Professional Documents
Culture Documents
Discourse & Society
Discourse & Society
http://das.sagepub.com/
Published by:
http://www.sagepublications.com
Additional services and information for Discourse & Society can be found at:
Subscriptions: http://das.sagepub.com/subscriptions
Reprints: http://www.sagepub.com/journalsReprints.nav
Permissions: http://www.sagepub.com/journalsPermissions.nav
What is This?
Article
Monika Bednarek
University of Sydney, Australia
Helen Caple
University of New South Wales, Australia
Abstract
This article introduces a new framework for the analysis of news discourse to scholars in Critical
Discourse Analysis (CDA) and beyond. It emphasises the importance of news values for linguistic
analysis and encourages a constructivist approach to their analysis. The new methodological
framework is situated within what the authors call a ‘discursive’ approach to news values.
From this perspective, news values are seen as values that exist in and are constructed through
discourse, and the primary research interest is in how texts construct newsworthiness through
multimodal resources. This article first introduces resources that are used to construe news
values in English-language news discourse, before illustrating the framework through two case
studies of a 70,000-word corpus of British news discourse. The framework itself is intended for
both multimodal discourse analysis and corpus linguistic analysis, although this article focuses
more on the integration of corpus linguistic techniques. Thus, the discursive approach ties in well
with two recent trends in CDA – towards multimodal and towards corpus-assisted discourse
analysis. More specifically, the case studies show that corpus linguistic techniques can identify
conventionalised discursive devices that are repeatedly used in news discourse to construct and
perpetuate an ideology of newsworthiness. They further show that such techniques can provide a
useful indication of the discursive construction of newsworthiness around a specific topic, event
or news actor. The article concludes with an outline of further applications of the framework for
(critical) linguistic analyses of news discourse.
Corresponding author:
Monika Bednarek, Department of Linguistics, F12 – Transient Building, University of Sydney, Sydney, NSW
2006, Australia.
Email: monika.bednarek@sydney.edu.au
Keywords
Corpus linguistics, corpus-assisted discourse analysis (CADS), Critical Discourse Analysis,
critical linguistics, journalism, multimodality, news discourse, news factors, news values
A brief introduction
The aim of this article is three-fold: first, to bring news values to the attention of critical
linguists; second, to encourage a constructivist approach to news values; and third, to
introduce a new framework for the analysis of news values in the hope that critical lin-
guists will both apply it and develop it further. By Critical Linguistics or CDA – we use
these terms interchangeably here – we mean any approach that is interested in using
discourse analysis (however defined) to uncover the (re)production of ideology. By news
values we mean the values of newsworthiness (e.g. Negativity, Proximity, Eliteness).
The new framework for the analysis of such values, which we introduce in this article by
means of two case studies, is explicitly aimed at the analysis of text/discourse, by which we
mean not just linguistic text, but also other semiotic systems such as images, layout and typog-
raphy. Due to space restrictions, however, we limit our discussion in this article to the analysis
of verbal resources (but see Bednarek and Caple, 2012a, 2012b; Caple, 2013, for a discussion
of images). Our approach is thus in line with a recent trend within CDA towards multimodal
discourse analysis (e.g. Machin and Mayr, 2012). It is also in line with another 21st-century
trend within CDA, towards combining quantitative and qualitative analysis, in particular the
use of corpus linguistic techniques (e.g. Baker et al., 2008; Mautner, 2000). Indeed, the latter
will be the main focus of our two case studies, although we will also have something to say
about multimodal discourse analysis. We will start by introducing news values and their con-
ceptualisation in Journalism/Communication Studies and Critical Linguistics before we con-
tinue with the illustration of our own discursive/constructivist approach to newsworthiness.
News values: What are they and why should they matter
for critical analyses of news discourse?
News values can be defined in many different ways, but in essence they determine what is
news(worthy). In the Journalism/Communication Studies literature, news values are typi-
cally defined as properties of events or stories or as criteria/principles that are applied by
news workers in order to select events or stories as news or to choose the structure and order
of reporting. Such values include Proximity (geographical or cultural ‘nearness’); Negativity
(negative aspects, e.g. conflict, death, disaster, accidents, negative consequences); Eliteness
(elite status); and Superlativeness (‘the more X, the more newsworthy’), to name but a few.
From a linguistic perspective, language can be seen as expressing, indicating, empha-
sising or highlighting news values (Bednarek, 2006; Bell, 1991; Conboy, 2006), or news
values can be regarded as becoming embedded in language (Cotter, 2010: 67). More
radically, news values have also been defined as values that are construed in and through
discourse (Bednarek and Caple, 2012a, 2012b). For instance, for a British audience, this
lead sentence from the Independent online:
Demolition has begun on the site of Britain’s worst nuclear accident, as scientists are
brought in to dismantle the infamous Sellafield chimney.
(http://tinyurl.com/ozlg579, accessed 12/9/2013)
construes the news value of Timeliness through tense and aspect (has begun, are brought
in); Proximity through Britain’s and Sellafield; Superlativeness and Negativity through
worst nuclear accident and infamous (‘famous for something bad’); and Eliteness
through scientists. In this view, news values can be seen as discursively constructed, and
newsworthiness becomes a quality of texts. News values are thus defined as the ‘news-
worthy’ aspects of actors, happenings and issues as existing in and constructed through
discourse. It should be emphasised that the view from Journalism/Media Studies is much
more widespread than this more recent discursive approach. In fact, the history of news
values in this area goes back to at least 1965, with Galtung and Ruge’s seminal publica-
tion in the Journal of Peace Research, although Walter Lippmann (1922) is widely
acknowledged as the first person to suggest a range of attributes, or news factors, that
lend news value to events.
