See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/283713718

Alba-Juez, Laura (2017). Evaluation in the headlines of tabloids and

broadsheets: A comparative study. In R. Breeze (ed.) Evaluation in media
discourse: European perspectives. Berl...

Article · November 2017

CITATIONS READS
0 592

1 author:

Laura Alba-Juez
National Distance Education University (UNED), Madrid, Spain.
47 PUBLICATIONS   111 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Studies on (im)politeness View project

Emotion and Language 'at work': The discursive emotive/evaluative function in different texts and work contexts View project

All content following this page was uploaded by Laura Alba-Juez on 12 November 2015.

The user has requested enhancement of the downloaded file.

 Evaluation in the headlines of tabloids and
broadsheets: A comparative study1
LAURA ALBA-JUEZ 

1. Introduction

This paper presents a comparative study of the use of the evaluative

devices used in the headlines of four different on-line British
newspapers: two broadsheets, BBC Online and The Guardian, and two
tabloids The Mirror and The Daily Mail. Some authors (e.g. Fowler
1991, Convoy 2004) have written about the distinguishing
characteristics of tabloids as opposed in many ways to those of
broadsheets. The main motivation leading to this study originated in the
observation that all newspaper articles (whether in tabloids or
broadsheets) seem to contain a given evaluation that is generally
condensed or made relevant in their headlines, where the main stance
of the article is thus reflected, a stance that may influence the reader’s
decision to make the effort of reading the whole article, or on the
contrary, to reject it in the first place.
The analysis carried out herein is multimodal, as not only the text of the
headlines of the articles, but also the images which very frequently
accompany them, were analyzed, given that these images interact with
the text in various and complex ways, and that consequently they may
also contribute to the evaluative meaning as a whole, as other authors
(e.g. Breeze, 2014) have pointed out. The headline semiotic space in an
article (including both text and pictures) normally seems to synthesize
the evaluative stance of the whole article, for it usually contains the
                                                            
1 The research done for this paper has been funded by the Spanish Ministry of

Economy and Competitiveness (MINECO), through the Emo-Fundett project

(FFI2013-47792-C2-1-P). I also want to sincerely thank Geoff Thompson for his
valuable comments on the first manuscript, as well as Lachlan Mackenzie for his
advice regarding the treatment of the syntactic constructions found in the headlines.
abstract (Labov, e.g. 1997) of the narration, and as Dor (2003: 695)
observes, it is “designed to optimize the relevance of their stories for
their readers”. Thompson and Hunston (2000: 19) note that even though
it is clear that evaluation tends to be found all throughout a text, there
are some strategic points where the pervading evaluation of the whole
text is condensed and can be easily identified. In the particular case of
journalistic articles, the headline seems to be one of those strategic
points.

Another of the motivations for the present study has arisen from the
observation that in 21st century online communication, the
‘conversationalization’ (Fairclough, 1995) of media discourse is
apparent in all the media, a fact which some authors believe to have
contributed to blur the distinctions between broadsheets and tabloids to
a certain extent. Connell (1998), for instance, writes about the
phenomenon of ‘tabloidisation’ to designate the process by which
supposedly rationalist discourses are transforming into sensationalist
ones. This is a homogenizing view which maintains that the journalism
which once seemed to be confined to the media for lowbrow consumers,
has now permeated all the other media. It should also be noted that the
old physical distinction between broadsheets and tabloids concerning
their size, i.e. the fact that broadsheets had ‘broader sheets’ than
tabloids, is no longer true (The Times and The Independent are tabloid
size, for instance), and for that reason the distinction is sometimes now
made by using the terms ‘quality’ vs. ‘popular’ newspapers. However,
in this work I will not use the latter pair of terms, considering that in
spite of the fact that the size distinction is no longer valid, in general the
terms ‘tabloid’ and ‘broadsheet’ have been maintained, and are
commonly used by English speakers. The comparative study carried out
herein, however, will exclusively look into the ‘evaluative blueprint’ of
both types of headlines, irrespective of other similarities or differences
that might exist between both types of press.
The main research question of this work, therefore, aims to elucidate
whether or not there are significant differences between broadsheets
and tabloids in the expression of evaluative meanings in the semiotic
space occupied by the headlines. Precisely, one characteristic usually
attributed to tabloids in contrast with broadsheets is that they focus
more on the emotional side of stories. Fowler (1991) has pointed out
that in the tabloids evaluation seems to be used as one more of the
linguistic and ‘poetic’ devices employed to build a sense of community
among their readers. Conboy (2004) remarks that an important feature
of the English used in tabloids is that “it shifts language from reporting
to an engaged and often enraged personalization of the political sphere”
(2004: 47), a shift that is prototypically carried out by means of the
stances taken and expressed by the journalists and/or the tabloid as a
whole. Furthermore, if we ask the layperson or search the internet for
opinions about the differences between a broadsheet and a tabloid, we
shall find that in most cases broadsheets are considered ‘more serious
and objective’, while tabloids are seen as ‘less serious and more
subjective’. Regarding the headlines of both, it is believed that those of
the tabloids are syntactically shorter and simpler, more dramatic and
emotive, and more biased. Those of broadsheets, in contrast, are
considered to be longer and syntactically more complex, truthful and
unbiased (O’Connor, 2011).

The definition of evaluation adopted for the analysis is Alba-Juez &

Thompson’s (2014: 13), which views evaluation as a dynamical,
intersubjective phenomenon that permeates all levels of linguistic
description, but which cannot be described or analyzed in ‘purely’
(whatever that may mean) linguistic terms, for it always occurs within
a given personal/ social/ emotional/ cultural/ etc. context, and therefore
reflects the values of a person, group, society or culture:

We […] define evaluation as a dynamical subsystem of

language, permeating all linguistic levels and involving the
 expression of the speaker’s or writer’s attitude or stance
towards, viewpoint on, or feelings about the entities or
propositions that s/he is talking about, which entails relational
work including the (possible and prototypically expected and
subsequent) response of the hearer or (potential) audience. This
relational work is generally related to the speaker’s and/or
hearer’s personal, group, or cultural set of values.

In the case of our corpus, the values inscribed or invoked (to use Martin
& White’s (2005) terminology for overt vs. covert evaluation,
respectively) in the headlines will normally not be ascribed to the
journalist’s own personal set of values, but to the newspaper, group,
society or culture s/he is representing. On occasions the headline just
poses a question, which might be interpreted as the decision on the part
of the editors to let the readers draw their own evaluative conclusion,
although these questions may be biased and therefore invoke or try to
provoke in the readers a given evaluation of the topic or people depicted
in the article. This constitutes one of the factors of the intersubjectivity
and dynamicity of the phenomenon, and raises the issue of the “Russian
doll” dilemma discussed by Thompson (2014), faced by the analyst
when having to categorize the type of evaluation found in any given
utterance or text. The dilemma has to do with the fact that an expression
of one category may function as a token (i.e. an indirect expression) of
a different category (and that token may even function as an indirect
expression of yet another category, and so on). Furthermore, as the
results of this study will show, in the same headline we may find co-
occurring inscribed evaluations having opposite polarity values, or the
picture accompanying the headline may contain an evaluative message
that is not totally in sync with that of the headline. I have certainly been
confronted with this dilemma not only at the stage of the qualitative
analysis, but more importantly, when trying to quantify the frequencies
of occurrence of some of the evaluative variables analyzed, such as the
linguistic level at which the evaluation is manifested, and, more
importantly, the subsystem of evaluation (within Martin & White’s
(2005) Appraisal model) where each particular headline should be
placed. In the upcoming sections I shall try to show how I have dealt
with this and other issues which inevitably showed up in the course of
research.

The research presented in this paper, therefore, is oriented towards

‘cracking the evaluative code’ of both broadsheet and tabloid headlines,
so as to be able to compare what might be called their ‘evaluative
blueprint’. The method used to achieve this aim involved looking into
the syntactic constructions used in the headlines and, more importantly,
the different variables that form part of the ‘evaluative functional
relationship’ (Alba-Juez, forthcoming).

