. Hutchison, C. (1995). The ‘Icp OnLine’: Jeux sans frontières on the CyberCampus. Journal of Computer-Mediated Communication, 1(1), Article JCMC114. https: / / doi.org/ 10.1111 / j.1083-6101.1995.tb00322.x Hymes, D. H. (1974). Foundations in sociolinguistics: An ethnographic approach. University of Pennsylvania Press. Janich, N. (2017). Genres in the business context: An introduction. In G. Mautner & F. Rainer (Eds.), Handbook of business communication: Linguistic approaches (pp. 42-61). De Gruyter. Jebahi, K. (2011). Tunisian university students’ choice of apology strategies in a discourse completion task. Journal of Pragmatics, 43, 648-662. https: / / doi.org/ 10.1016/ j.pragma .2010.09.008 Jefferson, G. (1972). Side sequences. In D. N. Sudnow (Ed.), Studies in social interaction (pp. 294-333). Free Press. Jensen, K., Lahn, L. C., & Nerland, M. (2012). Professional learning in the knowledge society. Sense Publishers. Johnson, J. J. (2021). Worldwide digital population as of October 2020 [Graph]. Statista. https: / / www.statista.com/ statistics/ 617136/ digital-population-worldwide/ 354 11 References <?page no="355"?> Jones, R. H., Chik, A., & Hafner, C. A. (Eds.). (2015). Discourse and digital practices: Doing discourse analysis in the digital age. Routledge. Jönsson, M. (2008). On compositionality. Media-Tryck. Jordan, M. P. (1998). The power of negation in English: Text, context and relevance. Journal of Pragmatics, 29, 705-752. https: / / doi.org/ 10.1016/ S0378-2166(97)00086-6 Jucker, A. H., & Taavitsainen, I. (Eds.). (2008). Speech acts in the history of English. John Benjamins. Jucker, A. H., Schreier, D., & Hundt, M. (Eds.). (2009). Corpora: Pragmatics and discourse. Papers from the 29th International Conference on English Language Research on Com‐ puterized Corpora (ICAME 29), Ascona, Switzerland, 14-18 May 2008 (Language and Computers, Vol. 68). Rodopi. Jucker, A. H., Schneider, K. P., & Bublitz, W. (2018). Methods in pragmatics. De Gruyter Mouton. https: / / doi.org/ 10.1515/ 9783110424928 Jacquemet, M. (2011). Crosstalk 2.0: Asylum and communicative breakdowns. Text & Talk, 31(4), 475-497. https: / / doi.org/ 10.1515/ text.2011.023 Kaneko, M. (2005). Marketization of higher education: Trends, issues and prospects. In Proceedings of the International Seminar on University Management and Higher Education Policies, University of Tokyo, 19-20 September 2005 (pp. 1-15). Centre for University Management and Policies. Karttunen, L. (1971). Some observations on factivity. Research on Language and Social Interaction, 4, 55-69. https: / / doi.org/ 10.1080/ 08351817109370248 Karttunen, L., & Peters, S. (1979). Conventional implicature. Syntax and Semantics, 11, 1-56. Keane, W. (2018). On semiotic ideology. Signs and Society, 6(1), 64-87. https: / / doi.org/ 10 .1086/ 695387 Kelly, M. P., & Charlton, B. (1995). The modern and the postmodern in health promotion. In R. Bunton, S. Nettleton, & R. Burrows (Eds.), The sociology of health promotion (pp. 78-90). Taylor & Francis. Kempson, R. M. (1977). Semantic theory. Cambridge University Press. Kerr, A., Cunningham-Burley, S., & Tutton, R. (2007). Shifting subject positions: Experts and lay people in public dialogue. Social Studies of Science, 37, 385-411. https: / / doi.or g/ 10.1177/ 0306312706068492 KhosraviNik, M. (2017). Right-wing populism in the West: Social media discourse and echo chambers. Insight Turkey, 19(3), 53-68. KhosraviNik, M. (2018). Social media techno-discursive design, affective communication and contemporary politics. Fudan Journal of Humanities and Social Sciences, 11, 427- 442. https: / / doi.org/ 10.1007/ s40647-018-0226-y 11 References 355 <?page no="356"?> KhosraviNik, M. (2022). Digital meaning-making across content and practice in social media critical discourse studies. Critical Discourse Studies, 19(2), 119-123. https: / / doi. org/ 10.1080/ 17405904.2020.1835683 KhosraviNik, M. (2023). Connecting the digital with the social in digital discourse: An introduction to the scene. In Social Media and Society: Integrating the digital with the social in digital discourse (pp. 1-14). John Benjamins Publishing Company. https: / / do i.org/ 10.1075/ dapsac.100.01kho KhosraviNik, M. & Esposito, E. (2018). Online hate, digital discourse and critique: Exploring digitally-mediated discursive practices of gender-based hostility. Lodz Papers in Pragmatics, 14(1), 45-68. https: / / doi-org.uaccess.univie.ac.at/ 10.1515/ lpp-20 18-0003 KhosraviNik, M., & Unger, J. W. (2016). Critical discourse studies and social media: Power, resistance and critique in changing media ecologies. In R. Wodak & M. Meyer (Eds.), Methods of critical discourse studies (3rd ed., pp. 205-233). SAGE. Kilgarriff, A., Baisa, V., Bušta, J., Jakubíček, M., Kovář, V., Michelfeit, J., Rychlý, P., & Suchomel, V. (2014). The Sketch Engine: Ten years on. Lexicography, 1, 7-36. https: / / doi.org/ 10.1007/ s40607-014-0009-9 Kim, S. J., Wang, R. J.-H., Masłowska, E., & Malthouse, E. C. (2016). “Understanding a fury in your words”: The effects of posting and viewing electronic negative word-of-mouth on purchase behaviors. Computers in Human Behavior, 54, 511-521. https: / / doi.org/ 1 0.1016/ j.chb.2015.08.015 King, B. (2009). Building and analysing corpora of computer-mediated communication. In P. Baker (Ed.), Contemporary corpus linguistics (pp. 301-320). Continuum. Knight, D. (2015). e-Language: Communication in the digital age. In P. Baker & T. McEnery (Eds.), Corpora and discourse studies: Integrating discourse and corpora (pp. 20-40). Palgrave Macmillan. Koller, V. (2005). Critical discourse analysis and social cognition: Evidence from business media discourse. Discourse & Society, 16(2), 199-224. https: / / doi.org/ 10.1177/ 0957926 505049 Koller, V. (2012). How to analyse collective identity in discourse: Textual and contextual parameters. Critical Approaches to Discourse Analysis Across Disciplines, 5(2), 19-38. Koller, V. (2013). Constructing (non-)normative identities in written lesbian discourse: A diachronic study. Discourse & Society, 24(5), 572-589. https: / / doi.org/ 10.1177/ 0957926 513486166 Koller, V. (2020). Discourse analysis and systemic functional linguistics. In C. Hart (Ed.), Researching discourse: A student guide (pp. 54-76). Routledge. Kopf, S. (2020). “Rewarding good creators”: Corporate social media discourse on mone‐ tization schemes for content creators. Social Media + Society, 6(4), 1-11. https: / / doi.or g/ 10.1177/ 2056305120969877 356 11 References <?page no="357"?> Kopf, S. (2025). Unravelling social media critical discourse studies (SM-CDS) - four approaches to studying social media through the critical lens. Critical Discourse Studies, 1-18. https: / / doi.org/ 10.1080/ 17405904.2025.2463622 Kress, G. R., & Van Leeuwen, T. (2001). Multimodal discourse: The modes and media of contemporary communication. Arnold. Kuzar, R. (2012). Sentence patterns in English and Hebrew. John Benjamins. Kwon, J. (2004). Expressing refusals in Korean and in American English. Multilingua, 23(4), 339-364. https: / / doi.org/ 10.1515/ mult.2004.23.4.339 Labov, W. (1972). Sociolinguistic patterns. University of Pennsylvania Press. Lacan, J. (2001). Écrits: A selection (A. Sheridan, Trans.). Routledge. (Original work published 1966) https: / / doi.org/ 10.4324/ 9781003059486 Lakoff, G., & Johnson, M. (1980). Metaphors we live by. University of Chicago Press. Lancaster University. (2009). Research ethics and research governance at Lancaster: A code of practice. Landert, D., Dayter, D., Messerli, T. C., & Locher, M. A. (2023). Corpus pragmatics (Elements in Pragmatics). Cambridge University Press. https: / / doi-org.uaccess.univie .ac.at/ 10.1017/ 9781009091107 Langacker, R. W. (1987). Nouns and verbs. Language, 63(1), 53-94. https: / / doi.org/ 10.230 7/ 415384 Lash, S. (2001). Individualization in a