Regardless of the way news values have been conceptualised, many researchers have
pointed to their ideological nature, including linguists. Thus, Fowler (1991) states that
news values are culturally and socially constructed rather than ‘natural’ (pp. 13, 15). Bell
stresses that ‘these are values. They are not neutral, but reflect ideologies and priorities
held in society’ (Bell, 1991: 156, italics in original). Cotter (2010) calls news values
important ‘ideological factors’ (pp. 8, 67) and points out that they reinforce or establish
‘an “ideology” about what counts as news’ (p. 67). News values are also ideological in
the sense that they can work to reinforce other ideologies (rather than just an ideology of
what is newsworthy). Richardson (2007: 93) mentions a study by the Glasgow Media
Group which showed that coverage of television news about developing countries
focused on negative happenings such as war, terrorism, disaster and conflict. This pre-
sents a narrow preconceived view of these countries, particularly if other countries would
not show the same foregrounding of the news value of Negativity. To give another exam-
ple, if it was found that stories about Central Asian republics like Kazakhstan, Uzbekistan,
Turkmenistan, etc., consistently foreground the news value of Novelty (unexpected
aspects), whereas stories about the USA and the UK foreground the news value of
Eliteness, this could be seen as reproducing an ideology of the ‘normal’ and ‘elite’ Us
versus the ‘exotic’ Them.
Despite these ideological aspects of news values, they have not yet been the focus of
critical linguistic analyses of news discourse. Instead, they are either ignored and only
mentioned in passing, or introduced as part of journalistic culture but not much used in
the actual analysis. For instance, there is no index entry for ‘news values’ in Fairclough’s
Media Discourse, and a search in the databases LLBA and MLA for ‘CDA’ or ‘Critical
Discourse Analysis’ together with ‘news values’ comes up with only two results. Among
the critical linguists who mention news values in passing, Baker et al.’s (2013) book,
Discourse Analysis and Media Attitudes, is a recent example that only refers to news
values four times in 270 pages (although there are additional mentions of newsworthy/
newsworthiness). Concerning those critical books on media or news discourse that intro-
duce news values, two classic examples are Van Dijk (1988b) and Fowler (1991). Van
Dijk (1988b) interprets news values as constraints that ‘have a cognitive representation’
(p. 121). These constraints are said to underlie the production of news, including selec-
tion and formulation: ‘the interpretation of events as potential news events is determined
by the potential news discourse such an interpretation (model) may be used for, and
conversely. News production seems circular: Events and text mutually influence each
other’ (p. 113). Despite making this connection to discourse/text, news values them-
selves are of a cognitive nature for Van Dijk (1988b), and thus do not feature much in his
analyses of news discourse (e.g. Van Dijk, 1988a, 1988b). Similarly, Fowler (1991) starts
off by explaining Galtung and Ruge’s (1965) original news values, but comes up with his
own cognitive conceptualisation. In this definition, news values are seen as socially con-
structed ‘intersubjective mental categories’ (p. 17), although Fowler also argues that they
are ‘qualities of (potential) reports’ and ‘features of representation’ (p. 19). Nevertheless,
Fowler does not discuss news values much in his book when analysing language, with
only a few mentions throughout. To be clear, both Fowler and Van Dijk do see news as
having discursive, social and cognitive dimensions and Van Dijk’s general work on
values and ideology certainly accounts for their discursive expressions.
A more recent addition to the Critical Discourse Analysis of newspapers is Richardson
(2007). News values are seen by him as one of several professional practices ‘that shape
journalism as a discourse process and therefore help to account for the products of news-
paper discourse’ (p. 182). Richardson explains news values as ‘the criteria employed by
journalists to measure and therefore to judge the “newsworthiness” of events’ (p. 91); he
talks about events (such as a war involving Britain) satisfying news values (p. 182), but
also writes that ‘news values are the (imagined) preferences of the expected audience’ (p.
94, italics in original). The relationship between these different conceptualisations is not
discussed at length or in depth, but if we interpret Richardson’s remarks correctly, this
means that in his view an event occurs, it satisfies particular news values, journalists
measure and judge this newsworthiness based on what they imagine their audiences find
newsworthy, and use this judgement to ‘select, order and prioritise the collection and
production of news’ (p. 91). Like the books of Fowler and Van Dijk, Richardson’s book
itself is not overly concerned with news values or their analysis. To name a second con-
temporary example, Machin and Mayr (2012) introduce an account of news values as
related to crime reporting in their introductory chapter, alongside other concepts such as
media panic, but they then only draw on news values a few times in the rest of their book.
These examples are not presented here as criticism of these (and other) linguists; they
just demonstrate that critical linguists have so far not found the notion of news values to
be very useful. But why is this the case? Why are critical linguists, who focus on uncover-
ing power relations and ideology, not interested in news values even though these values
have been called deeply ideological? One obvious reason is that such linguists have
focused on answering other valid research questions, such as the representation of agency.