2. Corpus and methodology of analysis

The corpus on which the analysis is based consists of 100 articles from
broadsheet newspapers (50 from BBC Online and 50 from The
Guardian), and 100 articles from tabloids (50 from The Mirror and 50
from The Daily Mail), published between October 2014 and May 2015.
The main research question was the following: Are there any significant
differences, both qualitatively and quantitatively, between the
evaluation found in the headlines of broadsheets and tabloids?
The act of evaluating someone or something can be done along several
different parameters, because there are many different variables that
intervene and interact in the phenomenon. The methodology followed
in this study aims at shedding light onto what seem to be the most
relevant knowledge resources or variables of the evaluative function of
language. Thus, the study consists in the comparison of the evaluation
found in the headlines of broadsheets and tabloids, by examining (both
qualitatively and quantitatively) the following aspects: 1) the syntactic
constructions used in the headlines; 2) the different ‘ingredients’ or
qualitative variables of the Evaluative Functional Relationship (Alba-
Juez, forthcoming), whereby evaluation is treated as a function of a
number of variables that interact with one another. In mathematical
symbols, the evaluation equation can be represented as consisting of
two terms (E and F), the variables being the arguments of the functional
relation F, as in (1):

(1) E = F(PH, LL, Deg, ContPos, P, Mo)

The arguments (or knowledge resources) of the evaluation equation are

the following: 1) Phase of the evaluation (PH); 2) Linguistic Level at
which the evaluation is realized (LL), 3) Degree of (in)directness of the
evaluative meaning (Deg), related to whether the evaluation is overt or
covert, 4) Position along the evaluation continuum (ContPos), which
has to do with the polarity of the evaluative act; 5) (main) Parameter of
evaluation (P), which in this study will be drawn from the Appraisal
model categories (Martin & White 2005); and 6) Mode of the evaluation
(Mo), a variable which contemplates the modes through which the
evaluation is expressed (e.g. linguistic, paralinguistic, or/and through
other visual/auditory means (images, videos, emoticons, etc.). These
are all qualitative variables that are thought to be (un)consciously
considered and appropriately weighed by the interlocutors for either the
expression or the interpretation of their evaluative act during discourse.
In the upcoming sections of this paper we shall explore each one of
them in relation to the newspaper headlines in the corpus. The
qualitative analysis and comparison of both broadsheet and tabloid
headlines will be complemented by a frequency analysis where the
significance of the differences (related to the variables) found between
one type of press and the other will be tested by means of the χ2 test,
given the fact that the variables are qualitative, and consequently the
statistical test used has to be of the non-parametric kind.
 3. Analysis

3.1. Syntactic constructions used in the headlines

Before delving into the analysis of the six variables contemplated in the
evaluation equation in (1), we shall look into the syntactic constructions
used in the headlines in order to test what seems to be the common
belief that broadsheet headlines are longer and more syntactically
complex than those found in tabloids (e.g. O’Connor 2011). This is
deemed necessary because the length or complexity of the headlines can
have certain effects on their evaluative content. One might think, for
instance, that if the headlines are longer and more syntactically
complex, they will contain richer and more varied evaluative forms and
meanings, or that there will be more room for semantic or discourse
prosody2 in a longer and more complex headline than in a shorter and
simpler one. My analysis, however, does not seem to confirm such
common belief. In order to compare the complexity of the headlines, I
analyzed them syntactically, and I found that some of the headlines
contained only phrases, while others contained simple, complex or
compound clauses3, or a combination of all these possibilities. My
assumption was that complex and compound clauses would be longer
and syntactically more complex. Thus I divided them into eight main
categories, based on what I found, as shown in Table 1, where their
frequencies of occurrence and one example of each are exhibited:

                                                            
2 The phenomenon of semantic/ discourse prosody refers to the frequent occurrence, in
a given text, of words or expressions containing the same evaluative valence (either
positive or negative) in relation to a certain expression that at first sight might appear
to be neutral.
3 I have used Downing & Locke’s (2006) definitions for the classification: “The

simple sentence consists basically of one independent clause…[…] The compound

sentence consists basically of two independent clauses, linked in a relationship of
coordination” […] The complex sentence consists basically of one independent clause
and one dependent clause, linked in a relationship of dependency…” (pp. 272-273).
Table 1: Comparison of the frequencies (and examples) of the main
categories of syntactic constructions found in the headlines
Type of Broad- Tabloids Examples and
construction sheets syntactic analysis
PHRASES 17 6 Charlie Hebdo hunt: Double
hostage crisis in France4

NP: NP[PrepP]5
SIMPLE 47 25 Andrew Flintoff makes a duck
CLAUSES on debut in Australia’s Big
Bash6

SPOCirc
COMPLEX 15 34 Cheaper oil could damage
CLAUSES renewable energies, says
Richard Branson7

[OSPO]PS (Reported
speech)
COMPOUND 2 10 Shocking crash splits car IN
CLAUSES HALF - but family somehow
walk away UNHARMED8

SPOCirc+but+SCircPCs
COMPOUND/ 0 9 'She passed away in my
COMPLEX arms': Mother tells how
healthy teen who died from
virulant mutated flu was
struck down by the strain
sweeping America9

SPCirc: SP[Owh-
cl[SPAg]]

                                                            
4 BBC Online, 9 January 2015.
5 See notation used in the Annex.
6 The Guardian, 21 December 2014.
7 The Guardian, 16 December 2014.
8 The Mirror, 23 November 2014.
9 The Daily Mail, 2 January 2015.
 PHRASES + 15 8 Charlie Hebdo: Gun attack
SIMPLE on French magazine kills
CLAUSES 1210

NP: SPO
PHRASES + 3 5 Brewers Unfayre: Family
COMPLEX told to pay £400 in parking
CLAUSES fines because their Christmas
lunch in the pub took too
long11

NP: SP[O to-inf

cl][CircsubSPCirc] (Elliptical
passive without be)
PHRASES + 1 3 Jesus Christ Superfast...
COMPOUND Andrew Lloyd Webber calls
CLAUSES for wi-fi in EVERY church -
and tells of his love for Cats
star Nicole Scherzinger12

NP+ SPOCirc+and+(S)PO
TOTAL 100 100

As can be seen just by looking at the raw frequencies, contrary to what

seems to be the common belief, the broadsheets contain more instances
of phrases and simple clauses than the tabloids, while there is a higher
frequency of complex and compound clauses in the tabloids. The results
of the statistical χ2 test13 show that these differences are significant, with
a p value of 0.00001 at p < 0.05.
But in order to check the complexity of the constructions, it was also
thought necessary to count the number of constituents found in each
one of the headlines14. The types of syntactic constructions found in
                                                            
10 BBC Online, 7 January 2015.
11 The Daily Mail, 4 January 2015.
12 The Daily Mail, 3 January 2015.
13 The tables obtained for the χ2 values are not included here for space reasons.
14 When counting the constituents, I included those of the main clause and those found

down to the first level of embedded clauses in complex sentences.

both broadsheets and tabloids were manifold, displaying a great variety
of combinations of constituents. Table 2 shows the number of
headlines, both in the broadsheets and the tabloids, that have a given
number of constituents (between 1 and 13). So, for instance, the table
shows that 27 of the headlines in the broadsheets had 3 constituents,
and that there were 15 headlines in the tabloids with 7 constituents. In
general, it can be seen that the tabloid headlines present a tendency
towards a higher number of constituents, as in effect, the χ2 test
significant results show (p value 0.00001 at p<0.05).

Table 2: Number of syntactic constituents in the headline clauses

Nº OF BROADSHEETS TABLOIDS
CONSTITUENTS IN
THE HEADLINES
1 7 3
2 12 5
3 27 1
4 27 25
5 15 14
6 6 15
7 5 15
8 0 10
9 1 6
10 0 4
11 0 0
12 0 1
13 0 1
TOTAL 100 100

It was also deemed necessary to look at the number of words in the

headlines, considering that the length was one of the characteristics
attributed to broadsheets which could also contribute to their
complexity. The results are shown in Figure 1, where it is obvious that
the majority of the broadsheet headlines have fewer words than the
tabloid ones, a result that was corroborated by the χ2 values obtained (p
value of 0.00001).

Figure 1: Nº of words in the headlines of broadsheets and tabloids

Nº of words in the headlines
120
98
100
80
60 50 50
40
20
2
0
5 to 15 words 16 to 33 words

BROADSHEETS TABLOIDS

The results shown in Tables 1 and 2, thus, present evidence to the

contrary of some authors’ previous assumptions that broadsheet
headlines are longer and more syntactically complex than tabloid ones.
Furthermore, it has also been observed that, even though both
broadsheets and tabloids include subheads which are longer and more
explanatory than the main headline, the subheads in the tabloids tend to
be longer, and in the case of The Daily Mail, the customary style is to
include a list of subheads, not just one15.
And precisely because of the length and higher syntactic complexity of
the tabloid headlines, it was found that the analysis of the evaluation
found in them was also more complex, for they normally include more
evaluative categories (nested or interrelated) than the broadsheet ones.
But irrespective of the length or complexity of the headline, in both

                                                            
15 The analysis of the subheads, however, goes beyond the scope of this paper.
broadsheets and tabloids the syntactic constructions used seem to serve
certain and various evaluative purposes. It was observed, for instance,
that the use of reported speech is not the only means used to allow for
dialogistic communication; many times journalists resort to
interrogative clauses instead of declarative ones, an option that is
equally located within the evaluative subsystem of Engagement (Martin
& White, 2005) of the Heteroglossic kind.
We shall now proceed to describe and examine the different variables
of which evaluation is a function, according to the methodology
proposed and anticipated in 2 above.

