Language and Characterisation in Television Series: A corpus-informed approach to the construction of social identity in the media 2022047242, 9789027212955, 9789027254665

Citation preview

Language and Characterisation in Television Series A corpus-informed approach to the construction of social identity in the media Monika Bednarek Studies in Corpus Linguistics

106 JOHN BENJAMINS PUBLISHING COMPANY

Language and Characterisation in Television Series

Studies in Corpus Linguistics (SCL) issn 1388-0373

SCL focuses on the use of corpora throughout language study, the development of a quantitative approach to linguistics, the design and use of new tools for processing language texts, and the theoretical implications of a data-rich discipline. For an overview of all books published in this series, please see benjamins.com/catalog/scl

General Editor

Founding Editor

Ute Römer

Elena Tognini-Bonelli

Georgia State University

The Tuscan Word Centre/University of Siena

Advisory Board Laurence Anthony

Susan Hunston

Antti Arppe

Michaela Mahlberg

Michael Barlow

Anna Mauranen

Monika Bednarek

Andrea Sand

Tony Berber Sardinha

Benedikt Szmrecsanyi

Douglas Biber

Elena Tognini-Bonelli

Marina Bondi

Yukio Tono

Jonathan Culpeper

Martin Warren

Sylviane Granger

Stefanie Wulff

Waseda University

University of Alberta University of Auckland University of Sydney Catholic University of São Paulo Northern Arizona University University of Modena and Reggio Emilia Lancaster University University of Louvain

University of Birmingham University of Birmingham University of Helsinki University of Trier Catholic University of Leuven The Tuscan Word Centre/University of Siena Tokyo University of Foreign Studies The Hong Kong Polytechnic University University of Florida

Stefan Th. Gries

University of California, Santa Barbara

Volume 106 Language and Characterisation in Television Series A corpus-informed approach to the construction of social identity in the media by Monika Bednarek

Language and Characterisation in Television Series A corpus-informed approach to the construction of social identity in the media

Monika Bednarek The University of Sydney

John Benjamins Publishing Company Amsterdam / Philadelphia

8

TM

The paper used in this publication meets the minimum requirements of the American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.

Cover design: Françoise Berserik Cover illustration from original painting Random Order by Lorenzo Pezzatini, Florence, 1996.

doi 10.1075/scl.106 Cataloging-in-Publication Data available from Library of Congress: lccn 2022047241 (print) / 2022047242 (e-book) isbn 978 90 272 1295 5 (Hb) isbn 978 90 272 5466 5 (e-book)

© 2023 – Monika Bednarek No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Company · https://benjamins.com

Table of contents

Acknowledgments

ix

Language notes

xi

Chapter 1 Televisual characterisation 1.1 Introduction 1 1.2 Creating characters: Processes and authorship  3 1.3 Characters and the audience  7 1.4 Features of the televisual character  9 1.4.1 Stability 9 1.4.2 Aspects of character identity  11 1.5 Frameworks for the linguistic study of televisual characterisation  12 1.5.1 Cognitive stylistics: Schemas and textual cues  13 1.5.2 Sociolinguistics: Telecinematic indexicalities  18 1.5.3 A mixed approach to televisual characterisation  22 1.6 This book’s case study approach  24 Chapter 2 Corpus linguistic analysis of televisual characterisation: Data and approach 2.1 Introduction 27 2.2 Corpus linguistic analyses of televisual characterisation  27 2.3 Corpus linguistic techniques and concepts  30 2.3.1 Frequency and keyness  30 2.3.2 Distribution  32 2.3.3 Collocation and concordance analysis  35 2.3.4 A note on statistics  37 2.4 Data 38 2.5 Units and levels of analysis  40 2.6 Concluding remarks  43

1

27

vi

Language and Characterisation in Television Series

Chapter 3 Character differentiation and character stability 3.1 Introduction 45 3.2 Gilmore Girls  45 3.3 Case study 1: Character differentiation  47 3.3.1 Lorelai as unique character  48 3.3.2 Lorelai as an emotional character  53 3.4 Case study 2: Character stability  58 3.4.1 Lorelai: An example of a “stable” televisual character?  58 3.4.2 Diachronic character stability re-visited  61 3.4.3 Intersubjective stability re-visited  64 3.4.4 Revisiting stability: The Gilmore Girls reboot  69 3.5 Conclusion 71 Chapter 4 Characters and stereotypes 4.1 Introduction 73 4.2 The Big Bang Theory  74 4.3 Nerd stereotypes/schemas  75 4.4 Methods  77 4.4.1 Concordance and key word analysis  77 4.4.2 Scene-based analysis  78 4.5 Results 80 4.5.1 Character insights from concordance analysis  80 4.5.2 Character insights from key word analysis  84 4.5.3 Character insights from scene-based analysis  90 4.5.4 Sheldon compared to other television characters  96 4.5.5 Beyond Sheldon: Nerds in The Big Bang Theory  97 4.6 Conclusion  102 Chapter 5 Flawed female characters 5.1 Introduction 105 5.2 Female gender representation in television series  106 5.3 Swear/taboo words  109 5.3.1 Swear/taboo words and televisual characterisation  109 5.3.2 Swear/taboo words and gender  113 5.4 Nurse Jackie, Weeds and Saving Grace  115 5.5 The use of swear/taboo words by the three “flawed” women  119 5.5.1 (Very) strong swear/taboo words  119 5.5.2 “Female” and “male” swear/taboo words  123 5.6 Conclusion 129

45

73

105



Table of contents vii

Chapter 6 Characterisation and the use of marginalised varieties of English 133 6.1 Introduction 133 6.2 Indigenous characters and Australian Aboriginal English  134 6.3 Marginalised and “non-standard” Englishes in telecinematic discourse  138 6.4 Methodology  140 6.4.1 Redfern Now  140 6.4.2 The RFN corpus  142 6.4.3 Approach  144 6.5 Findings 145 6.5.1 Key words and their range  146 6.5.2 Australian English/culture  147 6.5.3 Aboriginal ways of using English  149 6.5.4 Kinship terms  155 6.5.5 Additional AAE lexis  160 6.6 Conclusion  163 Chapter 7 Beyond individual series: Analyses of US and Australian television dialogue corpora 167 7.1 Introduction 167 7.2 Case study 1: The use of a stigmatised language feature in US television  168 7.2.1 Background: Ain’t as a stigmatised language feature  168 7.2.2 Ain’t and characterisation  169 7.3 Case study 2: AAE lexis in Indigenous-authored television series  175 7.3.1 Background: Indigenous representation and AAE  175 7.3.2 Using lexical profiling to analyse AAE lexis  179 7.3.3 Corpus and methodology  182 7.3.3.1 The Ngara corpus  182 7.3.3.2 Methodology  185 7.3.4 Results 189 7.3.4.1 AAE words: Frequency, range, and character diffusion  189 7.3.4.2 AAE words: Character ethnicity  191 7.3.5 Case study summary  198 7.4 Chapter conclusion  199

viii Language and Characterisation in Television Series

Chapter 8 Conclusion203 8.1 Introduction 203 8.2 Using corpus linguistic techniques to study televisual characterisation  204 8.3 Televisual characterisation and stylistics  207 8.4 Concluding remarks  209 List of TV series (and movies)

211

References221 Appendix243 Index of TV series (and movies)

261

General index

263

Acknowledgments

I would like to acknowledge the Gadigal people of the Eora Nation, the traditional custodians of the land on which this book was written and pay my respects to Elders past, and present. This book brings together for the first time multiple corpus linguistic studies of televisual characterisation that I have undertaken over a period spanning more than a decade. More precisely, the book draws on case studies of televisual characterisation originally published between 2010 and 2021. Chapter 3 is a blended and reworked version of sections from my 2010 book The Language of Fictional Television and from my chapter in the edited 2011 volume Telecinematic Discourse. Chapter 4 uses material from my 2012 article “Constructing ‘nerdiness’” (Multilingua 31/2–3) together with a small section from my chapter in the edited 2020 volume Telecinematic Stylistics, and Chapter 5 is based on work originally published in my 2015 article “‘Wicked’ women in contemporary pop culture” (Text & Talk 35/4). Chapter 6 uses material from case studies of the Australian TV series Redfern Now, namely a 2020 journal article (International Journal of Corpus Linguistics 25/4) and a journal article in the Journal of the European Association for Studies of Australia (Vol.12, No.1–2, 2021). The first case study in Chapter 7 comes from a section in my 2018 book Language and Television Series, while the second case study in that chapter has not previously been published. All case studies in Chapters 1–6 have been substantially recontextualised, updated, edited, and considerably developed and extended, incorporating new corpus linguistic analyses. I want to give special thanks to Susan Reichelt for allowing me to refer to her work on the Gilmore Girls “reboot” in Chapter 3, and for providing relevant figures. I am grateful to the various colleagues in Australia and internationally who influenced my work, assisted in my research, or commented on the publications on which this book builds. These colleagues are listed by name in the relevant prior publications. I also want to thank the reviewers who provided constructive feedback on these publications as part of the anonymous peer review process. Additional thanks go to the (anonymous) scholar who reviewed this book manuscript. Further, I am grateful to the book series editor Ute Römer and the whole team at John Benjamins for seeing this book through to publication. For contributions to corpus building, I would like to express my thanks to Mark Assad (character-based corpora in Chapter 3), David Caldwell, Ganna Veselovska, and students in Language and

x

Language and Characterisation in Television Series

Identity (corpora in Chapter 4). For assistance with corpus building for Chapters 6 and 7 I thank Kelvin Lee, Tim Bishop, Elena Sheard and especially Georgia Carr, who undertook the bulk of corpus transcription and editing. I am grateful to the University of Sydney’s sustained support and funding for my research which has enabled some of these corpora to be built. Additional thanks go to Kelvin Lee for assisting in checking and formatting references. Last but not least, I want to express my deepest thanks to my partner Helen Caple, for her continued support in all my research endeavours and beyond.

Language notes

In this book, the terms “standard” and “non-standard” are enclosed in quotation marks (scare quotes) to highlight their problematic and ideological nature. Aboriginal and Torres Strait Islander readers are advised that this book may contain the names of people who have died. In addition, Chapter 7 contains some examples of terms of abuse that may be experienced as confronting or triggering. Chapters 6 and 7 contain references to colonisation, stigmatisation, and racism. Readers should be aware that terms used in some chapters of this book such as Aboriginal, Torres Strait Islander, and Indigenous are labels introduced by the colonisers and do not capture “the diversity of languages, cultural practices, worldviews and experiences that exist across the continent now called Australia” (Roberts et al 2021: 1). I use these terms because I do not generally know the specific language groups, peoples, or nations of the television characters I refer to. I use Aboriginal and/or Torres Strait Islander people rather than peoples because I refer to individuals rather than nations. The term Indigenous is used to avoid too much repetition and typically means Aboriginal and/or Torres Strait Islander. In Chapter 6, I talk about Aboriginal characters rather than Aboriginal and/or Torres Strait Islander characters, on account of Sydney (Redfern) being the setting/location of the investigated series. In Chapter 7, I talk about Aboriginal and/or Torres Strait Islander characters, given the large number of diverse characters included in the 13 series in the corpus and the possibility that at least one of these characters could theoretically be Torres Strait Islander or both Aboriginal and Torres Strait Islander. (Actors playing the characters certainly include people who are both Aboriginal and Torres Strait Islander.) Importantly, the term Aboriginal and/or Torres Strait Islander is not meant to imply that these groups are the same, but is used in an inclusive sense given that the specific identity of characters is not always clear from the televisual narrative. There is not necessarily always full consensus around language use in this space, and often there will be a diversity of opinions on preferred terms. For further reading on terminology, readers may be interested in Carlson et al (2014); Public Health Association Australia (2017); Media Diversity Australia (2018); Roberts et al (2021); Canuto & Finlay (2021), or Latimore (2021).

xii Language and Characterisation in Television Series

Finally, I want to emphasise that when I use binary oppositions such as Indigenous and non-Indigenous, or African American and non-African American I do not aim to imply that they are internally homogenous or to essentialise people at the expense of diversity and variation. While I have aimed to adhere to relevant guidelines and advice regarding language use, it is possible that I have made errors and/or that some of the terminology will change over time in its appropriateness. If this is the case, I hope for understanding that these errors were not committed intentionally or with malice.

A note on the citing of television series My approach to citing television series (and movies) in this book is to only include in-text references for those series that are individually discussed (Little J & Big Cuz, The Warriors) or analysed (Gilmore Girls, The Big Bang Theory, Nurse Jackie, Redfern Now, Saving Grace, Weeds). Series that are only briefly mentioned or that are analysed as part of bigger or diverse corpora are instead included with all relevant information in the list of television series (and movies) provided at the end of this book.

Chapter 1

Televisual characterisation

1.1

Introduction

This is a book about televisual characterisation – more precisely, the linguistic construction of fictional characters in television series from the United States and Australia.1 Fictional characters are an important part of the televisual narrative because they are crucial for plot and narrative development (Pearson 2007; Selby & Cowdery 1995) as well as for audience engagement (Bednarek 2011a; Livingstone 1998: 119; Richardson 2010a: 63–84). As two scholars have put it: Nearly every successful television writer will point to character as the focal point of their creative process and how they measure success – if you can create compelling characters, then engaging scenarios and storylines will likely follow suit. (Mittell 2015: 118) The point of connection between ourselves and the text is, after all, primarily about the people we see on screen. We want to be like them, or we desire them or are amused or frightened by them: it is frequently character that dictates the drama, that makes things happen, things we are interested in precisely because we know and have gradually aligned ourselves with their character. The history of television fiction resonates with audiences because great characters made it so […] (Wickham 2007: 91)

We can examine how such characters are constructed from a variety of perspectives and using a range of approaches, from narratology to sociology to gender studies to linguistics, to name but a few. This book approaches televisual characterisation using the techniques of corpus linguistics and is informed by the linguistic disciplines of stylistics, sociolinguistics, and pragmatics. It revisits, expands upon, and brings together a range of studies of televisual characterisation that I have undertaken over a period of more than a decade. What do I mean by televisual characterisation? In literary studies, stylistics, narratology and media studies, the term characterisation typically refers to how 1. Engagement with research on TV series in other languages and cultures is beyond the scope of this book – some relevant research is included in Bednarek & Zago (2022), while Lee (2021) focuses on characterisation in Japanese anime. I use TV series as a cover term for both series and serials as well as hybrid forms. I also use TV narrative as a synonym.

2

Language and Characterisation in Television Series

writers imbue the “people” in their texts with certain characteristics, personalities or identities – how characters are constructed in discourse or how readers infer certain characteristics from discourse (Bednarek 2010: 98).2 Thus, televisual characterisation refers to the construction of characters in fictional television series – how characters are constructed and revealed to the televisual audience. Characters in television series are multimodal (semiotic) constructs in the service of a narrative aimed at a target audience, and therefore also exist in the minds of that audience as cognitive constructs. This approach to character integrates semiotic and cognitive approaches regarding the “ontological status” of characters (see Culpeper & Fernandez-Quintanilla 2017: 94; McIntyre 2014: 149–151). In this book, I sometimes refer to specific television characters (e.g. Lorelai in Gilmore Girls, Sheldon in The Big Bang Theory) just by their name, using statements such as “Lorelai uses X linguistic feature”. However, this is only ever a shorthand for the complex and collaborative process and authorship behind such language use (see Section 1.2). Ultimately, it is also less cumbersome to talk about the usage of linguistic features in this way, rather than consistently using formulations such as “the televisual character Lorelai is assigned the linguistic feature x” or “actor X performs the dialogue cue assigned to the televisual character Lorelai”. As mentioned, characters are multimodal constructs, i.e. multimodal cues construct or index (aspects of) character identity. This book focuses specifically on linguistic cues or indexes, i.e. on how TV dialogue functions for televisual characterisation. I define dialogue here similarly to my 2018 book: as shorthand for all character speech, whether this speech is by one, two or more speakers, and whether it consists of monologues, asides, voice-over narration, interaction, etc. Such dialogue is differentiated from screen directions, which may refer to elements such as location and time, camera angle, special effects, etc. This book does not investigate dialogue in scripts (screenplays) or subtitles but rather examines on-screen dialogue – the dialogue that the audience encounters when watching a TV episode. This dialogue has been collaboratively created, is uttered by the actors who are performing the characters, and fulfils multiple functions 2. It is beyond the scope of this book to compare televisual characterisation with characterisation in other fictional texts such as film, literary drama, novels, even though there are both similarities and differences to characterisation in television series. This is a matter for future research. In this chapter, then, I focus primarily on reviewing the linguistic literature on televisual characterisation. Overviews of characterisation in literary studies, stylistics, narratology and media studies are provided in Culpeper (2001) and – with respect to television – in Bednarek (2010, 2011a) and Richardson (2010a). Short (2014) and Lambrou (2014) present analyses of characters and character relations in novels, plays and films through turn-taking patterns, speech-acts, conversational maxims and politeness patterns. Kozloff (2000) discusses many examples where film dialogue functions to establish characters and their relationships.