But for us, there are three additional potential reasons for this neglect: first, the disci-
plinary origin of news values in Journalism/Communication Studies means that some
(critical) linguists are simply unfamiliar with the concept. Second, most (critical) linguis-
tic research tends to uncritically adopt the conceptualisation of news values from
Journalism/Communication Studies (e.g. as properties of events or selection criteria);
there is no real questioning or in-depth engagement with such conceptualisations. From
this, it follows that news values are not seen as relevant for linguistic analysis, but appear
to be regarded as outside the text. Third, there is no readily available linguistic frame-
work for analysing news values, in contrast to other well-developed linguistic tools such
as analyses of transitivity, active/passive voice, nominalisation, predication,
argumentation, modality, speech acts, metaphor, lexis, and so on, which are described in
general introductions to CDA and books on the critical analysis of news discourse
(Fowler, 1991; Richardson, 2007).
The first two points about the disciplinary origins of the concept and how it has been
treated outside its own discipline bring home not only the importance of interdisciplinary
research, but also its challenging nature. As is well known, critical linguistic approaches
(realist or not) take a constructivist approach to social reality. For example, social con-
cepts such as gender are clearly regarded in CDA as being discursively constructed, rather
than being properties of individuals. Given this constructivist stance, why do critical lin-
guists not also take such an approach to news values, regarding them as values that are
mediated through discourse?1 The answer may well be that it is more difficult to recon-
ceptualise concepts that originate from outside one’s own discipline and that one will tend
to adopt rather than reconfigure them. The final point – the lack of an appropriate system-
atic analytical framework for the linguistic analysis of news values – is what we want to
address in this article. Drawing and building on previous work by Bednarek and Caple
(2012a, 2012b), we introduce such a framework below. This is not meant to replace exist-
ing CDA tools such as transitivity analysis, but rather to be added to the toolbox.
The appendix presents a list of the resources that we have identified so far (revising
and updating Bednarek and Caple, 2012a, 2012b), with a focus on linguistic devices
only.2 This list is based on a survey of existing literature on news discourse (especially
Bednarek, 2006; Bell, 1991), as well as arising inductively from our analysis of a wide
range of English-language news stories. We also draw on research on evaluative lan-
guage (Bednarek, 2010; Martin and White, 2005). There is no claim here that these
resources would work across languages – this is clearly an area where we would encour-
age other researchers to develop their own lists. There is also no claim that these resources
are yet exhaustive – as we analyse more data, we will identify further resources; and as
we continue to teach students how to analyse news values, we will refine their categori-
sation and application. We invite anyone working with this framework to do the same.
The appendix lists the respective news value along with a brief definition in the left-hand
column, together with key linguistic devices in the right-hand column, exemplified by
authentic examples from the news.3
It must be stressed that this list of devices should not be taken as an automatic check-
list; rather, analyses should proceed in a context-based, interpretive way, using the guiding
question: Does this resource have the potential to establish aspects of actors and happen-
ings as newsworthy (e.g. as negative, novel, elite …) for the target audience? Are there
any other resources in the text that function in a similar way, even though they are not
listed in the appendix? Close attention needs to be paid to the likely potential effect/func-
tion of the linguistic resource as it is used in the text. Note also that we are not claiming
that such devices only function to construct newsworthiness – for instance, eyewitness
accounts and numbers construct truthfulness/credibility (Van Dijk, 1988b: 85, 93), tense/
aspect are important devices for constructing temporal relations between happenings, and
other devices contribute more generally to establishing aspects of the ‘“five W’s and an
H”: who, when, where, what, how, why’ (Bell, 1991: 175). In other words, these devices
can be multifunctional, but we focus here on their potential to construct newsworthiness.
Figure 1. Multimodal resources for the construction of Superlativeness (New York Post,
17 September 2010, p. 1).
Given that CDA has a history of in-depth, qualitative analysis, we assume familiarity
with this as a general approach and point the reader to Bednarek and Caple (2012b) for
an example analysis of news values and their foregrounding/backgrounding in a single
text. What we have not yet discussed in depth is how corpus linguistic techniques can
help with news value analysis.
Figure 2. Multimodal resources for the construction of Negativity (Metro, UK, 27 September
2013, p. 1).
Table 1. Possible stages in the corpus linguistic analysis of news values.
1 Analysis of frequent word forms, lemmas, clusters, including dispersion analysis and
qualitative analysis using concordancing/text view
2 Analysis of parts of speech using a tagged corpus: analysing those associated with
constructing a particular news value (e.g. personal pronouns → Personalisation;
Proximity; comparative/superlative, number, degree adverb → Superlativeness;
proper nouns → Personalisation; Eliteness; tense and aspect markers; adverbs of
time → Timeliness, etc.), including dispersion analysis and qualitative analysis using
concordancing/text view
3 Analysis of word senses using a semantically tagged corpus: analysing categories
associated with constructing a particular news value (e.g. from the USAS semantic
tagger: Damaging and Destroying → Negativity; Affect → Impact; Comparing →
Novelty; Consonance, etc.), including dispersion analysis and qualitative analysis using
concordancing/text view
4 Analysis of keyness concerning 1–3, with the help of an appropriate reference corpus
and/or comparison using ranked frequency lists (searching for big rank differences)
5 Analysis of collocates and phrases around significant words (e.g. words associated
with the topic)
6 Manual analysis of a subset of the corpus to identify additional resources for
constructing news values
7 Quantitative and qualitative analysis of a whole corpus for additional resources
are now 10 years old, the corpus, which we will call News2003, serves to illustrate the
approach, and we are more interested in investigating how corpus linguistic techniques
can help with news value analysis than in saying something about the corpus data per se.