3.2. Phase of the evaluation (Ph)

In Alba-Juez & Thompson (2014) we identified three phases of

evaluation in discourse: 1) the pre-realization phase (a purely cognitive
but active stage, in the sense that the speaker has a certain stance and/or
emotion s/he may choose to express in a verbal way or not); 2) the
textual phase (the actual verbal expression of the evaluation); and 3) the
metaevaluative phase (the interlocutor’s reaction phase, where s/he
expresses his/her stance related to what was said, which simultaneously
constitutes another textual phase)16. In this study the analysis was
centered on the textual phase, given that, on the one hand, it is
impossible to know the nature of the evaluation in the pre-evaluative
phase of the writer, and on the other, we have no access to the reaction
of the audience or readers, unless I made an analysis of my own
reactions when reading the headlines, which is neither pertinent nor
objective, and would besides go beyond the scope of the paper.

Thus in all 200 headlines analyzed, the focus of analysis was the
Textual Phase. The word textual here is used in its broader, discursive

                                                            
16
The characterization of the phases has been simplified here for space reasons. For a
more detailed account of the phenomenon see Alba-Juez & Thompson (2014).
sense, which entails looking into both text and context, the context
including not only the linguistic, but also the paralinguistic or pictorial
context given by the images accompanying the headlines, as well as any
other type of context (social, cultural, etc.) available to the reader of
these newspapers. Without the information given by the different types
of context, it would be very difficult, if not impossible, to interpret the
evaluative meaning of any message, including the headlines under
scrutiny. For instance, in the headline Russia says drivers must not have
'sex disorders'17, apart from the Heteroglossic Engagement made
evident by the use of reported speech, which makes it very clear that
this is not what the journalist or the newspaper says, we find the use of
quotation marks on the expression sex disorders. This has to be viewed
within the social and historical context of a country (Russia) wherein,
despite being in the 21st century, transsexual and transgender people are
considered mentally ill, and therefore not qualifying for a driving
license. The quotes, therefore, come across as ironic, this time showing
the stance of perhaps both the journalist and the newspaper editors,
who, like many people and organizations in the Western World, would
not accept (or at least view as ‘politically incorrect’) to label transsexual
and transgender people as having mental or sex disorders. It is therefore
clear to the eye of the reader that in this headline there is an invoked (or
covert) negative evaluation of the Russian government’s decision to
legally forbid these people to drive. Furthermore, the headline includes
a picture of a Moscow street jammed with cars, with the inscription
Moscow streets: various mental ‘disorders’ are seen as the cause of
road accidents, which again uses quotation marks not only to report
what the Russian government says of believes, but also to mark the
ironic stance taken by the newspaper writer and editors towards the
consideration of these conditions as disorders.

                                                            
17 BBC Online, 8 January 2015.
 3.3. Linguistic level at which the evaluation is manifested
(LL)

Evaluation can be manifested at each and every one of the levels of

linguistic description (phonological, morphological, lexical, syntactic
or semantic/pragmatic), a fact that is reflected in the corpus analyzed
herein as well. Even though the corpus is written, the mode of
evaluation carried out by means of capital letters or quotation marks has
been here labeled within the phonological level, as cases of ‘prosodic’,
paralinguistic evaluation, considering that these are strategies used in
written discourse to mark what in spoken discourse would be marked
by, for instance, a higher pitch or an ironic tone of voice. By way of
example, consider the headline TWO MILLION obese Brits to get free
gastric band operations on the NHS18, where the number of obese
people getting gastric band operations in Britain is capitalized, thus
showing the evaluative subsystem of Graduation at work (Martin &
White, 2005), not only through the subcategory of Quantification at the
lexical level (by specifying a number or quantity), but also through the
subcategory of Intensification (given by the capitalization).
Evaluation at the morphological level is generally manifested through
the pejorative or affectionate use of certain prefixes or suffixes. The use
of diminutives in English in words such as sweetie constitutes an
instance of morphological affective evaluation. In the headlines
analyzed herein, only one case of morphological evaluation was found
in the headline Brewers Unfayre: Family told to pay £400 in parking
fines because their Christmas lunch in the pub took too long19, whose
first phrase contains a pun on words related to the name of the pub
(Brewers Faye). Here a negative prefix (un-) and the letter r in the root
(faye has been transformed into fayre) have been purposely added in
order to make it sound like the word unfair. In this way the writer is
implying a negative judgement of the owners of the pub, who sent a
letter to their customers with a fine of £400 for having stayed a bit more

                                                            
18 The Mirror, 27 November 2014.
19 The Daily Mail, January 4th, 2015.
than three hours (the allowed time) having their Christmas lunch at the
pub.

Lexical evaluation is perhaps the most easily identifiable level at which

evaluation can be manifested, by simply using words or expressions
carrying an evaluative load, such as good, bad, beautiful or ugly, in
which case we are confronted with clear cases of inscribed or overt
evaluation, as illustrated in the headline Brian Cox's Human Universe
presents a fatally flawed view of evolution20, where the adjectival phrase
fatally flawed clearly presents an appreciation of the view of evolution
to be found in a BBC television program. Things are not always so
clear-cut for the analyst, however, and this is when evaluation at the
semantic/pragmatic level, which is normally context-dependent and
invoked or covert, comes into play and may co-occur with the overt
evaluation which is inscribed in some terms.

The syntactic level is also a fertile ground for the realization of

evaluation. For instance, thematically-marked constructions where the
normal or expected order of constituents in the mood structure of the
clause has been changed, may constitute a strategy for expressing a
given kind of evaluation. In our headlines it has been observed that such
may also be the case, for instance, with the use of a negative clause
construction where No is fronted as if it were the response to a yes/no
question (which does not appear in the headline, but is evoked), as in
No, Argentina's president did not adopt a Jewish child to stop him
turning into a werewolf21, where the fronting of the negative particle
focuses on the evaluation of the proposition contained in the clause as
not true.

                                                            
20 The Guardian, 4 October 2014.
21 The Guardian, 29 December 2014.
Thompson (2014: 51) illustrates the semantic/pragmatic22 kind of
evaluation by means of the utterance He wears sandals with socks,
which might seem an innocent statement to some people, but is
recognized by many British readers/hearers as “a token of scornfully
negative judgement of the sandal-wearer’s sartorial taste and, as an
extension, of the political and social beliefs that are assumed to be
reflected in his dress”. The evaluation here is totally context-dependent
on the values and assumptions of the particular social group or
discourse system in which the utterance occurs. An instance of
semantic/pragmatic evaluation is found in the headline Bisexual dad
beat boyfriend to death with table leg - then calmly watched England
match23, where we not only find the negative evaluation which is
already inscribed in the expression to beat someone to death, but also
the one invoked in the socially and morally sanctioned act of calmly
watching a football match after murdering someone. Furthermore, the
nominal phrase “bisexual dad” could also have very negative
connotations among a given group of readers having certain moral
values related to sexual orientation. This is precisely one of the cases in
which it was difficult to assign one category (lexical) or another
(semantic/pragmatic) to the evaluation found in the headline, and for
that reason, some practical decisions had to be taken for the
categorization, in order to avoid problems at the time of quantifying the
occurrences of the different kinds. Considering, then, that not all
headlines exemplified a single ‘pure’ category, those which exhibited
combinations of two or more categories were treated separately from
those containing only one of the five described above (phonological,
morphological, lexical, syntactic and semantic/pragmatic), as presented
in Table 3.

