Semiotics and Verbal Texts 2017

Semiotics and
STUDIES IN DISCOURSE
Verbal Texts
POSTDISCIPLINARY
How the News Media

Construct a Crisis
Jane Gravells
Postdisciplinary Studies in Discourse
Series Editors
Johannes Angermuller
University of Warwick
Coventry, United Kingdom
Judith Baxter
Halsecombe House
Minehead, United Kingdom
Aim of the series
Postdisciplinary Studies in Discourse engages in the exchange between
discourse theory and analysis while putting emphasis on the intellectual
challenges in discourse research. Moving beyond disciplinary divisions in
today's social sciences, the contributions deal with critical issues at the
intersections between language and society.
More information about this series at

http://www.springer.com/series/14534
Jane Gravells
Semiotics and Verbal

Texts
How the News Media Construct a Crisis
Jane Gravells
Aston University
Birmingham, United Kingdom
ISBN 978-1-137-58748-0 ISBN 978-1-137-58750-3 (eBook)

DOI 10.1057/978-1-137-58750-3
Library of Congress Control Number: 2016959556
© The Editor(s) (if applicable) and The Author(s) 2017

The author(s) has/have asserted their right(s) to be identified as the author(s) of this work in accordance
with the Copyright, Designs and Patents Act 1988.
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of
illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made.
Cover image © PM Images / Getty

Cover design by Oscar Spigolon
Printed on acid-free paper
This Palgrave Macmillan imprint is published by Springer Nature

The registered company is Macmillan Publishers Ltd. London
The registered company address is The Campus, 4 Crinan Street, London, N1 9XW, United Kingdom
To Molly Schroder
Foreword
I undertook the research for this book to satisfy an interest in the ways
in which language and business interconnect, and in particular to con-
sider the language surrounding companies in crisis. Many books have
been written from a business perspective on crisis communication, but
my concern was about how the “lay” person came to understand crisis
events. It seemed to me that major news events such as disasters and
crises came and went, and that our perception of them changed over
time—not just that we, or the media on our behalf, tired of them and
moved on to the next big thing, although this may be true. But also that
we understood them in a different way with the passage of time, that they
came to represent something other than an agglomeration of events. I
became interested in the BP Deepwater Horizon explosion of 2010 for
its wide-reaching effects including business, environmental, financial and
human outcomes. One of the things the book is “about” is certainly the
Deepwater Horizon explosion and crisis. However, this is not a hand-
book about the language of crisis communication. It explores how we
know what we think we know about crises such as the BP events. So
another thing the book is about is news media representation—how is
a story such as Deepwater Horizon mediated by different publications,
and over time?
Exploring a phenomenon as broad and disjointed as the media repre-
sentation of a crisis called for an ambitious research approach, and so the
vii
viii Foreword
main thing the book is about is considering ways in which we can inves-
tigate a linguistic representation and understand its way of making mean-
ing. This book is importantly about methodology—how, practically, can
we investigate such a fragmented, diverse and unbounded phenomenon
as the media coverage of a crisis? In particular, I was interested in ways of
looking at written coverage. Many tools and approaches were available for
me to do this, and I discuss some of these in Part I, but none were fully
suited to the flexible, emergent and holistic research process I had in mind.
I saw this as an area where an alternative and systematic research option
can be useful, and from the study of semiotics I sought to reclaim some of
the concepts and terminology which have most recently (and effectively)
been more the province of the study of visual and other signs. Barthes’
view that communication makes meaning at different levels—the sign,
the code, mythic meaning and ideology—provided a framework within
which to situate a flexible analysis, and I set out such a framework in Part
II. This section can be read as a “how to” guide to conducting research
within a broad semiotic perspective. Using a wide range of examples from
written news media texts about the BP Deepwater Horizon events, it dem-
onstrates how to conduct a written analysis of those discursive features
which are most relevant to the researcher’s own particular data set. I drew,
secondly, on Peirce’s view that signs can be understood as having iconic,
indexical and/or symbolic form, as an explanatory concept to describe and
interpret the analysis findings, and this is explored in Part III.
These several strands of semiotic theory, media social practice, real-
world illustration and direct linguistic application relate closely to the
concerns of this book series, which takes a postdisciplinary perspective on
the study of language. The work in this book relates to topics of interest
in the study of applied linguistics, media studies and business commu-
nication. It sets out a relationship between theory and practice, moving
from the understanding of a theoretical framework to its practical appli-
cation, to the explication of results through another theoretical approach.
This book would not have been written without Judith Baxter’s invalu-
able input and guidance and I thank her and Johannes Angermuller for
the opportunity to write for this series. Thanks to Chloe Fitzsimmons at
Palgrave Macmillan for her support in bringing the book to print and
to Jonathan Gravells for his reading of the first drafts and his constant
Foreword ix
e ncouragement. Any remaining failings in the book are my own. I am

very grateful to the Spectator for allowing the use of the Bernie cartoon,
which I found to my delight very early in the research process. I began
my work on the BP crisis with a set of four interviews—two in London
and two in New Orleans—with people who had been very close to the
BP events in different ways. My informants generously shared their time
and their insights about the events in the Gulf, allowing me to see the
crisis from perspectives other than those of the media texts I was working
with.
I would like to add a final word about the BP crisis. Over the six years
I have been involved with studying the media language of this crisis, it
could have become too easy to treat it as an abstract, though fascinating,
case study: an example of the multifarious ways in which we use language
to fulfil a social need. Indeed, my contention is that our understanding
of the crisis is, in one sense, a construction of language. Yet the explo-
sion had tragic material consequences—11 people lost their lives and
17 were injured. The spill caused significant damage to wildlife as well
as to the jobs and welfare of countless working people, particularly in
the industries of fishing and tourism. As part of my research I visited a
senior academic in marine environmental sciences at the University of
New Orleans, and a journalist who had covered the story from the first
day for a New Orleans newspaper, and I was left in no doubt about the
human and environmental cost of the events of April 2010. I hope I treat
these stories with respect in this book.
Alrewas
2016 Jane Gravells
xi
Contents
Part I Written language and semiotics1
1 Researching the Representation of a Crisis3
2 Semiotic Discourse Analysis27
Part II A Barthesian conceptualisation of written language43
3 Theoretical Foundations 45
4 Data Collection and Research Principles 75
5 A Barthesian Analysis of the BP Data in Four Stages 83
6 Stage 1: Contextualisation of the BP Texts 89
7 Stage 2: Preliminary Analysis of the BP Texts101
xiii
xiv Contents
8 Stage 3: A Depth Analysis at the Level of the Sign 111
9 Stage 3: A Depth Analysis at the Level of the Code127
10 Stage 3: A Depth Analysis at the Level of Mythic Meanings167
11 Stage 3: A Depth Analysis at the Level of Ideology 179
12 Stage 4: A Holistic Analysis of a Single Text191
Part III A Peircean conceptualisation of written language199
14 A Peircean Interpretation of the BP Data213
Part IV Concluding thoughts241
15 Other Events, Other Contexts243
References249
Index265
List of Figures
Fig. 2.1 Perspectives and tools of written text analysis 28

Fig. 3.1 A semiotic heuristic for considering written language 48
Fig. 3.2 Barthes’ and Fairclough’s views of language in context 72
Fig. 6.1 A genre categorisation of news texts 93
Fig. 7.1 Nine linguistic features for analysis 109
Fig. 9.1 Notional progression of text types in BP data 141
Fig. 9.2 Texts embedded intertextually “In Too Deep:
BP and the Drilling Race that Took it Down”
(M2 PressWIRE, 27.4.2011) 152
Fig. 13.1 Distance of sign forms from the object 208
Fig. 14.1 A linear view of iconic, indexical and symbolic phases 236
xv
List of Tables
Table 4.1 Sample of BP-related texts from Nexis UK database 79

Table 5.1 Four stages of data analysis 87
Table 6.1 Geographical source of items mentioning BP events 2010–12 90
Table 6.2 BP oil spill texts by genre 2010–12 93
Table 6.3 Proportion of media text dealing directly with
BP oil spill 2010–12 98
Table 8.1 Analysis of naming terms for the BP
Deepwater Horizon events 113
Table 8.2 Social actors in 2010–12 BP texts 116
Table 9.1 Instances of modality in the 2010–12 texts 155
Table 9.2 Summary of findings—modality and appraisal 165
Table 10.1 The occurrence of metonyms in the 2010–12 BP texts 168
Table 10.2 The occurrence of metaphors in the 2010–12 BP texts 174
Table 14.1 Summary of findings from depth analysis 214
Table 15.1 A comparative study by political party 245
Table 15.2 A comparative study of a political issue over time 246
Table 15.3 A comparative study by mediums 246
xvii
Part I
Written Language and Semiotics
1
Researching the Representation
of a Crisis
A Semiotic Account of a News Story

In this book, I make a broad claim about news media representation. I
suggest that whole sets of texts, whole representations of events, have lin-
guistic commonalities which can be investigated. In this view, the linguis-
tic picture constructed by the media immediately after an event is quite
different from the picture we get some years later, even taking account of
important variations by news genre, channel, publication style and so on.
It is these large flows of social meaning that I propose we can investigate
systematically using semiotic concepts. Consider these two examples of
news coverage concerning the BP Deepwater Horizon oil rig explosion
of 2010.
1. The oil is now about 20 miles (32 kilometres) off the coast of Venice,
Louisiana, the closest it’s been to land. But it’s still not expected to
reach the coast before Friday, if at all.
BP, which was leasing the Deepwater Horizon, said it will begin drill-
ing by Thursday as part of a $100 million effort to take the pressure
© The Author(s) 2017 3

J. Gravells, Semiotics and Verbal Texts, Postdisciplinary Studies
in Discourse, DOI 10.1057/978-1-137-58750-3_1
4 Semiotics and Verbal Texts
off the well, which is spewing 42,000 gallons (159,000 litres) of crude
oil a day. (Carleton Place (Canada), 27.4.2010)
2. The ditty by the two singers included the lines: “When I hear that BP
story, Green and yellow melancholy, Deepwater despair.” (Coventry
Evening Telegraph, 27.4.2012)
Example 1 is drawn from a Canadian newspaper (Carleton Place) one week

after the explosion, and deals with the ongoing attempt to cap the leaking
oil well. It is packed with information of a certain kind—times, places,
amounts of money and volumes of oil, as well as the reported voice of
BP. The second example appeared in the Coventry Evening Telegraph (UK)
on the same date two years later. It is still of the genre “news report”,
but this time its topic is a protest about BP’s environmental record. This
extract also draws on reported voices, but in this case the commentary
on the BP crisis is expressed through an artistic genre—the protest song.
The phrase “green and yellow melancholy”, alluding here to the BP logo
colours, is taken from Shakespeare’s “Twelfth Night” (Act II, Scene iv).
These two news pieces make meaning of the crisis in very different ways,
and in this book I will explore what these changes in meaning-making
are, how and when they occur, and how we can examine the shape of
media representations through systematic language analysis.
The book explores a number of connected themes. Its primary focus
is the semiotic analysis of written verbal text. I will explore whether con-
cepts common in the study of semiotics, and more usually applied to
non-verbal sign systems, can be deployed as frameworks for investigating
stretches of text (discourse) and shed light on how they make mean-
ing. In this conceptualisation, collections of texts such as news media
representations of a story can be regarded as signs with their own sets of
characteristics. In other words, I will consider whether text 1 above has
anything in common with other texts written at the same time on the
same topic, and if so, what that tells us about how the press made sense
of the BP events at that time. I propose that text 2 seems to me to be
different in kind from text 1. What evidence can I marshal to support
that instinct? How are the texts (demonstrably, analysably) different, and
what does that tell us about how the view of the crisis has changed? In
suggesting that semiotic frameworks have something to offer in the study
1 Researching the Representation of a Crisis 5
of written texts, I will be considering semiotic concepts as epistemologi-

cal foundations for a practical methodology.
I take it that written text is a sign—a privileged sign no doubt—but
still one type of sign system amongst many. Writing sits alongside speech,
still image, film, gesture, music and so on as a resource for making mean-
ing, and it has been the enterprise of semiotics to explore the ways by
which such resources are exploited, managed, combined and systema-
tised by their users for myriad communicative purposes in an infinite
variety of contexts. Following Barthes’ (1972) insights that diverse cul-
tural phenomena can be understood as a sign or representation, scholars
of multimodality have investigated the regularities (the “grammar”) of a
range of semiotic modes including visual images (Kress & van Leeuwen,
2006), typography (van Leeuwen, 2006), music (e.g. Monelle, 1999),
film (e.g. Machin & Jaworski, 2006), as well as the interaction of these
modes (e.g. Iedema, 2003).
It is perhaps surprising, however, that the scholar sitting down to
engage with written texts from a semiotic perspective—in other words,
asking how does this text make social meaning?—may find that her
options for analysis are constrained. In considering written text as one
mode amongst many she may find that text in multimodal artefacts, such
as advertising or websites, is analysed as much for its visual properties, for
example, font, position and layout, as for the contribution to meaning
inherent in its lexico-grammar. I call this approach the “text-as-graphic”
approach, and touch on this again in the next chapter. Alternatively,
Systemic Functional Grammar (Halliday & Matthiessen, 2004) provides
a comprehensive analysis framework within which analysts can give an
account not only of verbal language but also of other modes, in par-
ticular still and moving image (Kress & van Leeuwen, 2006). Systemic
Functional Grammar provides an account of verbal language within the
clause and sentence and, and inter-sentential connections through the
study of cohesion (Halliday & Hasan, 1976). However, the concerns of
a semiotic account of text representations may also be broader, for exam-
ple, an investigation of the role of metaphor, or intertextuality, or how
writers have chosen to name events.
I argue in this book for an approach which offers an alternative to
these ways of studying written text. I suggest that there are a number of
concepts in the field of semiotics which are useful as explanatory frame-

works. Suppose our scholar of the preceding paragraph has a broad inves-
tigative agenda. Her need is for an account which is:
• Emergent. She does not wish to approach the text with a presupposi-
tion of what she might find, but rather to let the data “speak for
themselves”.
• Comprehensive. Her analysis should give a picture of the text at all its
levels, from close text analysis to social meaning-making.
• Critical without an agenda for emancipation.
• Flexible in terms of analysis tools.
• Multimodal in that written text can be analysed within the same epis-
temological framework as other semiotic modes.
Using news data related to the BP Deepwater Horizon disaster spanning

a two-year period, I will illustrate how a number of theoretical concepts
drawn from semiotics can be useful as starting frameworks to build a
comprehensive, critical, situated description of written texts. I will move
from a micro-analysis of word choice to a macro-analysis of whole sets
of texts, arguing that these can be construed as signs in themselves, with
shared characteristics of ideological significance. I started this chapter
with two quotations from the news coverage of the BP crisis. In order
to show the interplay between semiotic theory and practical discourse
analysis, I will apply the semiotic principles I will describe to this real-life
example of a business crisis constructed through the news media.
The text extracts at the beginning of this chapter are drawn from a very
particular linguistic context—news media coverage—and this is another
key theme. This book is partly a story about a story: how it starts, grows,
develops and changes. The lives of stories in the news media are tightly
bound to the conventions and practices of news writing, and the lan-
guage usages I discuss here will be very different from the discursive con-
struction of the BP story, say, at the pub. The analysis of the story needs
to be grounded in an understanding of how news media construct news.
The principles in which I ground my analysis practices are, nevertheless,
transferable to other contexts, genres and registers. They could equally
be used to examine our pub conversation about the BP crisis, as long as
they take account of the conventions and practices of pub conversation.

And while the news media are still highly influential in shaping how we
understand events in the world around us, their influence is increasingly
fragmented by both the proliferation of professional news outlets and
the increase in “lay” interpretations of the news, particularly through the
Internet. In proposing that the news media make identifiable “pictures”
of news stories at given times, I also acknowledge that these pictures are
multifaceted and shifting mosaics made up of different media profes-
sional and individual voices.
The next section of this chapter sets out some general observations
about the characteristics of news media stories, practices and genres. My
specific illustration is the explosion on the BP-leased Deepwater Horizon
oil-drilling platform on 27 April 2010, and its business aftermath, and
this introductory chapter concludes with a brief account of how the news
story developed over the two subsequent years.
Researching News Stories

The scholarship on news stories is extensive. Recent work takes it as read
that far from being a representation of an external reality, news writing
is highly culturally-situated and dependent on conventional journalistic
practice. Stories are mediated at every level, through selection and choice
of emphasis, through journalistic practices and constraints, through their
structure, format and co-text, and in terms of their language. The ways
stories are presented vary considerably by type of publication (tabloid,
quality/broadsheet, online), channel (print, online, TV, radio) and genre
of news writing (“hard news” article, feature, editorial, financial report
and so on), as well as displaying idiosyncratic features of individual writ-
ers. There is no version of a news story which can be taken as definitive,
and choices made by writers about representation of a story are always
ideologically grounded, in that they are a product of culturally-specific
convention and are subject to influence by those groups and individuals
with access to “voice” (Blommaert, 2005).
A brief review of the four areas I mention above—story selection; jour-
nalistic practice; structure/format and language—as well as overarching
conventions of genre – places the approach I discuss here into context.

The semiotic approach I propose is not intended to replace but to com-
plement this work into the language of news representation. A semiotic
account of a news story only makes sense when we understand what news
stories “look like” and why.
Story Selection
Given that news outlets need to select what is included in publications

(they cannot cover everything), then certain characteristics make some
stories far more likely to be considered newsworthy than others. Galtung
and Ruge (1973) found that stories with certain criteria dominated news
publications. Based on these criteria, the BP story was an obvious can-
didate for being reported, being of significant size and intensity, unex-
pected, unambiguously catastrophic and involving elite nations. The
larger point is that what is considered news is not naturally predeter-
mined, but selected and prioritised according to journalistic codes. These
codes are culture-specific, and reflect the ideologies of politics, power
and social agreement that are at play within large institutions like media
organisations. While story-selection procedures are broadly shared by the
media (a story such as the BP events would be likely to appear in most
national news publications) they also vary by publication—tabloids and
quality newspapers regularly choose different lead stories, or select differ-
ent aspects of the stories they report on (Bignell, 2002).
Apart from the topic of the stories selected, certain individuals and
organisations have a strong influence over which stories are prioritised.
Fowler (1991) and others give accounts of the many stakeholders involved
in producing news articles, from the news outlet proprietors, editors,
journalists and other staff, to those routinely consulted about affairs in
the public eye (politicians, business representatives, non- commercial
organisations, community representatives, the police and so on) and
those only consulted expediently (e.g. eyewitnesses and victims of crime).
Of these, some will have greater and more regular access to representa-
tion than others, and it follows that these will have a greater influence on
which stories are chosen and how they are eventually presented for public
consumption. Van Dijk (1996: 86) summarises:
Most obvious and consequential are the patterns of access to the mass
media: who has preferential access to journalists, who will be interviewed,
quoted and described in news reports, and whose opinions will thus be able
to influence the public? That is, through access to the mass media, domi-
nant groups also may have access to, and hence partial control over, the
public at large. (Emphasis in original)
Accounts of influence outside and inside publications focus on money or

political interest or both. Proprietors and editors of newspapers may have an
overt political stance that is made more or less clear to their reading public,
and is relevant to story selection and emphasis. Also less obvious is the influ-
ence of advertisers, usually a major source of income for mass-media publica-
tions. Advertisers buy space in publications whose readership and stated values
already fit their own, but researchers have argued that there is evidence that
their money buys a degree of influence over content (Roberts & McCombs,
1994), and it is certainly the case that the perceived behaviour of publica-
tions affects advertising spend, as the demise of the News of the World in
the UK after phone-hacking scandals shows. Touching on economic factors
and the priority given to advertising, Cotter (2010: 193) writes of the “news
hole”—namely, “what is open to editorial content—news stories—after the
advertising has been positioned”. Other influence on content and language
use has been shown to come from pressure groups, PR agencies and corporate
communications departments, all of whom have close relationships with the
media as part of their function (e.g. Burt, 2012). Examining another dimen-
sion of influence, Scollon (1998) contends that journalists largely orient their
writing towards other journalists. In recent years the public has gained a voice
in story selection, as “what’s trending” on the Internet becomes an early indi-
cator of likely stories of interest, and Twitter provides reaction and feedback
to stories from known and unknown voices, shaping the future direction of
representation.
Journalistic Practice
The language of news media texts cannot be separated from the context
in which they are produced, and this study of how a particular story
comes to be constructed needs to take account of the writing and editing
practices that constrain news writing, from the collection and selection
of information, to the organisation of stories to meet the particular news
cycles and space constraints of the publication. Such time and space con-
siderations might range from the daily cycle of a print newspaper with its
relatively regular number of pages, to a 24-hour TV channel, to a news
website with regularly updated content and space considerations that are
only limited by writing resources.
Far from compiling information for news stories from scratch, news
organisations have at their disposal a complex, ongoing network of infor-
mation sources. These include press releases from corporations and insti-
tutions, and news agencies such as Reuters and Agence France Presse
which are positioned as neutral and unaffiliated in their standpoint.
Agencies distribute news stories via newswires to other news organisa-
tions, mainly newspapers, television and radio. The customer publica-
tion may use the output in full or in part, so repeated forms of words in
different publications are common. A large number of the BP stories in
the data set derived from news agencies, and repetition, in particular of
reported speech, is a notable feature of the data. Other sources include
the publication’s own reporters “on the ground” in various locations at
home and abroad, who often initiate the coverage of breaking news, as
well as contacts in business, Parliament, the police, pressure groups, uni-
versities and other groups and institutions. Bignell (2002: 88) calls these:
“accessed voices” to whom the media have access and who expect access to
the media. The discourses of these groups therefore become the raw mate-
rial for the language of news stories, since news language is parasitic on
their discursive codes and ideological assumptions.
As these relationships develop, some of the contacts develop considerable

journalistic skills themselves, with business communications departments
writing press releases in such a way that they can be used almost unaltered
(Jacobs, 1999), universities developing a sense of what lies within a theoret-
ical piece of research that makes it an item of general interest (Baxter, 2014)
and the police finding ways of using media access to the public that can
contribute to the effectiveness of their own work.
Some of the most interesting work in writing and editing practices

has taken an ethnographic approach to tracking the life cycle of articles,
and looking at what selections, deletions and other changes are made in
the process and why. Cotter (2010: 88) gives a comprehensive account,
including in particular the role of the story meeting in “[d]eciding what’s
fit to print”. She argues that these meetings are a less visible, but poten-
tially more revealing reflection of a publication’s priorities and values
than the editorial pages. Van Hout and Macgilchrist (2010) follow a
story from press release to publication, and find that framing decisions,
that is, the selection and emphasis of certain information elements at the
expense of others, can be due as much to technical and space constraints
as to ideological considerations.
Structure and Format
News writing is a highly recognisable “genre colony” (Bhatia, 1993), not

least because of its repeated structures. For the hard news report, one typi-
cal structure is the “inverted pyramid”. The use of a section at the start that
summarises all key points results in a story that is not chronological, but
rather has temporal shifts, and journalists learn to present information in a
conventional hierarchy, through which narrative sequences are necessarily
reordered. The inverted pyramid structure appears to facilitate a brief but
accurate assimilation of “the facts”; however, the headline and lead do not
summarise the story, but rather point to the issues of maximum societal
disruption (White, 1997). Bird and Dardenne (1988: 77) point out that
this structure can encourage a partial, and highly directed, reading.
Feature articles are more likely to follow a conventional narrative struc-
ture, with events often related in chronological order, which entails certain
other linguistic regularities, such as a high presence of deictics and con-
necting words (Fulton, Huisman, Murphet, & Dunn, 2005). The typical
narrative structure means that the “point of closure” (i.e. the outcome) of a
feature tends to be near the end, rather than at the beginning, as it is in the
hard news story. Another key news genre, the editorial, tends to be struc-
tured quite differently again, using patterns of rhetoric and argumentation
such as Problem–Solution or Problem–Denial–Correction (Winter, 1994).
The language of news is organised not only by structuring strategies

within the text, but graphically upon the page, and issues of medium and
formatting can be relevant to how a story is construed. The traditional
print medium is characterised by columnar formatting and clear framing
of text items. Online news, on the other hand, generally exhibits a much
looser layout and is less constrained by space, as well as having additional
affordances such as hyperlinks to related stories within and outside the
news website, multimodal options including still images, video clips and
sound files, and opportunities for interaction between news writers and
readers. The effects of these looser graphic formats on the verbal language
of news include longer stories, greater access to voice by non-professional
writers and deictic expressions to guide the reader to alternative informa-
tion options.
News Media Language
Language choices in the news media concern how information is com-

municated once selected. The canonical descriptive work on the register
of newspaper language is that of Biber, Johansson, Leech, Conrad and
Finegan (1999). Biber et al. were able to isolate through corpus analysis a
number of features which were characteristic of news texts, for example,
a high use of nouns and prepositional phrases, a low use of pronouns
and a high proportion of complex phrases compared with other text
types. They found the present tense to be used more than the past tense,
because, although news items tend to describe past events, they make
use of historic present tense commentary, as well as direct quotation.
However, Biber et al.’s work does not take account of the differences
between newspaper types (tabloid and quality) or types of news item
such as the hard news report, feature, soft news, editorial, all of which are
distinctive in their language characteristics. Even within one newspaper
type, the stories that are presented as “hard news” will have a different set
of linguistic conventions from “soft news” stories. Because language usage
by newspaper type (e.g. tabloid and quality) and by media sub-genre
(e.g. hard and soft news) can be so varied, Biber et al.’s concept of a
newspaper language has been challenged (e.g. Landert, 2014) for being
so broad as to be unhelpful as an explanatory category.
Studies of the collection, writing and editing practices discussed above

overlap with research into specific language usages within media and news
writing. Most research into this topic adopts a broadly critical perspective
on the language studied, and writers who work within the field of Critical
Discourse Analysis have found media texts to be a particularly productive
data source for the study of the exercise and maintenance of power. Key
works on the language of the media from this critical perspective include
Fairclough (1995b, 2000), van Dijk (1985, 1988) and Fowler (1991). A
recent challenge to the assumption of manipulative intent in the press
has been made by Martin (2004) and Macgilchrist (2007), who argue
that some media pieces challenge and reframe dominant ideologies in a
way that runs counter to expectation. They call this enterprise “Positive
Discourse Analysis”. Similarly, the ethnographic work mentioned above
emphasises the role of writing practices as much as ideological choices in
motivating the language of news stories.
Media language is usually the product of many voices. These will
almost certainly be those of the journalist and sub-editor, but can, as
discussed above, include interviewees, directly or indirectly quoted, and
other voices that are partly or wholly unattributed, drawn from general
debates, background research, and written input from interested parties.
This form of intertextuality in the news media is explored in work such
as Bednarek (2006), Oliveira (2004) and Macgilchrist (2007). Studies of
intertextuality in news articles also indicate the extent to which journal-
ists can converge around ways of presenting information (e.g. naming
practices) partly because so many of their sources are shared, and much of
their writing is drawn from others’ previous work on the same news item.
Ready access to global accounts both in print and on television increases
the resources that are available for journalists to draw on as source texts
to be either endorsed or challenged. Other work on intertexts considers
the extent to which writers variously align themselves with, or distance
themselves from, others’ texts and voices using speech presentation strate-
gies, modality resources (Roberts, Zuell, Landmann, & Wang, 2008) and
the Appraisal System (Martin & White, 2005; White, 1997).
Despite this frequent use of shared resources, the lexis used by tabloid
and broadsheet newspapers has been found to be specific not only to each
format (Conboy, 2007) but also to individual publications, through the
imposition of style guides (Cameron, 1995; Cotter, 2010). Each publica-

tion will identify with a certain audience and set of values, and will select
newsworthy items and language usages appropriate to this positioning.
Tabloid newspapers use a more oral and conversational style (Bignell, 2002;
Fowler, 1991), marked by contractions, slang, idioms, nicknames, indi-
cations of spoken emphasis such as italics or bold type and a restricted
set of vocabulary, which, while not always representing its readers’ own
language use, is recognisable and shared. This oral style connotes famil-
iarity, camaraderie and entertainment. Quality newspapers, on the other
hand, tend to use a style that connotes authority, formality and serious-
ness. Vocabulary and structures belong to a more “written” and formal
register, with longer sentences, no creative misspelling and a preference
for metaphor over puns. Quality newspapers, through their patterns of
modality and personal pronouns, are more likely to give an impression of
objectivity and even-handedness in contrast to the impression of subjec-
tivity and shared values given by tabloids.
As well as these differences by newspaper type, genres of news writing
have distinctive patterns of topic, structure and language choice, which
are both recognisable and constitutive of their type. These differences in
news genres are an important explanatory factor for changes in the BP
story, and are worth reviewing here.
The News Report
Scholars of news media regularly make a distinction between “hard news”

and “soft news” (Bird & Dardenne, 1988; Fulton et al., 2005; Martin &
White, 2005; White, 1997). Bird & Dardenne suggest that hard news is
“important” whereas soft news is “interesting”. Fulton et al. (2005: 143)
further describe the difference:
“Hard” news reports on politics, economics, the doings of the powerful

and international affairs—in other words, on those aspects of a nation’s
public life that are considered to have the greatest influence on the lives of
its citizens. It is contrasted with “soft” news, which is about “human inter-
est”, about celebrity, crime, the small scale and domestic.
It is not only the content of the story, but the discursive approach and
positioning that distinguish hard and soft news reports. In the case of
a disaster such as BP Deepwater Horizon, for example, the main news
story might present the latest overview of events, while a soft news story
still on the same topic might turn to family stories of victims, or the
heroic actions of a single participant.
Large-scale events such as political, business and financial crises, as
well as accidents and disasters, are generally reported as hard news, and a
number of features of style and structure are associated with these types
of report, for which I will use the term “news report”. Although the term
“hard” implies definitive, objective and factual, the news report genre
cannot be considered a representation of fact—as there is no single objec-
tive reality to be reported on. Rather, news reports are one of the many
discursive practices that mediate our perception of events. News writing
comprises a complex set of practices, constraints and choices, all of which
mitigate against a neutral representation. Fulton et al. (2005: 232) write
that “objectivity is a style, a set of linguistic practices that we have learned
to recognise as signifiers of factuality and impartiality”. The authors draw
attention to a number of these practices in factual reporting (2005: 233):
• High proportion of empirical information regarding dates, places,

times, amounts of money and so on.
• Third-person narration; narrative voice externalised and elided.
• Lack of modality; preponderance of declarative (i.e. not conditional)
verbs indicating certainty.
The aim of such strategies is to present an authorial (and institutional) voice

which presents itself as “neutral and anonymous and thus as directly and
mechanically determined by the events it portrays” (White, 1997: 129).
However, suggesting that a news report places an emphasis on empiri-
cal information does not imply that opinion and evaluation play no
part—on the contrary, comment from eyewitnesses and experts is an
essential feature of news reporting. What is significant is that such
comment is clearly positioned as being separate from the (neutral) fac-
tual report, usually through the use of direct and indirect quotation.
As Martin and White explain (2005: 168, emphasis in original), in
these types of text “those values of judgement which occur are always
mediated through attribution (the journalistic author is never their

immediate source)”. As judgements are overtly positioned as personal
opinion through attribution, the rest of the report is, by contrast, posi-
tioned as objective and neutral. This is related to the third point above,
where modality and (un)certainty are available to both the journalistic
voice or reported voices but, the authors argue, are less typical of the
journalistic voice in hard news. In Chap. 9 I will go on to problematise
further the assumptions about the use of modality in hard news.
The “objective” strategies mentioned above are not used uniformly
in news articles. White (1997: 108) finds that quality and tabloid pub-
lications alike commonly use what he calls “intensified lexis” (he cites
“axed” for “dismissed”, “shake-up” for “reorganisation” and “torrential
rains lashed” for “heavy rain fell”). Figurative language such as metaphor
can introduce “an element of modality or appraisal which enlivens (and
ideologically positions) an otherwise bald account” (Fulton et al., 2005:
232). And a hard news article following the inverted pyramid structure
can contain a section in narrative feature article style, which may serve
not only to engage the reader, but to emphasise by contrast the claims to
objectivity of the rest of the piece. One of the early BP texts tells the story
of the escape of a cook from the burning rig. The article in its entirety is
a hard news article, but the section which recounts the cook’s story is told
largely in reported speech, as a chronological narrative, adding a different
dimension to the unembellished prose. Further, the impersonal, objec-
tive style is not uniformly used across types of newspaper. As mentioned
earlier, tabloid discourse is typically more personal, conversational and
subjective than that of quality newspapers.
These tensions between impartiality and evaluation, and neutral and
subjective language reflect the challenges for news journalists in meeting
the conflicting aims of representing the world “as it is”, taking account of
institutional interests, and engaging the reader.
The Feature Article
In rhetorical terms, news reports are positioned as primarily realising a

descriptive act. Other types of mass media journalism are more evaluative,
and these can include feature articles, editorials, blogs and commentary
pieces as well as reviews and letters pages.
Feature articles position their information and participants in quite a

different way from news stories. Features are not tied directly to the news
of the day, and because of this can be “slotted in” when there is space
to be filled. Unlike news stories, they do not present “just the facts”,
but generally present human interest stories within a much broader, less
time-bound way (Cotter, 2010). Bird and Dardenne (1988) make a use-
ful distinction between the “chronicle” that is primarily descriptive, and
the “story” which follows a more subjective, narrative pattern. They point
out that while the former is more aligned with journalistic values that
foreground a perspective of objectivity, the latter approach is more suc-
cessful in engaging readers, and suggest that media writing typically seeks
a balance between the two, which can be more or less successful.
As well as the narrative structure already discussed, other generic char-
acteristics of the feature article are that represented participants are often
recognisable stereotypes brought to life by quotations, and that there can
be a conflict or parable, leading to the outlining of a (shared) moral posi-
tion. Landert (2014) notes a high usage of first- and second-person pro-
nouns in soft news. Where there is a third-person narrator, the viewpoint
of the writer is foregrounded, and he/she directs the attention towards
the desired reading.
The Editorial, Blog, Commentary and Reviews
In considering evaluative texts, a number of issues to do with language

choices are relevant. The language of judgement and evaluation is far
more likely to appear in articles that analyse events than articles that
report them. Martin and White (2005) propose that evaluative texts are
rich in expressions that communicate judgements in the areas of affect,
ethics and aesthetics. As discussed in earlier comments on the differences
between the quality and popular press, emotive and judgemental lexis
can be a feature of both news reports and opinion pieces in the popu-
lar press, whose concern is to reinforce values that they share with their
readers, whereas the quality press tend to avoid evaluative language in
news reports. Just as news reports are said to exhibit a lack of modality,
in evaluative texts, the writer is more likely to indicate his/her own view
on propositions through the use of modal expressions. This applies not

only to the expression of (un)certainty, (in)ability and permission/obliga-
tion, but to the degree of “modal responsibility” exhibited in the modal
expressions chosen (e.g. the choice between “it seems to me that” and “I
think that”) (Thompson, 2004). As a final point, one of the rhetorical
acts performed by editorials and blogs is to persuade the reader of the
validity of a point of view, and this genre frequently exhibits rhetorical
features associated with persuasive argument, such as end-focus or tri-
colon forms, as well as argumentation structures such as variations on
Problem–Solution structures.
I mention weblogs as a particular type of commentary writing partly
because they appear in a different channel from print news, and this
can imply different language characteristics, and partly because their
writers are less constrained by the influences and practices that are
typical of journalistic writing for institutional publications. Weblog
writers are not necessarily trained journalists, but can be lay individu-
als with a particular interest in a topic. Herring et al.’s (2004) genre
analysis of weblogs identifies diverse types, from those that share simi-
larities with editorials and/or letters to the editor, to those that have
commonalities with the personal journal, and those that invite an
exchange of views. The authors found overall that writers exaggerated
the extent to which weblogs offered an interactive, multimodal, exter-
nally-oriented form of participation in the news. Rather, they tended
to be “individualistic, intimate forms of self-expression” (Herring et al.
2004: 1). Weblogs which deal with the news enjoy easier access to
a voice by non-professional writers, fewer space constraints and are
typically less heavily-edited than their printed equivalents. These and
other characteristics of online news affect the linguistic output, often
resulting in a more informal, individual and provocative writing style,
and this was clearly exhibited in the BP texts. The recent demise of the
print edition of The Independent newspaper in favour of online only is
one indicator that non-institutional voices will be increasingly present
in news language given the interactivity of the digital medium.
In Summary: Difference and Sameness in News

Reporting
One of the critical aspects for the study of sets of news stories is the role
of these different genres and sub-genres in the overall representation. How
significant a presence are news articles in relation to features, or editorial
or arts reviews in the telling of the story? The choice of genre in itself has
semiotic meaning (why is a story considered to be a “news story” rather
than a “feature story” for example) and our analysis approach should take
account of choice of genre and how this changes over time. The conven-
tions of these genres will play a large part in shaping our understanding of
how to interpret the story. When the proportion of any one news genre
increases or decreases, then our overall sense of how we are to understand
the story changes.
I have argued to this point that news media language and structures
are highly varied by genre, challenging Biber et al.’s conception of a single
newspaper register. However, there is a case for sameness as well as difference
in newspaper writing, and a degree of standardisation of language across
publications is also evident. I have already touched on the intensified
lexis shared by tabloid and quality press. Conboy (2007: 108) shows how
repeated forms of words activate an understanding of shared values and
ideologies—he uses the example “our boys” to describe British soldiers on
active duty. Cotter (2010: 71) explains how reports of ongoing news sto-
ries recycle language expressions used in previous news reports in “boil-
erplate” formulations. The use of press releases and newswires encourages
repetition of the same, readily-available wording. For example, in my BP
data I repeatedly encountered this quotation from 2010, “We’ve never
seen anything like this magnitude” by George Crozier, “oceanographer
and executive director at the Dauphin Island Sea Lab in Alabama”. The
quotation appears six times in my full 2010 data set, which represents
only a fragment of what was written in the week following the explosion.
It is credible to suggest that George Crozier’s words became a resonant
part of the BP story as understood by the English reading public. Cotter
(2010: 192) writes:
News discourse is rule-governed, routinized, and follows on from

profession-specific norms and routines. Journalistically, standardization
involves the rules of writing for the appropriate modality […] the news-
gathering conventions of the profession […] and the proliferation of core
journalistic ideals such as balance of sources, attribution of sources, and
responsiveness to audience. (Emphasis in original)
It can be argued, then, that what we have termed “news media discourse”
is characterised by both homogeneity and fragmentation. Different
types of news item, news publication and channel entail differences in
tone, framing, story selection, structure and emphasis, depending on the
work each has to do and the audience each addresses. Within each spe-
cialist media output—an editorial in the online version of an English-
speaking newspaper in India, say, or the financial pages of the Mail on
Sunday—various patterns of language can be found which both reflect
and construct a version of reality consistent with their social purpose
and audience. Conversely, the significant (and increasing) sharing of
resources, agency accounts, eyewitness accounts, press conferences, previ-
ous news texts as well as conventions of language usage constitute a move
towards homogeneity.
I will draw on aspects of both sameness and difference in news media
writing as I present an analysis of the BP data. The rest of this chapter
introduces the BP news story: as archetype and as specific exemplar of a
business crisis.
Crisis and Communications
I turn first to the BP crisis as archetype. The choice of business crises
as objects for linguistic research is persistently relevant, as business crises
appear regularly on both the front and business pages of news publica-
tions (and their online equivalents). Over the last 30 years our perception
of global events has changed considerably: the speed of communications
technology and the proliferation of online and offline media channels have
allowed us virtually instant access to information about events across the
world. Events such as natural disasters that were once considered entirely
out of our control are now subject to scrutiny as we become aware of our
own part in environmental change. In the world of business, an aware-
ness of unpredictability has been incorporated into the fabric of corporate
strategic thinking. At a time when many organisations operate in a diverse
global environment, when the pace of technological development is accel-
erating and when corporate actions are more visible and more deeply scru-
tinised than ever before, most business organisations practise some form
of crisis planning. Business studies students are taught that flexibility will
be a key skill in the “real” world, and that business strategies that prioritise
ongoing learning are replacing the linear planned strategies of the twen-
tieth century, as the operating environment becomes more volatile and
unpredictable (Mintzberg, Ahlstrand, & Lampel, 1998).
However, even against a backdrop of chronic flux, certain business
crises stand out. Over the past few years alone, there has been consider-
able public attention paid to, for example, the BP Deepwater Horizon
oil spill, the collapse of various financial institutions, including Lehman
Brothers, AIG and Northern Rock, unethical practices within some divi-
sions of News International and fraudulent claims by Volkswagen about
the emission levels in its vehicles. Crises that stand out in the media are
often not only those that are significant in size or impact, but those that
appear to exemplify some particular aspect of modern life. For whatever
reason, certain crises are given more attention and move from controlled
private setback to major public affair. It is worth noting at this point
that I regard the word “crisis” as contentious. Since I take the viewpoint
that social phenomena are both shaped by and shape discourse, it follows
that my calling the phenomenon a “crisis” (rather than, say, “disaster”, or
“event” or “set of events” or any alternative descriptor) already implies a
set of presuppositions. I address this point through an investigation of
naming practices for the BP events.
Organisational crises are defined as events characterised by high conse-
quence and low probability, ambiguity and decision-making time pressure.
An organizational crisis is a low-probability, high-impact event that threat-

ens the viability of the organization and is characterized by ambiguity of
cause, effect, and means of resolution, as well as by a belief that decisions
must be made swiftly. (Pearson & Clair, 2008: 3)
Within this general definition, writers have found types of crisis to be

significant in determining business response. Coombs (2004) proposes
typologies based on attribution of responsibility: Victim Crisis (e.g. natu-
ral disasters); Accidental Crisis (e.g. technical errors); Intentional Crisis
(e.g. human error and misdeeds). In media coverage of the Deepwater
Horizon oil spill, all three of these have been suggested as relevant at
certain points, but the most applicable category would seem to be the
“Accidental Crisis”. In Accidental Crises, the organisation may be per-
ceived as potentially negligent but not wilfully damaging.
A crisis is not viewed as a single event, but as a process, consisting of
a number of phases (Hale, Dulek, & Hale, 2005; Pearson & Mitroff,
1993), most commonly characterised as crisis prevention, crisis response,
and recovery from the crisis. Fink (1986) amplifies the prevention stage
to propose five stages: crisis mitigation, planning, warning, response and
recovery, and in keeping with business priorities the focus of attention
here is on planning and prevention. My area of interest, which is media
coverage, only starts at the point when the crisis is made public, and
relates only to the final two of Fink’s five stages—response and recov-
ery. This view of progressive stages can imply that the set of responses
to a crisis follows a linear pattern from start to finish. However, Hale
et al. (2005: 123) suggest that an iterative pattern is more likely, whereby
responses are made, assessed, revised and made again. It is self-evident
that the language used to describe and inform stakeholders about a crisis
can be critical to how it progresses. These assessment and revision stages
of crisis response often relate to the success or otherwise of particular
communication strategies, as well as the technical and logistical handling
of crisis consequences.
With regard to communication and message strategies, it is clear that
the type of crisis will affect the organisational response in terms of tone
and content, as stakeholder views are generally determined by the compa-
ny’s perceived role in originating the crisis. These organisational responses
are typically communicated to the media through press releases, press con-
ferences, and television and press interviews. In the case of the BP events,
television and press interviews were frequent and high profile. What is
ultimately published in the media (our area of interest here) is of course
not under the control of the organisation, although research shows the
significant presence in media writing of corporate press releases, which

are often reproduced more or less verbatim. The tone of the organisa-
tional response is likely to be picked up and either endorsed or dismissed
by the publication depending on its political stance and likely readership
response. Response strategies are tailored to the intended audience, and
Stephens, Malone, and Bailey (2005) draw on stakeholder theory to dis-
cuss the shifting importance of varying stakeholders at a time of crisis.
The intended audience for crisis response communication is found to make
a difference to structure and tenor, as well as content and level of detail.
This is particularly crucial for crises that require technical explanation,
as is the case for BP Deepwater Horizon. Just as media writing is multi-
voiced, so business communication is co-created by individuals, includ-
ing the Chief Executive Officer, the Communications Director and other
writing professionals involved, as well as by the company culture, and the
norms of the industry within which it operates. The balance between the
corporate and the personal in crisis communications is a potentially inter-
esting aspect for language study. BP itself encountered well-publicised dif-
ficulties in finding an effective public communications approach.
The BP Story
As a specific exemplar of a crisis, BP Deepwater Horizon constitutes a
particularly appropriate case for research in that it can be seen as a single
set of events, and is therefore more manageable as a research topic than,
for example, the UK’s financial crisis that dates from 2007 and is still a
media topic today. Both causes and consequences of the BP crisis are mul-
tidimensional—encompassing human, technical, environmental, politi-
cal and business aspects. There is a great quantity of available data: much
was written in the immediate aftermath of the events of 20 April 2010,
as well as later with hindsight. It continues to be referenced widely today,
as an example of new challenges to the PR industry (e.g. Burt, 2012),
as an archetype of a multibillion dollar compensation case (e.g. in
Cahalan, 2012) and, in July 2012, in an HBO television series The
Newsroom, with a script by Aaron Sorkin and in a 2016 feature film. BP
Deepwater Horizon offers an exemplar of speakers and writers competing
for the right to define the history of a crisis.
The media story is as follows. On 20 April 2011 an oil-drilling rig in

the Gulf of Mexico, which was operated by Transocean on behalf of BP,
exploded; 11 rig workers were killed, and 17 were injured. BP’s report
on the crisis (BP incident investigation team, 2010), as well as a US
National Commission report for the president (National Commission
on the BP Deepwater Horizon Oil Spill and Offshore Drilling, 2011),
described how a series of faults and failures contributed to the cata-
strophic event, each of which in isolation might have been controlled.
These included the suitability of the design of the well, faults in the
construction (particularly the failure of the cement fixing), the failure of
various early warning systems, the failure of the blowout preventer and
human error. In his book telling the story of the BP Deepwater Horizon
explosion and BP’s subsequent handling of the crisis, Bergin (2011) sug-
gests that this range of direct causes can be situated within two broad
background contexts—a management and organisational culture within
BP that encouraged productivity improvements at the expense of health
and safety, and a US regulatory system for the industry made lax by the
historically strong influence of oil interests on the US government.
Heading the handling of the media story personally, Tony Hayward,
the CEO of BP at the time of the explosion, had very early tried to dis-
sociate BP from the failures leading to the disaster (Bergin, 2011) and
named its partners on the rig, particularly Transocean, as being primar-
ily to blame, while accepting BP as ultimately “responsible” in the sense
defined by the US Environmental Pollution Act of 1990. The BP report
names eight failures, of which seven can be attributed to sub-contractors.
The National Commission report highlighted nine causal factors, which
were very similar to those BP identified, but placed the blame “at BP’s
door” for seven or possibly eight of these.
From a media perspective, the focus of the story changed over the con-
siderable period in which it was covered by news outlets, and these myriad
aspects of the story make it of particular interest for research, as the crisis
is shown in the media to have resonance from a number of socio-cultural
perspectives. The search for the 11 missing men was abandoned within a
few days, and Tony Hayward and his team expressed their deep condo-
lences for the tragic loss of life. Crisis teams, both technical and commu-
nications, were assembled immediately, and work on both fronts carried
on day and night. There was significant uncertainty about controlling the
oil spill; the rig was working in deeper water than had been attempted
before, and there had been so many equipment failures in the lead-up to
the explosion that standard options for containment were no longer avail-
able. From the onset of the crisis, Tony Hayward insisted personally on
being the face and voice of BP during this crisis period (Bergin, 2011: 228).
The BP communications team was trying to control speculation, but had
limited information to impart, and journalists working on the story at the
time recount the frustration within the media community as questions
at press conferences were heavily restricted. These restrictions had the
outcome that journalists sought interpretations from a range of sources.
After a week of daily press releases concerning the efforts to block the oil
spill, Quarter 1 financial results for BP were issued. Some of the news
stories mentioned the financial results in passing, while financial reports
and business pages mentioned the oil spill in terms of its likely impact on
the business health of the company.
The Deepwater Horizon oil well was finally capped in July of 2010 after
12 weeks of oil spillage. In September 2010 Tony Hayward was forced to
resign, owing to concerns about his management of the spill, and his han-
dling of the media. He was replaced as CEO by American Robert Dudley,
previously head of BP’s joint Russian venture TNK-BP. BP accepted
legal responsibility for the spill; that is, that in a specific legal sense they
accepted that they were liable for clean-up and other compensation costs.
With the cooperation of local agencies, they coordinated a significant
clean-up operation. Compensation mechanisms were set in motion for
individuals and communities affected by the spill, and compensation dis-
putes, as well as more positive stories, were the focus of much media
coverage. Compensation payments ran concurrently with legal proceed-
ings to determine the causes of and specific responsibility for the spill.
The results of BP’s own investigation into the incident were published
in September 2010 (BP incident investigation team, 2010), and those of
the National Commission on the BP Deepwater Horizon Oil Spill and
Offshore Drilling were published the following January (2011). Apart
from their different readings of the “facts” mentioned above, the tone and
presentation of the two reports are substantially different. The BP report
reads something like a UK Government report, using plain language that
suggests objectivity. The National Commission Report from the USA,

in contrast, is given an informal title (“Deep Water”), is illustrated with
non-technical photographs and is dedicated on the first page to those
who died and their families. These two texts on their own provide an
early insight to how even broad agreement on the “facts” of the case can
give rise to significantly different discursive constructions of the crisis.
Once clean-up operations in the Gulf were finished, the waters of the Gulf
and its wildlife continued to be monitored by BP and other US agencies.
The costs of the clean-up and compensation, estimated at over $40 billion
by March 2014, as well as the sale of assets to fund them, left BP in a fragile
financial position, exacerbated by falling oil and gas prices in early 2012.
Large legal settlements were still being made in 2015. BP entered into several
high-profile sponsorships, most notably as an official partner of the London
2012 Olympic Games and as sponsor of the World Shakespeare Festival.
This overview of events indicates how, over the two years for which I
examined media data, the core story of the explosion and the subsequent
environmental disaster was supplemented by other related events and
actions, for example, the clean-up operation and the legal process of com-
pensation. But there were more profound changes in the story than can be
accounted for by this additional content. The language used by the media
to construct the story of the crisis changed in fundamental ways: in the
representation of certainty, for example, in referencing the crisis and in
associating it with other events. Taken together, these changes can be under-
stood as shifts in media representation from one perspective to another.
Understanding these shifts involves taking a broad and holistic view of the
data and I suggest that conceptualisations from the study of semiotics can
be used to provide a fresh perspective on the study of written texts.
2
Semiotic Discourse Analysis
Researching Written Text

The key concern of this book is to offer a way of looking at written v erbal
text through the lens of semiotics; in other words, it will consider how
meaning is made in the particular sign system of writing. To illustrate
this approach, it will investigate a certain event—the BP Deepwater
Horizon crisis—within a particular context: written news media output.
Having considered both illustration and context in Chap. 1, I turn now
to the main focus of the book—a semiotic understanding of written text.
It is useful to locate this proposed approach alongside other analytical
approaches to written text. The diagram in Fig. 2.1 depicts some of the
more widely used approaches to the analysis of written text. It is not an
exhaustive account, and the approaches are not necessarily discrete. Not
only do more perspectives and methods exist, but language researchers
can and do make use of multiple methodological resources to answer
their research questions. When researchers handle written text, they
often start from a theoretical perspective or standpoint such as Critical
Discourse Analysis (CDA) or a semiotic perspective, and proceed to
examine data using diverse “tools” (methods) for analysis, such as content

in Discourse, DOI 10.1057/978-1-137-58750-3_2
Fig. 2.1 Perspectives and tools of written text analysis
analysis or grammatical analysis. So, while I regard the “perspectives” in

the diagram as being relatively discrete, the tools are not. For example,
scholars working under the broad umbrella of Potter and Wetherell’s
Discourse Analysis have used content analysis (Potter & Reicher, 1987),
Conversation Analysis (Speer, 2001) and poststructuralist approaches
(Wetherell, 1998) in their work. Corpus analyses can be used in the ser-
vice of critical accounts, in the analysis of computer-mediated discourse,
and even to support and contextualise other “bottom-up” methods such
as grounded analysis. An essential element of the approach I will outline
in this book is that it has the flexibility of assembling a “toolkit” from a
range of linguistic research methods.
The following brief overview of the “tools” and “perspectives” named
in Fig. 2.1 is intended to locate the approach I propose in a wider view
of text analysis. All of the approaches and tools mentioned have been
used to examine media data and all would have been possible options
for examining the BP data. None, however, fully addressed the breadth
of the research questions with the degree of flexibility I sought, and I
explore this position further, as I discuss alternative approaches below.
2 Semiotic Discourse Analysis
29
Tools
On the right-hand side of Fig. 2.1 is a set of tools or methods which can
be used singly or in combination to investigate aspects of language use.
Corpus Linguistics
While corpus linguistics has always been concerned with the study of large
data sets, or corpora, the advent of computer technology has enabled work to
be carried out on millions of items quickly and accurately, and corpus anal-
ysis can now be undertaken by all researchers with computer access through
accessible software programmes. The key benefit of corpus approaches to
language is that large-scale analysis can uncover patterns of language use that
are not easy to see with the human eye. Some of these prove to be counter-
intuitive, and challenge what we believe we see from experience or instinct.
In this way, quantitative studies can provide fresh insights as well as locating
our qualitative observations within patterns we can demonstrate. Corpora
are of three main types—general linguistic corpora (such as the British
National Corpus), specific linguistic corpora which contain examples from
particular genres of language (for example, journal articles or medical writ-
ing) and self-compiled corpora (of which my set of BP texts could be an
example). Evidence from specific or self-compiled corpora can be compared
with that from general linguistic corpora to indicate whether a particular
pattern of use conforms to a general norm, or appears to be somehow dif-
ferent. This kind of work is often carried out to identify language differ-
ences between registers and genres, for example Biber, Johansson, Leech,
Conrad and Finegan’s (1999) study mentioned already of four types of lan-
guage use: conversation, academic, fiction and newspaper language. Biber
et al. proposed clusters of features which differentiated the four registers,
placing them along six dimensions (involved vs information, primarily nar-
rative and primarily non-narrative and so on). Biber et al.’s work highlights
not only that it is sets of features co-occurring rather than individual features
which can differentiate different text types, but also that such differences are
tendential rather than binary absolutes. Common patterns of language use
easily obtained using corpus tools include frequencies of words (significant
in comparison with a norm), concordance lists of words (lists which show a

word of interest, or keyword, with a number of words either side) and collo-
cations (the most common words which co-occur with the keyword). Such
quantitative output of a corpus analysis requires qualitative interpretation
of what findings are significant for the research question and why. Corpus
scholars also make a distinction between research which is corpus-driven
and corpus-based. In corpus-driven research the findings are emergent; for
example, certain words or word classes prove to be significantly more rep-
resented than predicted by general corpus norms. In corpus-based research,
particular terms, words or grammatical features are investigated according
to a research question or hypothesis.
Corpus linguistics is a field of study in its own right, but such analysis
is a tool which can be used in multi-method linguistic work. It has been
used both in constructivist and more positivist paradigms as a robust
starting point for other, qualitative, analysis. For the purposes of the BP
research I chose to ground my qualitative findings in quantitative work
of a different kind—a quantitative overview of the data set and feature
counting by hand—but a corpus approach would have been possible and
potentially effective in scoping out the broad traits of my data.
Content and Narrative Analysis Tools
Under this broad heading falls a group of largely emergent analysis meth-
ods. Content analysis, thematic analysis, grounded theory and narrative
analysis have different theoretical and methodological approaches, but
share a focus on what is communicated rather than how it is communi-
cated. Content analysis can be both quantitative and qualitative: in its
quantitative form, it is the systematic description of the content of often
large data sets, via a set of chiefly predefined codes. Qualitative content
analysis places more emphasis on the emergent and recursive derivation
of coding categories, where codes are proposed, applied, tested for rele-
vance and reapplied usually on smaller amounts of data (Bryman, 2004).
Thematic analysis according to Braun and Clarke (2006: 79) is “a method
for identifying, analysing and reporting patterns (themes) within data”.
Braun & Clarke emphasise the flexibility of thematic analysis as a tool
which can be used to support research of different paradigms, and across
31
different research methodologies. They emphasise that it does not require

the commitment to theory-building which is a pre-requisite of grounded
theory, and so can be carried out in a wider range of research environ-
ments, and by researchers with greater and lesser levels of experience.
As Braun and Clarke suggest, the principles of grounded theory are
more exacting than content or thematic analysis, as well as harder to
define, having been revisited by a number of scholars since their incep-
tion by Glaser and Strauss (1967). Broadly, grounded theory is a qualita-
tive method which is iterative and recursive with regard to its data, in
the same way as content and thematic research. It places great emphasis
on being exhaustive in its coverage; in other words, categories are built
and refined until all the content of the data is accounted for. In addi-
tion, it has a declared commitment to theory-building. This can be seen
as problematic—it is not always clear that a grounded analysis of data
results in theory which is applicable in a wider context (Bryman, 2004).
As with all recursive analysis, it can be difficult to suspend assumptions
until all options are exhausted. Finally, narrative analysis places emphasis
on how individuals make sense of their relationship to social phenom-
ena. Through the collection and analysis of people’s accounts of their
lives, researchers build a picture of how events, organisations and other
phenomena are both understood and constructed. Narrative analysis can
make use of content, thematic or grounded analysis approaches.
These iterative, bottom-up approaches are particularly useful for research
questions which require (at least initially) a description of the data. I shall
go on to describe an “immersion” phase of my data analysis which used
an emergent and recursive method to identify key language features in the
items. However, my interest was in the use of language features rather than
their semantic content—the “how” rather than the “what”. The approaches
briefly described here are adopted for descriptions of the “what”.
Traditional Grammar
By the term “traditional” grammar, I refer to grammar that is not Systemic