Case study 1
A first step in much corpus linguistic analysis is to examine word frequency (of word
forms, lemmas, clusters). As far as news values are concerned, this can give us insight
into how happenings are ‘sold’ to us as newsworthy through conventionalised ways of
saying that are repeated frequently. We will focus here on the top 100 most frequent
words and two-word clusters (bigrams5) in News2003, without distinguishing if any
instances are attributed to sources or not, since both attributed and non-attributed text
contributes to the construction of newsworthiness. We will not provide a complete and
exhaustive analysis here, but just point to some of the key results and identify avenues
for follow-up analysis. Among the top 100, we can indeed identify many words and
phrases that could potentially be used in the text to construct news values.
Eliteness
First, there are the names of people and institutions that would have been considered
prominent in 2003 for many members of the British public, and which hence construe
Eliteness: Mr Blair/Blair (then prime minister); Duncan/Duncan Smith/Mr Duncan
(then leader of the opposition); Ferdinand (Manchester United footballer Rio Ferdinand);
Manchester United; the FA/the FA’s (the Football Association, the governing body of
English football). Borderline examples are Burrell (the former butler of Princess Diana)
and Mr Barrett (then chief executive of Barclays) – these might only be known/consid-
ered prominent by some readers, although we would expect there to be an introduction
with an explicit role label in the text at first mention. Concordancing shows that this is
indeed the case. Here, for example, are the immediate co-textual environments for the
first mention of Barrett’s name in the 10 newspapers:
1 The boss of Barclays Bank admitted yesterday that […]. The astonishing confession came
as Matthew Barrett […]
2 Matt Barrett, the man ultimately responsible for Barclaycard, Britain’s biggest credit
card business
3 […] the boss of Britain’s biggest credit card company yesterday stunned his customers
with […]. Matt Barrett, the £1.7m-a-year chief executive of Barclays, […]
4 […] Matt Barrett, the chief executive of Barclays […]
5 Chief executive Matthew Barrett […]
6 THE boss of Barclays Bank gaffed yesterday by admitting […]. Matthew Barrett –
Britain’s highest paid banker admitted that […]
7
Barclaycard boss Matt Barrett admitted yesterday that […]
8 THE boss of Barclays made an amazing gaffe yesterday […] Millionaire Matt Barrett
revealed […]
As can easily be seen, various role labels and descriptions (underlined) are used in all
10 newspapers to construct Barrett as an ‘elite’ news actor from the start.6 It is not sur-
prising that we can also find other role labels and titles in the top 100 from News2003:
chief/chief executive, prime minister, the princess, party (in all but seven of 81 instances
used in the sense of ‘political party’). Further concordancing of the lemma CHIEF
uncovers other titles such as: chief whip, police chief, chief executive(s), chief constable(s),
chief of the Paris vice squad, chief of staff, jewellery chief, Barclays chief, HBOS chief,
police chief(s), Manchester United chiefs, banking chiefs, party chiefs, or just chiefs (e.g.
appalled chiefs). With respect to dispersion/distribution, CHIEF occurs across eight of
the 10 topics and is used by all 10 newspapers. It may thus belong to a conventionalised
repertoire of rhetoric of newsworthiness in the British news.
Superlativeness
Second, the symbol # (standing for a number) and various bigrams including it, along
with the word forms and bigrams per cent, two, more/more than/than, the most, all and
only have the clear potential to construe Superlativeness. This would depend on how
high the numbers are (with numbers and per cent) and looking at the co-text. Even a low
number such as two can be used in expressions that construe Superlativeness, e.g. less
than two months earlier, having waited almost two months since, in the past two months
alone, fallen by more than two-thirds. Conversely, not every instance of than or only
would construct Superlativeness. For instance, concordancing of than reveals that some
of its usages are likely to construct Superlativeness (e.g. more than, larger/bigger/
cheaper than, no fewer than, less than TIME PERIOD), while others are not (e.g. rather
than).
Proximity
Third, the top 100 include the names of locations that would be considered geographi-
cally or culturally near by a British audience: England, Manchester, Greater Manchester.
The first-person plural pronoun we is also frequent, but we would need to undertake
concordancing to identify whether any instances are examples of an inclusive usage that
addresses a community that also includes the reader. Some grammatical words in the top
100 can partake in location references, e.g. prepositions such as at/in, and would also
deserve a closer look.
Negativity
Further, the negative word form attack and the names of groups that are likely to be nega-
tively evaluated by the audience (Islamic Jihad, the IRA) also turn up among the top 100
in News2003. There are also examples of words that might be associated with negativity
in other ways: police/the police (in terms of dealing with crime), credit/credit card/credit
cards (could potentially be associated with debt), the Israeli (we might hear about con-
flict). These word forms are not negative per se, but it would be worth identifying the
co-textual environments of these words/bigrams to see if any negativity is indeed set up.
Timeliness
The top 100 also include potential temporal references, namely yesterday, last/last night/
night. Yesterday and last night clearly construct Timeliness by locating the reported event
as recent. Concordancing illustrates that last in general often constructs Timeliness in
News2003, occurring in phrases such as last week/last week’s, last weekend, last Saturday.
But in other phrases it is used for temporal positioning with respect to events that are less
‘recent’: last month, last year/last year’s, last October, last November, last autumn, and
last is also used for other purposes (e.g. England’s last clean sheet). This reinforces the
importance of concordancing rather than a pure reliance on frequency results.
While the top 100 also include several instances of high frequency verbs, which are
marked for tense and aspect (e.g. has been, had been, is, are, being), an analysis of part-
of-speech tags would be necessary for a more accurate approach to Timeliness.