                                                            
22 Thompson does not use the term pragmatic, but I have added it in order to
differentiate it from the ‘purely’ semantic kind of Evaluation found at the lexical
level, and because this is a clear example of invoked evaluation, where pragmatic
considerations play an crucial part in the interpretation of the utterance as negatively
evaluative.
23 The Mirror, 3 December 2014.
Table 3: Corpus occurrences of evaluation at the different linguistic
levels
LINGUISTIC LEVEL (LL) BROAD- TABLOIDS
SHEETS
PHONOLOGICAL 1 0
(Prosodic)
MORPHOLOGICAL 0 0
LEXICAL 66 50
SYNTACTIC 8 8
SEMANTIC/PRAGMATIC 17 8
Lexical & Prosodic 2 13
Lexical & Pragmatic 6 7
Lexical & Syntactic 0 11
COMBINED Lexical, Prosodic & 0 1
Syntactic
Lexical, Syntactic & 0 1
Pragmatic
Syntactic & Prosodic 0 2
Syntactic, Pragmatic & 0 1
Prosodic
Morphological & 0 1
Pragmatic

The table shows that the level at which evaluation is most frequently
expressed is the lexical one, both in the broadsheets and the tabloids,
and that while in the broadsheets there seems to be a tendency for the
evaluation to be expressed at a single level of analysis, in the tabloids
there is a tendency towards a greater variety and complexity of
combinations. A relevant difference is observed as well in the use of
semantic/ pragmatic evaluation, which presents a higher frequency of
occurrence in the broadsheets than in the tabloids. This might indicate
a preference, on the part of the broadsheet journalists, to leave the
evaluation of the topic or people depicted in the headline to the ‘free’
interpretation of the reader, leaving fewer ‘inscribed prints’, so that they
may be labelled as more objective than their tabloid colleagues.
These differences are reflected in the significant results obtained
through the χ2 test, showing a p value of 0.00001 for the LL variable.
Within the headlines which showed a combination of levels at which
the evaluation is manifested, there are also significant differences (a p
value of 0.0139 was obtained when comparing the Lexical & Prosodic
Lexical & Pragmatic, and Lexical & Syntactic combinations
separately). The combination of Lexical & Syntactic, and Lexical &
Prosodic level evaluation presents a higher frequency of occurrence in
the tabloids, the latter being a phenomenon which, as we shall see in
3.6., seems to be partly related with the more frequent occurrence of
evaluation within the subsystem of Graduation, given the fact that the
prosodic type (manifested in the corpus mainly by the use of capitals
and inverted commas) in the tabloids generally co-occurs with some
kind of intensification, upscaling, or maximization24.

3.4. Degree of (in)directness of the evaluative meaning

(Deg): Inscribed vs. invoked evaluation

Evaluative meanings may be expressed either in an overt or a covert

manner. To use Martin & White’s (2005) terms, the evaluation may be
either ‘inscribed’ –and therefore, directly and openly expressed in the
text– or invoked (i.e. ‘hidden’ behind the literal meaning of the words
used), and thus conveyed in an indirect manner. As Figure 2 illustrates,
in the majority of the headlines examined, both in broadsheets (74%)
and tabloids (79%), the stance or evaluation was inscribed in some of
the words or expressions used. The frequencies are much lower for
invoked evaluation, (22% of the broadsheet headlines and 10% of the
tabloid ones), which are nevertheless higher for broadsheets. In some
of the headlines both types of evaluation co-occur, and therefore a third

                                                            
24 These are three of the categories within Graduation in the Appraisal Model.
category reflecting this fact was deemed necessary, which, as the figure
shows, seems to be slightly more frequent in the tabloids.

Figure 2: Percentage of occurrence of the Deg variable in broadsheets and

tabloids

90 79
80 74
70
60
50
40
30 22
20 10 11
10 4
0
Inscribed Invoked Both inscribed and
invoked

Broadsheets Tabloids

The results of the χ2 test for this variable show a significant p value of
0.0189, which indicates that the use of inscribed and invoked evaluation
varies significantly from broadsheets to tabloids in a general way. If the
variables are tested separately in pairs however, the results show that,
as the observed frequencies indicate, the difference is not significant for
the Inscribed cases of evaluation (p value: 01075), while it is significant
for the Invoked and the Both Inscribed & Invoked categories (p value:
0.0068). The tendency for broadsheets to resort to invoked evaluation
more often may be related to the intention of not leaving ‘traces’ of
overt evaluation so as to appear ‘objective’, as other results within this
study seem to point to.
 3.5. Position along the evaluation continuum (ContPoss):
Polarity.

The simplest and most accessible parameter that comes to mind when
dealing with evaluative meanings is the positive-negative one. As
Thompson & Hunston (2000: 25) note, “the most basic parameter, the
one to which the others can be seen to relate, is the good-bad
parameter”. These authors also point out that evaluations of good and
bad are dependent on the value-system underlying the text. This can be
clearly observed in the headlines, some of which have a clearly positive
orientation, such as Real-life Santa 16-year-old brings Christmas joy to
needy in memory of great-grandmother25 (bringing Christmas joy to the
needy is considered as something good in Western Christian society),
and some others a clearly negative one, such as I’m being emotionally
abused by my husband26 (being emotionally abused by one’s husband
is considered bad in Western and other cultures).
However, given the complexity of the human mind and language, it is
not always the case that an evaluative act can be clearly labelled as
either totally good/positive or totally bad/negative. The linguistic
phenomenon of evaluation is viewed herein (as in Alba-Juez & Attardo
2014) as a continuum where intermediate or mixed stances can be
identified. Put simply, evaluative language can be found at any of the
different points of a continuum that includes different positions from
one pole to the other, the central one being the neutral stance. In the
corpus examined for this study, some of the headlines display a neutral
stance, such as Kieron Richardson says Hollyoaks wedding was 'dress
rehearsal' for his own nuptials27, where there is evaluation within the
system of Engagement (Heteroglossic) but there seems to be no sign as

                                                            
25 The Mirror, 21 December 2014.
26 The Guardian, 26 December 2014.
27 The Mirror, 15 December 2014.
to whether what is reported is considered good or bad. Some other
headlines express both a positive and a negative evaluation at the same
time, as shown, for instance, in Shocking crash splits car IN HALF - but
family somehow walk away UNHARMED28, where we find an invoked
negative evaluation of the crash (which is normally associated to fatal
or undesired consequences), intensified by the capitals used in the
phrase IN HALF, in opposition and co-occurrence with the inscribed
positive evaluation contained in the word UNHARMED, which is also
intensified by the use of capital letters, but in an obviously different
direction from those of IN HALF.
Taking all the above aspects into consideration, the sub-variables
considered for both the qualitative and quantitative analysis of the
(ContPos) argument of the Evaluative Functional Relationship were
four: positive, negative, neutral, and both positive and negative, whose
frequencies of occurrence in the corpus are illustrated in Figure 3:

Figure 3: Percentages of occurrence of the ContPos variable (evaluative

polarity) found in the headlines of broadsheets and tabloids
80 76

70 65
60
50
40
30 22
20 16
7 9
10 4
1
0
BROADSHEETS TABLOIDS

Positive Negative Neutral Both pos & neg

                                                            
28 The Mirror, 23 November 2014.
As the figure shows, the most frequent kind of evaluation found in both
the broadsheets and the tabloids in terms of polarity is the negative one,
a fact that correlates with ‘newsworthiness’29, and that will not surprise
anyone who is used to reading the news on a regular basis. Another
result that would have normally been expected is that, considering the
sensationalism normally ascribed to tabloids, they would contain more
negatively-biased evaluation in the news than the broadsheets, an
expectation that has not been confirmed by the results of this study, for
as can be seen, the percentage of negatively-evaluative meanings in the
broadsheets is higher than that of the tabloids. The tabloids contain a
slightly higher percentage of occurrences of positively-charged
headlines than the broadsheets, the amount of neutral headlines in both
types of press is similar (7% in the broadsheets and 9% in the tabloids),
and the tabloids contain a slightly higher percentage of instances in
which the headlines contain a combination of both positive and negative
evaluation. The results of the χ2 test for the ContPos variable show a p
value of 0.3660, which presents strong evidence in favor of the null
hypothesis. The polarity of the evaluation is predominantly negative in
both tabloids and broadsheets, with the positive stance in second place,
the neutral one in third place, and the combination of both positive and
negative constituting a very small portion of the total number of
occurrences, inferior to 5% in both cases.

3.6. The evaluation Parameter (P) variable of the evaluative

functional relationship: The Appraisal Model.

Different authors have explored different patterns and categories of

evaluation. Traditional approaches (e.g. Cruse 1986), for instance, have
mainly worked with the basic positive-negative parameter presented in
3.5. Labov (1972) has looked into the evaluation found in narrative
                                                            
29See Bell (1991), who places negativity in the list of factors that make a story
newsworthy.
structure. Winter (1982) points out that clauses have two kinds of basic
information: what we know and what we feel about it, the latter being
crucial for the manner in which we deliver such information. Hunston
(1993) provides a list of values for the judgment of concepts and
activities in academic discourse. Bednarek (2006, 2008a) puts forward
a taxonomy with ten parameters of evaluation, seven of which she
considers to be “core” parameters, and three of which she labels as
peripheral30. Among all these and others, Martin & White’s (2005)
Appraisal Theory stands out as the most elaborate and fully developed
model of evaluation in the literature to date, and is the model chosen for
the analysis of the (P) term of the evaluative functional relationship in
this study.
Because the reader is expected to be familiar with Appraisal Theory (or
at least be able to refer to it), I shall give a very brief account of the
model. Appraisal is located by Martin & White as “an interpersonal
system at the level of discourse semantics” (2005: 33). The three main
interacting domains or resources on which Appraisal draws are the
following.