Functional Grammar (SFG). We can make distinctions between the genera-
tive grammar of Chomsky (1965) and SFG that largely concern different
emphases on either form or function. Generative grammar supposes an ideal
competence which is never realised in imperfect performance. Its emphasis is

on structure and form. By contrast, SFG (covered in more detail later in this
chapter) works from the principle that language is always functional: “language
is as it is because of what it has to do” (Halliday, 1978: 19). However, what we
might call traditional grammar is no longer concerned exclusively with fixed
and abstract structures. Crystal (2004: 7) writes that in the recent teaching of
English grammar “the concern [is] to integrate the study of language structure
with that of language use”. Handbooks of grammar such as Crystal’s not only
address social and functional issues including context-dependent variations of
language in use, and the legitimacy of non-standard Englishes, they also bor-
row concepts from Hallidayan grammar to explain issues of language which
were not previously addressed in traditional grammars, for example the con-
cept of “given” and “new” information to explain sentence organisation and
information focus. In its turn, SFG has always drawn on many terms used in
traditional grammar (parts of speech and so on) while renaming others. Both
traditional grammar and SFG give us the vocabulary to refer to and the tools
to identify and explain features of language. In my own analysis, concepts
from both traditional grammar and SFG were used to identify and explicate
features such as modality and direct/indirect speech, as well as other phenom-
ena within the broader area of news media genres.
Perspectives
On the left-hand side of Fig. 2.1 are approaches to the study of language
which start from a particular theoretical standpoint. Each approach is
grounded in a particular view of language and its relationship with soci-
ety, by which I mean both context (however defined) and participants.
These “perspectives” lean towards using methods which support either a
more descriptive or a more critical aim.
Critical Discourse Analysis
CDA is an approach to texts which uses diverse analytical tools to uncover the
covert enactment of power through language. Analysts in this tradition are
explicit about having an agenda (Fairclough, 1989, 1992a; van Dijk, 1985,
33
2008; Weiss & Wodak, 2003; Wodak, 1996, 2000). They argue, rightly in
my view, that any analysis process, even one which purports to be purely
descriptive, starts from a point of view, and the analyst of necessity selects an
approach and a data set which reflect a particular research intent. Given this,
writers in the CDA tradition find it appropriate to make this intent a prin-
ciple of the approach, as an overt declaration of reflexivity. Critical work in
the field of media representation has shown effectively how news stories, far
from being an objective representation of “what is out there”, offer a version
of the world that is shaped by political, financial, institutional, personal and
temporal constraints. However, it has been argued (e.g. Graham, 2005) that
in following its agenda for change, much work in CDA is inclined to make
its own claims to truth and objectivity. Widdowson (1995), in a frequently
cited debate with Fairclough about the theoretical validity of CDA as an
analytic approach, criticises Critical Discourse Analysts for bias in seeking
out examples of text that support their own political viewpoint.
While CDA is characterised by its attitude to text analysis rather than
by a specific method, different CDA approaches (Fairclough, 1989; van
Dijk, 2008; Weiss & Wodak, 2003; Wodak & Meyer, 2001) offer fully
developed analysis frameworks, which have in common that they place
the individual text at the heart of a set of social practices and cultural
norms that tend to both reflect and perpetuate the interests of those with
power and access to voice. Writers such as Fairclough and Kress use SFG
as a tool for the close analysis of written text, and I return to SFG as
analysis tool and perspective later in the chapter.
For the BP research, although I started from the principle that
media representation does not reflect “the truth”, but is instrumental in
constructing one (albeit influential) version of the truth, I did not wish
to pursue an agenda for change from the outset, as is the declared aim for
CDA. It was important to me that I came into the research process with
as few preconceptions as possible. I fully expected to find evidence of
power and convention rooted in interest in the texts I studied. However,
in developing a research approach that started with the identification of
patterns in the texts, however derived, I intended a necessary critical per-
spective to follow, rather than precede, my analysis. I agree with Coupland
and Jaworski’s (2001: 145) observation that “In all but its blandest forms,
such as when it remains at the level of language description, discourse
analysis adopts a ‘critical’ perspective on language in use”. The research

approach I propose in this book offers a critical view of representation,
but one that emerges from its descriptive aims, rather than from an eman-
cipatory agenda. In its understanding of text and socio-cultural context,
my approach owes more to the work of Barthes than of Fairclough, and I
discuss in Chap. 3 a comparison of the Barthesian framework I draw on
and the Faircloughian conceptualisation of discourse.
Poststructuralist Discourse Analysis
Like other perspectives on the analysis of written language, Poststructuralist

Discourse Analysis (PDA) is an umbrella term to cover a range of research
approaches. The work of Foucault, in particular “The Archaeology of
Knowledge” (Foucault, 1972), is a significant influence for analysts in this
tradition, who share an understanding of the “opacity of discourse, neither
reducible to ‘langue’ nor to social or psychological instances” (Maingueneau
& Angermuller, 2007: np). Poststructuralist Discourse Analysts problema-
tise the view that meaning is centred in the speaking subject, rather arguing
that there is no hidden meaning to be uncovered in text, but that the prac-
tice of discourse constitutes objects, institutions and identities. In Foucault’s
words, discourses are “practices that systematically form the objects of
which they speak” (Foucault, 1972: 49). Writers in the tradition of PDA
(e.g. Angermuller, 2011, 2014; Baxter, 2002, 2008; Weedon, 1987) reject
the possibility of a definitive reading of a text, and the discourses that it may
encode, acknowledging reflexively within their theoretical approach the sit-
uated, deferred and partial nature of the analysis of discourse(s). However,
to accept this perspective is not to accept that there is an infinite number of
possible interpretations of a text, and “preferred readings” (Hall, 1980) can
be indicated through socially shared codes and contexts.
This centrality of discourse for Foucault did not translate into pro-
posed methodologies for the analysis of discourse; indeed he regarded
linguistics, with de Saussure, as the science of “langue” (the system of lan-
guage rather than language in practice). Nevertheless, PDA researchers
have drawn on a range of areas for language study to investigate discourse
from a poststructuralist perspective, including deixis, modality, tense,
speech acts and genre studies. For written text analysis, one recent area
35
of study, enunciative pragmatics, has been particularly productive for the

investigation of multiple subjectivities (Angermuller, 2011, 2014; Nølke,
2006, and see the ScaPoLine project). Building on the work of Bakhtin
(1984) and Ducrot (1984) on polyphony in discourse, researchers in this
area propose analytical models for recovering traces of the points of view
of multiple “enunciators” encoded in the linguistic output of the “locu-
tor” (speaker/writer). In this way, myriad subjectivities can be systemati-
cally explored in naturally occurring written texts.
I note the importance of poststructuralist writers (particularly Barthes
and Baudrillard) to my own semiotic perspective, and I share with PDA
an understanding of representations as partial, unfixable and subject to
the negotiation of writer and reader. However, in proposing a method of
discourse analysis drawing on the work of Barthes, I need to clarify that I
will deploy some semiotic concepts as frameworks for language investiga-
tion, rather than claim to realise his ideas as discourse analysis practice.
Barthes wrote extensively on the analysis of texts of all kinds, including
visual texts (1972), written fiction (1977b) and narratology (1977a). This
book does not extend these methods of textual analysis, but rather uses
Barthes’ conceptualisation of language as a starting point for the system-
atic examination of written representations at different levels of meaning.
Potter & Wetherell’s Discourse Analysis
An alternative approach to the investigation of written text is the form of

Discourse Analysis developed by Potter and Wetherell (1987). Discourse
Analysis does not adopt an overtly emancipatory agenda in the same way
as CDA, but is nevertheless a critical approach, concerned as it is with
repeated patterns of language which serve to entrench certain social posi-
tions. Discourse Analysis has its roots in Discursive Psychology rather
than linguistics, and was developed from studies by Gilbert and Mulkay
(1984) into how scientific research work has been constructed and posi-
tioned within the scientific community. Writers in the Discourse Analysis
tradition have more frequently used conversations and interviews as data,
but the approach has also been used in the study of written media texts
(e.g. Potter & Reicher, 1987).
Linguists of this school seek to identify “interpretative repertoires”,

which are “broadly discernible clusters of terms, descriptions, common-
places … and figures of speech often clustered around metaphors or vivid
images and often using distinct grammatical constructions and styles”
(Potter, Wetherell, Gill, & Edwards, 1990: 212). These clusters of lan-
guage usages are understood to be flexible and “selectively drawn upon
and reworked according to the interpersonal context” (Wooffitt, 2005:
154). Wooffitt is inclined to draw a distinction between Foucauldian
discourses and interpretative repertoires based on this variability, but
other writers who use this approach do not. For example, Talja (1999:
461) suggests that the two concepts of “interpretative repertoires” and
“discourses” are equivalent. Potter & Wetherell’s Discourse Analysis is
appropriate and frequently used for media investigation. Its definition
of interpretative repertoires chimes with my interest in varying language
choices by writer and genre.
Ethnography of Written Text
While a key area for ethnographic study is speech (consider Hymes’ (1974)
“SPEAKING” model), there is still an important role for ethnographic
approaches to writing, where the concern is to place writing outputs (writ-
ten texts) in their social contexts. Such written texts can include tradition-
ally “written-like” genres such as academic texts, institutional documents
and newspaper writing of all kinds, but also much less formal writing prod-
uct such as shopping lists and diaries, as well as the written-like/speech-like
output of SMS, chat rooms, Twitter and so on. The term “ethnography”
has been used simply as a shorthand for research approaches where subjects
are interviewed about their practices, and certainly the approach places
emphasis on the role of the investigator in the research, but ethnography
can make use of a great range of methods in its quest to understand the role
of written texts in the lives of their producers. As Cameron and Panović
(2014: 64) point out, “Ethnography is not so much a kind of discourse
analysis as an approach that may be combined with discourse analysis”.
Importantly, the ethnography of written texts addresses the social prac-
tices of writing and the effect they have on the shape and content of the
37
text. In the area of written media texts, for example, work has been carried
out on press releases (Jacobs, 1999, 2000a), showing how these are shaped
by their purposes. Catenaccio et al. (2011) and Van Hout and Macgilchrist
(2010) have studied the routines and practices of newsrooms in order to
explain the effect these have on the final texts. Their work considers the
complex intertextuality of the news writing process and makes use of diverse
ethnographic methods including interviews, observation, text analysis and
the use of keystroke logging software (Van Hout & Macgilchrist, 2010) to
build up a rich picture of how news stories are created. My interest in this
project was in the output of writing practices, the texts themselves, rather
than the processes, although the work of ethnographers informed in par-
ticular my understanding of the genres of media writing.
Computer-Mediated Discourse
Not included in my diagrammatic summary, but nevertheless impor-

tant, is the analysis of Computer-Mediated Discourse (CMD). I do not
regard CMD analysis as either a tool or a perspective for the analysis
of written text, but hold rather that the digital medium is a context for
discourse analytical work, and in this I differ from some other scholars.
Nevertheless, it is an increasingly important context.
CMD has as its unifying feature the digital medium, and computer-
mediated text can be accessed on a wide range of devices (computers,
tablets, smartphones and so on). Beyond this commonality of channel,
CMD has become so ubiquitous as to defy description as a distinctive
language variant (Androutsopoulos, 2006). Crystal’s (2001) early project
to define such (a) language variant(s) resulted in a tentative outlining of
what he called “Netspeak”, which he proposed as a hybrid of spoken and
written language. The work was swiftly overtaken from two main per-
spectives—firstly, the exponential increase in the use of the Internet in
almost every aspect of life made it almost impossible to isolate common-
alities of a broad Internet language variant, and secondly, Crystal’s work
at that time focused primarily on linguistic form rather than function in
context. Other early work focused on the speech-like qualities of written
asynchronous or quasi-synchronous computer conversation, often using
Conversation Analysis to study turn-taking, interruptions and other

conversational strategies. Such work has now been placed within the
broader study of online communities (del Teso-Craviotto, 2006; Herring
& Paolillo, 2006; Siebenhaar, 2006).
Although research work now tends to focus on social practices of
CMD, the affordances of the digital medium can still be said to fos-
ter certain characteristics of language. Properties such as potential ano-
nymity, global reach, virtual community and new linguistic freedoms all
have considerable impact on language usage. However, the use of digital
devices for such diverse social purposes as trade, advertising, education,
social networking and personal finance, to name only a few, entails that
language usages are extremely varied. Just as there is no single “newspaper
register” or language variant, there is no single “computer register” or
language variant. All of the tools and approaches mentioned in Fig. 2.1
have been used for the analysis of computer-mediated language. In addi-
tion, some tools and approaches have been developed and adapted for the
particular context of CMC, for example Androutsopoulos’ Discourse-
Centred Online Ethnography (2008), which uses mixed-methods ethno-
graphic approaches for the study of discursive identities on the Internet.
Researching Written Text in Semiotic Studies

Two analysis areas shown in Fig. 2.1 are outstanding—semiotics and
multimodality, and SFG, and I come to these now to illustrate how semi-
otics has addressed the task of analysing written language. I will argue for
an alternative to two typical approaches described below that will offer a
more flexible and broader approach.
The “Text-as-Graphic” Approach
In the study of semiotics, verbal language (written and spoken) is a sign

like any other, sitting alongside image, dress, gesture, music and so on as
a semiotic resource for making meaning. Verbal language constitutes only
a restricted part of communication. As Wilden (1987: 137) writes, “all
39
language is communication but very little communication is language”.

Verbal language has had more scholarly attention than any other sign
type because of its ubiquity, complexity and relative interpretability
(Kress & van Leeuwen, 2006). By interpretability, I mean that verbal
language offers a more complex, wider and better agreed potential for
meaning than other sign systems. Work in the area of multimodality (e.g.
Kress & van Leeuwen, 2001, 2006; Machin, 2007; Machin & Jaworski,
2006) contrasts language with other sign systems. Reviewing the case
for regularities of interpretation in visual and other modes, writers have
recognised that modes such as typography (van Leeuwen, 2005), colour
(Gage, 1999) and still photography (Harrison, 2003) have limitations
in the extent to which they can offer agreed meanings, even taking into
account cultural understanding and contextual cues. Critiquing Kress
and van Leeuwen’s (2006) project to outline a “visual grammar”, Machin
(2007: 186) argues that photographs, for example, have less clear com-
municative intention than language, no obvious separate components
(with which to form a “grammar”), an unpredictable set of rules that may
or may not be activated in each viewing, and an inability to communi-
cate, say, negatives, conditionals or questions effectively. In other areas,
of course, modes other than language have different and arguably richer
types of meaning potential—music, for example, communicates in quite
a different way from language (Monelle, 1999).
Until relatively recently there was a neglect of academic interest in
semiotic modes other than writing and speech, partly because of their
fluid and contested meanings, and partly because, prior to the rise in
digital communication, the role of image, sound and film in everyday
texts and mediums was restricted. The situation has changed consider-
ably, with multimodal communication becoming the norm in educa-
tion, personal communication, multi-language instructional texts and
many other areas in a way it had not been previously. This has meant
that research into modes other than speech and writing has been “play-
ing catch-up”, which in turn has entailed that in much multimodal work
based on semiotics, verbal text has been of interest primarily with respect
to other elements of a text, for example its positioning, its typography
and its formatting and in particular its relationship to images in terms
of sense-making, an approach I call the “text-as-graphic” approach. This
approach is encapsulated in Barthes’ (1977a) discussion of the relation-

ship between written text and image in multimodal representations,
primarily focusing on three relationships he terms “anchorage”, “illustra-
tion” and “relay”. He calls “anchorage” the relationship where the text
clarifies or explains the meaning of the image, directing the reader to
a particular desired reading of the image. Conversely, “illustration” is
where the image “anchors” the text, directing the reader to a particu-
lar desired reading of the verbal text. The third relationship of “relay” is
where the text and the image carry different (although complementary)
information—both text and image are important in the understanding
of the ensemble. Van Leeuwen (2005) describes and develops Barthes’
taxonomy of text-image relations by analogy with two types of clause
relations: elaboration and extension. He argues that Barthes’ “anchorage”
and “illustration” are types of elaboration, in that either text or image is
performing a role of specification. On the other hand, “relay”, where each
mode offers complementary information, is a kind of extension.
These conceptualisations do not have as their purpose the close analy-
sis of written verbal text. Nevertheless, they inform much work on mul-
timodal texts in which written language—for so many years the most
scrutinised semiotic system—is in this view important for its shape, font,
format and relationship with the image, rather than for its own patterns
and content. Works on multimodal analysis (e.g. Kress & van Leeuwen,
2001; Machin, 2007) tend to focus on visual rather than verbal modes,
and handbooks of written discourse analysis which include chapters on
semiotics and multimodality (for example Cameron & Panović, 2014;
O’Halloran, 2004) pursue this focus. O’Halloran’s overview is typical.
For example, in addition to linguistic choices and their typographical

instantiation on the printed page, multimodal analysis takes into account
the functions and meaning of the visual images, together with the meaning
arising from the integrated use of the two semiotic resources. (O’Halloran,
2004: 1)
This book, on the other hand, offers a different semiotic perspective on

written language, which uses the concept of the written sign as a starting
point for the explanation of a broader interpretation of meaning.
41
Systemic Functional Grammar
I have suggested that those studying multimodality can sometimes

consider written text primarily for its graphic qualities, yet many scholars
active in the field of multimodal studies would argue that when closer
text analysis is required, there is a fully developed functional grammati-
cal analysis framework to hand, namely the SFG proposed by Halliday
(Halliday & Matthiessen, 2004) and developed in work by Droga and
Humphrey (2003), Martin (2000), Thompson (2004) and others. The
grounding principle of SFG is that all verbal communication functions
simultaneously to represent the world (the representational/ideational
metafunction) and to establish and maintain relationships (the interper-
sonal metafunction), as well as exhibiting features which realise logical and
textual relations (the textual metafunction) (Halliday & Hasan, 1985).
The principle that the properties of verbal language are primarily func-
tional, rather than formal or structural, directly connects lexico-gram-
mar with wider contexts, and indicates that the expression of messages
varies according to both the communicative situation and the wider
socio-cultural context. Halliday makes these connections explicit in par-
ticular through his work on register where the significance of field (con-
tent), mode (channel, medium) and tenor (interpersonal relationships) is
investigated. The relationship between the text and the communicative
situation is extended further to the wider socio-cultural context by writ-
ers in the field of Social Semiotics (Hodge & Kress, 1988; van Leeuwen,
2005) and CDA (e.g. Fairclough, 1989), who use SFG to uncover how
linguistic choices are instrumental in establishing and maintaining power
relations which obtain in the wider socio-cultural context.
These principles have been applied effectively to modes other than
speech and writing. Kress and van Leeuwen’s (2006) key work on the
“grammar” of images uses Halliday’s communicative metafunctions—
ideational, interpersonal and textual—as the basis for their analysis of
the social functions of imagery. As an example, one assumption for tex-
tual organisation in SFG holds that in verbal language the start of the
sentence or tone unit is likely to offer “given” information, referring to
what we already know, and move towards “new” information at the end.
This left-to-right organisation (in Western writing) is used to explain the
t endency of images to present “given” information to the left and “new” (or
disputed, or p
roblematic) information to the right. As another example, in
the discussion of the potential meanings of colour, Kress and van Leeuwen
(2002) argue that colour can have a representational function (for example,
the use of colour in company logos—the green and yellow of the BP logo
stands for BP in the second of the two quotations which begin this book),
an interpersonal function (for example, pale green as a calming colour for
hospital decor) and a textual function (for example, the colour coding of
headings and sub-headings in school textbooks). In this way, SFG offers
unifying frameworks for the study of verbal text and other modes.
In Summary
A wide range of options exists for the analysis of written text, and I have
made a distinction between perspectives or approaches to written text
analysis, such as CDA or semiotic approaches, and the tools they use, such
as corpus linguistics or grammar analysis. Locating my own approach in
the area of semiotic and multimodal perspectives, I suggested that it offers
an alternative to two main perspectives on written texts which are preva-
lent in the field of semiotics. The first explores written text primarily with
regard to its visual properties and its relationship to other modes such as
images in multimodal texts. The second is SFG, which gives an account
of text at the semiotic level of the code (see Chap. 3). This book intends
to explore ways in which small signs (words) are connected to large signs
(sets of texts). My interest is in presenting a framework for analysis which
provides linguistic evidence of these connections. This conceptualisa-
tion can make use of SFG as an analysis tool while exploring ideological
implications, including notions of power, without a CDA transformative
agenda. Further, it can utilise other analysis resources to investigate such
linguistic phenomena as intertextuality, rhetorical tropes, naming choices
and discourses. The following section sets out the theory and practice of
this analysis framework.
Part II
A Barthesian Conceptualisation of
Written Language
3
Theoretical Foundations
Aspects of Roland Barthes’ Work

Roland Barthes’ work sits within the European tradition of “semiology”,
which took as its starting point the theories of Ferdinand de Saussure
(1959). De Saussure was concerned with identifying deep structures in
language. He understood languages as arbitrary systems in which signs
carry meaning only in their relation to other signs in the same system.
This entails that signs themselves have no fixed meaning, but meaning
derives from their differentiation from other signs. To explicate the fact
that day-to-day language can show extreme variation yet remain compre-
hensible, de Saussure posited the existence of an underlying and complete
system for each language—the langue—which finds varied expression in
the day-to-day usages of speakers—the parole. In his work on signs, de
Saussure proposed a dyadic model, consisting of a signifier, or the form
taken by the sign, and the signified or the concept referred to by the signi-
fier. He did not concern himself with the real-world referents that signs
may represent—both signifier and signified are psychological constructs.
De Saussure’s own work primarily focused on language as a sign system,

in Discourse, DOI 10.1057/978-1-137-58750-3_3
although the terms signifier and signified have been appropriated by later
semioticians in the discussion of visual, linguistic and other signs.
Barthes continued to explore the concept of the sign, but diverged
from de Saussure’s thinking in a number of respects. He did not conceive
of language as fixed langue and imperfect parole, but rather as a continual
process of meaning-making, where the word sign is inseparable from its
social context and thus constantly shifting: he refers to a “floating chain
of signifieds” (Barthes, 1977a: 39). Further, he broadened his investiga-
tions considerably from de Saussure’s focus on language, to myriad other
cultural phenomena and representations. In Mythologies (1972) he pro-
poses that such diverse phenomena as steak and chips, soap powder and
the hairstyles of Romans as depicted in films are understood as carrying
more than everyday functional meaning, and that such meaning is cul-
turally constructed.
The starting point of these reflections was usually a feeling of impatience at

the sight of the “naturalness” with which newspapers, art and common
sense constantly dress up a reality which, even though it is the one we live
in, is undoubtedly determined by history … I wanted to track down, in the
decorative display of what-goes-without-saying, the ideological abuse which,
in my view, is hidden there. (1972: 11, emphasis in original)
The fact that we think that the meaning of many everyday phenomena
remains at a denotative or literal level is a concern for Barthes, as he argues
that a range of additional meanings are encoded at a connotative level, in
other words that we bring to our interpretation of signs all the social, cul-
tural and personal associations we have collected and absorbed through
our lifetimes. These are the myths or “mythologies” to which the title of
his book refers, and he suggests that such cultural myths are commonly
perpetuated in the service of particular power interests. The less evident
the accumulation of associations becomes, the more ostensibly straight-
forward yet potentially deceptive is the sign. Barthes calls this process
naturalisation. This thinking led Barthes in his later work (1974: 9) to
suggest that there is no such thing as denotation, as it is impossible to
divorce any representation, however neutral-seeming, from the cultural
associations with which it has become invested.
Denotation is not the first meaning, but pretends to be so; under this
illusion, it is ultimately no more than the last of the connotations (the
one which seems both to establish and to close the reading), the superior
myth by which the text pretends to return to the nature of language, to
language as nature. (Emphasis in original)
Barthes argues here that the more naturalised the meaning (“what-goes-
without-saying”) the more denotative it appears, and this seemingly circular
process of apparently transparent to densely associative and back to appar-
ently transparent meaning is one that is of great interest in the study of repre-
sentations, particularly over time. If we accept his concept of naturalisation,
we can seek evidence for the naturalisation process in the language we analyse.
At the heart of Roland Barthes’ work is a concept of signs which is
holistic—where signs, including words, are inseparable from meaning,
and meaning is inseparable from text context and infused with social
context. His was an integrated view: the sign made no sense without a
knowledge of the codes or systems within which it was situated, these
codes being multifarious and including social, textual and ideological
codes, for example, language, dress or artistic genres. For Barthes, codes
are not neutral or “given” but rather constructed and political. They
naturalise—make self-evident—societal “myths”, which themselves are a
product of the dominant (or resistant) ideologies of the time and place in
which the communication is produced. The significance of this holistic
view—from individual sign to social ideology—is that, if it can be inves-
tigated through a practical analysis methodology, it has the potential to
offer a comprehensive description of how text(s) work to offer representa-
tions. Barthes’ terms can be used as a heuristic to organise the analysis of
texts. The different semiotic levels of sign, code, myth and ideology refer
to semiotic concepts at levels of increasing abstraction (from micro to
macro level), and can be directly attributed to the work of Barthes.
Sign, Code, Myth, Ideology

Barthes’ work, then, provides us with an epistemological starting point
for the understanding of textual representation. We will assume that a
discursive representation is realised at four levels of sign, code, myth and
Fig. 3.1 A semiotic heuristic for considering written language
ideology. Figure 3.1 represents this idea diagrammatically, but we have

noted that, crucially, each level is interdependent, so that signs can only
be understood in the context of codes, signs and codes can realise myths
and ideologies and so on. These four levels provide a framework within
which we can eventually locate the different levels of our text analysis,
which range from close examination of the text to a consideration of the
discourses that run through them which are connected to the culturally-
situated beliefs and attitudes that constitute ideology.
An analysis of the sort I propose does not seek to treat these different
levels as entirely discrete. However, it is useful for the sake of complete-
ness to envisage that we are building up a complete picture or “map” by
addressing relevant language issues at each level—looking at choices of
sign, paying attention to the codes in which they operate and moving
from there to build a picture of connotation, myth and ideology.
Chapters 8, 9, 10, and 11 will demonstrate what this means in terms
of the practical analysis of the BP data. This chapter gives theoretical
examples of how sign, code, mythic meanings and ideology can be con-
ceived specifically in the case of written language. Not all of the examples
I give here were relevant to my data set. In the methodology I propose,
an early stage consists of immersion in the data and it is at this point in
the research process that relevant language features are identified. It is the
data themselves that will drive the identification of relevant features at
each level.
The Level of the Sign
When I look at this first semiotic level of sign, I want to consider primarily
what de Saussure (1959: 128) calls the “isolated sign” or group of signs. (I
will go on to argue that entire representations themselves can be regarded
as “signs”.) If we conceive of language signs as “building blocks”, we can
discuss the meaning-making potential of individual words or groups of
words. This understanding of “sign” encompasses lexical choices of all
kinds, and these include, importantly, naming practices. Fowler (1991)
discusses how the names and attributions given to both people and events
shape how these people and events are understood, and in an exploration
of how crises are constructed through language, naming practices are of
central interest. Such practices help us to organise our world: “We man-
age the world, make sense of it, by categorizing phenomena” (Fowler,
1991: 92). Srivastva and Barrett (1988: 34–35) call attention to the fact
that the practice of naming entities is not simply organisation for conve-
nience, but can affect action towards the entities:
The process of giving language to experience is more than just sense-

making. Naming also directs actions toward the object you have named
because it promotes activity consistent with the related attribution it car-
ries. To change the name of an object connotes changing your relationship
to the object and how one will behave in relationship to it because when
we name something, we direct anticipations, expectations, and evaluations
toward it.
I have already mentioned in Chap. 1 that my referring to the BP events

as a “crisis” says something about my attitude and relationship to the
events, the other phenomena with which I would associate them, the
theoretical constructs I might use to explicate them (the business man-
agement theories I outlined, for example) and the way in which I would
wish others to understand them. In writing about crises, the importance
of naming strategies directed at both the events themselves and the par-
ticipants involved has been frequently explored (e.g. in Butchart, 2011;
Lischinsky, 2011; McLaren-Hankin, 2007). Because of their reach, the
media have an important role to play in what events and people come to
be called and this is frequently a site of struggle between the media and
organisations’ communications teams.
In the case of the BP data, I researched naming practices of two kinds.
First of these was the naming of the BP events themselves—how did writ-
ers refer to them and did this change? This was likely to be a fruitful area
for analysis for language research into any catastrophic event. The second
type of naming practice concerned reference to social actors—were these
identified or anonymous, individuals or part of a group, real or imagined?
In both cases of naming practices, it was possible to trace changes over
the span of the data, and these findings are outlined in Chap. 8.
An extension of naming practices concerns the association of an entity
(in this case the BP events) with other entities, a process I term “cat-
egorisation”. It can be valuable to explore the development of associative
relationships through an examination of listings and groupings. Once an
individual or phenomenon is named or labelled, it is much more easily
placed into groups of other entities with similar characteristics, in a pro-
cess of categorisation. The rhetorical device “classification”, as described
by Connor and Lauer (1985: 314–315), “involves putting the subject
into a general class and showing the implications of the subject’s mem-
bership of that class”. We draw inferences about the subject based on the
category in which it is placed. Fowler (1991: 58) examines the impor-
tance of categorisation in building consensus, and for organising and
managing the understanding of reality.
Experience is sorted into agreed categories in conversational exchange, and

these categories are then the “taken-for-granted” background in ongoing
conversation.
The point that these “taken-for-granted” categories are socially con-

structed is made strikingly in Foucault’s The Order of Things (1970: xvi)
where he cites Jorge Luis Borges’ hoax classification of animals from a
“Chinese Dictionary”. Borges’ classification of animals begins: “(1) those
that belong to the emperor, (2) embalmed ones, (3) those that are trained,
(4) sucking pigs, (5) mermaids, (6) fabulous ones”. The sheer unexpect-
edness of Borges’ groupings is intended to exemplify that our categori-
sations are not objective or natural, but rather agreed and constantly
reinforced through use. The categories commonly used in news writing

become a shorthand to positioning individuals and entities, and leading
the reader to understand them in particular ways.
News reporting makes a distinction between the categories associated with

hard news (politicians, experts, government representatives, business lead-
ers) and those with soft news (celebrities, “ordinary” people, occupational
groups, minority groups), implying a hierarchy of social positioning …
Categorisation is therefore a powerful way of naturalising social divisions
and hierarchies that are the effects of cultural and economic factors, includ-
ing the institutional conventions of media reporting. (Fulton, Huisman,
Murphet, & Dunn, 2005: 249)
The implication of Fulton et al.’s observation is that categories are

intended to be understood as relatively homogenous. To take politicians,
for example, assumptions are made about characteristics shared by the
group, for example, their warrant to speak, their likely knowledge about
certain topics, an assumed political agenda. Readers are expected to have
a shared conception of what a “politician” is like, built up through media
and other texts, and this will affect their understanding of the category
“politicians”.
So the processes of naming, defining and grouping are crucial to how
human beings make meaning of their surroundings and communities.
These processes are not neutral acts of organisation, but are socially
agreed, and indicate how we choose to or are expected to respond to enti-
ties and phenomena. When we conceive of a list or group, we locate enti-
ties according to our understanding of the world. This location of a given
entity alongside other entities says much about which aspects of their being
we perceive to be important, and in what context we wish them to be
understood. An analysis of lists, groups or categories involves considering
which aspects of our entity are foregrounded in the process of aligning it
with other group members. My intention was to explore where the BP
events were located in relation to pre-existing phenomena, and whether
and how this changed over time.
These three processes of naming of events, naming of participants and
categorisation are important indicators of how the BP crisis is c onstructed
in news media language. In de Saussure’s terms, analysis of these three

features over time can show us one way in which changes in the signifier
can alter our understanding of the signified.
The Level of the Code
Individual signs can only have meaning within a system, and the concept
of “codes” in semiotics refers to the many types of system that provide the
necessary framework for our understanding and interpretation of signs.
Since the meaning of a sign depends on the code within which it is situ-
ated, codes provide a framework within which signs make sense. Indeed,
we cannot grant something the status of a sign if it does not function
within a code. (Chandler, 2001: np)
Codes are specific to places and times, but are widely shared within a
culture. Because of this, they can appear entirely natural, and so virtually
invisible. They provide a set of “rules” to aid the interpretation of signs,
and in this way they can both guide and restrict the understanding of
texts. In other words, although multiple interpretations of signs are avail-
able, readers are likely to find a “preferred reading” (Hall, 1980: 124)
based on their knowledge of the code within which the sign is presented,
as well as other contextual cues. There are not limitless possibilities for
interpretation—we interpret according to the codes to which we have
access. Chandler (2001: np) suggests a broad taxonomy for codes.
• Social codes. Language codes of all kinds, including grammar.

Paralanguage, gesture, gaze. Fashion. Behaviour.
• Textual codes. Scientific codes, including mathematics. Genre, aes-
thetic and stylistic codes. Mass media codes, including those applica-
ble to photography, film, newspaper.
• Interpretative codes. Perceptual codes (e.g. of visual perception).
Ideological codes such as feminism, capitalism, materialism (and resis-
tant codes: anti-feminism, anti-capitalism and so on).
Under textual codes are the special codes we recognise as belonging to

particular genres. The understanding within newspaper publications

of what constitutes a newsworthy story, which I mentioned earlier, is
an example of a genre-specific code, but generic codes can be stylistic,
structural or technical. We can also include under “codes” the notion of
intertextuality or the influence of other texts upon any given text. The
implications of intertextual relations are so broad that they can be con-
sidered at the levels of sign, code, myth and ideology, but we can practi-
cally place accounts of intertextuality at the level of code. This is because
the network of other texts and other voices are just one of the systems
or frameworks within which we locate, understand and interpret indi-
vidual signs. As Chandler writes, “every text and every reading depends
on prior codes” (2001: np). There is also a close relation between genre
and intertextuality which justifies them being discussed as interdepen-
dent theories. One aspect of intertextuality is the concept that prior texts
constitute what we recognise as genre, and created texts serve to construct
and modify our future understanding of genre (Bhatia, 2002; Johnstone,
2008). Finally, under language codes, Chandler points to systems such as
syntax, phonology, prosody and so on as interpretative frameworks. Thus
the analysis of grammatical choices and systems can also be located at the
semiotic level of “code”. A more detailed account of genre, intertextuality
and the grammatical system of modality follows.
Genre
The concept of genre in language texts has been borrowed from that of
genre in literature and art. While art forms have long been readily labelled as
belonging to a certain genre (cartoon, Romantic poetry, detective novel and
so on) considerable recent academic attention has been paid to how institu-
tional and other written genres might similarly be identified. I have already
assumed broad agreement on the interpretation of a “genre” in my overview
of news media genres in Chap. 1. Swales (1990), Bhatia (1993) and Dudley-
Evans (1994) suggest that genres can be characterised by commonalities
across a range of aspects, including purpose, audience, structure, content and
features of style. Bhatia’s (1993: 13) definition of a genre is typical:
It is a recognisable communicative event characterised by a set of communicative

purposes identified and mutually understood by the members of the profes-
sional or academic community in which it regularly occurs. Most often it is
highly structured and conventionalised with constraints on allowable contribu-
tions in terms of their intent, positioning, form and functional value. (My
emphasis)
More recent writing on genre (Bazerman, 2004; Beghtol, 2001; Bhatia,

2002, 2004; Herring et al. 2004; Kessler, Nunberg, & Schűtze, 1997)
has expanded the area of scholarly interest from the relatively narrow
“professional or academic” contexts referred to by Bhatia above to a much
wider understanding of genre as a type of text regardless of institutional
status. Commonalities of purpose are partly identified through an under-
standing of what type of rhetorical act or generic value (Bhatia, 2002) is
realised in a text, for example, argument, narrative, description, explana-
tion, instruction, persuasion, evaluation. In the area of news media, two
particularly prominent rhetorical acts are those of description (in news
reports, for example) and evaluation (in editorials and letters). Texts that
are primarily evaluative might also feature the rhetorical acts of argu-
ment, persuasion or instruction. These overarching rhetorical aims will
affect the kind of language choices made within texts.
Defining what counts as a genre can be complex: Yates and Orlikowski
(1992: 303) point out that genres exist at various levels of abstraction, so,
for example, a business letter is of a different order of genre from a letter of
recommendation. Bhatia (2004: 59) proposes a comprehensive hierarchy
of genre from genre colony (his example is promotional genres) to genre,
defined by specific communicative purpose (e.g. advertisements, book
blurbs and so on) to sub-genres, definable by medium (TV, print), and/
or product (car, holiday) and/or participants (business travellers, holiday
travellers). An equivalent for our study of news texts would be the genre
colony of journalism or news writing, the genres of news story, editorial,
feature and so on, with sub-genres by medium (print, radio, TV), “prod-
uct” (travel features, movie reviews) and participants (businesspeople,
non-professionals). Considering newspapers as a particular type of over-
arching genre, Hoey (2001) labels these as “colonies” with embedded
“sub-colonies”. According to Hoey, colonies have the characteristics that
they are not in order and do not form continuous prose. They need a
framing context (e.g. a title), they tend to acknowledge either no or mul-
tiple authors and have components that may be accessed separately, with
many components serving the same function.
Despite varying positions on the study of genre, writers tend to agree
that genres are fluid and permeable. There may be prototypical charac-
teristics, and these are isolated and foregrounded in texts such as tem-
plate letters or CVs, or in spoofs or parodies of film or television genres.
However, most texts exhibit some generic characteristics while remaining
unique. Genres are subject to continuous change. Not only does each
instance of a text written in a certain genre serve to construct an ever-
changing interpretation of that genre, but new communication needs,
new technologies and creative play with generic texts all play a part in
redefining the generic landscape. The fact that the boundaries of genres
are permeable is shown in the prevalence of “genre-mixing” and “genre-
blending”: respectively, the former being the overlap between genres (e.g.
business leader reports that have both an informative and a motivational
function) and the latter being the deliberate use of two or more genres
to create a different kind of text. This is common in many areas, for
example, film (the comedy-western, the sci-fi thriller), and advertising
and business, for example, advertorials (Cook, 2001; Fairclough, 1995b),
and most recently “advergames” (BBC, 2014). Forms arising from new
technologies, such as web pages or blogs, may draw on prior genres, such
as newspapers, posters or personal journals, but introduce new character-
istics that are appropriate to the medium, the audience and the purpose
of the text (Herring et al. 2004; Johnstone, 2008).
The importance of an understanding of genre in studies of representa-
tion cannot be overstated, and there are two overriding considerations for
analysis. Firstly, an analysis of language cannot be divorced from context,
be it media language, academic language, business, private, healthcare
discourse and so on. The job writers are doing with language and for
whom shapes the language in every way. It is an invaluable first step for us
to know, as discourse analysts, whether the text we are studying is typical
or not of its genre and why it is or is not. This consideration feeds into the
synchronic analysis of texts, as the relative dominance of any one genre
will imbue the overall data set with its particular characteristics.
The second consideration becomes clear in a diachronic analysis. If

a set of texts (a representation) consists of a number of genres (in the
BP case, say, news articles, financial reports, editorials, features) then
any change in the relative importance of the genres over time will bring
with it a change in the language profile. To take an example, editorials
are characterised by various linguistic markers constructing a personal
and subjective perspective. Hard news articles typically exhibit markers
constructing an impersonal and objective stance. An increase in the fre-
quency of subjective markers at the expense of objective markers is likely
to be due to a generic shift from hard news articles to editorials, rather
than to a shift in perspective on the part of any given writer. Thus text
genre in the data set will need to be named and identified at a broad level,
as well as explored at text level.
Intertextuality
The study of intertextuality explores the relationship of texts with other

texts that inform and shape them, and texts that they will in turn influ-
ence and shape. As Allen (2011: 1) explains:
Meaning becomes something which exists between a text and all the other
texts to which it refers and relates, moving out from the independent text
into a network of textual relations. The text becomes the intertext.
The term “intertextuality” and the foundation of the area of study are the
work of Kristeva (1980) following principles set out by Bakhtin (1981).
Bakhtin proposed that all utterances (and writings) are infused with the
traces of previous utterances, as well as with the anticipation of a response
from the audience. Both writers suggest that all texts are dialogic, respond-
ing to and being shaped by multiple previous texts of all kinds, and antici-
pating the response of either a present or an imagined receiver.
This view of intertextuality suggests that all the texts read, spoken or
written by the writer, as well as conventions of genre, and the constraints
of social practices are potential influences on any given text. For the
purposes of our analytical work, this means finding ways in which to
acknowledge the role of any past (and potential) text on our researched
texts. Seen in this way, the notion of intertextuality is so diffuse as poten-
tially to defy analysis. Allen (2011: 59) raises the problem of definition
and understanding: “Is intertextuality a manageable term, or is it essen-
tially unmanageable, concerned with finite or infinite and overwhelming
dimensions of meaning?”
Research into analysis of intertextuality must account for the fact that
textual influence can be taken to exist in myriad forms: from the most
overt and specific (e.g. an exact sourced quotation in an academic essay) to
the virtually untraceable, for example, general shared language resources,
a 50-year-old conversation, or a commonly held belief. At this very non-
specific end of the spectrum we can see the role that these almost uncon-
scious usages play in constituting Barthes’ “what-goes-without-saying”.
Conceptual frameworks for intertextuality have frequently reflected
the fact that the majority of input from other texts will be irretrievable.
Bazerman (2004: 83) uses the metaphor of a “sea of words”.
We create our texts out of the sea of former texts that surround us, the sea of
language we live in. And we understand the texts of others within that same
sea. Sometimes as writers we want to point to where we got those words from
and sometimes we don’t. Sometimes as readers we consciously recognize
where the words and ways of using words come from and at other times the
origin just provides an unconsciously sensed undercurrent. And sometimes
the words are so mixed and dispersed within the sea, that they can no longer
be associated with a particular time, place, group, or writer.
Expanding on this final observation, Bazerman (2004: 87) refers to “[b]

eliefs, issues, ideas, statements generally circulated”. The concept of texts
that cannot be pinned down was articulated by Riffaterre (1980: 239),
who acknowledged that not all intertexts would be identifiable, but that
it would be sufficient for analysis that we recognise that they exist: he calls
this “the presupposition of the intertext”.
Studies of intertextuality have offered diverse perspectives and clas-
sification frameworks which have aimed to identify this most elusive of
language phenomena. Kristeva (1980: 65–66) envisaged two axes, one
running from subject to addressee and the other from text to context.
The coincidence of these axes shows, according to Kristeva, that “each

word (text) is an intersection of word (texts) where at least one other
word (text) can be read”. Unlike Kristeva, whose work was mainly set
within a poststructuralist paradigm, Genette (1997) pursued the study
of intertextuality from the structuralist perspective that texts are part of a
(closed) system, and that it is the job of the analyst to uncover their place
in the system. Genette’s five types of “transtextuality” related primarily to
literary texts. More recently, Bazerman’s (2004) work is directed towards
the practice of discourse analysis of a broad range of texts, including
media discourse. He outlines six types of intertextuality, proposing them
as points along a cline, moving from an extremely explicit to an extremely
broad understanding of the phenomenon.
I discuss here finally Fairclough’s (1992a, b) taxonomy of intertextual-
ity, which is broader than those mentioned so far. Fairclough makes a
distinction between “manifest intertextuality”, which encompasses more
or less explicit allusion to other texts, and “constitutive intertextuality”,
which draws upon conventions of different genres and discourses.
I shall draw a distinction between “manifest intertextuality”, where specific

other texts are overtly drawn upon within a text, and “interdiscursivity” or
“constitutive intertextuality” … On the one hand, we have the heteroge-
neous constitution of texts out of specific other texts (manifest intertextu-
ality); on the other hand, the heterogeneous constitution of texts out of
elements (types of convention) of orders of discourse (interdiscursivity).
(Fairclough, 1992a: 85)
Fairclough’s approach is more consolidated than either Genette’s or

Bazerman’s, and points to a fundamental difference between what might
perhaps be termed “conscious” and “unconscious” borrowing. Fairclough
writes that certain types of “manifest intertextuality” are also a feature of
certain types of “interdiscursivity”. To take an example from news media,
we might assume a directly quoted witness statement in a news report to
be a clear case of manifest intertextuality—as it is a “specific other text
[…] overtly drawn upon within a text”. Indeed, Fairclough cites “dis-
course representation” (or “speech reportage”) as an instance of manifest
intertextuality. However, the witness statement is so common in the news
report genre as to be considered an interdiscursive feature—a marker of

this particular type of writing—and so is also in some way constitutive.
Both of Fairclough’s categories refer to intertextual relations that are rela-
tively retrievable, either as recognisable prior texts or as identifiable char-
acteristics of genre or discourses, and he does not move beyond this to
general characteristics of language. Fairclough’s notion of “interdiscursiv-
ity”, referring to discourses as well as to conventions of genre, relies upon
his belief that discourses can be retrieved and identified systematically
(1989, 1995a), even if not always linked with specific texts.
The approaches above, summarised and simplified, indicate three
broad areas of intertextuality:
• Intertextuality as irretrievable texts. Historical and current language

resources, beliefs, issues ideas, “presupposed intertext”.
• Intertextuality as identifiable style/register/genre. Generic and stylistic
characteristics of both form and content.
• Intertextuality as retrievable texts. Reported voices, quotations, allu-
sions, parody.
These categories considerably simplify the complexity of intertextual

research, but provide a starting point from which to consider a frame-
work for media text analysis, which I will expand on in a later chapter.
I have already discussed the second category of generic characteris-
tics, noting that genre and intertextuality are closely interconnected.
Retrievable texts (the third category above) that regularly appear in
the mass media include reported voices, which have been explored in
depth in work by Fairclough (1995b), Fowler (1991) and van Dijk
(1988). Jacobs (1999, 2000a) has written extensively about the role of
press releases as contributing texts. The role of images in news report-
ing has been explored in the work of Bignell (2002), van Leeuwen
(2005), Kress and van Leeuwen (2006) and Machin and Jaworski
(2006). Other retrievable texts include official reports of all kinds, pre-
vious news reports and the idiosyncratic use of cultural quotation and
allusion by individual journalists.
Grammatical Codes: Modality and the Appraisal System
One of the most pertinent types of code for the study of written verbal
text is that of grammatical systems. The organisation of language signs
into grammar systems has been conceptualised in many diverse ways, and
Chap. 2 mentions two—the formal approach characterised in the work
of Chomsky (1965) and others and functional approaches such as that of
Halliday (Halliday & Matthiessen, 2004). Within the theories which aim
to offer comprehensive accounts of the structures of languages, are myr-
iad sub-systems, each of which are candidates for analysis, as potentially
revealing about representation choices. The researcher might identify as
revealing a study of verb voice, for example, or clause types, or cohesive
devices. One sub-system which emerged as meaningful for the BP data
was modality. Perhaps unsurprisingly, the expression of doubt, certainty,
(in)ability, willingness and necessity was a fruitful area for investigation
in media texts about crisis (although, as it turned out, modality usages
did not conform to my expectations). Given this, I give here an account
of modality as just one exemplar of a grammatical system, otherwise
called a textual code.
Modal expressions are one of the resources used by journalists to signal
their commitment to the validity of the propositions they make in their
news reports, and their presence or absence can be a marker of subjec-
tive or objective styles. At its simplest level, the system of modality is
said to “construe the region of uncertainty that lies between yes and no”
(Halliday & Matthiessen, 2004: 147). The resources of the modal system
allow the speaker or writer (in the case of this data, the journalist, blogger,
etc.) to interpose his/her own judgements about the propositions they
are making. For Halliday, the modal system is part of the interpersonal
function of language, that is, the resources used by language to facilitate
interpersonal communication. Halliday and Hasan (1976: 27) character-
ise the interpersonal component of language as “the speaker in his role
as intruder”. However, this “intrusion” can be realised in ways that allow
speakers and writers to associate themselves with or dissociate themselves
from responsibility for the views they are expressing. This is referred to as
“orientation” in Halliday’s terms (Halliday & Matthiessen, 2004: 619),
or, more usefully I feel, as “modal responsibility” by Thompson (2004:

69). Degrees of modal responsibility can be expressed along two dimen-
sions—objective–subjective and implicit–explicit—and these dimen-
sions can provide useful ways of commenting on degrees of engagement
of the journalist with his/her materials.
This “region … between yes and no” contains different sorts of uncer-
tainty, in the areas of probability, usuality, obligation and inclination, and the
resources in the English language used to express these four areas of interper-
sonal perspective are substantial. Writing just about probability and usuality,
Chen (2010: 28) suggests that “over 350 lexical devices are found and used
for expressing doubt and certainty in English”. These resources include modal
auxiliaries, adverbials, adjectives, nouns in modal phrases and clauses. While
modal auxiliary verbs, such as “would”, “ought to” and so on, are a relatively
restricted group, many other verbs can be used with modal import, for exam-
ple, “try to”, “hope to”, “refuse to” and so on (Roberts, Zuell, Landmann,
& Wang, 2008). Within each of these grammatical realisations, and for
each type of modality, there are also degrees of strength, from high through
medium to low. So, for example, in the area of probability, the following are
possible. A mix of grammatical realisations is shown for each:
High modality: certainly, surely, certainty, it is definite that.

Medium modality: probably, in my opinion, fairly, tend to suggest.
Low modality: possibly, possibility, could, would appear to suggest.
The choice of grammatical realisation is meaningful in itself, for exam-

ple, modal clauses and phrases can suggest the writer/speaker is less of
an “intruder” than some other grammatical choices, for example, modal
auxiliary verbs, as Droga and Humphrey (2003: 61) argue:
Modal clauses and phrases (interpersonal metaphors) are a more indirect

way of expressing modality and are therefore often used to make texts seem
more objective and difficult to argue against.
Choice of modal expression is part of a mix of linguistic strategies that

can signal the construction of objectivity or subjectivity, and patterns of
modal choice vary by news media genre.
The modal areas of probability, usuality, obligation and inclination

can be defined in terms of two main types of modality: what Jespersen
(1922) calls “propositional modality” and “event modality”. Propositional
modality, which consists of epistemic and evidential modality, expresses
the speaker’s judgement of the proposition (probability and usuality). Of
the two types of propositional modality, epistemic modality is by far the
more common, with evidential modality found to occur relatively rarely
(de Haan, 1999). In propositional modality, there is no element of will.
On the other hand, event modality, which comprises deontic and dynamic
modality, expresses the speaker’s attitude to a potential future event and
does contain an element of will. Deontic modality includes permission
and obligation, while dynamic modality is concerned with ability and
willingness or inclination. These three main types of modality—epis-
temic, deontic and dynamic—provided a categorisation for the analysis
of modality in the BP texts.
A linked perspective on writer judgement and the related issue of sub-
jectivity and objectivity is offered by work in Appraisal Theory. Appraisal
Theory has been most fully developed by Martin and White (2005) and
has at its core many of the same issues as the study of modality. Indeed,
Halliday and Matthiessen present appraisal categories alongside modality
categories in direct correspondence. Martin and White (2005: 1) describe
the topic as follows:
This book [the Language of Evaluation] is concerned with the interpersonal

in language, with the subjective presence of writers/speakers in texts as they
adopt stances towards both the material they present and those with whom
they communicate.
Where Appraisal Theory offers additional analytic resources to modality

studies is in the consideration of lexis, and here an overlap between the
sign (word) and the code (system) is evident—although Appraisal Theory
is characterised by Martin and White (2005: xi) as a “framework”, and
therefore a system. The framework examines evaluative lexis expressing
attitudes of three broad kinds: affect, which deals with emotional reac-
tions, judgement, which concerns the evaluation of behaviour and ethical
issues, and appreciation, which relates to aesthetic evaluation. Appraisal

Theory systematises ways of analysing the force (how strong) and the
focus (how typical) of terms used in texts of all kinds. It also provides
ways of assessing the engagement of the speaker or writer by looking
at processes of attribution, modality, proclaiming and disclaiming. This
analytical procedure has been used effectively to comment on the ways
crises have been depicted in the media (e.g. White, 1997).
The Level of Mythic Meanings
The next level in the semiotic heuristic that structures this analy-
sis approach is that of “mythic meanings”. Myths, in the sense that
Barthes (1972) conceived them, are the ideas, beliefs and attitudes
shared by cultural groups that generally go unremarked, so taken-
for-granted are they. One of Barthes’ endeavours was to uncover and
analyse these taken-for-granted myths, or “what-goes-without-say-
ing” (1972: 11), and thereby make clear the ideologies and interests
that they support and perpetuate. Barthes recognised myths not only
within verbal language (he argued that myth is itself a language) but
in many forms of cultural practice which we do not readily recognise
as ideologically informed. In fact, he purposely selected objects for
study that were as far as possible from literature and the literary appli-
cation of the word “myth”.
Although semiotic “codes” have been presented as being at a different
level in the hierarchy from mythic meanings, this does not imply that
their constitution does not serve an ideological purpose. On the contrary,
as outlined earlier, the frameworks within which we understand signs
are culturally and temporally specific, and thus potentially as subject to
myth-making as all communication practices. It is simply a convenience
to discuss a semiotic view of language by following a line of thought:
building blocks → systems → meaning potential → ideology behind
meaning. Put simply, both building blocks and systems in language are
already myth and ideologically laden.
Connotation
One of the ways in which Barthes was able to expose and describe
myth was by making a distinction in his early work between two types
of meaning for signs, namely, “denotation” and “connotation”, and I
described how “denotation” can be considered the literal, face-value
“meaning” of a sign and “connotation” the additional associations, both
socio-cultural and personal, that the sign generates for the person who
perceives it. Barthes conceptualises this such that the signifier and signi-
fied, being the denotative meaning, together form another signifier, to
which the receiver of the sign attaches a further signified; this “second-
order semiological system” is termed “myth” (Barthes, 1972: 114). This
process can lead to an infinite chain of signification—Barthes’ “floating
chain of signifieds” (1977a: 39).
Barthes argues not that denotative and connotative meaning are on
a continuous spectrum and difficult to disentangle (although this may
also be true) but rather that the ultimate connotation may appear to be
the most “innocent”, to use Barthes’ term (1972: 61), by being appar-
ently the least burdened with non-literal association. As Lacey (1998: 68)
writes: “Myths are connotations that appear to be denotations.” Indeed,
it has been argued that there are no denotative meanings at all, as “lit-
eralness” and “simplicity” are as much constructs as representations that
appear to have complex and diverse meaning. Hall (1980: 122) suggests
that the distinction between denotation and connotation is no more than
an analytical convenience, with denotative meanings simply being those
that attract a wider consensus than connotative meanings.
It is useful, in analysis, to be able to apply a rough rule of thumb which

distinguishes those aspects of a sign which appear to be taken, in any
language community at any point in time, as its “literal” meaning
(denotation) from the more associative meanings for the sign which it
is possible to generate (connotation). But analytic distinctions must
not be confused with distinctions in the real world. There will be very
few instances in which signs organized in a discourse signify only their
“literal” (that is, near-universally consensualized) meaning.
My earlier observations about the construction of fact and objectivity

in the media serve to illustrate this point. What is presented as a simple
description of reality is mediated through a set of discursive practices
that we have learned to interpret as simple description. In effect the news
report genre presents itself as denotation, where, in Barthes’ terms (1974:
9), it may be “no more than the last of the connotations”.
If we accept Hall’s analytical convenience in order to consider language
which is more connotative than denotative, we are dealing with language
as it communicates in a figurative rather than a literal way. Here language
is not primarily representative, but has an intended alternative mean-
ing beyond “what it says”. In the study of rhetoric, figurative language
can belong to one of four main tropes: irony, metaphor, metonymy and
synecdoche. Each trope represents a desired meaning via a different and
non-literal relationship between word (signifier) and concept (signified),
and these relationships are understandable through codes. Again, any or
all of these figurative tropes might prove of interest in the study of a par-
ticular data set, as indeed might connotative meanings which do not fall
into the categories of these tropes. In the BP data, metonym, synecdoche
and metaphor emerged as significant, and I use these as illustrations of
language features at the level of mythic meaning.
Metonym and Synecdoche
Both metonym and synecdoche are forms of language that create mean-
ing by association. This association is seen in part–whole relationships, for
example, “two heads are better than one”, where “heads” stands for “peo-
ple” or “people’s ideas”, and in relationships where an associated entity
stands for the actual referent, for example, “the White House made a
statement today” where “the White House” stands for the US president
or his/her spokespersons. The metonymical relationship covers a very
broad range of actual instances, but its distinguishing feature from meta-
phor is that both entities (the metonym and the referent) are drawn from
the same domain, or semantic field. In metaphor, on the other hand,
one entity is described in terms of another from a different domain.
Some linguists, including Lakoff and Johnson (1980), have argued that
metonymy is simply a type of metaphor. There are certainly areas where
the two interact, and the work of Goossens (1990) theorises “metaphor
from metonymy” and “metonymy within metaphor”, also coining the
word “metaphtonomy”.
The distinction between metonym and synecdoche is not always agreed,
but many linguists (e.g. the early Jakobson [Jakobson & Halle, 1956])
would see synecdoche as falling under the larger heading of metonymy.
In my own work, I agree with Lock’s (1997: 323) droll assertion: “I shall
follow the early Jakobson and treat synecdoche as a synecdoche of meton-
ymy.” Synecdoche is generally considered to be a part–whole or whole–
part relationship, while metonymy also includes other relationships of
association. These examples are taken from Radden, Köpcke, Berg, and
Siemund (2007) and Wales (1989):
PLACE FOR EVENT— “he was shocked by Vietnam”—metonymy.