Chapter 1.  Televisual characterisation

for the televisual narrative (see Bednarek 2018a). Characterisation or “character revelation” (Kozloff 2000: 43) is just one of these functions, albeit a crucial one. Characterisation communicates aspects of the narrative to viewers, including who the characters are, their histories and backgrounds, their individual and social character traits, their relationships to other characters and the role they play for the narrative (e.g. villain vs hero). Dialogue functions to “make characters substantial, … hint at their inner life” (Kozloff 2000: 43). Aspects other than dialogue also contribute to characterisation, but are not the focus of this book. This emphasis on dialogue also means that a corpus linguistic approach is very useful. Television dialogue can be compiled into corpora of different sizes and of different types, which lend themselves to diverse linguistic investigations. For instance, it is possible to compile corpora of a whole series or a whole season of a television narrative, it is possible to collect dialogue corpora for specific television characters, and it is possible to design corpora that tell us something about characters across different television series. These corpora can then in turn be interrogated using corpus techniques such as quantitative analysis of frequency, distribution and statistically salient words, or qualitative analysis of the meanings and patterns of particular linguistic resources. As further explained in Chapter 2, this book uses a range of corpora and corpus methods. It supplements quantitative results with qualitative analyses, and combines the corpus approach with relevant theories and concepts from stylistics, sociolinguistics, and pragmatics. This in turn can offer novel insights into televisual characterisation as a linguistic phenomenon. The following sections provide an overview of selected aspects of televisual characterisation. Section 1.2 examines the processes and authorship at play in the creation of characters, while Section 1.3 focusses on the relationship between characters and the audience. Section 1.4 introduces different features of the televisual character, while Section 1.5 introduces the linguistic frameworks for televisual characterisation that are most relevant to this book. Finally, Section 1.6 explains this book’s case study approach. 1.2

Creating characters: Processes and authorship

Television characters are produced collaboratively by writers, producers, actors, and other screen creatives (Mittell 2015: 119). This collaborative authorship may involve story producers working with writers and/or actors through dedicated workshops. For instance, in the case of the Australian TV series Redfern Now (Dale & Dear 2012–2013) – discussed in Chapter 6 – British story producer Jimmy McGovern worked closely with Indigenous screenwriters (Nelson 2013). This collaborative process was explained in a news item (Elliot 2012) as follows:

3

4

Language and Characterisation in Television Series

Starting in December 2010, McGovern held a series of workshops with the indigenous [sic] writers, honing their stories, mining their minds, searching for what McGovern calls “emotional truth”. “We drove the writers into the ground,” he says. “Two four-hour sessions every day, just talking story; burrowing down, asking, ‘is this exciting? Is this convincing’? And because part of the exercise was to bring on Aboriginal writers, we had to make sure that they wrote every word.”

Workshop participants for a particular series are sometimes (but not always) listed in the end credits and can then be identified by the researcher. For instance, the credits for the Australian football drama The Warriors (Briggs & Connolly 2017) lists individual “workshop participants” under “Early Development”, namely six Indigenous writers, screen creatives, and footballers. In other cases, there may be “bootcamps” dedicated to intensive writing sessions, which then become the basis for content development for a series, as happened for the Australian Indigenous-authored sketch comedy series Black Comedy (O’Toole & Shelpher 2014–2020) according to executive producer Kelrick Martin (cited in Looking Black 2022). Such collaboration may include the creation of dialogue with respect to character identity, for example for ethnicity. For the Australian animated children’s TV series Little J and Big Cuz (Lander 2017-present), script meetings, table readings, and writers’ workshops were used to create “a standardized Aboriginal vernacular” (e.g. the variants fulla and pela were standardised to fella) and to check expressions, idioms, code-switching, the inclusion of Dreaming stories, characterisation, etc (Ned Lander Media 2017). As script producer Clare Madsen further recalls: We set up table readings of every script at first or second draft stage with one of the writers and an actor brought in specifically for this process. […] It gave producer Ned Lander and myself as script producer the opportunity to hear the dialogue and ask questions about vernacular. It also gave us the chance to get a sense of our emerging characters, their obsessions and passions, and their sense of humour. We could also check on consistency of characters – for example: was Nanna too cranky in this episode or too soft in that episode? (Ned Lander Media 2017: 14–15)

This collaborative process and the specific involvement of screen creatives depends on the respective television series and there is not necessarily a uniform approach that is always used. Actors, for instance, “have varying degrees of creative authority and collaborative ownership of their ongoing characters” (Mittell 2015: 119). Thus, some showrunners (executive producers, head/lead writers) do not allow actors to make any changes to the dialogue in scripts, while others permit or even encourage it. Mittell argues that the performance of characters is a “collaborative creative act” that is “most typically developed through pre-production work between actors and showrunners” (Mittell 2015: 119). The showrunner’s importance is shown in the fact that they are the head writer and have overall managerial responsibility, while



Chapter 1.  Televisual characterisation

the goal of other writers in the team is to “mimic the showrunner’s voice in an effort for stylistic consistency” (Mittell 2015: 91). As one screenwriting manual puts it: “Your job as a staff writer on any TV series is to execute the vision of the Show Runner [sic]” (Bull 2007: kindle loc 3455). The showrunner may do major rewrites of scripts and sometimes also performs as actor in their own series. Screen creatives that are particularly important in the construction of characters are the series creator(s) and/or the writer(s) of the pilot episode – this may be the same person(s) as the showrunner. Screenwriting manuals advise their readers to study and capture the voices of existing characters (Bednarek 2018a: 212).3 This means that those who created the original characters in the pilot episode are an important influence on the subsequent creation of characters through dialogue. It is in the pilot that the writer/creator goes about “actually creating the characters and finding their unique voices” (Cook 2014: 241). As screen creative Jane Espenson explains: “A lot of that [building a character’s personality or individuality through language] happens at the pilot stage, which is before the staff is hired” (cited in Bednarek 2019a: 39). As summarised in Richardson (2010a) and Bednarek (2018a), the writers’ room is particularly important for the collaborative production of television dialogue, especially in the United States. It is often here that the writing team maps out character arcs. Screenwriters write their own episodes (on the basis of an approved outline) but also workshop and rewrite other writers’ outlines or drafts. As far as character dialogue is concerned, some lines may be suggested in the writers’ room, and may already be included in the outline, but the bulk of the dialogue is written when script drafts are produced and revised. As mentioned, the showrunner ensures that the script fits the standards, style, and arcs of the series. Consultants, network and studio executives also provide notes during the whole process, including on characterisation. Once the script has been finalised, rewriting may occur in rehearsals or table readings based on input from actors and other screen creatives or even as late as during the filming, but the extent to which this happens varies. During post-production, further changes can be made, for instance editing or cutting certain lines of dialogue. All of this has implications for the authorship of characters, including who is responsible for creating a particular language feature that is used for characterisation. In addition, events in the real-world involving the actors who perform core characters (e.g. pregnancy, illness, etc) can have impacts on character arcs and storytelling (Mittell 2015: 120–122).

3. This book uses the terms screenwriting and screenwriters, rather than scriptwriting and scriptwriters which are also sometimes used for television.

5

6

Language and Characterisation in Television Series

While writers’ rooms are typically associated with the Unites States, they are also used in Australia, where we can find different types of writers’ rooms, roles, and processes. While Australian writers’ rooms differ in various ways from those in the US, here, too, participants work collaboratively to “create or enhance characters” (Maloney & Burne 2021: 187), and in ongoing series they “work with given character […] conventions” (Maloney & Burne 2021: 189). In linguistics, the collaborative creation of television dialogue and its multiple authorship is clearly recognised. Dynel (2011: 313) uses the term collective sender for this “author”, for instance. In this book, I use screen creatives for all those involved in the creation of dialogue. However, I believe that screenwriters (including the lead/head writer or showrunner) have a key role to play in the creation of dialogue, so often just refer to the screenwriters. Even when I do so, this is not meant to imply that it was only the writers who were responsible for the dialogue. Ultimately, to attribute authorship to a particular linguistic feature that occurs in a character’s dialogue we would need to either observe the process used to create this particular linguistic feature or interview the screen creatives involved in the particular episode (and hope that they report accurately who was responsible). Another question – which this book is not particularly concerned with – relates to the extent to which screenwriters consciously use particular language features or are aware of specific resources studied in linguistics (see Reichelt 2018: 54ff; Richardson 2010a: 63–84). My past research has included interviews with Hollywood screenwriters and surveys of television screenwriting manuals (Bednarek 2018a, 2019a). I concluded that writers are aware of several aspects of language use, but that this awareness is primarily at play when they are in what we might call an “analytical” frame of mind – when they try to deconstruct television writing in manuals for the benefit of others or when they are asked to explicitly reflect upon dialogue in interviews. It is not necessarily drawn upon to the same extent while they are writing TV dialogue. With certain exceptions (e.g. swear/taboo words), screenwriters primarily seem to draw on their intuition, talent, and “tacit understanding, knowing how” (Richardson 2010a: 64, emphasis in original) rather than on analytical thinking about language. To conclude this section, I want to briefly note that the creative processes and employment conditions of screenwriters in streaming services (Netflix and the like) differ from those in network and cable television (Phalen 2021). The series that are analysed in this book do not include characters in series that were originally created for streaming services. However, the analysed series may now (also) be available through such services – e.g. Gilmore Girls and Redfern Now are available on Netflix in Australia.



1.3

Chapter 1.  Televisual characterisation

Characters and the audience

As has become evident, televisual characters occur in collaboratively-constructed narratives aimed at a target audience. Like in other mediated texts, there is a “double articulation” (e.g. Lorenzo-Dus 2009: 161), that is, an interaction between the on-screen televisual characters on the one hand (communication within the diegesis), and an interaction between the characters and the viewers on the other hand (communication with the narrative’s external audience). Most linguistic research on TV series draws on relevant work in literary and media studies (e.g. Pfister 1988; Scannell 1991; Short 1981) in recognising this “double” communicative interaction plane, although different terms are used for the layers.4 Relevant reviews are provided in Bubel (2008) and Locher and Jucker (2021). This book is not so much concerned with theorising the different layers and their contribution to characterisation, but does proceed from the assumption that characters communicate both with each other and with the audience. In relation to conceptualising this audience, many linguists have drawn on Goffman’s (1976, 1979) model of participation structure in their study of film (e.g. Bubel 2006, 2008) or television (e.g. Brock 2015; Dynel 2011, 2012). There is general agreement that the viewer is an overhearing, but ratified participant for whom the dialogue is ultimately designed. However, there is some debate about how to best conceptualise and label this role – whether as overhearer (Bubel 2006, 2008, 2011; Bubel & Spitz 2006), eavesdropper (Richardson (2010a) or recipient (e.g. Dynel 2011). Brock (2015: 28) suggests that these approaches can be reconciled by taking into account the different communicative layers: the television viewer is ultimately the intended recipient of television dialogue, but can also be assigned different listener roles at the level of the fictional world (most frequently that of the overhearer, sometimes that of the eavesdropper or speaker). Messerli (2017) discusses the differences between the models of participation structure offered by these and other linguists, while Locher and Jucker (2021) propose a new model for the more general category of “performed fiction”. While I do not use the participation framework explicitly to study processes of characterisation or character relationships (see Gregori-Signes 2020), the “audience design” of television dialogue provides 4. Direct viewer address (e.g. asides) disrupts the two levels, rendering their separation porous and explicitly acknowledging the audience (Sorlin 2016: 203). It could be argued that television series with a narrator (e.g. Jane the Virgin) include an additional “narratorial level”. Such a level is typically recognised in narrative genres such as novels or short stories (see Culpeper & Fernandez-Quintanilla 2017: 97). Further, Locher and Jucker (2021: 53–54)’s model suggests that where (live or pre-recorded) audience laughter occurs (e.g. in many sitcoms), this is situated on a “supradiegetic level”, which differs from the intra- and extradiegetic level. I would suggest that voiceover dialogue contained in television title sequences is also situated on this level.

7

8

Language and Characterisation in Television Series

an important backdrop to this book. The general assumption is that characters are designed with the target audience in mind, aiming for common ground (Bednarek 2010: 14–17; Bubel 2006: 57). While mass-mediated products are generally highly accessible, audience members that are not targeted by the screen creatives may have occasional trouble with comprehensibility (Bell & Gibson 2011a: 563). In this respect, Locher and Jucker (2021: 44–48) make a useful distinction between the core recipients (addressee recipients = target audience; meta recipients = critics/ academics) and the incidental and accidental recipients of television series. Fans are arguably a particular type of recipient, who have “sincere emotional attachments to characters” (Mittell 2015: 128). Fans may take on the identity of a character (e.g. role playing, cosplaying, tweeting/blogging as character), address characters (in live-tweets), take up certain character catchphrases, comment on or interpret character personalities, speculate about character reactions, display characters and their dialogue on T-shirts or purchase other character-based merchandise, create character gifs, remix videos, or create fan fiction and fan art about characters and their relationships (see overviews in Bednarek 2017a; b; Richardson 2010a). Mittell (2015) connects such fan activities to parasocial relationships, and there are also other ways in which viewers can form social and interpersonal connections with televisual characters, including (wishful) identification and affinity/ liking (Cohen 1999; Giles 2002). In media psychology, characters are thus said to “mediate a range of television effects, through the processes of imitation, identification, role modeling and parasocial interaction” (Livingstone 1998: 119). However, audiences may also react negatively to character dialogue, for example critiquing stereotypical, inaccurate or inauthentic representations of accent or dialect (see Bell & Gibson 2011b; Bleichenbacher 2012; Knooihuizen 2019; Mitchell 2015; Planchenault 2017; Queen 2015). Mittell (2015: 127–132) discusses in detail how television series allow audiences to recognise, align with, and forge allegiances with television characters. For example, allowing audiences access to the interior states of characters (their emotions, morality, thoughts) through exterior markers (appearance, action, dialogue) permits audiences to align with such characters. Locher and Jucker (2021: 192) provide an overview of different levels of emotional connection with fiction, ranging from emotional recognition to empathy. Culpeper and Fernandez-Quintanilla (2017: 115– 117) provide further discussion of audience perception of character and point out that this requires empirical audience research. This is beyond the scope of this book, which is not intended as a contribution to reader- or viewer-response research. Nevertheless – keeping the above points in mind – this book proceeds from the assumption that characters are important for television viewers, that viewers care about characters, that they can and do form parasocial relationships with characters, that they interpret dialogue in terms of character identity, that they have evaluative



Chapter 1.  Televisual characterisation

reactions to the construction of character identity, and that televisual characters are designed with the (target) audience in mind. The next section discusses the different aspects of character identity that may be at play in such processes. 1.4

Features of the televisual character

The two aspects of television characters that are particularly relevant to this book involve the stability of the televisual character (Section 1.4.1) and the question of what aspects of character identity can be distinguished (Section 1.4.2). In addition, Bal suggests that there are four principles “which work together to construct the image of a character” (Bal 1997: 126), and several of the case studies in this book will draw on these principles. They are: the principles of repetition (repeated characteristics), accumulation (of different characteristics), character-relations, and transformation (changes in character). The issues covered in Section 1.4.1 could be related to Bal’s principles of repetition, accumulation and transformation, as the section discusses how characteristics of a character are accumulated and repeated to create stable rather than changing characters. Bal’s character-relations are covered in Section 1.4.2 as an aspect of character, alongside individual and social character traits (which could also be seen as associated with the principle of accumulation). 1.4.1 Stability Televisual characters are generally described in terms of depth, complexity and dimension rather than change and transformation. They are discussed as being “stable” characters that do not change radically over time. However, long-running programs may show developments in characters and/or relationships with other characters. Pearson (2007) elaborates on this latter aspect as follows: The repetitive nature of the television series dictates a relative state of stability for its characters […] In television, it’s more accurate to talk about character accumulation and depth than it is to talk about character development. The long-running American television drama can create highly elaborated characters of greater accumulation and depth than any contemporary medium. (Pearson 2007: 56)

Television characters are thus regarded as stable figures who accumulate life events, experiences and relationships but do not change drastically from them. Douglas (2011: 14) speaks of vertical instead of horizontal development. The serial nature of television series allows for the gradual revelation of different (but consistent) character attributes, while characters’ changing interactions with other characters

9

10

Language and Characterisation in Television Series

(conflicts, friendships, romances, etc) create the illusion that the characters are fluid and dynamic (Mittell 2015: 136–137). The stable nature of televisual characters may be caused by the fact that they need to sustain a whole television series and are important for audience engagement. In Mittell’s (2015: 141–2) words: The desire for stable characters with consistent traits and personalities is a major draw for serial storytelling, as we want to feel connected to such characters through parasocial relationships and might be quite disappointed if they changed in ways that violate their initial connections and appeals.

Character consistency is indeed listed by Richardson (2010a: 95) among the standards that fans ask television creatives to uphold. Reichelt (2018: 25; 47) discusses how viewers’ familiarity with a television character’s stable presentation helps them not only to distinguish different characters performed by the same actor (e.g. in the Canadian series Orphan Black), but also to recognise special cases such as out-of-character moments or characters’ imitating other characters. There are also industrial incentives that help explain why television series and audiences place importance on a stable and safe core cast of characters (Mittell 2015: 126). Even if they are relatively consistent and stable, multifaceted televisual characters with accumulated depth and dimension could be described as “round” or “complex” characters (sometimes antiheroes, see Mittell 2015: 142–163; Schubert 2017a). In other words, stable characters are not necessarily “flat” characters (see Culpeper 2001: 52–57 for the literary-stylistic categorisation of flat vs round characters and Richardson 2010a: 134–146 for discussion in relation to television characters). In addition, certain characters may well undertake a “journey” within particular television series (see Mandala 2007: 63; 2011). Reichelt and Durham (2017) show that changing linguistic patterns can indicate character developments within a series, including shifts in characters’ stereotyping. Sorlin (2016: 22–23) suggests that changes in a television character’s status in a power hierarchy can be associated with changed linguistic behaviour. Reichelt (2018: 243–245) demonstrates that a character’s language use may change when the same character occurs in a spin-off series. Non-linguistic examples and general possibilities of character changes are discussed in Mittell (2015: 134–142), including shifting allegiances, changed moral beliefs, and psychological/emotional growth. In sum, stable characters are presumed to be frequent in television series, as they allow high degrees of viewer interaction, engagement, and potential for bonding and identification.