non-linear mode. In U. Beck & E. Beck-Gernsheim (Eds.), Individualization: Institutionalized individualism and its social and political consequences (pp. 9-15). SAGE. Lash, S., & Wynne, B. (1992). Introduction. In U. Beck, Risk society: toward a new modernity (pp. 1-13). SAGE. Lawton, J. (2003). Lay experiences of health and illness: Past research and future agendas. Sociology of Health & Illness, 25(S1), 23-40. https: / / doi.org/ 10.1111/ 1467-9566.00338 Leech, G. (2004). Adding linguistic annotation. In M. Wynne (Ed.), Developing linguistic corpora: A guide to good practice. Retrieved August 2025, from http: / / users.ox.ac.uk/ ~ martinw/ dlc/ chapter2.htm Leech, G. N. (1983). Principles of pragmatics. Longman. Leech, G. N. (1992). Corpora and theories of linguistic performance. In J. Svartvik (Ed.), Directions in corpus linguistics (pp. 105-122). Mouton de Gruyter. Lemke, J. L. (2000). Intertextuality and the project of text linguistics: A response to de Beaugrande. Text & Talk, 20(2), 221-226. https: / / doi.org/ 10.1515/ text.1.2000.20.2.221 Leppänen, S., Kytölä, S., Jousmäki, H., Peuronen, S., & Westinen, E. (2014). Entextualiza‐ tion and resemiotization as resources for identification in social media. In P. Seargeant & C. Tagg (Eds.), The language of social media (pp. 112-136). Palgrave Macmillan. htt ps: / / doi.org/ 10.1057/ 9781137029317_6 11 References 357 <?page no="358"?> Leppänen, S., Møller, J. S., Nørreby, T. R., Stæhr, A., & Kytölä, S. (2015). Authenticity, normativity and social media. Discourse, Context & Media, 8, 1-5. https: / / doi.org/ 10.1 016/ j.dcm.2015.05.008 Leppänen, S., Westinen, E., & Kytölä, S. (Eds.). (2017). Social media discourse, (dis)identi‐ fications and diversities. Routledge. Leppänen, S., & Sultana, S. (2023). Digital multilingualism. In A. De Fina, J. Androutsopou‐ los, M. Martin-Jones, & P. Seargeant (Eds.), The Routledge handbook of multilingualism (2nd ed., pp. 175-190). Routledge. Levinson, S. C. (1983). Pragmatics. Cambridge University Press. Lew, R. (2009). The web as corpus versus traditional corpora: Their relative utility for linguists and language learners. In P. Baker (Ed.), Contemporary corpus linguistics (pp. 289-300). Continuum. Li, E. C.-Y. (2021). Fandom. In M. Baker, B. B. Blaagaard, H. Jones, & L. Pérez-González (Eds.), The Routledge encyclopedia of citizen media (pp. 163-169). Routledge. https: / / d oi.org/ 10.4324/ 9781315619811-28 Little, M., Jordens, C. F. C., & Sayers, E.-J. (2003). Discourse communities and the discourse of experience. Health, 7(1), 73-86. https: / / doi.org/ 10.1177/ 1363459303007001619 Love, R., Dembry, C., Hardie, A., Brezina, V., & McEnery, T. (2017). The Spoken BNC2014: Designing and building a spoken corpus of everyday conversations. International Journal of Corpus Linguistics, 22(3), 319-344. https: / / doi.org/ 10.1075/ ijcl.22.3.02lov Lüders, A., Dinkelberg, A., & Quayle, M. (2022). Becoming “us” in digital spaces: How online users creatively and strategically exploit social media affordances to build up social identity. Acta Psychologica, 228, 103643. https: / / doi.org/ 10.1016/ j.actpsy.2022.10 3643 Lukin, A. (2016). Language and society, context and text: The contributions of Ruqaiya Hasan. In Society in language, language in society (pp. 143-165). Equinox. Lyons, A. (2018). Multimodal expression in written digital discourse: The case of kineticons. Journal of Pragmatics, 131, 18-29. https: / / doi.org/ 10.1016/ j.pragma.2018.0 5.001 Maaranen, A., & Tienari, J. (2020). Social media and hyper-masculine work cultures. Gender, Work & Organization, 27(6), 1127-1144. https: / / doi.org/ 10.1111/ gwao.12450 MacBride, F. (2020). Truthmakers. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Spring 2020 ed.). https: / / plato.stanford.edu/ archives/ spr2020/ entries/ trut hmakers/ Mackiewicz, J. (2010a). Assertions of expertise in online product reviews. Journal of Business and Technical Communication, 24(1), 3-28. https: / / doi.org/ 10.1177/ 10506519 09346929 358 11 References <?page no="359"?> Mackiewicz, J. (2010b). The co-construction of credibility in online product reviews. Technical Communication Quarterly, 19(4), 403-426. https: / / doi.org/ 10.1080/ 10572252 .2010.502091 Make a Meme.org (n.d.). I’m the expert, I know everything about everything! I am an expert in everything, even things I can’t pronounce… [Meme]. Retrieved August 2025, from htt ps: / / media.makeameme.org/ created/ im-the-expert-40a906390e.jpgSource: https: / / me dia.makeameme.org/ created/ im-the-expert-40a906390e.jpg Make a Meme.org (n.d.). Not an expert, still an authority figure [Meme]. Retrieved January 2020, from https: / / makeameme.org/ meme/ not-an-expert Malinowski, B. (1935). Coral gardens and their magic (Vol. 2). Allen & Unwin. Maqbool, A. (2020, July 9). Black Lives Matter: From social media post to global movement. BBC News. https: / / www.bbc.com/ news/ world-us-canada-53273381 Markham, A., & Buchanan, E. (2012). Ethical decision-making and Internet research: Recommendations from the AoIR Ethics Working Committee. Association of Internet Researchers. https: / / aoir.org/ reports/ ethics2.pdf Marko, G. (2015a). Heart-healthy or cardioprotective: A critical analysis of medicalization, health promotion and the discourse of self-help books on cardiovascular diseases (Un‐ published Habilitation thesis). University of Graz. Marko, G. (2015b). Making informed healthy lifestyle choices: Analysing aspects of patient-centred and doctor-centred healthcare in self-help books on cardiovascular diseases. In J. Romero-Trillo (Ed.), Yearbook of corpus linguistics and pragmatics 2015: Current approaches to discourse and translation studies (pp. 65-88). Springer. Marko, G. (2017). Power and status by definition: Preliminary theoretical considerations on a study of defining as a metadiscursive strategy in lay-to-lay interaction. AAA: Arbeiten aus Anglistik und Amerikanistik, 42(2-3), 243-279. Martin, J. R. (1981). How many speech acts? UEA Papers in Linguistics, 14-15, 52-77. Martínez, I. P. (2013). Non-standard negation in modern English: A corpus-based study of four salient features. ES Review: Spanish Journal of English Studies, 34, 211-226. ht tps: / / revistas.uva.es/ index.php/ esreview/ article/ view/ 2183 Maruenda-Bataller, S., Mercé, L., & Ortolà, E. C. (2024). Discourse, gender, and violence: Insights from news and social media texts. Peter Lang Verlag. https: / / doi.org/ 10.3726/ b 21783 Marwick, A. E. (2014). Gender, sexuality and social media. In J. Hunsinger & T. M. Senft (Eds.), The social media handbook (pp. 59-75). Routledge. Marwick, A. E., & boyd, d. (2011). I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience. New Media & Society, 13(1), 114-133. ht tps: / / doi.org/ 10.1177/ 1461444810365313 11 References 359 <?page no="360"?