OBJECT FOR USER—“the sax has flu today”—metonymy.
AUTHOR FOR WORK—“I love Proust”—metonymy.
PART FOR WHOLE—“strings [stringed instruments]”—synecdo-
che → metonymy.
WHOLE FOR PART—“England thankful to avoid serious injury”—
synecdoche → metonymy.
The fact that metonymy (used henceforth as the umbrella term)

draws both of its elements (the signifier and the signified) from the same
domain, and that the two are already associated in experience, can sug-
gest that metonymy is somehow less figurative and more realistic than
metaphor. Jakobson (Jakobson, 2002; Jakobson & Halle, 1956) argued
that the tropes are quite different: his metonymic “pole” associated met-
onym with prose and writing in a realistic tradition, whereas metaphor
was associated with romanticism and invention. Metonymic usages may
be seen as less creative than metaphoric ones, but they share with meta-
phor the fact that they select certain aspects of the signified to foreground,
and discard others. For example, both proverbs “Two heads are better
than one” and “Many hands make light work” use parts of the body
synecdochically to mean “people”, but select those parts that are most
relevant to the meaning (connections with “intelligence” and “strength/
skill”, respectively). This selection can be meaningful, and can direct the
audience’s attention to particular desired readings, while discarding or
suppressing others, and these emphases can be ideologically significant.
In his review of metonymy in a corpus of business texts, Cornelissen
(2008) finds the metonym ORGANISATION FOR MEMBER (e.g. “BP
announced …”) to be widespread and significant, and discusses how it
works together with the common business metaphor A COMPANY IS A
HUMAN BEING, to direct the reader to envisage that the company is a
person with single goals and purposes who can speak with a unified voice.
This selective direction is of particular significance in journalism, where
the principle of economy of expression is important (Bhatia, 1993; Biber,
Johansson, Leech, Conrad, & Finegan, 1999).
Metaphor
Unlike metonymy, through which signifier and signified have a relation-

ship of association or contiguity based on a single domain, metaphor pro-
poses a relationship of likeness between the target (entity being described)
and the source (entity used to describe) across different domains. It is this
relationship of likeness which caused Peirce (1931–58: 2.277) to classify
metaphor as an Iconic sign: a view that I will problematise in Chap. 13.
Metaphors are extreme instances of figurative language, and as such can
be a source of innovation and creativity. Van Leeuwen (2005) describes
metaphor as one of the two main agents of semiotic change, alongside
connotation. It is this aspect of metaphor that Jakobson emphasises when
he writes that the metaphoric pole is the dimension relevant to poetry,
surrealism and romanticism. This drive to innovation is clearly evident
in the case of fresh metaphors that use unexpected juxtapositions to shed
new light on existing concepts. Yet the broadest definition of metaphor—
describing one entity in terms of another—encompasses a vast number of
metaphorical usages that are so commonplace that their literal meaning
is no longer recognised.
These “taken-for-granted” usages are the subject of Lakoff and Johnson’s
“Metaphors we live by” (1980) in which the authors argue that most
metaphors we use are not unusual, creative and poetic, but rather funda-
mental ideas, often with an experiential basis, that structure our thinking
in particular shared ways; they are at the very root of how we conceptual-
ise experiences, entities and practices. Once a metaphor is shared widely
enough, it ceases to have the power to make us understand experience in
new ways, but rather reinforces shared ways of seeing (and in that sense
has a mythical or ideological dimension). Because of this ubiquity, it can
become difficult for the analyst to isolate what is a metaphorical usage
and what is not. To illustrate this problem, Lakoff and Johnson call forth
an imaginary “objectivist” who argues with them that in many cases a
word used in a metaphorical sense is actually a homonym of the literal
term with an alternative dictionary definition.
Researchers can choose to view these commonplace metaphors in two
ways: one is that they are so routinised as to simply have the status of
a literal meaning. The other view would hold that such metaphorical
usages should be identified and scrutinised critically; on account of this
very ability they have to normalise a way of seeing that is a product of
society and culture, and that may reinforce power relations of certain
kinds. This is an argument made by Koller (2003) in her examination
of how common metaphors in business can legitimise unequal gender
relations. Like metonyms, metaphors are selective in the sense that they
choose one aspect of an entity or phenomenon to describe by compari-
son with another entity. This aspect becomes the focus of attention, and
excludes other aspects of the phenomenon. This concept is illustrated
in the Milne, Kearins, and Walton (2006) article, which shows how the
metaphor BUSINESS IS A JOURNEY is used by organisations in con-
nection with their environmental commitments. By conceptualising
business in this way (rather than as, say, WAR or SPORT), organisations
can imply that they are making progress, without committing to a final
goal. In this case, metaphors do not shed new light on abstract issues but
rather delimit them. As Tsoukas (1991: 582) writes
Metaphors tend to be used as substitutes for deeper knowledge, and they

tend to be constitutive of, and prescriptive in relation to, the social phe-
nomena they are connected with.
Tsoukas makes a three-way distinction between “dead”, “dormant” and

“live” metaphors. His definition of dead metaphors broadly corresponds
to Lakoff and Johnson’s “conventional metaphors”:
Frozen or dead metaphors have become so familiar and so habitual that we

have ceased to be aware of their metaphorical nature and use them as literal
terms. (Tsoukas, 1991: 568)
The category of dormant metaphors is potentially useful for analysis, in

that it encompasses terms that are widely used, with commonly shared
meaning, but that are still recognisable as metaphorical:
Dormant metaphors are quasi-literal terms through which we restrict our-

selves to seeing the world in particular ways; however, the metaphorical
nature of these terms can be easily exposed. (Tsoukas, 1991: 568)
Dormant metaphors can be reawakened to shed productive new light on

the target, or can go on to be dead metaphors, only used in a literal sense.
Live metaphors are those that are more unusual and creative, where the
writer has as his/her purpose to encourage the reader to conceptualise one
entity in terms of another in a way that gives unconventional insight into
the target domain (the concept or entity that is being described). It is live
metaphors, or rather metaphors in the “live” part of their life cycle, that
are agents for semiotic change.
The aspect of juxtaposition of one concept with another gives the met-
aphor its mythical quality in the Barthesian sense. Selection of both the
aspect to be compared and the entity with which it is to be compared
foregrounds ways of looking at objects and ideas that cannot help but be
value-laden, whether the metaphor is conventional or innovative. Indeed,
conventional metaphors can be considered those in which these particu-
lar ways of looking at things have been naturalised.
The Level of Ideology
I have mentioned ideology in the previous two sections, in the context

that both codes and mythic meanings serve to naturalise ideologies. An
ideology is a set of beliefs and attitudes that structures all kinds of social
practices, and allows for group identity. Coupland and Jaworski (2001:
144) write:
We understand the term ideology as a set of social (general and abstract)

representations shared by members of a group and used by them to accom-
plish everyday social practices: acting and communicating.
This definition suggests that ideologies can only be maintained and dis-
tributed through “representations”, that is, through language and other
meaningful social practices. Further, ideologies are ubiquitous and neces-
sary—ideologies are what bind groups together, allow for group identities
and allow for the workings of society to be intelligible to its members. It
is understood that any number of competing ideologies may operate in
the daily social life of individuals. However, not all ideologies have access
to the same resources for awareness and understanding. Certain groups
will have a greater level of power and access to voice, for example, politi-
cians, media publications or celebrities. These more powerful groups are
better equipped to establish their views, attitudes and ideas than are less
powerful groups. Uncovering the working of power behind the represen-
tation of ideas is the project of work by Barthes and Foucault, as well as
later writers in the school of Critical Discourse Analysis. Essential for the
operation of power is compliance (Foucault, 1980) and what concerns
critical linguists is the extent to which the workings of power are hidden,
in ideas that are “taken-for-granted” or naturalised. As Eagleton (1983:
117) puts it: “Ideology seeks to convert culture into Nature, and the
‘natural’ sign is one of its weapons.”
I asserted earlier that the news media are one of the powerful groups
that has special access to voice, and named some of the other groups that
have regular and unquestioned access through media channels to the gen-
eral public. The news media do not only have the power to express their
own institutional opinions in the form of editorial and opinion pieces,
but they also have the power to select and interpret the views of others
and to police what counts as news. As Parmentier (2009: 146) notes,
“Anyone who manipulates or regiments the flow of interpretants thereby
indexes social power or cultural capital.” Through connotation and inter-
textuality the press can access shared social meanings that can indicate
an ideological position. For example, the now well-used device of attach-
ing “-gate” to words to indicate a scandal (“Monicagate”, “Cheriegate”,
“Plebgate”) by analogy with the Watergate affair can quickly imply a
shared moral position on some sets of events (Conboy, 2007: 98). (But
interestingly not on others. I have yet to see the formulation “BP-gate”
used in media texts, despite extended coverage of alleged safety weak-
nesses.) Investigating the linguistic strategies that encode ideologies is an
important element of a holistic semiotic analysis.
In Chap. 2, I gave a brief overview of Critical Discourse Analysis
(CDA) and intimated there and earlier in the book that I make a distinc-
tion between critical traditions which start with an emancipatory agenda
and others which do not, arguing that most discourse analysis is criti-
cal to an extent. I need to make this position more explicit with regard
to my own approach, based on the thinking of Barthes, because of the
important role CDA has played in the study of discourses, particularly in
media texts. There is no doubt that Barthes himself was troubled by the
hidden exercise of power through language—he makes this point explic-
itly (1972: 11) when he writes of “ideological abuse” in “newspapers, art
and common sense”. His project in the book, including his now famous
deconstruction of two images—a Paris Match cover and an advertisement
for Italian food—is to problematise naturalised concepts.
The difference between Barthes’ and Fairclough’s understanding of
texts in context, then, is one of emphasis and interpretation, rather than
fundamentally differing worldviews. Comparing the Barthes heuristic
with Fairclough’s (1992a: 73) three-dimensional framework for Critical
Discourse Analysis illustrates these different emphases. The visualisation
of Barthes’ levels here is the same as the one shown in Chap. 3, but
redrawn by analogy with Fairclough’s CDA approach (Fig. 3.2).
There are three issues of divergence I perceive between the Faircloughian
approach to critical text analysis and the Barthesian approach I propose.
Firstly, I suggest that Barthes’ approach to levels of language is more text-
focused at all levels than Fairclough’s. Fairclough’s approach includes
analysis of the production and interpretation of texts within both the
immediate context and the socio-cultural context. At these levels, the
emphasis is on a description and interpretation of the immediate and
wider context and their effects on the text. Barthes is more concerned
with how the sign reflects, embodies and realises the context. Secondly,
I argue that although Barthes’ perspective is entirely critical, it has a less
explicit emancipatory agenda than CDA. My final point is connected
Fig. 3.2 Barthes’ and Fairclough’s views of language in context
with this. In seeking out a politically motivated driving force for the
understanding of language, Fairclough proceeds from a particular view
and expectation, based on the thinking of Marx and Gramsci, rather
than the shifting, fragmented and competing view of discourses typical
of poststructuralist thinkers such as Barthes.
This final point is central to the approach I propose. This approach
starts with description of the text, with no preassumptions other than a
generally critical standpoint. The findings of the work, which I will go on
to outline in the following chapters, suggest discourses of representation
which co-exist, change unevenly and even move in different directions
(simultaneously “opening up” and “closing down” meaning potentials).
These patterns are better understood within a frame of reference which is
open-ended rather than agenda-driven.
Discourses
Foucault (1980) proposes that ideologies are maintained and distributed

through language via discourses. These are regularly articulated sets of
ideas in circulation at a particular time and place. In Mills’ (2003: 53)
definition “a discourse is a regulated set of statements which combine
with others in predictable ways”. In using the word “regulated”, Mills

refers to the dimension of power that is able to regulate what counts as
knowledge and what does not. Foucault argues that defining and exclud-
ing certain groups (such as the “mad” or the “sexually deviant”) limits the
possibilities for discourse. Other limitations he identifies include restric-
tions on who is allowed to speak, and exclusive distinctions between aca-
demic disciplines. Some discourses have much more support and wider
distribution than others, an understanding of which leads to the obser-
vation that some are “dominant”, although opportunities can occur for
“resistant” discourses to be heard, and even to become dominant them-
selves in time. The news mass media have a primary role to play in the
construction and distribution of discourses, both dominant and (less
often) resistant. Discourses serve to position subjects (Foucault, 1972); in
other words, discourses offer roles to individuals that they can accept or
reject. Rejection of a positioning is not without cost—as it can transgress
social expectation. Subject positions are multiple and changing, rather
than unified and fixed.
The semiotic heuristic proposes that signs, codes and mythic meanings
all contribute to the construction and perpetuation of ideologies. The fact
that discourses are myriad and fragmented suggests that researchers may
wish to restrict their analysis to certain types of discourse. In the case of
the BP texts, for example, there was ample data to investigate discourses
concerning oil—its perceived necessity or otherwise and its relationship
to the environment, business and politics. Such an investigation would
require a study of content, meaning a focus on what was expressed in the
text. My research interest was rather in how the crisis was expressed, in
other words, in the media representations of the crisis. Given this focus,
my aim was to identify and describe discourses of representation.
In Summary
Barthes offers an understanding of language as multilayered, whereby

meaning is encoded at different levels: in sign choices, in the combina-
tion of signs and in the understanding of signs as continually in flux,
and socio-culturally and historically determined. In this chapter I have
defined the terms “sign”, “code”, “mythic meaning” and “ideology”, in

each case giving examples of how these levels of meaning find expression
in language. This view allows us to investigate texts from a narrow to a
broad (a micro to a macro) perspective, that is, connecting small signs
to social contexts. What this means in practice is that we can identify
and discuss language features at different levels—individual words and
phrases at the level of the sign, systems of signs at the level of the code,
connotations and figurative language at the level of mythic meaning and
broad patterns of ideas at the ideological level. Taken together, these dis-
cursive elements will give us a rich picture of how meaning is constructed
in a text or set of texts.
4
Data Collection and Research Principles
The principles emerging from the Barthesian conceptualisation of language

described in Chap. 3 is that meaning is made at four interrelated levels (sign,
code, mythic meaning and ideology). The following chapters set out a tem-
plate for the analysis of sets of representations by the news media of the BP
events at these four levels. This approach entails firstly a proposal for data
collection and selection, and subsequently a four-stage methodology for data
analysis. This four-stage approach is highly flexible; it allows for complete
data sets to be described and set into context, and then for relevant language
features to be identified and analysed, using a toolkit of methods, according
to need and purpose. The features analysed and the choice of analysis tools
for BP are not necessarily the same as those that would be used for another
set of texts—they were selected for this specific project with its particular
research questions. The BP project was designed to investigate:
How has the BP Deepwater Horizon crisis been constructed in the media?
What patterns and characteristics of language can be identified within and
across media texts about Deepwater Horizon? What do these patterns
reveal about the process of crisis representation, and how we come to make
meaning of crises over time?

in Discourse, DOI 10.1057/978-1-137-58750-3_4
The following chapters describe how I addressed these questions, from

identifying researchable representations of the crisis to investigating pat-
terns and characteristics.
Choosing a Source for Texts

According to my view of a media representation, the texts in a data set
will have something in common with each other, although they might
be highly disparate in some respects. In the case of the BP media texts,
investigating media representations over time entailed collecting data
sets from different time points. I selected three: 2010 (just after the
explosion), 2011 and 2012. The texts had in common that they were
in English, that they were written and that they contained reference to
the BP events of 2010. On the other hand, they were of differing geo-
graphical origins, differing news media genres (e.g. news report, feature,
letter), differing lengths and differing channels (e.g. online and print).
Describing changes in representation of the crisis events over time would
mean identifying commonalities in the sets of texts which outweighed
the differences.
The BP Deepwater Horizon oil spill of 20 April 2010 was an event of
international significance, and was widely reported upon by the global
media from that date to today. While this intense media coverage makes
the oil spill an interesting and relevant topic for study, the main implica-
tion for researchers is that millions of texts on the topic exist. Thus the
main concern for data collection and selection was how to create a data
set that was manageable yet robust: it had some degree of internal coher-
ence which would bear the weight of claims made about it.
I chose the database Nexis UK as my source of media items for a num-
ber of reasons. Nexis UK is a database of media texts that covers hun-
dreds of media sources in many different countries. Publication types
include both press and web channels: broadsheet and tabloid national
and regional newspapers, international trade and industry magazines
and online newsletters, blogs and reports (Nexis UK, 2014). Items are
included from all sections of a given publication or website, so there is no
particular focus on, say, news or finance, and features, letters and reviews
are included. The database is reliably searchable, meaning that users are
able to specify dates of publication, language and topic keywords in order
to identify relevant texts. Individual texts are complete, in that they are
not edited or rearranged, although they do not include accompanying
images. The texts are also presented with additional information, such as
word length and a categorisation of publication type. Despite these ben-
efits, the database is not exhaustive. In particular, certain business blogs,
but not all, are included, and not all English language local newspapers
are included. This meant that I was restricting my data to those publica-
tions and web pages covered by Nexis UK. However, using the database
gave an extensive and varied set of data items for my purpose.
Compiling a Data Set

Initial work on identifying texts that mentioned the BP Deepwater
Horizon oil spill showed that there were many thousands in the Nexis
database alone. My aim was to extract a greatly reduced text data set in a
systematic way. I was potentially interested in both synchronic and dia-
chronic patterns of representation—in other words, how were the events
represented at a given time and did this change over time? One way of
deriving a data set that would enable me to address both of these ques-
tions, while at the same time reducing the text numbers to a manageable
level, was to look at all the texts on a number of single days. My first
selected date was 27 April 2010. The reasoning behind this choice of one
week after the Deepwater Horizon explosion was to allow reporting on
the events to have become widespread across a number of publication
types. Muralidharan, Dillistone, and Shin (2011) note that the first tweet
and Flickr photo from BP were both posted on 27 April. However, the
date chosen could equally well have been the day after the explosion, or
three or four days later, as long as the crisis period was still in train. What
I felt to be important, having chosen a preliminary date, was that sub-
sequent data sets should be drawn from the same date in the years up to
the time of analysis, that is, 27 April 2011 and 27 April 2012. This would
generate texts from three days, each separated by one year. Using all the
data from each date would allow for an exhaustive study of texts relating
to the BP events within a narrowly specified time frame, although this

does not entail that the dates are representative of general BP coverage.
Concurrently, in order to determine how many texts this time-based
search would generate, I needed to define what constituted “texts relating to
the BP events”. The main search term would be “BP” as there were no other
viable candidates (the former name “British Petroleum” was superseded by
“BP” in 1998, and is not used in news media texts). I used three additional
terms that my own reading of newspaper and online reports suggested were
reliably present each time the events were covered. These were “crisis”, “oil
spill” and “disaster”. A search for any one of these three plus “BP” should,
in my view, have found virtually all of the texts covering the BP events. I
therefore carried out an initial search based on the following terms:
• BP AND
• “crisis” OR
• “oil spill” OR
• “disaster”
From my previous reading of news reports on the BP oil spill, I felt that
these terms would return a near-complete set of texts; however, I was also
aware of other references to the events that paired names such as “Macondo”
(the company name for the well) and “Deepwater Horizon” (the name of
the Transocean rig) with “oil spill” and so on, referring to exactly the same
events. I wanted to test the possibility, particularly in shorter or later texts,
that BP might not be a default descriptor. Further, if I was going to examine,
for example, how the events were named, I needed to feel secure that I was
not simply replaying the search terms I myself had defined. To address these
issues, I performed an alternative search as follows:
• BP OR
• “Macondo” OR
• “Deepwater Horizon” AND
• “oil spill”
This alternative search produced almost identical lists to the original

search, ratifying my choice of search terms.
Table 4.1 Sample of BP-related texts from Nexis UK database

Texts including search terms “BP” AND “crisis” Sample for depth
OR “oil spill” OR “disaster” analysis
27 April 169 20
2010
27 April 94 20
2011
27 April 31 20
2012
I cleaned the data by removing highly similar texts through the search
filter and hand sorting the rest for exact duplication, retaining those where
a proportion of the text was similar or the same, but not all. In many
cases this partial duplication appeared to arise from the direct reproduc-
tion of the wording of press releases, which was an interesting point for
investigation in itself. The result of these selection decisions was three
separate data sets: 169 texts for 27 April 2010, 94 texts for 27 April 2011
and 31 texts for 27 April 2012, shown in the first column of Table 4.1.
I considered that the each full data set would contain a workable num-
ber of items for broad contextual analysis to be carried out, using quanti-
tative methods (although any number-based findings from 2012 would
need to be treated with caution as this is a small data set of 31 texts). The
majority of my analysis would be qualitative, for which I judged a smaller
sample was appropriate. I selected 20 texts from each data set for deeper
qualitative analysis, using systematic random sampling, that is, choosing
every nth text in the larger data set to give me 20 texts (this does not imply
that my sample is random in any other way). So, for example, from 94 texts
in 2011, I selected every fifth text plus the final text, which yielded 20 texts.
The final sample is shown in the second column of Table 4.1.
Text, Co-text and Context

The texts chosen varied considerably by origin and purpose, but also in
their relationship with the BP story. Some texts were entirely concerned
with the BP Deepwater Horizon story, others touched on the story as part
of a wider background (e.g. a business page piece about oil), still others
only mentioned it in passing. This observation raised the issue of texts and
co-text. Would a text always comprise the complete entity as presented by
Nexis, or would it be legitimate to reduce some items? One straightfor-
ward approach would have been to analyse only those sections of the data
items that were directly related to the BP events. However, it was also useful
to observe whether and how the importance of the BP story within texts
changed over time. Given the concern of semiotics with intertextuality and
the concept that all texts are acted upon by myriad other texts, I was inter-
ested to understand what the BP story was mentioned with. Further, it was
not always self-evident what texts were “about”. For example, in a letters
page, it was quite a straightforward task to separate out a letter about the BP
oil spill from a different letter on a different topic. On the other hand, it was
less straightforward to separate out coverage of the immediate effects of the
oil spill from a general discussion of the current energy picture in the USA.
Researchers will arrive at a definition of a text for analysis purposes,
according to their research question. In the case of the BP data, I considered
all texts in their entirety for a contextual analysis of genre, source, geographi-
cal location and salience of the BP story. For the detailed qualitative analysis,
some texts were reduced to focus on the segment that related to BP. To take
an example, one transcript of a news programme in my data ran to 7345
words, of which only 634 dealt with the BP story. I chose to use just this seg-
ment in my sample for depth analysis. Nevertheless, it was still important to
retain co-text that was, however, indirectly, part of the overall BP story. This
allowed me to form a view about a particular aspect of the construction of
the crisis through language: once the oil spill has ceased to be a main news
story, in what contexts is it still considered to be relevant?
Research Approaches: Micro, Median

and Macro
A significant innovative feature of the methodology I describe is that it
sits at a medial level of data description that is uncommon in language
research. By this I mean that my methodology is situated in between
the decontextualised, large-corpus approach characteristic of Corpus
Linguistics and the highly detailed analysis of one or a few texts which
characterises much discourse analysis. This middle-level approach is

apparent in both data selection and data analysis. I largely worked with
three data sets of 20 texts, this number falling between the norms for
Corpus Linguistic and close text analysis studies mentioned above. My
analysis approach combined semi-quantitative counting methods with
depth analysis of particular texts and text extracts in order to investigate
and contextualise the findings suggested by frequency counting. This
medial level of investigation has several advantages for research questions
that are broad in scope. It allows the researcher to construct reasonably
robust “maps” of language patterns through the analysis of frequencies of
language-feature occurrence, without the loss of rich contextual detail.
The approach is effective both to confirm that the patterns of language
features proposed are indeed consistently present (the usual province of
Corpus Linguistics) and to describe what these features “look like” in the
context of the news media genres in which they appear.
Research Approaches: Qualitative

and Quantitative Methods
It has been typical (Bryman, 2004) that quantitative research is used for
research within a positivist or post-positivist paradigm, and that qualita-
tive research is more frequently used for research within an interpretivist
or constructionist paradigm. However, linguists such as Wodak (Weiss
& Wodak, 2003) and Cook, Robbins, and Pieri (2006) use quantita-
tive methods such as surveys, feature counts and corpus methods as
part of their range of methodological tools for investigating linguistic
phenomena within a constructionist worldview. The purpose of quantita-
tive and semi-quantitative work within my research was not to identify
and closely describe language features, but rather to provide one way of
looking at patterns across data sets. There are two types of pattern that a
semi-quantitative analysis is appropriate to uncover:
• Patterns in text types. How many texts are there on the selected dates
that refer to the BP events? To what news genres do they belong? What
is their country of origin? How important (how big) is the BP story
within them?
• Patterns of linguistic features. Does the researched feature occur in the

data set, and how often? If there are different sub-categories of the
feature, which are present and in what proportions? Do these occur-
rences change over the timespan of the data sets?
Approaching these questions via semi-quantitative methods gave a

degree of confidence that the patterns I detect in my data sets were reli-
ably present. Quantitative research did not tell me why these patterns are
occurring, or whether the changes were peculiar to certain contexts, or
indeed whether they went hand in hand with other phenomena. It was
not enough to find that a feature type increased or decreased over time:
that movement needed to be described and explored using qualitative
approaches to explain the function of a language feature in context, and
in interaction with other features. However, quantitative findings of this
kind were one way of demonstrating confidence that such changes are
occurring, and provided a firm foundation for further questioning.
5
A Barthesian Analysis of the BP Data
in Four Stages
Before I outline the four stages of my research approach, I will dis-

cuss in more detail a concept I have mentioned a number of times
already—that entire media representations can be conceived of as signs
in their own right. This contention is important to the rest of my argu-
ment. I hold that sets of texts, such as my three data sets of BP texts,
have enough characteristics in common that they are describable as an
entity—a “language map”.
Semiotic Discourse Analysis

and the “Language Map”
A “language map” is a holistic view of a representation which connects
small signs to large signs. When we describe speech or writing as a col-
lection of signs, or semiotic modes, or resources, we are perceiving signs
as small elements, perhaps as “building blocks” which become invested
with meaning only through the ways in which we put them together,
our agreement about what this “code” or system means, our recollection

in Discourse, DOI 10.1057/978-1-137-58750-3_5
of the ways in which words have been used before and our knowledge
that alternative words might have been used in their place but were not.
Individual words are signs in that they have a signifier (the written word
or spoken sound pattern) and a signified (their “meaning” or mental con-
cept in context). This is the definition of de Saussure’s “isolated sign”
(1959: 128) and it is one I will use as I develop my argument. However,
there is also a broader interpretation of language-as-sign that I would like
to introduce here. De Saussure suggests that larger stretches of language
can also be signs:
As a rule we do not communicate through isolated signs but rather through

groups of signs, through organized masses that are themselves signs. In lan-
guage everything boils down to differences but also to groupings. The
mechanism of language, which consists of the interplay of successive terms,
resembles the operation of a machine in which the parts have a recipro-
cating function even though they are arranged in a single dimension.
(de Saussure, 1959: 128, my emphasis)
The implication of de Saussure’s assertion above is that language can func-

tion as a sign at different levels—that individual words, groups of words
and individual texts are all signs, and, by extension, that sets of texts
may be considered as signs. De Saussure uses the metaphor of a machine
with its various interconnected parts, each essential to the efficacy of the
whole. I imagine a landscape or a map, where individual features appear
in varying quantity, configuration and distribution to give a distinctive
yet recognisable landscape. The concept of a map sits well with the notion
of “representation”. The question I address in this book is not “what does
the BP crisis ‘look like’” but “what does the representation of the BP crisis
‘look like’”.
The idea of representation as map is one explored by Baudrillard
(1994) in Simulacra and Simulation. Baudrillard posits that society has
moved through successive periods in which representations have become
increasingly disconnected from the reality they depict. Writing of the
“image” (which covers a broad range of representation types) he identifies
four successive phases (1994: 6):
1. It is the reflection of a profound reality.

2. It masks and denatures a profound reality (art imitates life).
3. It masks the absence of a profound reality.
4. It has no relation to any reality whatsoever: it is its own pure
simulacrum.
Baudrillard here presents progressive stages from signs that offer some
reflection of reality to signs that refer only to other signs and have no
relation to reality at all. In discussing relationships between simulacra
(loosely, “signs”) and “reality”, he draws on concepts of both physical
resemblance and authenticity. He alludes in his work to the premodern
and modern periods which he identifies as having at least some relation-
ship, however distorted, with the real and original, but his main argu-
ments deal with the nature of late twentieth-century society, where he
suggests that not only are the real and original unrecognisable, but that
they have evaporated. To illustrate this argument, Baudrillard draws on
a fable told by Borges (1999: 325) in which, in an ancient Empire, “the
Cartographers Guilds struck a Map of the Empire whose size was that of
the Empire, and which coincided point for point with it”. Eventually the
map rotted away, leaving only a few remaining shards in remote places.
Baudrillard claims that society has reached a stage where it is the territory
itself that has crumbled away, leaving only the map as a representation
of a reality that no longer exists. Baudrillard denies in his work the exis-
tence of any material reality, and this is an extreme view. However, this
notion of signs being the outward representation of a shifting, unstable
and ungraspable reality is one that is fundamental to semiotic thinking.
The discourse analysis approach I offer is one way of describing these
outward representations.
In the case of the BP data, I have sought to map three sets of texts, each
at a time point a year apart. I propose that, although each set will comprise
texts which are very disparate, it will, nevertheless, have important things
in common which make it different from the others in the way it makes
meaning. I discussed in Chap. 1 that news media texts are highly dispa-
rate, yet also in some ways relatively circumscribed in terms of their rep-
resentation: by external technical, political and financial considerations,
by space and time constraints, by a strong set of generic expectations and
by the demands of collaborative processes. It is this sense of what news

representations of a story have in common rather than their differences
which leads me to propose the idea of language landscapes or maps. By
this I mean that a synchronic representation of a news story can be inves-
tigated by considering a range of texts about the story at a certain point
in time. Although the publications, the writers and the genre of the text
may be different, nevertheless, a description of the language will show
that the combined representation has certain characteristics which make
the representation describable as an entity. This is a language map or a
larger sign. It is by drawing up a set of language maps of the coverage of
the BP events that I hoped to understand how the media were construct-
ing our shared meaning of these events over time.
Four Stages of Analysis

I return to my research question: “What does the language of representa-
tion of the BP crisis ‘look like’?” Such a broad aim called for an open-
ended and interactive approach, which would allow the data themselves
to suggest areas of particular linguistic interest. This emergent approach
to analysis can be summarised as having four stages:
1. A contextualisation stage. A broad description of the data sets. What

sort of texts were they, and where did they come from?
2. An immersion stage—Preliminary analysis by reading and rereading
the texts in order to identify language features of interest for analysis,
at Barthes’ four semiotic levels.
3. A depth analysis stage—Investigating these language features and
analysing them for frequency, function and change.
4. A holistic analysis stage. What does an in-depth analysis of a single
text tell us about how these features interact? Can this exercise describe
the “language map” of the representation?
Table 5.1 summarises the stages of the analysis process for BP:
Table 5.1 Four stages of data analysis

Research
BP data method Analysis aim
Stage 1 Full data sets: Semi- Overview of country
Contextualisation 2010 (169 texts) quantitative of origin, genres,
stage 2011 (94 texts) salience of BP story
2012 (31 texts)
Stage 2 Data subsets: Qualitative “Immersion” in data
Preliminary analysis 2010 (20 texts) to identify
stage (immersion) 2011 (20 texts) significant features
2012 (20 texts) of language use
Stage 3 Data subsets: Semi- Analysis of frequency
Depth analysis stage 2010 (20 texts) quantitative and type of
2011 (20 texts) Qualitative significant features.
2012 (20 texts) Analysis of language
usages in context
Stage 4 Data subsets: Qualitative Single text analysis
Holistic text analysis Single texts as in-depth—language
required usages in combination
Chapters 6, 7, 8, 9, 10, 11, and 12 describe explicitly how each of

these four stages of analysis can be applied to data, using the example of
the BP texts. At each stage, findings from the data analysis are outlined
in brief. This not only gives an idea of the kinds of information generated
by the analysis approach, but also sets out a source of data which can be
further interpreted using semiotic concepts.
6
Stage 1: Contextualisation of the BP
Texts
The first stage of analysis is one of broad description of the data sets. At
this stage, I dealt with the full sets of texts (169, 94 and 31) before reduc-
ing the text numbers for closer examination. The purpose of this stage of
analysis was to provide a contextualising description of the sets of texts
that mentioned the BP oil spill across a number of broad dimensions.
Nexis UK provides certain information on the articles in its database,
including word length, name of publication and type of publication (e.g.
newspaper, magazine or weblog). However, I was particularly interested
in three other pertinent categories. The first was the country of origin of
the item. From the Nexis information, it was generally possible to deter-
mine the country of origin of the text, although this is not always clear
for online publications. The second was genre: my depth analysis needed
to take account not only of the range of language features characteristic of
the many text genres in the data set, but also of the shifts by genre across
the time period of the data. I wished to know whether the news genres
in which the BP story appeared were different immediately after the cri-
sis from two years later. While the Nexis database provided information
on type of publication, this did not necessarily map on to analysable

in Discourse, DOI 10.1057/978-1-137-58750-3_6
definitions of media genre. The third was some means of addressing the
salience of the BP story in the texts in which it appeared. Did the story
constitute the news item, or was it only a small part of a longer text, per-
haps mentioned as contextualising information in another story? For this
analysis, I determined by word count the proportion of the item which
directly concerned the BP story, and classified the texts in bands—less
than 25% of the item (by word count) concerned BP; between 26% and
50% of the item concerned BP; between 51% and 75% and between
76% and 100%.
The contextualisation analysis therefore consisted of:
1. Country of origin.
2. Genre of texts.
3. Salience of the BP story.
These three analyses were intended both to offer an initial picture of the
kinds of text covering the BP oil spill and to support and enhance the
later depth analysis of language features.
Country of Origin
Table 6.1 shows a breakdown of the country of origin of the texts that
mention the BP oil spill. Only texts in English were included in the
sample.
Table 6.1 Geographical source of items mentioning BP events 2010–12

27 April 2010 27 April 2011 27 April 2012
No. of No. of No. of
texts % texts % texts %
USA 76 45 56 60 19 61
UK 27 16 11 12 4 13
Rest of world 56 33 11 12 3 10
Unable to determine 10 6 16 17 5 16
(Internet text)
Total 169 100 94 101 31 100a
a
Note that low numbers make percentages indicative rather than firm findings
The site of the BP oil spill—the USA—begins and remains the

main country of origin for English-speaking media coverage, and this
reflects not just the direct interest in the events, but also the size of
the USA and the number of local and regional newspapers operating
there. However, the fact that the events are perceived to be of global
rather than just national interest is reflected in the fact that half the
English-speaking coverage in 2010 is outside the USA, primarily
from the UK. The UK has a specific interest in events, given that BP,
while a global company, is British in origin. At the time of the crisis,
President Barack Obama positioned BP firmly as a British (rather than
global) company (Burt, 2012) and by extension directed the blame
and responsibility in the direction of the UK. By 2012, the USA rep-
resented an even greater proportion of coverage at 61% than it did in
the first days of the disaster, with both the UK (at 13% on a low base)
and the rest of the world showing decreased coverage as a proportion in
2011 and 2012. This picture suggests that concerns with the physical
effects of the spill in the USA, both real and potential, are of greater
ongoing interest than concerns with moral and financial culpability
in the UK. That said, the other area of increase over the years by per-
centage is the geographically indeterminate category, which consists
mainly of Internet and blog coverage. This rose from 6% in 2010 to
16% in 2012, which implies a continuing interest in the story at a
more general, societal level.
The other issue of relevance to the geographical origin of texts is
that linguistic analysis is situated temporally and culturally—that is,
when country of origin changes, then style and social practice change.
This principle is central to a social semiotic view of communication
and is a critical component of the study of language choices—any
story of linguistic representation is also the story of the situated style
and social practice from which it arises. It can be important in analy-
sis to indicate that the writer is not just a journalist, but a British
or US journalist; that the newspaper is national or local; and that it
originates in one of the affected Gulf States, rather than an unaffected
US state.
Genre of Texts
Definition and Analysis Method
In order to carry out a quantitative analysis of the genres of the texts, I

required a definition and a categorisation of media genres. Establishing
a categorisation involved a strategy for identifying media genre types.
In presenting genre as one of Barthes’ semiotic “codes” in Chap. 3, I
characterised text genres as being identified through any or all of the fol-
lowing: common communicative purpose, structural regularities, stylistic
regularities and similarities in content. I have yet to find a systematic
taxonomy of media genres, and it is possible that none such is available.
According to Chandler (1997: 1), “There are no undisputed ‘maps’ of the
system of genres within any medium”, no one text displays all the features
of a genre and there is considerable overlap between genres. I required a
genre categorisation that was suitable for my purpose, that was pragmatic
and workmanlike. The two principles I used for genre categorisation were
that, where possible, the genres should be recognised in scholarly litera-
ture, and, secondly, that they should be on the same level as each other
in a notional taxonomy. By this I mean that they are (relatively) mutually
exclusive, and one is never subsumed by the other.
Following these principles of previous recognition as a genre, and sin-
gle hierarchical level, and using Bhatia’s (1993) terminology, I used the
categorisation shown in Fig. 6.1.
The central column, “genres”, provided the basis for the genre analysis
shown in Table 6.2.
Findings from the BP Data
Table 6.2 shows how the genres to which the selected texts belong change
over the three years of the data. In 2010, 73% of the texts were either
print or TV news reports, with the remaining 27% being largely financial
reports (27 April was Quarter 1 results day). By 2012, news reports had
reduced to less than half of the texts (45%). There is an increase over the
years (by proportion) in business and market reports, feature articles and
Fig. 6.1 A genre categorisation of news texts
Table 6.2. BP oil spill texts by genre 2010–12

27 April 2010 27 April 2011 27 April 2012
News report (written) 117 69 24 26 13 42
News report (TV and radio) 7 4 1 1 1 3
Financial report 34 20 38 40 2 6
Spoken interview (TV or radio) 3 2 3 3 0 0
Feature article 0 0 0 0 2 6
Editorial or opinion piece 6 4 13 14 3 10
(print/online)
Arts review 0 0 1 1 5 16
Letters 0 0 3 3 1 3
Business or market report 2 1 11 12 4 13
Total 169 100 94 100 31 100a
a
arts reviews of various kinds. The following is a brief description of the

texts in my BP data sets, according to genre.
News Reports
In 2010, the news reports on the BP oil spill (73% of texts) are mainly
progress reports that discuss announcements by BP concerning the victims
and the work being carried out to stop the oil escape, which was still uncon-
trolled and unpredictable at this stage. As well as quotations from press
releases and news conferences, other inputs from eyewitnesses and experts
in the fields of oil drilling, the environment and food industries are evident.
By 2011, only 27% of texts are news reports. Directly related news stories
deal mainly with the progress of compensation and legal action. However,
very many of the news reports cover stories that are not directly related to
the BP oil spill, but concern other events or companies, where the BP oil
spill is mentioned in a minor capacity. For example, one 2011 text (The
Irish Times, 27.4.2011) deals largely with the new chairman of commodi-
ties trader Glencore and his contentious remarks about women employees,
also mentioning that Glencore had appointed Tony Hayward as a senior
independent director, but noting that “his reputation has been shredded by
the Deepwater Horizon disaster”. A 2012 text (Lewiston Morning Tribune
[Idaho], 27.4.2012) reports that an oil spill in the Yellowstone River has
created a need for fish testing. Laboratories, however, are still “backed up
processing specimens collected in the BP oil spill in the Gulf of Mexico”.
As well as written news reports, the 2010 data set includes a number of
television or radio news reports. Even more than reports in newspapers,
television news focuses on breaking news. It is therefore unsurprising that
while seven of the texts in 2010 were transcripts of television interviews,
this dropped to one in each of 2011 and 2012, as the story moves from the
immediate explosion to longer-term issues such as compensation.
Financial Reports
In 2010, financial reports account for 20% of texts, and in these the
news of BP’s first quarter results competes with the ongoing news of the
oil spill; 27 April 2011 is again results day, and the first quarter results
become the main story in texts mentioning the BP events, accounting for
40% of items. The spill is mentioned (this is more or less the first anniver-
sary) but in the context of Quarter 1 results. By 2012, only 13% of texts
about the BP events fall into the financial reports pages.
Spoken Interviews
Of the six spoken interviews in the full data sample, four are Fair
Disclosure interviews in which annual company earnings are discussed
in conference calls and are presented in transcript by Nexis. These did
not all concern BP solely, but BP was mentioned in all. The other two,
both in 2011, are an interview with Michael Greenberger, of the Center
for Health and Homeland Security, and a transcript of a speech made by
Assistant Attorney General Tony West at the University of Chicago Law
School. Both make reference to BP only in passing, in the first case as an
instance of a crisis to be dealt with at a national level, and in the second
as an example of a large litigation case.
Feature Articles
This media genre appears only twice in the entire data set, and this is
in the form of travel features, both, coincidentally, dealing with New
Orleans and both written in 2012. Both use the theme of the city’s renais-
sance in the wake of a series of disasters (one shared with the rest of the
USA, that is, the financial crisis, and two that were much more localised,
that is, Hurricane Katrina and the BP oil spill). One of the items goes
even further back in time and scope:
The nearly 300-year-old city has had to rebound from centuries of disasters
including fires, plagues, hurricanes and most recently, the BP oil spill in the
Gulf of Mexico. (The New Zealand Herald, 27.4.2012)
One writer uses a personal narrative approach, comparing New Orleans

with her home town of Sarasota, and the other uses an impersonal
observer approach. In both cases, the tone is largely descriptive, with the
point of closure at the end, as is typical of the genre.
Editorial or Opinion Pieces
The texts in this category in the sample are of two main types—editorial/
journalistic comment and blog comment. Over the three-year span, the
proportion of editorial or opinion pieces increases strongly to 2011 and
remains at a similar level in 2012, indicating a shift in focus from report-
ing to commenting on and evaluating the events. Comment tends to be
presented as more impersonal in quality papers, as more personal and
emotional in the popular press and often as highly charged in Internet
blogs, as two example titles from 2012 exemplify: “Should we kill the
politicians before they kill us?” (Phil’s Stock World, 27.4.2012) and
“‘Crucify them’: the Obama way” (Right Wing News, 27.4.2012). How
commentary is variously constructed as more or less “personal” is a topic
for attention in the detailed analysis chapters.
Arts Reviews
The purpose of the arts review genre is to inform the reader about and
comment upon particular items of art, fiction and non-fiction literature
and so on. This genre does not appear amongst the 2010 texts, and is
represented by only one text in 2011, but by 2012 5 of the 31 texts in
the data set are of this kind. This is commensurate with the length of
time many artworks (in the broadest sense) take to complete, but might
also suggest that there is a length of elapsed time deemed appropriate or
decent for works of this nature to appear, and also indicates that a process
of assimilation may have taken place, that it takes an amount of time for
the wider significance of events to be understood and this is necessary for
a work of art to have resonance. These texts imply a move of the repre-
sentation of BP events from being part of the world of “reality” to having
an alternative existence in the world of artistic representation. Examples
of works referred to in the data are films, books (both fiction and non-
fiction) and documentaries.
Letters
Three letters appear in the sample in 2011, with a further one in 2012.
These are sent from both those in the public eye (e.g. by the Florida
Attorney General), with the purpose of publicising engagement with the
crisis, and members of the general public, challenging politicians about
their response to the crisis, and in 2012 addressing the continued effect
on sealife of the spill.
Business or Market Reports
Business or market reports have in common that they represent events

from the sole perspective of their effect on the business world. They can
be argued as a genre, in Bhatia’s terms, as they have purpose and audience
in common, yet stylistically they can be rather different. They can range
from somewhat resembling news reports to offering summaries, evalua-
tion and commentary on diverse business-related events. Business reports
account for only 1% of the texts in 2010 but increase to 12% and 13%
in 2011 and 2012, as writers are able to gain better insight with time into
the implications of the BP events for the oil and gas industry as well as
for wider business practice. Examples in this data set are a web piece on
petrol price rises (Theflyonthewall.com, 27.4.2011), a report on offshore
drilling (Greenwire, 27.4.2011) and a report in Campaign on “Building
brands through behaviour” (Campaign Middle East, 27.4.2012).
In Summary: The Importance of Genre
I suggested earlier that an understanding of media genres is central to any

work which seeks to identify meaningful patterns across a widely varied
data set. It was important to understand which of the language features
I examined were generally characteristic of the media genre under study,
and which were characteristic of writing about crisis events, or indeed
the BP crisis in particular. However, I return to my earlier point that

the choice of genre has semiotic meaning in itself. If the BP events are
described in a way that is typical of a particular genre, then this is a cru-
cial part of how meaning is made by journalists and received by readers.
In this way, the shift of the genres and sub-genres within which the story
is located is as important as the individual expressions or structures used
to describe the BP events.
Salience of the BP Story

One of the measures of how a story is being represented concerns not just
how many media items are written about it, but how much of a given
item is directly connected to the story. A reduction in the number of
stories is indicative of a loss of media interest, and there being less going
on in terms of the story. However, a reduction in how big the BP story
is within a given piece is also of interest, and this is shown in Table 6.3.
In 2010, more than two-thirds of the media items in the sample dealt
with the BP spill alone. The majority of the remaining texts are those that
deal with financial results, in which BP is only one of the organisations
reported on, and TV news programmes, where BP is one story of several.
By 2011 just one-third of texts are primarily about the disaster. By 2012,
this figure is only about one-fifth, and BP is mentioned only minimally
in nearly three-quarters of the texts.
Table 6.3. Proportion of media text dealing directly with BP oil spill 2010–12
27 April 27 April 27 April
2010 2011 2012
% of text relating to 0–25% 37 22 52 55 22 71
BP story 26–50% 12 7 9 10 3 10
51–75% 4 2 1 1 0 0
76–100% 116 69 32 34 6 19
Total 169 100 94 100 31 100a
a
As time goes on, stories that mention the BP events tend to mention
it not as a story in itself but as an example that illuminates another phe-
nomenon. Consider the following text from 2012. In a 3029 word blog,
the following fragment appears:
Some of the most profitable of all corrupt activities involve energy.

Remember Dick Cheney’s secret energy meetings? Those led directly to
electricity deregulation scams, corporate welfare for energy producers,
fracking, the BP oil spill, gas pipeline explosions, high gas prices, faulty
nuclear reactors, and an unreliable grid. (Phil’s Stock World, 27.4.2012,
my emphasis)
Here the BP oil spill is used as an illustration of the results of “corrupt”

activities involving energy. This is the only place the BP oil spill is men-
tioned in the entire piece, and in this way it represents a typical example
of the pattern of coverage of the oil spill as time moves on. This pattern
of a reducing proportion of coverage within stories may indicate some-
thing more than media loss of interest in the BP oil spill. This may be
an assimilation of the concept of the BP oil spill into the way we look at
the world, where the BP events have become a shorthand or an index of
something else, which is jointly understood, and which casts light on yet
other social phenomena in a process of unlimited semiosis. This idea is
further explored in analysis.
The Contextualisation Stage in Summary

The overview analysis in this chapter indicates a picture of a news story
that becomes increasingly a matter of US interest. The genre breakdown
shows the story moving from newswires and TV coverage to newspa-
pers, and from the front pages of newspapers to the features, comment
and review pages. Meanwhile, other forms of commentary on the events
become more important—particularly that in Internet blogs and the
business press. From being the major focus of news media items, the BP
story becomes largely a “footnote”, where it is drawn on as a supporting
illustration for the main stories represented.
The contextualisation stage served a number of useful purposes:
1. It provided an overview of the BP story as represented in the selected

texts. I was able to establish the kinds of stories which were being writ-
ten about the events, how they changed and the relative importance of
the BP events in the articles in which they were mentioned.
2. It allowed me to trace the movement of genre over the timespan,
which I have already suggested is critical for the understanding of the
language in the texts. Patterns of change in news genres were reliably
established through this quantitative exercise, before being further
analysed qualitatively.
3. It suggested a number of areas of interest for analysis which could be
further investigated in Stage 2. For example, what role did the story of
BP play in other news stories once it was no longer the main topic of
coverage?
7
Stage 2: Preliminary Analysis of the BP
Texts
The first stage of analysis (contextualisation) served to set the overall data
set into context, giving a description of the kinds of texts in the set and
the role of BP within them. The second stage (preliminary analysis, or
immersion) entailed a close reading of 20 texts in each year, in order
to identify features of interest for analysis. I have described my analy-
sis approach as emergent, and this second stage involved a reading and
rereading of the texts to generate a longlist and then a shortlist of lan-
guage features for further investigation. The first task was to read the BP
texts repeatedly, questioning the data for significant features and patterns.
This process was reflexive—I was cautious not to overburden infrequent
features with a weight of significance they did not justify, while remain-
ing open-minded to features I had not expected to find. A close reading
of the texts should suggest specific questions that lead to analyses for
patterns. I set out below three texts, to show how this process can be
used to identify language features that are potentially of interest. To avoid
“cherry-picking” significant texts, I show the first text in each data set of
20, as returned by Nexis UK. These texts did not necessarily illustrate all

in Discourse, DOI 10.1057/978-1-137-58750-3_7
102 Semiotics and Verbal Texts: How the News Media Construct a Crisis
nine of the features for analysis; rather, I model below the process used in
all 60 texts, which resulted in the selection of the final features.
An Example Text from 2010

This text from 2010 (24/7 Wall St, 27.4.2010) is extremely short, at
only ten words, including the attribution. It is from an online weblog
under the title “Media Digest 4/27/2010 Reuters, WSJ, NYTimes, FT,
Bloomberg”. The full original text contained 28 entries, of which only
one, shown below, relates to the BP oil spill. This single entry constitutes
my data item for analysis. As the title of the blog suggests, this is a quick
way to alert interested audiences to what the authors deem to be the
main stories from respected news sources. Readers can follow up items
they wish to learn about in more detail. In the item below, WSJ stands
for Wall Street Journal.
WSJ1: The2 BP oil spill3 could4 reach land within days5.
1. General question: writer/audience, who and why? What does

this short online piece have in common with longer print news
reports, and where does it differ?
2. “The” signals presupposition. What forms does presupposition
take?
3. “BP oil spill” indicates a particular naming choice. What naming
choices are made and what do they suggest?
4. “Could” is an epistemic modal form. Whose opinion, whose
uncertainty? How is modality used?
5. “Within days” is one use of ‘facts and figures’. Where are these
specific and where (as here) indefinite and why?
This briefest of texts at the beginning of the first data set suggests a
number of questions of potential interest to the research, and provides
a point of reference for questions of the subsequent texts. I will use the
7 Stage 2: Preliminary Analysis of the BP Texts 103
five questions generated above to illustrate how this initial reading and
data interrogation can ultimately be translated into a framework for
analysis.
1. General questions. Questions relating to the purpose and structure of

the text and comparison with other similar and dissimilar texts relate
to the study of genre. Consideration of the genre of texts is critical, as
genre entails and explains many features of style and structure I am
likely to encounter: genre is therefore adopted as a feature for study.
2. Presupposition in media texts is a potential topic for study. However,
it entails the study of a large number of sub-categories (Zare,
Abbaspour, and Nia [2012] identify eight categories) and I judge that
space does not allow for this level of detail as well as commentary on
a wider range of features.
3. Naming choices are potentially interesting in the construction of rep-
resentations: these are adopted as a feature for study.
4. Modality is likely to be a fruitful area for enquiry, both from the per-
spective of (un)certainty and (in)ability and as a way of looking at the
extent of writer involvement in or distancing from the text: this is
adopted as a feature for study.
5. The presentation of empirical facts is a feature of this and other texts.
It can be conceptualised as a signal of an “objective” style of writing
that in turn is characteristic of the news reporting genre: for this rea-
son facts and figures may be subsumed into the analysis of genres.
I have taken advantage of the brevity of the 2010 text to illustrate in some
detail the steps between initial reading and drawing up a final analysis
plan. The following texts, for 2011 and 2012, are analysed in less detail,
and used mainly to show how the patterns that emerged from the lan-
guage can suggest a final line of enquiry.

The 2011 text below (Journal of Counterterrorism and Homeland Security
International, 27.4.2011) is a transcript of an interview between a repre-
sentative of the International Association of Counterterrorism and
Security Professionals (IACSP) and Michael Greenberger, Director of the
Center for Health and Homeland Security. Again, the original full text is
longer, but has been reduced for analysis in line with the principles out-
lined earlir.
This text differs from that of 2010 in that it is no longer describing
events but discussing and evaluating them. The BP oil spill is not the
IACSP: Looking back at Katrina, the BP oil spill and other events1,
it does not appear that2 we3 are prepared to respond effectively or to
recover from natural disasters4, let alone a major terrorist attack. In
your view, is America prepared to withstand a chemical, nuclear or
another 9/11 style terrorist attack?
Greenberger: Well, in all candor5, I don’t think we’re prepared to
withstand a nuclear attack6. It would be such a devastating thing
that our resources would be overrun. I’m very worried about our
ability to7 withstand a chemical attack. Next to a nuclear attack,
chlorine or some other dangerous stored chemical that’s released
would be very devastating. As you get into emergencies that are not
as huge as a large weapon of mass destruction attack, I think the
country is much better prepared. If you are talking about 9/11 or
a repeat of the anthrax scare, I really think that the infrastructure
has gotten better. The public health infrastructure is better. People
know what to do. I would say also for a dirty bomb we’re better
prepared. But anything that involves large parts of the population,
I worry about our ability to respond. I don’t rule out a weapon of
mass destruction, but most of the things we are facing are conven-
tional weapons.8
1. This is an interesting list of events. What do Katrina and BP have

in common, and what might “other”, presumably similar events
be? What other lists and groups does BP appear in, and does this
change?
2. Epistemic modality—why is this used here and what does it tell
us about the speaker?
3. “We”? US or the global population?
4. Is the BP oil spill a “natural disaster”? In what way? How do oth-
ers perceive it?
6. This is a conversational discourse marker. Is this simply a feature
of spoken text, or is the speaker doing Nuclear attack is set up in
contrast to the BP oil spill (signalled linguistically by “let alone”
in the first paragraph). However, does the general discussion of
disasters invite comparison rather than contrast?
7. Dynamic modality—the role of modality in general, and differ-
ent types of modality in particular.
8. Greenberger’s turn makes use of considerable repetition. Again, is
this a feature of conversation, or are the repeated parts linked
with Greenberger’s area of professional interest? What does this
passage tell us about the US view of future crises like BP? What
is the dominant discourse here?
only topic, but one of many topics of conversation. This text generates
a set of questions that are rather more complex than the largely gram-
matical questions generated by the 2010 text, and may be addressed
at the semiotic level of myth and ideology as well as sign and code.
Alongside this broadening of interest comes an interrogation of con-
tent alongside language usage. My questions of this piece have taken
on a critical rather than descriptive flavour, and I note that this is
something I need to be reflexive about, particularly where I discuss
discourses and ideology.