Chapter 1.  Televisual characterisation

1.4.2 Aspects of character identity Moving on to the different aspects of character identity that can be recognised (or accumulated), for film scholar Kozloff (2000: 43–46) these include class, intelligence, personality, character psychology, emotional state, or motive. Similarly, television scholar Pearson (2007), developing concepts by Bordwell (1985), proposes six elements of character identity: psychological traits/habitual behaviours; physical characteristics/appearance; speech patterns; interactions with other characters; environment; and biography. It has also been pointed out in television studies that certain genres use characters to contrast diverging ideologies between lifestyles (Feuer 2001). Linguists have examined many aspects of televisual character identity, including gender, sexuality, national identity, affluence, character relationships, character/ actant roles, and others (see survey in Bednarek 2017c). While not focussing on television, Culpeper (2001: 75–76; 87) distinguishes character categories into personal categories (e.g. preferences, interests, habits, goals, abilities, personality), social role categories (e.g. kinship, occupation, relational role), group membership categories (e.g. age, gender, class, nationality, religion) and genre-based dramatic role categories (e.g. heroes, villains). Building on this body of linguistic and non-linguistic research, Bednarek (2018a: 47) suggests that televisual characterisation includes at least the following aspects: I197 

I279 

I65 

I151 

– – – – –

The identity of characters as expressed through names The biographies (personal histories) of characters The roles of characters in the narrative (e.g. hero, opponent, helper) Characters’ individual traits (e.g. personality aspects, character psychology) Characters’ social traits (age, gender, sexuality, ethnicity, nationality, occupation, etc) – Characters’ expressive identity (their emotions, attitudes, values, and ideologies)5 – Character’s relationships to others (e.g. kinship, friendship, hierarchical status, hostility)

5. The notion of expressive character identity is used in Bednarek (2011a) to refer to character traits that concern emotions, attitudes, values, and ideologies. This model argues for the importance of such character traits in televisual characters and proposes that verbal and non-verbal expressive behaviour by characters construes certain kinds of expressive character identities, e.g. being “emotional”, “a whinger”, “upset”, “harmony-seeking”, being “feminist” or “conservative”. Further, such expressive resources can be considered at different levels (micro-, meso- and macro), depending on whether we consider their individual occurrence, strategies, or temporary/ stable identities and identity types. Since the focus of this book is not solely on expressive identity, it does not employ these different levels of analysis, although it does draw on the notion of expressive identity and expressive resources.

11

12

Language and Characterisation in Television Series

As far as these aspects of character are concerned, gender and sexuality have attracted most attention in linguistics (e.g. Baker 2005; Bednarek 2008a; Bubel 2006, 2011; Bubel & Spitz 2006; Gregori-Signes 2007, 2017; Paltridge et al. 2011; Queen 2012; Raymond 2013; Rey 2001), but there is also research on character relationships (e.g. Bednarek 2008a; Bubel 2006, 2011; Mandala 2007, 2008; Sorlin 2016) as well as on national identity (Walshe 2011), affluence (Queen 2012) and on expressive identity (Bednarek 2011a; Sorlin 2016: 48ff). Only some studies have focussed on narrative/actant roles (e.g. Sorlin 2016: 33–48 on House of Cards; Wodak 2009: 164–183 on The West Wing). The way regional, social and ethnic categories are represented in television (and film) through dialect has also attracted linguistic attention – overviews are presented in Stamou (2014); Planchenault (2017) and Bednarek (2018a: 23–28), and recent additions to this field include Hayes (2021) and Valleriani (2021). Hodson (2014) is an introduction to the representation of dialect in film and literature, with some comments on television series. Of these diverse aspects of character, this book focusses mainly on characters’ individual and social traits, as well as expressive character identity (Chapter 3). The social categories that I analyse include the “nerd” (Chapter 4) and “flawed woman” personae (Chapter 5) as well as gender (Chapter 3 and 5) and ethnicity (Chapter 6 and 7). Additional relevant background will be presented in the respective chapters, while the next section briefly introduces the principal linguistic frameworks used in the analyses. 1.5

Frameworks for the linguistic study of televisual characterisation

It is beyond the scope of this chapter to provide an overview of all linguistic research on televisual characterisation (see entries in Bednarek & Zago 2022) or on characterisation in other types of fiction (see e.g. Culpeper & Fernandez-Quintanilla 2017; Leech & Short 2007: 296–298; McIntyre 2014). Instead, this section focuses primarily on introducing important concepts from cognitive stylistics (Section 1.5.1) and sociolinguistics (Section 1.5.2) and the mixed approach that will be drawn upon in this book (Section 1.5.3). The basic underlying assumption is that dialogue works for “character differentiation” (Bednarek 2010: 132) – i.e. to create different and unique characters. This is also in line with media studies research and screenwriting advice, as further discussed in Chapter 3.



Chapter 1.  Televisual characterisation 13

1.5.1 Cognitive stylistics: Schemas and textual cues In the last decade or so, stylistics has increasingly focussed on language use in film and television (e.g. Bednarek 2011a; b; Hoffmann & Kirner-Ludwig 2020; McIntyre 2008).6 Approaches to characterisation informed by cognitive stylistics are particularly common and are also relevant to this book. Cognitive stylistics draws on concepts and frameworks from cognitive science (including cognitive linguistics), such as conceptual metaphor, prototypicality, blending, mind models, schemas/ frames, etc. With respect to characterisation, cognitive stylistic models typically assume that the interpretation of character involves cognitive processes and draws on the prior knowledge of viewers. The focus is predominantly on “how we form impressions of characters in our minds” (Culpeper 2001: 2), including how readers/ viewers’ attention, expectations and knowledge structures might be manipulated or exploited to create particular impressions. Of particular interest to this book is Culpeper’s (2001) influential cognitive stylistic model of characterisation. Culpeper focuses primarily on “how a representation of character (an impression of character) might be constructed in the mind during the process of reading” (Culpeper 2001: 34), and most of his examples come from Shakespearean plays. However, he also includes several examples from films and television series, indicating that this model is intended to apply to ‘telecinematic’ (Piazza et al. 2011) narratives as well. Since it ultimately aims to provide insights into how audiences comprehend a text (cognitive processes), Culpeper’s model is rather complex, involving components such as a control system, prior knowledge, a situation model, a textbase, and the surface structure of the text (Culpeper 2001: 34–38), and he also draws on attribution theory to explain inferencing (Culpeper 2001: 113–155). This section introduces only those concepts of his model that are relevant to this book: the notion of textual cues and the concept of schemata. Textual cues are “the particular linguistic choices attributed to characters” (Culpeper 2001: 37) which “give rise to information about character” (Culpeper 2001: 163), while schemata (schemas) are “structured bundles of generic knowledge” (Culpeper 2001: 28). Schemata may relate to different aspects of the world, including genre knowledge (e.g. knowledge about the character types in particular genres), world knowledge (e.g. knowledge about a restaurant visit) and social knowledge (e.g. schemata about social categories, called social schemata). Examples from television drama 6. A few early stylistic studies of television drama from the late 1990s are reviewed in Richardson (2010a), who also offers a general (now somewhat outdated) overview of interdisciplinary research in this area. Bednarek and Zago (2022) provides a bibliography of linguistic research on film and television, which is updated on a regular basis.

14

Language and Characterisation in Television Series

are schemata for categories such as “American detective, hospital nurse, hapless middle-class parents” (Richardson 2010a: 32). In essence, Culpeper (2001) proposes that audiences form impressions of characters based on an interaction between bottom-up cognitive processes (influenced by “textual cues” on the surface structure) and top-down processes (influenced by activated background knowledge, or schemata). McIntyre (2014) provides an example from a British television series (Pennies from Heaven) where the audience’s “initial schema for a married couple […] is tuned as a result of the textual triggers encountered in the dialogue, demonstrating how the characterisation of Arthur and Joan is achieved through a combination of top-down and bottom-up processing” (McIntyre 2014: 161). In other words, TV dialogue can first activate schemata and subsequently show whether characters conform to this prior understanding (Richardson 2010a: 138). Discourses that surround television series (e.g. trailers, reviews) may even activate character schemata before a series is broadcast (Richardson 2010a: 136). Social schemata (based on our knowledge of real people) play an important role in this process, as Culpeper argues: first impressions of characters are guided by the implicit models offered by social schemata. Such schemata, once activated, offer a scaffolding for incoming character information. Moreover, they allow us to make further knowledge-based inferences and thereby flesh out our impressions of character. (Culpeper 2001: 86–87)

However, schemata relating to genre-based dramatic roles also guide our perception of characters: Genres may activate viewer knowledge and expectations about particular dramatic roles (e.g. regarding a typical hero in a Western), which then play a role in character comprehension. While audiences thus interpret fictional characters from a “humanising” perspective, drawing on their knowledge about “real life” people, there are also clear differences to real life in how audiences infer the characteristics of fictional characters (Culpeper & Fernandez-Quintanilla 2017: 95–96). This includes the fact that audiences imbue character behaviour with particular relevance or significance because of the “double” communicative interaction plane of literary (and televisual) texts (Culpeper 2001: 145). As Richardson, puts it, viewers “are supposed to form character impressions in this context. The writing/performance/production is designed to promote particular impressions/inferences” (Richardson 2010a: 128). In addition, Culpeper posits that fictional characterisation involves character behaviours that are complete, since “by reading the whole text we have access to that character’s whole life – complete and finite” (Culpeper 2001: 145). However, while this may be the case for the literary characters that Culpeper focusses on, this is not necessarily the case for television characters which audiences may encounter



Chapter 1.  Televisual characterisation 15

episode-by-episode, season-by season. Recaps and updates may help to (re)introduce characters to viewers (Richardson 2010a: 137) and textual cues may be reiterated throughout the TV narrative (Thompson 2003: 27). A further aspect of this model is that characters are located on a scale: depending on whether they are inferred by accessing category-based processing – where there is great reliance on schemata – or whether they are inferred through person-based processing – where individual pieces of information are combined to form a character impression (Culpeper 2001: 121; Culpeper & Fernandez-Quintanilla 2017: 103). Culpeper (2001: 120–121) notes that each type of character impression comes with its own expectancies, whether these are category-based (schema-based) or target-based (person-based). Correspondingly, “flat” characters are described as highly stereotypical, schema-based characters that are “schema reinforcing”, while “round” characters are considered as characters that are “person-based” (Culpeper 2001: 93) and “typically schema refreshing” (Culpeper 2001: 95). In the context of television with its opportunity for character accumulation and depth (see Section 1.4.1 above), it is a matter for debate whether televisual characters based on well-known schemata remain flat characters, as Culpeper (2001: 94–96) appears to suggest. In principle, television characters can certainly be round rather than flat characters (Richardson 2010a: 137), and an initial schematic/flat character impression “can give way to textually determined, round character impressions” (Richardson 2010a: 143). Finally, it is important to underline that this model does not assume uniform, unvaried interpretation by viewers. As Culpeper (2001: 68), Snell (2006: 63) and Richardson (2010a: 137–138) all point out, people with different schemata may arrive at different reactions and interpretations of characters depending on variables such as their age, ethnicity, education, class, culture, etc. In this book, I only partially apply Culpeper’s (2001) model. I do draw on the cognitive linguistic concept of schema, but only in Chapter 4, where I explore how dialogue cues attributed to a television sitcom character align with a “nerd” schema. In other words, I use Culpeper’s model to examine whether a particular television sitcom character possesses the kind of character identity that fits a social schema (cf. Richardson 2010a; Snell 2006). This is similar to Culpeper’s (2001: 268–283) analysis of the character Katherina in Shakespeare’s The Taming of The Shrew, although I will be using different techniques. In the other chapters, I do not use schema theory, but I do draw on Culpeper’s concept of textual cues, which I will now discuss in more detail.7 As noted above, 7. Culpeper (2001: 164) explicitly notes that he follows Pfister (1988) in his organisation of textual features. His classification of explicit/implicit cues also has similarities with suggestions made by other researchers such as Ewen (1971, cited in Rimmon-Kenan 2002: 59–60), Bal (1997: 131),

16

Language and Characterisation in Television Series

it is textual cues that give rise to character information. These cues can be explicit, implicit and authorial (see Table 1.1). Explicit cues are those where “a character … provides explicit information” (Culpeper 2001: 167) about character. These cues include self- and other-presentation where characters provide information about themselves or other characters (Culpeper 2001: 167). In contrast, implicit cues are “cues that are important in conveying … character information which has to be derived by inference” (Culpeper 2001: 172), for example, conversational structure, lexical and syntactic features, accent and dialect, paralinguistic features, etc. Authorial cues, finally, are “cues over which the character notionally has no power of choice” (Culpeper 2001: 229) – for example, stage directions – although ultimately all characterisation cues could be considered as authorial and could instead be distinguished according to the communicative plane on which they operate (McIntyre 2014: 157). This idea of textual cues for characterisation can clearly be applied to televisual characterisation (see also Reichelt 2018: 57–62). As Table 1.1 shows, textual cues include linguistic cues (language choice), paralinguistic resources, kinesic features, and non-linguistic aspects such as clothing or a character’s company and setting. Most of these cues would be relevant for televisual characterisation, even though access to certain authorial cues (e.g. screen directions) may be indirect – audiences get to see the results of screen creatives’ authorial cues as performed on television. We could also analyse such screen directions if we examined scripts (screenplays) rather than on-screen dialogue. However, unlike literary plays, viewers are, on the whole, unlikely to encounter television characters by reading a script rather than watching a television series. In Richardson’s words, “it seems appropriate, when the production rather than the written script is the canonical text, for analysts to focus on the former rather than the latter” (Richardson 2010a: 39). Analysis of scripts would be useful for providing further insights into production and creation processes, however.

Lothe (2000: 81–85) and Toolan (2001). The distinction between Self- and Other-presentation appears similar to Bal’s (1997: 130) distinction between characters talking about themselves, characters talking about others, and narrators talking about characters. The idea that viewers reconstruct or infer character from explicit or implicit evidence and that they draw on real-world experiences in so doing appears in line with statements by Chatman (1978, cited in Richardson 2010a: 129–130). From a non-linguistic perspective, Mittell (2015: 130–131) suggests that as part of the cognitive viewing process, audience members infer television characters’ interior states through explicit exterior marks (dialogue, actions, appearance), the dramatic context, and their knowledge of characters – this in turn broadly aligns with Culpeper’s (2001) approach.



Chapter 1.  Televisual characterisation 17

Table 1.1  Explicit, implicit and authorial cues in characterisation (after Culpeper 2001) Explicit cues

Self-presentation (character gives explicit information about self) Other-presentation (character gives explicit information about other character[s]) Implicit cues Conversational structure (e.g. turn length, turn-taking, turn allocation, topic shift, topic control, incomplete turns/hesitations, interruptions) (Non)adherence to conversational maxims; conversational implicature Lexis (Germanic vs. Latinate, lexical richness/diversity, surge features/affective language, terms of address…) Syntactic structure Accent and dialect Verse and prose Paralinguistic features (e.g. tempo, pitch range/variation, loudness, voice quality) Visual features: kinesic features and appearance (e.g. stature, clothing, facial expression, posture) Context: a character’s company and setting (Im)politeness strategies Authorial cues Proper names Stage directions

Of the various cues listed in Table 1.1, this book focuses on explicit and implicit linguistic cues in character dialogue – henceforth called dialogue cues. A wide range of such dialogue cues have been examined in relevant linguistic studies of television characters. For instance, Bubel’s (2006) thesis investigates interactional/ conversational patterns (see also Lawson in press), while Reichelt’s (2018) thesis examines selected hedges, general extenders, modal adverbs, intensifiers, and pragmatic markers (see also Reichelt 2020a). Other dialogue cues have also been studied, including code-switching (e.g. Mandala 2008), impoliteness (e.g. Bruti 2018; Dynel 2012; Mandala 2011; Richardson 2010a), linguistic innovation (e.g. unusual -y adjectives, Mandala 2007; non-codified affixed items, Reichelt 2021), naming practices (e.g. Gregori-Signes 2020; Lawson in press), taboo/euphemism (e.g. Bednarek 2008a; Sánchez Ruiz 2018), language variation (e.g. Fine & Anderson 1980; Queen 2015; Walshe 2011), (dirty) joke-telling (e.g. Bubel & Spitz 2006), emotionality (e.g. Bednarek 2008a; Lawson in press), interactive/involved versus informational language (e.g. Rey 2001), and others (see entries in Bednarek & Zago 2022). While Beers Fägersten (2016a) is primarily an introduction to linguistics with the help of televisual data, in so doing it also illustrates the range of linguistic cues that are used in television series for the construction of characters and other stylistic effects. These studies do seem to highlight a clear association between the use of particular linguistic features and character’s individual and social traits and their relationship to other characters.