> Marwick, A. E., & boyd, d. (2014). Networked privacy: How teenagers negotiate context in social media. New Media & Society, 16(7), 1051-1067. https: / / doi.org/ 10.1177/ 14614 44814543995 Mattson, M., & Hall, J. G. (2011). Health as communication nexus: A service-learning approach. Kendall Hunt. Mautner, G. (2005a). The entrepreneurial university: A discursive profile of a higher education buzzword. Critical Discourse Studies, 2(2), 95-120. https: / / doi.org/ 10.1080/ 1 7405900500283540 Mautner, G. (2005b). Time to get wired: Using web-based corpora in critical discourse analysis. Discourse & Society, 16(6), 809-828. https: / / www.jstor.org/ stable/ 42888717 Mautner, G. (2007). Mining large corpora for social information: The case of elderly. Language in Society, 36(1), 51-72. https: / / doi.org/ 10.1017/ S0047404507070030 Mautner, G. (2009). Corpora and critical discourse analysis. In P. Baker (Ed.), Contempo‐ rary approaches to corpus linguistics (pp. 32-46). Continuum. Mautner, G. (2010). Language and the market society: Critical reflections on discourse and dominance. Routledge. Mautner, G., & Koller, V. (2004). Computer applications in critical discourse analysis. In A. Hewings, C. Coffin, & K. O’Halloran (Eds.), Applying English grammar. Arnold. McBride, D. (2011). Cancer survivors find blogging improves quality of life. ONS Connect, 26(4), 20. McCloskey, D. (2018). An examination of the boundary between work and home for knowledge workers. International Journal of Human Capital and Information Technology Professionals, 9(3), 25-40. https: / / doi.org/ 10.4018/ ijhcitp.2018070102 Mehlenbacher, A. R. (2019). Science communication online: Engaging experts and publics on the internet. Ohio State University Press. MEME. (n.d.). I’m not an expert myself, but the experts are wrong [Meme]. Retrieved January, 2020, from https: / / me.me/ i/ im-not-an-expert-myself-but-the-experts-are-wr ong-e9914076fa1647a696703062656d05e8 Mendes, K., Ringrose, J., & Keller, J. (2019). Digital feminist activism: Girls and women fight back against rape culture. Oxford University Press. https: / / doi.org/ 10.1093/ oso/ 9 780190697846.001.0001 Merchant, G. (2006). Identity, social networks and online communication. E-Learning and Digital Media, 3(2), 235-244. https: / / doi.org/ 10.2304/ elea.2006.3.2.235 Meredith, J. (2017). Analysing technological affordances of online interactions using conversation analysis. Journal of Pragmatics, 115, 42-55. https: / / doi.org/ 10.1016/ j.pra gma.2017.03.001 Meredith, J. (2019). Conversation analysis and online interaction. Research on Language and Social Interaction, 52(3), 241-256. https: / / doi.org/ 10.1080/ 08351813.2019.1631040 360 11 References <?page no="361"?> Merkley, E. (2020a). Anti-intellectualism, populism, and motivated resistance to expert consensus. Public Opinion Quarterly, 84(1), 24-48. https: / / doi.org/ 10.1093/ poq/ nfaa00 8 Merkley, E. (2020b, March 19). Many Americans deeply distrust experts. So will they ignore the warnings about coronavirus? The Washington Post. https: / / www.washin gtonpost.com/ politics/ 2020/ 03/ 19/ even-with-coronavirus-some-americans-deeply-di strust-experts-will-they-take-precautions/ Merriam-Webster. (n.d.). Come down from. In Merriam-Webster.com dictionary. Retrieved September 2025, from https: / / www.merriam-webster.com/ dictionary/ come%20down %20from Meyer, M. (2001). Between theory, method, and politics: Positioning of the approaches to CDA. In R. Wodak & M. Meyer (Eds.), Methods of critical discourse analysis (pp. 14-31). SAGE. Mittal, B. (2006). I, me, and mine - How products become consumers’ extended selves. Journal of Consumer Behaviour, 5, 550-562. https: / / doi.org/ 10.1002/ cb.202 Moffitt, B. (2016). The global rise of populism: Performance, political style, and representa‐ tion. Stanford University Press. https: / / doi.org/ 10.1515/ 9780804799331 Mondada, L. (2013). The conversation analytic approach to data collection. In J. Sidnell & T. Stivers (Eds.), The handbook of conversation analysis (pp. 32-56). Wiley-Blackwell. https: / / doi.org/ 10.1002/ 9781118325001.ch3 Montgomery, M. (2017). Post-truth politics? Authenticity, populism and the electoral discourses of Donald Trump. Journal of Language and Politics, 16(4), 619-639. https: / / doi.org/ 10.1075/ jlp.17023.mon Moreau, M.-P., & Leathwood, C. (2007). Graduates’ employment and the discourse of employability: A critical analysis. Journal of Education and Work, 19(4), 305-324. http s: / / doi.org/ 10.1080/ 13639080600867083 Morzy, M. (2013). Evolution of online forum communities. In T. Özyer, J. Rokne, G. Wagner, & A. Reuser (Eds.), The influence of technology on social network analysis and mining (Lecture Notes in Social Networks, Vol. 6, pp. 269-282). Springer. https: / / doi. org/ 10.1007/ 978-3-7091-1346-2_27 Mudde, C. (2004). The populist zeitgeist. Government and Opposition, 39(4), 541-563. htt ps: / / doi.org/ 10.1111/ j.1477-7053.2004.00135.x Mudde, C., & Rovira Kaltwasser, C. (2012). Populism in Europe and the Americas: Threat or corrective for democracy? Cambridge University Press. Muñoz, P., Bellogín, A., Barba-Rojas, R., & Diez, F. (2024). Quantifying polarization in online political discourse. EPJ Data Science, 13(39). https: / / doi.org/ 10.1140/ epjds/ s136 88-024-00480-3 Musolff, A. (2023). Permacrisis, conspiracy stories and metaphors. Półrocznik Językoz‐ nawczy Tertium, 8, 5-21. https: / / doi.org/ 10.7592/ Tertium.2023.8.1.246 11 References 361 <?page no="362"?> Nardone, C. (2018). ‘Women and work’: A cross-linguistic corpus-assisted discourse study in German and in Italian. Critical Approaches to Discourse Analysis Across Disciplines, 10(1), 167-186. https: / / ugp.rug.nl/ cadaad/ article/ view/ 41932 Nau, C., Quan-Haase, A., & McCay-Peet, L. (2022). Defining social media and asking social media research questions: How well does the Swiss army knife metaphor apply? In A. Quan-Haase & L. Sloan (Eds.), The Sage handbook of social media research methods (2nd ed., pp. 13-26). SAGE. Ng, C. W. (2018). Critical multimodal discourse analyses of news discourse on Facebook and YouTube. The Journal of AsiaTEFL, 15(4), 1174-1185. https: / / doi.org/ 10.18823/ asi atefl.2018.15.4.22.1174 O’Callaghan, C. (2020). Postpolitics and post-truth. In A. Kobayashi (Ed.), International encyclopedia of human geography (2nd ed., pp. 339-345). Elsevier. https: / / doi.org/ 10.1 016/ B978-0-08-102295-5.10683-3 Ochs, E. (1996). Linguistic resources for socializing humanity. In J. Gumperz & S. Levinson (Eds.), Rethinking linguistic relativity (pp. 407-437). Cambridge University Press. Ogiermann, E. (2009). Politeness and in-directness across cultures: A comparison of English, German, Polish and Russian requests. Journal of Politeness Research, 5(2), 189-216. https: / / doi.org/ 10.1515/ JPLR.2009.011 O’Halloran, K. (2013). A corpus-based deconstructive strategy for critically engaging with arguments. Argument & Computation, 4(2), 128-150. https: / / doi.org/ 10.1080/ 194 62166.2012.729861 OpenAI (2025, July 11). ChatGPT [Large language model]. https: / / chat.openai.com/ Page, R. (2012). Stories and social media: Identities and interaction. Routledge. https: / / doi .org/ 10.4324/ 9780203148617 Page, R. (2014). Hoaxes, hacking and humour: Analysing impersonated identity on social network sites. In P. Seargeant & C. Tagg (Eds.), The language of social media: Identity and community on the Internet (pp. 46-64). Palgrave Macmillan. https: / / doi.org/ 10.10 57/ 9781137029317_3 Page, R. (2018). Narratives online: Shared stories in social media. Cambridge University Press. https: / / doi.org/ 10.1017/ 9781316492390 Page, R. (2019). Group selfies and Snapchat: From sociality to synthetic collectivisation. Discourse, Context & Media, 28, 79-92. https: / / doi.org/ 10.1016/ j.dcm.2018.10.003 Page, R., & Thomas, B. (2011). New narratives: Stories and storytelling in the digital age. University of Nebraska Press. Page, R., Barton, D., Unger, J. W., & Zappavigna, M. (2014). Researching language and social media: A student guide. Routledge. Pannofino, N. L. (2024). The “global” deception: Flat-Earth conspiracy theory between science and religion. Genealogy, 8(2), 32. https: / / doi.org/ 10.3390/ genealogy8020032 362 11 References <?page no="363"?