The first text of the 2012 data set is from the trade journal Marketing
Management (27.4.2012). Only part of the article is included for analy-
sis, as only part relates to BP. It is part of a multiauthored article on the
subject of crisis management. The article is intended to be of general
interest to its audience, but its topic is a particular service—a quantita-
tive research product aimed at identifying, and hence more effectively
dealing with, the onset of a business crisis. In this sense, the article is also
serving as sales promotion.
When Does a Crisis1 Begin?

In the regular course of business, the increase of familiarity is a
good thing. When marketing campaigns are working well, famil-
iarity with the company will grow. We see a crisis begin when the
firm's familiarity rises, but this kind of increase in familiarity is the
result of the company being in the news. The starting point of a cri-
sis is usually fairly easy to identify, because familiarity will suddenly
spike and favorability will drastically2 decline for reasons outside the
normal marketing efforts.3
When the Gulf of Mexico platform of BP, the energy company,
blew up2 and the safety valves didn’t work,4 the situation was pretty
much5 guaranteed to become a crisis6. Familiarity went up because
the burning platform was broadcast immediately around the world;
favorability for BP went down when word began to spread that
there was likely to be a catastrophic2 oil leak (see Figure 17).
However, many events that might be thought of as a crisis, such
as the multiple Toyota recalls6, do not actually meet the criteria
of one.
1. The authors choose to use the word “crisis” and identify the BP
events in terms of business crises. How do writers in other genres
name the crisis?
2. Highly affective language. How does this relate to genre, includ-
ing the purposes of the authors?
3. This first paragraph is scene-setting rather than directly related to
BP—there is a decision to be made here on text and co-text.
4. This appears to be a deliberate, almost dismissive over-
simplification of the complex situation. What work is this doing,
in relation to the article’s purpose? Does it tell us anything about
views of the events, two years afterwards?
5. “Pretty much” is a modal item, and is also markedly conversa-
tional in tone. Is this a generic feature, or does it relate to my
comment in note 4?
6. Listing and grouping: is “business crisis” another group to which
the events belong (like “natural disasters” above)? BP meets the
criteria of a ‘crisis’ by the authors’ very specific definition, but
Toyota (last paragraph) does not. As in 2011, both comparison
and contrast are in play.
7. Reference to supporting text is a form of intertextuality. This
instance is routine, but intertexts such as press releases are crucial
to representation.
As in the 2011 text, the BP events are being used here to illustrate the
phenomenon “business crises”, which in this article have a very specific
definition. Once again, the use of a categorisation for different purposes
is suggested by the data as an area for further exploration. The descriptor
“crisis” is surely significant here and gives further support for an anal-
ysis of the feature “naming choices”. The use of modal items such as
“pretty much” is again of interest, in its interrelated functions of signal-
ling degrees of “truth” and confidence, and doing interpersonal work.
Intensified lexical choices such as “drastically” and “catastrophic” indicate
that looking at affect and appraisal in language as well as modality might
be productive.
Features Chosen for Analysis

The three sections above represent this preliminary engagement with the
data as a set of questions suggested by the language of each of the texts.
The process modelled here was repeated in the case of all 20 texts in each
data set. The process now became interactive and recursive. The long-
list of questions generated was followed by preliminary analysis, some
of which proved fruitful, and some less so. Some of the features noted
turned out to be relatively rare. Others did not appear to generate a pro-
ductive analysis. For example, a review of sentences that pivoted on the
conjunction “as” (for example, in the headline “BP profits soar as oil giant
accelerates spill clean-up” (The Evening Standard [London], 27.4.2010))
showed that the construction was quite common, but otherwise unin-
teresting. Some of the features which were identified as significant were
well-identified and researched, for example, the grammatical analysis of
modality. Some appeared to be less standard, for example, my particular
understanding of listing and categorisation in this data, which required
additional identification and description work.
Through this process of identification and preliminary analysis, I drew
up a list of nine features for particular investigation for this set of texts,
and these were:
1. Naming of events.
2. Naming of people.
3. Categorisation.
4. Genre.
5. Intertextuality.
6. Modality.
7. Metonym.
8. Metaphor.
9. Discourses.
The number nine is not significant and the list is not exhaustive. Three
data sets offering a total of 60 texts provide a rich source of linguistic data
with potential for analysis against a wide range of dimensions. Five, ten
or twenty features might have been equally appropriate for investigation,
depending on the nature of the data set and the purpose of the research
Signs
1.Naming of events
2.Naming of people
3.Categorisation
Codes
4.Genre
5.Intertextuality
6.Modality
Mythic meanings
7.Metonymy
8.Metaphor
Ideology
9.Discourses
Fig. 7.1 Nine linguistic features for analysis
work. Nevertheless, the final nine features were the ones that not only
seemed to be pertinent to an understanding of how the BP events were
represented linguistically in the media but also showed movement and
development across the data.
The term “features” is a convenient shorthand; in fact, the linguistic
areas suggested as important are disparate including grammatical systems
such as modality; rhetorical usages such as metaphor; pragmatic and
semantic processes such as naming and categorisation; and supra-textual
concerns such as genre, discourses and intertextuality. It is at this point
in the analysis process that the chosen features could be reconnected with
Barthes’ framework of semiotic levels. Those features that operated at the
level of a word or word group (such as the naming of events) were con-
sidered to be a single “sign”, in the sense of a building block for meaning.
Those features that represented systems or codes were considered to be
at the level of “code”; for example, the modality system, genre and inter-
textuality. At the level of “myth” were two rhetorical tropes—meton-
ymy and metaphor—that serve to add additional connotative meaning
to denotative representation. Finally, an analysis of discourses sought to
uncover the ideological motivation of some of the language choices at
the other semiotic levels. Figure 7.1 shows how the selected features for
the BP data relate to the framework of Barthes’ semiotic levels first intro-
duced in Chap. 3.
Of these nine feature types, some would probably be common to most
linguistic studies of this kind—for example, an analysis of social actors,
genre, intertextuality and discourses is likely to be relevant to most data.
Other features such as categorisation or modality might be more relevant
to some data than to others.
The Immersion Stage in Summary

This second immersion stage served two main purposes:
1. Gaining familiarity with the data. Reading and rereading the data gave
an invaluable sense of the texts which, while not pre-empting the find-
ings from the depth analysis, suggested connections and hypotheses
relating across texts and data sets as well as within individual text items.
2. The output of Stage 2 was a defined set of discursive features for
study, which were suggested by the data themselves, rather than
predetermined.
8
Stage 3: A Depth Analysis at the Level
of the Sign
The next stage of the research was the depth analysis of the nine selected fea-
tures. The first task at this stage was to determine an analysis process for each
of the features that would allow me to produce a description and under-
standing of patterns of language over the timespan of the data sets. Patterns
within and across media texts can be investigated by looking at features in
three ways: the feature in context in a particular year, patterns of the feature
across three years and commonalities across all nine features in a single year.
In order to identify these patterns, the following questions are relevant:
• How can I identify the feature?

• How frequent or significant is the language feature within the data
sets?
• What form does it take, and what function is it performing in
context?
• Does its frequency, form or function change across the data sets over
time? How?
Therefore, for each feature I needed a workable definition and a means of

investigating frequency over time. In addition, I needed to be sensitive to

in Discourse, DOI 10.1057/978-1-137-58750-3_8
the functions of the features—what work was the feature doing in context,
and did this change? With a set of very disparate features for BP, analysis
methods ranged from the very straightforward to the rather complex.
To illustrate the approach with the BP data, for each of the features
selected for study I outline the definition I used and my plan for investi-
gating frequency and function. This is followed by an indicative overview
of my findings for that feature from the BP data. The definitions and
analysis approaches are not necessarily exhaustive, and alternatives could
have been suggested. The critical issue is that they were applied consis-
tently, across all the texts in the data sets and all the time periods. In this
way, internally consistent patterns could be identified. From this point
onwards, my analysis is based on the reduced set of 20 texts for each year
of the data, so any findings such as counts of features are based on like-
for-like comparisons across the same number of texts.
The first level of the semiotic framework is that of the sign, and three
of my nine identified features belonged at this level: naming of events,
naming of people and categorisation.
Feature 1: Naming of Events

Analysing the feature “naming of events” was relatively straightforward.

My concern with this linguistic feature was to identify and discuss the
different ways in which writers refer to the events of the BP Deepwater
Horizon explosion and oil spill. The task was to identify the expressions
within the texts that refer to these events. In some cases these were single
nouns, and in other cases a noun phrase including determiners, adjectives
and prepositional phrases. This raised the question of how much of an
expression to count as a naming choice. My procedure was to count the
whole noun group without the determiner, that is, the noun head and any
accompanying adjectives and adverbials. The following is an example of
a naming choice:
British energy giant BP said Tuesday that first-quarter profits rocketed on

higher oil prices but admitted that the news was overshadowed by last
week’s tragic accident at a rig in the Gulf of Mexico. (Agence France Presse,
27.4.2010, my emphasis)
Here I counted the noun head “accident”, the adjective “tragic” and prep-
ositional phrases of time or place “last week” and “in the Gulf of Mexico”.
I carried out an analysis of the length of naming phrases by calculating
the average number of words in the naming expressions within each data
set. Once I had a set of naming choices for each year, I was further able
to analyse them by whether they were generally neutral or generally nega-
tive in tone. In the first example above, while I would consider the term
“accident” to be a neutral choice on its own, the presence of the adjective
“tragic” would suggest a negative shading.
Using this definition and analysis approach, the data were analysed in 20
texts from each year. The results across the three years—summarised in
Table 8.1—showed a number of significant features.
The number of times the events are named at all drops from 89 in 20 texts
in 2010 to 41 in 20 texts in 2012. By 2012, the BP story has become part of
a bigger picture, and is often mentioned only once per news item. The three
most mentioned descriptors (excluding “spill”) are consistently “oil spill”,
“disaster” and “explosion”. In 2010, these terms accounted for 38 of the 89
terms used (43 %). By 2012, they accounted for 29 of the 41 terms used
(71 %), suggesting an ever closer alignment within the media concerning
how the events should be named. The average length of the nominal group
Table 8.1 Analysis of naming terms for the BP Deepwater Horizon events
2010 2011 2012
Number of Number of Number of
naming naming terms naming
terms (per (per 000 terms (per
000 words) % words) % 000 words) %
Neutral 73 (8.0) 82 39 (5.6) 60 23 (3.2) 56
Negative 16 (1.8) 18 26 (3.7) 40 18 (2.5) 44
Total 89 (9.8) 100 65 (9.3) 100 41 (5.8) 100
increased from 2.2 words in 2010 to 3.2 words in 2012, as the noun head
is accompanied by a greater number of adjectives and adverbials. This type
of density in noun groups is typical of journalistic prose where a lengthy
noun group concentrates a considerable amount of information in a small
space. What is important about these dense nominal groups is that they are
selective: they organise and categorise the events in a particular way, which
includes certain features or evaluations and excludes others.
Typical of 2010 are names such as:
“tragic incident”
“oil rig spill”
“potential environmental disaster”
By 2012, naming choices have become longer and more descriptive:
“the BP Deepwater Horizon nightmare in the Gulf of Mexico in 2010”

“the US Gulf of Mexico oil spill disaster in 2010”
“the 2010 BP Macondo rig disaster”
“the BP Deepwater Horizon oil disaster”
The qualifying adjectives and adverbials alongside the noun head are
commonly temporal and spatial, but the nouns are often highly evalua-
tive. Naming terms used in 2012 are even more likely to be negative (by
proportion) than they are in 2011, and this follows a pattern from 2010.
“Disaster” is now as commonly used as “oil spill”, increasing the number
of negative references that together now account for 44 % of mentions.
In Summary
Over the three-year period, naming terms:
• Are drawn from a smaller pool of terms, with the consensus clustering
around “oil spill” and “disaster”.
• Increase in terms of number of words used in the descriptor. This is
judged to be largely due to journalistic convention (the greater the
distance from the events in time, the greater the need for specific
identification). However, importantly, an increase in the length of the

naming term does not entail an increase in the specificity or the accu-
racy of the overall description of events. Rather, these longer terms
replace detailed descriptions, thereby reducing depth and accuracy and
offering a simplified depiction of the events.
• Are increasingly likely to be negative in tone. This finding should be
read in conjunction with the finding on modality, which suggests an
increasing level of certainty about the nature of events and how we are
to understand them.
The clustering around negative descriptors is unsurprising in itself, given the

serious nature of the explosion and oil spill, and the media here acknowledge
rather than in any sense downplay that seriousness. The question remains of
how this negative naming of the events by the media is positioned: whether
these negative shorthand terms are contextualised to suggest that events such
as Deepwater Horizon are either regrettable but inevitable or aberrant and
preventable. The regular use by the media of certain shared terms, however
negatively shaded, can serve to familiarise and compartmentalise deviant
phenomena such as crises, as much as mark them out as shocking.
Feature 2: Naming of People

As in the case of the naming of events, an analysis of the naming of people

involved identifying mentions of human participants in the texts, count-
ing them and looking at the types of descriptor involved. Participants were
either individuals or groups of people. They were either specifically named
(e.g. “Alison Reed”) or referred to by an indefinite noun, a title or a group
name (e.g. “a young man”, “company officials”, “fishermen”). However, I
omitted personal pronouns (e.g. “she”, “they”) from the analysis. Multiple
descriptors were sometimes used, for example, “Committee chairman and
TDC member Marty McDaniel”. Although the great majority of partici-
pants were real, I also included in my analysis people who were fictional,
such as book characters. In the same spirit, I included in analysis people
and groups who were hypothetical and representative rather than specific,
Table 8.2 Social actors in 2010–12 BP texts

2010 % 2011 % 2012 %
BP staff 44 29 12 10 4 2
US agencies 37 24 10 9 10 6
Universities and private agencies 22 14 4 3 1 1
Workers, public, local businesspeople 17 11 13 11 32 34
BP management 14 9 23 20 2 1
Politics 9 6 8 7 33 19
Business and media—comment 3 2 32 27 19 11
Finance 3 2 7 6 1 1
Lawyers 1 1 2 2 1 1
Art-related (writers, artists and 0 0 3 3 42 24
fictional characters)
Others 3 2 3 3 3 2
Total including repeated mentions 153 100 117 100 148 100
Total per 1000 words 16.8 16.7 20.8
Total excluding repeated mentions 61 59 89
for example, “your grandkids” in the expression “What are you going to
say to your grandkids when they say …”, or “designated managers” who
would be dealing with a hypothetical crisis.
Having gathered all instances of human actors, I looked for logical
groupings, for example, “BP employees” at each of executive and staff levels,
representatives of agencies and so on. From this analysis I was able to inves-
tigate the salience of different groups to the story, and how this changed
over time. I was able to look at the proportion of named and unnamed
participants, and what groups each belonged to: and for both named and
unnamed people, I could see what descriptors were used to define them. I
was able to look at whether participants were real, fictional or hypothetical
and whether this changed over time. Finally I could identify movements in
participants’ proximity to the BP story, looking at how closely related the
people mentioned in the news texts were to the actual Deepwater Horizon
events, and whether and how this changed over the three years.
In 2010 and 2011, the number of actors mentioned is nearly 17 per 000
words. By 2012 there is an increase to about 21 per 1000 words. Put
simply, more people are being drawn into the BP story, or rather the BP
story is being widened to include more people. The proportions of types

of social actor are also considerably changed, as Table 8.2 shows, with
some types increasing and others disappearing from the story.
The largest groups of people mentioned in 2010 are (1) BP staff (2)
US agencies and (3) universities and private agencies. By 2012, the three
largest groups of people, accounting for three-quarters of actors, are (1)
members of the public and the community (2) writers, other artists and
fictional characters and (3) those in politics. This represents a consider-
able shift in the cast of stakeholders from both 2010 and 2011 to 2012.
By 2012, mentions of BP management and staff are almost completely
absent. Where the name of Tony Hayward does appear, it is either as the
neutral but familiar “ex-boss of BP” (Campaign Middle East, 27.4.2012)
or the evaluative “BP’s hapless chief executive, Tony Hayward” (The
New York Times, 27.4.2012). The one named BP employee (Kurt Mix)
is mentioned in the context of his arrest for deleting electronic evidence
in relation to the disaster. Those mentioned in the areas of business and
media comment are still evident, but these are now secondary to those
mentioned in an artistic context.
Members of the public are increasingly mentioned in broad groups—for
example, “thousands of people”, “Americans”, “the public”, “consumers”—
and are ever further away from the events, for example, people who live in
a town (not near the Gulf) whose work situation has been affected by the
Gulf oil spill, or the brother- and sister-in-law of the journalist visiting New
Orleans to report on the state of the town after recent disasters. Similarly,
politicians, who have an increased presence in the 2012 texts, are less closely
connected to events than is the case in 2010. At that time, key figures include,
for example, the Chair of the Energy committee, and Governors of the Gulf
States. By 2012, they include figures much further away from events, for
example, George W. Bush, UK Chancellor George Osborne and President
Vladimir Putin. This is another indication of the shift of the representation
of the crisis from being something highly situated and local, to something
representative and global (even though we know that newspaper coverage is
located increasingly in the USA). The considerable proliferation of people
mentioned is shown in the analysis which excludes repeated instances. This
indicates that 89 unique people or groups are mentioned out of 148 total
participant-mentions—a higher proportion than in previous years, where
the same people tend to be cited repeatedly.
As indicated in the overview of text genres, a significant change in

2012 is the emergence of texts with a connection with fiction and non-
fiction writing, art and music. This brings with it a new cast of characters
whose part in the BP story is extremely diverse:
[Review of a film based on a Margaret Atwood book] Also thrown into the
mix are Conrad Black, the disgraced media mogul who went to prison for
mail fraud, a tattooed Canadian man serving time for robbery and abused
migrant tomato pickers in Florida. All are subjects worthy of discussion, but
tackling them in one film disrupts the movie’s momentum and short-
changes viewers. Baichwal could have devoted a single film to just BP’s
disgraceful behavior. (The New York Post, 27.4.2012, my emphasis)
[New Orleans’ recovery from Hurricane Katrina, the financial crisis and
the BP oil spill] Fast-forward to April 14, 2012. There were no musicians
enlivening the concourse as I arrived this time, but there would be hun-
dreds of them down along the sunny riverfront, where an estimated half-
million people make a pilgrimage each year to the French Quarter Festival.
(Sarasota Herald Tribune [Florida], 27.4.2012, my emphasis)
The references to written and performed art generally entail the BP events
being placed within a wider context, with no restriction on participants.
By 2012, readers and viewers are expected to understand what the events
might represent, or what social meaning they have, even as further
light is shed upon them through their juxtaposition with other social
phenomena. A number of the people included in the “Art” category are
not real but fictional.
In Summary
The pattern of participants in the texts over the years is one of increasing
fragmentation and dispersal. More and more people are mentioned in
connection with the BP story, and there are more unique mentions rather
than repeated individuals, yet they have weaker and more distant connec-
tions with the BP crisis. BP employees and the victims of the explosion
are virtually absent by 2012.
Feature 3: Categorisation
The observation from my preliminary analysis which suggested categori-

sation as a feature for study was that the BP oil spill is included in groups
and compared and contrasted with other entities both like and appar-
ently unlike itself.
Examination of the data indicated that this categorisation process
appears in a number of interrelated forms:
1. Comparisons
Events were referred to in constructions such as “one of …”, “the big-

gest …” and so on. Comparing crisis events explicitly and implicitly with
other similar events is one form of categorisation common in media rep-
resentation, and is intended to help the reader make sense of the current
situation.
2. Lists of entities
A number of entities were listed, for example, “Katrina, the BP oil spill
and other events”, and the BP crisis was on the list.
3. Using the events as an index of something else
This indexical process can be recognised where the events were a

“shorthand” for something else. This analysis linked with naming choices
for the events, where there began to be a socially agreed interpretation of
what names for the events stood for. In this way, the BP events and the
indexed phenomenon create a category together. Sometimes this process
was realised through one of the other practices mentioned here, such as
comparison or listing:
Whether you are the leader of one of the Arab Spring countries, the ex-boss
of BP, or a fashion designer prone to drunken, racist out-bursts—if you
don’t behave in the right way, people will remove you (Campaign Middle
East, 27.4.2012)
This list refers to an element of the BP story as an exemplar of, or

“standing for”, unacceptable behaviour.
4. Contrasts and absences: What the events are not
Sometimes the events were categorised with an emphasis on what

they were unlike rather than what they were like. In the 2010 data, the
expression “We’ve never seen anything like this magnitude” (Associated
Press Financial Wire, 27.4.2010) did not imply that nothing as big as
this has ever happened anywhere, but that this was an exceptional event
amongst the group “oil spills” or “man-made environmental disasters”.
Such exclusions, omissions and redefinition of categories were worthy of
examination.
Instances of what I call here “categorisation” are rare in the 2010 data set,
and only six are identified in the 20 texts:
We’ve never seen anything like this magnitude (Associated Press Financial
Wire, 27.4.2010)
We’ve never seen anything like this magnitude (BreakingNews.ie,
27.4.2010)
We’ve never seen anything like this magnitude (Trend Daily News
(Azerbaijan), 27.4.2010a, 27.4.2010b)
If we don’t secure this well, this could be one of the most significant oil
spills in US history (Carleton Place (Canada), 27.4.2010)
If the missing workers died, it would be the deadliest US offshore rig
explosion since 1968 (TendersInfo, 27.4.2010)
In 1990, a similar bid to change the rules failed, in part because it fol-
lowed the Exxon Valdez spill. Now observers think the Gulf of Mexico acci-
dent could do much the same. (The Globe and Mail (Canada), 27.4.2010)
Three of these instances are identical, where the words of the same
expert source have been quoted. The implication of “never seen anything
like this magnitude” is that the BP events at this point are so far not one
of a group, and I mention in describing my analysis approach that refer-
ring to what phenomena are not is one way of indicating what they might
be. In the fourth and fifth examples above, the use of the conditional (“If
… this could be”; “If … it would be”) signals uncertainty and as yet unre-
alised potential. In the sixth example, the comparison with the Exxon
Valdez spill is highly mitigated (“observers think … much the same”). So
in 2010, what categorisation exists is expressed in negative terms, realised
in the conditional, or highly mitigated. At this point, writers are not
attempting or only cautiously attempting to define what the events are,
and where they “fit” into already processed understanding.
By 2011, instances of categorisation increased to 17 in 20 texts, indicating
a greatly increased presence for this language feature. Categorising expres-
sions in 2011 locate the Deepwater Horizon events within a number of dif-
ferent groups. These include the perhaps surprising group of natural disasters
(given that the explosion and oil spill were not naturally occurring events):
Looking back at Katrina, the BP oil spill and other events, it does not
appear that we are prepared to respond effectively or to recover from natural
disasters, let alone a major terrorist attack. (Journal of Counterterrorism
and Homeland Security International, 27.4.2011)
Secondly, other texts place the events alongside business crises, that is,
events that disrupt the normal run of business, and are to be dealt with
and normalised:
Recently companies such as Toyota, BP, Johnson & Johnson and Hewlett
Packard have experienced crises that distracted management, cost millions
of dollars in time and resources, reduced shareholder value and resulted in
lawsuits that will take years to resolve. (Executive Counsel, 27.4.2011)
Thirdly, the 2010 oil spill is positioned as just one of a number of BP

problems, where news, finance and business reports in the 2011 data indi-
cate that the oil spill is not the only area of BP concern. As 27 April
2011 is another financial results day, there is media interest in the part
Deepwater Horizon is playing in the general business performance of the

company. The following is an example of a synchronic list, where a set of
events contribute to a state at a point in time:
A still-rising bill for the Gulf of Mexico disaster, lower production after
selling off assets to help pay for it and a hit from the Budget’s tax grab on
North Sea oil profits saw BP’s profits fall 2 % in the first three months of
the year. (The Evening Standard [London], 27.4.2011)
An example of diachronic categorisation is the listing of BP accidents and

disasters, of which Deepwater Horizon is the most recent:
In 2005, fifteen workers were killed when BP’s Texas City Refinery
exploded. In 2006, corroded pipes owned by BP led to an oil spill in
Alaska. Now, in 2010, eleven men drilling for BP were killed in the blow-
out of the Macondo well in the Gulf of Mexico. (M2 PressWIRE,
27.4.2011)
The majority of the lists and groups unsurprisingly set the events into
categories of negative experience, but there is one quite interesting excep-
tion. This is taken from a press release by BP of 25 April, acknowledged
openly in the text “According to a release”. The writing positions the BP
events more neutrally than is typical for this year’s data set:
... Scientific understanding of oil spill and dispersant impacts on ocean and
coastal systems in the Gulf region, as well as other ocean and coastal sys-
tems, and how these systems respond to oil and gas inputs, especially acci-
dental inputs. (Wireless News, 27.4.2011)
The language is rather opaque—in fact it is difficult to see that the research
referred to is a direct result of the 2010 Deepwater Horizon events,
although these are referred to indirectly twice as “oil spill and dispersant
impacts” and “oil and gas inputs, especially accidental inputs”. Oblique
forms of reference recur throughout the full text. The phrase “scientific
understanding” serves to position the research as academic and objec-
tive, and the listing of the Gulf region alongside other ocean and coastal
systems implies that the research is being carried out for the benefit of
the global community. Similarly “oil and gas inputs” generalises (makes
vague) the object of the research. The reference to “accidental” inputs
(unspecified) again has a role in distancing blame and responsibility. (In
making this observation, I do not imply that the 2010 blowout was in
any way deliberate, rather that the word “accident” implies unavoidable
chance rather than failures of responsibility.)
The bulk of categorisation in 2011, then, positions the BP events along-
side other business crises, natural crises, BP’s 2011 difficulties and oppor-
tunities to gain scientific knowledge for the future. These are expected
and conventional groupings, given the topic. This process has started to
construct the events as an exemplar of something, often something out-
side itself, for example, something to be dealt with by the International
Association of Counterterrorism and Security Professionals (IACSP).
Thus categorisation works in a metonymic way, with the part that is the
reference to BP serving to represent a whole that is a more graspable type
of phenomenon. We begin to see the process of myth-making described
by Barthes in action—that the signified has become in turn a signifier
of something else. These events are no longer unique, no longer distinc-
tive—media confidence in representation has moved on from the 2010
phrasing “We’ve never seen anything like this magnitude” to positioning
the events alongside other phenomena that it is suggested to resemble.
The detail about the events that characterised the 2010 reports is no lon-
ger a feature; the events are referred to in a shorthand through selected
naming practices (analysed earlier), and in lists, comparisons and group-
ings that indicate to the reader how he/she is expected to understand and
locate this particular phenomenon.
In 2012, categorisation remains a significant feature of media writing,
with 18 instances appearing across the 20 texts. The categories into which
the BP events are placed include some that are similar to those in 2011—
what I have called “expected groupings” such as BP business problems:
BP has had a torrid time of late dealing with the costly repercussions of the
Deepwater Horizon disaster and the failure of its Russian Arctic venture.
(CompaniesandMarkets.com, 27.4.2012)
Another expected group is the oil spill as the representation of the cat-
egory “disaster”. In the fragment below, these are disasters that have
affected the city of New Orleans:
After the storms, New Orleans endured the 2008 economic crash and the
BP oil spill. (Sarasota Herald Tribune [Florida], 27.4.2012)
However, a particular characteristic of categorisation emerges in the 2012

data. There is an increasing unexpectedness and creativity of the group-
ings and listings. In 2012, the BP crisis is being made an exemplar of an
increasingly disparate set of social phenomena. The following examples
are illustrative:
Whether you are the leader of one of the Arab Spring countries, the ex-boss
of BP, or a fashion designer prone to drunken, racist outbursts—if you
don’t behave in the right way, people will remove you, and the weapon they
will use is social media. (Campaign Middle East, 27.4.2012)
This [the complex issue of debt, both moral and financial] includes BP’s
failure to deal with its environmental transgressions, and the years-long
dispute between two poor rural clans that keeps the members of one family
virtual prisoners in their own home. (The New York Post, 27.4.2012)
The first example in particular is illustrative of the changes in some

of the instances of categorisation in 2012. The category in this case is
individuals who “don’t behave in the right way”, and so is broad and
only indirectly related to the BP oil spill. To understand the point being
made by the writer, in this case a businessperson writing in the advertis-
ing trade journal “Campaign”, the reader needs to have knowledge of a
range of unconnected events in the news. To relate to the thought, the
reader needs to share a perspective that recognises these three examples
as instances of bad behaviour. So the social and cultural resources the
reader needs to bring to an understanding of the first example above are
far greater than those needed in previous years at an earlier stage of the
representation process. This later type of categorisation demonstrates a
form of meaning-making that depends less on ostensibly objective, fac-
tual representation than on construction within a wider, culturally agreed
and evaluative context.
In Summary
By 2012, the positioning of BP events within groups can be either

expected or unexpected, and can serve as an exemplar for a number of
phenomena, in diverse arguments. The movement from 2010 to 2011
to 2012 is that from a low level of tentative, provisional categorisation,
through a relation with oil spills, crises and BP problems to finally a posi-
tioning with unrelated events, the world’s ills and general political issues.
Part of the reason for a new creativity in categorisation is that the
events are being categorised within the context of evaluative or persuasive
rather than descriptive texts. Texts such as editorial and opinion pieces
use language that is high in affect, with the result that the process of cate-
gorisation appears to be less a listing of phenomena, as in 2011, and more
an accumulation of evidence within an (often impassioned) argument.
Taking the examples above, in the first fragment the phrases “drunken
racist outbursts”, “behave in the right way”, “weapon they will use”, all
express judgement, as does “failure” in the second text.
9
of the Code
Chapter 8 discussed three language features at the semiotic level of the

sign. The depth analysis continues with a further three features which are
situated at the level of the code: genre, intertextuality and the system of
modality.
Feature 4: Genre
I have considered issues of genre in two places in my analysis. I outlined

at the contextualisation stage those genres which were represented in the
data, and indicated the frequency and relative importance of the genres
within the data sets, and how these changed over three years. In order to
do this, I offered a definition and a categorisation of news media genres,
found in Fig. 6.1, and these remained the basis for my consideration of
genre at this depth analysis stage. The findings which follow focus on the
extent to which typical genre characteristics were evident in my data, and

in Discourse, DOI 10.1057/978-1-137-58750-3_9
how the dominant genres in each data set influence the ways we under-
stand the crisis representation.
In 2010 texts are primarily news reports and financial reports. By 2012
the text genre types are much more varied, and include news reports, edi-
torial pieces, market reports and arts reviews. This change can be summed
up as both a widening of coverage in terms of genre and a shift from pri-
marily descriptive genres towards more evaluative genres. I outline some
key findings from the data for the following genres:
1. News reports.
2. Financial reports.
3. Features.
4. Editorials or opinion pieces.
5. Arts reviews.
6. Business or marketing reports.
Genre 1: News Reports
In Chap. 1, I drew upon Fulton, Huisman, Murphet, and Dunn’s (2005:

233) characterisation of the genre news reports as exhibiting a number of
language features, including:
1. A high proportion of empirical information.

2. Third-person narration where the narrative voice is externalised and
elided.
3. Lack of modality, preponderance of declarative verbs indicating
certainty.
4. Use of the inverted pyramid structure.
These typical generic features are largely present in the 2010 BP news
reports.
9 Stage 3: A Depth Analysis at the Level of the Code 129
Many of the texts contain a significant use of empirical information,

including the detailed naming of people and clear specificity of time and
place. At this point in the BP events, one week after the explosion, the
news media is still aiming to answer the canonical questions of news
reporting: “who, what, when, where and why?” The following text extract
is typical in respect of its focus on “the facts”:
The oil is now about 20 miles (32 kilometres) off the coast of Venice,
Louisiana, the closest it’s been to land. But it’s still not expected to reach the
coast before Friday, if at all.
BP, which was leasing the Deepwater Horizon, said it will begin drilling
by Thursday as part of a $100 million effort to take the pressure off the well,
which is spewing 42,000 gallons (159,000 litres) of crude oil a day.
Company spokesman Robert Wine said it will take up to three months to
drill a relief well from another rig recently brought to the site where the
Deepwater Horizon sank after the blast. Most of the 126 workers on board
escaped; 11 are missing and presumed dead. No cause has been deter-
mined. (Carleton Place [Canada], 27.4.2010, my emphasis)
This extract makes use of an accumulation of facts and figures including

distances, volumes, temporal expressions, names and titles and employee
numbers. In some respects, there is a high degree of specificity. The con-
versions of miles to kilometres and gallons to litres connote an atten-
tion to detail, and signal an orientation towards both national (in this
case, Canadian) and international readers. The source of information
(“Company spokesman Robert Wine”) is clearly named and titled, and
the numbers of missing and surviving employees are given exactly. In
support of this objective presentation, the future tense declarative mood
is mainly used, indicating firmness of intent. For example, in paragraph
2: “BP…said it will begin drilling” and in paragraph 3, “Robert Wine
said it will take up to three months”.
Nevertheless, there is a degree of mitigation in the presentation of
facts. Distances are approximate (“about” 20 miles in paragraph 1) and
temporal expressions are vague (“before Friday, if at all”, “by Thursday”,
“up to three months”, “recently”). This can be plausibly explained by the
fact that at this stage in the development of the crisis, information was
limited or absent, but I would also suggest that a degree of imprecision
can support rather than undermine the representation of an objective

reality. Not only does the reader not need to know the exact number of
gallons of crude oil being lost by the well, but a figure of, say, 42,367 gal-
lons would invite incredulity and challenge. In fact, the figure of 42,000
gallons, presented with an unmitigated declarative (“is spewing”), was far
from a generally agreed figure, being at the lowest end of BP’s declared
estimate and less than a tenth of BP’s internal estimates (Bergin, 2011:
171).
Fulton et al.’s reference to the elision of the authorial voice is manifest
in a number of forms in the data. Attributed speech is one way of show-
ing that any opinion and judgement within the piece are not the writer’s
own and individuals and institutions quoted are usually presented as hav-
ing a warrant to speak. In 2010, the warrant is either (1) expert status,
often technical or environmental, or (2) eyewitness status, as shown in
the following data extracts:
1. “That system has been deployed in shallower water, but it has never
been deployed at 5000 feet of water, so we have to be careful,” he
[Doug Suttles, chief operating officer of BP Exploration and
Production] said. (BreakingNews.ie, 27.4.2010)
2. “We can only hope that they can make that sucker stop very soon,”
said Wilton “Tony” Sturges, a retired Florida State University ocean-
ographer. The winds that would push the spill toward Tampa Bay’s
beaches do not normally start until midsummer, he noted. (St.
Petersburg Times [Florida], 27.4.2010)
Both illustrations are examples of personal opinion, in both cases using

modal auxiliaries to signal uncertainty—“we have to be careful” and “we
can only hope”—which in this genre is typically expressed through speak-
ers other than the journalist, while the journalist usually uses declaratives,
and is sparing with modal expressions.
The inverted pyramid in Western news reporting is a key marker of the
“objective” style, as it is interpreted primarily as emphasising the factual
nature of the story, rather than offering evaluations or telling personal
stories. This structure is recognisable where key facts are presented in
summary first, followed by detail later. In this way, the point of closure
comes at the beginning of the piece, rather than at the end (as is generally
the case in a narrative structure). This structure is evident in many of the
texts, for example:
BP revealed a 135 % rise in first-quarter profit to $5.6 billion (£3.64 bil-

lion) today as the oil giant announced it is accelerating its clean-up of the
Gulf of Mexico oil spill. (The Evening Standard [London], 27.4.2010)
This introductory section flags up the two key topics to be explored in

the subsequent article; firstly, the BP first quarter results, with a dense
summary of the key points (fact of increase, size of increase, size of profit)
and, secondly, the oil spill (new news on the oil spill, acceleration of
clean-up efforts).
There is an interesting exception to this typical inverted pyramid struc-
ture in the 2010 texts. In the following text (Associated Press Financial
Wire, 27.4.2010), there is an example of a narrative structure. It is pos-
sible that this strategy was used because the reporting organisation had
an exclusive interview with a direct participant in the events, a cook on
the rig who was one of the survivors, and this was felt significant enough
to warrant an atypical approach. Unlike the inverted pyramid structure,
narratives are characterised by a sequential structure, and by choices
in lexis, personal reference and evaluative elements that serve the pur-
pose of engagement and entertainment rather than that of information
and description. The following extract follows in written form Labov’s
(Labov & Waletzky, 1967) spoken narrative structure of abstract, orienta-
tion, complicating action, evaluation and resolution, and thus gives the
impression of recounting the story as it was told to the journalist.
Cook on La. oil rig that exploded recalls escape ABSTRACT
[headline]
Oleander Benton, a cook on an oil rig that exploded ORIENTATION
off the Louisiana coast, was sitting at a laundry
room table with a friend when the lights went out.
Then, there was the blast. COMPLICATING ACTION
The Deepwater Horizon platform shuddered, debris
fell from the ceiling and Benton hit the floor, as she
had been trained to do. She scrambled through
hallways littered with rubble, following a man in a
white T-shirt.
“I could not see anything but that man. He just kept
on saying ‘Come this way, come that way.’ It was like EVALUATION
he was coaching me to my lifeboat, because I
couldn’t see,” she said.
She made it across the sweltering, mud-caked deck to RESOLUTION
a lifeboat one of 115 people to safely escape the
platform after the explosion a week ago.
The personal story presented above forms only part of the report. After
these paragraphs, the journalist reverts to a more typical news structure
with the following statement:
Benton, 52, recalled her tale as crews used a remote sub to try to shut off
an underwater oil well that’s gushing 42,000 gallons a day from the site of
the wrecked drilling platform.
This extract above marks the transition between the narrative section
and the more typical news report section, where facts and figures (“52”,
“42,000 gallons a day”) work to portray an objective reality. Thus the
narrative section is embedded within a more typical news report structure
and language features. Fulton et al. (2005: 146) suggest that the narrative
style, when found within or alongside news reports, may have a specific
function—that of showcasing the objective style as a contrast:
A narrative model also allows more “attitude” to be expressed: evaluations

of behaviour and outcomes are coded into the narrative structure. Using
the narrative model for some items therefore can have the effect of posi-
tioning the information model as objective and neutral by comparison.
In the same way as the tentativeness noted in the presentation of facts

and figures, small shifts away from expected patterns of “objective” style
or structure can support rather than undermine the representation of
objectivity.
By 2011, news reports do not all primarily focus on the BP Deepwater
Horizon events in the way that the 2010 news report texts did. Fewer
of the texts in 2011 offer “new news” on Deepwater Horizon and its
aftermath, and these include “BP expects to resume Gulf drilling this
year” (The Associated Press, 27.4.2011), “Long legal battle ahead over
Macondo” (EI Finance, 27.4.2011) and “GRI research board announces

request for proposals for BP’s $500 million Gulf of Mexico research ini-
tiative” (ENP Newswire, 27.4.2011). The rest of the news reports refer
to the BP events only in passing. “Unemployment falls in 80 pct. of large
cities” (The Associated Press, 27.4.2011) is a good news story about the
recovery of the economy, but points out that seven of the ten cities with
the largest increases in unemployment are in the area most affected by
the BP oil spill. Another text, “Rubio: National debt can no longer be
ignored” (States News Service, 27.4.2011), reports on a speech made by a
Florida senator that includes comments on compensation claims against
BP for the effects of the oil spill. In these cases the proportion of each
story relating to BP is small, and these texts show how the BP story relates
to phenomena that are outside the events themselves.
By 2012, news reports still represent 42 % of texts; however, as was the
case in 2011, the BP story is not the focus for the majority of these. An
oil spill in Yellowstone (Lewiston Morning Tribune [Idaho], 27.4.2012)
has given rise to a need for fish testing; however, specialist laboratories for
this work are still backed up with work from the Deepwater Horizon oil
spill. In two reports, Congressman Frank Pallone makes a statement on
seismic testing in the Atlantic Ocean, which makes reference to the BP
oil spill, and Interior Secretary Ken Salazar defends the Obama admin-
istration’s record on energy. However, two years on from the spill, there
are two new stories that directly relate to the BP spill: one is the arrest of
a BP engineer in connection with the spill and the other concerns chal-
lenges to the administration of the compensation payments made by BP
to individuals and organisations after the spill.
Genre 2: Financial Reports
Like news reports, financial reports have the function of presenting

a great deal of factual information in limited space. They typically
draw on commentary from industry experts (those with a warrant to
speak) although not from members of the public. They are unlike news
reports in two main respects: firstly, that they tend to be addressed to a
more specific and knowledgeable audience than general news reports.
In terms of language choice, the expert audience allows for specialist

and technical language to be used without supporting explanation. For
example the following extract is from the London Evening Standard
(27.4.2010):
Still, BP’s $5.6 billion replacement cost profit which trips out fluctuations
in the value of oil inventories smashed City expectations of $44.8 billion.
As well as its use of technical lexis (e.g. “replacement cost profit”), this

text makes no concession to the reader by converting from dollars to
pounds sterling. Further, the figures are very precise. The second specific
feature of financial reports is the significant use of spatial and physical
metaphor, which is commonplace enough to count as “dead” or “dor-
mant” metaphor in most cases, but which still gives a dynamic dimension
to otherwise straightforward accounts. This issue will be explored in more
detail in discussion of metaphor, but is exemplified by the choice of “trips
out” and “smashed” in the text above. The same text offers the metaphors
“soar”, “rise”, “depth”, “boost”, “flat” and “lower”.
In both respects (factual information and spatial metaphor), the finan-
cial reports of the 2010 data set are typical of the genre. What is some-
what less typical is the amount of space given to the Deepwater Horizon
events in this specialist type of report, and this is an indication of the
seriousness of the events, and the size of the likely impact on the business
(as yet unknown).
The financial report genre is still significant within the 2011 texts. The
oil spill is still a major part of BP’s financial picture, but the account of
the events is more likely to be shortened, summarised and taken as read,
as shown in the following extract (Agence France Presse, 27.4.2011):
In London, BP shares gained 1.43 percent to 470.85 pence after the energy
group posted a 17-percent jump in first-quarter net profits.
Earnings after taxation leapt to $7.124 billion (4.9 billion euros) on the
back of surging oil prices, one year after being hit by the US oil disaster.
However, BP also upgraded the cost of last year’s devastating Gulf of
Mexico spill to $41.3 billion. That compared with previous guidance of
$40.9 billion. (my emphasis)
The BP events constitute only part of the BP financial picture and the
BP financial report itself is placed in the wider context of a report that
also comments on other companies’ financial results. In this respect, the
2011 financial reports continue a process begun in 2010, in that they
characterise the Deepwater Horizon explosion as an explanatory factor
for BP financial and business performance. In this respect the financial
reports resemble 2011 news reports that have started to marginalise the
story. In 2012, 27 April is not Quarter 1 results day, and there are only
two instances of financial reports in the full sample.
Genre 3: Feature Articles
Generic characteristics of feature articles are that they present their stories
as universal rather than time-bound, they often include human interest
elements, their structure tends to be narrative rather than the inverted
pyramid, which means that the point of closure is towards the end of
feature articles rather than the beginning, and they present a subjective
rather than objective point of view. The two feature articles in the 2012
data set are typical of the genre. Both, coincidentally, report on New
Orleans, and its renaissance since Hurricane Katrina and the BP oil spill.
One contains a typical example of categorisation.
After the storms, New Orleans endured the 2008 economic crash and the
BP oil spill. (Sarasota Herald Tribune [Florida], 27.4.2012)
The text is presented from a first-person perspective: the journalist

intrudes upon the story, making her visit and her family part of the com-
mentary (“As my sister and brother-in-law and I drove there from the
airport”) and structuring the report around a comparison between New
Orleans and her home town of Saratoga. The article closes with (has as its
point of closure) its key argument.
There it is: the sense of place that sets a city apart from all the rest. Maybe
we, too, will get there someday.
The article is illustrative of two related shifts from the news texts of 2010:
in the type of reporting from impersonal and objective news reports to
more evaluative and personal reports of various kinds such as editorials,
reviews and feature articles; and in the absorption of the BP story into
other stories, where it both illustrates and forms part of particular phe-
nomena—in this case, the tribulations undergone by New Orleans that
form the backdrop to a good news recovery story.
Genre 4: Editorial or Opinion Piece
In terms of rhetorical acts, the 2010 texts were primarily descriptive. By

2011, there is an increase in texts that are to some degree evaluative. The
increase in texts in the editorial or opinion piece genre in 2011 suggests
that writers have begun to consider that they are now able to put the
2010 events into context, with the purpose of making judgements about
them (and the first anniversary is a motive to do so). The opinion piece
genre is characterised in genre literature by the use of lexis that is rich in
emotion and judgement, modal expressions that interpose the writer’s
view on the propositions made and rhetorical persuasive features.
The editorial and opinion pieces in the 2011 data set are typical of the
genre in that they exhibit judgement on the part of the writer. In these
pieces it becomes more common that the writer takes responsibility for the
proposition, rather than attributing it to another source, as is characteris-
tic of news reports. The following extracts are examples of opinion pieces:
If the BP oil spill hadnt happened¦ | Deep Sea News #dsn #ocean
RT: One year after the BP disaster, tell the President no new drilling: #B.
(The Right Blue via Twitter, 27.4.2011, my emphasis)
Ironically, while American oil companies are banned from drilling in the
Gulf of Mexico, other countries are not. (CaptainKudzu, 27.4.2011, my
emphasis)
OPA [Oil Pollution Act] requires our state—like any other party harmed
by the oil spill—to present a claim to BP before resorting to a lawsuit.
Although Florida has at least three years from the date of the oil spill to
assert its legal rights under OPA, we intend to file a claim with BP this
summer. If BP does not do the right thing and pay that claim, I will not
hesitate to take BP and any other responsible party to court. (Tampa Bay
Times [Florida], 27.4.2011, my emphasis)
The first example is from a news digest piece that gathers together tweets
and retweets them, the second is a blog and the third is a fragment of a
letter to a newspaper. These three pieces are interesting in that they convey
judgement and evaluation through diverse strategies. The first tweet of the
first example makes use of the modal clause “If (only)…then”, but uses
only the “If ” to indicate modal intent, perhaps reflecting the necessary
economy of Twitter communication. The second retweet uses an impera-
tive with no specific addressee—“tell the President”—to convey the writ-
er’s view. In the blog, the comment adjunct “ironically” serves to indicate
the writer’s view, while the rest of the proposition is an unmitigated declar-
ative. The letter from the Florida Attorney General takes a much less con-
versational but no less direct tone. The lexis is formal and legal (“OPA”,
“party”, “assert its legal rights”, “file a claim”) but the text is unequivocal
in its expression of judgement through social sanction (Martin & White,
2005) in the phrase “If BP does not do the right thing”. In the three pieces
the authorial position is being conveyed partly through the resources
of modality, as might be expected, and partly through other linguistic
resources—here choice of mood and the Appraisal System (lexis).
Earlier, I mentioned argumentation and persuasion strategies as char-
acteristic of evaluative genres, and these are also found in the 2011 edito-
rial/opinion texts. The following piece is an online newsletter reflection
on the rising price of oil:
Looking for reasons why benchmark Brent crude is trading around $120
per barrel? There are plenty. Rapid economic recovery in emerging econo-
mies in Asia, political turmoil in the Middle East and North Africa and
constrained supplies from the deepwater Gulf of Mexico are a few. (EI
Finance, 27.4.2011)
A number of persuasive rhetorical strategies are exhibited in this short

extract. The piece exhibits a Preview–Detail structure (Hoey, 2001;
Winter, 1994) where the topic of reasons for high oil prices is raised, and
reasons are enumerated. The fact that three reasons are chosen gives a
balanced tricolon pattern to the final sentence. It is typical of persuasive

argument (Cockcroft & Cockcroft, 2005) that the writer asks a question
(“Looking for reasons…?”) that his subsequent argument will answer.
The reasons are presented as fact, not opinion, with no mitigating expres-
sions. This style of writing continues into the 2012 texts, and a similar
tone is found in the extract below:
We’re living in a world of radical transparency. It’s changing the rules of

marketing and offers enormous opportunity for those who get it right, and
public humiliation for those who don’t. (Campaign Middle East,
27.4.2012)
Again, this is an evaluative piece, offering an assessment of the current

marketing environment in the form of unmitigated declaratives, with no
modal terms. This phenomenon might be called “opinion-as-fact”.
These examples of opinion pieces exhibit extreme variation in both
channel and tone, but all share the generic aim of expressing opinion and
evaluating phenomena. To achieve this, writers use not only modal expres-
sions, but also lexis of affect and judgement and rhetorical strategies.
Genre 5: Arts Reviews
The arts review genre is virtually absent in 2010 and 2012, but is sig-
nificant in 2012 in texts mentioning BP. In this year, the BP story is
referenced in a part-fictionalised documentary film, and two non-fiction
accounts of oil companies as well as an unrelated film about events whose
PR handling is compared with that of the BP oil spill (BP events as exem-
plar of a “disastrous” PR campaign). In addition, a different item in the
news report genre covers a protest at the Royal Shakespeare Company
and refers to a song written about the BP oil spill events: an excerpt from
this article is cited at the start of this book. In these examples, the arts
review genre exhibits many of the characteristics of evaluative writing: the
writer is strongly present in the texts, and the lexis shows high levels of
affect, judgement and appreciation (Martin & White, 2005).
Based on Margaret Atwood’s book of the same name, writer-director

Jennifer Baichwal’s film explores the complex issue of debt, both moral and
financial. This includes BP’s failure to deal with its environmental transgres-
sions, and the years-long dispute between two poor rural clans that keeps
the members of one family virtual prisoners in their own home. (The
New York Post, 27.4.2012, my emphasis)
In this book’s more than 600 pages you may sometimes be tempted to
utter, as did BP’s hapless chief executive Tony Hayward, disastrously, during
the Deepwater Horizon disaster, “I’d like my life back.” (The New York
Times, 27.4.2012, my emphasis)
In both of these fragments the journalists are exhibiting judgement not

only about the works they are reviewing (in Martin & White’s (2005)
term, “Appreciation”) but about the crisis events themselves (what Martin
& White would classify as “Judgement: social sanction”). Reviewers
choose whether to endorse or distance themselves from the viewpoints
expressed in the works they review, which may or may not be in line
with the political perspective of their publication. So both dominant and
resistant views are given voice in review pages, but clearly positioned as
non-factual and non-news by being placed outside the news pages. This
is also the case with letters pages, where readers’ views are positioned as
“other” (Cook, Robbins, & Pieri, 2006). I mentioned earlier that the
choice of news media genre in itself has semiotic meaning. Publications
as a whole might offer a platform for a range of voices and views about the
crisis, but placing some within the front “hard” news pages and others in
review and letters pages is a way of bracketing, or making “other” certain
voices and the opinions they express.
Genre 6: Business or Market Report
Academic literature on the genre of the business/market report is rare.

However, a number of the BP texts are of this type, and I set out here
some observations based on the texts themselves. Business or market
reports can be of rather different kinds, ranging from those texts that
are similar to news reports but aimed specifically at the business com-
munity and characterised by markers of objectivity, to texts that serve to
synthesise information and arrive at an evaluative conclusion, to other

texts that serve a more argumentative or persuasive purpose (e.g. busi-
ness journal articles that may partially serve a selling function). Because
of this, the genre of business or market reports may be better viewed as
a collection of texts with a unified audience, but which have stylistic
features of other news genres. The complete text below resembles a news
report, although it appears specifically on the business pages:
MIAMI—Carnival Corp., operator of Carnival Cruise Lines, is seeking

compensation for damages and losses it incurred as a result of last April’s
Deepwater Horizon oil rig explosion that caused a major oil spill in the
Gulf of Mexico, according to The Miami Herald. The Miami-based cruise
line filed suit against BP Plc. and several other companies related to the oil
rig’s operations. (Palm Beach Post [Florida], 27.4.2011)
Although the term “major oil spill” is somewhat negatively evaluative, the
lexis is generally neutral, and some of the terms (“compensation”, “dam-
ages and losses”, “filed suit”) are drawn from a legal technical register.
There are no modal items, and statements are expressed as unmitigated
declaratives. These are all features that, as in news reports, convey an
impression of factual objectivity. The following text is rather different in
terms of grammatical choices, and belongs to the second type of business
report I mentioned above, of synthesis and evaluation.
The situation may be further complicated by the approaching election in

Russia, since a huge payment to the AAR billionaires for what used to be
state assets is not something the incumbent administration will want to
defend as it works to win support and secure another term in office. A more
modest settlement would suit BP, too, given its commitments in the wake
of the Gulf of Mexico disaster. (European Gas Markets, 27.4.2011)
This text is rich in modal expressions (“may be”, “will want to”, “would
suit”), where, importantly, the writer is taking modal responsibility. The
lexis in this text is more evaluative (“complicated”, “huge”, “disaster”), as
the writer interprets, evaluates and directs the reader. The piece resembles
an opinion piece. The final example below is of the third type, which seeks
ostensibly to inform, but also contains elements of persuasive language.
Whatever the reason, responsible companies and thoughtful boards must

realize that no business, no matter how well-managed or low-profile, is
immune from a crisis. When it comes, it might not be as damaging and
public as BP’s or Toyota’s, but that does not mean it can’t cause damage.
(Executive Counsel, 27.4.2011)
A number of features, such as the deontic expression “must realise” and

lexis with high affect such as “damaging and public”, indicate an overt
involvement by the writer that tends to be avoided in hard news reports.
Familiar or conversational language is shown in the contraction “can’t”,
and rhetorical (persuasive) constructions appear in the repetition in “no
business, no matter how well-managed or low-profile”, and the argument
structure of “it might not be…but that does not mean”. This text has the
indirect purpose of selling consultancy services.
So, while some of the business or market reports do the work of a
news report in updating the business community on the latest events in
the BP story from a business perspective, the majority are doing the work
of evaluative or opinion pieces that serve to place the events into other
business contexts, such as the BP–Rosneft share swap deal in the second
example above, and the BP events as signifier of a business crisis in the
third example.
In Summary
I have discussed generic features at some length, proposing that shifts in

genres can explain patterns in language features, from those that con-
struct a factual and objective picture of the BP story to those that relate
to rhetorical acts of evaluation and persuasion. An analysis of the genres
of the BP texts suggests the broad progression shown in Fig. 9.1.
writing which
Deepwater writing which writing which
evaluates other
Horizon describes evaluates
texts about
events events events
events
Fig. 9.1 Notional progression of text types in BP data

Figure 9.1 positions the Deepwater Horizon events as the object of

representation (and I will go on to discuss a more explicit understanding
of the term “Object” in Chap. 13). The “writing which describes events”
is news reports and financial reports (largely from 2010). The evaluation
of events takes place in feature articles, editorials and blogs within the
sample texts, but importantly also outside of the texts, in official reports
and artworks connected with the topic. These are the subject of the final
genre shift to “writing which evaluates other texts about events”. Here
the topic is less the events themselves than what has been written about
them. In this way, the coverage becomes increasingly distanced from the
events, not only temporally, but in terms of the closeness of the author to
the object under discussion. This observation leads us to the importance
of other texts in the BP story, and I turn to this in the following section.
Feature 5: Intertextuality
In my discussion of intertextuality, I described different frameworks for

considering types of intertextuality (e.g. Bazerman, 2004; Fairclough,
1992b; Genette, 1997), and based on this work, I suggested that there
were three broad types of intertextuality:
• Intertextuality as irretrievable texts.