18

Language and Characterisation in Television Series

In this book, I primarily discuss dialogue cues that arise inductively from corpus linguistic techniques such as key word analysis (see Chapter 2), while also analysing some dialogue cues that were pre-selected for the specific insights they can provide into character identity (e.g. swear/taboo words and impoliteness strategies). I refer to such cues as constructing (aspects of) character identity. This terminology is not meant to imply that this type of linguistic analysis can show us how viewers actually perceive characters or what cognitive processes are at play while they form impressions of televisual characters. In addition to saying that such cues “construct” character identity, it is also possible to say that they “index” character identity. The next section explains how we can look at character through this lens of indexicality, a concept often used in sociolinguistic approaches to film and television and also relevant to this book. 1.5.2 Sociolinguistics: Telecinematic indexicalities While some sociolinguists dismiss media language as “artificial”, others have interrogated such language closely, suggesting for example that the narrative mass media play a significant role in establishing, reflecting, recycling and changing language ideologies, language attitudes and sociocultural norms (e.g. Coupland et al. 2016; Meek 2020; Queen 2015). In addition, sociolinguists who study film or television have drawn on concepts such as style/stylisation (e.g. Bucholtz & Lopez 2011; Coupland 2007; Gibson 2011), code (e.g. Androutsopoulos 2012b), social networks (e.g. Mandala 2007; Reichelt 2018, 2021), and audience/referee design (e.g. Bell 2011) to name but a few. Androutsopoulos (2012a) is an edited special issue of Multilingua dedicated to sociolinguistic studies of film and television. Bell and Gibson’s (2011a) special issue on the Sociolinguistics of Performance focuses on phonological features and includes articles on film and television (Bell 2011; Bucholtz & Lopez 2011; Gibson 2011). Queen (2015) is a book-length introduction to the use of sociolinguistics in the study of the narrative mass media, while Richardson’s (2010a) book offers a sociolinguistic study of television dramatic dialogue and Reichelt’s (2018) thesis on televisual characterisation is centrally informed by a sociolinguistic approach (see also Lee 2021). In my own study of characterisation in this book, I draw on sociolinguistic approaches in two key ways. First, I modify Androutsopoulos’s (2012b: 297) “three levels of analysis of sociolinguistic difference”: repertoire analysis, character analysis, and scene analysis. How I use different levels and units of analysis in this book will be explained in Chapter 2. Secondly, I draw on the sociolinguistic concept of indexicality. While this originally comes from semiotics, in relation to televisual characterisation it is strongly associated with sociolinguistic, sociocultural linguistic, or linguistic anthropological approaches. Indexicality refers to the complex



Chapter 1.  Televisual characterisation 19

association between linguistic features and social meanings, including the links between linguistic resources and identity categories such as stance, social personae, gender, or ethnicity (Bucholtz & Hall 2005). For example, the repeated use of a particular linguistic feature (e.g. a type of quotative) may be used by speakers to display or signal an identity category (e.g. being nerdy/popular, see Bucholtz & Hall 2005: 597). Dialects, too, are indexically associated with identity categories – that is indexicality runs the gamut from micro to macro categories of identity (Bucholtz & Hall 2005: 598). Matters are more complex than this brief overview suggests, as shown in theorisations about orders of indexicality (Silverstein 2003) and indexical fields (Eckert 2008), but these concepts will not be drawn upon in this book. The media have been recognised as important for the study of how indexical associations are enregistered and circulated (Bell & Gibson 2011b: 558–559; Craft et al. 2020: 396; Gibson 2011: 603). In the context of television drama, viewers “know that they should […] pay attention to the indexical signs, that this is a requirement for successful engagement with the production” (Richardson 2010a: 128, italics in original). Coupland argues that “TV in particular has put mediated linguistic diversity in front of the viewing public far more pervasively and with much richer and more saturated indexical loading than face-to-face social reality” (Coupland 2010: 69). Performances of characters in the media are important because they associate particular linguistic resources with social types/personae (Bell & Gibson 2011b: 558). Such mediated performances use established linguistic resources to index social meanings such as ethnicity, class, nationality as well as personae such as the “femme fatale” or the “tireless romantic” (Bell & Gibson 2011b: 569). Viewers draw on their prior knowledge to understand such performances and may also create new indexical links between linguistic features and social categories by watching and listening to such representations (Gibson 2011: 603). Thus, staged performance is central to processes through which styles that index social personae or figures become engraved or “enregistered” (Bell & Gibson 2011b: 561, drawing on Agha 2003). Androutsopoulos proposes the term cinematic indexicalities as “a cover term for the various layers of indexical meaning that sociolinguistic difference can articulate in film” (2012b: 302). These differences rely on indexicality in everyday life but are recontextualised and function for particular narrative purposes such as characterisation or creating a setting. They rely on language-ideological assumptions by screen creatives about how particular social types are expected to use language and can be interpreted differently by audiences (Androutsopoulos 2012b: 303–304). In general, the linguistic features that are used for characterisation in fiction tap into indexical fields that are both time- and culture-dependent (Locher & Jucker 2021: 108, 118). These points also apply to television series such that we can speak of telecinematic indexicalities if we want to cover both types of narrative mass media.

20 Language and Characterisation in Television Series

Queen (2015) applies the concept of indexical meaning to the study of both film and television, distinguishing between the indexing of social variables or social personae (type indexicality) and the indexing of personality features or stances (trait indexicality).8 As she explains: Type indexicality is the meaningful connection of language to the broad set of social kinds that characters often inhabit. We can think of these both as their social demographics, like race, gender, age, sexual orientation, and class, as well as specific kinds of personae, like nerds, jocks, and girly girls. Trait indexicality, on the other hand, is the connection of language to variability within categories of social demographics and personae, such as being quiet, cheerful, melancholy, or energetic. (Queen 2015: 176, italics in original)

An example Queen provides from the movie Boyz n the Hood is the different frequency of multiple negation and null copula in the dialogue of two African American characters who are similar in age, gender, and neighbourhood, but who are distinguished in their orientation or stance towards their circumstances (Queen 2015: 177–178). The different frequency of these features of African American Vernacular English (AAVE) thus indexes the two characters’ different orientation/ stance (trait indexicality). Fine and Anderson’s (1980) analysis of AAVE features in three US sitcoms further showed that different frequencies of relevant features are associated with identity variables such as sex, age and socio-economic status as well as with stereotypes. For example, urban teenage middle-class characters may use such features much less frequently than urban teenage gang members, and use of these features can also index class (Fine & Anderson 1980: 403–405). Type and trait indexicality may play different roles in characterisation depending on characters and series: While social categories such as gender may be a defining category for some television characters, individual character traits such as their stance or attitude to life may be more important for other characters (Reichelt 2018: 256). Reichelt’s (2018) thesis provides many examples for how particular language resources index social demographics (age, gender, nationality, socio-economic status), personae (e.g. a Valley Girl persona), and personality features (e.g. self-assurance, pragmatic competence). In some cases, it is only 8. The underlying idea that television characters can be considered in terms of personality traits (~trait indexicality) as well as social variables (~type indexicality) is in line with Selby and Cowdery (1995) who follow a structural framework and put character traits on a continuum for variables such as intelligent, sociable, pleasure oriented, excitable, warm (Selby & Cowdery 1995: 177) and also analyse characters with the help of social variables such as age, ethnicity, class, domestic situation (Selby & Cowdery 1995: 176). These authors also recognised that certain television genres feature social types (e.g. efficient secretary) and stereotypes (e.g. domineering wife) (Selby & Cowdery 1995: 108).



Chapter 1.  Televisual characterisation 21

particular features that index a character trait (with certain characters preferring particular variants), but in other cases frequency patterns are repeated or consistent across features and may point to stereotyping or typification (Reichelt 2018: 253). Importantly, conventionalised indexicalised meanings are not only associated with one or two linguistic features but can also become attached to packages or clusters of linguistic features. In this case, an entire bundle of features makes up an “indexical package” and “points to a character with traits that are culturally recognizable” (Queen 2015: 127). The way indexical meaning works in film and television is influenced by the goal “both to differentiate characters from one another in a believable way and to minimize the degree of variability in the interpretation of those characters” (Queen 2015: 129). Telecinematic indexicality therefore draws on easily recognisable and ideologically salient or iconic cues as well as familiar associations made in the media (Queen 2015: 129; 165). An example for the latter would be the association of Received Pronunciation (RP) with the dramatic role of the sophisticated villain in Hollywood films (Culpeper 2001: 207–208). However, this tendency does not necessarily lead to sameness across the television landscape. While televisual characterisation does build on recognisable or salient patterns and stereotypes and some of these do recur in different series, there is nevertheless also evidence of variation across series – even when similar character types or personae are performed by the same actor (Reichelt 2018: 18, 107, 248, passim). Although I am using the terms type indexicality and trait indexicality in Queen’s sense in this book, I use the term character traits as a cover term for both social and individual aspects of character (see Section 1.4.2). In this terminology, linguistic features or feature packages index social and individual character traits. For example, the use of norm-breaking swear/taboo words may index a character’s social persona (see Chapter 5), the use of specific lexis may index a character’s ethnicity (see Chapters 6, 7), or the use of particular interjections may index a character’s age/ generation (see Chapter 3). As explained above, this indexical meaning is shaped by the “double articulation” of telecinematic discourse, depends both on the context of occurrence and on the listener and can therefore be interpreted differently by audiences (Androutsopoulos 2012b: 304; Queen 2015: 128). Thus, speakers of different language varieties or different first languages may not share the same indexical associations. This “makes the nature of the meaning of the linguistic variation in the film dependent on the viewer” (Queen 2015: 130). In this book, any mention of indexicality must be read with this caveat in mind – these are potential associations which may differ among viewers. This is similar to the point made above in relation to schemata which may also differ among viewers. In addition, associating a language feature with one character trait simplifies matters considerably, since several linked character traits can be indexed by the same feature.

22

Language and Characterisation in Television Series

1.5.3 A mixed approach to televisual characterisation This book applies a mixed approach to televisual characterisation, incorporating the concepts introduced in Sections 1.5.1 and 1.5.2 above. Arguably, the sociolinguistic and the cognitive stylistic approach to televisual characterisation are not incompatible. In fact, Culpeper’s (2001) model combines insights from sociolinguistics, literary theory, and social and cognitive psychology. Richardson’s (2010a) study is explicitly identified as sociolinguistic, but uses Culpeper’s model of characterisation. Both models argue that there is a non-random relationship between language and different character aspects. Both models argue that we need to consider both the audience and the narrative context in which characterisation occurs. Both models assume that viewers’ knowledge or experience of the “real” world influences how they perceive and categorise characters. And both models allow us to investigate types of characters that are construed to be easily recognisable by the audience. In cognitive stylistic terms we might talk about these in terms of social schemata (stereotypes) and in sociolinguistic terms, we would refer to these as social personae (created on the basis of language-ideological assumptions). Each model, however, emphasises or foregrounds different aspects of televisual characterisation and makes specific connections between language and character identity. While the approaches are not radically incompatible, they provide different conceptual lenses into televisual characterisation. The case studies in this book use concepts, terminology, and assumptions from both models. Of course, these two frameworks are by no means the only approaches used in linguistics for the study of television characters. Pragmatic or pragma-stylistic approaches provide frameworks to help explain how particular aspects of character identity are achieved through implicit dialogue cues (see Culpeper 2001: 175–182 and Table 1.1 above). Locher and Jucker (2021) is a new textbook dedicated to The Pragmatics of Fiction and includes a chapter on character creation, which applies to characters in television series as well as other types of fiction. Sorlin (2016) takes a pragma-stylistic perspective to the analysis of House of Cards, its protagonist Frank Underwood, and occasionally other characters, drawing on pragmatic frameworks such as Gricean principles, speech act theory, facework, and (im)politeness to focus on power/authority and manipulation. Statham’s (2015) analysis of The Sopranos makes use of the participation framework as well as Gricean principles to examine characters’ participant roles and interactions in crime talk, albeit without an explicit focus on characterisation. The mixed approach to televisual characterisation adopted in this book integrates pragmatic approaches to impoliteness, since (im)politeness features “can be a central textual technique in characterisation” (Culpeper 2001: 261). The pragmatic framework for analysis of impoliteness will be introduced in Chapter 4. Pragmatic



Chapter 1.  Televisual characterisation 23

analysis can clearly be integrated into a sociolinguistic or cognitive stylistic approach. Reichelt (2020a), for example, analyses pragmatic competence but from a sociolinguistic perspective. Bubel (2006) uses Conversation Analysis in her development of Culpeper’s approach with respect to what she calls relationship impression formation – that is, how viewers form impressions of relationships between characters. Sorlin’s (2016) above-mentioned pragma-stylistic study also partially draws on psychological and cognitive research. Other mixed approaches also exist. For example, Schubert (2017a) combines approaches from register theory, conceptual blending theory, and Culpeper’s characterisation model in his study of immoral villains in Breaking Bad, House of Cards, and Dexter, while Schubert (2017b) integrates cognitive linguistic concepts in his Critical Discourse Analysis of Mexican and Mexican American characters in the same series. Harrison (2020) combines concepts from film studies and cognitive stylistics in a multimodal analysis of the “split” selves of the character June/Offred in The Handmaid’s Tale. The chapter on character creation in Locher and Jucker (2021) mentions Culpeper’s explicit/implicit/authorial cues alongside various sociolinguistic concepts including indexicality. Richardson (2010a) also uses theories from sociolinguistics, stylistics and other linguistic sub-disciplines. Clearly, the linguistic study of televisual characterisation profits from such sub-disciplinary combinations, as this allows us to illuminate television characters from different perspectives within linguistics. As such, I see this book as a contribution to mixed-methods “telecinematic stylistics” (Hoffmann & Kirner-Ludwig 2020) rather than as a contribution specifically to, say, cognitive stylistics, pragma-stylistics, or sociolinguistics. Before concluding this section, a brief comment is necessary on the multimodal construction and analysis of televisual characters: Television series are instances of “performed fiction” (Locher & Jucker 2021: 5), where actors perform characters in multimodal contexts. As already mentioned above, televisual characterisation does not just consist of language choice. It also involves what I have elsewhere (Bednarek 2010: 18–21) called “multimodality in the product” (camera angle, shot size, lighting, etc) and “multimodality in characters” (gaze, gesture, body position, facial expression, clothing, etc). In Culpeper’s model, textual cues include paraand non-linguistic cues. Televisual dialogue is performed rather than just written on the page, and multimodal analysis of characters in film and television is clearly a worthwhile endeavour (see e.g. Bell & Gibson 2011b; Culpeper 2001; Harrison 2020; McIntyre 2008; Reichelt 2020a; Richardson 2010b; Toolan 2011). For reasons of scope, this book does not include any analysis of such multimodal aspects, but I have incorporated this previously (e.g. Bednarek 2010: 143–179, 2015a). As such, the book is not intended as a contribution to multimodal stylistics or multimodal corpus-assisted discourse analysis.

24

Language and Characterisation in Television Series

In terms of the topology for corpus and discourse analysis proposed by Bednarek and Caple (2017a; b), the analyses in this book are hence purely “intrasemiotic” (monomodal). For the most part, they are also “intertextual”, focussing on linguistic patterns that can be found across the multiple texts included in a corpus. (The term intertextual does not refer here to literary intertextuality/allusion, but rather refers to patterns across texts). However, examination of selected scenes will allow some analysis of “intratextual” patterns, e.g. impoliteness within character interactions (Chapter 4). In addition, the studies in this book are examples of what Reichelt (2018: 33) calls “intra-source studies”, in that the aim is not to systematically compare television dialogue with naturally-occurring language, but rather to study television series in their own right. For this intra-source investigation, the book draws on a range of relevant concepts, as introduced above, including character differentiation, character stability, expressive character identity (Chapter 3), characters and stereotyped schemas (Chapter 4), complex (flawed) characters and gender norms (Chapter 5), and the indexing of ethnicity (Chapters 6 and 7). 1.6

This book’s case study approach

This book draws primarily on case studies of televisual characterisation published between 2010 and 2021 (see details in the Acknowledgments). Each study has been edited, updated, and incorporates new analyses. The book also includes a previously unpublished study of a new Australian corpus. Brought together in this book for the first time, I hope that these analyses constitute more than the sum of their parts. Crucially, I consider them as case studies for exploring different aspects of televisual characterisation and using different data, methods, and approaches in its analysis. Chapter 3 presents two case studies of the character Lorelai from the dramedy Gilmore Girls (Sherman-Palladino 2000–2007), focusing on character differentiation and character stability, respectively. Chapter 4 consists of a case study of the character Sheldon from the sitcom The Big Bang Theory (Lorre & Prady 2007–2019), examining dialogue cues in relation to the stereotypical nerd schema. The case study in Chapter 5 compares “flawed” female televisual characters from three different series (Nurse Jackie [Brixius et al. 2009–2015], Weeds [Kohan 2005–2012], and Saving Grace [Miller 2007–2010]), and looks at the characters’ use of swear/ taboo words through the lens of norm-breaking and gender identity. Chapter 6 moves from the US to the Australian national context and presents a case study of the drama anthology Redfern Now (Dale & Dear 2012–2013), examining its use of dialogue cues for indexing Aboriginal identity. Finally, Chapter 7 goes beyond the analysis of one or more individual series, consisting of one case study of a



Chapter 1.  Televisual characterisation 25

stigmatised language feature (ain’t) across 66 different US TV series and a second case study of the use of Australian Aboriginal English words across 13 Australian TV series. These two case studies provide insights into characterisation from a wide-lens angle, with a focus on social variables such as ethnicity. The corpus linguistic techniques used for the various case studies in Chapters 3 to 7 are introduced in the next chapter, while Chapter 8 offers a critical reflection on corpus linguistic methods for the study of televisual characterisation.

Chapter 2

Corpus linguistic analysis of televisual characterisation Data and approach

2.1

Introduction

As mentioned in Chapter 1, this book uses corpus linguistics to analyse televisual characterisation. This chapter introduces the corpus linguistic techniques, types of data, and units and levels of analysis that are used in the case studies. All analyses were undertaken using versions of Mike Scott’s WordSmith Tools corpus software (e.g. Scott 2004, 2019, 2020), with the exception of lexical profiling analysis which was undertaken using AntWordProfiler (Anthony 2013), and collocation analysis/visualisation which was undertaken using LancsBox/GraphColl (Brezina et al. 2015) as well as WordSmith. Before presenting the relevant corpus linguistic techniques and concepts (Section 2.3), data (Section 2.4) and units/levels of analysis (Section 2.5), I first provide a brief overview of other corpus linguistic studies of televisual characterisation (Section 2.2).9 2.2

Corpus linguistic analyses of televisual characterisation

Corpus linguistic techniques have been applied in the study of television series across a range of fields ranging from register studies to second language pedagogy to translation studies and more (see Bednarek & Zago 2022). Most relevant to this book are studies that concern televisual characterisation in English-language programs. This includes some studies by sociolinguists who would perhaps not identify their work as corpus linguistic, but who do use TV dialogue datasets and frequency measures in their analyses. Typically, such studies transcribe and/or code particular features and then analyse their distribution. For example, Queen (2012) includes quantitative analysis of distributional patterns of gender-indexical language strategies (e.g. questions, imperatives, hedges, tags, compliments, taboo words, endearments) across nine episodes of a daytime drama, comparing male and 9. This overview excludes the case studies that are revisited in this book in Chapters 3–7.