> Parsell, M. (2008). Pernicious virtual communities: Identity, polarisation and the Web 2.0. Ethics and Information Technology, 10(1), 41-56. https: / / doi.org/ 10.1007/ s10676-008-9 153-y Pavlova, A., & Berkers, P. (2020). Mental health discourse and social media: Which mechanisms of cultural power drive discourse on Twitter? Social Science & Medicine, 263, 113250. https: / / doi.org/ 10.1016/ j.socscimed.2020.113250 Payne, J. R. (1985). Negation. In T. Shopen (Ed.), Language typology and syntactic description (Vol. 1, pp. 197-242). Cambridge University Press. Pendry, L. F., & Salvatore, J. (2015). Individual and social benefits of online discussion forums. Computers in Human Behavior, 50, 211-220. https: / / doi.org/ 10.1016/ j.chb.201 5.03.067 Petroni, S. (2019). How social media shape identities and discourses in professional digital settings: Self-communication or self-branding? In P. Bou-Franch & P. Garcés-Conejos Blitvich (Eds.), Analyzing digital discourse: New insights and future directions (pp. 251-281). Palgrave Macmillan. https: / / doi.org/ 10.1007/ 978-3-319-92663-6_9 Phillips, P., Barnes, S., Zigan, K., & Schegg, R. (2016). Understanding the Impact of Online Reviews on Hotel Performance: An Empirical Analysis. Journal of Travel Research, 56(2), 235-249. https: / / doi.org/ 10.1177/ 0047287516636481 Pichler, H. (2010). Methods in discourse variation analysis: Reflections on the way forward. Journal of Sociolinguistics, 14(5), 581-608. https: / / doi.org/ 10.1111/ j.1467-984 1.2010.00455.x Pihlaja, S. (2017). More than 50 shades of grey: Creative content and copyright on social networking sites. Applied Linguistics Review, 8(2-3), 213-228. https: / / doi.org/ 10.1515 / applirev-2016-1036 Pípalová, R. (2008). Thematic organization of paragraphs and higher text units. Univerzita Karlova v Praze. Postill, J. (2018). The rise of nerd politics: Digital activism and political change. Pluto Press. Preece, J., Maloney-Krichmar, D., & Abras, C. (2003). History of emergence of online communities. In K. Christensen & D. Levinson (Eds.), Encyclopedia of community: From village to virtual world (pp. 1023-1027). SAGE. Radden, G., & Dirven, R. (2007). Cognitive English grammar. John Benjamins. Radin, P. (2006). “To me, it’s my life”: Medical communication, trust, and activism in cyberspace. Social Science & Medicine, 62(3), 591-601. https: / / doi.org/ 10.1016/ j.socsci med.2005.06.022 Rampton, B. (2017). Interactional sociolinguistics. In K. Tusting (Ed.), The Routledge handbook of linguistic ethnography (pp. 13-25). Routledge. https: / / doi-org/ 10.4324/ 97 81315675824 Rayson, P. (2008). From key words to key semantic domains. International Journal of Corpus Linguistics, 13(4), 519-549. https: / / doi.org/ 10.1075/ ijcl.13.4.06ray 11 References 363 <?page no="364"?> Reddit. (2025). Reddit by the numbers: As of June 30, 2025. https: / / redditinc.com/ press Reed, M. I. (1996). Expert power and control in late modernity: An empirical review and theoretical synthesis. Organization Studies, 17(4), 573-597. https: / / doi.org/ 10.1177/ 01 7084069601700402 Regulation (EU) 2016/ 679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons with regard to the processing of personal data and on the free movement of such data, and repealing Directive 95/ 46/ EC (General Data Protection Regulation) (2016). Official Journal of the European Union, L 119, 1. h ttps: / / eur-lex.europa.eu/ eli/ reg/ 2016/ 679/ oj Reisigl, M., & Wodak, R. (2016). The discourse-historical approach (DHA). In R. Wodak & M. Meyer (Eds.), Methods of critical discourse studies (3rd ed., pp. 23-61). SAGE. Richardson, K. P. (2003). Health risks on the internet: Establishing credibility online. Health, Risk & Society, 5(2), 171-184. https: / / doi.org/ 10.1080/ 1369857031000123948 Riedel, R. (2017). Populism and its democratic, non-democratic and anti-democratic potential. Polish Sociological Review, 3(199), 287-298. Riordan, M., & Kreuz, R. (2010). Emotion encoding and interpretation in computer-medi‐ ated communication: Reasons for use. Computers in Human Behavior, 26(6), 1667-1673. https: / / doi.org/ 10.1016/ j.chb.2010.06.015 Ritzer, G., & Jurgenson, N. (2010). Production, consumption, prosumption: The nature of capitalism in the age of the digital “prosumer”. Journal of Consumer Culture, 10(1), 13-36. https: / / doi.org/ 10.1177/ 1469540509354673 Roitmann, M. (Ed.). (2017). The pragmatics of negation: Negative meanings, uses and discursive functions. John Benjamins. Romero-Trillo, J. (2008). Pragmatics and corpus linguistics: A mutualistic entente (Vol. 2). Mouton de Gruyter. Rooduijn, M. (2014). The mesmerising message: The diffusion of populism in public debates in Western European media. Political Studies, 62(4), 726-744. https: / / doi.org/ 10.1111/ 1467-9248.12074 Rosa, H. (2013). Social acceleration: A new theory of modernity ( J. Trejo-Mathys, Trans.). Columbia University Press. Ross, A. S., & Rivers, D. J. (2017). Digital cultures of political participation: Internet memes and the discursive delegitimization of the 2016 U.S presidential candidates. Discourse, Context & Media, 16, 1-11. https: / / doi.org/ 10.1016/ j.dcm.2017.01.001 Roter, D. L., & Hall, J. A. (2006). Doctors talking with patients/ patients talking with doctors. Praeger Publishers. Rudolf von Rohr, M.-T., Thurnherr, F., & Locher, M. A. (2019). Linguistic expert creation in online health practices. In P. Bou-Franch & P. Garcés-Conejos Blitvich (Eds.), Analyzing digital discourse: New insights and future directions (pp. 219-250). Palgrave Macmillan. https: / / doi.org/ 10.1007/ 978-3-319-92663-6_8 364 11 References <?page no="365"?> Rueger, J., Dolfsma, W., & Aalbers, R. (2021). Perception of peer advice in online health communities: Access to lay expertise. Social Science & Medicine, 277, 113117. https: / / doi.org/ 10.1016/ j.socscimed.2020.113117 Rühlemann, C., & Aijmer, K. (2014). Introduction - Corpus pragmatics: Laying the foundations. In K. Aijmer & C. Rühlemann (Eds.), Corpus pragmatics: A handbook (pp. 1-26). Cambridge University Press. https: / / doi.org/ 10.1017/ CBO9781139057493.001 Rühlemann, C., & Clancy, B. (2018). Corpus linguistics and pragmatics. Journal of Pragmatics, 134, 1-7. https: / / doi-org/ 10.4324/ 9780429451072 Rupert, D. J., Moultrie, R. R., Read, J. G., Amoozegar, J. B., Bornkessel, A. S., O’Donoghue, A. C., & Sullivan, H. W. (2014). Perceived healthcare provider reactions to patient and caregiver use of online health communities. Patient Education and Counseling, 96(3), 320-326. https: / / doi.org/ 10.1016/ j.pec.2014.05.015 Russell, B. (1905). On denoting. Mind, 14(56), 479-493. https: / / doi.org/ 10.1093/ mind/ XIV .4.479 Sacks, H. (1972). On the analyzability of stories by children. In J. J. Gumperz & D. Hymes (Eds.), Directions in sociolinguistics: The ethnography of communication (pp. 325-345). Rinehart & Winston. Sacks, H. (1984a). Notes on methodology. In J. M. Atkinson & J. Heritage (Eds.), Structures of social action: Studies in conversation analysis (pp. 21-27). Cambridge University Press. Sacks, H. (1984b). On doing ‘being ordinary’. In J. M. Atkinson & J. Heritage (Eds.), Structures of social action: Studies in conversation analysis (pp. 413-429). Cambridge University Press. Saini, M. (2025, March 31). 50+ key Reddit statistics you must know in 2025. CropInk. ht tps: / / cropink.com/ reddit-statistics Sbisà, M. (2002). Speech acts in context. Language & Communication, 22(4), 421-436. htt ps: / / doi.org/ 10.1016/ S0271-5309(02)00018-6 Schegloff, E. A. (1988). Goffman and the analysis of conversation. In P. Drew & T. Wootton (Eds.), Erving Goffman: Exploring the interaction order (pp. 89-135). Polity Press. Schegloff, E. A. (1992). In another context. In A. Duranti & C. Goodwin (Eds.), Rethinking context (pp. 191-229). Cambridge University Press. Schmiedel, T., Müller, O., & vom Brocke, J. (2019). Topic modeling as a strategy of inquiry in organizational research: A tutorial with an application example on organizational culture. Organizational Research Methods, 22(4), 941-968. https: / / doi.org/ 10.1177/ 109 4428118773858 Schneebeli, C. (2017, November). The interplay of emoji, emoticons, and verbal modalities in CMC: A case study of YouTube comments [Conference presentation]. VINM 2017: Visualizing (in) the New Media. Neuchâtel, Switzerland. https: / / shs.hal.science/ halshs -01632753 11 References 365 <?page no="366"?