1) Attitude: the subsystem concerned with our feelings, including

emotional reactions, judgements of behaviour and evaluation
of things). It is divided into three further subsystems: Affect,
Judgement and Appreciation.
2) Engagement: the subsystem dealing with the source of
attitudes and the play of voices in discourse.
3) Graduation: the subsystem concerned with adjusting the
gradability of an evaluation.

                                                            
30The seven core parameters are comprehensibility, emotivity, expectedness,
humorousness, importance, possibility/necessity and reliability, and the three
peripheral ones evidentiality, mental state, and style.
Though extremely thorough and detailed (or perhaps because of this),
the Appraisal model, as Thompson (2014) and Macken-Horarik and
Isaac (2014) point out, presents some difficulties when dealing with
varied textual contexts, one of them being the “Russian doll” syndrome
(Thompson 2014), a term used in a metaphorical way to refer to the
problem of the different layers of evaluative meanings that may be
found in the same utterance or text under consideration.  It is very
common, as has been the case when analyzing the headlines of
broadsheets and tabloids, to find instances where an expression of one
kind of Attitude is nested inside another kind, or where an expression
of one kind of Attitude serves to invoke another kind of Attitude. In
order to facilitate the task of analysis within this model, then, Thompson
proposes a way of analysis which draws on the principle of “trusting
the text” (using Sinclair’s 2004 term) by “tracking the layers of
appraisal ‘outward’”, in such a way as to make explicit “that each step
represents a further move into interpretation” (2014: 62). This way of
analysis is exemplified in (2), where Thompson analyzes an utterance
which contains an inscribed expression of Judgement (about Henry
James’ qualities as novelist and short-story writer), found within a
wider co-text which critiques James’ short stories. In context (the
preface to his short stories), it is primarily negative judgment (he isn't a
good short-story writer) and it functions as a token (t) of negative
appreciation of his short stories, supporting the inscribed negative
appreciation of the stories that follows in the text:

(2) Henry James is a greater novelist than short-story writer

because he always needed space.
t - - appreciation [-judgement] (2014: 62)

In my analysis of the (P) variable in the headlines, I have followed

Thompson in his view that one should ‘trust the text’ wherever possible,
and therefore the wording was taken as the basis for the initial
assignment of categories. Thompson explains that

“the constraint of taking the wording as the basis for the

analysis on at least the initial categorization seems essential if
an examination of appraisal in a text is to retain as much of a
footing in replicable linguistic analysis as possible, rather than
being a subjective commentary on one person’s reading of the
text” (2014: 58)

In an attempt to make the analysis in this work as objective as possible,

then, I have tried to ‘trust the text’ and give an account of the different
layers or co-occurring types of evaluation, first looking into any traces
of inscribed evaluation in the wording, and then considering any
possible invoked evaluative meaning nested or in co-occurrence with it.
In some of the headlines, however, and as was shown in 3.4., the
evaluation is exclusively of the invoked kind: to use Jordan’s (2001)
terminology (in italics), an experiential Basis is given, and the
newspaper readers are left to reconstitute the Assessment for
themselves, an assessment that will depend on the values that are
assumed by the editors and journalists to be shared with their
readership. For instance, in the headline What difference would it make
if Ched Evans said sorry?31, the writer/journalist uses a question,
instead of an assertion, to make the evaluation ambiguous and dialogic
-as if it were a conversation with the readers- however guiding them to
believe or think that it would really make a difference if Evans (a
football player who returned to the field after serving two-and-a-half
years in jail for rape) apologized for what he did.

In my examination of the Appraisal subsystems present in each of the

headlines, I have followed Thompson in his philosophy of analysis,

                                                            
31 BBC Online, 7 January 2015.
although I have not used the same notation. (3) and (4) illustrate the
notation used in square brackets, as well as two of the ways in which
the three subsystems of Appraisal interact in the headlines:

(3) Father fatally [Attitude (negative, inscribed Appreciation]

shook baby [Attitude (negative, invoked Judgment)] in
momentary [Graduation (downscaling)] loss of self-control
[invoked, negative Judgement], court told.32 [Whole headline:
Engagement (Heteroglossic)].

(4) Unlucky [(Negative) Attitude (Judgement)] goalkeeper breaks

neck for the SECOND [Graduation (Quantification and
intensification)] time - but still plays on33 [(positive, invoked)
Judgement] [Whole headline: Engagement (Monoglossic)]

In (3) we find a combination of Heteroglossic Engagement (the

journalist is reporting what the Court said), two types of Attitude (an
Appreciation of the way in which the father shook the baby, and an
invoked Judgment of the father as a murderer). There is also Graduation
(the father’s loss of self-control is assessed as “momentary”, perhaps in
an attempt to minimize the murdering intention, in the sense that it was
not premeditated but only the result of a moment of loss of self-control)
and invoked negative Judgement in the expression “loss of self-
control”. The headline in (4) is an example of Monoglossic Engagement
which contains an instance of Judgement related to the goalkeeper, as
well as an instance of Graduation (quantification and intensification) in
the modifier SECOND, both by specifying the number of times the goal
keeper has broken her neck, and by using capital letters, which adds to
the intensification of the Graduation of the unlucky accident. There is
also invoked positive Judgement in the subordinate clause but still plays

                                                            
32 The Guardian, 18 December 2014.
33 The Mirror, 28 November 2014.
on, suggesting that the fact that she continued playing even after having
broken her neck is something to be praised and admired (because we
later know that this fact aided in obtaining a better score for her team).
As can be seen, the analysis can become rather complex because most
of the headlines contain more than one type of appraisal, and for that
reason, when performing the quantitative analysis of this variable, it
was deemed appropriate to group the types according to the different
combinations of the subsystems found in each of the headlines. And
since both the (ContPos) and the (Deg) variables (related to the polarity
of the evaluation and to whether the evaluation was inscribed or
invoked, respectively) have already been analyzed in previous sections,
in dealing with the (P) variable, only the combinations of types of
appraisal will be listed in the comparative table (Table 3).
It should be noted that some of the headlines contain neither instances
of Attitude nor of Graduation. The one subsystem that is always at
work, however is that of Engagement. This makes perfect sense if we
take into account that in their model, Martin & White (2005) express
their agreement with Stubbs (1996: 197) regarding the view that “when
speakers (or writers) say anything, they encode their point of view
towards it”. This view originates in Bakhtin’s (1981) notion of
heteroglossia and his idea that all verbal communication is dialogic, i.e.
that speaking or writing always reveals the influence of what has been
said or written before, and simultaneously anticipates the responses of
actual, potential or imagined interlocutors. Thus, when analyzing the
data, I have taken these considerations into account; nevertheless, I
have not labelled all the headlines as heteroglossic, because even
though it is true that all kinds of text are dialogic and intertextual, I have
followed Martin & White’s (2005) two-way categorization of
Engagement, classifying as Monoglossic those headlines that make no
reference to or recognition of other voices, and as Heteroglossic those
which invoke or allow for dialogistic alternatives. Thus (5) is an
example of a Monoglossic headline, while (6), which acknowledges the
voice of someone other than the writer (the gamers) illustrates a
Heteroglossic one:

(5) Sri Lanka's Rajapaksa suffers shock election defeat34.

(6) Gamers warned overusing Nintendo Wii could lead to chest

bleeds and even strokes35.

Let us now examine the ten most frequent combinations of Appraisal

subsystems found in the corpus, summarized in Table 4. The table
shows that in both broadsheets and tabloids the most frequent
evaluative combination is that of Heteroglossic Engagement and the
Judgment subsystem within Attitude. The difference between both
types of newspaper here is not significant (20% for broadsheets and
22% for tabloids), which means that both of them resort to this
possibility as a productive one. It is interesting to see how this
combination allows the journalist/ writer to negotiate Attitude by means
of Engagement, since the Judgement is normally attributed to other
people, in such a way as to apparently maintain the writer’s
‘objectivity’, when in fact the journalist is guiding the reader towards a
given (desired) evaluation. The headline in (7) exemplifies this fact:

(7) Oh, Rob Lowe, what were you thinking of?36

Once more, an interrogative clause is used in order to show recognition

of voices other than the writer’s own, thus allowing her to avoid
responsibility for the invoked negative Judgment of Rob Lowe as a
frivolous person, who does not in fact think much before talking and

                                                            
34 BBC Online, 9 January 2015
35 The Mirror, 16 December 2014
36 The Guardian, 28 December 2014.
acting (he is later in the article depicted as having had “monumentally
stupid outbursts”).

The second most frequent combination in both broadsheets and tabloids

(13% in both cases) is the one involving Monoglossic Engagement and
Judgement (within Attitude). An example is found in (6), where the
writer of the article says that a teenager attacked a woman on New
Year’s Eve (taking responsibility for his assertion to be true, thence its
labelling as Monoglossic), at the same time that he guides the reader
towards both an inscribed (Yob) and invoked negative judgment of this
teenager (the fact that he attacked a woman is socially sanctioned as
something bad or undesirable), reinforced by the word ‘shamed’ at the
beginning of the headline.

