Reading: Critical Discourse Analysis

(This document is intended as course reading for part of the module Introduction
to English Linguistics 2 taught at the Faculty of Linguistics and Cultures of English
Speaking Countries – ULIS, VNU – 2015)

Prepared by Nguyen Thi Thu Ha (Ph.D)

Division of English Linguistics
Faculty of Linguistics and Cultures of ESC – ULIS, VNU

1. What is CDA?

Critical Discourse Analysis (CDA) emerged in the 1970s, reflecting a turn in the
interest of linguistic research from the purely structural dimension to the functional aspect
of language as it relates to the social. CDA studies language use in its socio - political
context and regards ‘language as social practice’ (Fairclough and Wodak, 1997). CDA
pays particular attention to the relation between language and power, aiming to unearth
the ideologies underlying discourse. Quoted in Wodak and Meyer (2001), the following
passage by van Dijk best captures some of the main aims and goals of CDA:

Beyond description or superficial application, critical science in each domain

asks further questions, such as those of responsibility, interests, and ideology.
Instead of focusing on purely academic or theoretical problems, it starts from
prevailing social problems, and thereby chooses the perspective of those who
suffer most, and critically analyses those in power, those who are responsible,
and those who have the means and the opportunity to solve such problems.
(van Dijk, 1986:4)
2. Main tenets of CDA

Phillips and Jorgensen (2002) summarize five CDA tenets and principles as follow.

First, situating language use in social networks, CDA perceives discourse, or

discursive practice, as social practice. It is undeniable that language is a social
phenomenon and is part of society, not somehow external to it (Fairclough, 2001). Linguistic
phenomena are social phenomena of a special sort, and social phenomena are (in part)
linguistic phenomena. According to Eckert and Mc Connell-Ginet (2003: 43), the term ‘social
practice’ refers to human activity when emphasizing the conventional aspect of activity and
its relation to social structure. While social structure constrains practice, it does not
determine it. It is important to highlight what discourse refers to as this term is often ‘left
undefined, vague or confusing’ (Mills, 1997; Wodak, 1997). In one sense, discourse refers to
the communicative process in which language in the form of texts is produced and

1
interpreted in a social communicative setting. Discourse encompasses not only spoken and
written linguistic texts but also other forms of communication such as visual or audio-visual
modes. A current view of discourse sees that discourse in most cases is multi-modal (Kress
and van Leeuwen, 1996); hence, a comprehensive analysis of discourse must include forms
of communication other than language. In a more post-structuralist sense, discourse is a way
of representing and interpreting social realities. Discourses exist in relation to other
discourses and these discourses can be supporting as well as competing or conflicting. As
Phillips and Jorgensen (2002) put it, the aim of CDA is to shed light on the discursive
dimension of social and cultural phenomena and processes of change.

Second, discourse is both socially constituted and constitutive. As mentioned in the

first principle, discourse is a social practice; consequently, it complies with social conventions
in reflecting social reality. In this sense, discourse is conditioned by the social in two
dimensions. First, discourse signifies the extra-discursive, conditioned by other already
existing discourses. Second, how discourse reflects reality is conditioned by socially
acceptable ways of doing it. Take the example of classroom discourse: How students and
teacher are arranged in the classroom, what they are supposed to do, and what topics or
subjects are legitimately appropriate in this discourse comply with certain norms, reflecting
the pre-existing social arrangement at the institutional level and societal level. In a nutshell,
discourse is socially constituted in both the what and the how dimensions. The relation
between discourse and the social, however, is not a one way process but a dialectical one
(Fairclough, 2001). Discourse is constitutive at the same time because discourse also
contributes to the shaping and reshaping of social structures. In other words, discourse has
effects upon social structures and contributes to the achievement of social continuity or
social change. In a sense, social structures are partly a product of discourse. Foucault (1972)
emphasizes the potential constitutiveness of discourse by defining it as ‘the practices that
systematically form the objects of which they speak’. However, CDA is less post-structuralist
in that it does not negate the material that exists independently from discourse as most post-
structuralists maintain. Discourse is not omnipotent, and is conditioned by social reality or
other discourses. As Chouliaraki and Fairclough (1999: 6) put it, CDA recognizes the social
import of discourse without reducing social life to discourse.