> Schneider, K. P., & Barron, A. (2008). Variational pragmatics: A focus on regional varieties in pluricentric languages. John Benjamins. Scott, M. (2008). WordSmith Tools (Version 5). Lexical Analysis Software. Sealey, A. (2012). “I just couldn’t do it”: Representations of constraint in an oral history corpus. Critical Discourse Studies, 9(3), 195-210. https: / / doi.org/ 10.1080/ 17405904.201 2.688212 Sealey, A., & Carter, B. (2001). Social categories and sociolinguistics: Applying a realist approach. International Journal of the Sociology of Language, 152, 1-20. https: / / doi.or g/ 10.1515/ ijsl.2001.012 Seargeant, P., & Tagg, C. (Eds.). (2014). The language of social media: Identity and community on the Internet. Palgrave Macmillan. https: / / doi.org/ 10.1057/ 97811370293 17 Seargeant, P., Tagg, C., & Ngampramuan, W. (2012). Language choice and addressivity strategies in Thai-English social network interactions. Journal of Sociolinguistics, 16(4), 510-531. https: / / doi.org/ 10.1111/ j.1467-9841.2012.00543.x Searle, J. R. (1969). Speech acts: An essay in the philosophy of language. Cambridge University Press. Seering, J., Wang, T., Yoon, J., & Kaufman, G. (2019). Moderator engagement and community development in the age of algorithms. New Media & Society, 21(7), 1417- 1443. https: / / doi.org/ 10.1177/ 1461444818821316 Sierra, S. (2023). The epistemics of authentication and denaturalization in the construc‐ tion of identities in social interaction. Language in Society, 52(4), 571-594. https: / / doi .org/ 10.1017/ S0047404522000161 Sillence, E. (2010). Seeking out very like-minded others: Exploring trust and advice issues in an online health support group. International Journal of Web Based Communities, 6(4), 376-394. https: / / doi.org/ 10.1504/ IJWBC.2010.035741 Silverstein, M. (1992). The indeterminacy of contextualization: When is enough enough? In P. Auer & A. DiLuzio (Eds.), The contextualization of language (pp. 55-75). John Benjamins. Silverstein, M. (1993). Metapragmatic discourse and metapragmatic function. In J. A. Lucy (Ed.), Reflexive language (pp. 33-58). Cambridge University Press. Silverstein, M. (2001). The limits of awareness. In A. Duranti (Ed.), Linguistic anthropol‐ ogy: A reader (pp. 382-401). Blackwell. Sims, M. (2016). Like my status: Identity construction through online community building. Harvard Political Review, October 2. Sinclair, J. (2004). Trust the text: Language, corpus and discourse. Routledge. Sinclair, J. M., & Coulthard, M. (1975). Toward an analysis of discourse: The English used by teachers and pupils. Oxford University Press. 366 11 References <?page no="367"?> Sindoni, M. G. (2019). ‘Of course I’m married! ’ Communicative strategies and transcrip‐ tion-related issues in video-mediated interactions. In P. Bou-Franch & P. Garcés-Con‐ ejos Blitvich (Eds.), Analyzing digital discourse: New insights and future directions (pp. 71-103). Palgrave Macmillan. Skovholt, K., Grønning, A., & Kankaanranta, A. (2014). The communicative functions of emoticons in workplace e-mails: : -). Journal of Computer-Mediated Communication, 19(4), 780-797. Sosnowy, C. (2014). Practicing patienthood online: Social media, chronic illness, and lay expertise. Societies, 4(2), 316-329. Sperber, D., & Wilson, D. (1986). Relevance: Communication and cognition. Blackwell. Spilioti, T. (2012). Beyond genre: Closings and relational work in text messaging. In C. Thurlow & K. Mroczek (Eds.), Digital discourse: Language in the new media (pp. 67-93). Oxford University Press. Spitzmüller, J. (2015). Graphic variation and graphic ideologies: A metapragmatic ap‐ proach. Social Semiotics, 25(2), 126-141. https: / / doi.org/ 10.1080/ 10350330.2015.101032 3 Spitzmüller, J., & Warnke, I. (2011). Diskurslinguistik: Eine Einführung in Theorien und Methoden der transtextuellen Sprachanalyse. De Gruyter. Statista. (2025). Number of social network users worldwide from 2019 to 2025 (in billions) [Graph]. In Statista. Retrieved August 14, 2025, from https: / / www.statista.com/ statis tics/ 278414/ number-of-worldwide-social-network-users/ Steele, C. K. (2017). Black bloggers and their varied publics: The everyday politics of Black discourse online. Television & New Media, 19(2), 112-127. https: / / doi.org/ 10.117 7/ 1527476417709535 Stehr, N. (1992). Knowledge and expertise. In N. Stehr & R. V. Ericson (Eds.), The power of knowledge: Inquiries into contemporary societies (pp. 108-111). De Gruyter. Stehr, N., & Grundmann, R. (2017). Experts: The knowledge and power of expertise. Routledge. Sterling, B. (2017, November). Dead media beat: CompuServe forums. Wired. https: / / ww w.wired.com/ beyond-the-beyond/ 2017/ 11/ dead-media-beat-compuserve-forums/ Stubbs, M. (1983). Discourse analysis: The sociolinguistic analysis of natural language. Blackwell. Stubbs, M. (2007). On texts, corpora and models of language. In M. Hoey, M. Mahlberg, M. Stubbs, W. Teubert, & J. Sinclair, Text, discourse and corpora: Theory and analysis (pp. 127-161). Continuum. Suiter, J. (2016). Post-truth politics. Political Insight, 7(3), 25-27. https: / / doi.org/ 10.1177/ 2041905816680419 Swales, J. M. (1990). Genre analysis: English in academic and research settings. Cambridge University Press. 11 References 367 <?page no="368"?> Sykora, M., Elayan, S., & Jackson, T. W. (2020). A qualitative analysis of sarcasm, irony and related #hashtags on Twitter. Big Data & Society, 7(2). https: / / doi.org/ 10.1177/ 205 3951720972735 Taeuscher, K., Bouncken, R. B., & Pesch, R. (2021). Gaining legitimacy by being different: Optimal distinctiveness in crowdfunding platforms. Academy of Management Journal, 64(1), 149-179. https: / / doi.org/ 10.5465/ amj.2018.0620 Tagg, C. (2012). Discourse of text messaging: Analysis of SMS communication. Continuum. Tagg, C., & Lyons, A. (2018). Mobile messaging by migrant micro-entrepreneurs in contexts of superdiversity. In A. Creese & A. Blackledge (Eds.), The Routledge handbook of language and superdiversity (pp. 312-328). Routledge. Tagg, C., Seargeant, P., & Brown, A. A. (2017). Taking offence on social media: Conviviality and communication on Facebook. Palgrave Macmillan. Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behaviour. In S. Worchel & W. G. Austin (Eds.), Psychology of intergroup relations (pp. 7-24). Nelson-Hall. Talmy, L. (n.d.). Introspection as a methodology in linguistics. https: / / www.acsu.buffalo.e du/ ~talmy/ talmyweb/ Handouts/ introspection2.pdf Tanner, S., & Gillardin, F. (2025). Toxic communication on TikTok: Sigma masculinities and gendered disinformation. Social Media + Society, 11(1). https: / / doi.org/ 10.1177/ 20 563051251313844 Taylor, A. E. (1971). Plato: The sophist and the statesman (R. Klibansky & E. Anscombe, Eds.). Dawsons. (Original work published 1926) Taylor, D., & Bury, M. (2007). Chronic illness, expert patients and care transition. Sociology of Health & Illness, 29(1), 27-45. https: / / doi.org/ 10.1111/ j.1467-9566.2007.00516.x Taylor, J. R. (1995). Linguistic categorization: Prototypes in linguistic theory (2nd ed.). Oxford University Press. Teixeira, P. N., & Dill, D. (2011). Public vices, private virtues? Assessing the effects of marketization in higher education (Vol. 2). Sense Publishers. https: / / doi.org/ 10.1007/ 9 78-94-6091-466-9 The Internet Society. (2017). Understanding your online identity: An overview of identity. ht tps: / / www.internetsociety.org/ wp-content/ uploads/ 2017/ 11/ Understanding-your-On line-Identity-An-Overview-of-Identity.pdf Thomas, J. (1995). Meaning in interaction: An introduction to pragmatics. Longman. Thurlow, C., Lengel, L. B., & Tomic, A. (2004). Computer mediated communication: Social interaction and the Internet. SAGE. Titscher, S., Meyer, M., Wodak, R., & Vetter, E. (2000). Methods of text and discourse analysis. SAGE. 368 11 References <?page no="369"?