(8) Named and shamed: Yob who attacked a woman on New Year's
Eve identified as 19-year-old scaffolder37

It is apparent, then, that the subsystem of Judgment within Attitude is a

very frequent and productive one in both broadsheets and tabloids, used
in different combinations with other subsystems of Appraisal, since it
is the resource of Appraisal which is most frequently used in the
headlines.

Because of the common belief that tabloid discourse is ‘more emotive’

than broadsheet discourse, the number of occurrences of evaluation
within the subsystem of Affect were counted independently of the
combinations in which they appear, considering that Affect is the field
devoted to the “emotive dimension of meaning” (Martin & White,
2005: 42). The results yield exactly the same percentage of occurrences
of Affect (24%) for both types of newspaper, thus presenting no
evidence in favor of this hypothesis. However, I adhere to Thompson’s

                                                            
37 The Daily Mail, 3 January 2015.
(2015) view that both Emotion and Emotional talk38 may overlay the
other areas of Attitude (Judgment and Appreciation), as well as the
other two subsystems of Appraisal (Engagement and Graduation),
which make the picture of emotion much more complex, and that is the
reason why it was decided that looking into the different combinations
of Appraisal systems would give a better and more complete picture of
how the evaluative resources are used to obtain certain effects. Due to
space restrictions, however, not all the combinations found can be
discussed here, but in Table 4 it can be observed that there is a certain
balance between broadsheets and tabloids with respect to the other
combinations as well. The third one is the combination of Monoglossic
Engagement together with the subsystem of Appreciation within
Attitude, which seems to be a bit more frequent in broadsheets than in
tabloids (14% as compared to 9%). The remaining seven combinations
also show minor differences in frequency of occurrence. One
interesting difference, however, is that there is a higher occurrence of
Graduation (which is normally of the upscaling or maximization type)
in the tabloid headlines than in the broadsheet ones, which may be an
indicator of one of the commonly alleged ‘flaws’ of tabloids: their
exaggeration resulting in sensationalism.

Table 4: Most frequent combinations of the Appraisal subsystems in the

corpus
Most frequent
combinations BROADSHEETS TABLOIDS
of Appraisal (%) (%)
subsystems
1. Heteroglossic
Engagement / 20% 22%
Attitude
(Judgement-
social sanction)

                                                            
38 These terms are drawn from Bednarek (2008b, 2009).
2. Monoglossic
eng / Attitude 13% 13%
(Judgement:
social sanction)

3. Monoglossic
engagement / 14% 9%
Attitude
(Appreciation)

4. Heteroglossic
engagement 13% 5%

5. Monoglossic
engagement / 12% 7%
Attitude (Affect)

6. Monoglossic
engagement 9% 5%
7. Monoglossic
eng / Graduation 5% 8%
(force)

8. Heteroglossic
eng. / 3% 9%
Graduation
(force)

9. Heteroglossic 8% 3%
Attitude (Affect)

10. Monoglossic 2% 8%
eng. / Attitude
(Judgement) +
Graduation

Other (infrequent 1% 11%

and varied)
combinations
The χ2 test results for this variable present a p value of 0.0499, which is
very close to the level of confidence (p< 0.05), therefore instantiating a
marginal case, where they could be interpreted either way (as
significant or not). It could be thought, however, that if we looked
exclusively into the two main types of Engagement, we could find some
significant results, and for that reason the bare occurrences of
Monoglossic and Heteroglossic Engagement –irrespective of their
combinations– were counted as well, as shown in Figure 4. The figure
shows that broadsheet headlines contain a higher number of
Monoglossic cases than the tabloids. This would mean that broadsheets
present their headlines as unquestionable more often than tabloids do,
and tabloids, on the contrary, resort more often to a negotiable stance.
However, the statistical results of the χ2 test for these two variables
show that the difference is not significant (p value of 0.1949 at p <
0.05), which again leaves us with uncertain results concerning the P
variable of the evaluative functional relationship. Further research on a
larger corpus would be necessary to accept or reject the null hypothesis,
but in any case, the statistically marginal results for the combinations
in Table 4 and the non-significance of those in Figure 4 confirm the
previous observation that, as far as the evaluative subsystems are
concerned, the differences between broadsheets and tabloids are not so
clear-cut or easily identified.

Figure 4: Occurrences of the two main Engagement subsystems in the

headlines
 Engagement in the headlines
80
64
55
60 45
36
40

20

0
BROADSHEETS TABLOIDS

MONOGLOSS HETEROGLOSS

We now turn to the last one of the arguments of our evaluation equation,
i.e. the Mode of the evaluation.

3.7. Mode of the evaluation (Mo) in the headlines

The last variable to be examined in the evaluation of the headlines has

to do with the mode in which the evaluation is conveyed. I have
followed Kress and van Leeuwen (2001: 21) in their definition of mode
as “semiotic resources which allow the simultaneous realization of
discourses and types of (inter)action”. Language, music, or images are
examples of modes. In the newspaper headlines, this variable may have
one, or a combination of any two, or of the three following values:

1) Linguistic evaluation (L), i.e. evaluation carried out by means of the

language used

2) Paralinguistic (PL), which in the headlines is realized by means of

the use of capital letters or quotation marks. In this particular corpus,
all the examples of paralinguistic evaluation coincide with those labeled
as ‘prosodic’ within the phonological level of the (LL) variable
examined in 3.3. For this reason, no further examples will be analyzed
in this section.

3) Evaluation conveyed through images and/or symbols other than

writing (e.g. photos, drawings, emoticons, videos, etc.) (Im). In the
corpus used for the present analysis, the majority of the headlines, in
both broadsheets and tabloids, were accompanied by images, whose
evaluative meaning interacts with that of the language used.

Table 5 shows the percentage of headlines in the corpus which were

accompanied by pictures, as well as the different combinations of
Modes of evaluation found. In addition, in the last row the figures tell
us the percentages of the images whose evaluation seemed to match that
of the linguistic part of the headlines, and those of the images whose
evaluation did not seem to be in agreement with the prevailing
evaluative tone of the language used. I am aware of the fact that the
assessment of this kind of evaluation contained is bound to be more
subjective than that of the wording in the headlines39. However, and
even though this is not the main concern of the present study, an
examination of the evaluation conveyed by these images was carried
out, in order to see if it was coherent with the evaluation expressed by
linguistic means, taking into account previous studies (e.g. Breeze
2014, Richardson & Meinhof 1999). Breeze (2014: 305) points out that:

“It is possible that an article which offers a balanced assessment

of an issue might be found to have a headline with negative
overtones, or a photograph that appears to show a positive

                                                            
39 Stökl (2010), for instance, points out that images, though immediate in their

cognitive and relational effect, are semantically vague and open-ended.

 vision of a particular issue, simply because of the time pressures
in the production process.”

In my analysis of the images (which is much less delicate and detailed

than Breeze’s, but was inspired by it), I examined their main and most
salient features (such as the colors or light in the picture, the possible
codification40 of some meanings in certain images, the expressions in
the faces of the people portrayed, or the clothes or costumes they were
wearing) in order to gain additional insight regarding the equal or
different status of the different modes found in the headlines, which
would in turn constitute a clue as to the evaluative coherence of the
whole semiotic space devoted to the headline.

Table 5: MODE of the evaluation found in the headlines

MODE of the BROADSHEETS TABLOIDS
evaluation
Nº of 84 (84%) 73 (73%)
headlines
with pictures
Linguistic 17 (17%) 26 (26%)
mode
Linguistic & 0 (0 %) 1 (1%)
paralinguistic
modes
Linguistic 78 (78%) 59 (59%)
and Image
modes
Linguistic, 5 (5%) 14 (14%)
paralinguistic
and image
modes

                                                            
40 See Forceville & Clark (2014), who argue that some nonverbal behaviors, such as

pictures, can contain coded meanings, which would allow for the possibility of non-
linguistic explicatures.
 Does ling eval Yes: 67 out of 84: 79.8% Yes: 72 out of 73: 98.6%
coincide with
the image? No: 17out of 84: 20.2% No: 1 out of 73: 1.4%