• Intertextuality as identifiable style/register/genre.
• Intertextuality as retrievable texts.
In my BP text analysis, I did not address this first category above, although
I presumed a considerable weight of this kind of intertextual influence.
The second category above concerns structural and stylistic characteristics
of genres, and these have been outlined in my analysis of genre in the
previous section. In this section, therefore, I will discuss only the third
category—that is, references within the analysed texts to other texts that
have a more or less transparent provenance. In the BP data, the most
prominent of these other texts were:
1. Direct and indirect quotation, for example, eyewitness reports.

2. Press releases.
3. Other sources—news reports, news agencies and other reports.
4. Artistic and literary references.
5. Co-text.
An analysis of how and why such texts are incorporated in news media
reports contributed to an understanding of how news writers use avail-
able linguistic resources as support in constructing a particular view of
the BP crisis.
Intertextuality Type 1: Direct and Indirect Quotation
Both direct and indirect quotation of speech are present extensively in

the 2010 texts, primarily from senior BP staff, technical experts and
involved members of the public (e.g. the cook’s interview narrative
mentioned earlier). Quoted speech is also a feature of press releases.
The purposes of direct and indirect quotation in news reports are
varied, including to add credibility, to personalise a story, to invite
reader identification with the quoted individual and to distance the
writer from the propositions made in the quotation. In most cases in
the 2010 texts, quotation is used to present information from BP as
a credible source, and to gain reactions to the oil spill from experts
other than BP, such as the Coast Guard, environmental and ocean
studies experts and local fishermen. The following excerpt presents
the BP perspective:
Paragraph 1 BP plans to collect leaking oil on the ocean bottom by

lowering a large dome to capture the oil and then pump-
ing it through pipes and hoses into a vessel on the surface,
said Doug Suttles, chief operating officer of BP Exploration
and Production.
Paragraph 2 It could take up to a month to get the equipment in place.
Paragraph 3 “That system has been deployed in shallower water, but it

has never been deployed at 5000 feet of water, so we have
to be careful,” he said. (BreakingNews.ie, 27.4.2010)
This fragment shows evidence of a range of reporting strategies. The first

paragraph is shown in Free Direct Speech (Leech & Short, 1981)—that
is, no quotation marks, but a clear reporting clause (“said Doug Suttles”).
The use of Doug Suttles’ full name and title (see earlier comments on
naming actors in the events) connotes that this is someone with a war-
rant to speak, as both a BP representative and a technical expert. The
third paragraph (“That system has been deployed…”) is an example of
Direct Speech with full use of quotation marks, reporting clause and spe-
cific attribution. However, the second paragraph “It could take up to a
month…” is neither in quotation marks nor attributed. The reader can
make the assumption that this is still a version of Doug Suttles’ words,
particularly as the concluding “he said” refers back across paragraph two
to Suttles’ full name in paragraph one, but there is no other linguistic
evidence of quotation. The fragment “It could take up to a month…”
takes on the colour and authority of Suttles’ pronouncements through its
placement (and may indeed be his words). However, it allows for econ-
omy (no need for reporting clauses or additional attributions if relevant),
stylistic variation and a sense of journalist as expert.
This mix of representation strategies is common within news report-
ing, and it is sometimes difficult to unravel whether markers of (un)
certainty can be attributed to the writer or the quoted source. In news
report writing, journalists seek ways of varying the presentation between
their own consolidated but unattributed understanding from a range of
sources, direct quotation from experts and eyewitnesses and something
in between, as above.
By 2011, quotations are fewer in number. These are more likely to appear
in connection with BP financial results than in news report items: journal-
ists are no longer using quotations as a channel to express uncertainty in
relation to the Deepwater Horizon events. I have noted that direct and
indirect quotation is one strategy by which journalists distance themselves
from reported propositions. This process is observable in the 2011 texts,
where certain voices are quoted directly (e.g. Michael Greenberger in an
interview and the views of the Florida Attorney General in a letters page)
but such quotations are considerably rarer than in the 2010 data.
Direct quotation is certainly present in the 2012 texts, but used in a
different way from the 2010 texts. In 2010, it was a way of representing
the crisis as an immediate, recently witnessed occurrence, of inviting the
reader to feel somehow present at an unfolding event. By 2012, direct
and indirect quotations fulfil a number of different functions.
• Referencing the event, in order to place it in a context, as here, in a

business article about crises that quotes the originator of a spoof BP
Twitter account: “The best way to get the public to respect your brand?
Have a respectable brand.” (Campaign Middle East, 27.4.2012)
• Direct quotation from members of the public, but indirectly related to
the oil spill (here in the context of a feature about an oil town in
Dakota.) “After the spill in the Gulf, it was really getting hard to know
if you have a job or not.” (Canwest News Service, 27.4.2012)
• Reports of political speeches, dealing with energy matters
– “Hood says as many as 200,000 individuals and businesses who
signed the deals should qualify to receive payments for future dam-
ages and any increased damage payments.” [indirect quotation]
(Greenwire, 27.4.2012)
– “He [Secretary of the Interior Ken Salazar] blasted the Republican-
controlled House of Representatives for not acting to codify
offshore-drilling regulations adopted since the 2010 BP Deepwater
Horizon oil spill, failing to quickly approve a US–Mexico agree-
ment on offshore development, not making tax credits of renewable
energy permanent, and not adopting new clean energy standards.”
[indirect quotation] (Foster Natural Gas/Oil Report, 27.4.2012)
• Literary quotations
One reference of interest is to what was possibly the best-known quota-

tion of the Deepwater Horizon crisis, that of Tony Hayward in May 2010
mentioned earlier:
In this book’s more than 600 pages you may sometimes be tempted to utter,
as did BP’s hapless chief executive Tony Hayward, disastrously, during
the Deepwater Horizon disaster, “I’d like my life back.” (The New York
Times, 27.4.2012)
The quotation was clearly suggested to the writer by his topic matter, in
this case a book on oil companies, primarily Exxon Mobil and BP. But
it also shows how recognisable texts from the events (Tony Hayward’s
words) are appropriated into the public domain for general-purpose use,
in this case to make a humorous observation.
Intertextuality Type 2: Press Releases
Press releases are one type of intertext that is more or less retrievable from
the items in which they appear, and they are interesting for my analysis
because they are one means by which participants other than journalists
play a part in shaping the media representation of events. Press releases
are written with the primary aim of being repeated verbatim in news
reports (Jacobs, 1999, 2000a). They are intended by their originators to
present a version of the “truth” that is easily accessible and repeatable, and
so eventually becomes the version of the truth. Jacobs demonstrates how
frequently sections of press releases are inserted in their entirety into a
news report, and how the writers of press releases encourage this by their
use of the third person (“BP” rather than “we”), their formulaic struc-
tures, their submission of direct quotations and their deliberate mirroring
of newspaper house styles. In this way, the originator’s perspective on the
story is far more likely to find its way into the media.
The direct use of press releases as copy is demonstrated in the case of
inadvertent error:
Chief executive Tony Hayward said better weather in the area around the
rig disaster that is believed to have killed 11 men had increased confidence
we can tackle this spill offshore. (The Evening Standard (London),
The failure to make the changes from direct to indirect quotation (one of
which is the change from first to third person) is unlikely to have arisen
from the reporting of a face-to-face interview with Tony Hayward, and

more likely to have been taken from the following press release:
“Given the current conditions and the massive size of our response, we are
confident in our ability to tackle this spill offshore,” Hayward added. (BP,
2010b, 24 April)
Press releases are also an efficient way of communicating complex infor-

mation, such as technical information or numerical data. The following
press release is dated two days after the explosion (the list continues for a
further four items):
BP has mobilised a flotilla of vessels and resources that includes:
• significant mechanical recovery capacity;

• 32 spill response vessels including a large storage barge;
• skimming capacity of more than 171,000 barrels per day, with more
available if needed;
• offshore storage capacity of 122,000 barrels and additional 175,000
barrels available and on standby (BP, 2010a, 22 April).
The communication of facts and figures, from press releases, ostensibly an

objective exercise, is however, subject to ideological influence. For exam-
ple, many of the 2010 BP reports make mention of a particular proposi-
tion from a press release, that, as at 27 April, oil may reach land in three
days. This fact is positioned by reporters in various ways.
• “as little as three days” (Trend Daily News (Azerbaijan), 27.4.2010)

• “expected to reach land by Saturday” (theflyonthewall.com, 27.4.2011)
• “within days” (24/7 Wall St, 27.4.2010)
• “It’s still not expected to reach the coast before Friday, if at all”
(Carleton Place (Canada), 27.4.2010)
• “may reach land in just days” (CNN, 27.4.2010)
These reformulations of the same basic information reveal a position-

ing that ranges from the (relatively) neutral “expected to reach land by
Saturday” to the more urgent “as little as three days”, “just days” and
“within days”. An alternative, and more reassuring, version is presented
in “not expected…before Friday, if at all”. The choice of contrasting posi-
tive and negative formulations in “expected” (second example in the list)
and “not expected” (fourth on list) again indicate differing modal posi-
tioning of identical original information.
The part press releases play is considerably reduced in 2011, with the
number of press releases published by BP in April 2011 being much
smaller than that in April 2010. Whereas BP issued one release per day
from the explosion on 20 April to 27 April 2010, only six releases were
issued between 1st and 27th April in 2011. Only three of the six are
related to the BP events, and these indirectly. The first concerns sales of
assets, and this issue is referred to in several of the financial report texts
of 2011 in connection with the need to fund compensation for the oil
spill. The second refers to environmental projects along the Gulf Coast.
The third concerns the Gulf of Mexico Research Institute (GRI), and is
reported in great detail in one of the 2011 texts, with much of the original
text from the press release unchanged. (This text has already been noted as
of interest for its use of categorisation, because it positions the Deepwater
Horizon events (unnamed) as only one of many similar events, and as
backgrounded to the global benefits of the work of the GRI.)
The number of press releases made by BP in the month to 27 April
2012 is again six. Three are unrelated to the Deepwater Horizon crisis
(one of these concerns BP’s involvement in the 2012 London Olympic
Games). The other three are connected to ongoing legal issues, and these
are reflected in references to legal texts in several of the 2012 items. From
a BP perspective, press releases are concerned partly to continue to con-
trol the accuracy of the Deepwater Horizon story, and partly to show
how the company is moving on away from Deepwater Horizon through
undertakings such as the Olympic sponsorship.
Intertextuality Type 3: Other Sources—News Reports, News

Agencies and Other Reports
In 2010, apart from the press releases and external sources mentioned
above, there is evidence of use by writers of other sources close to the
news-making process, including historical news stories, reports from

news agencies, websites, official reports and so on. Some of these sources
are explicitly acknowledged, for example:
…Hayward said in the email, obtained by AFP. (Agence France Presse,

27.4.2010)
WSJ [Wall Street Journal]: The BP oil spill could reach land within days.
(2010, Text 1/20)
The first extract credits both Tony Hayward as the writer of the email
and the news agency (who wrote the piece) for obtaining it. The second
item abstracts information in very short form, directing the reader to
the source for further detail. However, in other news items, portions are
very likely to have been grounded in previous newspaper reports or other
sources, but these remain unacknowledged.
The Nexis UK texts are rich in what Chandler (2007) calls “hypertex-
tuality”, a category of intertextuality he proposes to take account of the
ability of the Web to refer to other texts through hyperlinks. In the case
of the Nexis texts, it is possible to gain additional information on compa-
nies with an attached hyperlink from, for example, the New York Stock
Exchange. The website of the SEC is referred to in “other filings with the
Securities and Exchange Commission (SEC), which are available free of
charge on the SEC’s website at www.sec.gov” (Market News Publishing,
27.4.2010). Other documents are directly or indirectly referred to: there is
a presumption of a previous text in the phrase “in a recent strategy update”
(Kuwait News Agency [KUNA], 27.4.2010), and “according to data from
the Minerals Management Service” (TendersInfo, 27.4.2010). Some of the
data items refer to visual texts, such as aerial photographs by NASA (the
North American Space Agency) of the oil slick. Transcripts of TV reports
make reference to accompanying film footage. Finally there is a reference
to an absent text: “No word on when British Petroleum is expected to cap
the leaking well” (The Richmond Democrat, 27.4.2010). Poststructuralist
analysis proposes that absences as well as presences are significant (see, for
example, Sunderland [2006], Wetherell [1998], Oliveira [2004]). Here
the absence of the text is flagged; the key piece of information awaited by
journalists is unforthcoming. Reference by the writers of the 2010 texts to
a rich background of source texts deliberately hints at a complex reality,

from which the writer has pieced together a credible narrative on behalf of
the reader, using his/her journalistic expertise.
In 2011 a shift begins from news items that report on the BP events
to news items that report on the documentation of the events. Various
documents related to the events are mentioned and commented upon in
the 2011 texts:
• Documents related to legal processes: “legal red tape”; “lawsuits”;

“filing”
• Political agreements: “moratorium on deepwater drilling in the Gulf ”
• Financial reports and statements: “a new report from Oppenheimer &
Co”; “reported Bloomberg”; “Income statements”; “conference call to
analysts”; “BP plc reported Wednesday that profits rose”
• Reports on the explosion and its aftermath: “Coast Guard report”,
“Research Initiatives report”
So from 2011 onwards, there is an increasing focus on the ways in

which the BP events are reported, analysed and codified into docu-
ments or other written texts. These other texts include those related
to the legal process to determine responsibility and compensation, the
drilling moratorium and official reports on the explosion and oil spill,
amongst others. In short, the BP texts are dealing less with events and
increasingly with other texts. Even more than interviews and eyewit-
ness reports, this kind of written documentation serves to reify certain
versions of events, clarifying areas of confusion or disagreement, and
excluding alternative versions. The text examples above are supple-
mented, particularly in 2012, by a number of intertexts which can be
described as artistic and literary.
Intertextuality Type 4: Artistic and Literary Texts
The 2011 texts contain, for the first time, reference to a number of con-
tributing texts from the worlds of art and literature. The non-fiction book
“In too deep”, by a journalist with a long relationship with BP, has been
mentioned above (M2 PressWIRE, 27.4.2011). Arts texts also appear in
several guises in an article which is only indirectly concerned with the BP

events (The News Herald [Panama City, Florida], 27.4.2011). This report
deals with a meeting of the local tourist council in Panama City Beach,
Florida, discussing their Autumn marketing plans. These plans (partially
funded by a grant from BP) comprise such current and proposed texts
including TV and print advertising, and a potential booking of the Zac
Brown band and their song “Knee Deep”. References such as these are
an indication that the BP events are being placed within social contexts
that are increasingly less related to the events themselves, although they
are still connected to the outcomes of the events. In 2011 these social
contexts are beginning to include references to the arts.
The number of texts with a direct artistic connection increases consid-
erably in 2012, and in these the level of intertextuality is high. One text
(The New York Post, 27.4.2012) is a review of a documentary that is itself
based on a book. Another (Coventry Evening Telegraph, 27.4.2012)
begins with a headline based on a well-worn pun “BP or not BP? That
is the question” and goes on to report the actions of a campaign group
objecting to BP’s sponsorship of the Royal Shakespeare Company’s World
Shakespeare Festival. Their protest takes the form of a song, performed
(as an “unexpected musical prologue”) on the stage at the start of the
opening night of the play “Twelfth Night”. I mentioned earlier that, by
striking coincidence, the phrase “green and yellow melancholy” from the
play (Act II, Scene iv) contains reference to the colours of the BP logo,
and is repurposed in the protest song.
Even in texts that do not themselves deal directly with artistic works,
literary reference is present.
“The first thing we do, let’s kill all the lawyers.” William Shakespeare,
Henry the Sixth [Blog dealing with a range of, in the writer’s view, political
shortcomings] (Phil’s Stock World, 27.4.2012)
Truman Capote once wrote of the “gilded baritone” of its hourly bells,
drifting over Jackson Square and joining the “solitary grieving howl of a
far-off shiphorn”. [travel feature on New Orleans] (Sarasota Herald Tribune
[Florida], 27.4.2012)
Alongside the shift in genres noted earlier, we can conceptualise a similar

movement through the years where texts act upon other texts. Figure 9.2
Book on
[BP company ‘ten years Publication
Deepwater
practices] reporting on the announcement
Horizon and
company’ about book
suggested causes
Fig. 9.2 Texts embedded intertextually “In Too Deep: BP and the Drilling
Race that Took it Down” (M2 PressWIRE, 27.4.2011)
gives the example of a 2011 text that announces a book about Deepwater
Horizon (M2 PressWIRE, 27.4.2011). The example shows how a range
of texts interconnect in news reporting, and how news items can become
progressively distant from the original events on which they are based.
We can see how one type of text, “news reporting”, is nested into
another, “non-fiction writing”, which is nested into another, “publication
announcement”. These densely intertextual items set the BP events in
sometimes unexpected contexts, providing new perspectives from which
to understand their meaning.
Intertextuality Type 5: Co-text
The co-text surrounding the BP story is of two kinds. One is the sur-
rounding text in a print or online newspaper. Newspapers are what Hoey
(2001) calls a “colony” of texts, where otherwise unrelated stories and
genres are loosely linked by placement, headings and publication styles.
In the colony, each of the items provides a context for the others that will
affect to a certain extent the reading of any given piece. More important to
my own investigation is the co-text within a particular story. As noted in
my Stage 1 analysis, in 2010 the BP story accounts for virtually all of the
news items in the majority of the cases. In other words, there is no co-text
alongside the BP story in most of the 2010 texts. The exception is financial
reports, where BP is one of the companies reported upon, and the co-text
comprises financial information about other companies. In these cases,
the presence of the co-text gives a different impression from that given by
news reports that position the BP oil spill as a serious and unique news
story. In financial reports, placing the BP oil spill in the context of Quarter
1 results, and in the further context of general financial news, works to
minimise its significance, suggesting that it is a business crisis disrupting

the “normal” flow of continuous production and profit progress.
The proportion of each text relating specifically to the BP events
becomes smaller in 2011, which entails an increase in associated co-text.
To return to two examples discussed above, the “In too deep” text (M2
PressWIRE, 27.4.2011) mentions the history of BP, other serious acci-
dents and the management style of its CEOs, in addition to the events of
2010. The “Panama City Beach” text (The News Herald [Panama City,
Florida], 27.4.2011) discusses marketing plans in general, the composi-
tion of the committee and the reaction of local residents, as well as the BP
contribution. The amount and nature of co-text in 2011 indicates both
the decreasing importance of the BP story in terms of newsworthiness
and the absorption of the story into a bigger social picture. The co-text
positions the events within a wider social context: for “In too deep” as
an example of the unwelcome effects of particular management cultures
and for “Panama City Beach” in the context of the effect of the oil spill
on the daily lives of the residents and local businesspeople in a particular
community.
This continues to be the case in 2012, where co-text continues to serve
the dual functions of constraining the size and significance of the BP
events, and positioning them in diverse contexts in which they are pre-
sented as meaningful.
In Summary
Given the complexity of the Deepwater Horizon story in 2010, and

the number of journalists working on it, there is perhaps a surprising
degree of homogeneity in the intertexts referred to in the news reports.
Primary sources are press releases, interviews and press conferences. Press
releases provide a source of factual information (as offered by BP), and
some “direct quotations”, almost exclusively attributed to Tony Hayward.
Interviews and press conferences also generate direct quotations, but
these same quotations are repeated frequently from half a dozen actors in
the story, chief amongst them Rear Admiral Mary Landry of the Coast
Guard. In addition, a number of other texts are implicitly or explicitly
drawn on, such as photographs, websites and TV footage. The impres-

sion is of a set of texts teeming with a cast of informed voices, but closer
analysis shows these to be relatively restricted and repeated. They are,
with few exceptions (one being the voice of the cook Oleander Benton),
what Bignell (2002: 88) calls “accessed voices”, namely, those with regu-
lar and unquestioned access to the news media.
By 2011, press releases and witness voices are drawn upon to a much
smaller extent, but documented accounts of the events and the ensuing
legal processes are far more evident. These are texts reporting upon other
written accounts rather than events themselves. So along with the tempo-
ral distance from events comes a distancing through the use of intertexts.
Not only does this practice tend to mitigate the intensity of accounts, but
it also functions to create shared accounts that draw on each other, rather
than presenting multiple individual perspectives.
The final year of data draws from a number of intertext types, includ-
ing for the first time in this data, direct literary quotations. These are
used to evoke particular social worlds, to which the reader needs to bring
a degree of shared cultural understanding. This is particularly the case
where well-known quotations are modified (“BP or not BP?”) or used
out of context to make a humorous point (“let’s kill all the lawyers”).
This kind of reference does not use the words of those with a warrant to
speak about the topic, but rather draws on shared experience and cultural
knowledge to present events as understood, as having a degree of collec-
tive meaning.
Feature 6: Modality
In an overview of the modality system in Chap. 3, I noted three key types

of modality:
1. epistemic modality, expressing degrees of certainty.

2. dynamic modality, expressing ability and willingness.
3. deontic modality, expressing permission and obligation.
A fourth type, evidential modality, which includes sensory and reporting

items, is found in research and in my texts to be very rare and for this
reason not included. I further noted that modal resources are very broad,
comprising not only modal auxiliaries, but adverbials, adjectives, nouns,
phrases and clauses. I described how modal expressions can vary in
strength, with speakers taking strong, weak or medium positions in rela-
tion to their propositions. Investigating these different types of meaning
potential for diverse modal resources will contribute to an understanding
of how writers position themselves in relation to their construction of the
crisis. I also discussed the Appraisal System, which represents an alterna-
tive, but related, method for analysing the interpersonal function of texts.
Its relevance for my work is that it provides a means for the systematic
description of evaluation, particularly in the case of lexical usages.
To analyse modality I started with a count of instances of modal expres-
sions. In line with the categories above, this comprised a frequency count
of all kinds of modal usages, according to whether they had epistemic,
deontic or dynamic force. I then considered these usages in context, for
example, whether an expression of uncertainty represented the perspec-
tive of the writer, or a participant in the story. I further considered the
strength of modal commitment and the relative subjectivity of the writer.
Comparison of modal choices across time indicated the extent to which
this positioning alters with increasing familiarity with the material.
Table 9.1 shows the findings for modal usages over the three years of data.
Table 9.1 Instances of modality in the 2010–12 texts

2010 2011 2012
Numbers % Numbers % Numbers %
Epistemic modality 141 56 52 50 43 51
Dynamic modality 97 38 36 34 17 20
Deontic modality 14 6 17 16 25 29
Totals 252 100 105 100 85 100
Total per 000 words 27.6 15.0 11.9
In 2010, epistemic and dynamic modality are both strongly present in

the texts, with epistemic modality occurring more frequently than dynamic
modality. The expression of deontic modality (indicating permission and
obligation) is extremely low at only 6 % of all modality expressions. These
findings are perhaps not surprising. Epistemic modality is the system through
which certainty, uncertainty, likelihood and probability is conveyed, while
dynamic modality concerns ability as well as willingness. In a set of texts that
deal with an event of almost unprecedented magnitude and the attempts
of participants to deal with it, expressions of uncertainty and (in)ability
are likely to feature strongly. The examples below contain linguistic items
showing epistemic modality:
The BP oil spill could reach land within days. (24/7 Wall St, 27.4.2010, my
emphasis)
He also expressed “tremendous sorrow” when it became clear that the
missing workers were probably dead. (Agence France Presse, 27.4.2010, my
emphasis)
Such a well could help redirect the oil, though it could also take weeks to
complete, especially at that depth. (Associated Press Financial Wire,
The 2010 texts are deeply concerned with the ability of participants to
manage the crisis situation, and with their intention and determination
to do so. These are expressed through dynamic modality.
“We are going to do everything we can” (Agence France Presse, 27.4.2010,

my emphasis)
Crews were trying to keep oil out of the Pass A Loutre wildlife area, a
115,000-acre preserve that is home to alligators, birds and fish near the
mouth of the Mississippi River. (BreakingNews.ie, 27.4.2010, my
emphasis)
Efforts so far have failed to shut off the flow of oil nearly 5000 feet
(Carleton Place (Canada), 27.4.2010, my emphasis)
(The expression “going to” in the first example above could be analysed
either as a simple future tense or, as I have, as a modal expression convey-
ing something like “we intend to”.) Words such as trying, failing, hoping,
intending, wanting, being committed to—and 20 other verbs expressing
dynamic modality—are employed in the 2010 texts, and this conveys a

picture from both the writer’s and his/her reported speakers’ perspective
of effort and intent rather than achievement.
It is worth exploring here the potential tension between the uncertainty
of the situation and the (in)ability of the actors to have an effect on it,
and the portrayal of objectivity which is one key characteristic of the
news report genre. White (1997: 5) observes that in the typical hard news
report, the reporter “avoids or at least minimises interpersonal meanings”.
Yet the writers of the 2010 texts make significant use of modal resources,
which would appear to run counter to the canonical expectation for news
reports of low modality. This raises two important questions:
1. Are the BP reports unusual compared to other news reports for some
reason?
2. Whose (un)certainties, and (in)abilities are being expressed?
To take the first question—the BP events in themselves may have been

unusual, but they are by no means unique in terms of crises. Both natu-
ral and man-made crises are characterised by a degree of confusion or
uncertainty in the early stages. In fact, it is disruption of the social order
in some way that makes an event considered newsworthy. The BP events
of 2010 are extreme in certain respects, but I would suggest represent
no more uncertainty than, for example, the Tsunami in South-East Asia
of 2004 or the disaster at the Fukushima nuclear plant in 2011. So the
degree of confusion about the events is not atypical, and other aspects,
such as cause, responsibility and the unpredictable movements of the spill
were the subject of intense speculation from the start.
What is a more likely explanation for the high levels of modality in the
2010 media coverage is that even where sources are not overtly indicated,
the expressions of uncertainty and inability are not the voice of the jour-
nalist himself/herself, but of his/her sources. The “objective” picture here
is one of (un)certainty and (in)ability, and it is not an intrusion of the
writer to depict this. The following short fragment shows an example (St.
Petersburg Times (Florida), 27.4.2010, my emphasis).
Paragraph 1 The marshes of southern Louisiana and Mississippi appear

to face the most immediate risk from the spill because they
are closest to it, said George Crozier, director of the
Dauphin Island Sea Laboratory in Mobile, Ala.
Paragraph 2 What happens after that depends on how quickly the
owners of the rig can shut off the flow of oil. On Sunday
they began using robot submarines to try to shut off a valve
called a blowout preventer on a leaking pipe deep under-
water. If that fails, then they will drill new wells on either
side of the leak to relieve the pressure there—a process that
could take months.
Paragraph 3 “If it goes on for four months, then yeah, we’ve got a prob-
lem,” Crozier said. “But if they’re able to shut it down after
a day or two, then the risk is minimal.”
Here attribution is shown in two ways. The first paragraph consists of

free direct speech by George Crozier. Despite the lack of quotation marks
within the piece, the reporting clause (“said George Crozier”) makes it
clear that these are either the actual words or a close paraphrase of Mr.
Crozier’s interview with the journalist. Paragraph 3 is even clearer. We are
intended to understand that these are the exact words of the interviewee.
In fact, the interpolation of the discourse marker “yeah” in the middle of
the words makes their proposed accuracy even more convincing—we feel
as readers we are hearing unmediated speech. The use of modality within
these attributed portions of the texts points to the views and perspectives
of participants other than the journalists.
Paragraph 2 is slightly different. The use of epistemic modality in
“could” and dynamic modality in “can” and “try to” appear to be the
interpolation, the “intrusion” in Halliday’s term, of the writer himself (in
this case the writer is male). These expressions, and the use of conditional
clauses, suggest a tentative view towards the propositions made in the
paragraph. But my interpretation of this is that we are to understand this
paragraph too as a report of information, itself tentative, gained from
external sources, rather than the opinion of the writer about the feasibility
or otherwise of the actions undertaken.
The same motivation seems to be at work in the few cases of deontic

modality in 2010. These again tend to occur where there is an implicit
(first example below) or explicit (second example below) reporting of
another’s opinion, rather than the stance of the journalist:
If crews cannot stop the leak quickly, they might need to drill another well
to redirect the oil. (Associated Press Financial Wire, 27.4.2010, my
emphasis)
“That system has been deployed in shallower water, but it has never been
deployed at 5000 feet of water, so we have to be careful,” he said.
(BreakingNews.ie, 27.4.2010, my emphasis)
The obligation and permission relates to external circumstances, deriving

from the crisis itself—a set of uncontrollable forces obliging a course of
human action. They are not the interpolation of the journalist’s perspec-
tive on the propositions he sets out. My interpretation of these findings
about the level of modality in my 2010 texts is that where news reports
include a high level of directly or indirectly reported speech, the occurrence
of modality will be high overall. And in the case of crisis representation,
where (un)certainty and future intention are critical to a representation of
the story, the level of modality is likely to be higher, even in the authorial
voice, than it might be for other types of news report.
The overview of the frequency of modal markers appearing in the
2011 texts indicates a marked decrease, from 27.6 instances per 000
words in 2010, to 15.0 per 000 words in 2011. By 2011, there was a
greater degree of confidence by both journalists and other participants
in discussing causes and outcomes, together with a lower occurrence of
reported speech. In my earlier analysis of genre, I drew attention to the
presentation of opinion and evaluation in language that was markedly
low in modal expressions—that is, opinion presented as “fact”—and
noted that this is sometimes accompanied by highly evaluative lexis. I
suggest that these three processes together—increased confidence, reduc-
tion in reported speech and “opinion as fact”—account for the decrease
in modal items in the data.
Alongside this marked drop in frequency, there is a shift in the types
of modal item. Both epistemic and dynamic modality represent a lower
proportion of instances of modality than was the case in 2010, while

deontic modality appears more frequently. By 2011 there is a reduced
need to express levels of (un)certainty and (in)ability (through epistemic
and dynamic modality) in relation to the events themselves. The oil spill
itself has long since been contained, and media attention has turned to
investigation, compensation, subsequent business performance and evalu-
ation. The instances of epistemic modality halve from 2010 to 2011 (nor-
malised for the size of corpus). Where epistemic modality was used in 2010
to convey uncertainty about the outcome of the ongoing events, by 2011 it
is more likely to indicate uncertainty about BP’s business future, for exam-
ple, the scale of compensation and Deepwater Horizon-related costs:
But the court cases are likely to take years and BP could face tens of billions
of dollars more in fines and penalties if it is prosecuted. (The Associated
Press, 27.4.2011, my emphasis)
However, the majority of uses of epistemic modality do not deal directly

with the BP events at all, but rather with issues arising from them:
Meanwhile, deepwater costs are expected to rise significantly worldwide as

governments enhance safety requirements after last year’s Macondo disaster
in the US Gulf (EI Finance, 27.4.2011, my emphasis)
Committee chairman and TDC member Marty McDaniel said a fall
roster of events, which could include a Zac Brown Band concert if a deal
can be reached, would be in place by Memorial Day. (The News Herald
(Panama City, Florida), 27.4.2011, my emphasis)
The cases above demonstrate again the shift of the BP events from being
the central focus of events to having a less direct role in news stories.
Changes can also be seen in the function and use of dynamic modal-
ity in the 2011 texts. This kind of modality indicates aims, targets and
intent. In 2010 these related to dealing with the spill and its effects, but
in 2011 they are related to moving forward away from the crisis events
and towards a more optimistic and more stable future:
“BP is in the midst of major change as we work to reset focus for the com-
pany and begin the task of rebuilding long-term sustainable value for our
shareholders,” Chief Financial Officer Byron Grote said on a conference
call to analysts. (The Associated Press, 27.4.2011, my emphasis)
The implication of the example above is of optimism and a positive

movement.
Unlike the first two modal types, the role for deontic modality, which is the
system for expressing obligation and permission, increased in 2011, both in
number and proportion of occurrences, although it remains the least used of
the three types. Deontic modality is being used by writers to express views
about what BP or other similar companies should do in terms of handling
the crisis, or how readers should understand the argument in the news item:
To find out when the country will get relief from these high gas prices, it is
first necessary to understand why they are rising in the first place.
(CaptainKudzu, 27.4.2011, my emphasis)
capital outlays, like F&D costs, must be evaluated over longer spans than
one year to account for “lumpy” investments in major projects. (EI Finance,
The April 20th blowout of 2010 was only the latest of a series of BP
accidents that should have served as warning signs to company executives
and regulators. (M2 PressWIRE, 27.4.2011, my emphasis)
The movement towards statements of obligation and necessity can be

related to the shifts in text genre and the purpose of the texts analysed.
The 2010 texts primarily have the function of conveying facts and report-
ing the opinions of others. By 2011, texts of an evaluative nature have
a greater presence, and the opinions and evaluation of the authors are
frequently articulated as suggestions or instructions.
In 2012, an analysis of modality markers shows a further drop from
15.0 per 000 in 2011 to 11.9 per 000, meaning that by 2012, occur-
rences of modal expressions are approximately two-fifths of the level in
2010. Instances of epistemic and dynamic modality fall again, but deon-
tic modality continues to become more important, in terms of both abso-
lute numbers and proportion. By 2012, epistemic modality indicators are
at about one-third of the 2010 level. The mood used in the journalistic
writing of 2012 is more likely to be declarative—“this is how it is”—

than modalised—“this is how it might be”. There are very few reinforcing
modal adjuncts such as “totally” or “absolutely”. Yet evaluative language is
certainly present, as the following examples show:
In the regular course of business, the increase of familiarity is a good thing.

(Marketing Management, 27.4.2012, my emphasis)
The company’s fortunes were ravaged in 2010 by an explosion on the
BP-leased Deepwater Horizon rig that killed 11 workers, sent millions of
barrels of oil spewing into the sea and left it with huge compensation costs.
(Agence France Presse, 27.4.2012, my emphasis)
They sang an acapella protest blasting BP sponsorship of the World
Shakespeare Festival. (Coventry Evening Telegraph, 27.4.2012, my
emphasis)
The examples above contain no instances of modality. However, the words

“good”, “ravaged”, “spewing”, “huge”, “blasting” are all highly evaluative.
The interplay between modality and Appraisal is discussed further below.
In 2012, dynamic modality has reduced in number of occurrences, but
in some texts has taken on a more absolute character; that is, there are
fewer expressions such as “trying to”, “an effort to” and “working to” from
2010. In fact, dynamic terms such as “could/could not”, “try to”, and “in
hopes of ”, all of which appear frequently in the 2010 set, are entirely
absent in 2012. The following examples illustrate where dynamic modal
expressions are still present, but have a more definite tone:
Without ongoing brand benchmarking prior to a crisis, it is impossible to chart

how the crisis has influenced the brand’s decline, how long the crisis is lasting,
when the crisis has past [sic] and what steps are needed to restore corporate
brand equity. (Marketing Management, 27.4.2012, my emphasis)
People understand that business needs to make a profit, but they want to
know what the business stands for. (Campaign Middle East, 27.4.2012,
my emphasis)
“These days it’s hard to connect to the horror of what’s being done—pos-
sibly in our name—by oil companies.” (Coventry Evening Telegraph,
(Note in the first example that “impossibility” can also fall into the cate-
gory of epistemic modality, but here it is taken to refer to ability and futu-
rity.) If we accept Halliday and Matthiesen’s categories of high, median
and low modality values (Halliday & Matthiessen, 2004: 620), we can
generally categorise the 2012 expressions of dynamic modality as high in
value, with 68 % of dynamic modal expressions being of high modality
compared to 42 % of expressions in 2010. While the earlier texts related
to the difficulties inherent in making an impact on a physical entity, the
later expressions of dynamic modality take on the characteristics of defi-
niteness and conviction.
In contrast, deontic modality increases in 2012 in both numerical and
proportional significance, with the increase in evaluative and argumenta-
tion texts. These frequently take a didactic or instructional tone.
For any company, taking the decision to act and transform itself is a critical
step, but the key challenge, once this decision is made, is to identify and
define what it should do, how it should act and where. (Campaign Middle
East, 27.4.2012, my emphasis)
“but I do think there is some low-hanging fruit that should and could be
passed even this year.” They are:
1. Congress should move immediately to codify the reforms implemented
since the Deepwater Horizon disaster. (Foster Natural Gas/Oil Report,
In this last extract, the instances of deontic modality are offered in both
direct and indirect representations of the voice of Interior Secretary Ken
Salazar. The reversal of the more canonical “could and should” to “should
and could” throws particular emphasis on the deontic rather than the
epistemic modal item. The findings concerning the shift in modal types
recall Thompson’s (2004: 73) observation that texts such as leader articles
can show a progression from epistemic to deontic modality as the writer
moves from describing a situation to advising how to address it. In a sense
a similar progression is reflected across my entire data sets.
So by 2012, there appears to be a decrease in the mitigation and hedg-
ing that were characteristic of the reporting in previous years, primarily
in the direct and indirect reporting of other people’s (not the writers’)
views. At the same time, the increase in deontic modality appears largely
in the authorial voice, in suggestions of what participants in (or read-
ers of ) the texts should do. So while there is an increase in one form of
modal resources (deontic modality) to indicate the more overt intrusion
of the speaker, this is the only area of increase. The overall decrease in
both frequency and strength of modality indicates an increase in linguistic
certainty.
In Summary
My genre analysis might have generated the hypothesis that, because the
texts are moving from more “factual” towards more “evaluative”, they
would move from a state of certainty (this is what is happening) to a
state of uncertainty (but what does it all mean?). We might have expected
that an aim for objectivity would give way to a more subjective view of
events, and thus a greater intrusion by the writer. In fact, the movements
go in the opposite direction. This move to certainty could rather be seen
as a move from a state of “newness” to a state of “givenness” (Halliday &
Matthiessen, 2004). With the passage of time, writers are presenting their
observations in such a way that they are incontrovertible. A statement
without modality is the most unobtrusive way, grammatically, to signal
unequivocal information.
However, the analysis also suggests an increase in evaluative language,
and here a distinction between the processes at work in the modality sys-
tem and those at work in the Appraisal System (Halliday & Matthiessen,
2004; Martin & White, 2005) are worth making. Following the work of
Hunston and Thompson, Martin and White (2005: 38) comment that:
Opinions about entities are canonically attitudinal and involve positive

and negative feelings; opinions about propositions on the other hand are
canonically epistemic and involve degrees of certainty.
From this perspective, we see a discernible move from a highly modalised

(“canonically epistemic”) means of expression towards one that is high in
force in terms of the Appraisal System (“positive and negative feelings”).
Table 9.2 Summary of findings—modality and Appraisal

Descriptive genres Evaluative genres
Generic Low modality High modality
expectation
Findings from High Lower
data • Modality mainly attributed to • Opinion presented as
other writers/speakers fact
• Some authorial modality • Use of other
evaluative strategies
Type of modality Epistemic Deontic
found “It might be” “You should”
Findings for Present High
Appraisal • “Semantics of intensification” • Judgement: social
sanction
This could indicate that the crisis moves from a set of disparate propo-
sitions about the relevant events, presented using the resources of the
modal system, towards being a single entity with a recognisable set of
characteristics, presented using resources that are more relevant to atti-
tude and judgement.
Table 9.2 summarises these findings on modality and Appraisal across
my texts, as they move from more descriptive genres to more evaluative
genres.
The findings indicate that news media writers have become increas-
ingly clear, certain and judgemental in their approach to the topic of the
BP Deepwater Horizon events. In effect, they are asserting by this stage:
“we know what this is”.
10
of Mythic Meanings
My analysis in this chapter moves to a consideration of the linguistic

features related to figurative language and connotation, that is, to expres-
sions at the level of mythic meanings. For the BP data, metonymy and
metaphor were of interest.
Feature 7: Metonymy
I presented an outline of the rhetorical trope metonymy in the context of

the semiotic level of connotation or myth. My definition of metonymy
includes both the specific concept of metonymy (substitution accord-
ing to contiguity) and synecdoche (whole for part and part for whole
substitution). My analysis strategy for metonymic expressions was firstly
to identify all instances of metonymy within my data sets. In this pro-
cess of identification, I used for support and terminology the sum-
mary lists and categorisations of metonyms found in, amongst others,
Chandler (2007), Lakoff and Johnson (1980), Cornelissen (2008) and

in Discourse, DOI 10.1057/978-1-137-58750-3_10
Radden, Köpcke, Berg, and Siemund (2007), placing metonyms in groups

such as ORGANISATION FOR MEMBER, CONTROLLER FOR
CONTROLLED and so on. As well as using the guidelines offered in these
works, I also followed the general principle outlined by Dirven (2002) that
metonymic expressions can often be recognised by the use of an inanimate
subject with a verb usually requiring an animate subject. Dirven gives as
examples the verbs “mean” or “use”, and there are many usages of this type
in my own data, for example “The US Census Bureau recently declared
…” (Canwest News Service, 27.4.2012). In discussing the functions of
metonyms, I took into account that metonymic expressions select an aspect
of the entity described for attention, and in doing this necessarily disregard
other aspects of the entity that might have been selected.
Table 10.1 shows the fall of the occurrence of metonyms from 9.7 per
000 words in 2010 to 3.2 instances per 000 words in 2012. I follow the
convention of referring to metonyms in the format “X FOR Y” and in
capitals.
The pattern of metonymy in the 2010 data is relatively straightforward.
Metonymic expressions (which can be words or phrases) occur nearly
ten times for every 000 words, as Table 10.1 shows. By far the major-
ity of metonymic expressions in 2010 are those that use an organisation
Table 10.1 The occurrence of metonyms in the 2010–12 BP texts

2010 2011 2012
[BP] ORGANISATION FOR MEMBERS 36 23 4
[OTHER] ORGANISATION FOR MEMBERS 33 15 9
MACHINE FOR CREW 13 0 0
INSTRUMENT FOR PRODUCER 1 7 2
CONTROLLER FOR CONTROLLED 2 0 0
ORGANISATION FOR INDEX 0 1 0
PLACE FOR PERSON/PEOPLE 0 3 3
Other metonymic expressions 3 3 3
Extended metonyms 0 0 2
Total 88 52 23
Total per 000 words 9.7 7.4 3.2
10 Stage 3: A Depth Analysis at the Level of Mythic Meanings 169
name to indicate a person or persons, namely, ORGANISATION FOR

MEMBERS. These expressions typically include a verb that would nor-
mally require a human subject, for example, in this data extract:
British energy giant BP said Tuesday that first-quarter profits rocketed on

higher oil prices but admitted that the news was overshadowed by last
week’s tragic accident at a rig in the Gulf of Mexico. (Agence France Presse,
27.4.2010)
Unsurprisingly, the most common organisation used in this way is BP,

whose corporate comments and actions are central to the story. Other
organisations frequently used metonymically are the Coast Guard and
rig-owner Transocean. Using the metonymy ORGANISATION FOR
MEMBERS can have different kinds of effect. Cornelissen (2008) makes
the point that metonymy is used in this way primarily for convenience.
However, he goes on to argue (2008: 90) that using the company name
with a verb normally requiring an animate subject supports the wide-
spread metaphor by which the company is seen as a person.
While the initial motivation for the metonymy may have been primarily
referential as shorthand for a relative clause, the use of this kind of
metonymy also cues a metaphorical image of a company as a person or
human being.
According to this thinking, by using metonymy in constructions such as

“BP said it is committed to doing everything in its power” the company
“is imbued with a certain ‘corporate personality’ or ‘corporate identity’”
(2008: 90). Cornelissen presents this observation neutrally. However, I
would argue that it is generally to a company’s advantage to represent
itself as a single entity. In particular in the case of a crisis situation, it is
considered crucial that the company “speaks with one voice” and acts as
one with a single aim and purpose (Burt, 2012). In this sense, the conve-
nience of using “BP said” rather than naming a person whose identity is
not relevant might serve the additional purpose of contributing towards
a positive sense of a unified entity.
The corollary of this observation is to conceive the ORGANISATION

FOR MEMBERS metonymy as obscuring agency. In the 2010 situation
where the question of fault, blame and responsibility was at issue but as
yet unexpressed, the use of the designation “BP”, with all the weight of
its brand values and corporate identity, has a different role in meaning-
making than “Doug Suttles, chief operating officer of BP Exploration
and Production”. In this way, when “BP” is used metonymically for
actors within the company, it works both to represent an existing reality,
by indexing all the pre-existing associations readers have with the organ-
isation, and to construct reality, by representing the company as a human
being. This applies also to the other organisations used metonymically
within the data set.
In the case of BP, however, there was a departure from the typical pub-
lic relations model. CEO Tony Hayward decided to step out from behind
the metonym “BP” and become the public face of the crisis. Many of
the 2010 texts include quotations from Hayward personally, from press
conferences, press releases and direct from the shores near the oil spill. As
Bergin (2011: 166) writes:
This constituted Hayward’s third big PR mistake: he had decided to front

the response effort himself. If not the most fatal, it was certainly the most
public of his mistakes in handling the crisis. As the CEO of a rival would
later tell him, “You stopped being the CEO and slipped into chief operat-
ing officer mode.”
Because of this, it is possible that the frequency of occurrence of BP as

ORGANISATION FOR MEMBERS is lower than it might have been
in a business crisis where key figures seek to present comments as emanat-
ing from the company as an organisation rather than as being the words
of particular individuals.
The use of metonymic expressions drops from 9.7 per 000 words in
2010 to 7.4 in 2011, with some marked shifts in type of metonym. The
fall in absolute numbers of metonyms is due to the fact that the metonym
type ORGANISATION FOR MEMBERS drops by more than half: the
2011 texts do not feature as many quotations from BP or other organ-
isations as was the case in 2010, meaning that it is less common that
the organisation stands in whole-for-part relation for the spokesperson.

Where this type of usage does occur, it is more likely to be in relation to
the 2011 Quarter 1 results, than the Deepwater Horizon events them-
selves, as in the following example:
After BP PLC reported Wednesday that net profits rose 16 percent in the
first quarter, company officials acknowledged the company has applied for
permits to restart drilling in the Gulf. (The Associated Press, 27.4.2011,
my emphasis)
The category ORGANISATION FOR MEMBERS covered a limited

number of organisations in 2010, focusing on direct participants. In
2011, this group of metonyms refers to a wider spread of organisations,
including the US government, the Obama administration, Research and
Markets (a publisher) and Shell. This observation accords with the frag-
mentation of types of actor in the data. Similarly, metonymic usages of
the type PLACE FOR PERSON appear to show a movement of interest
from the core location of the events in the Gulf to further afield, referring
in 2011 to Florida, the White House, and “sister states”.
The category INSTRUMENT FOR PRODUCER appears more sig-
nificantly in 2011 than in 2010, and relates largely to reports and legal
documents, in constructions such as the following:
Coast Guard report slams Transocean over Deepwater Horizon. (SNL Daily
Gas Report, 27.4.2011, my emphasis)
The Times editorial seems to ignore the applicable legal background. (SNL
Daily Gas Report, 27.4.2011, my emphasis)
The legal remedy that promises to give Florida the maximum recovery in the
shortest time is the federal Oil Pollution Act, which makes BP and any
other responsible party strictly and fully liable for such harm. (Tampa Bay
Times (Florida), 27.4.2011, my emphasis)
These are typical examples of inanimate entities associated with verbs

that call for an animate subject, but they also exemplify the process of
representation I mentioned earlier in my discussion of intertexts, by
which reports and documentation of various kinds, instead of direct par-
ticipants, are called upon to witness and make meaning of the events.
By 2012, very few metonymic expressions are used. It is under-

standable that BP would feature far less frequently in the construction
ORGANISATION FOR MEMBERS than it did in 2010. What is inter-
esting is that it has not been replaced by the use of any other organisations
in this whole-for-part way, and this reflects the limited role that organ-
isational statements of any kind are now playing in the BP story. PLACE
FOR PERSON has stayed at about the same level, albeit low. The instances
featured in the 2012 data set are “the State [estimates]”, “the city [is catch-
ing up]” and “the City [expects]”, reflecting the longer-term implications
of the effects of the spill not on directly affected areas but those affected
indirectly. Here the metonymic references are to North Dakota (state) and
Williston, North Dakota (city) and London’s financial centre (the City).
The category INSTRUMENT FOR PRODUCER is lower in 2012 than
2011, but, as in the previous year, helps demonstrate the increasing role of
reporting texts mentioned in 2011 in the phrases “social media [has taken
corporate responsibility]” and “the report [guides people]”.
In Summary
Key changes in the use of metonymy can be summarised as follows:
1. A drop in metonymic expressions over the timespan of the data, due

largely to the decrease in the form BP FOR MEMBERS.
2. Metonymic expressions become more likely to be used in the
contexts:
• INSTRUMENT FOR USER, especially in 2011 to refer to reports

of various kinds.
• PLACE FOR PERSON where the places are increasingly distant
from the Gulf States.
The findings of the metonymy analysis provided additional evidence

for the distancing of the story from its original sources on a number of
dimensions—distance from BP, reliance on written accounts and the spa-
tial distance of the geographical locations mentioned.
Feature 8: Metaphor
Like metonymy, metaphor is a rhetorical trope, and one that is identi-

fied as a particular source of innovation in language. Chapter 3 offers a
broad overview of scholarly work on metaphor, and of Jakobson’s (2002)
critical distinction between the metaphoric and metonymic dimensions
in representation. In the same way as for metonymy, my research process
consisted of identifying metaphors, grouping them by type (in this case,
sense domains) and examining their frequency, the nature of source and
target domains, and changes over the time period of the data set.
The first stage was to identify all metaphors in order to gain an over-
view of the use of metaphorical language and how this changes over the
three-year period. This was an emergent approach to the data, where no
assumptions were made about the type of domain of metaphors encoun-
tered, and this approach was aligned to the non-directive nature of my
research questions. In other words, I did not seek to explore the frequency
and usage of specific metaphorical domains, as is the case in, for example,
Koller and Davidson (2008) who examine social exclusion; Koller (2003)
on business and war/sport; Milne, Kearins, and Walton (2006) writing
about business as a journey and White (2003) on economics and growth.
Rather, I sought to identify metaphorical usages of any kind, and investi-
gate themes that emerge: the recurrent domains that are used by journal-
ists to conceptualise the BP crisis and its related outcomes.
The task of identifying all metaphorical usage is complex. The con-
ventional denotation X IS Y (e.g. BUSINESS IS WAR) can imply that
nominal metaphors are most typical (Cameron, 1999); however, in ver-
bal language, metaphors span all word classes. Although metaphors are
usages where one entity or process is described in terms of another, for
some metaphors (“dead” metaphors), the presence of two sense domains is
weak or no longer discernible. Nevertheless, I included dead and dormant
metaphors in my analysis, on the grounds that even these non-creative
usages can contribute, albeit weakly, to a dominant view of the phenom-
enon in question. Lakoff and Johnson (1980) argue that metaphors we
scarcely notice still structure our shared thinking in fundamental ways.
In a business context, Koller (2003: 88–89) argues that weak or dead

metaphors, nevertheless, give additional weight to dominant themes.
Further, the distinction between live, dead and dormant metaphors is not
always easy to define. As is the case in much linguistic analysis (e.g. the
identification of genres) prototypical examples are clear, but much other
identification and categorisation can be a matter of judgement.
In the BP texts, metaphors either belonged to a recurrent group (e.g.
the common group BUSINESS IS WAR) or were one-off usages. In the
case of recurrent groups, I included all three categories of live, dormant
and dead metaphors, on the grounds that dead metaphors may still sup-
port a particular metaphorical line of thought in a congruent way.
Table 10.2 shows the frequency of metaphorical uses from 2010 to 2012.
A frequency count is a relatively superficial way of investigating meta-
phor, but a number of interesting changes were evident over the period of
the data, which formed part of the evidence for wider patterns. These were:
Table 10.2 The occurrence of metaphors in the 2010–12 BP texts