Prepositions such as at/in can also be part of temporal expressions and the top 100 modal
verbs will/would/could and associated bigrams can refer to the future (and might also
construct Impact). Further concordancing would be necessary to investigate these
grammatical devices in more detail.
As far as the foregrounding or backgrounding of news values is concerned, it must be
kept in mind that tense is a grammatical category in English and must hence always be
specified. What a systematic analysis can tell us, however, is first, whether or not there are
any time references that explicitly establish the event as recent/immediate or about to hap-
pen soon (e.g. last night, today, this weekend) and second, how far the respective happen-
ing is constructed as past, present, future and as impacting on the present (e.g. via aspect).
While news values such as Superlativeness or Novelty lend themselves to repetition (i.e.
they can be emphasised in different ways throughout a story), it is not conventional to
keep repeating an explicit time reference throughout a story, once that time has been
specified. Thus, repetition cannot be used as an indication of foregrounding in general.
Personalisation
Potential pointers to Personalisation are hard to identify, but we could look at those per-
sonal pronouns in the top 100 in News2003 that are only used to refer to human referents
(I, we, he, she, plus associated bigrams). Examining their co-text would tell us which
news actors they refer to. Other top 100 words such as people and Mr may also be useful
to look at because they might construct ordinary people as news actors.
Lower frequency items, such as the names of ordinary individuals (rather than politi-
cians, stars, etc.), can also point to Personalisation – for instance, the female name Deborah
can be found at rank 247 in the corpus. The proper noun DEBORAH occurs 39 times in
eight of the 10 newspapers, all concerning the same story (the sentencing of Mohammed
Dica, who was found guilty of infecting two women with the HIV virus). Deborah is
variously described as victim, naïve and unsuspecting – a mother of two children, who
‘fought back tears’, ‘sobbed’ and ‘wept with relief’. In such instances, the story is clearly
given a ‘human’, ‘personalised’ face through references to an ordinary individual and her
experience. In total, the tabloids use the proper noun DEBORAH 35 times (~10.7 per 10.000
words) and the broadsheets six times (~1.6 per 10.000 words), the difference being statisti-
cally significant (LL = 26.69). This may suggest that the tabloids emphasise the news value
of Personalisation in the construction of this happening more than the broadsheets, although
it does not capture anaphoric reference via the pronoun she, and other ways of referring to
the women (e.g. as his first victim), etc. There are also individual differences between news-
papers’ use of DEBORAH: Daily Mirror (12.32 hits per 1000 words) > Sun (12.20) > Daily
Star > (7.08) > The Guardian (5.23) > Financial Times (4.20) > Daily Telegraph (4.06) >
Daily Express (3.96) > Daily Mail (3.34). There are no occurrences in The Times and The
Independent. This could indicate that the newsworthiness of the same happening is con-
structed differently depending on the newspaper and its imagined audience.
Novelty
Among the top 100 bigrams only one is clearly used to construe Novelty: the first. Of 41
occurrences, 23 construe Novelty (Figure 3). Although they relate to only two of the 10
topics, they occur across all 10 newspapers in the corpus.
Interestingly, usages such as the ones exemplified in Figure 3 (e.g. the first + histori-
cal comparison) appear to belong to a common repertoire in news discourse (c.f.
Bednarek and Caple, 2012a: 241–242). To identify whether there are any other resources
in the corpus that construe Novelty, we would need to either consider lower frequency
items or work with semantic tags (e.g. the UCREL Semantic Analysis System (USAS)
category Comparing: usual/unusual). In fact, from the manual analysis of evaluation in
this corpus (Bednarek, 2006), we know that the corpus includes expressions such as
amazing, astonishing(ly), bizarrely, curious, dramatic(ally), extraordinary, fully,
sensational(ly), spectacular(ly), strikingly, stunning, unexpected(ly), unprecedented,
unusually, which function to express evaluations of unexpectedness.
Other top 100 word forms/bigrams and their potential contribution to the
construction of newsworthiness
As to be expected, the top 100 of News2003 also include a large amount of function
words and associated bigrams that we have not yet discussed. It is not straightforward
to associate such grammatical words with a particular news value, but they should not
be ignored either. For instance, the conjunction but and markers of negation (not, no,
only) are associated with counter-expectation and could in certain contexts be used to
construct Novelty. Similarly, to can occur in LEAD to (potentially constructing
Impact) and, together with up and numbers, can certainly construct Superlativeness
(Figure 4).
Other words/bigrams from the top 100 that would be worth investigating further
include Mr, said, told, plus various bigrams with said and one with added, all of which
do not construe newsworthiness per se, but could nevertheless yield interesting insights
into news values: Mr would be followed by the name of the respective news actor and the
reporting verbs would be associated with particular speakers. We could then analyse
these news actors/sources further for Eliteness or Personalisation. We could go even
further and identify all reporting expressions (nouns, verbs, etc.) in the frequency list and
examine their associated sources.
Even seemingly innocuous words can be interesting to explore. Content words that
are frequent in a corpus indicate an important topic – analysing such words in their co-
text can then show us what news values are associated with them. Candidates in the top
100 in News2003 include letter, hotel, test, car. Looking at the dispersion of car, for
example, shows that it occurs in 24 texts (five out of 10 topics), although most occur-
rences are in two stories – one about a suicide bomber and one about Princess Diana.