As the table shows, in the majority of cases, both in the broadsheet and
tabloid headlines, the evaluative message of the picture seemed to
match that of the wording. However, the percentages show that in this
respect the tabloids were more coherent than the broadsheets, for only
one case (within the 100 tabloid headlines explored) was found in which
these two modes had a different evaluative status, while in the
broadsheets the number of incongruities in that respect amounted to 17.
An example of coherence between words and picture is found in the
headline Ashley Roberts flashes a LOT of sideboob in sexy plunging
dress41, which is accompanied by a picture of Ashley where, in effect,
she is wearing a dress with ample cleavage that allows her to show a
great part of her breasts. In contrast, an example of incoherence between
words and image can be observed in the article entitled Vaccines move
to Ebola frontline, published in BBC Online on 9 January 2015: While
the wording of the headline inspires hope in the readers, telling them
that vaccines will be taken to West Africa (the Ebola frontline) to be
administered among the population, the image42 accompanying the
headline seems to inspire fear instead, because it shows a man or
woman whose face cannot be seen because s/he is completely covered
by and dressed in a hazmat suit, walking through a dark and very narrow
alley between rocky walls, in the foreground of which a blurred image
of some African people can be glimpsed. If we follow Forceville and
Clark’s (2014) idea that some images or pictorial elements contain

                                                            
41 The Mirror, 16 December 2014. The picture cannot be reproduced here for
copyright reasons, but can be seen at http://www.mirror.co.uk/3am/celebrity-
news/ashley-roberts-flashes-lot-sideboob-4823860
42 The picture can be seen at http://www.bbc.com/news/health-30742769
coded meanings, it could be argued that a hazmat suit encodes all kinds
of negative concepts related to disasters such as nuclear war or
dangerous diseases. In addition, dark alleys evoke difficult or dangerous
situations which one would hardly associate with hopeful thoughts.
This mismatch between text and picture might reflect the writer’s
conscious or subconscious skepticism about the power of these vaccines
to finally eradicate the Ebola epidemic, or about the extent to which
these measures will be taken in West Africa. In fact, this fear is
somehow verbalized by the writer of the article later on, when he writes
that “It is still unclear how much protection against Ebola, or for how
long, the vaccines might provide”, in which case it can be deduced that
the picture is more in agreement with the information given in the body
of the article than with that of the headline, which, if read in isolation,
will most likely provide the reader with an excessive dose of optimism.

As to the comparison of the frequencies of each of the modes of

evaluation in broadsheets and tabloids, the χ2 results show a general p
value of 0.0205, which is significant at p < 0.05. A more detailed
analysis (carried out by isolating the variables where there were greater
differences in the observed frequencies) shows that the modes which
vary significantly (but only slightly, with a p value of 0.0460, which
makes them marginal cases) from one type of press to the other are the
Linguistic Mode (which is significantly more frequently used as a
unique mode in the tabloid headlines), the Linguistic and Image Mode
(which appears to be a more frequent combination in the broadsheet
headlines) and the Linguistic, Paralinguistic and Image Mode (which is
more frequent in the tabloids, a result which is in-sync with that
obtained for the subsystem of Graduation in 3.6, considering that the
paralinguistic effect is obtained in the headlines by means of its
different categories, such as intensification or quantification ).

Finally, the χ2 results obtained (p value 0.0021) for the question in the
last row of Table 4 suggest significant differences in the coherence
between picture and wording in the headlines, the tabloids presenting a
more coherent whole than the broadsheets. This may be due to the fact
that broadsheets tend more towards ambiguity or ‘objectivity’ by letting
the reader draw their own conclusions and evaluations, while tabloids
are more consistent in matching their words with the realistic43 images
included.

3.8. The ‘evaluative blueprint’ of the headlines captured in

the ‘evaluation equation’: An example.

After assigning the qualitative values to each of the variables of the

evaluative functional relationship, where many different aspects
affecting the final result of the evaluation of a given utterance or text
have been included, we obtain what might be called the ‘evaluative
blueprint’ or general evaluative picture of each of the headlines in
question. By way of example, (7) shows the summarized complete
characterization of the variables for the headline in (1) above:

(9) Father fatally shook baby in momentary loss of self-control,

court told.44
E = F (T, Lex & Sem-prag, Ov & Cov, Neg, [H Eng, Att (Judg
& Appr), Grad], Ling)

The equation synthesizes the information related to the fact that

Evaluation is a Function of the variables whose values are specified
between parenthesis, which tell us that this function (E) in this
                                                            
43 By saying that the pictures in the tabloids are more ‘realistic’ I mean that, rather
than suggesting or ‘implying’ a given meaning, they tend to show the real thing being
depicted, such as the picture of a car-crash with the blood stains of the injured people,
or the bruised face of a woman who has been beaten by her husband, without much
restriction.
44 The Guardian, 18 December 2014.
particular headline: a) is found in the Textual Phase; b) is found at both
the Lexical and the Semantic-Pragmatic Levels of linguistic analysis;
c) presents a combination of both the overt (Inscribed) and the covert
(Invoked) kinds; d) is negatively-oriented; e) presents a combination of
Appraisal subsystems and categories which includes Heteroglossic
Engagement, two of the subsystems of Attitude (Judgement and
Appreciation) and Graduation; and f) is only found in the Linguistic
Mode (this is one of the minority cases in which no picture accompanies
the headline).

This characterization is to be viewed as the evaluative ‘skeleton’ of the

headline as it were, which in all cases was complemented by the
detailed analysis and categorization of each evaluative expression
found, as was shown in (1) above. Thus, and considering that all the
steps taken have proved to be useful and insightful, the method
followed in this study is proposed here as a possible method of analysis
of the evaluative content of any type of text.

4. Summary and conclusions

In this paper I have presented a comparative analysis of the evaluative
resources used in the semiotic space occupied by the headlines of
broadsheets and tabloids. The method followed in the analysis has
mainly consisted in scrutinizing each one of the six variables or
knowledge resources of evaluation which constitute the arguments of
the evaluative functional relationship E = F(PH, LL, Deg, ContPos, P,
Mo), and its results and conclusions can be summarized as follows:

- As far as the complexity and length of the headlines are

concerned, significant results, both qualitative and quantitative,
have been obtained in favor of the thesis (which is contrary to
 what other authors have stated) that the headlines of the
tabloids are longer and syntactically more complex than those
of the broadsheets. The majority of the broadsheet headlines in
our corpus were shorter in number of words and contained
fewer constituents than the tabloid headlines. However, an
important point to take into account for the purposes of this
study is that some of the syntactic constructions used tended to
favor certain types of evaluation, irrespective of their length.
Interrogatives, for instance, are very frequently used as a means
to negotiate Attitude by means of Engagement.
- The Ph variable looked into has been exclusively the Textual
one, in both tabloids and broadsheets, given the nature of this
particular text type.
- Regarding the LL variable, the results have shown that Lexical
Level evaluation is the most frequent one in both types of
headlines. However, both the qualitative and the quantitative
results have thrown light on some relevant differences in the
management of this evaluative resource. Tabloids show a
greater tendency towards prosodic evaluation (at the
Phonological Level, manifested in written form mainly through
the use of capital letters and quotation marks), as well as
towards the combination of different linguistic levels of
evaluation in the same headline, while broadsheets present a
higher tendency towards the use of pragmatic evaluation and a
lower tendency towards combining evaluation at different
linguistic levels. This might indicate a preference, on the part
of the broadsheet journalists, to leave the evaluation of the topic
or people depicted in the headline to the ‘free’ interpretation of
the reader, leaving fewer ‘inscribed prints’, so that they may be
labelled as more ‘objective’.
- As to the Deg variable, it has been shown that the use of
inscribed and invoked evaluation varies significantly from
broadsheets to tabloids, the invoked type being more frequent
 in the broadsheet headlines (a result which is in agreement with
the higher incidence of Pragmatic Level evaluation found for
the LL variable), and the mixed type (both Inscribed &
Invoked) being more frequent in the tabloid ones (though not a
very productive combination in any of them). The Inscribed
type is the most frequent option in both types of headlines.
- In terms of the polarity of the evaluation (ContPos variable),
Negative evaluation is by far the most frequent and exploited
type in both broadsheets and tabloids, the statistical test
showing a strong result in favor of the null hypothesis. There is
very little doubt, therefore, as to the fact that both types of press
focus more on negative news stories than on positive or neutral
ones.
- For the treatment of the P variable, a qualitative method for the
analysis of the headlines in the line of Thompson’s (2014) was
followed and defended, so as to try to capture as many nuances
as possible regarding the interaction of the different Appraisal
subsystems. The quantitative results have shown that there are
no significant differences between tabloids and broadsheets in
the use of the two main subsystems of Engagement
(Monoglossic and Heteroglossic), and that the most frequent
combination of Appraisal subsystems in both tabloids and
broadsheets is that of Heteroglossic Engagement together with
the Judgment subsystem within Attitude, which displays a
general tendency in both types of newspapers to use
Engagement to negotiate Judgement. The possibilities here are
infinite and complex, this fact probably being the reason for
obtaining statistical results which show a marginal case, with
weaker evidence both towards accepting the null hypothesis
and its counterpart (which would argue in favor of the existence
of significant differences regarding the P variable). If, however,
the mere occurrences of the different subsystems of Appraisal
were counted, it was observed that there is a greater tendency
 for tabloids to resort to Graduation than broadsheets (a fact that
might be related with the tabloids’ alleged tendency to
exaggerate and maximize evaluation and emotion),
- Significant differences have also been found in the Mode
variable, both qualitatively and quantitatively. Pictures
accompany the words of the headlines very often in both
broadsheets and tabloids, but the Paralinguistic Mode is more
often used in the tabloids, the latter being a result which is in-
sync with the results of the LL variable (showing a greater
occurrence of prosodic evaluation in the tabloids as well). It has
been interesting to discover, through the results of the χ2 test
that, with respect to the evaluative content of the images
accompanying the headlines, the tabloids present a more
coherent picture of evaluation than broadsheets. This may be
an indication that tabloid journalists are more open in their
evaluative intention than broadsheet ones, who might want to
appear more ‘objective’ by being ambiguous and guiding the
readers indirectly towards their ‘own’ evaluative conclusions.