28

Language and Characterisation in Television Series

female characters and tying the identified sociolinguistic package to affluence as an aspect of character. Walshe (2011) also looks at the distribution of features, in this case grammatical, lexical and discourse features of Irish English in the sitcom Father Ted. In addition to transcribing and then tallying total number of occurrences, he also notes their distribution across episodes, thus “preventing a high number of appearances of a feature in one episode from distorting the general impression […] created for the whole series” (Walshe 2011: 129). The focus of Walshe’s study is not so much on characterisation but rather on authenticity and humour, but national identity can be considered as an aspect of character (see Chapter 1). Other sociolinguistic studies (Mandala 2007; Reichelt 2021; Reichelt & Durham 2017) use token distribution to identify language use by characters and character groupings in Buffy the Vampire Slayer. Mandala (2007) examines the linguistic variable of marked -y suffix adjectives (e.g. Heart-of-Darkness-y) and its distribution across the speech of individual characters (Buffy, Willow, etc), linking its use to the status of characters within their social network(s), for example as a marker of in-group identity (friendship). Reichelt (2021) broadens this study to the distribution of non-codified affixed items, while Reichelt and Durham (2017) focus on the distribution of tokens of adjective intensification and include aspects of character such as gender, age, nationality and group membership. Together, the results indicate that language patterns are used for characterisation in Buffy the Vampire Slayer, both in relation to individual characters as well as character groupings (e.g. in- and out-groups). These and similar sociolinguistic studies are important because they show that the quantitative distribution of linguistic resources in television series is linked to characterisation, both in relation to indexing one or more character traits (e.g. age, gender, nationality) and in relation to social networks/group membership (e.g. friendship groups) and character shifts (becoming friends). However, results are not always based on transcripts of the whole corpus and different techniques for presenting and normalising features are typically used. In corpus linguistics, one of the earliest relevant studies is Rey (2001). The study uses the Multi-Dimensional Analysis (MDA) framework developed by Biber (1988) to explore changing patterns in language use by male and female characters in three different Star Trek series from the 1960s to the 1990s. In line with changing gender roles in society, she identified a shift from involved toward more informational language choices in the dialogue of female characters. Csomay and Young (2021) revisit this study using key word analysis (see Section 2.3.1 below) to analyse both the “aboutness” of the series (including hierarchical character relationships) and changing gender roles over time. While neither of these studies are studies of televisual characterisation per se, gender is clearly an important aspect of character,



Chapter 2.  Corpus linguistic analysis of televisual characterisation 29

and we will therefore revisit these and other relevant linguistic studies of gender in Chapter 5. Key word analysis is also used in Baker (2005) which includes one chapter on the US sitcom Will & Grace, examining key semantic and grammatical categories in the speech of four characters. Although the main focus of this study is on discourses rather than televisual characterisation, it shows how key word analysis can be used to analyse and compare the language choices by different characters. Results are in turn linked to the construction of different character traits (e.g. self-assured, playful, self-absorbed, high-maintenance, needy), character relationships (e.g. friendship) and stereotypes (e.g. the “camp gay man” stereotype; stereotypical femininity). Lawson (in press) also uses key word analysis to study masculinities in a sitcom (Brooklyn Nine-Nine), which provide insights into the identities of each character, their relationships (family, personal, work relationships) and societal notions of masculinity. Bruti (2018) combines quantitative key word analysis with qualitative analysis with a focus on impoliteness and character traits such as power, aggression and disrespect in the drama series The Young Pope. In addition to key word analysis, other techniques are sometimes used in corpus linguistic studies of television characters. For instance, Baker (2005)’s study of Will & Grace also includes analysis of words relating to sex, love and sexuality, which is partially linked to relationships between the characters, character attitudes (e.g. towards homophobia) and the expression of gay identity. Bednarek (2010: 180–227) includes an analysis of the frequency and use of search terms relating to meat and vegetarian food in the dramedy Gilmore Girls, associating this with characters’ attitudes as well as the series’ ideology. Gregori Signes (2020) analyses names and other expressions referring to the victim in Twin Peaks, examining their frequency, distribution and the speakers who utter them, with the aim of studying character relationships. Together, such corpus linguistic studies of television characters demonstrate that corpus linguistic techniques can be used to retrieve language choices linked to individual and social character traits as well as character relationships and social stereotypes and norms. Key word analysis seems to be a particularly promising methodology for the retrieval of implicit and explicit character cues, and is also used in most of the case studies included in this book. This technique is introduced in the next section along with other important concepts in corpus linguistics.

30

Language and Characterisation in Television Series

2.3

Corpus linguistic techniques and concepts

2.3.1 Frequency and keyness An important first concept for all the case studies in this book is that of frequency. Frequency information will be relevant in relation to words (word forms such as look, looks, looked, looking), word categories (e.g. interjections) and longer structures (n-grams). Importantly, the actual number of occurrences (raw frequency) must be normalised to allow comparison of corpora of different length. Typically, frequencies are normalised to a given number of words which is dependent on the corpus size (e.g. frequency per 10,000 words, per 100,000 words, per 1 million words). Corpus linguistic software programmes and interfaces permit retrieving the frequency of a given item or the frequency of all items (frequency list). Frequency information can also be used to establish thresholds which have to be reached before a word is included in the analysis (e.g. requiring a minimum frequency of two). Frequency information is the basis for analysis of keyness, using a corpus linguistic technique that is called key word analysis (also spelt keyword analysis). While there are different ways in which the term key word has been used (see Stubbs 2008), in relation to corpus linguistics the term most often refers to a word that occurs with unusual frequency in a given text or collection of texts (variously called the node, target, focus, study corpus or corpus of interest) in comparison to a second corpus (called the reference corpus). Such a key word can be unusually frequent (“positive” key word) or infrequent (“negative” key word), as determined through statistical measures (Scott 1997; Scott & Tribble 2006). In Culpeper’s (2009: 34) words, “‘keyness’ is a matter of being statistically unusual relative to some norm”. In essence, key word analysis works by compiling word frequency lists of the two corpora, comparing both lists, and applying statistical testing to identify those words that are “key” in the node corpus. The calculation takes into account the different sizes of the corpora. Because key words are retrieved on the basis of comparing two corpora, the choice of reference corpus – which provides a baseline or norm – influences the key words that are identified (Culpeper 2009: 34–35). However, it has been suggested that “above a certain size, the procedure throws up a robust core of KWs [key words] whichever the reference corpus used” (Scott & Tribble 2006: 64). Importantly, key words vary in their raw frequency and distribution across corpus files (e.g. Culpeper 2009; Egbert & Biber 2019). While key word analysis is a staple technique in corpus linguistics, it has recently been debated how key words should be identified. One discussion concerns the question of which statistic measure to use in the retrieval and sorting of key words. Various measures have been suggested (see Brezina 2018: 85; Egbert & Biber 2019: 80; Gabrielatos 2018; Gries 2021; Jeaco 2020), although log likelihood (LL;



Chapter 2.  Corpus linguistic analysis of televisual characterisation 31

G2 – Dunning 1993) has traditionally been the most widely used, and was also applied in the case studies in this book. The log likelihood measure is able to reveal features of thematic prominence, style, and markedness in (collections of) texts (Jeaco 2020: 150). It tells us if the frequency difference between two corpora is statistically significant by providing a LL value which corresponds to a particular p-value. For example, a LL of 10.83 or higher is significant at p  2)

Lorelai Luke Rory Sookie Emily Richard

aw, for the love of god, (oh) thank god, rats, ugh, whew, whoa, yay aw/oh man, crap, geez/jeez, damn (it) bummer, wow, whoa, yuck, yikes ew, god, oh god, oh my god, (oh) thank god for x’s sake, (oh) my goodness my god

Starting with the female characters, Lorelai’s interjections are informal/young (rats, yay) and, according to dictionary definitions (see Bednarek 2010: 88), potentially signal emotions of annoyance (rats), pleasure (yay, (oh) thank god), disapproval, protest, sympathy (aw), anger/impatience (for the love of god), surprise/relief (whew), and disgust (ugh). Thus, she is quite varied emotionally, which would be in line with her construction as an emotional character. Rory’s interjections are informal/young (bummer, whoa, wow, yuck, yikes), and can construct emotions such as disappointment (bummer), surprise/admiration (whoa, wow), disgust (yuck), and surprise/fear (yikes) – again relatively varied emotionally. Sookie’s interjections predominantly involve god (surprise, shock, annoyance), and indicate her not being constructed as very varied emotionally. Emily’s interjections are more formal/“older” than that of the other female characters and potentially express emotions of annoyance/impatience (for x sake) and surprise (my goodness). Finally, Luke’s interjections seem associated with negative emotions and swear/taboo words: man (surprise, anger, impatience), geez (surprise/shock/anger), damn (it) (annoyance/disappointment), crap (associated with displeasure).15 To some extent, these interjection patterns are related to constructing stereotypical femininity/masculinity (see Chapter 5; see Bednarek 2010: 131). However, since there are differences in interjection use within the female characters, it is clear that interjection usage is not only associated with the construction of (stereotypical) gender identity. Rather, patterns of interjection usage appear to be multifunctional cues with respect to characterisation: They 15. Richard’s interjection (my god) only occurs three times so is not discussed here.

58

Language and Characterisation in Television Series

construct relationships between characters as well as informing us about both social and individual character traits (i.e. type and trait indexicality), including but not limited to characters’ expressive identity. In summary, this first case study has shown that Lorelai is constructed as a unique character through repeated linguistic patterning, allowing viewers to engage with her as a specific character. Part of her characterisation includes the fact that she is an emotional character, as seen for example in her usage of emotive interjections. It is hypothesised that this uniqueness of character dialogue is the case for many contemporary television characters, since unique characters allow viewers to emotionally engage and identify with who is portrayed on the screen. Further, such characters attract viewers to television narratives, and are a way of creating viewer engagement. Indeed, Reichelt (2018) presents further evidence that fictional characters across series do exhibit unique patterns of language use. 3.4

Case study 2: Character stability

3.4.1 Lorelai: An example of a “stable” televisual character? In the previous case study, the focus was on character differentiation. We now move on to the analysis of character stability, here defined in terms of variation in a character’s language use. As mentioned in Chapter 1, television characters are usually described as stable and consistent. Screenwriting advice commonly states that character voices (their dialogue) should be “consistent” (e.g. Cook 2014: 255). A TV series will only succeed if “it is consistent from week to week. The characters have to remain the same people” (Bull 2007: kindle loc 3465). Some advice goes as far as stating that “The golden rule is that characters must be consistent” (Smethurst 2016: kindle loc 1439, emphasis in original). Two aspects of stability and variation will be explored using the corpora described in Section 3.2 above: – How much does Lorelai’s language vary across seasons 1 to 7? – How much does Lorelai’s language vary according to who she is talking to? The analyses thus investigate both diachronic and intersubjective stability. With respect to the latter, it has been argued that: Characters can be shown to adjust their modes of expression depending on whom they are talking to, to be more or less articulate, more or less equipped to cope with particular interactional circumstances, and more or less in control of how the interaction unfolds and of the outcomes that result from it. (Richardson 2010a: 106)



Chapter 3.  Character differentiation and character stability 59

In a pilot study, I explored these two aspects of variation using the standard default settings of WordSmith, i.e., the Log Likelihood statistic, a minimum frequency of three, and a p-value of 0.000001. First, to investigate how Lorelai’s language changes over the seasons, key words and n-grams were produced of Lorelai’s dialogue in individual seasons (1, 2, 3, 4, 5, 6, 7 respectively), with the reference corpus comprising her dialogue in all seasons. The results indicated that Lorelai’s language does not vary much across seasons, as there were only a few positive and negative key words and n-grams per season (Table 3.5). The differences between seasons mainly relate to the names of characters (e.g. Rachel, Kirk, Logan) and variants of want/wanna or going/gonna, which might be the result of inconsistent transcription (see Chapter 2). In any case, they do not seem to indicate a big shift in Lorelai’s language across seasons. Table 3.5  Key words (# and spelling variants excluded; p = 0.000001; no minimum frequency)  

Positive key words and 2-grams

Negative key words and 2-grams

S1 S2 S3 S4 S5 S6 S7

Rachel ugh Lorelai, Jess, wanna, Rory Jason, during Logan Paul, Anka, going, during, going to, Paul Anka, June third gonna, know, snakes, rocket, Christopher, you know, want to, I mean

Kirk Lorelai, going to want to – wanna, gonna gonna going to

To investigate how much Lorelai’s language varies according to what other character she talks to, WordSmith was again applied to different GiGi sub-corpora. Lorelai’s complete dialogue was used as a reference corpus, whereas node corpora include files with dialogue addressed to (a) Christopher, (b) Emily, (c) Luke, (d) Richard, (e) Rory, (f) Sookie. Different relationships with Lorelai are enacted by these characters: Christopher is Rory’s father and Lorelai’s friend/partner/husband; Emily is Lorelai’s mother; Luke is Lorelai’s friend/partner; Richard is Lorelai’s father; Rory is Lorelai’s daughter, and Sookie is Lorelai’s best friend. The analyses showed that there is slightly more variation than when considering Lorelai’s dialogue across seasons, with key words/n-grams classifiable as names (e.g. Chris, GiGi, Sherry, Luke, Michel), terms of address (dad, mom, honey), reference to family roles (dad, mom), pronouns (+ human: he, her, she’s, you), interpersonal and hesitation markers (hey, hi, thanks, um), and content words (snakes, diner, seventy). However, some of these only occur a few times, are not distributed across texts, and purely reflect what the characters talk about in one episode. For instance, all twelve instances of snakes

60 Language and Characterisation in Television Series

occur in episode 7.4, where Lorelai and Christopher talk about the movie Snakes on a Plane. This key word certainly does not reflect a recurring change in Lorelai’s language when talking to this character. Again, not many key n-grams can be found, with the only key n-gram (mom I) occurring in Lorelai’s dialogue with her mother. Thus, the analyses using standard default WordSmith settings seem to indicate a relative stability of the character of Lorelai, but is the low p-value setting appropriate? The WordSmith help page notes: “In the case of key word analyses, where the notion of risk is less important than that of selectivity, you may often wish to set a comparatively low p value threshold such as 0.000001 […] so as to obtain fewer key words.” I take this to mean that in most searches a p-value of 0.000001 still results in a considerable number of key words. These analyses, then, appear to indicate that the language of the televisual character Lorelai does remain fairly stable, in particular when considering her dialogue across seasons. In other words, no significant “style shift” – “variation within the speech of an individual” (Bell 1991: 104) – can be detected in terms of key words and n-grams. While this finding might be the result of applying a quantitative, statistically-based methodology to characterisation which focusses on word forms and n-grams, it may also be tied to characterisation. For instance, Lorelai and her daughter Rory are supposed to be best friends. This may result in Lorelai’s style not shifting greatly when talking to her. It may also be part of Lorelai’s character identity that she does in fact not talk very differently to, say, her mother and father, than to other characters. We can revisit Example (1) above in this respect. As we have seen, Lorelai does not change the informal way she speaks (what just went down, score a deal, bagged the Swede), although she talks to her “proper” father, Richard, who plainly has difficulties understanding what she means. Despite this being made explicit in Richard’s turns (down where?; one doesn’t score deals), she continues not to shift her style (You bagged the Swede?). Thus, the relatively low occurrence of intersubjective style shift may in fact contribute to Lorelai’s characterisation, constructing a particular aspect of her identity in that she does not “accommodate” much. An additional reason for the low occurrence of style shift might lie in the nature of televisual data as such. A television series often features only a limited number of settings, topics, and participants (Quaglio 2009). The main characters of Gilmore Girls, for instance, are not very socially diversified, and when we consider the potential contexts in which Lorelai’s dialogue with the investigated characters takes place, there are many similarities in terms of power/status, contact/solidarity, ethnicity, and the type of interaction (casual, spoken). Further, all dialogue is similar in terms of being a scripted version of casual conversation, which is designed for the target audience (see Chapter 1). In terms of this audience design, a stable character arguably helps viewers to identify with television characters, creating viewer loyalty.



Chapter 3.  Character differentiation and character stability 61

Contemporary television series are frequently character-driven and a stable character adds to serial stability. With respect to Gilmore Girls in particular, producer, creator and writer Amy Sherman-Palladino has stated that “[i]t’s very important that it feel like the same show every week, because it is so verbal. It’s not about car crashes or vampires or monsters or suspense. It’s really about people talking to each other and the way they talk to each other, which is very specific.” (Sherman-Palladino 2005, my italics). To conclude this section, while the original key word/n-gram analyses suggest a relatively high degree of character stability, it is worth exploring in more detail the variation that does occur, both with respect to diachronic and intersubjective stability. Sections 3.4.2 and 3.4.3 revisit both aspects through my analysis of Lorelai’s dialogue using modified WordSmith settings (Bednarek 2011b). 3.4.2 Diachronic character stability re-visited In order to explore character stability in further detail, the p-value was adjusted to 0.05 in a subsequent analysis of diachronic and intersubjective stability. Minimum frequency thresholds were applied to exclude key words/n-grams with a low distribution across the corpus.16 Table 3.6 shows a potential categorisation of the resulting key words and n-grams that distinguish Lorelai’s dialogue in different seasons. In order to classify the items in Table 3.6 clearly and unambiguously, further analyses would be necessary. For instance, while some linguistic items (e.g. please, honey) are classifiable as interpersonal markers, others (e.g. primary verbs, pronouns, names, references to family roles) could have a range of uses. Thus, names and references to family roles (dad) may be terms of address (examples: Mom, I’m 16. The frequency thresholds used for analysing Lorelai’s dialogue across seasons were set as follows: – key words: raw frequency ≥ 50 (about 0.1%) – key 2-grams [clusters]: raw frequency ≥ 40 (about 0.08%) – key 3-grams [clusters]: raw frequency ≥ 20 (about 0.04%) As n-grams occur less frequently than words, the threshold for n-grams was lower. In all searches only the top 50 in terms of keyness (both positive/negative) were analysed, and the following settings were used: – Index: default settings (but minimum frequency = 2, no thorough concordancing, n-gram [cluster] size 2–5); – N-gram (cluster) settings: default (but all words, cluster size first 2–2, then 3–3, minimum frequency: 1, omit any containing #). – Key word settings: default (but p-value set to 0.05 and frequency/keyness thresholds applied) The spelling variants ok/okay were excluded from the analysis.