Concordancing shows us how car is being talked about in these stories:
•• accident/crash: car accident, car crash, car smash, car blast, and references to
plot an accident and a death in a car; drive car into hotel, suddenly turned into the
hotel, smashed into, careered into
Here, we can easily see the construction of the news values of Negativity (nega-
tive lexis, e.g. accident, fired, car bomb) and Superlativeness (intensified lexis, e.g.
smash, blast, hurl, engulf, blaze).7 Thus, it may be useful to produce concordances
for the most frequent content words in the text/corpus, as this will offer valuable
insights into how a particular happening or news actor is constructed as newsworthy.
The value of this approach depends on the corpus construction – for instance, the
words car, letter, hotel, test are in the top 100 of News2003 because the same 10
topics are repeated across the corpus. If each of the 100 texts covered a different
topic, these words would not necessarily be among the top 100, as they might be
pushed down the frequency list by words that are repeated across topics (the same
goes for some of the other devices identified as frequent above, e.g. police,
Manchester). But this approach should work for corpus-linguistic CDA studies,
since these often concern one particular topic (e.g. Europe, in Mautner, 2000, refu-
gees/asylum seekers, in Baker et al., 2008; Islam/Muslims, in Baker et al., 2013),
and topic-specific corpora are likely to contain frequent topic-specific content words.
In addition to concordancing such words, automatic collocation and cluster analysis
can also be undertaken (cf. Table 1).
To sum up this section, we have illustrated how a researcher might proceed with cor-
pus linguistic techniques and have discussed how such techniques could help with news
value analysis. Since we used a non-topic-specific corpus here, we gained first insights
into a conventionalised repertoire of rhetoric of newsworthiness in the British news
rather than into how a particular happening’s newsworthiness was constructed. A much
larger, more representative corpus would be necessary to confirm the conventionalised
nature of the identified resources and to add others.8 What is missing from the above
frequency analysis of News2003 is a comparative element. It is generally useful to com-
pare frequencies between news texts or corpora. For instance, in a previous investigation
(Bednarek and Caple, 2012b) we compared the top 20 frequent word forms in one par-
ticular news story (about the 2011 Queensland floods) with the top 20 frequent word
forms in the one-million-word news subset of the BNC Baby (four million words of
British English, extracted from the British National Corpus). Exploring the co-text of
those words that occurred much more frequently in the flood story (in terms of rank)
proved highly productive for news value analysis. Another way in which a comparative
element can come in is through automatic keywords analysis. This corpus linguistic tech-
nique provides the researcher with a list of key words or clusters in a text or corpus –
those words/clusters that are, statistically speaking, unusually frequent (or infrequent) in
relation to a comparison corpus. We undertook such a keywords analysis for our second
case study.
Case study 2
While the first case study used a corpus that is perhaps slightly atypical of CDA because
it covers several different topics, the second case study on keywords will briefly consider
stories about a particular happening. For this keywords analysis case study, we compared
10 stories about a suicide bombing in Iraq in News2003 with another 90 stories about
nine other topics. Without going into detail or offering up an exhaustive analysis, the
keywords that the software identified are indeed highly informative as far as newswor-
thiness is concerned. They point to Negativity (blast, explosion, exploded, bomb, attacks,
killed, soldiers, wounded, injured, suicide, bomber, attack…) and Eliteness (US, CIA,
Bremer, Krivo, American, Americans, officials, governing, administrator, agents . . .).
Names such as Bremer and Krivo, which are likely to be unfamiliar to many in the audi-
ence, are accompanied by titles emphasising their eliteness (similar to the Matt Barrett
example earlier) (Figures 5 and 6).
Interestingly, the word form ‘al’ is also a keyword. Concordances show that this is used
to refer to people or organisations (e.g. A member of the US-appointed Iraqi Governing
Council, Mouwafak al-Rabiyeh; Al-Qaeda), to localise the event (e.g. crashing down into
al-Sadoun Street; Some of the wounded, [. . .] were taken to nearby al-Kindi hospital), and
also occurs in the co-text of eyewitness accounts (e.g. Ghaith al-Qassi, who was sitting at
a tea shop 100 yards away, said: […]). A close reading of the 10 stories reveals that all
newspapers except the Financial Times cite at least one eyewitness. In summary, the key-
words analysis appears to suggest that overall this particular event was constructed as
negative and as involving elite news actors and that Personalisation (the human face of the
event, through eyewitness accounts from ‘ordinary’ people) also plays a role. That is, the
British media sold the event to its readers by emphasising these particular news values,
rather than others, although there appear to be differences between the popular/quality
press and individual newspapers. A dispersion/distribution analysis and in-depth manual
analysis would be necessary to identify such differences in detail.
Concluding remarks
We hope that this article has achieved our three aims: to raise awareness of news values
in CDA; to promote a constructivist approach to such values; and to introduce a new
analytical framework that can be used in their analysis. Because of their ideological
nature, we believe that a systematic analysis of news values should belong to the
standard procedure of critical linguistic analyses of the news.
Our first case study demonstrated that a systematic linguistic analysis can tell us what
kind of discursive devices are repeatedly used in, say, the British press, to construct dif-
ferent news values and to perpetuate the ideology of newsworthiness in itself – namely,
that negativity is interesting, that elite people are worth listening to, that what is near to
us is more important than what is far from us, and so on. It could be interesting to under-
take a diachronic analysis of news discourse over time to investigate any changes in this
general ideology and in its conventionalised rhetoric.