2010 2011 2012
OIL SPILL IS MALEVOLENT/WILD CREATURE 23 0 3
FINANCIAL ITEMS MOVE IN SPACE 20 39 4
WEATHER IS ANIMATE 9 0 0
Metaphors in source domain of MYTHS AND LEGENDS 1 4 8
BUSINESS IS SPORT 1 10 8
BUSINESS IS A JOURNEY 0 1 15
BUSINESS IS WAR 0 12 7
OIL IS WATER 0 0 6
Metaphors in source domain of THEATRE AND ART 0 2 5
Metaphors in source domain of CRIME 0 0 5
BUSINESS GROWTH IS LIKE BUILDING 0 6 4
BUSINESS GROWTH IS ORGANIC 0 1 2
COMPANY IS HUMAN 0 4 2
Other live metaphors 4 8 23
Other dormant metaphors 14 37 50
Other dead metaphors 14 26 41
Totals 86 150 183
Totals per 000 words 9.5 21.4 25.8
1. The occurrence of metaphors of all kinds increased over the span of

the data.
2. The source and target domains changed over the span of the data.
Firstly, the general rise in frequency of metaphorical usage may have a

number of drivers. Studies indicate (e.g. Krennemayr, 2011) that news
writing is quite metaphorical, and that, if anything, hard news writing
uses more metaphor than soft news. These findings relating to occur-
rence of metaphor run counter to my own analysis of the BP texts, where
the pattern of text type shifts from primarily news and financial reports
(Krennemayr’s “hard news”) towards evaluative writing such as editorials,
travel pages, reviews, letters, business articles and personal blogs, which
are more aligned to her definition of “soft news”. The shift is not whole-
sale, but Krennemayr’s findings would predict a drop, rather than a rise
in metaphor. Since I find that genre is an important explanatory factor
for linguistic movement in this BP corpus over the three-year period, and
since Krennemayr’s work is relevant, providing a specific review of how
metaphor works in news within a substantial corpus, it is worth taking
this apparent discrepancy seriously. It would seem that both the higher
incidence of metaphor in Krennemayr’s texts and the higher incidence of
metaphor in hard news compared with soft news can largely be explained
by a single categorisation difference. Krennemayr takes the formulation
ORGANISATION FOR MEMBERS to be a metaphor, which she labels
personification. Indeed, she cites this as being a key explanation for the
“unexpected result” (2011: 123) of the relative overuse of verbs in their
metaphorical sense in news. I classify such usages as metonymy, with the
justification that they signify a whole–part relation, and in doing this I
follow Chandler (2007) and Cornelissen (2008). If this category were
added into the analysis of metaphor, it would certainly change the pat-
tern to be more in line with Krennemayr’s expected behaviour for the
data. Her overall point remains very pertinent to this work, namely, that
news writing is highly metaphorical, and that metaphor has a number
of functions that are specific to the genre—including making complex
or abstract concepts more accessible to the reader, a cohesive function to
create a satisfying whole, a rhetorical persuasive function and a way of
creating humorous effects (Krennemayr, 2011).
Secondly, there is noticeable change in the clusters of metaphors

that appear regularly. Target domains in 2010 are the oil spill itself, the
weather and financial items such as shares and profits. Both the oil spill
and the weather are represented through metaphor as animate, threaten-
ing or uncontrolled.
Louisiana-based BP spokesman Neil Chapman said 49 vessels—oil skim-

mers, tugboats, barges and special recovery boats that separate oil from
water—are working to round up oil as the spill area continues to expand.
(Carleton Place (Canada), 27.4.2010)
The sunken BP and Transocean oil rig is spewing 42,000 gallons of crude
a day. (NewsWatch: Energy, 27.4.2010)
And the Coast Guard unfortunately admitting that no matter how much
cooperation they get from the currents as well as the winds, it probably will
not be able to stop that 1800-mile slick from splashing onto shore by the
weekend. (CNN, 27.4.2010)
By 2011, targets remain financial items, given the first quarter results, but
metaphors relating to business become more predominant, particularly
those that offer a combative view of business, using the source domains
WAR and SPORT. This was found in context to relate mainly to BP’s
business struggles one year after Deepwater Horizon, which concern not
only the oil spill, but also other difficulties such as BP’s business deal-
ings in Russia. By 2012, as 27 April does not fall on results day, financial
texts have dropped in number and proportion, and related metaphors
are greatly reduced. Business is still a target for metaphorical expressions,
and those relating to WAR and SPORT are still in evidence, although
there are fewer, and in the area of business some alternative metaphorical
constructions have emerged in the areas of business growth and business
conceived as a JOURNEY.
Various metaphors for business growth have been identified in litera-
ture, including parenting (Dodd, 2002) building (Dodd, 2002; White,
2003) and organic growth. Within the 2012 data set, examples of the
BUILDING metaphor appear, in phrases such as the following.
Building brands through behaviour. (Campaign Middle East, 27.4.2012,

my emphasis)
Shell turned a profit of $7.7bn (nearly £5bn) in the first three months of
this year and the trading performance will buttress already good market
sentiment around its recommended acquisition of Cove. (The Scotsman,
The examples above are rather conventional usages, and only indirectly
related to the BP events: in the first case the metaphor refers to a theoreti-
cal or hypothetical brand (advice for building a brand: don’t do what BP
did!), and in the second to BP rival Shell.
This widening of application is true of the other cluster of business
metaphors that are more conspicuous in 2012, namely, BUSINESS IS A
JOURNEY. Of the 15 instances of the metaphor identified in 2012, only
one refers to BP itself, and this is the third shown below.
Brands that embrace this new honest and responsible world have an excit-
ing future. And agencies that can help their clients understand, navigate
and deliver in this new world will be more important than ever. (Campaign
Middle East, 27.4.2012, my emphasis)
The next step is to find the idea that can be used as a strategic compass not
only to communicate this externally but also to galvanise the organisation
itself. An idea that lies between the two biggest trends impacting business
today: social responsibility and social media. (Campaign Middle East,
The troubled energy major, which is seeking to move on from the US
Gulf of Mexico oil spill disaster in 2010, had returned to profit last year
with net annual earnings of $23.9 billion. (Agence France Presse,
Once again these metaphors are conventional, but I suggest that choice
of conventional metaphors can still be telling. The third extract above
refers specifically to BP, “the troubled energy major”, proposing that busi-
nesses need to move on from crisis situations. It would be reasonable to
assume that BP, and other oil companies mentioned, could still be por-
trayed as “battling”, “manoeuvring” and “using weapons”, but there is a
more “questing”, “building”, “pathfinding” tone to the metaphors used
by 2012. Milne et al. (2006) explore the JOURNEY metaphor in the
context of business writing on sustainability, finding that in that context,
the metaphor depicts a journey without a destination, and hypothesising

that the journey metaphor is a useful device for avoiding commitment
to a definite end goal. It is possible that a similar ambiguity is relevant in
speculation on the future of BP post-Deepwater Horizon.
Similarly interesting is the emergence of MYTH AND LEGEND as
a source domain in the 2012 texts. The concept of myth and legend is
realised in the following examples:
Attendees of the United States Energy Association’s (USEA’s) membership

meeting, taking place simultaneously at the Washington DC-based club,
were invited to listen in as the Secretary blasted unnamed Washington
insiders for perpetrating “fairy tales” about imagined obstacles to oil and gas
drilling and expansion on US offshore and onshore federal properties.
(Foster Natural Gas/Oil Report, 27.4.2012, my emphasis)
“Like presidents of both parties before him, however,” Mr. Coll writes,
“he lacked the depth of conviction, the political coalitions and the scien-
tific vision to do more than toss relative pennies into a wishing fountain.”
(The New York Times, 27.4.2012, my emphasis)
These metaphors are not about BP’s progress itself, but appear in co-text.
It would be overburdening the findings to suggest that the metaphors of
journey, myth and art somehow recontextualise our perception of the BP
events. However, these metaphor clusters, emerging as they do in 2012,
may be a small part of a general pattern that places the BP into a less
concrete and more abstract context.
In Summary
There is evidence to suggest that the use of metaphor within the BP texts
increases. This is a slightly unusual pattern, given that news reports in
particular are held to be highly metaphorical, but can partly be explained
by an overlap in definitions of metaphor and metonym. The other change
is in metaphor domains, where metaphors relating to the uncontrollable
weather and oil spill give way to metaphors in the domains of journey,
myth and legend, which add a different (more symbolic) dimension to
the BP story, albeit to a limited extent.
11
of Ideology
So far in this depth analysis stage, I have covered eight of the nine features
of discourse relevant to my data set. These eight features have been situ-
ated at the semiotic levels of “sign”, in the case of naming and categorisa-
tion choices, “code”, in the case of the explorations of modality, genre
and intertextuality, or “myth” in the case of the rhetorical tropes of meta-
phor and metonym. An identification of discourses should draw on any
or all of the eight language features in order to identify persistent patterns
that point to the presence of ideological positions. A consideration of
discourses is one way of visualising what my disparate set of language fea-
tures might “add up to”. I was open to the possibility that new discourses
may emerge year by year, to reveal a continually shifting pattern, or dis-
courses may overlap the data sets, or co-exist in a single year.

in Discourse, DOI 10.1057/978-1-137-58750-3_11
Feature 9: Discourses
My particular research interest was in news media representations of

the BP events, and the ideologies that were important to my account
were those that informed the ways in which the story of the Deepwater
Horizon events was represented to the reading public. In investigating
significant discourses, I sought patterns and characteristics of the repre-
sentation of the BP events which suggested shared ways of thinking, as
well as changes in thinking over the time period of the data. Discourses
have been found problematic to analyse systematically, and the work of
Foucault does not provide specific methodological guidance (Graham,
2005). Sunderland points out (2006: 166) that discourses are “partial”,
“non-finite” and “non-ubiquitous”—and subject to the perspective of the
individual analyst.
Different discourses are accordingly likely to be “spotted” by different

social groups of readers and analysts—for example those who favour a
feminist perspective and those with a more traditionalist perspective—even
when looking at exactly the same textual set of linguistic traces.
Sunderland defines a number of linguistic features as relevant to discourse

identification, including lexical choices, verb forms and moods, speech
acts and collocations. She points out that the analyst should be alert to
what may be omitted as well as what is present. Fairclough’s (1989, 1992a,
1995a) approach to identifying the presence of ideologies focuses strongly
on analyses within the framework of Systemic Functional Grammar,
where linguistic choices concerning, for example, actor, mood and word
order are held to be significant in positioning speakers and audience in
certain desired relationships. A great diversity of individual and mixed
method approaches have been used in the identification and analysis of
discourses, including content analysis (Potter & Reicher, 1987), gram-
matical analysis (Fairclough, 1989), conversation analysis (Speer, 2001),
poststructuralist approaches (Baxter, 2008; Wetherell, 1998) and corpus
methods. Many analysts propose a mixed method approach in their selec-
tion of tools for analysing discourses (e.g. Baxter, 2010; Cook, Robbins,

& Pieri, 2006).
To identify patterns in the BP data relating to the ninth feature—dis-
courses—I drew on my findings for the other eight discursive features. In
my analysis so far, I had already identified a number of recurrent themes.
I refer to these now as evidence of four representational discourses.
Discourse 1: Objective Factuality
In 2010 the dominant text genres were the news report and the financial
report, and I have shown in analysis that the texts within these genres
largely conformed to genre type in the ways they use language to give
an impression of objective factuality, which I suggest was the dominant
discourse in the 2010 texts.
Several of the individual analyses described above indicate prototypi-
cal features of the representation of “objective reality”: the prevalence of
facts and figures, a use of modality attributed to reported voices, the mar-
shalling (and repetition) of supporting statements from a limited num-
ber of sources perceived as credible and metonymic usages that replace
people with institutions as social actors. The typical structure of these
reports is the inverted pyramid structure, which is one of the conven-
tional markers signalling objective reporting. Alongside these expected
stylistic and structural language features are others that are less charac-
teristic of the prototypically “objective” news report. Some are explicable
in the context that my selected news texts appeared very early in the
course of events. There is evidence of uncertainty in naming choices for
the events, and the categorisation of the oil spill amongst world events is
as yet tentative. Facts and figures are sometimes unspecific. While modal
expressions are generally a feature of quoted sources in the texts, other
instances of modality can be seen in authorial voices, serving to qualify
the unmodalised declarative mood which is typical of the news report
genre, and indicating the level of uncertainty surrounding the events at
this stage. This is evident in the metaphors used to describe the spill and
the weather, which represent these as uncontrollable forces. There is occa-

sional use of a narrative or storytelling structure, which is less typical of
“factual” reports. While these features would be unusual in quantity for
news report writing, in the frequency in which they occur I judge them
to be further indicators of an objective–realistic presentation, in that
they indicate an acknowledgement of a messy and uncertain reality that
accords with readers’ experience. An excess of certainty and an entirely
declarative style would be neither credible nor easily readable.
In describing the “factuality” of news as a construct, I accept that
representations of “fact” are ideologically grounded. In aggregate, these
result in versions of news stories that are aligned variously to the interests
of media organisations, readers, other journalists and (sometimes) par-
ticipants in the story. Where these interests are in conflict, there can be
a struggle for control over the information presented. In the 2010 texts
there is direct evidence from my texts of BP’s attempts to control both
the content and expression of information via impersonal linguistic con-
structions in press releases, interspersed with personal (non-metonymic)
communications from CEO Tony Hayward. Alternative voices, such as
environmental and anti-globalisation interests, have a limited or miti-
gated presence in 2010, as there is considerable uncertainty about the
scale, reach and duration of the spill.
Discourse 2: Positioning
Typical markers of the discourse of objectivity continue to be discernible

in many of the 2011 texts, including the presentation of empirical facts,
neutral lexis and some distancing of journalists from commentary on
their material through the use of reported speech. At the same time, other
2011 texts show a greater degree of engagement by journalists, who have
started to express their own opinions overtly, and to summarise, synthe-
sise and interpret information about the oil spill. As a crucial part of this
process, journalists place and locate the unknown (the BP events) within
the known, in other words, positioning it as a certain entity. While the
discourse of objective factuality continues to be a feature of the 2011
texts, a discourse of positioning is the characteristic feature in this year.
The positioning of the events is recognisable particularly in the

linguistic process of categorisation, which has increased significantly
in 2011 over 2010. Through groups, lists and comparisons, BP is posi-
tioned, unsurprisingly, as a “disaster” (to use one of the key terms from
the 2011 texts) and journalists define the events by analogy with other
events, which in their turn have been previously defined and categorised.
I have commented that the 2011 lists and groups are what we might
term “expected”. So by 2011, journalists follow a relatively proscribed
pattern of writing about disasters. Amongst other strands, the following
topics are typically of interest (Arpan, 2002; Stephens, Malone, & Bailey,
2005):
1. The cost: for restoration and compensation.

2. Blame and responsibility: for the cause of the disaster (not applicable
to natural disasters, but applicable to BP) and the subsequent handling
(applicable to both natural and man-made disasters).
3. The future: when the immediate effects of the disaster (e.g. oil on the
beaches) will have been tackled. Any changes arising from the disaster
(e.g. in oil drilling policy).
These are only three of a number of regularly rehearsed ideas about disas-
ters of this kind. These shared ideas raise a series of expectations about
media coverage (that the questions inherent in them will be answered),
and the texts in my data set show evidence that this is the case. In 2011,
the texts increasingly address the questions of cause, blame and respon-
sibility, as well as compensation. They address issues of recovery, such
as the form this is taking, and how long it will be before normality is
resumed. Addressing the questions that arise from generally held assump-
tions is part of a longer-term process towards a sense that we now under-
stand what the events mean. The 2011 texts fulfil a critical role before
this can happen, and this role is one of definition and location. Here, the
relevant information is “bigger/smaller than what?”, “like/unlike what?”,
“as expensive/not as expensive as what?”
Categorising, listing and grouping are the key features of a discourse of
positioning, but three other processes are also relevant. Firstly, the analy-
sis of the news report genre in 2011 shows that the BP story increasingly
forms part of other news stories; 27 % of the texts are news stories, but
only 15 % relate primarily to the BP events. The rest mention BP as part
of another story. Secondly, represented events are positioned by means
of intertexts. That is, it becomes more common that the researched texts
comment on other texts about the events rather than the events themselves.
In this way, an agreement about the meaning is being fixed and shared,
through the use and spread of common, legitimised sources. Thirdly, this
involves temporal positioning—texts start to place the BP events within a
diachronic context. Four broad time bands are visible as strands through-
out these texts: DISTANT PAST → RECENT PAST (2010 EVENTS)
→ CURRENT EVENTS (2011) → FUTURE. Current events, as cov-
ered in the 2011 texts, are more or less directly related to the 2010 events
of a year ago—these are the touchpoints for the 2011 commentary. The
more distant past is evoked in phrases such as “the biggest oil spill in US
history”, which is used several times in the 2011 data set, serving to place
the oil spill in the context of events further back than 2010. At the other
end of the time spectrum, a number of the texts refer to planning for the
future. These include the Panama City Beach text (marketing planning),
an item on predicting future oil prices, the text headlined “BP expects to
resume Gulf drilling this year” and another whose headline is “Prepare in
Advance for the Inevitable Crisis”.
Discourse 3: Redeployment
The 2011 texts also begin to show evidence of another discursive feature:
a high tolerance for non-neutral and critical voices, for example, a blog
and an editorial about the high price of oil, a book about BP’ s alleged
management failings and Twitter criticism one year after the oil spill. In
the same way as many 2011 texts are still part of the 2010 discourse of
objective factuality, I suggest that others featuring alternative or resistant
voices also belong to a discourse other than the positioning discourse—that
of redeployment—which becomes even more evident in the 2012 texts.
In 2012, I suggest that two types of representational discourse are evi-
dent—one of which, rather than continuing the process of positioning in
a further linear direction of fixing meaning, serves instead to question
fixed meanings and offer different meaning potentials, and this is what I
mean by a discourse of redeployment. In this discourse, once a meaning
for the BP events has been agreed, or partially agreed, it can be regarded
as “known” and redeployed as a resource with a (temporarily) fixed mean-
ing. This is, I propose, what can be observed in three significant features
of the 2012 texts—firstly, the increase in the exploration of the meaning
of BP events through creative works; secondly, the tolerance for alterna-
tive voices (resistant discourses) and, thirdly, the use of a creative form of
categorisation to play with meaning.
I showed in my analysis how creative works about the BP events
became part of the media representation of the phenomenon. These cre-
ative works use the new signified, or the positioned concept of the events,
as a starting point for exploration of meaning at a societal level. It is only
once there is a shared concept of the BP oil spill, that it becomes available
for interpretation in this way. In reporting and commenting on these
creative treatments, the media texts mark them as “other”, and are able to
position these texts for the reader, through selection, approval and disap-
proval, in ways they choose. However, these challenging artistic represen-
tations—if widely seen and adopted—can serve to shift the positioning
of the events described above, in an iterative development of meaning.
The challenge presented in creative works is often through the articu-
lation of alternative or resistant discourses. Artistic interpretations often
represent resistant discourses, frequently challenging how the mass media
has positioned the phenomenon—through humour, by adding layers of
complexity or by setting it alongside other phenomena in a productive
interplay of comparison and contrast intended to yield new insights. So,
for example, the Margaret Atwood book mentioned in one of the 2012
texts takes the idea of the BP events and their subsequent outcomes as
moral failure, rather than, say, the inevitable consequence of the risky but
heroic venture of delivering oil to allow global progress. Blogs, letters and
reports of political speeches also show evidence of resistant discourses:
issues of compensation, financial impacts and future drilling agendas in
the Gulf and elsewhere are exposed and aired in print and online texts in
the data set. I have mentioned in the introductory chapter that control
by traditional media institutions has been increasingly broken down by
mass access to the Internet, and one of the outcomes is a considerable
increase in voice for minority and alternative views. Within the 2012
(and some 2011) texts, critical voices stand in contrast to, for example,
financial reporting, in which the oil spill tends to be mentioned as an
explanatory footnote to Quarter 1 results. However, my analysis shows
that while critical voices, or resistant discourses, are given a place in the
texts, particularly if they belong to politicians or others with ready access
to a voice, they are frequently “bracketed”. By this I mean that they are
placed in non-news pages (so are not endorsed as factual), or they are
quoted directly whereby the journalist does not take responsibility for
them, or they are positioned as non-mainstream (e.g. they are part of
artworks) or they are backgrounded through rhetorical choices. So while
institutional and non-institutional voices contribute to this conversation,
non-institutional voices can be positioned as to one side of the discus-
sion. As Conboy (2007: 97) notes:
Hierarchies of news value can be varied to allow unfamiliar and even con-
tentious voices to be heard as long as they do not destabilize longer term
patterns of meaning.
Thirdly, the concept of creative categorisation refers to the observation

that the linguistic feature of categorisation is employed in a rather differ-
ent way by the 2012 texts than it was in 2011. Where in 2011, the events
were placed in expected categories for the purpose of positioning, by 2012
they are mentioned in contexts that can be quite distant from oil spills,
and placed alongside quite other social phenomena in a way that suggests
an exploration of the boundaries of the partially agreed meaning of the
concept “BP oil spill”.
Discourse 4: Naturalisation
At the same time as, and in tension with, the discourse of redeployment is
the tendency to fix and normalise meaning by presenting interpretations
of the crisis as “given”. This discourse is aligned with Barthes’ concept of
the naturalisation of representation, and can be seen as a progression of the
discourse of positioning, whereby once events are located within known
frameworks, they are presented as naturally understood in certain ways.
So throughout the three data sets, but particularly by 2012, sit

discourses that are concerned to normalise the catastrophic events, and
absorb them into the general consciousness. These discourses are part
of a longer-term naturalisation of the understanding of events (Barthes,
1972), by which the meaning of the BP oil spill is seen as presumed and
unarguable. They are evident in a number of linguistic patterns. There is
a continued decrease in modal expressions of uncertainty, suggesting that
propositions are not open for question or challenge. The range of nam-
ing terms for the events is smaller, and the terms are longer, incorporat-
ing shorthand descriptions of the events that propose brief and selective
ways of understanding them. The BP story is increasingly reduced to an
explanatory footnote or an illustrative example within a different story.
For the BP oil spill to have any resonance as explanatory footnote or
illustrative example suggests that a process of naturalisation is under way.
Strategies by writers to present the events as naturalised can be seen
to serve both business and socio-cultural interests. With time, events are
positioned as no longer new and unique, but past and “other”. From
a specifically business perspective, a return to business as usual is the
key aim for business management (de Cock, Cutcher, & Grant, 2010;
Stephens et al., 2005), for whom crises are seen as part of the nature of
business development. Coverage in my set of texts draws upon a number
of entrenched assumptions about business in general, and the oil business
in particular.
1. That business needs to continue to grow, whether or not this is logical

from the point of view of sustainability (Western, 2010).
2. That business involves inevitable risks and setbacks. These need to be
“dealt with”, “beaten” and “recovered from”. Indeed, the business cri-
sis can be construed as an opportunity to expedite desired change or
renegotiate leadership practices (Mitroff, 2005; O’Reilly, Lamprou,
Leitch, & Harrison, 2013).
3. Business in general, and the oil business in particular, is regarded as
adventurous, exciting and pioneering. Seen in this way, risk (although
not its negative outcomes) is potentially both exciting and admirable.
So texts such as those of the business and market reports genre present
the events from the perspective that crises such as oil spills are inevitable,
but that measures have been taken to mitigate their likely future occur-
rence. In two of the researched texts consultancy services advise on cri-
sis management and restoring business reputations. I have mentioned in
analysis texts which deal with BP’s forward planning, as well as calls for
the resumption of deepwater drilling in the Gulf.
Outside the business context, media coverage refers to the BP events
within the context of known social phenomena. The events are no lon-
ger the exclusive province of business and the environment, but part of
a shared history, and aligned with quite disparate concepts through the
processes of redeployment. One difference between business and non-
business media coverage is that outside the area of business, there is less
concern to naturalise the events by treating them as part of the “rough
and tumble” of business. Rather, they are a resource to be drawn on as
a representative token of a certain type of human experience, be it an
environmental issue, a huge lawsuit or an example of complacency about
safety. This is a more complex reading of the representation than the
business reading. By 2012, resistant discourses are still evident (environ-
mental protest being the most obvious example) and at that time it was
still clear that discourses of naturalisation are by no means complete and
meanings are still being argued, and their boundaries contested.
In Summary
I found four discourses of media representation across the three data sets:
1. Objective factuality.
2. Positioning.
3. Redeployment.
4. Naturalisation.
While there was a broad progression from each of these discourses to

the next over the course of the media coverage, nevertheless, there was
also evidence of overlap and interplay between them within each of the
data sets. The progression was not necessarily fully sequential or system-
atic, and I explore this observation in Chaps. 13 and 14. Meanwhile, the
fourth stage of analysis is to investigate how the findings from the Stage
3 depth analysis can be understood within single texts, and the following
chapter offers an example of this type of holistic analysis.
12
Stage 4: A Holistic Analysis of a Single
Text
The analyses I have just described have been relatively decontextualised,

and the final stage, which I have called a “holistic analysis stage”, serves to
reconnect some disparate observations with a complete text, as it would
be encountered by a reader. A micro-analytic overview of a single text
investigates the interplay of the selected features. This textual overview
can serve to show how signs (lexical choices), codes (systems) and mythic
meaning (connotations) are combined at different stages in the represen-
tation. The holistic analysis stage was particularly useful for the complex
process of identifying discourses, which, as I argue above, can be realised
through a wide range of language choices and strategies. Here is a text
from 2010, briefly analysed as an illustration of the first discourse of
representation: objective factuality. This is one example, but any number
of texts can be analysed in this way, to demonstrate either a specific or a
general analysis point.

in Discourse, DOI 10.1057/978-1-137-58750-3_12
Analysis of a Full BP Text

The text below is shown largely as presented in Nexis UK, with a couple
of small changes: an error in transcription of an apostrophe has been cor-
rected (“Iâ€[TM]m” appeared in Nexis instead of “I’m”), the font has been
changed to fit with the rest of this book and the highlighting of my search
terms using emboldened red type has been removed. As well as the com-
plete news report, the database also shows paratextual information, such as
the load date (28 April) as well as the original publication date (27 April),
the logo and name of the source (in this case the news agency Agence
France Presse [AFP]), the length of the piece (in this case 481 words), the
language (English) and the publication type (here, a newswire).
[Agence France Presse logo]

Agence France Presse—English
April 27, 2010 Tuesday 10:45 AM GMT
BP’s soaring profits overshadowed by oil rig tragedy
LENGTH: 481 words
DATELINE: LONDON, April 27 2010
British energy giant BP said Tuesday that first-quarter prof-
its rocketed on higher oil prices but admitted that the news was
overshadowed by last week’s tragic accident at a rig in the Gulf of
Mexico.
Europe’s biggest oil company said net profit soared 137 percent
to 6.08 billion dollars (4.5 billion euros) in the three months to
March compared with the same period in 2009.
Adjusted net profit on a replacement cost basis soared 135 per-
cent to 5.6 billion dollars.
The replacement cost figure, which excludes the effect of changes
in the value of oil and gas inventories, is closely watched by the
market and compared with analyst expectations for profits of 4.81
billion dollars.
Production in the three month period was little changed at 4.01
million barrels of oil equivalent per day. BP Chief Executive Tony
Hayward, in an email to staff, acknowledged that the strong results
12 Stage 4: A Holistic Analysis of a Single Text 193
were overshadowed by the “tragic accident” and continuing oil spill

from a BP well in the Gulf of Mexico.
The Deepwater Horizon oil rig, operated by BP and owned by
Transocean, sank last Thursday—two days after a massive explosion
left 11 workers missing and presumed dead.
“We are going to do everything we can—firstly, to control the
well; secondly, to ensure there is no serious environmental conse-
quence; and thirdly, to understand how this has occurred and ensure
that it never occurs again,” Hayward said in the email, obtained by
AFP.
He also expressed “tremendous sorrow” when it became clear
that the missing workers were probably dead.
“I’m sure, like me, you have all experienced a whole range of
emotions over the course of the last week,” he told BP staff.
“Shock and, indeed, anger that the accident could happen.
Tremendous sorrow when it became evident that the 11 people
missing had probably died in the initial explosion.
“And great sorrow and sympathy for the families and friends of
those who lost their lives.”
Hayward, who has been in the United States since late last week
because of the incident, added: “We have a great team in the Gulf
of Mexico leading this response.
“I have every confidence that we are doing everything in our
power to contain the environmental consequences of this incident.”
BP deployed robotic underwater vehicles on Monday to try to
cap the leaking well and prevent a growing oil slick from developing
into an environmental disaster.
Satellite images showed the slick had spread by 50 percent in a
day to cover an area of 600 square miles (1550 square kilometres),
although officials said some 97 percent of the pollution was just a
thin veneer on the sea’s surface.
The group has dispatched skimming vessels to mop up the oil.
Hayward said improved weather conditions were helping the
recovery effort.
“This, combined with the light, thin oil we are dealing with, has
further increased our confidence that we can tackle this spill off-
shore,” he said.
LOAD-DATE: April 28, 2010
LANGUAGE: ENGLISH
PUBLICATION-TYPE: Newswire
The fact that this is a newswire tells us that this piece was written for
onward distribution to news outlets, such as newspapers or radio or TV
news, who will pay organisations such as AFP for prewritten content that
they can slot, often more or less intact, into their coverage. The writer is
uncredited, which is often but not always the case in newswires; how-
ever, AFP goes on to assert its journalistic credentials in the following
way. The report partly concerns an email, and the writer uses the phrase
“the email, obtained by AFP”. This implies that the information con-
tained in the email, ostensibly at least, is not in the public domain like
a press release, for example. In fact, it is said to be an internal email to
employees, which suggests that AFP has a source inside the company. As
it happens, the quotations “obtained by AFP” are similar to expressions
of sorrow and sympathy expressed in publicly available press releases, and
indeed the final quotation “This, combined with the light thin oil…” is
taken directly from a BP press release of 26 April 2010. Why, then, use
the employee email as a source rather than the press releases if it is not
substantively different in content? It is possible that AFP wishes to “add
value” to its service by demonstrating that it does not simply copy and
paste sections from press releases, but carries out original investigative
journalism and seeks out new sources.
In terms of genre, this text is something of a hybrid. It combines a
financial report with a news report—this is evident in the headline where
the implication is that what is usually good news (“soaring profits”) will be
mitigated with other information. As noted earlier, these genres construct
factuality in two slightly different ways. The early sections on financial
information are typical of the genre financial reports, with substantial use
of numbers presented in different ways (percentages, dollars, euros) and
covering alternative key measures (net profit and replacement cost profit).
There is use of expert technical lexis (“adjusted net profit”, “oil and gas
inventories”) and spatial metaphor (“soared”, “rocketed”) which are typi-
cal of financial reports, but evaluative markers, for example, modality,
affective or judgemental lexis, are lacking.
The move to the second section of the text (beginning “BP Chief
Executive Tony Hayward…”) is a clear break. This is not a text that
blends two genres, but rather exploits two genres held loosely together
by the headline and the summarising first paragraph, which is a signal of
the inverted pyramid structure. The second section of the text deals with
the highly emotive subject of sympathy for the families and colleagues
of the victims of the explosion, but presented in a way that communi-
cates factuality and objectivity. All the emotive content is presented as
direct quotation and explicitly attributed to a source, in this case Tony
Hayward. The writer is careful not to interpose his/her own perspective
on what is being said, as shown in the following fragment, where readers
are intended to understand that “tremendous sorrow” is a direct quota-
tion for which the writer of the piece takes no responsibility:
He also expressed “tremendous sorrow” when it became clear that the miss-
ing workers were probably dead.
Otherwise the information presented directly by the journalist tends

to be restricted to descriptive narrative of events in time and space, for
example:
Hayward, who has been in the United States since late last week because of
the incident, added…
BP deployed robotic underwater vehicles on Monday to try to cap the
leaking well and prevent a growing oil slick from developing into an envi-
ronmental disaster.
As quotations are used selectively, the reader has no opportunity to piece

together the full text of Hayward’s internal email, but it is evident even
from these sections that Hayward is doing interpersonal work with his
own text. He prioritises the human aspect of the tragedy with his expres-
sion of sorrow and sympathy and he aligns himself explicitly with his
staff: “I’m sure, like me, you have all experienced a whole range of emo-
tions…”. Yet he also needs to demonstrate that he (as the face of BP—the
“we” in “we are going to do everything we can”) is in control of the situ-
ation. This is an interesting development from earlier in the week, where
BP press releases positioned BP as supportive of Transocean, but not as
leading the response:
BP today offered its full support to drilling contractor Transocean Ltd. and
its employees after fire caused Transocean’s semisubmersible drilling rig
Deepwater Horizon to be evacuated overnight, saying it stood ready to assist
in any way in responding to the incident. (BP press release, 21 April 2010)
In this text of 27 April, Hayward appears to have assumed control, as he

appends to his expressions of sorrow some strongly worded demonstra-
tions of confidence. He inserts the phrase “I have every confidence [that
we are doing everything in our power]” to emphasise the positive impres-
sion of his message and he expresses optimism in being able to tackle the
spill, with the weather now on their side. His use of “light thin oil” is
intended to address fears about the physical nature of the spill (compare
here my finding that metaphors about the oil slick in 2010 conversely
emphasise its uncontrollable nature). Finally, Hayward’s use of logical
structural relations in his “utterance” suggests a rational and coherent
strategy for dealing with the events at hand. Specifically, he uses a pre-
view–detail structure (Winter, 1994) in the following text quotation:
We are going to do everything we can—firstly, to control the well; secondly,

to ensure there is no serious environmental consequence; and thirdly, to
understand how this has occurred and ensure that it never occurs again.
So we have here a mixed genre text that is typical of the texts of that year,
in that news reports and financial reports between them account for 90
% of the texts mentioning the BP events on 27 April. The text shows
the different characteristics of its two parent genres; however, both have
in common that they are presented in such a way as to appear objective
and factual. Although part two of the text (typical of the news report
genre) gives clear indications of the interpersonal work Tony Hayward
is doing—namely, expressing sorrow, and reassuring the reader that he is

in control—this interpersonal work is distanced from the writer through
the consistent use of attribution—using quotation marks and words such
as “said”, “told” and “added”.
Features of the speech situation are noted overtly in the Nexis para-
textual framing, and alluded to in the phrase “obtained by AFP”. But
otherwise the text is presented to the reader in such a way as to suggest it
is authoritative, that it is the reading of the events, within its limitations,
rather than a reading of events.
Summary of Section II
Section II has focused on an analysis approach rooted in Barthes’ view
of how meaning is made in texts of all kinds. This section has offered the
following theoretical and practical support for written text analysis:
• A theoretical grounding, introducing a four-level heuristic for the

understanding of texts.
• A guide to data collection and research principles, based on the research
of the BP Deepwater Horizon story in the news media.
• A plan for a four-stage analysis of data:
– Stage 1: contextualisation of the BP texts.
– Stage 2: a preliminary analysis of the BP texts (immersion stage).
– Stage 3: a depth analysis of the data, based on the four-level
heuristic.
– Stage 4: a holistic analysis of a single text.
• Illustrative findings from the BP crisis coverage in the media for each
of the four stages above.
In Section III, I take these findings and make use of another semiotic
conceptualisation, drawn from the ideas of C.S. Peirce, to synthesise and
interpret the data.
Part III
A Peircean Conceptualisation of
Written Language
13
Theoretical Foundations
Aspects of Charles Sanders Peirce’s work

I have shown how discourse analysis based on a Barthesian concept of
language can generate a micro- and median-level analysis of the language
of representation of the BP crisis. This analysis generated a considerable
amount of data for interpretation at the level of discursive features. At a
broader level, however, sits the “language map”, and in order to capture
a sense of the full linguistic picture for each data set, I drew on some
aspects of the work of Charles Sanders Peirce on semiotic concepts.
Earlier in the twentieth century, but partly in parallel with the work of
de Saussure and Barthes in France, Charles Sanders Peirce was developing
alternative theories of the sign. He named his area of study “sem[e]iotics”,
which became a more widely used term than de Saussure’s “semiology”.
Peirce’s definition of what constitutes a sign is extensive. For Peirce, a
sign is “something which stands to somebody for something in some
respect or capacity” (1931–1958: 2.228). This definition appears to give
the scope for full linguistic representations or data sets to be regarded
as signs in themselves, much as de Saussure’s definition. Seeking to

in Discourse, DOI 10.1057/978-1-137-58750-3_13
encompass all possible instances of representation, Peirce eventually the-

orised over 59,000 sign types (Cobley & Jansz, 1999: 30). From this
extremely complex logic system, only a few of Peirce’s taxonomies are
regularly drawn upon in current scholarship, and I will discuss here two
key ideas which offered a useful explanatory perspective on the BP data.
The first is Peirce’s understanding of elements of a sign, and the second is
his taxonomy of three sign forms—Icon, Index and Symbol—which are
ways of expressing three relationships that the sign or representation has
with a real-world referent.
Object, Representamen, Interpretant

Unlike de Saussure’s dyadic view, Peirce’s view of the sign was triadic:
the element he acknowledged in addition to de Saussure’s conception
was that of some real-world referent, which he termed the Object. This
referent did not need to be material; it could also be concepts, theo-
ries or ideas (Peirce, 1931–1958). Further, Peirce did not presuppose an
external reality in which the Object was a fixed, unalterable entity. To
explain Peirce’s Object of the sign, Chandler (2007) uses an elegant anal-
ogy of the sign being a labelled box containing the Object—the point
being that the Object is only knowable via the sign, remaining, as it does,
hidden inside the box. In addition to the Object, Peirce considered that
the sign consisted of a Representamen, or formal sign (broadly equivalent
to de Saussure’s signifier), and the Interpretant (broadly equivalent to de
Saussure’s signified). Peirce (1931–1958: 2.228) explains the Interpretant
as follows: “[A sign] addresses somebody, that is, creates in the mind of
that person an equivalent sign, or perhaps a more developed sign. That
sign which it creates I call the interpretant of the first sign.” Cobley and
Jansz (1999: 23) explain this as “the sign in the mind that is the result
of an encounter with a sign”. The importance of this conception is that
the “sign in the mind” is then open to further interpretation, creating
a further Representamen and so on, in a process that Eco (1976: 69),
with reference to the work of Peirce, Barthes and Derrida, calls “unlim-
ited semiosis”, in other words, an endless and unfixable generation of
meanings.
To use the example of BP: the BP crisis is conceived as the Object,

or the “real-world” referent, the media representations are the sign or
Representamen which stand for or “mean” the Object, and the Interpretant
is the “sign in the mind” which is a result of our understanding of the
media representation. We could conceive that the initial Interpretant in
our collective mind from the media Representamen would be that of an
oil rig explosion. However, our developing conception of the crisis might
include the damage and destruction caused by the explosion in all its
aspects, which might also call to mind other explosions, or indeed crises
of different kinds, perhaps business crises, natural disasters and so on.
What does it mean to say that the crisis is the Object for this
Representamen? It should be clear already that I do not claim that there
is an objective reality of the events that the media reports are managing to
describe more or less accurately, although this is not to deny the material
consequences of loss of life, injury and destruction of the environment
caused by the explosion. Apart from the question of objective reality,
there is the question of scale and scope—what are the parameters of the
phenomenon I wish to investigate? Butchart (2011: 291) alludes to the
complexity of defining “events”.
In what does a happening consist in order for it to obtain the status of an

object of knowledge? Is there a difference between a set of related occur-
rences and the identity of their concept? Is what happens distinct from
what comes before it and everything else that codetermines it?
In the case of my research, a broad definition of the hidden Object is

an aggregate of the Deepwater Horizon explosion and oil spill, the reac-
tions of participants to it and the crisis that BP found itself dealing with
for many years. More specifically, it is the object[s] of discussion by the
media that I find when I consider the terms “BP”, “disaster”, “crisis” and
“oil spill”. This discussion surrounding the Object changes substantially,
and this both describes and constitutes the changing “meaning” of the
Object. In the way of unlimited semiosis it will become impossible to
discuss the BP oil spill without reference to the web of discussions that
have invested it with additional meaning, just as it is impossible to dis-
cuss, say, Nelson Mandela as “a man” without reference to the vast range
of images, films and writings that constitute our concept of him. I will go
on to hypothesise how the representation of events will develop beyond
the time of my data sets, drawing on the ideas of Baudrillard (1994).
In Baudrillard’s terms, representations at the present historical moment
have come to have no relation to any reality whatsoever, but are pure
simulacra. In my argument, I conceive that there is, in fact, an Object,
albeit hidden and unknowable, and we move away from this Object and
towards a simulacrum. In my research, I have endeavoured to trace the
linguistic evidence of this movement.
Icon, Index, Symbol

Relationship to the Object
Peirce proposed three forms of the sign related in different ways to the
Object, and labelled these Icon, Index and Symbol. Prototypically, an
Iconic sign is one that relates to its Object via a relationship of likeness
(Peirce, 1931–1958: 2.276). It represents the Object by looking like it, as
a portrait might its sitter, an engineering diagram its realised article or the
Underground map of London its system of train lines. These representa-
tions may be different kinds of likeness, but their form is recognisably
related to their Object, and all are labelled as Iconic by Peirce. Under the
heading of Iconic signs, Peirce also placed the metaphor, on the grounds
that metaphors posit a relationship of likeness between one entity and
another. Peirce’s classification of a metaphor as an Iconic sign was not
motivated by the fact that many metaphors are dead—that is they have
ceased to rely on unexpected pairings of entities. Rather, he saw it as a
logical outcome of the fact that targets resemble their sources in a rela-
tionship of likeness. Unfortunately, Pierce did not write extensively about
metaphor, and there is little further guidance or support for his decision
other than his mention of metaphor as one type of Icon. Other semioti-
cians have recognised that the disparity between the source and target
domains may require interpretation through codes and rules that suggest
that there the metaphor can also be Symbolic. Some scholars (e.g. Haley,
1995; Ponzio, 2010; Sørensen, 2011) have argued that both sign forms
are present in metaphor. Chandler (2001: np) writes:
The basis in resemblance suggests that metaphor involves the iconic mode.
However, to the extent that such a resemblance is oblique, we may think of
metaphor as symbolic. (Emphasis in original)
An Indexical sign is characterised as being in a relationship of contiguity

with its Object (Peirce, 1931–1958: 2.276) rather than a relationship of
resemblance, as is the case for Iconic signs. Peirce’s notion of Indexical
signs can be elusive to grasp, as this relationship of contiguity can be man-
ifest in a number of different ways (Eco, 1976; Grutman, 2010; Lock,
1997; Sørensen, 2011). Contiguity can take the form of cause-and-effect
relations, such as that between smoke and fire or footprints and the pres-
ence of a person. It can be realised in metonymic and synecdochic rela-
tions, where a part of an entity stands for the whole, or a single instance
can stand for an entire class, such as a sign showing a coffee cup with a
red line through where the coffee is a metonym for “all drinks”. Whole-
for-part relations are also synecdochic, and thus Indexical, shown in lan-
guage by expressions such as “BP announced”, where the organisation
here stands for a person/people within it. A sign can be Indexical when it
points to or indicates the presence of something else, so that arrows and
pub signs, for example, are Indexical. These prototypical instances are
relatively straightforward, but some signs are more difficult to locate, or
show complex relationships with their Object, for example, a Jaguar car
can be seen as both an Index and a Symbol of wealth. A photograph may
be regarded as an Icon because it shows a resemblance to its Object, or an
Index because it represents a point-to-point correspondence with reality
(Chandler, 2007: 38–39).
A Symbolic sign is one that is related to its Object only arbitrarily
and by convention (Peirce, 1931–1958: 2.249). It neither looks like the
Object, nor is it related diagrammatically, nor is it associated causally
or through part–whole relations. It is understood only through social
agreement; it has acquired meaning through the development of conven-
tional systems, which have to be learned to make sense to the receiver.
For Peirce, Symbols had meaning through rules and laws rather than
through instinct and observation. Words are prototypical Symbolic signs,
as they (generally) have no relation to their Object apart from that which
has been conventionally agreed. Peirce’s view largely accords with that
of de Saussure, which is that language is a set of arbitrary, rather than

motivated signs. Other Symbolic signs include mathematical symbols or
literary or artistic symbols (such as a lamb in a painting, intended to
represent innocence).
I describe the definitions above as prototypical because signs need not
be (indeed seldom are) pure versions of Icons, Indexes or Symbols. Most
Symbols have at root an Icon and/or Index—for example, red is consid-
ered a symbol of danger through convention, but this convention is likely
to be based on an association of red with blood, or fire, or the fact that
red is a highly visible colour. These associations are primarily in Indexical
relationship with the concept “danger”. Similarly, (alphabetically) written
signs, which are seen as arbitrary and Symbolic by Peirce’s definition, and
which de Saussure argued only have meaning in relation to each other,
are often originally based on systems of pictograms that have an Iconic
basis—for example, a tree or a snake (Singleton, 2000). A single sign
can be argued as Iconic, Indexical and Symbolic. If I pursue the example
of Jaguar, the bonnet ornament on Jaguar cars is Iconic in that it is a
direct likeness of the animal; however, we attach Indexical associations of
power, speed, sleekness and beauty, which we transfer from the animal
to the car itself. But we also understand the ornament as a Symbol of the
car through the cultural knowledge that the signifier “Jaguar” represents
not only the signified “animal” but also “a car manufacturer”. Further, we
might interpret the bonnet ornament as an Index of wealth, or indeed a
Symbol of wealth. As mentioned above, writers developing Peirce’s tri-
chotomy in the area of metaphor stress the interconnectivity of the three
sign forms in creating a metaphor (Abrams, 2002; Haley, 1995; Ponzio,
2010; Sørensen, 2011). As Merrell (2001: 37) writes:
Now, everything I have written in this section suggests that a sign can be in
varying degrees iconic, indexical, and symbolic, all at the same time. A
sign’s evincing one sign type does not preclude its manifesting some other
sign type as well. There are no all-or-nothing categories with respect to
signs. As one sign type is, another sign type can become, and what that sign
was may become of the nature of the first sign that the second sign now is.
Merrell illustrates that Peirce’s three sign modes are separate only in the-
ory: not only can a sign manifest more than one sign type, but also one
sign type can become another. This quality of mutability is potentially

significant for my diachronic study of representation.
Distance from the Object
In their prototypical instances, then, Peirce’s three sign modes have dif-
fering relationships with the unfixable Object. These relationships are to
different degrees either motivated by the Object or in arbitrary relation-
ship with it. So Iconic signs are said to be rather strongly motivated by
the Object, in that they somewhat resemble it. Indexical signs have a rela-
tionship with the Object through experience, but one which is less direct.
Symbolic signs have no experiential relationship with their Object. We
can therefore discuss sign forms in terms of their conceptual distance
from their Object.
My concept of the BP data depends on regarding Peirce’s modes
of sign: Icon, Index and Symbol, as progressively distant in this order
from the Object that they represent. In other words an Iconic sign
would be most “motivated” by the Object, an Indexical sign next and
a Symbolic sign least “motivated” by the Object. (I acknowledge the
complexity that all signs can properly be seen to have elements of each
of these modes but focus for the moment on archetypal signs). The
order outlined above is not one endorsed by all scholars. For exam-
ple, Chandler (2007: 35) suggests that Indexical signs are closer to
the Object than Iconic signs, citing Peirce as support, because they
“direct the attention to their objects by blind compulsion” (Peirce,
1931–1958: 2.306). However, elsewhere, Peirce seems to imply that
the Icon is more closely related to the Object than the Index, not least
because he orders them in this way in his discussions of the quali-
ties of phenomena in general. One of Peirce’s basic categorisations of
phenomena was into the domains of “Firstness”, “Secondness” and
“Thirdness”. “Firstness” relates to the quality of things, or their pos-
sibility. “Secondness” introduces the idea of relationships between
things. “Thirdness” is the area of laws, or the mental realm. In Peirce’s
theory, Icons belong to Firstness, Indexes to Secondness and Symbols
to Thirdness. The view that a cline from Icon to Index to Symbol
represents an increasingly arbitrary relationship with the Object is

certainly one held by a number of other semiotics scholars. Hodge and
Kress (1988: 26–27) suggest that Iconic signs are closer to the Object
as they involve “direct perception”, whereas Indexical signs are “based
on an act of judgement or inference” and are therefore lower in modal-
ity, that is, they have a lesser claim to realism or truthfulness. Writers
in the field of metaphor also endorse this ordering, and these include
Sørenson (2011), Ponzio (2010) and Haley (1995).
We can therefore conceive of the sign forms as distant from the Object
as shown in Fig. 13.1.
As I develop this conceptualisation of my data with specific reference
to Peirce’s terminology, it is worth relating it also to Baudrillard’s concep-
tion of representation, discussed in Chap. 2. In Baudrillard’s thinking,
representation has moved from some degree of connection with reality
(in Peirce’s terms, motivation by the Object) to the modern-day limited
or no connection with reality (in Peirce’s terms an arbitrary connection
to the Object through laws).
Icon, Index, Symbol and Verbal Language
Peirce’s three sign forms are intended to be comprehensive, that is, all phe-
nomena which are signs by Peirce’s broad definition should be able to be
understood as one or more of the sign forms. The sign forms have tended
to be used as explanatory concepts more for visual signs and iconography
of all kinds, than in the particular area of language. Nevertheless, there
are various ways in which language has been envisaged as Icon and Index,
as well as purely Symbolic.
Fig. 13.1 Distance of sign forms from the object

Language as a Symbolic System
In the terms of Peirce’s definitions of Iconic, Indexical and Symbolic

signs, language is considered to be a Symbolic system on the grounds
that its relation to its Object is arbitrary and unmotivated, a product of
conventional systems or rules that can vary according to culture. (This
observation is to be distinguished from the idea of symbolic language,
which is language that deviates from the naturalistic, using general and
literary symbols [Wales, 1989: 446].) This notion of arbitrariness is most
easily illustrated by the fact that different words are used in different
languages to denote the same referent. De Saussure, although he did not
use the word “symbol”, shows that meaning is made within a language
system only by reference to other elements within the same system—so
“cat” differs from “rat”, “Katze” from “Ratte”, “chat” from “rat”.
Everything that has been said up to this point boils down to this: in lan-
guage there are only differences…Whether we take the signified or the
signifier, language has neither ideas nor sounds that existed before the lin-
guistic system, but only conceptual and phonic differences that have issued
from the system. (de Saussure, 1959: 120)
Peirce wrote less extensively about language as a system than did de

Saussure, but he was clear that in terms of his sign forms it was Symbolic:
“All words, sentences, books and other conventional signs are symbols”
(Peirce, 1931–1958: 2.292). Yet the idea of the complete conventionality
of language has been challenged in certain aspects. Writers have argued
that there are ways in which language is not a completely arbitrary sys-
tem, and can be conceived as either Iconic or Indexical.
Iconicity in Language
In my earlier discussion of the multiple forms of signs under the Peircean sys-
tem, where Iconic origins can be detected within Symbolic representations,
I mentioned the case of written alphabetic systems that emerged from pic-
tograms that themselves were originally Iconic representations of real-world
referents. What appears now to be purely arbitrary—Symbolic—shapes such

as letters, may have at root an Iconic representation (Singleton, 2000: 93). As
these Iconic pictograms developed to become more stylised, partly because
of the practicalities of the materials used, and partly because of the desire to
express increasingly complex concepts, they moved in a process from Iconic
to Symbolic.
In spoken language, a further challenge to arbitrariness is the case
of onomatopoeia, where sounds are intended to represent their Object
through a relationship of aural likeness. As a retort to this challenge,
de Saussure (1959: 69) pointed to the wide variation in onomatopoeic
words within different languages, arguing that “they are chosen somewhat
arbitrarily, for they are only approximate and more or less conventional
imitations of certain sounds (cf. English bow-bow and French ouaoua)”.
Related to onomatopoeia in the field of literary linguistics is sound sym-
bolism, where “certain sounds or sound clusters are felt to ENACT or to
be in some way appropriate to the meanings expressed” (Wales, 1989:
426, emphasis in original). Sounds have been described metaphorically
as “lighter” or “darker”, and this can be used to create particular effects
beyond the meaning of the words themselves, particularly in poetry or
poetic prose. Rhyme can be said to work in a similar way. While these
poetic effects can be argued as both intended and understood, Simpson
(2004) points out that they are still contextually determined (e.g. we read
poetry with a heightened expectation of sound symbolism that we may
not bring to, say, the reading of a textbook). Cook (2001) points to the
graphological presentation of words so that they have an Iconic relation
to their intended meaning; he discusses this particularly in relation to
advertising but it is also evident in some poetry (“visual poetry”).
At a syntactic level, Iconicity has been argued in various ways (Givón,
1995; Nöth, 2001; Radwańska-Williams, 1994; Simone, 1995). One
of the most obvious manifestations of Iconicity in language is word
order, where sequential ordering suggests temporal unfolding of events.
“Veni, vidi, vici” (“I came, I saw, I conquered”) is an example (Cobley &
Jansz, 1999: 145). Jakobson argues (Caton, 1987: 237) that the form of
words often relates directly to meaning in a relationship of likeness. His
examples are of Russian verbs, where tense aspects indicating expanded
meaning are almost always longer than tense aspects indicating restricted
meaning. Similar claims are made in English for, for example, related
words such as “big”, “bigger”, “biggest”, where the longest form indi-
cates the most expansive concept. Similarly, Iconicity has been claimed
in the use of grammatical subordination and focus, where the location of
information can indicate its importance. The examples of Iconicity given
above show that the connection between Peirce’s category and language
and linguistic representation has been researched primarily in terms of
phonological, morphological and structural features. That is, researchers
have focused on the form rather than the function of language. Language
is held to be iconic if, in its form, it somehow “look[s] like the things it
stands for” (Simone, 1995: vii).
Indexicality in Language
There are other perspectives from which language has been considered
Indexical, rather than Symbolic. Jakobson proposes that deictics are in
Indexical relation to the speaker, because they have a “pointing function”
indicating time, place, person or specificity, all of which can change with
the context of the utterance (Caton, 1987). Jakobson refers to such expres-
sions as “I”, “you”, “here”, “that” as “shifters”, following Jespersen (1922).
I have discussed at length another linguistic realisation of Indexicality,
namely, that of metonym. As Wales (1989: 297) writes: “In semiotic
terms, metonymy is an indexical sign: there is a directly or logically con-
tiguous relationship between the substituted word and its referent.”
Scholarly work concerning the Indexicality of language (and some of
the work on Iconicity) does not deny the fundamental conception of
language as primarily Symbolic and arbitrary. Rather, writers are using
Indexicality (and Iconicity) as an analogy to certain types of discursive
function. So, for example, using the metonym “red tape” and declaring
it to be in Indexical relationship with a concept labelled “bureaucracy”
does not deny that the words “red” and “tape” are Symbols. The Indexical
qualities sit at a different level from the words which are still part of a
Symbolic system. Peirce’s Icon–Index–Symbol trichotomy has been more
widely applied to visual than linguistic signs, but it also offers a perspective
from which to look at verbal language. Peirce’s aim for his logic system was
to offer a comprehensive understanding of all types and modes of sign, of
which language was just one.
Understanding a Set of Texts as a “Sign”—The

Peircean Perspective
I have already introduced the idea that one way of conceiving extended
media representations (e.g. a set of texts from 27 April 2010 concerning
the BP crisis) is as one discrete sign. This is not to deny that it is made up
of myriad smaller signs, and it is this recognition that led me to use the
metaphor of a “language map”—a single recognisable landscape made up
of a number of distinctive features. I do not suggest that, say, 169 texts
from 27 April 2010 are any more definitive a representation of the BP
crisis at that time than another set of texts from another day. However, I
do suggest that there is value in examining an agglomeration of represen-
tations about an entity as a sign in itself, and Peirce’s extensive definition
of a sign cited at the start of this chapter would allow for such a view.
In summary, I propose that “signs” can refer to entire verbal represen-
tations of an entity, so a set of texts can be seen as a “sign” just as much as
a single word, although the process for analysing large and disparate signs
is more complex than that for analysing word- or phrase-level signs. In
investigating the BP media texts, I was interested in how representations
of a business crisis might change over time. This suggested examining the
sets of texts (“signs” in my broad sense) in order to explore whether and
how language choices differed between them. The procedures of this text
examination have been explicated in detail—features of language have
been identified and analysed at all four Barthesian levels of meaning.
What the Peircean perspective of analysis permits is the broad envision-
ing of these sets of features as a sign.
At the most uncomplicated level, English-speaking people with access
to news media on 27 April 2010 would have had access to a particular
configuration of meaning about the BP crisis from the press. I describe
this as a certain type of sign, in a Peircean sense. If we understand the
representation of the crisis as a semiotic sign, which Peircean sign modes
come to the fore with the passage of time? Does this way of looking at the
linguistic representation of phenomena shed light on how shared cultural
understanding is constructed in the case of a crisis?
14
A Peircean Interpretation of the BP Data
From Data Analysis to “Language Maps”

If we accept that sets of texts can behave as single representations, or
signs, then it is reasonable to attempt to describe these signs from a lin-
guistic perspective. I have suggested that it is possible to draw up a lin-
guistic “map” of different representations. Comparing these “maps” gives
an overview of what kind of sign is, jointly, being constructed through
written texts.
The initial research output from the Barthesian analysis was a number
of observations about the nine selected features at the three time points
represented by the data sets. A summary of the findings in Table 14.1
gives a starting point for broader analysis at the representational sign level
(what I have called the “Peircean conceptualisation”). Viewing the find-
ings in tabular form makes clear that they have possible meanings both
synchronically, for example, what does the totality of the 2010 language
features “add up to” in terms of a 2010 sign?, and diachronically, for
example, what changes in modal choices do we see across the three time
points? It is not possible fully to separate these synchronic and diachronic

in Discourse, DOI 10.1057/978-1-137-58750-3_14
214
Table 14.1 Summary of findings from depth analysis

2010 2011 2012
Naming of events Low numbers of names, Increase in number of names, Instances of naming fall: 80 %
but wide range. Primarily but smaller range. Increase include just two head terms.
neutral in negative shading Increase in length of descriptor.
Increase in negative shading
Naming of people Focus on direct Numbers remain similar, but Continued focus on commentators,
participants—senior BP increase in commentators also writers, other artists,
Semiotics and Verbal Texts
personnel, relief workers/ and decrease in direct fictional or hypothetical

agents, affected public participants characters. Wider range of actors,
less directly related
Categorisation Little categorisation, usually More instances of grouping Level as 2011, but disparate
relating events to other and categorisation. Primarily groups, related less often to
oil spills business-related categories business/oil, more to wider social
phenomena
Genre Primarily news report genre Fewer news reports, increase 16 % of texts now arts-related,
in financial and market and more evaluative articles and
reports blogs, although news reports still
evident
Intertextuality Key intertexts are press Fewer press releases and Layered intertexts:
releases, direct quotation witness quotations, documentaries < news
increased reference to reports < direct quotation.
BP-related documents/artistic Indirectly related texts
works
Modality Apparent high level of Decrease in modal expressions Decrease in total modal
modality, in particular overall. Specifically, decrease expressions; decrease in
epistemic and dynamic in epistemic and dynamic but epistemic/dynamic, increase in
increase in deontic modality deontic. High use of appraisal
resources
Metonymy Significant use of Decrease in metonymic Further decrease in metonymic
metonymy, primarily expressions, due to reduction expression
ORGANISATION FOR in ORGANISATION FOR
MEMBERS MEMBERS constructions
14
Metaphor Metaphorical use relates Increase in metaphorical Further increase in metaphor.