62

Language and Characterisation in Television Series

Table 3.6  Key words and n-grams in Lorelai’s dialogue across seasons Potential categorisation

Key words and n-grams

primary verb (do, be, have)

is, was, did, didn’t, has, have, don’t, do, to be

(+ human) pronouns [plus primary verb]

she, she’s, he, he’s, her, him, I, me, you, we, they, we’re; I was, you are, I had, I am, I don’t, I have, do you

(− human) pronouns [plus primary verb]

it, it’s, this, this is [that]

names

Dean, Jess, Lorelai, Rory, Jason, Kirk, Luke, Michel, Paul, Paul Anka

reference to family roles

my mother, dad

interpersonal markers (terms of endearment, greetings, politeness markers, intensifiers, markers of agreement, evaluative language, interjections, mental process verbs, discourse markers, negation, wh-questions)

honey, hello, hey, hi, please, could, really, yeah, yes, right, good, have to, have to go, because, oh, [like], wanna, want, want to, I don’t want, thought, know, don’t know, you know, I mean, mean, ugh, uh, no no, why don’t you, why

content (time, place, people, things, actions)

tonight, here, there, now, life, guy, coffee, house, look, come, tell me

other (e.g. prepositions, articles, conjunctions, going to, quantifier, if )

if, at, for, of, the, a, to, into, down, back, gonna, going, going to, not going to, I’m going to, so, [that], one, of the; if you, if I

pregnant!; Rory… this is to commemorate all those articles of ours that you carefully, thoughtfully ripped to shreds) or terms of reference (examples: As I was saying, in recognition of Rory’s hard work and devotion to the daily news, we have a few tokens of our appreciation; Yeah, my mom’s not really into baking or booms). Key words/n-grams such as look, come, tell me can be used in imperatives (Tell me!) or in declaratives (e.g. He wanted to tell me about…). Thus, the classification in Table 3.6 is hypothetical, with only those key words/n-grams categorised as interpersonal markers that seem reasonably and frequently associated with constructing interpersonal relationships between characters. As a detailed investigation of all key words/n-grams is beyond the scope of this case study, especially since some have very high raw frequencies and an analysis based on extrapolation is not ideal, I will only offer some general hypotheses based on Table 3.6, and a selected analysis of some lower frequency content key words. First, the shift in names is presumably related to the focus of particular seasons. For instance, season 1 (Dean positive key word, Luke negative key word) is more focused on the Rory/Dean relationship and how Lorelai deals with it, than the Lorelai-Luke relationship. Similarly, Jess and Dean are positive key words in season 2 because the focus is on the Rory/Dean/Jess development. In season 6, Paul is a positive key word because Lorelai’s dog is called Paul Anka. Such key words arguably

Chapter 3.  Character differentiation and character stability 63

reflect more a change in the narrative focus than a change in the character’s identity itself. However, they may reflect changes in relationships between characters: Rory is a negative key word in season 6 because this is a season where Lorelai and Rory fight, and do not talk as much with each other as in other seasons. The positive key word honey in season 7, where Lorelai and Rory reconcile, may also have to do with their changing relationship (see Section 3.4.3). The change in (+ human) pronouns may also be related to this shift in narrative focus (see Section 3.4.3). That is, certain pronouns will be positive key words in a particular season (e.g. she, her in season 1; we, we’re in season 4), while others will simultaneously be negative key words (e.g. they, we, we’re, he in season 1; I, me, her in season 4), because different characters are referred to more frequently as a result of shifting narrative focus. Further, variants occur as positive and negative key words respectively (e.g. hey positive, hi negative; yeah positive, yes negative; gonna/wanna positive, going to, want to negative). This may have to do with who characters talk to most in a particular season (see Section 3.4.3). In other words, many changes seem to concern not the development of the character itself but rather the development of character relationships. As pointed out in Chapter 1, television series incorporate both stable and changing elements: “stable” characters encounter new and changing environments and happenings, and changes in their relationships with other characters may occur. Other key words may have to do with the nature of the narrative in television series, where what comes earlier in the course of a series is important in establishing characterisation. For instance, coffee is a positive key word in season 1. Of the 64 occurrences, the majority are associated with Lorelai either wanting coffee (Coffee, please; I need (some) coffee (6x); can I just have some coffee; I would love some coffee; please can I have coffee; I’m going to find coffee) or offering it to others. Instances of coffee also include innovative uses, which contribute to characterising Lorelai, such as: – – – – –

Oh finally, the coffee cavalry arrives Coffee hunt This is a coffee morning Oh yeah, I’m Mrs coffee. … Luke’s where the happy coffee is

As noted in Section 3.3.1, Lorelai’s taste for coffee is important for constructing her identity. That coffee occurs more frequently in season 1 than in the following seasons does not indicate a change in Lorelai’s character. Rather, early episodes and series are particularly important in establishing characters and their behaviour (see Chapter 2). This set-up of character need not be reinforced to the same extent in later seasons. Summing up, the identified key words seem to reflect shifts in narrative focus, changes in character relationships, and the serial nature of the televisual narrative.

64 Language and Characterisation in Television Series

3.4.3 Intersubjective stability re-visited We will now return to and explore intersubjective stability in more detail, again adjusting the WordSmith settings so that they are less strict. Table 3.7 shows a potential categorisation of the top 50 (according to keyness) positive/negative key words/n-grams with respect to intersubjective stability, with a p-value of 0.05 and a minimum frequency of 0.1%.17 Table 3.7  Key words and n-grams relating to intersubjective stability Potential categorisation

Key words and n-grams

Content (time, place, people, things, actions)

business, woman, dinner, work, ask, tell, to go, come, [there], here, night, time, home, day, food, inn, the inn

Names

Chris, GiGi, Christopher, Sherry, Sookie, Jackson, Rory, Rory’s, Dean, Yale, Luke

Reference to family roles

Mom, dad, you and dad

(+ human) pronouns [plus primary verb]

she, you’re, she’s, her, they, we, he, he’s, him, I’ll, your, you, my, you don’t, I, you are, we’re, I don’t, he was, and he, he was, are you

Interpersonal markers (terms of address/endearment, greetings, politeness markers, intensifiers, markers of agreement, evaluative language, interjections, mental process verbs, discourse markers, negation, wh-questions)

mom I, mom you, mom this is, dad I, hi mom, hey dad, hi dad, honey, hello, hey, please, can, very, all, [so], really, just, yeah, yes, right, sure, I’m sure, I love, I love you, I have to, I got to, because, need, I need, fine fine fine, great, good, oh, I want, I want to, don’t want to, want, wanted, and I know, I know that, you know, mean, I mean, well, um, uh, not, no, what, what’s, why, when

Other (pronouns [-human], prepositions, articles, conjunctions, going to, quantifier), also including primary verbs (do, be, have)

this, that, it’s, that’s something, at, a, and, the, going to, gonna, for, five, five thousand dollars, seventy five thousand, and a half, the, an, if, [so], one is a be, did, is a, got, had

17. As characters have more or less talking time, the corpora differ in size, which means that proportional (rather than raw) frequency thresholds were used: – – –

key words: proportional frequency ≥ 0.1% (10 occurrences per 10.000 words); key 2-grams [clusters]: proportional frequency ≥ 0.08%; key 3-grams [clusters]: proportional frequency ≥ 0.04%.

As this corresponds to a low minimum raw frequency for small corpora, a wide distribution across corpus files (episodes) is not entirely ensured for these sub-corpora (Lorelai-Richard and Lorelai-Chris dialogue).

Chapter 3.  Character differentiation and character stability 65

What has been said about the classification of key words/n-grams in Table 3.6 also applies to Table 3.7, with linguistic items needing more investigation and certain items being ambiguous. Again, I will only briefly comment on some of these key words/n-grams, and I will not discuss those in the final row (“Other”).

Content words Content words may often concern typical topics of talk between certain characters, for instance dinner in Lorelai-Emily dialogue (Lorelai and Rory spend every Friday evening dinner with Emily and Richard); business, work in Lorelai-Richard dialogue (Richard runs his own business) or food, (the) inn in Lorelai-Sookie dialogue (Sookie is a chef, who co-owns an inn with Lorelai at some stage). Such content words can be related to different character traits such as occupation or areas of expertise and interest. Names It is not surprising that names (whether used as terms of reference or terms of address) are key words in Lorelai’s dialogue with individual characters. For instance, when talking to Christopher, Lorelai’s key words include his name Chris/ Christopher, his (other) daughter’s name GiGi and his girl-friend’s name Sherry. This is simply because Lorelai talks to Christopher and addresses him by name, and because she is more likely to talk with him about his daughter/girl-friend than with other characters. Reference to family roles Family or kinship terms (mom, dad) can be used to address characters (classified as interpersonal marker: term of address) or be a reference to family roles. The key n-grams mom I, mom you, mom this is, dad I, hi mom, hey dad, hi dad, however, clearly point to use as term of address and have thus been solely classified as interpersonal markers. While mom, dad are ambiguous in this way, the 3-gram you and dad points to simultaneous usage as term of address and term of reference. More specifically, it is used by Lorelai to address her mother (Emily) and refer to her father (Richard), though sometimes Richard may be present and thus also be indirectly addressed (see Figure 3.1, concordance line 4). Both mum and dad are positive key words only in Lorelai’s dialogue with her parents – not surprisingly, as she is likely to either address them with mom/dad or to mention one to the other. Thus you and dad is a positive key n-gram in Lorelai’s dialogue with her mother.

66 Language and Characterisation in Television Series

Figure 3.1  Concordance lines for you and dad

(+ Human) pronouns The frequency of certain pronouns presumably relates to the referential use of names and family/kinship terms discussed above, as third person personal pronouns are typically employed to anaphorically refer to already introduced participants. For instance, she, she’s and her are positive key words in Lorelai’s dialogue with Christopher where female names are also positive key words. Similarly, Lorelai’s dialogue with her best friend Sookie features the key words/n-grams he, he’s, him, and he, he was, which seems to suggest that they often talk about male characters, with the n-grams and he and he was simultaneously indicating a narrative/story. The telling of stories has frequently been connected to involvement or friendship, and the telling of (inter)personal matters seems to occur frequently in female friendship, including the talking through of crises or troubles with partners (cf. Bubel 2006, 2011). The key words/n-grams with the male singular pronoun may thus work to construct Lorelai and Sookie as traditionally heterosexual female friends. In summary, names, references to family roles, and pronouns give insight into the personal relationships between Lorelai and other characters by reference to who she talks to and about. Interpersonal markers Interpersonal markers include a variety of devices that are used to negotiate relationships with others (terms of address/endearment, greetings, politeness markers, intensifiers, markers of agreement, evaluative language, interjections, mental process verbs, discourse markers, negation, wh-questions). Many of these can reasonably be linked to alignment practices (establishing dis/affiliation between speakers), e.g. discourse markers, accounts, orientation to affective stance, self-disclosure,



Chapter 3.  Character differentiation and character stability 67

terms of address/endearment, questions (Bubel 2006). Evaluative and emotional language further reflects the close relationships between characters in Gilmore Girls. It is not surprising that the majority of key words and n-grams for intersubjective stability fall into this category. That is, interpersonal markers are crucial in characters positioning themselves with respect to other characters because it is their linguistic function to do so. Among the identified interpersonal markers I will briefly discuss terms of address, as we have already encountered them above where we have seen that names and kinship terms can be used as terms of address. Together with terms of endearment (e.g. honey), they can be classified as familiar terms of address (Bubel 2006: 153). In fiction, terms of address “can be an important means of signalling social information” (Culpeper 2001: 193), indicating the social relationships between characters. In her conversational analysis of Sex and the City, Bubel (2006) shows how televisual characters exhibit different frequency patterns in their use of first name address and endearments. This also concerns “the recipients of these forms of direct address” (Bubel 2006: 165). This is also the case in Gilmore Girls, as the corpus linguistic analyses show. We have already seen this with names/references to family roles and will now briefly look at honey as a term of endearment. Endearments typically indicate an intimate relationship between characters and are used by “close female family members, sexual partners and ‘favourite’ people” (Culpeper 2001: 193). Honey is a key word in Lorelai’s dialogue with Christopher (Rory’s father) and with Rory, and shows the special relationship between Lorelai and Rory. Of the 275 occurrences of honey in Lorelai’s dialogue, 141 are addressed to Rory. Figure 3.2 shows ten selected examples.18

Figure 3.2  Ten concordance lines for honey

18. As a reminder, this sub-corpus does not comprise Rory’s dialogue or speaker names, rather it includes only Lorelai’s language addressed to Rory. For instance, the line No, honey, that was a joke is a response to Rory’s question You didn’t … with the principal, did you? (episode 1.01, Pilot).

68 Language and Characterisation in Television Series

A closer examination of the concordance lines shows that honey often occurs in the co-text of other features related to intimacy, such as expressions of emotion (oh, honey – 22 occurrences); familiar greetings (hi, honey; bye, honey; night, honey); other familiar terms of address (Rory, honey; sweetie, baby, cookie, honey) or positive evaluations (that’s fantastic honey; very pretty honey; good thinking honey). That is, it frequently works as an affiliative practice for Lorelai and Rory. A familiar term of address, even a term of endearment, can also be combined with a disaffiliative practice such as a command, signalling affiliation at the macro rather than the micro level (Bubel 2006: 161–162). This also occurs in the Gilmore Girls data, e.g. a co-occurrence of honey with imperatives or disagreements. In sum, the analyses of Gilmore Girls suggest that terms of endearment are used in television dialogue both for constructing characters’ individual style (certain characters prefer particular terms of address), and for constructing relations between characters. One further point to be noted is that some of the key words seem to differ in terms of in/formality (e.g. hello vs. hi, hey; yeah vs. yes). For instance, the colloquial hey and great are negative key words in Lorelai’s dialogue with her “proper” and older parents Emily and Richard, and hello is a positive key word in her dialogue with Emily; you are (more formal than the contracted you’re) is a negative key word in Lorelai’s dialogue with her daughter Rory; very is a negative key word in her dialogue with friend/partner and diner owner Luke. On the other hand, yeah, sure and you know are, surprisingly, negative key words/n-grams in Lorelai’s dialogue with Rory. Further research would be needed to explore this inconsistency and to investigate the role of in/formality in characterisation further (see Culpeper 2001: 182–188). In summary, and discounting possible transcription, corpus, software, or statistical errors, we have seen that Lorelai is constructed as a relatively stable character in terms of diachronic and intersubjective style shift. In particular, while there is some intersubjective variation in terms of topics and interpersonal positioning, there is much less of a diachronic shift in Lorelai’s dialogue across seasons, and that which occurs reflects changes in her relationship with other characters, shifts in narrative focus, or can be linked to the nature of Gilmore Girls as a serial television narrative. This case study therefore confirms assumptions made on the nature of characters in television series in general, namely that television characters are relatively stable, and that they do not change drastically, although some biographical and personal development is possible.

Chapter 3.  Character differentiation and character stability 69

3.4.4 Revisiting stability: The Gilmore Girls reboot When I originally undertook the case study of Lorelai as a stable character (Bednarek 2011b), I operated under the assumption that the series (and corpus) was fully representative of the narrative. However, as mentioned above, Gilmore Girls: A Year in the Life was released on Netflix in late 2016 as a so-called “reboot” or “revival” (comprising four 88-to-102-minute episodes). This provides us with a unique opportunity to revisit stability in televisual characters. Rory, 16 at the start of the original Gilmore Girls series is now 32 (Lorelai’s age then), and Lorelai is correspondingly in her late 40s. The question is whether Lorelai’s (and Rory’s) language in the reboot is still similar to the original, indicating character stability despite the characters’ changes in age. Reichelt (2019) in fact addressed this question by investigating stability and change with a focus on particular linguistic features: Using a corpus of transcripts containing the dialogue of individual characters, she first examined the characters’ normalised frequencies of the hedge kind of (Figure 3.3) and of the discourse marker I mean (Figure 3.4). I will focus on Lorelai and Rory in my summary of Reichelt’s findings, although she also considered Emily. series Gilmore Girls A Year in the Life

kind of per 1000 words

0.6

0.4

0.2

0.0

Rory

Lorelai character

Figure 3.3  The hedge kind of in Rory’s and Lorelai’s dialogue across original and reboot (Figure courtesy of Susan Reichelt)

70 Language and Characterisation in Television Series

series Gilmore Girls A Year in the Life

I mean per 1000 words

1.5

1.0

0.5

0.0

Rory

Lorelai character

Figure 3.4  The discourse marker I mean in Rory’s and Lorelai’s dialogue across original and reboot (Figure courtesy of Susan Reichelt)

Reichelt argues that Rory’s reduced use of the hedge kind of in the reboot (evident in Figure 3.3) likens her style to Lorelai’s in the original series, reflecting the character’s aging. In contrast, Lorelai’s use of the hedge remains fairly stable across original and reboot. Rory’s use of the discourse marker I mean in the reboot is also similar to Lorelai’s use of this discourse marker in the original series. However, Lorelai’s use of I mean is significantly reduced in the reboot rather than remaining stable. Both Figure 3.3 and Figure 3.4 are simply based on average frequencies. To investigate the stability of Lorelai’s use of discourse markers in more detail, Reichelt therefore examined discourse values for three discourse markers (you know, I mean, like). These values describe “the ratio (in percentage) of pragmatic functions a linguistic form has with reference to their grammatical function” (Reichelt 2018: 216), i.e. a d-value of 100% means that a marker is “used exclusively in pragmatic contexts” (Reichelt 2018: 217). The value thus shows whether discourse markers “expand in use and contexts over time” (Reichelt 2019). Figure 3.5 shows Lorelai’s d-value and indicates that her relative use of the three markers is in fact fairly stable across the years, with only minor variation. Reichelt (2018) links this minor variation to the special nature of initial episodes for establishing characters and to a “presumed increase in shared knowledge between characters and audiences” (Reichelt 2018: 247), as episodes continue. In the case of the reboot, the results appear to be somewhat inconsistent, with a lower d-value for I mean but a higher d-value for like and a similar d-value for you know. In any case, we can see that the values for the reboot are not radically different to the values for the original series.