> Tocci, J. (2009). Geek cultures: Media and identity in the digital age (Publication No. 953) [Doctoral dissertation, University of Pennsylvania]. ProQuest Dissertations Publishing. http: / / repository.upenn.edu/ edissertations/ 953 Tomlinson, M. (2017). Student perceptions of themselves as ‘consumers’ of higher education. British Journal of Sociology of Education, 38(4), 450-467. https: / / doi.org/ 10 .1080/ 01425692.2015.1113856 Tong, C. (2025). Unite or divide? Biased search queries and Google Search results in polarized politics. In Proceedings of the 17th ACM Web Science Conference 2025 (pp. 158-168). ACM. https: / / doi.org/ 10.1145/ 3717867.3717888 Ucros, M. (2018, February 7). The 10 areas of expertise in data science, and why you should choose one: Understanding your career options. Medium. https: / / medium.com/ @mel odyucros/ interested-in-data-science-heres-a-list-of-of-10-specializations-to-choose-f rom-cd342c53b673 UK Government. (2018). Data Protection Act 2018. https: / / www.legislation.gov.uk/ ukpga / 2018/ 12/ contents/ enacted Unger, J. W. (2020). Digitally mediated discourse analysis. In C. Hart (Ed.), Researching discourse: A student guide (pp. 180-200). Routledge. Unger, J. W. (2025, in press). Critical discourse studies and the rise of social media: Methodological debates and challenges. In B. Forchtner & F. Zappettini (Eds.), Edward Elgar handbook on critical discourse studies. Edward Elgar Publishing. Unger, J. W., Wodak, R., & KhosraviNik, M. (2021). Critical discourse studies and social media data. In D. Silverman (Ed.), Qualitative research (pp. 263-281). SAGE. van Dijk, T. A. (1993). Principles of critical discourse analysis. Discourse & Society, 4(2), 249-283. https: / / doi.org/ 10.1177/ 0957926593004002006 van Dijk, T. A. (1997). Cognitive context models and discourse. In M. I. Stamenow (Ed.), Language structure, discourse and the access to consciousness (pp. 189-226). John Benjamins. van Dijk, T. (2001). Discourse, Ideology and Context. Folia Linguistica, 35(1-2), 11-40. ht tps: / / doi-org/ 10.1515/ flin.2001.35.1-2.11 van Dijk, T. A. (2003). The discourse-knowledge interface. In G. Weiss & R. Wodak (Eds.), Critical discourse analysis: Theory and interdisciplinarity (pp. 85-109). Palgrave Macmillan. van Dijk, T. A. (2008). Discourse and context: A sociocognitive approach. Cambridge University Press. van Dijk, T. A. (2009). Society and discourse: How social contexts influence text and talk. Cambridge University Press. van Dijk, T. A. (2011). Discourse studies: A multidisciplinary introduction (2nd ed.). SAGE. 11 References 369 <?page no="370"?> van Dijk, T. A. (2017). Socio-cognitive discourse studies. In J. Flowerdew & J. E. Richardson (Eds.), The Routledge handbook of critical discourse studies (pp. 26-43). Routledge. Vandenberghe, F. (2015). Globalisation and individualisation in late modernity: A theo‐ retical introduction to the sociology of youth. Idéias, 5(1), 115-172. https: / / doi.org/ 10 .20396/ ideias.v5i1.8649449 Varela Pérez, J. R. (2013). Operator and negative contraction in spoken British English: A change in progress. In B. Aarts, J. Close, G. Leech, & S. Wallis (Eds.), The verb phrase in English: Investigating recent language change with corpora (pp. 256-285). Cambridge University Press. Vásquez, C. (2014). Usually not one to complain but…: Constructing identities in user-generated online reviews. In P. Seargeant & C. Tagg (Eds.), The language of social media: Identity and community on the Internet (pp. 65-90). Palgrave Macmillan. https : / / doi.org/ 10.1057/ 9781137029317_4 Vásquez, C. (2022). Research methods in linguistics. Research methods for digital discourse analysis. Bloomsbury. Vásquez, C., & Chik, A. (2021). How to study online restaurant reviews. In J. Leer & S. Krogager (Eds.), Research methods in digital food studies (pp. 127-140). Routledge. Vásquez, C., & China, A. S. (2019). From “My manly husband…” to “…sitting down to take a pee”: The construction and deconstruction of gender in Amazon reviews. In P. Bou-Franch & P. Garcés-Conejos Blitvich (Eds.), Analyzing digital discourse: New insights and future directions (pp. 193-218). Palgrave Macmillan. Vermeulen, I. E., & Seegers, D. (2009). Tried and tested: The impact of online hotel reviews on consumer consideration. Tourism Management, 30(1), 123-127. https: / / doi.org/ 10. 1016/ j.tourman.2008.04.008 Vertovec, S. (2007). Super-diversity and its implications. Ethnic and Racial Studies, 30(6), 1024-1054. https: / / doi.org/ 10.1080/ 01419870701599465 Vertovec, S. (2015). Introduction: Migration, cities, diversities ‘old’ and ‘new’. In S. Vertovec (Ed.), Diversities old and new: Migration and socio-spatial patterns in New York, Singapore and Johannesburg (pp. 1-22). Palgrave Macmillan. Villadsen, K. (2024). The subject and power - four decades later: Tracing Foucault’s evolving concept of subjectivation. Foucault Studies, 36, 293-321. https: / / doi.org/ 10.2 2439/ fs.i36.7220 Virtanen, T. (2013). Performativity in computer-mediated communication. In S. Herring, D. Stein, & T. Virtanen (Eds.), Pragmatics of computer-mediated communication (pp. 269-290). De Gruyter. Vygotsky, L. (1978). Mind in society. Harvard University Press. Wagner, D., Wenzel, M., Wagner, H.-T., & Koch, J. (2017). Sense, seize, reconfigure: Online communities as strategic assets. Journal of Business Strategy, 38(5), 27-34. 370 11 References <?page no="371"?> Wagner, P. (2012). The democratic crisis of capitalism: Reflections on political and economic modernity in Europe (LEQS Paper No. 4). London School of Economics and Political Science. http: / / www.lse.ac.uk/ european-institute/ research/ leqs/ discussion-papers Walton, D. (1997). Appeal to expert opinion: Arguments from authority. Pennsylvania State University Press. Watson, A. (2014). Who am I? The self/ subject according to psychoanalytic theory. SAGE Open, 4(3). https: / / doi.org/ 10.1177/ 2158244014545971 wedü. (n.d.). When nobody’s an expert, everybody’s an expert. WEDU. Retrieved October 20, 2019, from https: / / wedu.com/ blog/ when-everybodys-an-expert-nobodys-an-expe rt/ Wehrle, M. (2023). Can the “real world” please stand up? The struggle for normality as a claim to reality. Philosophy & Social Criticism, 49(2), 151-163. https: / / doi.org/ 10.1177/ 01914537221147852 Wessendorf, S. (2014). Commonplace diversity: Social relations in a super-diverse context. Palgrave Macmillan. Wetherell, M. (1998). Positioning and interpretative repertoires: Conversation analysis and post-structuralism in dialogue. Discourse & Society, 9(3), 431-456. https: / / doi.org / 10.1177/ 0957926598009003005 Widell, P., & Harder, P. (2019). Formal semantics and functional semantics. In K. R. Christensen, H. Jørgensen, & J. L. Wood (Eds.), The sign of the V: Papers in honour of Sten Vikner (pp. 735-757). Aarhus University. Wierzbicka, A. (2003). Cross-cultural pragmatics: The semantics of human interaction (2nd ed.). Mouton de Gruyter. Williams, G. (2014). Lay expertise. In W. Cockerham, R. Dingwall, & S. R. Quah (Eds.), The Wiley Blackwell encyclopedia of health, illness, behavior, and society (pp. 1283-1288). Wiley-Blackwell. Wirtz, V., Cribb, A., & Barber, N. (2006). Patient-doctor decision-making about treatment within the consultation: A critical analysis of models. Social Science & Medicine, 62(1), 116-124. https: / / doi.org/ 10.1016/ j.socscimed.2005.05.017 Wodak, R. (2015). The politics of fear: What right-wing populist discourses mean. SAGE. Wodak, R., & Meyer, M. (Eds.). (2016). Methods of critical discourse studies (3rd ed.). SAGE. Wooffitt, R. (2005). Conversation analysis and discourse analysis: A comparative and critical introduction. SAGE. Xiao, R., & McEnery, T. (2010). Corpus-based contrastive studies of English and Chinese. Routledge. Xiao, R., McEnery, T., & Qian, Y. (2006). Passive constructions in English and Chinese: A corpus-based contrastive study. Languages in Contrast, 6(1), 101-149. 