to physical entities—oil, usages overall, particularly Change in business metaphors
weather financial and business towards “journey” and so on.
spheres Increase in metaphors of theatre,
stories and myth
Discourses “Objective factuality”: “Positioning”: move towards “Redeployment”: artistic reference
representations of events, “business-as-usual”: events and resistant voices.
constructed through facts/ placed in context of BP “Naturalisation”: representation
figures, witness quotation global activities and “big of events “fixed” though
and so on picture” for US oil linguistic mechanisms; broad and
Symbolic “ownership” of events
A Peircean Interpretation of the BP Data
215
readings. For example, we can only confidently suggest that there is a

relatively low level of modal expressions in the 2012 “map”, by compari-
son with the relatively higher level in 2010. My first set of interpretations
of the findings abstracts patterns and processes through the data—from
left to right of Table 14.1—drawing from all nine examined features. This
initial work makes it possible to propose descriptors for the three years of
signs, reading from top to bottom of the table. We can finally return to
the diachronic perspective of development across the table, speculating
on what the changes in Peircean sign modes can suggest about how we
make social meaning of catastrophic events through written texts.
Patterns and Processes
Table 14.1 sets out very brief summaries of the findings about each lan-
guage features studied for each of the data sets 2010, 2011 and 2012.
These shifting behaviours of a number of linguistic features—both singly
and in aggregate—suggest a number of patterns across the data.
Five possible patterns are:
1. Shorthand. There is an increase in linguistic devices that enable both

writer and reader to access the BP events in ways that simplify their
multifarious nature.
2. Spread. This is discernible in the type of text in which the events are
mentioned, the intertextual environment and the social actors and
contexts introduced in the texts.
3. Categorisation. This pattern concerns the placing of the BP events in
lists and groups, and shows the events placed in increasingly disparate
contexts.
4. Art. Texts reveal that BP events are increasingly captured, explored or
fixed in artworks.
5. Discursive shifts. Discourses of representation appear to move from the
ostensibly factual towards discourses of naturalisation.
14 A Peircean Interpretation of the BP Data 217
Shorthand
In using the term “Shorthand” I wish to convey both a restriction

on meaning conveyed by linguistic terms and the idea of presupposi-
tion (Fowler, 1991; Zare, Abbaspour, & Nia, 2012), or “taken-for-
grantedness” (Martin & White, 2005). A number of the linguistic
strategies analysed have the effect of limiting and directing perspec-
tives, whether consciously or unconsciously, on the part of the writer.
For example, both of the tropes metonym and metaphor select certain
aspects of the entity described for attention, and disregard other aspects,
encouraging readers to see the entities from the perspective indicated
by the writer. Processes of naming both people and events presuppose
some agreement about how we should define, view and ultimately
behave towards them. Similarly, the groups and categories in which
we place people and events help us to make sense of them in a context
proposed by the writer, and later to attach meaning through processes
of collocation and association. This latter phenomenon is discussed in
a separate section below, and is one of the ways in which we collec-
tively attribute shared meaning through a “shorthanding” process. We
are encouraged to locate new concepts and ideas within existing known
dominant and resistant discourses, which notion is explored in more
detail below. Ideas can be presented as either “given” or “available for
discussion” through modal resources, particularly those relating to cer-
tainty and those relating to obligation or instruction, constituting fur-
ther ways in which potential meaning is restricted. Without repeating
my analysis findings in detail here, I will draw on them to illustrate how
shorthand entails presupposition in the case of the BP data.
Metonymy
The use of metonym decreases over the life of the data, and its shorthand
role is more evident at the start of the data period than the end. At the start,
metonyms that position BP as animate (BP “says”, “confirms”, “tries” and
so on) were noted in my analysis to have two effects that are potentially
in tension. On the one hand, metonyms work to humanise the company

by suggesting that it can speak and act as though it were a human being,
and, on the other, they remove agency and responsibility from individu-
als, by suggesting that it is the corporation, rather than people, performing
actions. The positioning conveyed by the frequent metonymic use of “BP”
seems to be that BP as a corporation is closely involved in all aspects of the
salvage and clean-up operation, having an active and above all consistent role
in the financial, technological and environmental spheres affected by the
spill. This would fit with the guidelines on crisis management outlined in
business communications literature (Cornelissen & Harris, 2001; Massey,
2001; Muralidharan, Dillistone, & Shin, 2011) and reinforced in two texts
from 2011 and 2012 that offer crisis consultancy advice. Later metonymic
usages show a focus on reports and other written materials as actor (such as
“the Coast Guard report slams”, 2011), positioning official and written texts
as agreed and definitive versions of the story.
Metaphor
Metaphors are used throughout the data sets to position the events in dif-
ferent ways. A number of metaphors in the early data (2010) present the
events—that is, the explosion, oil spill and the weather conditions that
affected operations—in ways that foreground that these phenomena were
out of control, or difficult to control. Metaphors used in the later data sets
position events and their subsequent business outcomes as part of a jour-
ney (BUSINESS IS A JOURNEY), a process, or a story (MYTHS AND
LEGENDS). This is a feature of the process of symbol creation, as the events
move towards a somewhat-agreed meaning, perhaps that events are “part of
life’s/business’s rich pattern”, but it also serves to background those elements
of the events that are troubling or fragmented, and were early on metapho-
rised as being unmanageable.
Naming of Events
My analysis of the naming of the events showed that qualifying terms of

time and place increased in number and complexity, the range of head terms
reduced, and that “oil spill” and “disaster” became the most commonly used
over time. This reduction in the variety of terms suggests a degree of conver-
gence on the part of the media, while still indicating two slightly different
evaluative stances. The word “disaster” already encodes negative shading,
and the term “oil spill”, in itself arguably neutral, is generally accompanied
by qualifying terms that evaluate the events negatively.
Of note is the use of the terms “accident” and “incident” in the 2010
texts, which appear to have been the preferred labels in the early BP press
releases. “Accident” implies no, or unclear cause, and “incident” implies
a neutral stance towards events. Both are absent in later texts, suggesting
that the widespread practice will be either a negative head (disaster) or a
neutral head (oil spill) + negative qualifiers. These naming practices do
not at all attempt to downplay the seriousness of the explosion and oil
spill, although there is a reduction over time in richness of signification,
whereby a complex noun phrase serves to replace a longer retelling of the
story, and in this way selects and restricts available meanings. The con-
ventional use by news media of certain negative formulations can serve to
make these terms familiar yet “other”, and in that way “safe”.
Modality and Certainty
In considering how aspects of modality relating to (un)certainty would

progress through the data sets, I hypothesised that there would be a move
from certainty to uncertainty, characterised by an increase in modal items.
I imagined an analysis of the early data would show a limited use of modal
expressions, as the genre of news reports is dominant in 2010, offering an
account of observable facts, and constructing an impression of objective
reality rather than personal opinion. By 2012, I hypothesised, an increase
in evaluation and speculation about the meaning of the events would sug-
gest an increase in modal expressions, with journalists increasingly intrud-
ing into their writing. Instead, I uncovered the reverse trend, namely, that
modal indicators reduced substantially between 2010 and 2012. I propose
that this change has a number of causes. The first is the initially high level
of uncertainty about the “reality” of the situation. Crises of all kinds can
be characterised at the outset by chaos and communication failures. The
Deepwater Horizon explosion in particular took place in uniquely diffi-
cult circumstances (depth of water, previously untried drilling techniques)
and any facts given to journalists were unclear and often unsubstantiated.
This uncertainty is reflected in the 2010 texts in the high levels of epis-
temic and dynamic modality occurring particularly in reported speech. This
is not a feature of later texts where a move away from uncertainty towards
apparent certainty is realised by a reduction in modality (although journal-
ists and other writers still make some use of modal resources, particularly
deontic modality, through which they suggest what should be done, or how
the reader should understand their assessment of the BP situation). The
resources of modality are one mechanism through which propositions are
presented as either “given” or “available for discussion”, and the later texts
are increasingly characterised by bare assertions of “opinion as fact”. Martin
and White (2005: 100) argue that monoglossic formulations (bare asser-
tions) demonstrate “taken-for-grantedness” in that they “make no reference
to other voices and viewpoints”. When modal items are absent, as is increas-
ingly the case over time in the BP data, it is strongly indicative that the possi-
bilities for “other voices and viewpoints” are being closed down. Martin and
White also observe that opinions about propositions are typically expressed
through epistemic modality, while opinions about events are attitudinal
and expressed through Appraisal resources. As the concept of Deepwater
Horizon is reified, it may become more of an “event” and therefore the
object of the lexis of judgement rather than the resources of modality.
Spread
Spread, or diffusion, relates to the observation that accounts of the events

appear in increasingly disparate text genres, comprise an increasing range
of contexts, and mention an increasingly wide circle of social actors.
These first two overall themes, shorthand and spread, may appear to be
in some tension, as the first is concerned with reduction, simplification
and selection, while the second is concerned with expansion, diffusion
and heterogeneity. I suggest there is a distinction between the events and
their context. It appears to be the case that the events themselves—in
de Saussure’s terms, the signified, in Peirce’s terms, the Interpretant—are
undergoing changes in representation that in some ways make them
simpler and more taken-for-granted. On the other hand the contexts in
which they appear and the entities with which they are associated are
multiplied and dispersed. Simply put, the events have a more singular
sense in a far greater range of applications. Spread is noted across a range
of language features, but primarily genre, intertexts and social actors.
Genre and Intertextuality
As shown in analysis, the early texts in my data set belong more firmly to
the genre of news report whereas the later texts are attached to a greater
range of genres, including commentary, features, reviews and business jour-
nal articles. This process alone accounts for much of the linguistic varia-
tion noted in my analysis, as different genres are characterised by different
purpose, structure and content. The news report genre tends to draw on
particular types of external text, including the eyewitness report, the inter-
view or the photograph, whereas the multiple later genres draw on a wider
range of external texts, for example, previous news reports on BP, associated
but not directly related news stories, artistic texts and evaluative articles, as
well as more generalised beliefs, issues and statements that remain unat-
tributed (Allen, 2011; Bazerman, 2004). As the original observable story
is absorbed, naturalised, tamed, consumed, through a range of linguistic
strategies, so the web of unseen texts with influence on my texts can be
imagined to become ever more extensive, and ever less tangible.
Social Actors
At a more detailed level, I observed the range of social actors to become

increasingly wide and less directly connected to the source events. From a
relatively restricted pool of experts, local people, company employees and
agencies in 2010, the metaphorical “cast of characters” associated with the
events becomes larger, more distant and widespread, ranging from previ-
ous chairpersons of BP and their associates to politicians outside the USA
and unconnected individuals who have gone through quite different crises.
More notable still is the introduction of fictional and hypothetical charac-
ters, including those appearing in books that cover the events, and imag-
ined future businesspeople dealing with as yet unrealised crises.
Categorisation
Both naming and grouping (which are separate phenomena) are acts with
consequences in terms of how entities are perceived and thus treated.
Categorisation, as I define it, comprises processes of listing and grouping by
which an entity (here the BP events) is characterised through its contiguous
relationship with other entities. This kind of grouping is fluid over time:
the entity, by being placed adjacent to other entities, both absorbs some-
thing of their character and represents something of its own within the
group. There is something in the process resembling metaphor, whereby
a particular feature of one entity (but not all) is selected as resembling a
feature of another entity (but not all). The process is also one of a met-
onymic nature, where an aspect of the entity comes to represent or point to
a wider phenomenon, just as the OED word of 2013 “selfie” might come
to index a society with an obsession with individual self-representation, at
the expense of unrecorded activity, or group activity. Cook (2004: 109)
construes this process as ultimately one of a metaphorical nature.
The physical nature of a new phenomenon, in other words, does not con-
stitute the entirety of its role in contemporaneous discourse. It is appropri-
ated and amplified, but also further understood, as a metaphor. It then
both draws upon, and contributes to, understanding of other arenas.
I see the concept of categorisation as representing the fluid process of

representation where an entity is not part of a stable category, but moves
between groups, absorbing and leaving behind traces, such that its own
nature becomes more and more a product of social and cultural agree-
ment. This concept is a commonplace in semiotic theory, but what I am
suggesting here is that we can see it happening through linguistic analy-
sis. In 2010, this feature of categorisation is barely present, indicating
an entity, or signified, that is represented as being itself, singular, with
unique characteristics. By 2011, the entity is represented as being part
of known groups, such as business crises, or environmental catastrophes.
By 2012, in a further process of “symbolisation”, it is sometimes located
in very different groupings that have no direct connection with business
crises or environmental disasters, for instance, to represent anti-social
behaviour, or as one of several themes in a literary work.
Art
The role of artistic representation has been treated so far under a discussion
of genre, but I want to identify it here as an important feature in itself. Like
categorisation, this feature was virtually absent in the first year of data, pre-
sumably for the understandable reason that at a time of immediate crisis,
writers have neither the time for reflection nor yet the insight with which to
create art, nor might it seem an appropriate first reaction. Examples of texts
I include under the broad heading of artistic representation are:
1. Reviews of books/films/documentaries about the events.

2. Reviews of books/films/documentaries that touch on the events, but
that are not primarily “about” them.
3. Reports of commentary about the events in artistic form (e.g. specially
written protest song).
Outside my own data sets, treatment of the events in artistic form has
been quite widespread. Many non-fiction books have been written solely or
partly about the BP oil spill (e.g. Bergin, 2011; Burt, 2012), oblique comic
reference was made in the sketch show “Ruddy Hell! It’s Harry and Paul!”
(2012) and US animation South Park satirised Tony Hayward’s apology
to victims (Coon 2: Hindsight, 2010). These treatments serve to fix or pin
down versions of the events: in the case of comic treatments, usually a ver-
sion that is subversive or resistant. On the one hand, the increase in refer-
ences to such texts within my data set is surely significant—journalists are
acknowledging that these texts and treatments are now an intrinsic part of
how the crisis is understood. On the other hand, the nature of my primary
texts—news reports, reviews and so on—means that “artistic” construc-
tions of the crisis appear to sit outside mainstream representation. These
potentially resistant versions tend to be “bracketed”—they are reported
on but do not form the substance of the report. This is a natural conse-
quence of choosing mass-media journalism as an object of study (rather
than, say, artistic representations themselves). However, my analysis shows
that in some instances other, dominant, representations of the crisis are pre-
sented as given, while artistic representation is clearly presented as “other”.
I develop this argument further in relation to the ideological implications
of the discourses I identify below.
Discursive Shifts
An account of discourses is of a somewhat different order from the linguis-

tic patterns and characteristics above, although it draws upon all of them.
My interest in this book lies in discourses of representation, although there
is much evidence in my data set that reveals discourses about oil explora-
tion, the environment, US nationalism, risk and reward and other social
concepts. To recap, a discourse is a set of shared ideas that appear natural
to a particular time and place, but that are, nonetheless, a product of, and
specific to, that time and place, and are in that sense conventional. These
ideas are taken for granted although not permanently fixed: dominant
discourses are challenged, and regularly deposed, by alternative resistant
discourses. However, it is through “taken-for-granted” ideas that power
interests are served. Individuals and institutions with power (including
the media) are able to influence what counts as knowledge, what reaches
the public domain and how it is positioned. In this way certain ideologies
are embedded in texts and practices and their ubiquity and appearance of
“common sense” allows them to go unremarked. Partly because of this,
identifying discourses is not a straightforward process. Discourses can be
identified through both what is said (content) and how it is said (form).
The significance of an analysis of discourses to this book is that it situates
the micro-analysis of text discussed earlier within an account of broader
social beliefs and practices. The research in this book does not concern itself
primarily with “what happened” after the BP explosion. Indeed, in relation
to Peirce’s hidden Object, I hold that we cannot understand “what hap-
pened” other than through language. My concern is rather with how “what
happened” was represented and how that representation evolved over time.
This is best understood within the context of the discourses of mass-media
representation.
From the earliest stages of the analysis, the representations in the three
data sets seemed to me to be of rather different kinds with different aims
and outcomes, and fuller, more detailed analysis confirmed this view. Put
very broadly indeed, it seemed that the language of the texts at the point
of crisis in 2010 had the aim primarily to describe the events, that language
usages in 2011 seemed to be concerned to place the incident, while by
2012, I was observing processes of two kinds—one by which the events
were explored through creative expression (including creative categorisa-

tion and artistic works) and the other by which events were presented as
understood or normalised. I took the presence of these two processes in
2012 as indicating that one phase (creative expression) was in flow and
the other (normalisation) was emerging. In making these observations, I
do not wish to imply either that the sets of texts were entirely discrete or
that these aims were necessarily concerted or conscious. They were partly
an outcome of journalistic practices by which developing news stories
move from the front pages to the middle pages of newspapers (or from
the home page to sub-pages of the website) and eventually become source
material for editorials and features.
In identifying discourses of representation, however, I was continu-
ally challenged by the apparent co-existence and conflict between the
“opening up” and the “closing down” of meaning. Processes that seemed
to allow for the “opening up” of meaning—that is, those that allowed
for discussion, challenge, alternatives and uncertainty—included the
modality of uncertainty, the use and reporting of voices that contested
the status quo and the depiction of the events through art. These
processes were evident, in different forms, through all three years of
data. “Closing down” processes that served to restrict and fix meaning
included the modality of certainty, the bracketing or “othering” of con-
testing voices and the location of the events within the known. These
were more present in the second two years of data, but importantly
were observable alongside “opening up” processes, and even preceded
them. The challenge for interpretation is not that both sets of processes
are observable, but more that they did not follow a consistent pattern,
for example, an opening up followed by a closing down of meaning.
One interpretation of this mixed picture is that there is a continued
contestation of meaning, a back-and-forth or opening up and closing
down between questioning and reassertion of meaning, which ends up
ultimately with some form of social agreement. For another, I return to
an observation I made earlier which is the distinction between the rei-
fication of the Object (the BP events) through “shorthanding” and the
fragmentation of its deployment as a concept through “spread”. What
is “closed down” is our understanding of the Object; what is “opened
up” are the situations in which it has resonance. In other words, it is
only when we feel we have a grasp of what the Object is that we can
deploy it as a resource in a number of social practices, including artistic
representation and other symbolic activities.
Following this line of argument, and taking into account that there is
constant interplay between these “opening up” and “closing down” strate-
gies, I do not propose that any one discourse correlates precisely to any
one data set, but rather that there are four discourses at play at various
times through the data sets, and hypothetically beyond, that can and do
co-exist, but that demonstrate a certain progression towards a naturalised
meaning for the BP events.
1. A discourse of objective factuality, in which events are described and

presented as news, and as a set of complex, but factual phenomena.
2. A discourse of positioning of the events, in which the events themselves
are located, associated and categorised with other events. At this stage,
they can also themselves become an exemplar of a category (oil spills,
business disasters, major lawsuits and so on). Their meaning becomes
placed within the known.
3. A discourse of redeployment in which this (partly) agreed version of
events becomes a resource for exploration and creative expression.
4. A discourse of naturalisation in which exploration is concluded, and
meaning is more or less fixed, less complex and more routine.
I have written in some detail in Chap. 11 about how each of these

discourses is realised through diverse discursive means. I have com-
mented that these discourses can be complex rather than straightfor-
ward, showing, for example, how the discourse of objectivity can be
given credibility rather than undermined by the (limited) presence of
uncertainty, disfluency and imprecision. I have also emphasised that
the emergence of these discourses of representation is far from linear or
inevitable—how “objective factuality” remains a feature of some sto-
ries in 2012, while some naturalisation, in the form of marginalisation
of the BP story in financial reports, is already evident in 2010. That
said, these four discourses were present in varying degrees in the data,
becoming prominent in broadly the order shown above.
Representations as Sign Modes

Referring back to summary Table 14.1, the five diachronic patterns described
above present patterns which can be discerned over the three time points of
the data set by reading the table from left to right and drawing in elements
from each of the nine individual linguistic features. I return now to my
notion of “linguistic maps” to describe the data at three single time points.
Is there something about the 2010 data in aggregate that characterises it,
and distinguishes it from the 2011 data, and again from the 2012 data? To
consider this question, we can read the table from top to bottom for a syn-
chronic view. In particular, the findings just described on the progression of
discourses are useful aids to interpretation.
Returning to Peirce’s classification of sign modes, I propose that there is
a way of conceptualising whole representations as sign modes, along a spec-
trum which runs from Iconic (strongly motivated by its Object) to Symbolic
(connected to its Object only by social agreement). This analogy encourages
us to explore the form that social agreement takes about a phenomenon,
and from this to suggest conclusions about how we, jointly, make meaning.
My question asks whether it is justified and productive to understand the
representation of the BP crisis as a semiotic sign, with different orders of the
sign (Icon, Index or Symbol) being more significant at different times as the
representation progresses. In my discussion of discursive shifts I have already
raised some of the points that will contribute to this argument—issues con-
cerning media representation of events ostensibly as they are, and ultimately
as what they have come to represent as a result of cultural agreement.
I would like to reconfigure this argument by conceptualising the repre-
sentations in the three data sets as three meaningful Signs, in the Peircean
sense discussed in Chap. 13. In the particular trichotomy Icon–Index–
Symbol, the critical relationship is that of the Sign (in this case the mass-
media representation) to the Object (in this case the BP oil spill). I have
observed that this is broadly different in each year: in 2010 the rela-
tionship is one of resemblance, in 2011 it is one of association, whereas
in 2012 it depends to a far greater extent on cultural knowledge and
agreement. It is this observation that leads me to suggest that these three
phases of semiotic representation might be conceptualised, respectively,
as Iconic, Indexical and Symbolic.
Argument for an “Iconic Phase”
In Peircean terms, the Iconic sign is one that is connected to its Object
via a relationship of likeness. I am suggesting that this relationship is evi-
dent in my data in the sense that the drive for linguistic choice is towards
producing a representation that looks like the Object of the BP events.
All the markers of “objectivity” and “factuality” that we have noted in the
analysis chapters, and that I have referred to in my discussion of a dis-
course of objective factuality, are pressed into service to express something
like “this is what it would look like if you were there, on the ground,
moment to moment, with all its uncertainties, emotions, facts, half-facts
and unwitting players”. In this view, we see not only evidence of a form
of “scientific objectivity” through the devices of highly specific naming,
expert witness and hard facts and figures. We also see an acknowledge-
ment of uncertainty and fragmentation in the use of epistemic modality,
the (reported) emotive language of non-expert witnesses and the insta-
bility of names, descriptors and categories. This uncertainty is, however,
mediated by the writer or journalist through devices such as direct and
indirect quotation and the presentation of contrasting views. In this way
he/she is selecting and combining information to make meaning, but the
meaning made is not intended to be a social understanding of the impli-
cations of the events, but rather a marshalling of the available informa-
tion to create the most credible picture for the reader of what is actually
happening, in other words a picture that looks most like the Object. The
picture is not the Object, but one selective version of it. As van Leeuwen
points out (2005: 160, emphasis in original), “Linguists and semioti-
cians therefore do not ask ‘How true is this?’ but ‘As how true is it repre-
sented?’” Fowler (1991: 170) makes a similar point about the apparent
representation of likeness by news reports:
As a working principle in discourse analysis or critical linguistics, we

assume that the ostensible subject of representation in discourse is not what
it is “really about”: in semiotic terms, the signified is in turn the signifier of
another, implicit but culturally recognizable meaning.
By using the term “culturally recognizable”, Fowler indicates that news

reports (and other texts) are Symbolic artefacts, being only understand-
able within the time and place that produced them. When I suggest that
certain types of news writing are Iconic in nature, I do not lose sight of
the fact that they are constructed to be so. There is a potential tension in
arguing that news reports propose a direct likeness but acknowledging
that any likeness is highly modified according to situated practice and
stakeholder interests. If this version of reality is a construct, then why is
it not a Symbol, like other culturally agreed-upon signs? I suggest that
the argument for Iconicity holds from two perspectives. Firstly, Peirce
never argued that an Iconic sign was in any way the Object, only that
its connection to the Object was one of apparent resemblance. And,
as illustrated in Chandler’s analogy of the Object hidden in the box,
the Object remains unknowable. Yet, crucially, and this is my second
point, the texts in 2010 are presented as resembling reality—this is their
explicit and implicit purpose. The texts are concerned to assemble data
of very different kinds into a multifaceted facsimile of an event that the
reader is meant to recognise as reality. In short, the writer has created
an Iconic sign.
Argument for an “Indexical Phase”
Early in my analysis it seemed clear to me that the language of the first phase
of data intended to represent a reproduction of reality, while the language of
the final phase of data was evaluative, highly intertextualised and connected
to artistic representation, as I shall go on to discuss in the next section. It
seemed appropriate to characterise these beginning and end states as display-
ing features of Iconic and Symbolic signs, while the middle phase could quite
well have been simply transitional: a point in the process of movement from
one state to the other, rather than a phase in its own right. The year 2011 is,
in regard to most of the features studied, a linguistic midpoint for increas-
ing or decreasing movements from the first to the last data sets. This might
indeed be the sole source of interest in the 2011 data, were it not for the
importance of the particular researched phenomenon, categorisation, which
I propose as evidence for an Indexical phase of representation.
Peirce’s Indexical signs are those based in relationship of contiguity

rather than a relationship of resemblance. This association can take the
form of cause-and-effect relations, part-for-whole and whole-for part
relations and a pointing or indicating function. Fowler (1991: 170) dis-
cusses media representations as entire Indexical signs, in the sense that
I am arguing here, where references to events become “a shorthand, a
metonymy, for an underlying ‘it’ of a more abstract kind”. Fowler’s point
here coincides with my own observations from the analysis of shorthand-
ing processes as well as the gradual move to use the BP events as an illus-
tration of wider phenomena.
Two questions of interest for the BP data arise from these definitions of
Indexicality. Firstly, since metonym is a linguistic realisation of Indexicality,
is there evidence from the BP data of a particular pattern in the use of
metonym in the 2011 texts? Secondly, are metonymic (Indexical) rela-
tions expressed in any other way? In other words is there evidence that
the BP events are somehow connected to, represent or stand for other events
or phenomena? In the first case, an investigation of metonymy as a purely
linguistic phenomenon did not reveal 2011 to be a special case. Instances
of metonymy in 2010 were predominantly of the kind mentioned above:
ORGANISATION FOR MEMBERS, by which BP and other organisa-
tion names were used to stand for particular spokespersons. This type of
usage decreased steadily over the years with the reduction in the preva-
lence of company statements, and the trend in 2011 was part of that
pattern. (The role of metonymy in selecting features of interest of the crisis
remained crucial.) However, there did prove to be an important connec-
tion with Indexicality in respect of the second question above.
Categorisation
The year 2011 showed a marked increase in a particular type of categorisa-

tion. This was the practice of placing BP within a particular category either
explicitly, by including it in an enumerated list, or implicitly, through
expressions such as “the biggest”, “the worst” and so on. The instances of
this type of usage increased substantially from 6 examples in 2010 (all ten-
tative or mitigated, as the analysis shows) to 16 (in only 20 texts) in 2011.
The groups or categories in which the events were placed were generally
predictable, including natural disasters, man-made disasters, business crises
and specific BP problems. Nevertheless, it was clear that the BP events were
no longer being represented as isolated, unique or irretrievably “messy”.
Rather, they were being used in a process of interactive meaning-making
that aligned them with other pre-existing or newly made groups. In this
way, either the events took on some of the qualities of group members or
other group members took on some of their qualities. If group members
are subsequently placed in different categories, they bring with them some
of the associations of their previous group membership. I will pursue this
line of thinking in my discussion of the Symbolic phase of 2012, when the
BP groups proved to be far more creative and disparate than in 2011.
BP Events as Index
The BP events are introduced as an Index of a number of other concepts in

2011. The data set includes texts from professionals offering crisis advice
services, whose interest was to hold up BP as a salient warning to those
who did not have crisis management protocols in place (BP events as cau-
tionary exemplar of a very serious business crisis). Other texts were writ-
ten from the perspective of the effect on other deep-water prospects (BP
events as obstacle). Others focused on BP’s safety practices (Deepwater
Horizon as Index of deep-rooted weaknesses in the BP culture). In these
and other texts, the BP events either stand for or point towards something
outside themselves. This can be seen as part of a process of meaning-
making whereby the phenomenon moves from a situation where it can
be described but not yet understood, to one (the Indexical phase) where
it is starting to be located within the known.
Argument for a “Symbolic Phase”
Peirce’s concept of the Symbolic sign is one that is related to its Object only
arbitrarily and by convention. Put another way, it is not, or only minimally,
motivated by its Object. I argue in this section that the way the BP events
are represented in the 2012 data set is increasingly a construct of convention
and agreement. Description of the events has given way to evaluation, and
the crisis is increasingly positioned as having meaning within a number of
social fields, including business and finance, the environment, politics and
the arts. By this I mean that the events are now a meaningful concept, or
signified with a largely agreed social meaning, that depends decreasingly on
the material events of 20 April 2010. Butchart (2011) would argue that that
the enterprise of describing crisis events is bound to fail, and for this reason
we turn to symbolising them. In contrast to the Indexical phase, indications
of the Symbolic nature of the 2012 representation can be found across a
range of linguistic dimensions, and these are set out below.
Symbolic Genres
From the contextualisation of the texts (Stage 1 of the analysis), we note

an increase in news text genres that sit outside the sphere of traditional
news reports. These include feature articles and online blogs, as well as
text types such as reviews that provide a link to art, music, film and docu-
mentary. I would argue that this generic shift is one of the mechanisms by
which representations of the BP events move from reporting and describ-
ing to evaluating and appraising. By 2012 the BP events are being drawn
upon as a cultural resource for making meaning. The genres that appear
in the 2012 data set are decreasingly those that purport to describe and
increasingly those that purport to comment and interpret. Such com-
mentary and interpretation is set in a context that arguably requires
increasingly sophisticated world knowledge and cultural agreement.
This idea of cultural resource is interestingly supported by an increased
number of literary and shared cultural allusions (e.g. to quotations from
Shakespeare) that are a feature of the 2012 texts.
So the move from Icon to Index to Symbol is reflected in the features of
the dominant genres of the years in question. According to this view, news
reports make use of primarily iconic characteristics of representation (they
describe), editorials use indexical features (they position, explain, draw
comparisons) and arts reviews and feature articles have symbolic features
of representation (they illustrate, integrate, draw on cultural references).
Thus the shift in genres goes hand-in-hand with changing sign forms.
Layering of Intertexts
Alongside this generic shift is a significant change in the type of source

text that is drawn upon as resources for the researched texts. I have
already noted a process of layering of intertexts which is evident by 2012,
for example, (a) a review of (b) a documentary based on (c) news reports
that in turn draw upon (d) press releases or eyewitness statements. From
this observed process of the layering of textual resources, we can postulate
that social and conventional meanings and understandings are constantly
being added to, amending and potentially obscuring the original Iconic
representation until it becomes a product of agreement and convention.
Symbolic Actors
Individuals and groups mentioned in the 2010 reporting of BP events

are closely attached to the events, from BP executives, (over-)specifically
named experts and public participants. Journalists take care to connect
these actors closely with the events: their importance lies in their capac-
ity to illustrate and validate the story. Their value lies precisely in the fact
that they are “real”. By 2012, the actors mentioned are considerably more
loosely attached to the story. Interestingly, an increasing number are “not
real”. These include the fictional characters in related artworks, and hypo-
thetical characters in scenario-building exercises such as those described
in crisis management manuals. An examination of actors reveals one way
in which the story is increasingly less anchored in the real world and
increasingly inhabits a Symbolic world.
Changing Metaphors
I outlined in the literature review chapter that Peirce (1931–1958: 2.277)

regards metaphor as an instance of an Iconic sign, relying as it does on a
relationship of likeness to its Object. Writers on metaphor from a Peircean
perspective (Abrams, 2002; Hiraga, 1994; Lattmann, 2012; Ponzio, 2010;
Sørensen, 2011) have problematised this view on the grounds of a greatly
reduced “immediacy of sign-object link” (Hiraga, 1994: 7). In the BP
data, I did not find that instances of metaphor decreased over time, which
might have neatly supported my hypothesis of a movement away from
Iconic representation. Instead, I observed an increase, and I make two
observations in this respect, one relating to the frequency of metaphor and
the other to the type of metaphor. With regard to frequency, I would like
to draw upon a different model of metaphor than that of Peirce, namely,
that of Jakobson (2002). Jakobson proposes that the “metaphoric order”
relies on choices that are paradigmatic, that is, they are concerned with
substitution and selection. In contrast, the “metonymic order” is syntag-
matic, relying on combination and contiguity. Each “order” or “pole” is
closely connected with different types of writing. Jakobson connects the
“metonymic order” with journalism, montage and ordered narratives such
as epic works. He suggests the “metaphoric order” is related to poetry,
romanticism, filmic metaphor and surrealism. I suggest that the observed
increase in metaphor relates to the move outlined above from a journalis-
tic to an artistic sensibility.
My second observation concerns the shift in both types of target and
types of source in the metaphor in my data sets. In 2010, the target for
metaphor is the material entities such as weather and the oil spill itself.
This tends to run counter to the pattern that the abstract is conceptual-
ised in terms of the concrete (Koller, 2003; Koller & Davidson, 2008; Ng
& Koller, 2013). By 2012, the target domain is more abstract, and source
domains also remain relatively abstract. By this stage, source domains
for metaphor have shifted to include those of theatre, stories, myth and
journeys. I wish only to touch lightly on this observation, as not all meta-
phorical usages are directly related to the BP events, some being a feature
of co-text. I do, however, suggest that there is a symbolic process of myth
and story creation relating to the BP events, which is realised at a number
of linguistic levels, of which metaphor is just one.
Categorisation
While categorisation was a key feature of the Indexical phase outlined

above, and occurs just as frequently in 2012 as 2011, there is an impor-
tant change in the nature of the groups in which the BP events were
placed. I have characterised 2011 lists and groups as “expected”, com-

prising events associated with business, finance and the environment. In
2012, group members are much less closely related to the oil spill itself,
and I see this as an indication that the concept “BP oil spill” has mean-
ings that are increasingly less dependent on the material events and more
related to our social construction of their meaning.
In fact, the linguistic choices described above all have in common that
they rely upon a shared understanding of what the BP events “mean” at
this stage of representation. The construction of meaning that has taken
place at the Iconic (descriptive) and Indexical (positioning) phases have
now reached a stage of relative arbitrariness, where full understanding
of the sign is only possible if we know the socio-cultural “code”. If we
accept this argument, the relevant order of the sign at this stage, follow-
ing Peirce, is the Symbol.
Process, State and the Blurring of Boundaries

The argument outlined above describes discrete and subsequent phases of
crisis representation, each with its own set of characteristics, which align
it with one of Peirce’s orders of the sign. This can only be a convenience:
the multifarious nature of my data resists such neat categorisation. Not
only does one proposed phase seem to move into the next gradually, but
elements of each phase are recognisable in each of the others. I noted
similarly that discourses co-existed and blended across the data sets. I
cannot even forcefully argue (to use a term from cognitive semantics, see
Singleton, 2000: 77–80) that each set of data is prototypically representa-
tive of the phases I propose, in other words, that each is an archetypal
example of Iconic, Indexical or Symbolic representation. The selection of
yearly spaced data sets was a device to allow me to explore the process of
representation and how it changed. It would be more than a coincidence
if I managed to alight on the day when, say, Iconicity, or Indexicality was
most prototypically represented in the texts. To take the Indexical phase
as an example, there may have been many more distinctive features of
what I call Indexicality on 4 November 2010 or 14 September 2011,
than there were on 27 April 2011.
Apart from the issue of prototypical representation, there is the issue

of overlap between the different phases. The data show that there are ele-
ments that I later describe as evidence of Symbolic representation in both
Iconic and Indexical phases, just as there are elements of both Iconicity and
Indexicality in the Symbolic phase. Those texts in 2012 that are of the news
report genre and cover the progress of the compensation cases or specu-
late on BP’s upcoming financial results include many of the features that I
have suggested are typical of making an Iconic representation, or descrip-
tive likeness (although I would argue that the Iconic news report changes
over time to accommodate Indexical and Symbolic features). To take an
example outside my data set, on 27 October 2010 the animated television
programme “South Park” aired an episode in which “Tony Hayward” apol-
ogised to the victims of the spill in a series of increasingly bizarre settings
(Coon 2: hindsight, 2010). This text is definitively representative of what
I term the Symbolic stage of representation, being artistic, evaluative and
culturally situated, but comes to wide public consciousness just six months
after the explosion, well before the date of my Indexical data set of 2011
and 18 months before my proposed Symbolic data set.
I do not present these variations as fatal flaws or even as serious prob-
lems for my argument: any synchronic linguistic description is partial
and imperfect, representing as it does a single, arbitrary point in a flow of
continual change. However, the discussion above does expose an interest-
ing challenge to conceptualising these representational phases. My first
view was that the three representational phases followed a linear pro-
gression from one point to another in time, represented by the diagram
in Fig. 14.1, albeit not entirely discrete as shown. In this view, once a
representation has reached the Indexical phase, it has left the Iconic stage
behind and so on.
Fig. 14.1 A linear view of iconic, indexical and symbolic phases

This thinking is unsatisfactory in two ways. One is that, as suggested

above, not all texts in the primarily Symbolic phase have purely Symbolic
features. Indeed, I argue that the Iconic phase is not one that is descrip-
tive likeness, but represents itself as descriptive likeness, and is in this way
Symbolic itself. Because of this, I find it difficult to argue for a complete
moving on from one phase to another. Secondly, the linear view implies
that the representation moves forward metaphorically to a greater and
greater distance from the Object. While this is to some extent what I
wish to convey, I will go on to reason in the following section that beyond
the Symbolic phase lies something that presents itself as a different kind
of likeness and so is in some sense closer to an Icon, which I shall call the
“naturalised Icon”.
Beyond Symbol to Simulacrum—The

Naturalised Iconic Phase
Beyond the three sign types I have discussed, we move away from the
evidence of the data to some degree of speculation concerning future
representation. I proposed a set of three synchronic readings, suggest-
ing that there are identifiable phases in the mass-media representation
of the BP Deepwater Horizon events (so far) that share characteristics
with Peirce’s three modes of sign—Icon, Index and Symbol. I sug-
gested that these are not discrete phases, but have blurred boundaries.
Despite the “messy” nature of the representations, I identified a ten-
dency for Deepwater Horizon accounts to move from Iconic in nature to
Symbolic in a diachronic process, and noted that this process is realised
through changes in the features of language used over the period of the
data. However, I raised again as a problem the simple linear view of
IconÎndex^Symbol^something else on several grounds. These included
the blurred time boundaries between the phases and the closely bound
nature of the sign modes. But my primary concern with a linear concep-
tualisation is that it implies a vector with a direction, and I suspect that
any direction will become less and less definable. This current section
of my discussion explores the issue of “what next”, and to address this
question, I turn to Baudrillard’s view of the “image” or representation.
As described in Chap. 5, Baudrillard identifies four successive phases of

the image (1994: 6) from signs that offer a close reflection of reality to
signs that refer only to other signs and have no relation to reality at all.
It would be convenient to suggest that my arguments map progres-
sively on to Baudrillard’s phases of the image, and I have certainly
explored how representations start by aiming for a resemblance to real-
ity and move towards an arbitrary relationship with reality. However, to
draw this parallel would be to misrepresent Baudrillard’s argument. In
fact, all three of the stages I outline, as well as my putative fourth stage,
belong to Baudrillard’s fourth order of the image. The fact that I have
argued strongly that my Iconic stage involves not factuality or objectivity
but what we might call “doing objectivity”, coupled with Baudrillard’s
argument that we belong to an age where it is no longer possible to attach
signs to a reality but only to other signs, can only mean that all three rep-
resentations are pure simulacra in Baudrillard’s terms; however, I choose
to sub-divide or classify them. A further divergence from Baudrillard’s
reasoning is one I have raised earlier in this chapter. My vision is one of
an increasingly elusive and distorted Object, but Baudrillard’s is one of
no Object at all.
Baudrillard’s proposed stages of the image imply a linear progression
through cultural epochs where representations become increasingly dis-
tant from what they purport to represent, such that the territory rots
away and the map of the territory is all that remains. His argument that
“theme park America” is somehow more “real” than the America outside
the theme park gates seems to refer to the fact that the less a representa-
tion is transparently a representation, the further it is from the Object.
This raises an interesting question for my own data. Is there a sense in
which, according to Baudrillardian logic, my Iconic data phase, by con-
structing what it intends to be understood as a “reflection of profound
reality” (Baudrillard, 1994: 6), is somehow further from the Object than
my Symbolic phase, that perhaps aims at a more fundamental authen-
ticity through art, critique and transparently expressed opinion—all of
which are openly acknowledged as being at some remove from reality?
How cultural representations progress has been a topic for academic
speculation. Chandler (2007) suggests that the move from Iconic to
Symbolic is typical. Parmentier (2009: 145), in a discussion of Peirce’s
trichotomies, sets out the case that although socio-cultural phenomena

are frequently held to develop in a way that exhibits increasing abstrac-
tion (he cites the example of money, whose form has evolved from direct
barter through gold coins and paper money to financial derivatives),
there exists an argument in the opposite direction. Sociolinguists Irvine
and Gal (2000) describe a process that they call “iconization”, by which
linguistic features that are actually indexes of a culture become seen as
natural (iconic) rather than culturally derived. This recalls Barthes’ dis-
cussion of “naturalisation”, which process is at its most complete when it
is least noticeable. As Parmentier sums up (2009: 145):
There are even situations where it seems that conventionalizers and natu-
ralizers engage in a direct semiotic confrontation—the bottom line perhaps
being that anyone who manipulates or regiments the flow of interpretants
thereby indexes social power or cultural capital.
Using the terms “conventionalizers” and “naturalizers”, Parmentier illus-

trates the two arguments—one that social phenomena progress to levels
of greater and greater abstraction, and the other that they become so taken
for granted that they are perceived as real and are unrecognisable as social
constructs. I do not see these two arguments as mutually exclusive, or even
contradictory, if one takes the view that naturalised social phenomena are
the most arbitrary signs of all. However, the argument expressed by the
“naturalizers” does introduce the possibility that distance from the Object
could be reconceptualised by imagining that an initial representation which
is recognised as Symbolic becomes increasingly seen as “the truth” (Barthes’
“what-goes-without-saying”) and in that sense more and more Iconic.
My preferred resolution of this tension is to conceive of these two
potential understandings of the term Icon as differing from each other,
as follows. The Peircean Iconic representation would be understood as
being at the early stage of high motivation by the Object and intended
to resemble the Object—however, flawed, culturally regulated and politi-
cally motivated that representation might be. The projected ultimate
form of the representational sign might therefore be termed the “natu-
ralised icon”. This is a representation that has been through the phases
of Icon, Index and Symbol to reach a different form of Iconicity that is
now so taken for granted that it is unrecognisable as a construct. It is this

concept that I wish to propose as my final stage—this is my proposed
answer to “what next?”
Pursuing this line of thought—we no longer have a linear progression to
an end point, but what might be conceived as a spiral, where representation
loops from Iconic back to Iconic status, but where the naturalised Icon is
on a different plane. This thinking still locates the representations within
Baudrillard’s final stage of “pure simulacrum”, but suggests that we can iden-
tify different processes within that understanding, and these follow a loosely
predictable pattern. This thinking is embedded in Barthes’ (1972) concepts
of the naturalisation of myths and ideologies. By the end of the period of the
data, there are signs that the “BP disaster” or the “BP oil spill” is part of the
common language, and it unites meanings from previous crises, meanings it
acquired at the time of the spill and an array of direct and indirect associa-
tions acquired since, in Peirce’s infinite series of signs. Its represented mean-
ing is now a “given”, at the same time both highly simplified and infinitely
fragmented. As the representation of oil spill reaches this stage, it becomes
a resource for our understanding of other oil spills and disasters. A trawl of
texts published on precisely the same day, 27 April, in 2013 (exactly one
year after the last data set), uncovers this text extract.
His work has immersed him in events that read like a roster of recent catas-
trophes, from 9/11 to the Gulf oil spill. Now, Kenneth Feinberg is adding
the Boston Marathon bombings to that list. (Associated Press, 27.4.2013)
This reference to the BP events is some evidence that our future under-
standing of the signified “catastrophe” in the first line will be informed
by the complex signified “Gulf oil spill”. I accept, following Barthes, that
the representation of the BP oil spill and its aftermath constructs versions
of reality that are more or less compliant with current discourses of repre-
sentation. However, I suggest that to reach a final state of naturalisation,
the representation undergoes a shape-shifting process, in which the sign
is variously constructed through language to resemble its Object, posi-
tion its Object, then test the boundaries of that position, finally present-
ing the Object in a new kind of likeness. This evolving relation of the
sign (media representation) to its Object (the BP events) is realised in a
complex network of interrelated language choices.
Part IV
Concluding Thoughts
15
Other Events, Other Contexts
In this book I have outlined a new approach to analysing verbal writ-

ten language in order to provide a template for other researchers aim-
ing to describe the linguistic characteristics of large written data sets.
The approach I describe is grounded in semiotic theory, but flexible to
respond to the emergent findings from specific data. It is adapted to make
use of language analysis tools according to need. It is critically aware but
not agenda-driven.
The particular interest of this book series “Postdisciplinary Studies in
Discourse” is in the connections between theories of language and their
application in practice. In my particular case study, the focus has moved
from theory to practice and back to theory in the following sense. Initially
the work drew primarily on the Barthesian view that language makes
meaning at a number of semiotic levels, and I suggested that an investiga-
tion which pays attention to these multiple levels can provide a holistic
account of a textual representation. De Saussure’s (1959: 128) concept of
“organized masses that are themselves signs” sits alongside the notion of
Peirce’s sign forms to suggest that a data set (in this case taken at a par-
ticular time point, but not necessarily so) can be seen as a representation

in Discourse, DOI 10.1057/978-1-137-58750-3_15
or “sign” with its own characteristics. These semiotic theories formed the
basis for the language analysis work. Four stages of analysis served to
contextualise the data, identify language objects for study at four semi-
otic levels, investigate the objects using diverse research tools and explore
how these features work together in a single text. The conclusions of this
project were based on the observation that the three data sets were analo-
gous to Peirce’s three main sign forms, Icon, Index and Symbol. What
seems to be useful in this conceptualisation is the notion of the proximity
and distance of each of these sign forms from the Object, whereby the
media representation relates to the BP events in changing ways over time,
finally returning to an apparent, but fully symbolic, proximity, in spiral
fashion. If this process can be shown to hold for other written repre-
sentations, then this Peircean conceptualisation provides an explanatory
theory which captures how meaning is created, positioned, played with,
naturalised and absorbed.
Wider Application of Semiotic Discourse

Analysis
The coverage of the BP events in the news media was a productive case
for investigation: much media attention was paid to the story over a
number of years and the outcomes of the explosion were significant, with
not only business but also environmental, social, political and financial
repercussions. The data offered evidence for examining our relationship
with a catastrophic event over time. However, this is just one form of
project. The concept of “language maps” proposes that any given instance
of communication will operate at each of the levels of sign, code, myth
and ideology, and that an analysis of the interplay of features at these
levels will provide a discursive picture of the data set.
This broad principle can be applied to modes other than language,
media other than the printed or online press and contexts other than news
media. Data sets could differ across dimensions including time (as for this
project), genre, geographical area and channel. Drawing up comparative
“maps” which differ across a chosen dimension gives an overview of what
kinds of large representative signs are being constructed.
15 Other Events, Other Contexts 245
To expand on how the approach could be applied, we can consider

what shape other research projects might take.
Example 1: Comparative Study of Policy Documents

by Political Party
In the study proposed in Table 15.1, as in the BP study, a single mode

(writing) is investigated, as well as a homogenous medium (political
documents, for example, manifestos). A similar approach could also be
used for political speeches, where the mode is that particular kind of
“written-like” speech that is characteristic of the genre. In this case, the
research is synchronic. The purpose of this particular example would be
to explore the linguistic constructions of a political issue, according to
parties engaged in the debate. Language features might include, as for BP,
the naming of the issue or of items related to it, modal constructions and
discourses. Rhetorical figures of different kinds or patterns of argumenta-
tion might prove interesting for study. As was the case for the BP data,
preliminary analysis would suggest the relevant discursive features.
Example 2: Comparative Study of Policy Documents

Over Time
In a similar study (Table 15.2), the emphasis is diachronic, investigating

how a single party has dealt with the issue of, say, Europe over time. This
design has something in common with the time-based approach to the BP
data. In this example, a timeline of external events which might affect atti-
tudes to Europe would be useful information to accompany the analysis.
Table 15.1 A comparative study by political party

Political issue, e.g. Britain’s newrole in Europe Party A Party B Party C
Language feature 1
Language feature 2
Language feature 3
Language feature n
Table 15.2 A comparative study of a political issue over time

Party A Party A Party A
Political issue, e.g. Britain’s Election year Election year Election year
role in Europe 2005 2010 2015
Language feature 1
Language feature 2
Language feature 3
Language feature n
Table 15.3 A comparative study by mediums

The construction of an issue e.g. child Magazine Printed
obesity or domestic violence (e.g. Bella) advertising Blog
Language feature 1
Language feature 2
Language feature 3
Language feature n
Example 3: Study of a Consumer Issue in Different

Written Mediums
In this study (Table 15.3), the medium of the communication is the focus
of attention. Analysis would seek to uncover which language choices were
typical of different mediums, using a particular topic as an illustration.
In this example, the mode is writing, but the mediums are very different,
and genre, purpose and audience will play a significant role in discussing
the verbal representation of a particular issue.
Example 4: A Different Crisis
The brief examples above addressed types of research design. A final exam-
ple touches on more specific aspects of language in the case of a different
crisis. On 1 October 2015, after a mass shooting in Oregon, President
Barack Obama made an emotional speech about gun control. Although
this is an instance of the spoken, rather than written, word, political
speeches of this kind, as mentioned above, are a type of communication
which Crystal (2003: 292) terms “writing to be read aloud”, in that they
are usually carefully pre-written, and generally read from autocue. They
15 Other Events, Other Contexts 247
frequently exhibit many of the features of writing discussed in this book.

A transcript of part of Obama’s speech follows:
Somehow this has become routine. The reporting is routine. My response

here at this podium ends up being routine. The conversation in the after-
math of it: we become numb to this. We talked about this after Columbine
and Blacksburg, after Tucson, after Newtown, after Aurora, after Charleston.
It cannot be this easy for somebody who wants to inflict harm on other
people to get his or her hands on a gun. And what’s become routine of
course is the response of those who oppose any kind of common-sense gun
legislation. (Obama, 2015)
A number of observations can be made about this brief extract which are
pertinent to the themes of this book. In just these few lines, Obama uses
several terms relating to discourse about the tragic events: “reporting”,
“response” (twice), “conversation” and “talked about this”. Obama’s
choice of words (or “signs”) emphasises the role of the discursive con-
struction of crises and other catastrophic events in how we anticipate,
manage, control and respond to such events. In a study of this kind of
text, one of the discursive features at the level of the sign generated by the
immersive Stage 2 of my approach might well be word choices relating to
discourse and communication.
Repeating the word “routine” in a tricolon construction (“Somehow
this has become routine. The reporting is routine. My response here at
this podium ends up being routine.”), Obama challenges a discourse
of the normalisation of mass shootings. For his part, he is attempting
to establish an alternative, resistant discourse about gun control, in
opposition to other discourses about “the right to bear arms” which are
widespread in the USA. One of the ways he presents this alternative
discourse as natural is by using the terms “of course” and “common-
sense” in relation to his own argument. If this speech were part of a
researched data set, an identification of discourses would be an impor-
tant area for investigation, where the discourses might be about disputed
notions regarding gun ownership rather than discourses of representa-
tion (as in the case of the BP research). As part of my immersion phase
of analysis, I would note from this excerpt rhetorical features at the level
of “code” (structuring figures such as the tricolon mentioned above)
and at the level of mythic meaning (rhetorical tropes such as metonym:

here we see a significant use of PLACE FOR EVENT in “Columbine”,
“Blacksburg” and so on). The use of modality is a further potential area
for analysis: in this fragment Obama’s words are largely unmodalised
declaratives expressing certainty, with only the modal auxiliary “cannot”
being used as a deontic (meaning “must not”) and having the force of a
directive, a pattern I have described as “linguistic certainty” and which
would warrant examination. I would also note the appearance and role
of categorisation (“after Columbine and Blacksburg, after Tucson, after
Newtown, after Aurora, after Charleston”).
Final Words
At the end of this book what I hope to have achieved through the detailed
analysis of the BP crisis is a demonstration of how to gather and investi-
gate language data in order to build a linguistic picture of a crisis represen-
tation. If we understand that the meaning of communication is located
in several places—in the intention of the producer, in the text itself and
in the understanding of the reader—then a key focus of my approach
has been on meanings located in the texts themselves. However, the texts
contain within them aspects of the socio-cultural world from which they
were generated and to which they are addressed. The study of the social,
textual and generic codes by which the signs are organised, the texts to
which they refer, their socially embedded connotations and the ideologies
they serve all give us access to their situated meanings, albeit partially and
imperfectly. Our shared conception of crises such as Deepwater Horizon
is a construct of (amongst other things) news media writing of different
kinds, whose language is shaped by the interests of a large number of
groups and individuals. The end stage of this representation—a natu-
ralised iconicity—may be the only way we can collectively cope with cri-
sis events of this magnitude.
References
Abrams, J. J. (2002). Philosophy after the mirror of nature: Rorty, Dewey, and
Peirce on pragmatism and metaphor. Metaphor and Symbol, 17(3), 227–242.
Allen, G. (2011). Intertextuality. Abingdon: Routledge.
Angermuller, J. (2011). From the many voices to the subject positions in anti-
globalization discourse: Enunciative pragmatics and the polyphonic organi-
zation of subjectivity. Journal of Pragmatics, 43, 2992–3000.
Angermuller, J. (2014). Poststructuralist discourse analysis: subjectivity in enuncia-
tive pragmatics. Basingstoke: Palgrave Macmillan.
Androutsopoulos, J. (2006). Introduction: Sociolinguistics and computer-
mediated communication. Journal of Sociolinguistics, 10(4), 419–438.
Androutsopoulos, J. (2008). Potentials and limitations of discourse-centred
online ethnography. Language@Internet, 5.
Arpan, L. M. (2002). When in Rome? The effects of spokesperson ethnicity on
audience evaluation of crisis communication. International Journal of Business
Communication, 39(3), 314–339.
Bakhtin, M. (1981). The dialogic imagination (C. Emerson & M. Holquist,
Trans.). Austin: University of Texas Press.
Bakhtin, M. (1984). The problem of Dostoevsky’s poetics (C. Emerson, Trans.).
Minneapolis: University of Minnesota Press.

in Discourse, DOI 10.1057/978-1-137-58750-3
250 References
Barthes, R. (1972). Mythologies (A. Lavers, Trans.). London: Vintage.