Chapter 3.  Character differentiation and character stability 71

item I mean you know like

d-values for pragmatic markers

1.00

0.75

0.50

0.25

0.00 season 1 season 2 season 3 season 4 season 5 season 6 season 7

A Year in the Life

Figure 3.5  D-values for pragmatic markers in Lorelai’s dialogue (Figure courtesy of Susan Reichelt)

Together, Reichelt’s (2019) results show that when there is a significant change in the age of characters – as in the case of reboots or revival long after the end of a series – this change may be reflected linguistically. This may especially be the case when the character shifts from being a teenager/young woman to being in her thirties or older. The linguistic changes, then, seem to be partially related to indexing age (type indexicality). At the same time, Reichelt’s results also suggest that stability of language use remains important for television characters. 3.5

Conclusion

When combined with scene-based analysis, corpus linguistic character-based analyses can potentially illuminate Bal’s (1997) principles of repetition (repeated characteristics), accumulation (of different characteristics), character-relations and transformation (changes in character). Thus, key word analysis was used to investigate the repeated linguistic behaviour of characters (repetition), and to identify the different linguistic key words/n-grams and traits that distinguish characters (accumulation). Both case studies suggested that key word analysis can provide insights into character-relations, while the frequency comparison of the original/ reboot indicated a potential character transformation (relating to maturity/age). Case study 1 also showed how a corpus analysis can provide information on the norms that are set up in texts, and how the breaking of such norms can be interpreted

72

Language and Characterisation in Television Series

by characters as a change in interpersonal relationships. In addition, case study 2 showed the insights that can be gained into character stability/transformation through key word analysis – both when considering diachronic and intersubjective stability. However, there are clear limitations to this corpus linguistic technique, as it will only identify any potential linguistic shifts that are associated with changes in the frequencies of key words and n-grams. Similarly, key word analysis will not identify all aspects of a televisual character, such as Lorelai’s tendency for joking remarks and intertextual references. Nevertheless, the results showed evidence for the stability of language use in televisual characters across the different seasons of a series and also indicated some of the scope for changes. Finally, we have seen that the existence of reboots or revivals of older series offers room for further linguistic studies into language change and stability. In addition to Gilmore Girls, many other series have also been “rebooted” – for instance The X-files, Full House, Will & Grace, Dexter: New Blood, and Sex and the City (And Just Like That) to name but a few. A comparison with movie remakes (Zago 2016) would be a further possibility.

Chapter 4

Characters and stereotypes

4.1

Introduction

This chapter examines televisual characterisation in relation to the interaction between character dialogue and audience stereotypes. Regarding the various concepts introduced in Chapter 1, the chapter mainly draws on the concepts of dialogue cues and investigates how they align with social schemas/stereotypes. More specifically, the chapter examines how the dialogue of the character Sheldon in the US sitcom The Big Bang Theory (Lorre & Prady 2007–2019) invokes a nerd/geek schema – i.e. a nerd stereotype – thereby constructing shared common ground with the audience.19 From the point of view of indexicality, it could be said that the relevant character dialogue indexes a nerd/geek persona. This is a social persona that “has been well represented in the US media since the 1980s” (Queen 2015: 172), and nerds are common character types in television sitcoms (Ankerstein 2019: 55). The focus on a sitcom is also justified because this genre highlights the relationship between fiction and the real world, regarding the production and negotiation of social and cultural meanings, values, and concerns (e.g. Murphy & Palma-Fahey 2018: 297–298). Regarding the techniques introduced in Chapter 2, this chapter uses concordance analysis, key word analysis, and frequency analysis. The unit of analysis is the season, with the focus mainly on the first season of The Big Bang Theory, given the importance of early episodes for establishing characters (see Chapter 2). Later seasons will be examined for occasional comparison. The case study combines different levels of analysis, including character-based analysis (contrasting Sheldon’s language usage with that of other characters) and scene-based analysis (Sheldon’s language usage in one interaction spanning several scenes). The scene-based analysis will draw on impoliteness theory (Bousfield 2008b). Again, comparison with the television landscape is incorporated through analysis of SydTV. Concepts and frameworks that were not explained in Chapters 1 and 2 will be introduced where relevant below (e.g. impoliteness). The next section introduces the series and 19. I use the term invoke rather than activate as I do not want to make any assumptions about viewers’ cognitive processes – as mentioned in Chapter 1, I am not fully adopting Culpeper’s (2001) model of characterisation.

74

Language and Characterisation in Television Series

corpora used in this chapter. Section 4.3 then presents relevant nerd stereotypes/ schemas, while Section 4.4 introduces the methodological details and Section 4.5 presents and discusses the results. 4.2

The Big Bang Theory

The Big Bang Theory is a recent US American sitcom which ran for 12 seasons, from 2007 to 2019; my focus here is on the first season. This season centres on two young physicists, Sheldon and Leonard, their scientist/nerd friends Howard Wolowitz and Raj Koothrappali, and Sheldon and Leonard’s new neighbour Penny, who wants to be an actress but starts out by working as a waitress. Other female characters become important in later seasons. A main source of humour is the contrast between these very different characters. As Queen explains, “Penny, who is fundamentally juxtaposed against the others as a non-nerd […] provides the relational component to the cast that helps foreground the nerd identities shared by the others” (Queen 2015: 175). The sitcom has been very successful both in terms of industry awards and audience figures. The actor Jim Parsons, who plays Sheldon, has been particularly successful in winning several industry awards for his portrayal of the character. Sheldon is also considered by many as the most noteworthy and popular character in the show (http://en.wikipedia.org/wiki/Sheldon_Cooper). It is for these reasons that the case study will focus on Sheldon, that is, the dialogue assigned to Sheldon will be compared to all other characters’ dialogue (character-based analysis). Using transcripts that were created by the same fan transcriber (retrieved from http://bigbangtrans.wordpress.com), three corpora were compiled: – – –

BBT: a 53,127 word corpus of transcripts for all 17 episodes in season 1 (including descriptions of settings, character actions and gestures, etc); SHELDON: a 14,896 word corpus of Sheldon’s dialogue only, extracted from the corpus above (excluding his name, scene descriptions etc); OTHERS: a 32,643 word corpus of dialogue by all other speakers, extracted from the corpus above and including speaker names but no scene descriptions etc.

The two sub-corpora containing the dialogue by Sheldon and by the other characters enable character-based analyses. In addition to these first-season-corpora, I occasionally make use of fan transcripts from later seasons (seasons 2–9) from the same website, which were similarly compiled into three different datasets: BBT 2–9 (like BBT including dialogue and descriptions) and two character-based datasets

Chapter 4.  Characters and stereotypes 75



that contain the relevant extracted dialogue of Sheldon and of the other characters, respectively.20 I also draw on SydTV to compare Sheldon with televisual characters in series other than The Big Bang Theory. These additional corpora and analyses were not available in the original study (Bednarek 2012). 4.3

Nerd stereotypes/schemas

As noted in Chapter 1, televisual dialogue is designed for the (target) audience (“audience design”). This point is relevant to this case study in various ways: in connection with the parasocial relationships that viewers can form with televisual characters; in relation to how viewers might interpret dialogue in terms of identity; and with respect to schemas or stereotypes. Following Culpeper (2001: 79) and Richardson (2010a: 178) schema and stereotype can refer to the same phenomenon, with the term stereotype carrying pejorative connotations. Culpeper and Fernandez-Quintanilla (2017: 99) explain that they use the term social schemata for the type of knowledge that is elsewhere referred to as cognitive stereotypes.21 Schubert (2018: 8) notes that stereotypes may have a long history in a particular country and constitute cultural models. Whether they are called stereotypes, social schemata or cultural models, such mental representations for social groups or personae are part of the relevant common ground that screen creatives aim for (e.g. Bubel 2008). In the first season of The Big Bang Theory, the most relevant social identity or persona at stake seems to be that of the nerd or geek. But what nerd/geek stereotypes or schemas do audience members have or are familiar with? There are several ways in which we can find out about potential features of such stereotypes/schemas, including dictionary definitions, academic research, and collocation information from corpora (Culpeper 2001: 265–268). I drew on a combination of dictionary entries from the Oxford English Dictionary, collocates in the Corpus of Contemporary American English (Davies 2008), Wikipedia entries, Google images, and related websites as well as academic literature. These resources suggest that Western audiences (and media) stereotypically associate geeks and nerds with the following traits:

20. Results from all character corpora are dependent upon the accuracy of the dialogue separation by research assistants and students (see Acknowledgments), which has not been checked manually. 21. An alternative cognitive approach to stereotype analysis makes use of prototype theory (see Schubert 2017b, 2018).

76

Language and Characterisation in Television Series

– intelligent, studious; – an interest in, obsession with, or knowledge of, all things technological or scientific, especially as relating to computers; – an interest in sci-fi and fantasy and related activities; – socially inept/awkward, loners/outsiders, reclusive, unsociable, having only online friends, often socially isolated or ridiculed, no conversational skills; – unattractive, e.g. in terms of weight (either very skinny or overweight), with glasses, weird clothing; – frequently white males (note the specific term nerd girls to describe female nerds and see Bucholtz 1999; Inness 2007 for further discussion of nerds/geeks and gender); – physically awkward or unfit, uninterested in sports; – sexually inactive/virgins. Nerds are also frequently shown as young (e.g. in the high-school context) and they can be linked to the autism spectrum. By way of illustrating some of these stereotypical aspects, compare Figure 4.1 and Figure 4.2 – the first images to come up when using Google Images to search for “nerd” (searches undertaken 16 February 2011 and 14 August 2021). I will draw on these stereotypical associations with nerds when discussing my analyses of Sheldon below.

Figure 4.1  Google images for “nerd” (16 February 2011)

Figure 4.2  Google images for “nerd” (14 August 2021)



Chapter 4.  Characters and stereotypes 77

It must be noted that while nerds/geeks are frequently negatively evaluated, the term can be reclaimed by speakers, and nerds/geeks are also associated with “economic fame and fortune” (Inness 2007: 4). Bucholtz argues that nerds “are not socially isolated misfits, but competent members of a distinctive and oppositionally defined community of practice” (Bucholtz 1999: 211). However, we need to make a distinction between the identity category of “nerd” as claimed by real-life participants, and the category or persona of the “nerd” as constructed in the media. What we are dealing with here is not a set of resources that is available to participants as a style that can be used to claim membership in a particular oppositional community; rather, we are dealing with scripted and performed identities that are presented to viewers as a particular construction of identity. 4.4 Methods 4.4.1 Concordance and key word analysis Concordance analysis was applied to both SHELDON and to OTHERS in order to explore explicit or implicit dialogue cues that characterise Sheldon. Examples for explicit cues from the corpora are: I’m a genius (self-presentation in Sheldon’s dialogue) and Sheldon is batcrap crazy (other-presentation in Leonard’s dialogue). Implicit cues only occur in the respective character’s dialogue (i.e. in SHELDON). An example for such an implicit cue is the utterance Actually, I don’t need a team, I could easily defeat you single-handedly, which points to Sheldon’s belief in his own superiority (in terms of conversational behaviour, face work or impoliteness). In order to investigate the construction of Sheldon’s character in both Sheldon’s and other characters’ dialogue, concordances for I and I’m in SHELDON and for Sheldon* in OTHERS were produced. Concordance lines were explored for the information they provide into Sheldon’s character in relation to character traits and their associations with the nerd stereotype. This analysis therefore focusses mainly on the content or function of these concordances rather than their linguistic form. In addition, key words (or “keyness”) analysis was undertaken to determine what n-grams Sheldon uses more frequently than other characters. For this analysis, SHELDON (node corpus) was compared with OTHERS (reference corpus), retrieving key n-grams (2-grams, 3-grams, 4-grams) with a minimum frequency of three and a p-value smaller than 0.05 (log likelihood statistics). To do full justice to all generated key n-grams is not possible within the scope of this chapter, which is therefore limited to pointing out some particularly relevant results. In addition to comparing Sheldon with other characters within the first season of The Big Bang Theory, I also newly compared Sheldon with televisual characters

78

Language and Characterisation in Television Series

outside his own narrative universe. More specifically, I undertook a key word analysis where SydTV was used as a reference corpus against which the dialogue of Sheldon (and Lorelai from the previous chapter) was compared. The details are presented in Section 4.5.4 below. Finally, going beyond Sheldon to get an overview of the stereotypical associations with nerds that are presented to viewers in The Big Bang Theory I undertook a new concordance and frequency analysis of the related identity labels NERD, GEEK and DORK, both in season 1 transcripts and in transcripts of later seasons. All caps are used to indicate that I do not just focus on the particular form but also include derived and inflected nouns (e.g. nerdvana) and adjectives (e.g. nerdy). In relation to Culpeper’s model (see Chapter 1), I would consider the use of these labels as explicit cues, since they are membership categorisation devices (in Conversation Analysis parlance) that directly relate to the “nerd” identity at stake. Bucholtz and Hall (2005) argue that the overt use of social category labels is the “most obvious and direct way that identities can be constituted through talk” (Bucholtz & Hall 2005: 594). The use of NERD category labels functions to classify someone or something as directly associated with “nerdiness”, although further details about the specific associations with the nerd persona may still need to be retrieved or inferred by viewers from the co-text and context. Examination of labels has been shown to be fruitful in other analyses of stereotypes in television series (e.g. gender stereotypes in Gregori-Signes 2017; Murphy & Palma-Fahey 2018). 4.4.2 Scene-based analysis Corpus linguistic concordance and key word analysis were complemented by scene-based analysis. As noted above, one aspect of nerd stereotypes concerns nerds’ supposed social awkwardness or lack of social and conversational skills. To investigate how this is linguistically constructed in particular scenes which consist of an interaction between Sheldon and Penny across different locations, I draw on analysis of “face-aggravating behaviour” (Locher & Bousfield 2008: 3), which can be linked to impoliteness as well as (lack of) pragmatic competence. Culpeper (2001: 248) suggests that impoliteness is particularly important for characterisation, while Reichelt (2020a: 167) argues that pragmatic competence is often a productive trait in character styling. Dynel (2017: 468) mentions the popularity of television characters whose dialogue features impoliteness. Eitelmann and Stange (2016) use The Big Bang Theory dialogue for introducing, explaining and illustrating various pragmatic principles, pointing out Sheldon’s “considerable problems with politeness conventions” (119). One of the disagreements or debates in impoliteness theory concerns the role of intention (for discussion in relation to telecinematic discourse, see Dynel



Chapter 4.  Characters and stereotypes 79

2017: 470–475). Thus, Bousfield (2008a) and Culpeper (2008) only count intentional face-aggravating behaviour as impoliteness, whereas Terkourafi (2008) sees impoliteness as unintentional, in contrast to (intentional) rudeness (which Culpeper sees as non-intentional). To avoid such terminological obstacles, I will not specifically talk about impoliteness or rudeness, but examine an interaction between Sheldon and Penny with a focus on non-cooperative behaviour on Sheldon’s part, more specifically his face-aggravating behaviour or “inappropriate verbal behaviour” (Garcés-Conejos Blitvich 2010: 550). From this perspective, face-aggravating behaviour concerns the omission of appropriate or the addition of inappropriate moves (Terkourafi 2008: 60). In the context of characterisation, such face-aggravating behaviour constitutes implicit cues to characterisation. Its importance for fictional characterisation has been noted by many researchers (e.g. Culpeper 2001: 247, 251) and it features in other analyses of television characters (e.g. Bruti 2018; Dynel 2012; Mandala 2011; Richardson 2010a, 2016; Sorlin 2016). To analyse Sheldon’s face-aggravating behaviour, many different analytical approaches could be adopted (see Garcés-Conejos Blitvich 2010). One key distinction is that between first- and second-order approaches (Locher and Bousfield 2008). The former approaches focus on participant assessments of what is face-aggravating; the latter focus on analyst assessments of what is face-aggravating (based e.g. on pragmatic theory). I use a combination of both: On the one hand, I draw on Bousfield’s (2008b) categorisation of impoliteness realisations to analyse the interaction between Sheldon and Penny (second-order approach). On the other hand, I also take into account how Penny takes up Sheldon’s behaviour (her reactions) in order to help determine whether or not Sheldons’ behaviour is face-aggravating (first-order approach). This includes her (emotional) reactions as well as her own increasingly face-aggravating behaviour. Concerning the second-order approach, I classify Sheldon’s contributions according to the impoliteness realisations in Bousfield (2008b: 99–143, based on Culpeper 1996). Such realisations include condescending, scorning or ridiculing the hearer, using obscure or secretive language (e.g. jargon), hindering/blocking the speaker (e.g. interrupting), or invading the other’s space (e.g. speaking about information that is too intimate). Although originally intended to cover only intentional face-aggravating behaviour, this classification has proven to be useful for investigating other types of face-aggravating behaviour (Archer 2008). It has been argued that interactions between characters are an important part of televisual characterisation (Pearson 2007: 45). With respect to The Big Bang Theory, analysing interactions where only Sheldon and Penny interact was seen as particularly illuminating, since Penny is the main “non-nerdy” character that is contrasted with Sheldon and his friends in the first season of the series. Such interactions occur only in seven episodes in the first season, of which the very first interaction

80 Language and Characterisation in Television Series

(in episode 4) was selected for analysis, because early episodes are especially significant for characterisation (see Chapter 2). The specific scenes that make up this interaction will be reproduced in Section 4.5.3 alongside their analysis. 4.5

Results

4.5.1 Character insights from concordance analysis To gain first insights into the construction of Sheldon as a nerd, we will examine the traits that are constructed in first-person statements of the character (self-presentation) derived from the concordances for I and I’m in the SHELDON corpus. While this analysis does not produce a full picture of the character, it does offer a considerable number of character traits, as Table 4.1 demonstrates. This table shows the retrieval of character traits from the content of these concordances in Sheldon’s own dialogue. All concordance lines were examined in their interactional context to clarify what aspects of identity they revealed, but for reasons of space this interactional context cannot be fully reproduced in Table 4.1. According to the character’s own dialogue, Sheldon is highly intelligent (a child prodigy) and believes in his own intellectual superiority, showing arrogance, is different from others in terms of not driving and/or enjoying social activities as well as struggling with social skills. Sheldon is also constructed as slightly health obsessed and as having issues with food/general health. We can further see that the character likes computer-related activities and is an expert in the area. In fact, utterances that indicate Sheldon’s expertise concerning computers and gaming show that while he may struggle with social skills as far as interactions with “non-nerds” (especially Penny) are concerned, he is clearly positioned as someone who has certain skills in the “nerd” community. At other times, however, the character is also seen struggling in his interactions with his “nerdy” friends (Leonard, Raj, Howard). As has become clear, these character traits are partial instantiations of the nerd stereotype, in terms of Sheldon’s intelligence, “difference”, interest in technology, lack of social skills, and physical unfitness. Sheldon’s dislike for change can be linked to obsessive-compulsive behaviour and his arrogance may be tied to his lack of social skills, since this type of behaviour is conventionally frowned upon. His difficulty with reading others’ emotions can be associated with audience expectations concerning people on the autism spectrum. As suggested above, at least parts of the audience would have these associations, considering the traits that Western audiences and media stereotypically associate with nerds. Other character attributes are perhaps less easily tied to “nerd” stereotypes, for instance, that Sheldon does not drive.