As our second case study indicated, we can use a linguistic analysis of news values
for a specific topic, event or news actor to establish how they are constructed as news-
worthy. This may also work to perpetuate existing ideologies, as when there is a particu-
larly high amount of conflict stories about Islam/Muslims, as found by Baker et al. (2013:
258):
[…] bad news stories tended to be viewed as having a higher news value than good news
stories. This is one argument that editors could make: that it is not that newspapers are
Islamophobic, [it’s] just that the media’s role is to report on bad news. However, to counter this,
we would point out that, when we compared our corpus of stories about Islam with other
corpora of more general news in Chapter 2, we found that there were more references to conflict
in the Islam corpus – a difference that was statistically significant. Even taking into account the
general press tendency to focus on bad news, the amount of conflict stories regarding Islam and
Muslims looked suspiciously high.
It is clear that Baker et al. (2008) are not taking a discursive approach to news values
here, but, interpreting their result from our perspective, we could say that the news value
of Negativity appears to be discursively foregrounded in such stories, more so than in
other news. (Our news value of Negativity includes what others have called the news
value of Conflict; see Cotter, 2010: 69.) But a systematic analysis of all news values,
including comparison with general news, would be necessary to find out how else the
construction of newsworthiness regarding Islam/Muslims differs from or is similar to
other subject matter. Indeed, some of the most frequent words in Baker et al.’s (2008)
corpus (see pp. 53–54) may point to the construal of particular news values. Thus, one
application of this framework lies in investigating whether a given topic is repeatedly
associated with particular news values and, if so, what the effect of this might be.
What are other potential uses of this framework for CDA? It would be interesting to
analyse how news cycles impact on the construction of newsworthiness and whether or
not different news values are emphasised, depending on whether we are dealing with
breaking news or follow-up reporting and analysis. With a longer timescale, we could
analyse the news values that are constructed across decades with respect to particular
issues, happenings or news actors, for instance gay marriage or the lesbian, gay, bisexual
and transgender (LGBT) community in general. Cross-cultural comparisons are also
possible, which would uncover whether news outlets from different countries construct
the newsworthiness of one event using the same news values.
If the researcher’s analytical stance is of a realist persuasion, they might be interested
in exploring the extent to which an event is made more newsworthy than it ‘deserves’,
which ties in with research on sensationalism/media panics (Fowler, 1991; Molek-
Kozakowska, 2013). Other researchers might be interested in investigating a particular
news outlet or output, such as the Sun newspaper, Fox News, BBC News, and so on.
How does this particular paper, organisation or programme sell the world to us as news-
worthy, and what ideologies are perpetuated? We could also undertake a systematic com-
parison of ‘popular’ versus ‘quality’ news (see Conboy, 2006: 15–16, on tabloid news
values).9
It is also possible to use news values analysis to gain insights into audience position-
ing: any story would be written in such a way as to sell the happening as newsworthy to
the target audience, imagining that they are interested in the particular construction of
newsworthiness that is applied. This also ties in with research into different news outlets
and outputs, as they each have their own imagined target audience.
To conclude, we hope that our two case studies have demonstrated that corpus linguis-
tic analyses of frequency (of word forms, clusters, lemmas) can provide a useful indica-
tion of the discursive construction of newsworthiness in a given text or corpus, without
being complete or exhaustive. This analysis should be complemented both by other cor-
pus analyses (POS tags, semantic tags) and in-depth discourse analysis, not just of lan-
guage but also of other semiotic systems. Ideally, an analysis of news values would
incorporate several strands:
Building on our existing work on language and images, we are currently working on
creating such an integrative approach, which we tentatively label corpus-assisted multi-
modal discourse analysis (CAMDA).
Acknowledgements
Our thanks go to Teun van Dijk for useful input on an earlier version of this article. This article was
produced while we were both on a Visiting Fellowship at the Reuters Institute for the Study of
Journalism (The University of Oxford) and we are grateful to the institute for this opportunity.
Funding
This research received no specific grant from any funding agency in the public, commercial or
not-for-profit sectors.
Notes
1. Incidentally, Cotter (2010: 86) argues that sociolinguists can regard the operation of news val-
ues ‘in the same sense as social variables like age, gender, and occupation’ (p. 86), although
she herself treats news values as becoming ‘embedded’ in the text. But a view where gen-
der and other social variables become ‘embedded’ would sit at odds with a performative or
constructivist view of gender.
2. Further explanation of news values, together with a discussion of the role of images, is pro-
vided in Bednarek and Caple (2012a, 2012b) and Caple (2013), and this article is best read in
conjunction with these publications on which it builds.
3. These news values are inspired by Bell’s ‘values in news actors and events’ (Bell, 1991:
156, italics in original), although our definitions differ slightly from his (e.g. we more or less
include Relevance in Impact, and Attribution in Eliteness) and we have excluded Unambiguity
and Facticity. Unambiguity (clarity) is more of a general objective of news writing for us,
rather than having to do with newsworthiness per se, and we are somewhat unsure about the
status of Facticity (the extent of facts and figures in stories, according to Bell, 1991: 158). For
us, this seems to be more about creating general credibility/truthfulness (see Van Dijk, 1988b:
85, 93), rather than establishing the happening as newsworthy (although figures are also used
to create Superlativeness and are an important resource for that news value). Thus, Facticity
seems more of a ‘meta’-value – that is, it concerns the amount of evidence gathered ‘about’ a
story and the credibility of the news organisation. The resources that are listed and examples
provided are non-exhaustive. For example, it is likely that most Graduation resources (Martin
and White, 2005) that upscale (increase the number/amount, size, duration, force or degree of
aspects of an event) would construct Superlativeness, and the whole language of cause–effect
relations could be relevant for Impact. Different sub-categorisations or groupings of devices
would also be possible.