Third, discourse should be analyzed within its social context. This principle derives
from the first and the second tenets, defining discourse as social practice and conditioned by
social structures. In other words, discourse is historical and can only be understood with
reference to its context or existing discourses. If we, for example, wish to analyze a political
speech by an influential politician on a controversial issue, we certainly need to consider his
(and other people’s) already-made speeches on the same issue as well. We also need to
relate the speeches to the social contexts when they are made. Hence, in Fairclough’s
model (2001), text analysis is not carried out separately but within discoursal interaction and

2
social context. Analyzing discourse means not just analyzing texts, nor analyzing processes
of production and interpretation, but analyzing the relationship between texts, processes, and
their social conditions. Social conditions include both the immediate conditions of the
situational context and the conditions of institutional and social structures on a higher level
(Fairclough, 2001). Beyond context, which refers to the extralinguistic factors such as culture,
political condition and ideology, CDA also introduces the notion of intertextuality and
interdiscursivity, which highlight the importance of seeing texts in relation to other existing
texts. Different from all other methods, CDA takes into consideration a wide range of factors
that influence the production and consumption of texts.

Fourth, discourse functions ideologically. At the heart of CDA is the claim that
underlying discourse is ideology or that ideology is mediated through language use.
(Fairclough, 2001, 1995). At this point, CDA shares with the Marxist tradition the view that
discourse can be a tool to disseminate and to suppress ideology. Discourse, hence, reflects
the interests of certain social groups, based on race, gender, age and so on. This principle is
not just applicable to political or mass media discourse, which is believed to have massive
influence on the public but also true with individual use of language. Since language is a
resource with which people constantly have to make choices in communication, they may be
unaware of the fact that their beliefs and viewpoints are embedded in their linguistic
selections. In other words, our language is far from neutral. According to Coates (1998: 302),
there is no neutral discourse; whenever we speak, we have to choose between different
systems of meaning and different sets of values. Language embodies our cultural and social
values; hence, when we speak, we do not just say words, we speak our culture (Goddard &
Patterson, 2000: 67). An example is, the Vietnamese can be very familiar with their
Resistance War Against America, which, however, is more often mentioned as the Vietnam
War by other peoples. Do you see any difference in the two expressions? CDA practitioners,
however, do not ask whether certain ideologies are deliberately or subconsciously
embedded; rather, they are interested in finding out what the underlying ideologies are. If
certain ideologies disadvantage a particular social group, then the uncovering of these
ideologies can raise people’s awareness, which is believed by CDA practitioners to be the
first step towards emancipation.

Fifth, CDA is research with a stance. Different from scientific research especially one in
the natural science domain which places importance on objectivity, CDA does not claim itself
as politically neutral and is always explicit about its own position and commitment. Critical
analysts must take a clear political stance from the beginning, aiming to uncover the
discursive practice that maintains the unequal power relations. CDA is committed to radical
social change for fairness and equality. CDA, hence, takes sides with the oppressed and less
powerful in discourse. For instance, feminists often use CDA to challenge discourses in
which women are disadvantaged and where gender inequality exists. However, this does not

3
mean that CDA is a biased interpretation or a political commentary where the authors
interpret meaning from their own experience on the basis of their ideological commitment. As
stated by Fairclough (2001:4), scientific investigation of social matters is perfectly compatible
with ‘opinionated’ investigators, and being committed does not prevent investigators from
arguing rationally and with evidence. CDA attempts to be a systematic kind of analysis which
bases on particular linguistic theories and social contextual information to draw out
conclusions. According to Chouliaraki and Fairclough (1999: 152), CDA is textually-oriented
discourse analysis, which means it anchors its analytical claims about discourses in close
analysis of texts. The term ‘critical’ is to be understood as having distance to the data,
embedding the data in the social, taking a political stance explicitly, and a focus on self-
reflection as scholars do research (Wodak and Meyer, 2001:9).

3. What is ideology, critical and power?

To better understand CDA as a linguistic approach, some key CDA concepts should be
understood. As far as CDA is concerned, the three notions of ‘critical’, ‘ideology’ and ‘power’
are often mentioned as the central circle of the subject. Basically, critical is to be understood
as having distance to the data, embedding the data in the social, taking a political stance
explicitly, and a focus on self-reflection as scholars doing research.

The concept of ideology is one of the most controversial and elusive academic issues.
According to Simpson (1993: 176), ideology is ‘a mosaic of cultural assumptions, political
beliefs and institutional practices’. Since language is regarded as the physical form of
ideology, language is thus an indispensable part of any attempt to study ideology
(Fairclough, 2001, 1995). Ideology, for CDA, is seen as an important aspect of establishing
and maintaining unequal power relations.