Chapter 4.  Characters and stereotypes 81



Table 4.1  Character information in SHELDON22 Character trait

Example dialogue

believes in his own intellectual superiority

They call me a genius because I’m a genius. You see, I’m a superior genetic mutation, an improvement on the existing mediocre stock. Actually, I don’t need a team, I could easily defeat you single-handedly. As I have explained repeatedly, unlike you, I don’t need validation from lesser minds. I have a masters and two PhD’s, I should not have to do this. Alright, but if we’re going to use flight metaphors I’m much more suited to being the guy from the FAA, analysing wreckage. You tell people I’m a rocket scientist? I’m a theoretical physicist.22 Penny, I have an IQ of 187, don’t you imagine that if there were a way for me to have had soup at home I would have thought of it? Don’t be ridiculous. I have no peers. [Preceded by Leonard: No, no. You gave me an explanation, it’s reasonableness will be determined by a jury of your peers.] I don’t guess. As a scientist I reach conclusions based on observation and experimentation I’m taking a sabbatical, because I won’t kow-tow to mediocre minds.

was a child prodigy

Yes, in fact I am the youngest person ever to win it [the Stephenson award]. No, it’s true, I did a series of experiments when I was twelve […] Before that I was in college, and before that, I was in the fifth grade. Not bad, I myself started graduate school at fourteen.

(continued)

22. This is part of a longer interaction, which clearly shows Sheldon’s belief in his intellectual superiority over others, even other scientists: Missy: Sheldon: Missy: Sheldon: Missy: Sheldon: Missy: Sheldon:

Yup, I’m always bragging to my friends about my brother the rocket scientist. You tell people I’m a rocket scientist? Well yeah. I’m a theoretical physicist. What’s the difference? What’s the difference? Goodbye Shelly. My God! Why don’t you just tell them I’m a toll taker at the Golden Gate Bridge? Rocket scientist, how humiliating.

82

Language and Characterisation in Television Series

Table 4.1  (continued) Character trait

Example dialogue

struggles with social skills

What would we talk about? We’ve no overlapping areas of interest I’m aware of, and you know I don’t care for chit-chat. [Preceded by: Sheldon: Hello Penny. Leonard just left. Penny: I know. I want to talk to you.] Oh! I don’t usually pick up on those things [people being upset]. Hunger? Indigestion, I’m sorry I’m really not very good at this [knowing what feeling Leonard is getting].

is different (in relation to social activities or fitting in)

I think a birthday party is a terrible idea. I envy Leonard for growing up without that anguish. Year after year, I had to endure wearing conical hats while being forced into the crowded sweaty hell of bouncy castles, not to mention being blindfolded and spun towards a grotesque tailless donkey as the other children mocked my disorientation. Have I pointed out that I am extremely uncomfortable with dancing, loud music and most other forms of alcohol induced frivolity? It’s about an Indian princess who befriends a monkey, who was mocked by all the other monkeys because he was different. For some reason I related to it quite strongly.

is health obsessed/ has food issues

Anyway, the local cuisine was a little more sausage-based than I’m used to, and the result was an internal blitzkrieg with my lower intestine playing the part of Czechoslovakia. I need to measure my fluid intake and output to make sure my kidneys aren’t shutting down. I shower twice a day and wash my hands as often as I can. Penny, I just want you to know that, you don’t have to live like this. I’m here for you [to clean her place]. When I was a little boy and got sick, which was most of the time […]

has an affinity for and knowledge of computer-related activities

I know, and I do yearn for faster downloads […]

does not like change

I like the hamburgers where we usually have hamburgers, you can’t make the assumption that I’ll like the hamburgers here.

I have a very wide circle. I have 212 friends on myspace. I know everything about this stuff [computers]. Good idea, I need my wrist brace, all this button pushing is aggravating my old Nintendo injury.

No, I sit there. [in response to Penny sitting on “his” seat on the sofa – a recurring topic is that Sheldon always has to sit in the same spot] It [the honorary Justice League of America membership card]’s been in every wallet I’ve owned since I was five. does not drive

Ah, because it’s in Long Beach, and I don’t drive. I don’t drive, and the only things available within walking distance are a Thai restaurant and a gas station.



Chapter 4.  Characters and stereotypes 83

With the help of additional concordance analysis we can investigate how far this characterisation is supported by the dialogue of other characters. Through exploring concordance lines for Sheldon* in OTHERS, it becomes apparent that while not all 190 occurrences provide information on Sheldon’s character, and even though there are only few examples where Sheldon’s character is explicitly described by others (i.e. Sheldon is …), the above characterisation of Sheldon is further confirmed. According to the other characters, Sheldon: – is highly intelligent (a child prodigy): So anyway, we’re eight years old, and Sheldon converts my easy-bake oven to some kind of high-powered furnace. (Missy, Sheldon’s sister); – believes in his own intellectual superiority: Hang on. Sheldon, is proving that you are single-handedly smarter than everyone else so important that you would rather lose by yourself than win as part of a team. (Leonard); – struggles with social skills: Sheldon, you have to let somebody else answer. (Leonard); – is “different”: Not to mention, Sheldon is batcrap crazy. (Leonard); – is health obsessed/has food issues: Penny, you’ll have to excuse Sheldon, he’s a bit of a germophobe. (Leonard) Other traits include being familiar with the sci-fi series Star Trek (Sheldon, what, do I need to quote Spock’s dying words to you. – Leonard) while we can also glean social relationships, i.e. his being Leonard’s and the others’ friend (Guys, let’s remember that Sheldon is still our friend and my room mate. – Leonard). Interestingly, a lot of dialogue concerns other characters explaining social conventions to Sheldon, for instance: – For God’s sake, Sheldon, do I have to hold up a sarcasm sign every time I open my mouth. (Leonard) – Sheldon, do you understand the concept of blackmail? (Penny) – Sheldon, are you going to introduce us? (Leonard) – Sheldon you just can’t dictate… (Leonard) – You know, Sheldon, you don’t have so many friends that you can afford to start insulting them. (Howard) – Sheldon, you can’t be selfish, we all paid for it, so it belongs to all of us. (Leonard) And – a fact that will become important later – Sheldon is shown willing to learn such unfamiliar conventions, as is the case in Example (1):

84

Language and Characterisation in Television Series

(1)

Penny: Sheldon: Penny: Howard: Sheldon: Howard: Sheldon:

Uh, Sheldon, I didn’t see your present. That’s because I didn’t bring one. Well why not? Don’t ask. The entire institution of gift giving makes no sense. Too late. Let’s say that I go out and I spend fifty dollars on you, it’s a laborious activity, because I have to imagine what you need, whereas you know what you need. Now I can simplify things, just give you the fifty dollars directly and, you could give me fifty dollars on my birthday, and so on until one of us dies leaving the other one old and fifty dollars richer. And I ask you, is it worth it? Howard: Told you not to ask. Penny: Well, Sheldon, you’re his friend. Friends give each other presents. Sheldon: I accept your premise, I reject your conclusion. Howard: Try telling him it’s a non-optional social convention. Penny: What? Howard: Just do it. Penny: It’s a non-optional social convention. Sheldon: Oh. Fair enough.  (The Big Bang Theory, season 1, episode 16, “The peanut reaction”)

We can see that concordance analyses are useful in showing us how features of televisual characters are established through their own and others’ dialogue, demonstrating Bal’s (1997) principle of accumulation (of character traits). The characterising effect of these dialogue cues is strengthened through narrative repetition (Lothe 2000: 84), as they are reinforced through self- and other-presentation. Hence, the analyses also demonstrate Bal’s (1997) principle of repetition (repeated characteristics). It has also become apparent that the televisual dialogue draws on mainstream nerd stereotypes and thereby constructs shared common ground with the audience. 4.5.2 Character insights from key word analysis Concordance analysis has proven to be a useful first tool for retrieving constructed character traits; this section will now briefly explore key word analysis to see whether findings align with the results presented so far. Sheldon’s key n-grams do indeed point to similar character traits as those already observed through concordance analysis. For example, there are expressions that relate to science (a nobel prize [r = 3], my research [r = 3]) and/or seem to show him speaking in words that

Chapter 4.  Characters and stereotypes 85



we may associate with formal academic writing rather than casual informal spoken language. As the range values (in brackets, abbreviated as r) demonstrate, several of these occur in two or more episodes: number of (r = 2), a series of (r = 2), the fact that (r = 4), lack of (r = 4), the result (r = 4), the possibility/possibility that (r = 3), your premise (r = 3), in addition (r = 2)

Table 4.2 shows frequencies for these n-grams in the Corpus of Contemporary American English (Davies 2008; at the time of use, over 425 million words of American English, evenly divided between spoken, fiction, popular magazines, newspapers and academic journals). Table 4.2  Frequencies in the Corpus of Contemporary American English Word/N-gram

Spoken

Fiction

Magazine

Newspaper

Academic

number of a series of the fact that lack of the result the possibility possibility that your premise premise in addition

17,663  2,744 16,124  3,543  2,029  3,078  1,300     31    741  2,939

 4,147  2,401  4,861  2,659  1,045  1,357    475      5    118  1,057

19,794  4,984  7,658  5,997  5,012  2,413    863      2    673  7,937

19,150  4,280  6,438  6,408  3,320  2,455    772      1    656  6,699

41,402   61,77 12,276 14,762   6,025   5,780   1,738      3   1,519 21,517

As Table 4.2 shows, with the exception of the fact that and your premise, all these n-grams are most frequent in academic journals and therefore clearly associated with academic discourse. Even the fact that is highly frequent in academic journals, more so than in any other written variety, and the word premise itself is clearly associated with academic discourse. In fact, your premise is so infrequent in spoken American English (raw frequency = 31; relative frequency = 0.034 per 10,000 words) that its occurrence in Sheldon’s dialogue is highly marked (raw frequency = 3; relative frequency = 2.02 per 10,000 words). Interestingly, “real-life” nerds have also been shown to use formal vocabulary (Bucholtz 2011a: 144), as part of using what Bucholtz (2011a: 151) calls superstandard English, which positions them as intelligent (Bucholtz 2011a: 153). Regarding Sheldon’s stance towards “standard” language, Ankerstein (2019) gives several examples where Sheldon makes prescriptive comments on good or proper language use. This includes advice against the use of final prepositions, although Ankerstein shows that Sheldon’s dialogue does in fact feature such instances, presumably because this language feature is not used systematically by screen creatives for his characterisation.

86 Language and Characterisation in Television Series

In any case, the key n-grams that are associated with formal academic writing position Sheldon as both intelligent and as a scientist. We can also find n-grams that seem to point to a tendency for Sheldon to define and evaluate things, for example this is a (r = 6) in Figure 4.3. Similarly, the key n-grams this is not a (r = 3) and if you will (r = 3) can be related to defining or labelling things.

Figure 4.3  Concordance for this is a in SHELDON

All of the above are associated with the stereotypical nerd’s interest in science and Sheldon’s identity as an (intelligent) “scientist”-nerd. Other n-grams seem to point to character traits that are related to Sheldon’s belief in his own superiority (his arrogance), as Table 4.3 illustrates. A further key n-gram is an interesting (r = 2: an interesting turn of events, an interesting development, an interesting thing about), a bigram which might link Sheldon intertextually to television characters such as the android Data and the human/Vulcan Spock in Star Trek who are emotionally stunted, but tend to find Table 4.3  N-grams indicating arrogance in SHELDON (range values in brackets) reference to own opinion

I hardly think so (r = 3)

tendency to give advice or show superior knowledge/ capacity

going to have to (r = 2): You’re going to have to call her; I can see we’re going to have to spell out everything for this girl; And if water is involved we’re going to have to ground the crap out of the thing.

it occurs to me (r = 2)

you might want to (r = 2): And that’s probably just a sinus infection, but it could be sleep apnoea, you might want to see an otolaryngologist. It’s a throat doctor; You might want to speak in a lower register.; Leonard, I’m not expert here but I believe in the context of a luncheon invitation, you might want to skip the reference to bowel movements. to point (r = 4): I was wrong to point it out; I do feel obligated to point out to you; I’ve hesitated to point this out luckily for you (r = 3) of course (r = 12)/well of course (r = 4) once again (r = 3) remind you (r = 3): let me remind you; need I remind you; I must now remind you



Chapter 4.  Characters and stereotypes 87

events fascinating or interesting (Martin & White 2005: 50). In fact, out of 14 occurrences of interesting in season 1, 50% are produced by Sheldon alone, the rest being split between Penny (3), Leonard (2), Howard (1) and Lesley (1). In the first season of The Big Bang Theory, this adjective is therefore more strongly associated with Sheldon than any of the other characters. Some n-grams (e.g. Leonard Leonard Leonard [r = 2]) also show Sheldon’s use of too much repetition in getting other characters’ attention and indicate impatience or unawareness of social conventions (Figure 4.4). Indeed, one of Sheldon’s trademark compulsive actions that gets cemented in later seasons is his tendency to repeatedly knock and, without pausing, say people’s names three times (e.g. three knocks Penny three knocks Penny three knocks Penny). For example, in the transcripts of Sheldon’s dialogue for seasons 2 to 9, the n-gram Penny Penny Penny occurs 58 times across 27 episodes, at least once in each season in Sheldon’s dialogue.

Figure 4.4  Concordance for Leonard Leonard Leonard in SHELDON

As a new follow-up study, I examined whether the season 1 key n-grams discussed above also occur in Sheldon’s dialogue in later seasons. To do so, I used WordSmith’s (Scott 2020) Concord tool to analyse their raw frequency, range (across individual episodes) and distribution across seasons 2 to 9 with the help of the additional fan transcript corpora mentioned above.23 I am not interested here in whether these n-grams are key; rather my interest is in the extent to which they continue to be used across episodes and seasons by Sheldon. Table 4.4 shows the results ordered according to raw frequency.

23. The search was for the n-gram as is, e.g. “my research” except for well of course where I searched for of course and then retrieved all instances of well of course manually and for Leonard Leonard Leonard where I searched for “Leonard Leonard Leonard/Leonard ? Leonard ? Leonard”. This retrieves instances with or without punctuation marks between the name. The same search was also used to retrieve the trigram Penny Penny Penny. Note that in both cases, there could be additional occurrences of the name, rather than this being restricted to the name being repeated exactly three times by Sheldon. No further analysis of contextual usage was undertaken, although one instance of point-to-point was excluded from the results for to point.

88

Language and Characterisation in Television Series

Table 4.4  N-grams across episodes and seasons N-gram of course this is a number of Leonard Leonard Leonard an interesting the fact that lack of going to have to a series of my research once again well of course the result the possibility a nobel prize if you will remind you possibility that I hardly think so your premise to point (out) this is not a it occurs to me you might want to in addition luckily for you

Raw frequency

Range (no episodes)

Seasons

130  36  21  13  11  11  10  10   8   7   7   7   5   5   5   5   4   3   3   2   2   2   2   2   1   0

 91  33  16  11  11  11  10  10   7   7   7   7   5   5   4   5   4   3   3   2   2   2   2   2   1   0

2–9 2–9 2–9 2, 3, 4, 6, 8, 9 2, 4, 5, 6, 7, 9 2, 3, 4, 5, 6, 7 2, 3, 5, 7 2, 3, 4, 5 2, 3, 5, 6 2, 3, 7, 8 2, 3, 6, 8, 9 3, 5, 8 2, 3, 6, 7 2, 3, 4, 8 2, 3, 5, 6 2, 3, 6 3, 4, 8, 9 2, 7, 8 4, 9 3, 4 4, 7 2, 6 2, 4 3 9 0

The results indicate that there is only one n-gram that is no longer used (luckily for you), only one n-gram that occurs in only one episode (in addition), and only two n-grams that occur in only one season (in addition, you might want to). All other n-grams (23 out of 26) occur in at least two episodes across at least two seasons, with 18 of these occurring in at least three seasons. Three n-grams occur in each season included in the dataset (season 2–9: of course, this is a, number of), with a further three n-grams occurring in six of the eight seasons (Leonard Leonard Leonard; an interesting; the fact that). One n-gram occurs in five seasons (once again) and eight n-grams occur in four seasons (lack of; going to have to; a series of; my research; the result; the possibility; a nobel prize; remind you). The vast majority of these n-grams are hence well-distributed and not just linked to individual episodes or seasons, and not restricted in their use to the first season.

Chapter 4.  Characters and stereotypes 89



For additional information, Table 4.5 shows the results of a comparison of seven frequent n-grams (rf ≥ ten; excluding the address term Leonard Leonard Leonard) in Sheldon’s dialogue (season 2–9: 144,387 words) and in the dialogue of other characters (season 2–9: 367,818 words).24 Log likelihood and LogRatio values were calculated using Lancaster’s Log-likelihood and effect size calculator (http://ucrel. lancs.ac.uk/llwizard.html). The LogRatio of the five significant results ranges from about one (twice as common in Sheldon’s dialogue than in others’) to about three (eight times more common). With the exception of the fact that and going to have to, these n-grams appear to be significantly overused in Sheldon’s dialogue across seasons and therefore continue to be important cues for constructing his identity as a (stereotypical) nerd. Taken together, this follow-up analysis provides some evidence for the assumption that (i) the character traits that are introduced in season 1 continue to be important for Sheldon, and (ii) at least some of the same dialogue cues are used for constructing these traits across episodes and seasons. Table 4.5  Frequent n-grams across characters (season 2–9)   of course this is a number of an interesting the fact that lack of going to have to

Raw frequency in Sheldon’s dialogue

Raw frequency in others’ dialogue

LL

LogRatio

130  36  21  11  11  10  10

175  40  13   9  24   3  17

  28.96***  12.51**   16.56***  6.29* 0.18  13.27** 0.99

0.92 1.20 2.04 1.64 0.22 3.09 0.58

One asterisk (*) indicates p