4. The idea of using semantic tagging to analyse news values was first suggested to us by
Amanda Potts.
5. Bigrams were not calculated across sentence breaks. An alternative approach that could be
pursued would involve looking at those word forms and bigrams that have the most range (i.e.
are the most distributed or dispersed among all the texts in a given corpus). Once the most
common conventionalised expressions for the construction of newsworthiness have been
identified (in a very large corpus of news reportage), it will also be possible to start search-
ing for these expressions directly. For instance, we know that first and new are associated
with Novelty, and indeed first occurs at rank 126 (in 38 texts) and new occurs at rank 163 (in
39 texts, with only three delexicalised instances: New Year, New York, New Labour). Thus,
a more inductive frequency analysis (if limited to the top 100) would miss these features –
although the bigram the first was identified.
6. The many references to Barrett’s status may also construct Novelty (a millionaire/the head of
Barclays not using credit cards), as well as criticise banks in general for being so expensive.
7. Of the remaining six instances of car, four can be associated with happenings that would be
evaluated as negative (e.g. racist behavior, being intoxicated, going on a shopping trip instead
of taking a drugs test) and two are neutral (e.g. car hire).
8. An unrelated study of word frequency in a corpus of about 2.6 million words of British tab-
loid and broadsheet news reports (from a wide range of sections) did indeed find that words
such as last, first, yesterday, England, minister and police, are statistically speaking, more
frequent in news reportage than in other varieties of English (Bednarek, 2008: 29).
9. Some of the top broadsheet/tabloid keywords in Baker et al.’s (2013) ‘Islam’ corpus could
certainly be interpreted as constructing different news values. For instance, the authors
note that the tabloids’ more frequent use of family/relationship words (e.g. mum, wife, kids)
‘suggests an emphasis on news stories that are at a more individual and personal level’
(Baker et al., 2013: 75). From our discursive perspective, we could rephrase this as an argu-
ment that the news value of Personalisation is discursively foregrounded in the tabloids in
this corpus.
References
Baker P, Gabrielatos C, KhosraviNik M, et al. (2008) A useful methodological synergy? Combining
critical discourse analysis and corpus linguistics to examine discourses of refugees and
asylum seekers in the UK press. Discourse & Society 19(3): 273–306.
Baker P, Gabrielatos C and McEnery T (2013) Discourse Analysis and Media Attitudes: The
Representation of Islam in the British Press. Cambridge: Cambridge University Press.
Bednarek M (2006) Evaluation in Media Discourse: Analysis of a Newspaper Corpus. London/
New York: Continuum International Publishing Group.
Bednarek M (2008) Emotion Talk across Corpora. Basingstoke/New York: Palgrave Macmillan.
Bednarek M (2010) Evaluation in the news: A methodological framework for analysing evaluative
language in journalism. Australian Journal of Communication 37(2): 15–50.
Bednarek M and Caple H (2012a) News Discourse. London/New York: Bloomsbury Publishing
(formerly Continuum).
Bednarek M and Caple H (2012b) ‘Value added’: Language, image and news values. Discourse,
Context & Media 1(2–3): 103–113.
Bell A (1991) The Language of News Media. Oxford: Blackwell Publishing.
Caple H (2013) Photojournalism: A Social Semiotic Approach. Basingstoke/New York: Palgrave
Macmillan.
Conboy M (2006) Tabloid Britain: Constructing a Community through Language. London/New
York: Routledge.
Cotter C (2010) News Talk: Investigating the Language of Journalism. Cambridge: Cambridge
University Press.
Fairclough N (1995) Media Discourse. London: Bloomsbury Academic.
Fowler RG (1991) Language in the News: Discourse and Ideology in the Press. London: Routledge.
Galtung J and Ruge MH (1965) The structure of foreign news. Journal of Peace Research 2(1):
64–91.
Lippmann W (1922) Public Opinion. New York: Harcourt Brace (reprint 1965).
Machin D and Mayr A (2012) How to Do Critical Discourse Analysis: A Multimodal Introduction.
London: SAGE.
Martin JRR and White PRR (2005) The Language of Evaluation: Appraisal in English. Basingstoke/
New York: Palgrave Macmillan.
Author biographies
Monika Bednarek is Senior Lecturer in Linguistics at the University of Sydney, with a research
interest in the linguistic analysis of media discourse. Her most recent books are: News Discourse
(Bloomsbury, 2012, with Helen Caple), Telecinematic Discourse (John Benjamins, 2011, co-
edited with Roberta Piazza and Fabio Rossi) and The Language of Fictional Television (Continuum,
2010).
Helen Caple is Senior Lecturer in Media, Communication and Journalism in the School of the Arts
and Media at the University of New South Wales, Sydney. Her primary research interests centre
on press photography and text–image relations. She is the author of Photojournalism: A Social
Semiotic Approach (Palgrave Macmillan, 2013) and co-author of News Discourse (Bloomsbury,
2012, with Monika Bednarek). She has also worked as a press photographer in the UK.
APPENDIX. (Continued)
APPENDIX. (Continued)
APPENDIX. (Continued)
*Just can express a range of meanings such as ‘only’ or ‘exactly’, and not all of its usages are used to intensify
(cf. the invented example: he’s just a student). Time relations of frequency/duration can also work to intensify
or quantify.
**Newness also has temporal aspects and could be grouped with Timeliness; the word novelty itself is
ambiguous, indicating ‘newness’ or ‘unusuality’.