All in all, it has been argued and shown here that the Evaluative
Functional Relationship E may be a useful ‘formula’ for deconstructing
and analyzing the different resources of evaluation in the corpus of this
study, as well as in any other type of text.

Both similarities and differences have been found between the

headlines of tabloids and broadsheets. The similarities might lead to the
thought that broadsheets are undergoing a process of ‘tabloidization’ or
at least that the differences between them are diminishing. And this is a
process that might very well apply more to the headlines than to the
body of the article, considering that the headlines are always meant to
be eye-catching. The differences found, however, seem to indicate that
this process is far from complete in the present state of affairs, since it
has been made apparent that the broadsheet headlines make use of
resources such as Invoked, Pragmatic Level evaluation, or pictures
whose evaluative message does not coincide with the wording, in order
to leave fewer inscribed marks than the tabloids, while the tabloids
make more use of the Graduation resources, and appear to be more overt
and coherent regarding their evaluative intentions. The marginal results
obtained for some of the variables might confirm, however, that the
distinction between tabloids and broadsheets is not so clear-cut or easily
identified as might have been expected.

No research results can boast of being totally complete or categorical,

and the ones obtained in the present study are no exception. While
allowing for the need of further research on the topic (by using a bigger
corpus, for instance), I hope this has been, however, a good start.

Annex

Notation used in the examples of syntactic analysis (Table 1)

NP: Noun Phrase […] : Embedded clause in

PrepP: Prepositional Phrase complex main clause.
S: Subject Cs: Subject Complement
(S): Unexpressed, implicit Ag: Agent
subject wh- cl: Wh- clause
P: Predicator to-inf cl: To-infinitive clause
O: Object sub: Subordinator
Circ: Circumstantial adjunct

References:
Alba-Juez, Laura (forthcoming). The ‘ingredients’ of the evaluative
functional relationship: The case of humorous discourse. In
 Ruiz Gurillo, Leonor (ed) Metapragmatics of Humor: Current
Research Trends. IVITRA Research in Linguistics and
Literature. Amsterdam: John Benjamins.
Alba-Juez, Laura / Thompson, Geoff 2014. Chapter 1: The many faces
and phases of evaluation. In Thompson, Geoff & Alba-Juez,
Laura (eds), Evaluation in Context. Amsterdam: John
Benjamins. 3-23.
Alba-Juez, Laura / Attardo, Salvatore 2014. The evaluative palette of
verbal irony. In Thompson, Geoff & Alba-Juez, Laura (eds),
Evaluation in Context. Amsterdam: John Benjamins. 93-116.
Bakhtin, Mikhail 1981. Discourse in the novel. In Holquist, M. (ed).
The Dialogic Imagination. Austin: University of Texas Press.
259-422.
Bednarek, Monika 2006. Evaluation and cognition: Inscribing, evoking
and provoking opinion. In Pishwa, Hanna (ed.), Language and
Memory. Aspects of Knowledge Representation. Berlin:
Mouton de Gruyter. 187-221.
Bednarek, Monika 2008a. ‘An increasingly familiar tragedy’:
Evaluative collocation and conflation. In Evaluation in Text
Types. Special Issue of Functions of Language 15.1.
Amsterdam: John Benjamins.
Bednarek, Monika 2008b. Emotion Talk across Corpora. New York:
Palgrave Macmillan.
Bednarek, Monika 2009. Emotion talk and emotional talk: Cognitive
and discursive perspectives. En H. Pishwa (ed.), Language and
Social Cognition. Expression of the Social Mind. Berlin:
Mouton de Gruyter. 395-431.
Bell, Allan (1991). The Language of News Media. Oxford: Blackwell.
Breeze, Ruth 2014. Multimodal analysis of controversy in the media.
In Thompson, Geoff & Alba-Juez, Laura (eds), Evaluation in
Context. Amsterdam: John Benjamins. 303-319.
Connell, Ian 1998. Mistaken Identities: Tabloid and Broadsheet News
Discourse. Tablodization and the Media, Vol. 5 - 1998, No. 3.
11-31.
Conboy, Martin 2004. Parochializing the global. Language and the
British tabloid press. In Aitchison, Jean & Lewis, Diana M.
(eds) New Media Language. London & New York: Routledge.
Cruse, D.A. 1986. Lexical Semantics. Cambridge: Cambridge
University Press.
Dor, Daniel 2003. On newspaper headlines as relevance optimizers.
Journal of Pragmatics 35 (2003) 695–721.
Fairclough, Norman 1995. Media Discourse. London: Arnold.
Forceville, Charles and Billy Clark (2014). Can pictures have
explicatures? Linguagem em (Dis)curso 14/3. 451-472.
Fowler, Roger 1991. Language in the News. London: Routledge.
Hunston, Susan / Thompson, Geoff (eds.) 2014. Evaluation in Text.
Authorial Stance and the Construction of Discourse. Oxford:
Oxford University Press.
Jordan, Michael 2001. Some discourse patterns and the signaling of the
Assessment-Basis relation. In Scott, Mike and Thompson,
Geoff (eds.) Patterns of Text: In Honour of Michael Hoey, ,
159-192. Amsterdam: John Benjamins.
Kress, Gunther / van Leeuwen, Theo 2001. Multimodal Discourse.
London: Hodder Education.
Labov, William 1972. The transformation of experience in narrative
syntax. In Labov, William (ed.), Language in the Inner City.
Philadelphia, University of Pennsylvania Press.
Labov, William 1997. Narrative Analysis: Oral versions of personal
experience. In Journal of Narrative and Life History, 7 (1-4).
3-38.
Macken-Horarik, Mary / Isaac, Ann 2014. Appraising Appraisal. In
Thompson, Geoff & Alba-Juez, Laura (eds), Evaluation in
Context. Amsterdam: John Benjamins. 67-92.
 Martin, J.R. & White, P.R.R. 2005. The language of Evaluation.
Appraisal in English. Hampshire and New York: Palgrave
Macmillan.
O’Connor, Evelin 2011. Broadsheets vs. Tabloids.
http://leavingcertenglish.net/2011/05/broadsheets-vs-tabloids/
Richardson, Kay / Meinhof, Ulrike. 1999. Words in Common? London:
Routledge.
Sinclair, John 2004. Trust the text: language, Corpus and Discourse.
Edited with Ronald Carter. London: Routledge.
Stökl, Hartmug 2010. Sprache-Bild-Texte lesen. Bausteine zur
Methodik einer Grundkompetenz. In Diekmannshenke, Hajo,
Klemm, Michael and Stöckl, Hartmut (eds), Bildlinguistik,
Berlin: Erich Schmidt. 43-70.
Stubbs, Michael 1996. Towards a modal grammar of English: a matter
of prolonged fieldwork. In Stubbs, Michael (ed) Text and
Corpus Analysis. Oxford: Blackwell.
Thompson, Geoff 2014. AFFECT and emotion, target-value
mismatches, and Russian dolls: Refining the APPRAISAL
model. In Thompson, Geoff & Alba-Juez, Laura (eds),
Evaluation in Context. Amsterdam: John Benjamins. 47- 66.
Thompson, Geoff 2015. Evaluation and Emotion. Talk given at the
Jornadas de Investigación (Proyecto Emo-Fundett), 4 February
2015, Madrid, Spain.
Thompson, Geoff / Alba-Juez, Laura (eds) 2014. Evaluation in Context.
Amsterdam: John Benjamins.
Thompson, Geoff / Hunston, Susan 2000. Evaluation: An Introduction.
In Hunston, Susan & Thompson, Geoff (eds.) Evaluation in
Text. Authorial Stance and the Construction of Discourse.
Oxford: Oxford University Press. 1-27.
Winter, Eugene 1982. Towards a Contextual Grammar of English.
London: George Allen and Unwin.

View publication stats