11 References 371 <?page no="372"?> Yang, Z., Liu, L., & Ge, L. (2025). Claiming queerness on Weibo: Public interaction discourse toward Chinese queer women athletes and their chugui. Women’s Studies International Forum, 110, 103082. https: / / doi.org/ 10.1016/ j.wsif.2025.103082 Yoo, K.-H., & Gretzel, U. (2009). Comparison of deceptive and truthful travel reviews. In W. Hopken, U. Gretzel, & R. Law (Eds.), Information and communication technologies in tourism (pp. 37-47). Springer. Yu, Y., Chan, T. F., & Huang, Q. (2023). Formulating the discourse of pro-work conserva‐ tism: A critical discourse analysis of Weibo posts in response to the implementation of the three-child policy. Feminist Media Studies, 24(3), 497-513. https: / / doi.org/ 10.10 80/ 14680777.2023.2200589 Zappavigna, M. (2011). Ambient affiliation: A linguistic perspective on Twitter. New Media & Society, 13(5), 788-806. https: / / doi.org/ 10.1177/ 1461444810385097 Zappavigna, M. (2012). Discourses of Twitter and social media. Continuum. Zappavigna, M. (2018). Searchable talk: Hashtags and social media metadiscourse. Blooms‐ bury. Zeng, J., & Yan, X. (2024). Understanding climate-related visual storytelling on TikTok: A cross-national multimodal analysis. Journal of Digital Social Research, 6(2), 66-84. https: / / doi.org/ 10.33621/ jdsr.v6i2.212 Zeng, J., Chan, C. H., & Fu, K. W. (2017). How social media construct “truth” around crisis events: Weibo’s rumor management strategies after the 2015 Tianjin Blasts. Policy & Internet, 9(3), 297-320. https: / / doi.org/ 10.1002/ poi3.155 Zulli, D., & Zulli, D. J. (2020). Extending the Internet meme: Conceptualizing technolog‐ ical mimesis and imitation publics on the TikTok platform. New Media & Society, 24(8), 1872-1890. https: / / doi.org/ 10.1177/ 1461444820983603 372 11 References <?page no="373"?> List of Figures Figure 5.1: Schematic representation of turn selection for inclusion in the corpus (arrows = sequential order). . . . . . . . . . . . . . . . . 125 Figure 6.1: Concordance list of product preference disclaimers (R1 sort; R1 = first word to the right of the node). . . . . . . . . . . . . . . . . 154 Figure 6.2: Contrastive conjunctions (xcj: con) following expert disclaimers (<nsi-EX>). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Figure 7.1: Overview of conceptual categories of identifying noun phrases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162 Figure 7.2: Conceptual categories of preference disclaimers (all lexical variants). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Figure 7.5: Word cloud of key identifying noun phrases. . . . . . . . . . . . 190 Figure 7.6: Word cloud of key conceptual domains of identifying noun phrases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Figure 7.7: Conceptual landscape of negative self-identifiers. . . . . . . . 192 Figure 10.1: Expert meme 1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328 Figure 10.2: Expert meme 2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 328 Figure 10.3: Expert meme 3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 329 <?page no="375"?> List of Tables Table 3.1: Results of the pilot discourse completion task by prompt type. 69 Table 3.2: Frequencies of variants of negative self-identifiers in the Spoken BNC2014. Optional elements (adverbs, adjectives) are shown in brackets. Figures are based on a search of 11,422,617 words from 1,251 texts in the Spoken BNC2014. The final variant (“I’m a/ n…”) is included for contrast and was not sorted, as it does not express negation. . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Table 4.1: Features of online contexts (adapted from Tagg et al. 2017: 37- 38), with additional elaboration under “Identification process” to highlight the relevance of NSIs. . . . . . . . . . . . . . . . . . . . . . . . 102 Table 5.1: Overview of variants of negative self-identifiers in the corpus, with token counts per variant. Searches were restricted to .uk domains via inurl: forum and inurl: thread. . . . . . . . . . . . . . . . . . . 122 Table 6.1: Formal-functional framework of NSIs developed for this study. 133 Table 6.2: Analyses conducted to answer Research Questions 1 and 2. . . 138 Table 6.3: Semantic categories of negative self-identifiers and corresponding tags. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 Table 6.4: Overview of categories annotated in the corpus. . . . . . . . . . . . 146 Table 6.5: Concordance lines of negative self-identifiers (<nsi-…>) preceded by discourse highlighters (<hl>), with semantic cores (<ic-…>) indicated. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Table 6.6: Thematic profile of forums from which negative self-identifiers were retrieved, and number of discussions. . . . . . . . . . . . . . . . . 150 Table 6.7: Excel matrix of negative self-identifiers with additive conjunctions (<xcj: +>). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Table 7.1: Criteria for conceptual classification of nouns and noun phrases. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 Table 7.2: Overview of categorization process with examples. . . . . . . . . 161 Table 7.3: Noun phrases in the category “preference” (+ fan). . . . . . . . . . . 163 Table 7.4: Noun phrases in the category “preference” (fan). . . . . . . . . . . 167 Table 7.5: Semantic categorization of preference (fan) disclaimers. . . . . 169 Table 7.6: Noun phrases in the category “habits”. . . . . . . . . . . . . . . . . . . . . 170 Table 7.7: Noun phrases in the category “expertise” (+ expert). . . . . . . . . . 171 Table 7.8: Noun phrases in the category “expertise” (expert). . . . . . . . . . 175 Table 7.9: Profile of expertise-related head nouns and modifiers by structural and conceptual category. . . . . . . . . . . . . . . . . . . . . . . . 178 <?page no="376"?> Table 7.10: Noun phrases in the category “professions”. . . . . . . . . . . . . . . . 179 Table 7.11: Noun phrases in the category “activities”. . . . . . . . . . . . . . . . . . 181 Table 7.12: Noun phrases in the category “characteristics”. . . . . . . . . . . . . . 182 Table 7.13: Noun phrases in the category “roles”. . . . . . . . . . . . . . . . . . . . . . 184 Table 7.14: Noun phrases in the category “usage/ consumption/ ownership”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188 Table 7.15: Conceptual categories of negative self-identifiers in the Spoken BNC2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 193 Table 7.16: Conceptual categories of negative self-identifiers: Corpus vs. the Spoken BNC2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 197 Table 8.1: Internal and external co-texts of negative self-identifiers analyzed. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 Table 8.2: Criteria for a general co-textual profile, including positional, structural, and functional criteria used to classify the preceding and following co-texts of negative self-identifiers. . . . . . . . . . . 204 Table 8.3: Clause-external co-text categories and their frequencies, with formal category tags and position frequencies for left (L1) and right (R1) contexts (L1 = first token to the left; R1 = first token to the right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 Table 8.4: Preand postmodifiers of negative self-identifiers. . . . . . . . . . . 