Barthes, R. (1974). S/Z (R. Miller, Trans.). Oxford: Blackwell.
Barthes, R. (1977a). Image, music, text (S. Heath, Trans.). London: Fontana.
Barthes, R. (1977b). Textual analysis of a tale by Edgar Poe. Poe Studies - Old
Series, 10(1), 1–12.
Baudrillard, J. (1994). Simulacra and simulation (S. Faria Glaser, Trans.). Ann
Arbor: University of Michigan Press.
Baxter, J. (2002). Competing discourses in the classroom: A post-structuralist
discourse analysis of girls’ and boys’ speech in public contexts. Discourse &
Society, 13(6), 827–842.
Baxter, J. (2008). Is it all tough talking at the top? A post-structuralist analysis
of the construction of gendered speaker identities of British business leaders
within interview narratives. Gender and Language, 2(2), 197–222.
Baxter, J. (2010). Discourse-analytic approaches to text and talk. In L. Litosseliti
(Ed.), Research methods in linguistics (pp. 117–137). London: Continuum.
Baxter, J. (2014). Language, gender and leadership: As seen on TV [inaugural
lecture, Aston University]. Retrieved March 7, 2016, from http://www.aston.
ac.uk/about/news/events/aston-inaugurals-2013-14/aston-inaugural-
2013-14-judith-baxter/
Bazerman, C. (2004). Intertextuality: How texts rely on other texts. In
C. Bazerman & P. Prior (Eds.), What writing does and how it does it: An intro-
duction to analyzing texts and textual practices (pp. 83–96). Mahwah, NJ:
Lawrence Erlbaum.
BBC. (2014). Advergames “target children with unhealthy products.” Retrieved
September 1, 2014, from http://www.bbc.co.uk/news/uk-27647445
Bednarek, M. (2006). Epistemological positioning and evidentiality in English
news discourse: A text-driven approach. Text & Talk, 26(6), 635–660.
Beghtol, C. (2001). The concept of genre and its characteristics. Bulletin of the
American Society for Information Science and Technology, 27(2), 1–5. Retrieved
September 1, 2014, from http://www.asis.org/Bulletin/Dec-01/beghtol.html
Bergin, T. (2011). Spills and spin: The inside story of BP. London: Random House.
Bhatia, V. K. (1993). Analysing genre: Language use in professional settings.
London: Longman.
Bhatia, V. K. (2002). Applied genre analysis : A multi-perspective model. Ibérica,
4, 3–19.
Bhatia, V. K. (2004). Worlds of written discourse: A genre-based view. London:
Continuum.
Biber, D., Johansson, S., Leech, G., Conrad, S., & Finegan, E. (1999). Longman
grammar of spoken and written English. Harlow: Longman.
References
251
Bignell, J. (2002). Media semiotics: An introduction (2nd ed.). Manchester:

Manchester University Press.
Bird, S., & Dardenne, R. (1988). Myth, chronicle and story: Exploring the nar-
rative qualities of news. In J. Carey (Ed.), Media, myths and narratives:
Television and the press (pp. 67–87). Newbury Park, CA: Sage.
Blommaert, J. (2005). Discourse: A critical introduction. Cambridge: Cambridge
University Press.
Borges, J. L. (1999). Collected fictions (A. Hurley, Trans.). London: Penguin.
BP. (2010a). BP initiates response to Gulf of Mexico oil spill. Press release 22nd
April 2010. Retrieved March 28, 2016, from http://www.bp.com/en/global/
corporate/press/press-releases/bp-initiates-response-to-gulf-of-mexico-oil-
spill.html
BP. (2010b). BP forges ahead with Gulf of Mexico oil spill response. Press release
24th April 2010. Retrieved March 28, 2016, from http://www.bp.com/en/
global/corporate/press/press-releases/bp-forges-ahead-with-gulf-of-mexico-
oil-spill-response.html
BP incident investigation team. (2010). Deepwater Horizon accident investiga-
tion report. Available at: http://www.bp.com/content/dam/bp/pdf/sustain-
ability/issue-reports/Deepwater_Horizon_Accident_Investigation_Report.
pdf [accessed 14th November 2016]
Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology.
Qualitative Research in Psychology, 3(2), 77–101.
Bryman, A. (2004). Social research methods (2nd ed.). Oxford: Oxford University
Press.
Burt, T. (2012). Dark art: The changing face of public relations. London: Elliott
and Thompson.
Butchart, G. (2011). An excess of signification: Or, what is an event? Semiotica,
187, 291–307.
Cahalan, P. (2012, July 1). Banks face lawsuits worth billions over Libor scam.
London: The Independent.
Cameron, D. (1995). Verbal hygiene: The politics of language. London: Routledge.
Cameron, D., & Panović, I. (2014). Working with written discourse. London: Sage.
Cameron, L. (1999). Operationalising “metaphor” for applied linguistic
research. In G. Low & L. Cameron (Eds.), Researching and applying metaphor
(pp. 3–28). Cambridge: Cambridge University Press.
Catenaccio, P., Cotter, C., De Smedt, M., Garzone, G., Jacobs, G., Macgilchrist,
F., et al. (2011). Towards a linguistics of news production. Journal of
Pragmatics, 43(7), 1843–1852.
252 References
Caton, C. (1987). Contributions of Roman Jakobson. Annual Review of

Anthropology, 16, 223–260.
Chandler, D. (1997). An introduction to genre theory. Retrieved September 1,
2014, from http://www.aber.ac.uk/~mcswww/Documents/intgenre/chandler_
genre_theory.pdf
Chandler, D. (2001). Semiotics for beginners. Retrieved September 1, 2014, from
http://www.aber.ac.uk/media/Documents/S4B/dutch/sem01.html
Chandler, D. (2007). Semiotics: The basics (2nd ed.). Abingdon: Routledge.
Chen, H. (2010). Contrastive learner corpus analysis of epistemic modality and
interlanguage pragmatic competence in L2 writing. Arizona working papers in
SLA and teaching (Vol. 17).
Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press.
Cobley, P., & Jansz, L. (1999). Introducing semiotics. Duxford: Icon Books.
Cockcroft, R., & Cockcroft, S. (2005). Persuading people: An introduction to
rhetoric (2nd ed.). Basingstoke: Palgrave Macmillan.
Conboy, M. (2007). The language of the news. Abingdon: Routledge.
Connor, U., & Lauer, J. (1985). Understanding persuasive essay writing:
Linguistic/rhetorical approach. Text, 5(4), 309–326.
Cook, G. (2001). The discourse of advertising (2nd ed.). London: Routledge.
Cook, G., Pieri, E., & Robbins, P. (2004). “The scientists think and the public
feels”: Expert perceptions of the discourse of GM food. Discourse & Society,
15(4), 433–449.
Cook, G., Robbins, P., & Pieri, E. (2006). “Words of mass destruction”: British
newspaper coverage of the genetically modified food debate, expert and non-
expert reactions. Public Understanding of Science, 15, 5–29.
Coombs, W. (2004). Impact of past crises on current crisis communication:
Insights from situational crisis communication theory. Journal of Business
Coon 2: hindsight. (2010). South Park. Series 14, episode 11. Comedy Central,
27th October, 2010.
Cornelissen, J. (2008). Metonymy in language about organizations: A corpus-
based study of company names. Journal of Management Studies, 45(1), 79–99.
Cornelissen, J., & Harris, P. (2001). The corporate identity metaphor:
Perspectives, problems and prospects. Journal of Marketing Management,
17(1–2), 49–71.
Cotter, C. (2010). News talk: Investigating the language of journalism. Cambridge:
Cambridge University Press.
Coupland, N., & Jaworski, A. (2001). Discourse. In P. Cobley (Ed.), The Routledge
companion to semiotics and linguistics (pp. 134–148). London: Routledge.
References
253
Crystal, D. (2001). Language and the Internet. Cambridge: Cambridge University

Press.
Crystal, D. (2003). The Cambridge encyclopedia of the English language (2nd ed.).
Cambridge: Cambridge University Press.
Crystal, D. (2004). Rediscovering grammar (3rd ed.). London: Longman.
de Cock, C., Cutcher, L., & Grant, D. (2010). Crisis, critique and the construc-
tion of normality: Exploring finance capitalism’s discursive shifts. Culture and
Organization, 16(2), 181–183.
de Haan, F. (1999). Evidential and epistemic modality: Setting boundaries.
Southwest Journal of Linguistics, 18(1), 83–101.
del-Teso-Craviotto, M. (2006). Language and sexuality in Spanish and English
dating chats. Journal of Sociolinguistics, 10(4), 460–480.
de Saussure, F. (1959). Course in general linguistics (W. Baskin, Trans.). New York:
The philosophical library.
Dirven, R. (2002). Metonymy and metaphor: Different mental strategies of
conceptualisation. In R. Dirven & R. Pörings (Eds.), Metaphor and metonymy
in comparison and contrast (pp. 75–112). Berlin: Walter de Gruyter.
Dodd, S. D. (2002). Metaphors and meaning: A grounded cultural model of US
entrepreneurship. Journal of Business Venturing, 17, 519–535.
Droga, L., & Humphrey, S. (2003). Grammar and meaning. Berry, NSW: Target
Texts.
Ducrot, O. (1984). Le dire et le dit. Paris: Minuit.
Dudley-Evans, T. (1994). Genre analysis: An approach to text analysis for
ESP. In M. Coulthard (Ed.), Advances in written text analysis (pp. 219–228).
London: Routledge.
Eagleton, T. (1983). Literary theory: An introduction (2nd ed.). Oxford: Blackwell.
Eco, U. (1976). A theory of semiotics. Bloomington: Indiana University Press.
Fairclough, N. (1989). Language and power. Harlow: Addison Wesley Longman.
Fairclough, N. (1992a). Discourse and social change. Cambridge: Polity Press.
Fairclough, N. (1992b). Intertextuality in critical discourse analysis. Linguistics
and Education, 4(3–4), 269–293.
Fairclough, N. (1995a). Critical discourse analysis. Harlow: Longman.
Fairclough, N. (1995b). Media discourse. London: Edward Arnold.
Fairclough, N. (2000). New labour, new language? London: Routledge.
Fink, S. (1986). Crisis management: Planning for the inevitable. New York:
AMACOM.
Foucault, M. (1970). The order of things (A. Sheridan, Trans.). New York:
Pantheon.
254 References
Foucault, M. (1972). The archaeology of knowledge (A. Sheridan, Trans.).

New York: Pantheon.
Foucault, M. (1980). Power/knowledge: Selected interviews and other writings
1972–77 (C. Gordon, L. Marshall, J. Mepham, & K. Soper, Trans.).
New York: Pantheon.
Fowler, R. (1991). Language in the news: Discourse and ideology in the press.
Abingdon: Routledge.
Fulton, H., Huisman, R., Murphet, J., & Dunn, A. (2005). Narrative and
media. Cambridge: Cambridge University Press.
Gage, J. (1999). Color and meaning: Art, science, and symbolism. Berkeley and
Los Angeles: University of California Press.
Galtung, J., & Ruge, M. (1973). Structuring and selecting news. In S. Cohen &
J. Young (Eds.), The manufacture of news: Social problems, deviance and the
mass media (pp. 62–72). London: Constable.
Genette, G. (1997). Palimpsests (C. Newman & C. Doubinsky, Trans.). Lincoln:
University of Nebraska Press.
Gilbert, G. N., & Mulkay, M. (1984). Opening Pandora’s box: A sociological anal-
ysis of scientific discourse. Cambridge: Cambridge University Press.
Givón, T. (1995). Isomorphism in the grammatical code: Cognitive and biologi-
cal considerations. In R. Simone (Ed.), Iconicity in language (pp. 47–76).
Amsterdam: John Benjamins.
Glaser, B., & Strauss, A. (1967). The discovery of grounded theory: Strategies for
qualitative research. Chicago: Aldine.
Goossens, L. (1990). Metaphtonymy: The interaction of metaphor and meton-
ymy in expressions for linguistic action. Cognitive Linguistics, 1(3), 323–342.
Graham, L. J. (2005). Discourse analysis and the critical use of Foucault. Australian
Association for Research in Education 2005 Annual Conference, Sydney.
Grutman, R. (2010). How to do things with mottoes: Recipes from the roman-
tic era (with special reference to Stendhal). Neohelicon, 37(1), 139–153.
Hale, J. E., Dulek, R., & Hale, D. (2005). Crisis response communication chal-
lenges: Building theory from qualitative data. Journal of Business
Haley, M. C. (1995). Iconic functions of the index in poetic metaphor. Journal
of Pragmatics, 24(6), 605–625.
Hall, S. (1980). Encoding/decoding. In S. Hall, D. Hobson, A. Lowe, & P. Willis
(Eds.), Culture, media, language (pp. 117–127). London: Unwin Hyman.
Halliday, M. A. K. (1978). Language as social semiotic: The social interpretation of
language and meaning. London: Edward Arnold.
References
255
Halliday, M. A. K., & Hasan, R. (1976). Cohesion in English. London: Longman.
Halliday, M. A. K., & Hasan, R. (1985). Language, context, and text: Aspects of
language in a social-semiotic perspective. Oxford: Oxford University Press.
Halliday, M. A. K., & Matthiessen, C. M. I. M. (2004). An introduction to
functional grammar (3rd ed.). London: Hodder Education.
Harrison, C. (2003). Visual social semiotics: Understanding how still images
make meaning. Technical Communication, 50(1), 46–60.
Herring, S. C., & Paolillo, J. C. (2006). Gender and genre variations in weblogs.
Journal of Sociolinguistics, 10(4), 439–459.
Herring, S. C., Scheidt, L. A., Bonus, S., & Wright, E. (2004). Bridging the gap:
A genre analysis of weblogs. Proceedings of the 37th Hawai’i International
Conference on System Sciences (HICSS-37). Los Alamitos: IEEE Computer
Society Press.
Hiraga, M. K. (1994). Diagrams and metaphors: Iconic aspects in language.
Journal of Pragmatics, 22(1), 5–21.
Hodge, R., & Kress, G. (1988). Social semiotics. Cambridge: Polity.
Hoey, M. (2001). Textual interaction: An introduction to written discourse analy-
sis. Abingdon: Routledge.
Hymes, D. (1974). Foundations of sociolinguistics: An ethnographic approach.
Philadelphia: University of Pennsylvania Press.
Iedema, R. (2003). Multimodality, resemiotization: Extending the analysis of
discourse as multi-semiotic practice. Visual Communication, 2(1), 29–57.
Jacobs, G. (1999). Preformulating the news. Amsterdam: John Benjamins.
Jacobs, G. (2000a). Self-reference in press releases. Journal of Pragmatics,
31(1999), 219–242.
Jacobs, G. (2000b). What’s in a crisis? A critical look at the field of crisis com-
munication. Document Design, 2(3), 225–235.
Jakobson, R. (2002). The metaphoric and metonymic poles. In R. Dirven &
R. Pörings (Eds.), Metaphor and metonymy in comparison and contrast
(pp. 41–48). Berlin: Walter de Gruyter.
Jakobson, R., & Halle, M. (1956). Fundamentals of language. The Hague: Mouton.
Jespersen, O. (1922). Language: Its nature, development and origin. London:
George Allen & Unwin.
Jespersen, O. (1924). The philosophy of grammar. London: George Allen & Unwin.
Johnstone, B. (2008). Discourse analysis (2nd ed.). Oxford: Blackwell.
Kessler, B., Nunberg, C., & Schűtze, H. (1997). Automatic detection of text
genre. Proceedings of ACL-97, 35th annual meeting of the Association for
Computational Linguistics (pp. 32–38), Madrid.
256 References
Koller, V. (2003). Metaphor clusters in business media discourse: A social cognition

approach. PhD thesis. University of Vienna.
Koller, V., & Davidson, P. (2008). Social exclusion as conceptual and grammati-
cal metaphor: A cross-genre study of British policy-making. Discourse &
Society, 19(3), 307–331.
Krennemayr, T. (2011). Metaphor in newspapers. Utrecht: LOT.
Kress, G. (1983). Linguistic processes and the mediation of “reality”: The poli-
tics of newspaper language. International Journal of the Sociology of Language,
40, 43–57.
Kress, G., & van Leeuwen, T. (2001). Multimodal discourse: The modes and
media of contemporary communication. London: Bloomsbury Academic.
Kress, G., & van Leeuwen, T. (2002). Colour as a semiotic mode: Notes for a
grammar of colour. Visual Communication, 1, 343–368.
Kress, G., & van Leeuwen, T. (2006). Reading images: The grammar of visual
design (2nd ed.). London: Routledge.
Kristeva, J. (1980). Desire in language: A semiotic approach to literature and art.
New York: Columbia University Press.
Labov, W., & Waletzky, J. (1967). Narrative analysis: Oral versions of personal
experience. In J. Helm (Ed.), Essays on the verbal and visual arts (pp. 12–44).
Seattle: University of Washington Press.
Lacey, N. (1998). Image and representation: Key concepts in media studies.
Basingstoke: Palgrave Macmillan.
Lakoff, G., & Johnson, M. (1980). Metaphors we live by. Chicago: University of
Chicago Press.
Landert, D. (2014). Personalisation in mass media communication. Amsterdam:
John Benjamins.
Lattmann, C. (2012). Icons of novel thought: A new perspective on Peirce’s defi-
nition of metaphor. Semiotica, 192(1/4), 535–556.
Leech, G., & Short, M. (1981). Style in fiction: A linguistic introduction to English
fictional prose. Harlow: Longman Group.
Lischinsky, A. (2011). In times of crisis: A corpus approach to the construction
of the global financial crisis in annual reports. Critical Discourse Studies, 8(3),
153–168.
Lock, C. (1997). Debts and displacements : On metaphor and metonymy. Acta
Linguistica Hafniensia : International Journal of Linguistics, 29(1), 321–337.
Macgilchrist, F. (2007). Positive discourse analysis: Contesting dominant dis-
courses by reframing the issues. Critical Approaches to Discourse Analysis Across
Disciplines, 1(1), 74–94.
References
257
Machin, D. (2007). Introduction to multimodal analysis. London: Bloomsbury

Academic.
Machin, D., & Jaworski, A. (2006). Archive video footage in news: Creating a
likeness and index of the phenomenal world. Visual Communication, 5(3),
345–366.
Maingueneau, D., & Angermuller, J. (2007). Discourse analysis in France: a con-
versation. Forum Qualitative Research, 8(2). Retrieved from http://www.qualita-
tive-research.net/index.php/fqs/article/view/254/559 [accessed 1st May 2016]
Martin, J. R. (2000). Design and practice: Enacting functional linguistics.
Annual Review of Applied Linguistics, 20, 116–126.
Martin, J. R. (2004). Positive discourse analysis: Solidarity and change. Revista
Canaria de Estudios Ingleses, 49, 179–200.
Martin, J. R., & White, P. R. R. (2005). The language of evaluation: Appraisal in
English. Basingstoke: Palgrave Macmillan. doi:10.1057/9780230511910.
Massey, J. E. (2001). Managing organizational legitimacy: Communication
strategies for organizations in crisis. Journal of Business Communication,
38(2), 153–182.
McLaren-Hankin, Y. (2007). Conflicting representations in business and media
texts: The case of PowderJect Pharmaceuticals plc. Journal of Pragmatics,
39(6), 1088–1104.
Merrell, F. (2001). Charles Sanders Peirce’s concept of the sign. In P. Cobley
(Ed.), The Routledge companion to semiotics and linguistics (pp. 28–39).
London: Routledge.
Mills, S. (2003). Michel Foucault. London: Routledge.
Milne, M. J., Kearins, K., & Walton, S. (2006). Creating adventures in wonder-
land: The journey metaphor and environmental sustainability. Organization,
13(6), 801–839.
Mintzberg, H., Ahlstrand, B., & Lampel, J. (1998). Strategy safari. New York:
Prentice Hall.
Mitroff, I. (2005). Why some companies emerge stronger and better from a crisis: 7
essential lessons for surviving disaster. New York: AMACOM.
Monelle, R. (1999). Music and the Peircean trichotomies. International Review
of the Aesthetics and Sociology of Music, 22(1), 99–108.
Muralidharan, S., Dillistone, K., & Shin, J.-H. (2011). The Gulf Coast oil spill:
Extending the theory of image restoration discourse to the realm of social
media and beyond petroleum. Public Relations Review, 37(3), 226–232.
258 References
National Commission on the BP Deepwater Horizon Oil Spill and Offshore

Drilling. (2011). Deep water: The Gulf Oil disaster and the future of offshore
drilling. Washington, DC.
Nexis UK. (2014). News database. Retrieved September 1, 2014, from https://
www.lexisnexis.com/uk/nexis/auth/bridge.do?rand=0.7553967985732511
Ng, C. J. W., & Koller, V. (2013). Deliberate conventional metaphor in images:
The case of corporate branding discourse. Metaphor and Symbol, 28(3), 131–147.
Nølke, H. (2006). The semantics of polyphony (and the pragmatics of realization).
Acta Linguistica Hafniensia: International Journal of Linguistics, 38(1), 137–160.
Nöth, W. (2001). Semiotic foundations of iconicity in language and literature.
In O. Fischer & M. Nänny (Eds.), The motivated sgn: Iconicity in language and
literature 2 (pp. 17–28). Amsterdam: John Benjamins.
Obama, B. (2015). PBS NewsHour. PBS, Tuesday 1st October 2015.
O’Halloran, K. (2004). Introduction. In K. O’Halloran (Ed.), Multimodal dis-
course analysis: Systemic functional perspectives (pp. 1–10). London:
Continuum.
Oliveira, S. (2004). The unthinkable unprecedented: Intertextuality in newspa-
per genres. Linguagem Em Discurso, 5(1), 9–28.
O’Reilly, D., Lamprou, E., Leitch, C., & Harrison, R. (2013). Leadership and
authority in a crises-constructing world: Towards a synthesis and clarifica-
tion. Leadership, 9(1), 145–147.
Palmer, F. (2001). Mood and modality (2nd ed.). Cambridge: Cambridge
University Press.
Parmentier, R. (2009). Troubles with trichotomies: Reflections on the utility of
Peirce’s sign trichotomies for social analysis. Semiotica, 117(1/4), 139–155.
Pearson, C., & Mitroff, I. (1993). From crisis prone to crisis prepared: A frame-
work for crisis management. Academy of Management Executive, 7, 48–59.
Peirce, C. (1931–1958). Collected writings (8 Vols.). (C. Hartshorne, P. Weiss, &
A. W. Burks, Eds.). Cambridge, MA: Harvard University Press.
Ponzio, A. (2010). Metaphoric image and iconic likeness. Semiotica, 181(1–4),
275–282.
Potter, J., & Reicher, S. (1987). Discourses of community and conflict: The
organisation of social categories in accounts of a “riot.”. British Journal of
Social Psychology, 26, 25–40.
Potter, J., & Wetherell, M. (1987). Discourse and social psychology: Beyond atti-
tudes and behaviour. London: Sage.
Potter, J., Wetherell, M., Gill, R., & Edwards, D. (1990). Discourse: Noun,
verb or social practice? Philosophical Psychology, 3(2), 205–217.
References
259
Radden, G., Köpcke, K.-M., Berg, T., & Siemund, P. (Eds.) (2007). Aspects of
meaning construction. Amsterdam: John Benjamins.
Radwańska-Williams, J. (1994). The problem of iconicity. Journal of Pragmatics,
22(1), 23–36.
Reynolds, M. (1997). Texture and structure in genre. Revue Belge de Philologie et
D’histoire, 75(3), 683–697.
Riffaterre, M. (1980). Syllepsis. Critical Inquiry, 16(4), 625–638.
Roberts, C. W., Zuell, C., Landmann, J., & Wang, Y. (2008). Modality analysis:
A semantic grammar for imputations of intentionality in texts. Quality &
Quantity, 44(2), 239–257.
Roberts, M., & McCombs, M. (1994). Agenda setting and political advertising:
Origins of the news agenda. Political Communication, 11(3), 249–262.
Ruddy Hell! It’s Harry and Paul. (2012). Series 4, episode 3. BBC, 11th November
2012.
Scollon, R. (1998). Mediated discourse as social interaction: A study of news dis-
course. Harlow: Addison Wesley Longman.
Siebenhaar, B. (2006). Code choice and code-switching in Swiss-German
Internet Relay Chat rooms. Journal of Sociolinguistics, 10(4), 481–506.
Simone, R. (1995). Iconicity in language. Amsterdam: John Benjamins.
Simpson, P. (2004). Stylistics: A resource book for students. Abingdon: Routledge.
Singleton, D. (2000). Language and the lexicon: An introduction. London:
Arnold.
Sørensen, B. (2011). The concept of metaphor according to the philosophers
C. S. Peirce and U. Eco—A tentative comparison. Signs: International Journal
of Semiotics, 5, 147–176.
Speer, S. A. (2001). Participants’ orientations, ideology and the ontological sta-
tus of hegemonic masculinity: A rejoinder to Nigel Edley. Feminism and
Psychology, 11(1), 141–144.
Srivastva, S., & Barrett, F. (1988). The transforming nature of metaphors in
group development: A study in group theory. Human Relations, 41, 31–64.
Stephens, K. K., Malone, R., & Bailey, C. (2005). Communicating with stake-
holders during a crisis: Evaluating message strategies. Journal of Business
Sunderland, J. (2006). Language and gender: An advanced resource book.
Abingdon: Routledge.
Swales, J. (1990). Genre analysis. Cambridge: Cambridge University Press.
Talja, S. (1999). Analyzing qualitative interview data : The discourse analytic
method. Library and Information Science Research, 21(4), 1–18.
Thompson, G. (2004). Introducing functional grammar (2nd ed.). London: Arnold.
260 References
Tsoukas, H. (1991). The missing link: A transformational view of metaphors in

organisational science. The Academy of Management Review, 16(3), 566–585.
van Dijk, T. (1985). Structures of news in the press. In T. van Dijk (Ed.),
Discourse and communication: New approaches to the analysis of mass media
(pp. 69–93). Berlin: Walter de Gruyter.
van Dijk, T. (1988). News as discourse. Hillsdale, NJ: L. Erlbaum Associates.
van Dijk, T. (1996). Discourse, power and access. In C. Caldas-Coulthard &
M. Coulthard (Eds.), Texts and practices: Readings in critical discourse analysis
(pp. 84–104). London: Routledge.
van Dijk, T. (2008). Discourse and power. Basingstoke: Palgrave Macmillan.
Van Hout, T., & Macgilchrist, F. (2010). Framing the news: An ethnographic
view of business newswriting. Text & Talk, 30(2), 169–191.
van Leeuwen, T. (2005). Introducing social semiotics. Abingdon: Routledge.
van Leeuwen, T. (2006). Towards a semiotics of typography. Information Design
Journal, 14(2), 139–155.
Wales, K. (1989). A dictionary of stylistics. Harlow: Longman.
Weedon, C. (1987). Feminist practice and poststructuralist theory. Oxford: Blackwell.
Weiss, G., & Wodak, R. (2003). Critical discourse analysis: Theory and interdisci-
plinarity. (G. Weiss & R. Wodak, Eds.). Basingstoke: Palgrave Macmillan.
Western, S. (2010). Eco-leadership: Towards the development of a new paradigm.
In Leadership for environmental sustainability (pp. 36–54). New York: Routledge.
Wetherell, M. (1998). Positioning and interpretative repertoires: Conversation
analysis and post-structuralism in dialogue. Discourse & Society, 9(3), 387–412.
White, M. (2003). Metaphor and economics: The case of growth. English for
Specific Purposes, 22(2), 131–151.
White, P. (1997). Death, disruption and the moral order: The narrative impulse
in mass-media “hard news” reporting. In F. Christie & J. Martin (Eds.), Genres
and institutions: Social processes in the workplace and school (pp. 101–133).
London: Cassell.
Widdowson, H. (1995). Discourse analysis: A critical view. Language and
Literature, 4(3), 157–172.
Wilden, A. (1987). The rules are no game: The strategy of communication. London:
Routledge & Kegan Paul.
Winter, E. (1994). Clause relations as information structure: Two basic struc-
tures in English. In M. Coulthard (Ed.), Advances in written text analysis
(pp. 46–68). London: Routledge.
Wodak, R. (1996). Disorders of discourse. Harlow: Addison Wesley Longman.
Wodak, R. (2000). Does sociolinguistics need social theory? New perspectives in
critical discourse analysis. Sociolinguistics Symposium, Bristol. Retrieved
References
261
September 1, 2014, from http://www.univie.ac.at/linguistics/forschung/

wittgenstein/unemploy/bristol5.htm
Wodak, R., & Meyer, M. (2001). Methods of critical discourse analysis. London: Sage.
Wooffitt, R. (2005). Conversation analysis and discourse analysis. London: Sage.
Yates, J., & Orlikowski, W. (1992). Genres of organizational communication: A
structurational approach to studying communication and media. The
Academy of Management Review, 17(2), 299–326.
Zare, J., Abbaspour, E., & Nia, M. (2012). Presupposition trigger: A compara-
tive analysis of broadcast news discourse. International Journal of Linguistics,
4(3), 734–743.
References for News Media Items

24/7 Wall St (27.4.2010) Media Digest 4/27/2010 Reuters, WSJ, NYTimes, FT,
Bloomberg.
Agence France Presse (27.4.2010) BP’s soaring profits overshadowed by oil rig
tragedy.
Agence France Presse (27.4.2011) European stock markets firm amid company
results.
Agence France Presse (27.4.2012) London market awaits major earnings next
week.
The Associated Press (27.4.2011) BP expects to resume Gulf drilling this year.
Associated Press Financial Wire (27.4.2010) Cook on La. oil rig that exploded
recalls escape.
Associated Press Worldstream (27.4.2010) BP Q1 profit more than doubles on the
year.
Benzinga (27.4.2011) BP Q1 Profit Slips.
BreakingNews.ie (27.4.2010) Massive oil leak threatens Gulf coast.
Campaign Middle East (27.4.2012) Building brands through behaviour.
Canwest News Service (27.4.2012) Sleepy North Dakota town transformed by
oil boom.
CaptainKudzu (27.4.2011) Why gas prices are going up … and how to stop
them.
Carleton Place (Canada) (27.4.2010) Crews may set fire to oil fouling Gulf of
Mexico after offshore drilling platform explosion.
CNN (27.4.2010) Taking on Wall Street: Dems First Bid at Reform Blocked by
GOP; Prescription for Waste: Hospital Overcharges in Billing Statement;
262 References
Taking on Wall Street; Arizona Immigration Law Protest; Bracing for a

Disaster; Docs Testing Bret Michaels.
CNN.com (27.4.2010) Seafood safe despite oil in Gulf of Mexico, experts say.
CompaniesandMarkets.com (27.4.2012) The Heavy Oil Market 2011–2021.
Coventry Evening Telegraph (27.4.2012) BP or not BP? That is the question:
Campaigners stage Stratford theatre protest at oil giant’s sponsorship of RSC’s
World Shakespeare Festival.
EI Finance (27.4.2011) The Cost of Success.
ENP Newswire (27.4.2011) GRI Research Board Announces Request for
Proposals for BP’s $ 500 Million Gulf of Mexico Research Initiative.
European Gas Markets (27.4.2011) Rosneft agrees extension to share-swap deal.
The Evening Standard (London) (27.4.2010) BP profits soar as oil giant a ccelerates
spill clean-up.
The Evening Standard (London) (27.4.2011) Triple whammy sees BP profits fall.
Executive Counsel (27.4.2011) Prepare in Advance for the Inevitable Crisis.
Foster Natural Gas/Oil Report (27.4.2012) Interior Secretary Salazar Takes It To
the Naysayers in Washington and Critics of the Administration’s Energy
Initiatives During the Past Three Years.
Fox News Network (27.4.2012) Discussion of President Obama’s Foreign Policy
and the EPA; ObamaCare Bill.
Global English Middle East and North Africa Financial Network (27.4.2011) BP
Q1 profit down 4%.
The Globe and Mail (Canada) (27.4.2010) The Gulf spill’s far-reaching impact
may spur doubts over effort to kill key Canadian safety rule.
Greenwire (27.4.2011) Offshore drilling: U.S. losing $4.7M a day from permit-
ting lag, group says.
Greenwire (27.4.2012) Gulf spill: Miss. AG pushes back against BP settlements
Guardian Unlimited (27.4.2011) Primark owner ABF sends FTSE lower ahead
of GDP figures.
Herald Sun (Australia) (27.4.2010) In brief.
The Irish Times (27.4.2011) Speaking mind on women lands Glencore chief in
trouble.
Journal of Counterterrorism and Homeland Security International (27.4.2011) An
IACSP Q & A With Michael Greenberger.
Kuwait News Agency (KUNA) (27.4.2010) Oil giant’s shell profits rise 135
percent.
Lewiston Morning Tribune (Idaho) (27.4.2012) Sampling fish for science: Trout,
suckers killed for testing to assess effect of oil spill in Yellowstone.
References
263
M2 PressWIRE (27.4.2011) Research and Markets: In Too Deep: BP and the

Drilling Race That Took it Down—The Truth behind the Greatest
Environmental Disaster in U.S. History.
Market News Publishing (27.4.2010) Provides Deepwater Horizon Update.
Marketing Management (27.4.2012) Crisis diagnostics: Assessing brand damage,
while restoring brand equity.
MarketWatch (27.4.2011) BP net higher, takes $400 million Gulf charge.
NewsWatch: Energy (27.4.2010) Coast Guard may try to burn off some of oil spill.
The New York Post (27.4.2012) Movie review.
The New York Times (27.4.2012) Oil’s Dark Heart Pumps Strong.
The New Zealand Herald (27.4.2012) New Orleans: Festivals mark city’s
renaissance.
The News Herald (Panama City, Florida) (27.4.2011) TDC forging fall market-
ing plan.
The Nightly Business Report (27.4.2010) Nightly Business Report.
PA Newswire: Corporate Finance News (27.4.2012) Week ahead.
Palm Beach Post (Florida) (27.4.2011) Biz briefs.
Phil’s Stock World (27.4.2012) Should We Kill The Politicians Before They Kill Us?
Plus Media Solutions (27.4.2012) Florida: Oil spill is prime suspect in hundreds
of dolphin deaths.
The Richmond Democrat (27.4.2010) Massive oil slick headed for wildlife
refuges.
The Right Blue via Twitter (27.4.2011) The Right Blue via Twitter.
Right Wing News (27.4.2012) “Crucify Them”: The Obama Way.
Sarasota Herald Tribune (Florida) (27.4.2012) New Orleans marches on.
The Scotsman (27.4.2012) Comment: Shell disposals are typical of industry
trend.
SNL Daily Gas Report (27.4.2011) Coast Guard report slams Transocean over
Deepwater Horizon.
St. Petersburg Times (Florida), (27.4.2010) Gulf oil spill could threaten Florida.
States News Service (27.4.2011) Rubio: “national debt can no longer be ignored”.
Tampa Bay Times (27.4.2012) Beyond Tampa Bay.
Tampa Bay Times (Florida) (27.4.2011) Immigrants make positive impact on
Florida.
Targeted News Service (27.4.2012) FBI’s Top Ten News Stories for the Week
Ending April 27, 2012.
TendersInfo (27.4.2010) United States : BP Says 1000 Barrels of Oil Leaking
Daily From Gulf Rig.
Theflyonthewall.com (27.4.2011) BP: Earnings.
Trend Daily News (Azerbaijan) (27.4.2010a) BP, GS, PUK, AZN, BA: Periodicals.
Trend Daily News (Azerbaijan) (27.4.2010b) Oil leak from sunken rig off La.
could foul coast.
The Virginian-Pilot (Norfolk, VA.) (27.4.2012) Your views.
Wireless News (27.4.2011) BP-Funded Gulf of Mexico Research Initiative
Reports Research Funding.
Index1
A Barthesian, 34, 69, 71, 75, 83–7,

Abrams, J., 206, 233 201, 212, 213, 243
Allen, G., 56, 57, 221 Barthes, R., 5, 34–5, 40, 45–7, 57,
anchorage, 40 63–5, 70–3, 86, 92, 109, 110,
Androutsopoulos, J., 37–8 123, 186, 187, 197, 201, 202,
Angermuller, J., viii, 34, 35 239, 240. See also Barthesian
Appraisal System, 13, 60–3, 137, 155, Baudrillard, J., 35, 84, 85, 204, 208,
164. See also Appraisal Theory 237, 238, 240
Appraisal Theory, 62, 63 Baxter, J., 10, 34, 180, 181
Arpan, L. M., 183 Bazerman, C., 54, 57, 58, 142, 221
arts reviews, 19, 93, 94, 96, 128, Bednarek, M., 13
138–9, 232 Beghtol, C., 54
arts texts, 150 Bergin, T., 24, 25, 130, 170, 223
Bhatia, V. K, 11, 53, 54, 67,
92, 97
B Biber, D., 12, 19, 29, 67
Bakhtin, M., 35, 56 Bignell, J., 8, 10, 14, 59, 154
Barrett, F., 49 Bird, S., 11, 14, 17
Note: Page numbers with “n” denote notes.

1

in Discourse, DOI 10.1057/978-1-137-58750-3
266 Index
blog(s), 16–18, 55, 76, 77, 91, 96, categorisation, 50, 51, 62, 77, 92,
99, 102, 137, 142, 151, 175, 107–10, 112, 119–25, 127,
184, 185, 232. See also weblogs 135, 167, 174, 175, 179, 181,
Blommaert, J., 7 183, 185, 186, 207, 216, 222,
Borges, J. L., 50, 85 223, 225, 229–31, 234–5, 248
BP. See British Petroleum (BP) Catenaccio, P., 37
Braun, V., 30–1 Caton, C., 210, 211
British Petroleum (BP) CDA. See Critical Discourse Analysis
crisis, 4, 6, 20, 23, 51, 84, 86, 98, (CDA)
118, 119, 124, 143, 173, 197, Chandler, D., 52, 53, 92, 149, 167,
201, 203, 212, 227, 248 175, 202, 204, 205, 207, 229, 238
Deepwater Horizon, 3, 6, 15, 21, Chen, H., 61, 99
23–5, 27, 75–7, 79, 112–14, Chomsky, N, 31, 60
132, 145, 197, 237 Clarke, V., 30–1
oil spill, 78, 80, 89–91, 93–5, 98, CMD. See computer-mediated
99, 102, 104, 105, 118, 119, discourse (CMD)
121, 124, 133, 135, 136, 138, Cobley, P., 202, 210
149, 152, 156, 185–7, 203, Cockcroft, R., 138
223, 227, 235, 240 Cockcroft, S., 138
texts, 16, 18, 29, 62, 73, 83, code, 8, 10, 30, 34, 42, 46–8, 52–3,
89–110, 139, 141, 150, 174, 60–3, 65, 69, 71, 73–5, 83, 92,
175, 178, 197 105, 109, 127–65, 179, 191,
Bryman, A., 30–1, 81 204, 235, 244, 247, 248
Burt, T., 9, 23, 91, 169, 223 Commentary (news media genre),
business or market reports, 6–8, 10, 4, 12, 16–18, 96, 97, 99,
15, 20–3, 25, 49, 51, 54, 55, 103, 133, 135, 182, 184,
67, 68, 73, 77, 79, 92, 97, 106, 221, 223, 232
107, 116, 117, 121, 123, 124, computer-mediated discourse
128, 134, 139–41, 145, 153, (CMD), 28, 37–8
160, 162, 170, 173–7, 187, Conboy, M., 13, 19, 71, 186
188, 212, 218, 221, 222, 226, Connor, U., 50
231, 232, 235, 244 connotation(s), 47, 48, 64–5, 67, 70,
Butchart, G., 49, 203, 232 74, 167, 191, 248. See also
connotative
connotative, 46, 64, 65, 109
C content analysis, 28, 30, 180
Cahalan, P., 23 contextualisation (stage of analysis),
Cameron, D., 14, 36, 40 86, 87, 89–100, 127, 197, 232
Cameron, L., 173 Cook, G., 55, 81, 139, 181, 210, 222
Index
267
Coombs, W., 22 Discourse Analysis, Potter &

cornelissen, J., 67, 167, 169, 175, 218 Wetherell’s, 28, 35–6, 180
corpus linguistics, 29–30, 42, 80, 81 Dodd, S. D., 176
co-text, 7, 79–80, 107, 143, 152–3, Droga, L., 41, 61
178, 234 Ducrot, O., 35
Cotter, C., 9, 11, 14, 17, 19 Dudley-Evans, T., 53
Coupland, N., 33, 169
crisis
business, 6, 20, 106, 107, 141, E
153, 170, 187, 212, 231 Eagleton, T., 70
communications, 20–4 Eco, U., 202, 205
management, 106, 188, 218, editorial, 7, 9, 11, 12, 16–20,
231, 233 54, 56, 70, 93, 96, 125, 128,
Critical Discourse Analysis (CDA), 136–8, 142, 171, 175, 184,
13, 27, 32–4, 70, 71 225, 232
Crystal, D., 32, 37, 246 ethnographic, 11, 13, 36–8
ethnography, 36–8. See also
ethnographic
D ethnographer, 37
Dardenne, R., 11, 14, 17
Davidson, P., 173, 234
de Cock, C., 187 F
de Haan, F., 62 “Factual objectivity” (discourse of), 140
del-Teso-Craviotto, M., 38 Faircloughian, 34, 71
denotation, 46, 47, 64, 65, 173. Fairclough, N., 13, 32–4, 41, 55, 58,
See also denotative 59, 71, 72, 142, 180. See also
denotative, 46, 47, 64, 65, 109 Faircloughian
depth analysis (stage), 80, 81, 86, 87, feature article, 11, 16–17, 92, 93,
89, 90, 110–25, 127–65, 95, 135–6, 142, 232
167–89, 197 financial report, 7, 25, 56,
de Saussure, F., 45, 46, 49, 52, 84, 92–5, 128, 133–5, 142,
201, 202, 206, 209, 210, 148, 150, 152, 175, 181,
220, 243 186, 194–6, 226
Dirven, R., 168 Fink, S., 22
discourse(s), 4, 6, 10, 13, 16, 20, 21, Foucault, M., 34, 50, 72, 73, 180
27–42, 48, 55, 58, 59, 64, 70–3, Fowler, R., 8, 13, 14, 49, 50, 59,
75, 81, 83–6, 105, 108–10, 158, 217, 228–30
179–89, 191, 201, 213, 216, 217, Fulton, H., 11, 14–16, 51, 128,
222–8, 235, 240, 243–5, 247 130, 132
268 Index
G Hiraga, M. K., 233

Gage, J., 39 Hodge, R., 41, 208
Galtung, J., 8 Hoey, M., 54, 137, 152
Genette, G., 58, 142 holistic analysis (stage), 86, 189,
genre, 3, 4, 6–8, 11, 12, 14, 15, 18, 191–7
19, 29, 32, 34, 36, 47, 52–6, Humphrey, S., 41, 61
58, 59, 61, 65, 76, 80, 81, 86, Hymes, D., 36
87, 89, 90, 92–100, 103,
107–10, 118, 127–42, 151,
152, 157, 159, 161, 164, 174, I
175, 179, 181, 183, 188, 194–6, Icon. See also iconic
219–21, 223, 232, 236, 244–6 iconicity, 209–11, 229, 235, 236,
Gilbert, G. N., 35 239, 248
Givón, T., 210 “naturalised icon”, 237–40, 248
Glaser, B., 31 iconic, 204–7, 209–11,
Goossens, L., 66 227–9, 232–40
Graham, L. J., 33, 180 ideologies, 8, 13, 19, 47, 48, 63,
grounded theory, 30, 31 69–73, 180, 224, 240, 248
Grutman, R., 205 ideology, 47–8, 53, 63, 69–72, 74,
75, 105, 179–89, 244. See also
ideologies
H ideological, 6, 7, 10, 11, 13, 16,
Hale, J. E., 22, 204, 206, 208 42, 46, 47, 52, 63, 67, 68, 70,
Haley, M. C., 204, 206, 208 71, 74, 109, 147, 179, 182, 223
Halle, M., 66 Iedema, R., 5
Hallidayan, 32 illustration, 7, 27, 40, 65, 99, 130,
Halliday, M. A. K., 5, 41, 60, 62, 191, 230, 246
163, 164. See also Hallidayan immersion, 31, 48, 86, 87, 101, 110,
Hall, S., 34, 52, 64 197, 247
hard news, 7, 11, 12, 14–16, 51, 56, index, 99, 119, 202, 204–11,
139, 141, 157, 175 222, 227, 231, 232,
Harrison, C., 39, 187 237, 239, 244.
Harris, P., 218 See also indexical
Hasan, R., 5, 40, 60 indexicality, 211, 230, 235–6
Hayward, Tony, 24, 25, 94, 117, indexical, 119, 205–7, 209, 211,
139, 145–7, 149, 153, 170, 227, 229–32, 234–6
182, 192, 193, 195, 196, 236 interpretant, 70, 202–4, 220, 239
Herring, S. C., 18, 38, 54, 55 intertextual, 53, 59, 142, 152, 216
Index
269
intertextuality, 5, 13, 37, 42, 53, Lauer, J., 50

56–9, 70, 80, 107–10, 127, Leech, G., 12, 29, 67, 144
142–54, 179, 214, 221. See also letters, 16, 18, 54, 55, 76, 80, 93,
intertextual 97, 139, 145, 175, 185, 210
intertext(s), 13, 56, 57, 59, 107, Lischinsky, A., 49
146, 150, 153, 154, 171, 184, Lock, C., 205
221, 233
inverted pyramid, 11, 16, 128, 130,
131, 135, 181, 195 M
Macgilchrist, F., 11, 13, 37
Machin, D., 5, 39, 40, 59
J Macondo, 78, 114, 122, 133, 160
Jacobs, G., 10, 37, 59, 146 Maingueneau, D., 34
Jakobson, R., 66, 67, 173, 210, Martin, J. R., 13–15, 17, 41, 62,
211, 234 137–9, 164, 217, 220
Jansz, L., 202, 210 Massey, J. E., 218
Jaworski, A., 5, 33, 39, 59, 69 Matthiessen, C. M. I. M., 5, 41, 60,
Jespersen, O., 62, 211 62, 163, 164
Johnson, M., 66–8, 121, 167, 173 McCombs, M., 9
Johnstone, B., 53, 55 McLaren–Hankin, Y., 49
journalistic practice, 7, 9–11, 225 media language, 12–14, 19, 52, 55
Merrell, F., 206
metaphorical, 67–9, 169, 173–6,
K 178, 221, 222, 234, 237
Kessler, B., 54 metaphors, 5, 14, 16, 36, 57, 61,
Koller, V., 68, 173, 174, 234 65–9, 84, 108, 109, 134, 167,
Krennemayr, T., 175 169, 173–9, 181, 195, 196,
Kress, G., 5, 33, 39–42, 59, 208 204–6, 208, 212, 217, 218, 222,
Kristeva, J., 56–8 233–4. See also metaphorical
metonymic, 66, 123, 167, 168,
170–3, 181, 182, 205, 218,
L 222, 230, 234
Labov, W., 131 metonymy. See also metonymic
Lacey, N., 64 metonymically, 169, 170
Lakoff, G., 66–8, 167, 173 metonyms, 65–8, 108, 109,
Landert, D., 12, 17 167–73, 175, 178, 179, 205,
“language map”, 83–6, 201, 211, 217–18, 230, 248
212–16, 244 Meyer, M., 33
Lattmann, C., 233 Mills, S., 72, 73
270 Index
Milne, M. J., 68, 173, 177 naming people, 49, 51, 108,
Mintzberg, H., 21 115–18, 129, 144, 169, 217
Mitroff, I., 22, 187 naming social actors, 50, 221
modal expressions, 18, 60, 130, 136, narrative
138, 140, 155, 159, 161, 162, analysis, 30–1
181, 187, 216, 219 structure, 11, 17, 131, 132
modality. See also modal expressions naturalisation. See also naturalise
deontic, 62, 154, 155, 159, 161, naturalised, 47, 69–71, 187, 221,
163, 220, 248 226, 239, 244
dynamic, 62, 105, 154–6, “naturalised icon”, 237–40, 248
158–63, 220 naturalise, 47, 69, 188
epistemic, 62, 105, 154–6, news agencies, 10, 143, 148–50, 192
158–64, 220, 228 news reports, 4, 9, 11, 12, 14–17,
modal, 18, 60–2, 102, 107, 130, 19–20, 51, 54, 58–60, 76, 78, 92,
136–8, 140, 148, 155–7, 159, 94, 97, 102, 128–33, 135, 136,
161–4, 181, 187, 213, 216, 138–44, 146, 148–50, 152, 153,
217, 219, 220, 245, 248 157, 159, 178, 181–3, 192, 194,
Monelle, R., 5, 39 196, 219, 221, 223, 228, 229,
Mulkay, M., 35 232, 233, 236. See also hard news
multimodal, 5, 6, 12, 18, 39–42 Nexis UK, 76, 77, 79, 89, 101, 149,
multimodality, 5, 38, 40. See also 192
multimodal Ng, C. J. W., 234
Muralidharan, S., 77, 218 Nølke, H., 35
myth, 46–8, 63, 64, 105, 109, Nöth, W., 210
123, 167, 179, 218, 234,
240, 244
mythic meaning, 48, 63–9, 73–5, O
167–78, 191, 248. See also myth Obama, B., 91, 96, 133, 171, 246–7
mythical, 68, 69 object, 20, 49, 63, 66, 69, 123, 142,
202–10, 220, 223–9, 231, 233,
237–40, 244
N objectivity, 14–17, 26, 33, 62, 65,
naming participants, 49–51, 115, 132, 139, 140, 157, 164, 182,
116, 118 195, 226, 228, 238
naming practices. See also naming construction of, 61
participants; naming social O’Halloran, K., 40
actors Oliveira, S., 13, 149
naming events, 21, 49–51, 108, O’Reilly, D., 187
109, 112–15, 218–19 Orlikowski, W., 54
Index
271
P Reicher, S., 28, 35, 180

Palmer, F., 258 Relay, 40
Panović, I. (2014). 36, 40 reported speech, 10, 16, 159,
Paolillo, J. C, 38 182, 220
Parmentier, R., 70, 238, 239 representamen, 202–4
Pearson, C., 21, 22 retweet, 137
Peirce, C. S., 67, 197, 201–2, 204–9, Reynolds, M., 258
211, 224, 227, 229–31, 233–5, rhetoric, 11, 65. See also rhetorical
237, 238, 240, 243, 244 rhetorical, 16, 18, 42, 50, 54, 109,
point of closure, 11, 95, 130, 135 136–8, 141, 167, 173, 175,
Ponzio, A., 204, 206, 208, 233 179, 186, 245, 247, 248
positioning (discourse of ), 182–4, Riffaterre, M., 57
186, 226 Roberts, C. W., 13, 61
Positive Discourse Analysis, 13 Roberts, M., 9
Poststructuralist Discourse Analysis Ruge, M., 8
(PDA), 34–5
poststructuralist analysis, 149
Potter, J., 28, 35, 180 S
preliminary analysis (stage), 86, Scollon, R., 9
101–10, 119, 197, 245. See also Semiotic Discourse Analysis, 27–42,
immersion 83–6, 244–5
press releases, 10, 19, 22, 23, 25, 37, semiotics, 4–6, 26, 27, 35, 38–42,
59, 79, 94, 107, 143, 146–8, 52, 80, 208
153, 154, 170, 182, 194, 196, SFG. See Systemic Functional
219, 233 Grammar (SFG)
Short, M., 144
Siebenhaar, B., 38
Q sign, 4, 5, 27, 38, 40, 45–53, 62,
quality newspapers, 8, 14, 16 64, 67, 70, 71, 73–5, 83–6,
quotation(s). See also reported speech 105, 109, 111–25, 179, 201–9,
direct, 12, 15, 143–6, 153, 195, 228 211–13, 216, 227–9, 231–3,
indirect, 15, 143–6, 228 235, 237, 239, 240, 243,
244, 247
signified, 45, 46, 52, 64–7, 84, 123,
R 185, 202, 206, 209, 220, 222,
Radden, G., 66, 168 228, 232, 240
Radwańska-Williams, J., 210 signifier(s), 15, 45, 46, 52,
“redeployment” (discourse of ), 64–7, 84, 123, 141, 202,
184–6, 188, 226 206, 209, 228
272 Index
Simone, R., 210, 211 Twitter. See also retweet

Simpson, P., 210 tweet(s), 77, 137
Singleton, D., 206, 210, 235 tweeting, 36, 137
“soft news”, 12, 14, 15, 17,
51, 175
Sørensen, B., 204–6, 233 V
Speer, S. A., 28, 180 van Dijk, T., 8, 13, 32–3, 59
Srivastva, S., 49 van Hout, T., 11, 37
Stephens, K. K., 23, 183, 187 van Leeuwen, T., 5, 39–41, 59,
story selection, 7–9, 20 67, 228
Strauss, A., 31
Sunderland, J., 149, 180
Swales, J., 53 W
symbol. See also symbolic Wales, K., 66, 209–11
symbolically, 178, 204–11, 226, Waletzky, J., 131
227, 229, 231–9, 244 weblogs, 18, 89, 102
symbolism, 210 Weedon, C., 34
symbolic, 178, 204–11, 226, 227, Weiss, G., 33, 81
229, 231–9, 244 Western, S., 187
synecdoche, 65–7, 167. See also Wetherell, M., 28, 35–6, 149, 180
synecdochic White, M., 173, 176
synecdochic, 205 White, P., 11, 13–17, 62, 63, 137–9,
Systemic Functional Grammar 157, 164, 217, 220
(SFG), 5, 31, 41–2, 180 Widdowson, H., 33
Wilden, A., 38
Winter, E., 11, 137, 196
T Wodak, R., 33, 81
tabloid newspapers, 7, 8, Wooffitt, R., 36
12–14, 76 Wright, E., 18, 54, 55
Talja, S., 36
thematic analysis, 30, 31
Thompson, G., 18, 41, Y
61, 163, 164 Yates, J., 54
traditional grammar, 31–2
Transocean, 24, 78, 171, 176, 193,
196 Z
Tsoukas, H., 68, 69 Zare, J., 103, 217

Semiotics and Verbal Texts 2017

Uploaded by

Copyright:

Available Formats

Semiotics and Verbal Texts 2017

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Semiotics and Verbal Texts 2017

Uploaded by

Copyright:

Available Formats

Semiotics and

How the News Media

More information about this series at

Semiotics and Verbal

ISBN 978-1-137-58748-0 ISBN 978-1-137-58750-3 (eBook)

Library of Congress Control Number: 2016959556

© The Editor(s) (if applicable) and The Author(s) 2017

Cover image © PM Images / Getty

Printed on acid-free paper

This Palgrave Macmillan imprint is published by Springer Nature

e­ ncouragement. Any remaining failings in the book are my own. I am

Part I Written language and semiotics1

1 Researching the Representation of a Crisis3

2 Semiotic Discourse Analysis27

Part II A Barthesian conceptualisation of written language43

4 Data Collection and Research Principles 75

5 A Barthesian Analysis of the BP Data in Four Stages 83

6 Stage 1: Contextualisation of the BP Texts 89

7 Stage 2: Preliminary Analysis of the BP Texts101

8 Stage 3: A Depth Analysis at the Level of the Sign 111

9 Stage 3: A Depth Analysis at the Level of the Code127

10 Stage 3: A Depth Analysis at the Level of Mythic Meanings167

11 Stage 3: A Depth Analysis at the Level of Ideology  179

12 Stage 4: A Holistic Analysis of a Single Text191

Part III A Peircean conceptualisation of written language199

13 Theoretical Foundations 201

14 A Peircean Interpretation of the BP Data213

Part IV Concluding thoughts241

15 Other Events, Other Contexts243

Fig. 2.1 Perspectives and tools of written text analysis 28

Table 4.1 Sample of BP-related texts from Nexis UK database  79

A Semiotic Account of a News Story

© The Author(s) 2017 3

Example 1 is drawn from a Canadian newspaper (Carleton Place) one week

of written texts, I will be considering semiotic concepts as epistemologi-

concepts in the field of semiotics which are useful as explanatory frame-

Using news data related to the BP Deepwater Horizon disaster spanning

they take account of the conventions and practices of pub conversation.

Researching News Stories

conventions of genre – places the approach I discuss here into context.

Given that news outlets need to select what is included in publications

Accounts of influence outside and inside publications focus on money or

As these relationships develop, some of the contacts develop considerable

Some of the most interesting work in writing and editing practices

News writing is a highly recognisable “genre colony” (Bhatia, 1993), not

The language of news is organised not only by structuring strategies

News Media Language

Language choices in the news media concern how information is com-

Studies of the collection, writing and editing practices discussed above

imposition of style guides (Cameron, 1995; Cotter, 2010). Each publica-

The News Report

Scholars of news media regularly make a distinction between “hard news”

“Hard” news reports on politics, economics, the doings of the powerful

• High proportion of empirical information regarding dates, places,

The aim of such strategies is to present an authorial (and institutional) voice

mediated through attribution (the journalistic author is never their

The Feature Article

In rhetorical terms, news reports are positioned as primarily realising a

Feature articles position their information and participants in quite a

The Editorial, Blog, Commentary and Reviews

e ncouragement. Any remaining failings in the book are my own. I am

Part I Written language and semiotics1

1 Researching the Representation of a Crisis3

2 Semiotic Discourse Analysis27

Part II A Barthesian conceptualisation of written language43

4 Data Collection and Research Principles 75

5 A Barthesian Analysis of the BP Data in Four Stages 83

6 Stage 1: Contextualisation of the BP Texts 89

7 Stage 2: Preliminary Analysis of the BP Texts101

8 Stage 3: A Depth Analysis at the Level of the Sign 111

9 Stage 3: A Depth Analysis at the Level of the Code127

10 Stage 3: A Depth Analysis at the Level of Mythic Meanings167

11 Stage 3: A Depth Analysis at the Level of Ideology 179

12 Stage 4: A Holistic Analysis of a Single Text191

Part III A Peircean conceptualisation of written language199

13 Theoretical Foundations 201

14 A Peircean Interpretation of the BP Data213

Part IV Concluding thoughts241

15 Other Events, Other Contexts243

Fig. 2.1 Perspectives and tools of written text analysis 28

Table 4.1 Sample of BP-related texts from Nexis UK database 79

A Semiotic Account of a News Story

Researching News Stories

News Media Language

The News Report

The Feature Article

The Editorial, Blog, Commentary and Reviews

In Summary: Difference and Sameness in News

Researching Written Text

Content and Narrative Analysis Tools

Critical Discourse Analysis

Poststructuralist Discourse Analysis

Potter & Wetherell’s Discourse Analysis

Ethnography of Written Text

Researching Written Text in Semiotic Studies

The “Text-as-Graphic” Approach

language is communication but very little communication is language”.

approach is encapsulated in Barthes’ (1977a) discussion of the relation-

Systemic Functional Grammar

Aspects of Roland Barthes’ Work

Sign, Code, Myth, Ideology

The Level of the Sign

The process of giving language to experience is more than just sense-

The Level of the Code

Grammatical Codes: Modality and the Appraisal System