207 Table 8.5: Conceptual categories of temporally modified negative self-identifiers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209 Table 8.6: Conceptual categories of negative self-identifiers adjacent to highlighters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Table 8.7: Formal-functional relations between negative self-identifiers and adjacent co-texts, based on conjunctions and conjunctive adverbs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 Table 8.8: Process-participant configurations in the immediate co-texts of negative self-identifiers, identified within the transitivity framework (Halliday & Matthiessen, 2014). The table lists Role-1 participants (Actor, Senser, Sayer, Carrier/ Token, Behaver, Existent), their associated processes, Role-2 participants where applicable, and notes on metaphorical realization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 Table 8.9: Transitivity analysis of contrastive and concessive co-texts of negative self-identifiers, listing referents, process types, verbs, and their frequencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 Table 8.10: Role-1 participant types in contrastive and concessive co-texts of negative self-identifiers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 232 376 List of Tables <?page no="377"?> Table 8.11: Role-1 configurations in contrastive/ concessive co-texts of negative self-identifiers, with frequencies in preceding (Pre.) and following (Fol.) positions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 233 Table 8.12: Mental process types in contrasting co-texts of negative self-identifiers (I as senser). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 Table 8.13: Categories of negative self-identifiers contrasted with mental processes (I as senser). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234 Table 8.14: Relational processes in contrastive co-texts of negative self-identifiers, subdivided into intensive, possessive, and circumstantial types. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 236 Table 8.15: Values and attributes in relational processes with third-person non-human Role-1 participants. . . . . . . . . . . . . . . . . . . . . . . . . . 236 Table 8.16: Transitivity analysis of causeand consequence-related co-texts of negative self-identifiers, listing referents, process types, verbs, and their frequencies. . . . . . . . . . . . . . . . . . . . . . . 240 Table 8.17: Role-1 participant and process types in causally related co-texts of negative self-identifiers, with frequencies in preceding (Pre.) and following (Fol.) positions. . . . . . . . . . . . . . . . . . . . . . . . . . . . 246 Table 8.18: Role-1 participant types in causal co-texts of NSIs. . . . . . . . . . . 247 Table 8.19: Categories of negative self-identifiers causally related to mental processes (I as senser). . . . . . . . . . . . . . . . . . . . . . . . . . . . 248 Table 8.20: Transitivity analysis of additively related co-texts of negative self-identifiers, listing referents, process types, verbs, and their frequencies. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 Table 8.21: Role-1 participant and process types in additively related co-texts of negative self-identifiers, with frequencies in preceding (Pre.) and following (Fol.) positions. . . . . . . . . . . . . . 253 Table 8.22: Categories of negative self-identifiers additively related to mental processes (I as senser). . . . . . . . . . . . . . . . . . . . . . . . . . . . 254 Table 8.23: Process types in co-texts with I as Role-1 participant, formally links to negative self-identifiers by different conjunction types (n = 443). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255 Table 8.24: Categories of negative self-identifiers in mental process co-texts (I as … senser). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 Table 9.1: Framework for functionally categorizing mental process co-texts (I as senser). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 266 Table 9.2: Functional categories of mental process co-texts (n = 192). Totals reflect clauses, not tokens of NSIs; a clause is counted once under its formal link (contrast/ cause/ addition). . . . . . . . . 270 List of Tables 377 <?page no="378"?> Table 9.3: Categories of negative self-identifiers in “knowledge/ understanding” co-texts (n = 31). . . . . . . . . . . . . . . . . . . . . . . . . . 271 Table 9.4: Categories of negative self-identifiers in “preferences/ habits” co-texts (n = 24). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273 Table 9.5: Categories of addressee-oriented co-texts of negative self-identifiers (n = 17). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275 Table 9.6: Framework for functionally categorizing relational process co-texts (I as carrier/ token). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279 Table 9.7: Functional profile of relational processes with different formal links to negative self-identifiers (n = 125). . . . . . . . . . . . . . . . . . 281 Table 9.8: Categories of negative self-identifiers in “knowledge” co-texts (n = 46). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 Table 9.9: Categories of negative self-identifiers in “experience” co-texts (n = 22). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 Table 9.10: Categories of negative self-identifiers in “preference/ habit” and “response” co-texts (n = 19). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 285 Table 9.11: Framework for functionally categorizing material processes (I as actor). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288 Table 9.12: Distribution of material process functions across formal links to negative self-identifiers (contrast/ concession, cause/ consequence, and addition; n = 84). . . . . . . . . . . . . . . . . . . . . . . . 290 Table 9.13: Functional categorization of “experience” co-texts (n = 22). . . 291 Table 9.14: Functional categorization of “habit/ principles” co-texts (n = 19). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292 Table 9.15: Functional categorization of “situative anchor” co-texts (n = 16). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295 Table 9.16: Functional profile of clauses with formal links to negative self-identifiers (n = 401). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 296 Table 9.17: Functional categories of negative self-identifiers, with descriptions and illustrative corpus examples. . . . . . . . . . . . . . . 299 Table 9.18: Functional profile of co-texts preceding negative self-identifiers, with definitions and illustrative corpus examples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 302 Table 9.19: Frequencies of functional categories in co-texts preceding negative self-identifiers (n = 376). . . . . . . . . . . . . . . . . . . . . . . . . 306 Table 9.20: Negative self-identifiers by functional category of preceding co-texts. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 308 378 List of Tables <?page no="379"?> www.narr.de Buchreihe zu den Arbeiten aus Anglistik und Amerikanistik ISBN 978-3-381-12821-1 Why do people in online forums say who they are not rather than who they are - especially when offline identities are unverifiable or irrelevant? And how do such linguistic moves reflect wider struggles over belonging and representation in digital publics? This book explores how negative self-identifiers - variants of the template “I’m not a…” - function as a linguistic strategy in Englishlanguage web forums, revealing how users present themselves while negotiating shared interests and epistemic positions. It starts from the premise that identity is something we do - a linguistic performance shaped by context - and takes a critical interest in how the interactional micromanagement of identity relates to its broader sociopolitical dynamics. Adopting a corpus-pragmatic approach, the study examines how negative self-identification positions speakers in relation to their claims and foregrounds identity features that matter in the moment, critically connecting micro-level linguistic practice with wider debates on discourse, identity, and the politics of digitally mediated representation. Triebl The Pragmatics of Negative Self-Identification: I Am What I’m Not Eva Triebl The Pragmatics of Negative Self-Identification: I Am What I’m Not A corpus-based study of UK web forums