The third concept of concern is power. In the society, people very often have different
power and CDA is interested in finding about such relations of difference, and particularly
about the effects of differences on social structures. Language is entwined in social power in
a number of ways: language indexes power, expresses power, is involved where there is
contention over and a challenge to power. Power does not derive from language, but
language can be used to challenge power, to subvert it, to alter distributions of power in the
short and long term (Fairclough, 2001).

For CDA, language is not powerful on its own – it gains power by the use powerful people
make of it. That’s why CDA often chooses the perspective of those who suffer and critically
analyses the language use of those in power, who are responsible for the existence of
inequalities and who also have the means and opportunity to improve conditions.

According to Fairclough (2001), ideologies are closely linked to power because the
nature of ideological assumptions is embedded in particular conventions, and the nature of
those conventions themselves, depends on the power relations which underlie the
4
conventions. Moreover, ideologies are closely linked to language, because using language is
the commonest form of social behavior where we rely most on ‘common sense’ assumptions.

4. Methods of doing CDA

As approaches to CDA are diverse, so are the methods of doing CDA. Different theorists
have proposed different methods, and it is apparent that there is no consistent CDA
methodology. Some examples of CDA methodologies include, for example, Siegfried Jager’s
content oriented steps of structure analysis and language oriented steps of fine analysis, or
van Dijk’s analysis that concentrates upon a long list of linguistic markers, as presented in
Wodak and Meyer (2001: 25). For the purpose of this introductory reading, I will present
Fairclough’s model (2001), which highlights the relation between discoursal processes and
social processes.

Fairclough (2001) suggests a three-step model in doing CDA namely description of

texts, interpretation of the relationship between the productive and interpretative processes
and explanation of the relationship between discourse processes and social processes. In
the first step, linguistic features and other modalities are closely examined. Fairclough
suggests analytical moves from the micro levels of vocabulary and grammar to a higher level
of textual structures. The set of textual features he lists is quite comprehensive, however, it
should be treated as a resource from which CDA practitioners may choose to focus on a
limited number of analytical tools, relevant to particular texts and for particular purposes. The
following is a list of guiding questions for text analysis (Fairlough, 2001):

A. Vocabulary

1. What experiential values do words have?

 What classification schemes are drawn upon?

 Are there words which are ideologically contested?

 Is there rewording or overwording?

 What ideologically significant meaning relations are there between words?

2. What relational values do words have?

 Are there euphemistic expressions?

 Are there markedly formal or informal words?

3. What expressive values do words have?

4. What metaphors are used?

B. Grammar

5. What experiential values do grammatical features have?

5
  What types of process and participant predominate?

 Is agency unclear?

 Are processes what they seem?

 Are nominalizations used?

 Are sentences active or passive?

 Are sentences positive or negative?

6. What relational values do grammatical features have?

 What modes (declarative, grammatical question, imperative) are used?

 Are there important features of relational modality?

 Are the pronouns we and you used? and if so, how?

7. What expressive values do grammatical features have?

 Are there important features of expressive modality?

8. How are sentences linked together?

 What logical connectors are used?

 Are complex sentences characterized by coordination or subordination?

 What means are used for referring inside and outside the text?

C. Textual structures

9. What interactional conventions are used?

 Are there ways in which one participant controls the turns of others?

10. What larger-scale structures does the text have?

(Fairclough, 2001: 92-93)

Notably, these guiding questions concern only the verbal aspects of texts. Fairclough
(2001: 22) acknowledges the presence of visual elements in discourse and their fast
increasing significance in meaning; however, visual features are not included as an analytical
point in the guideline for describing texts.

The second stage in Fairclough’s model is interpretation, which deals with discourse
processes and their dependence on background assumptions. Interpretations are generated
through what is in the text (formal features) and what is in the interpreter (Member’s
Resources-MR). In other words, formal features are 'cues' which activate elements of
interpreters' MR, and interpretations are generated through the dialectical interplay of cues
and MR, which is referred to as interpretative procedures (Fairclough, 2001: 118). There are

6
six major domains in the interpreting process in which two domains relate to the
interpretation of context and the rest relate to the interpretation of text.

In the interpretation of texts, the entry points include surface of utterance, meaning of
utterance, local coherence and text structure and points. These are the cues in the text and
interpreters will have to draw on their MR, which is their background assumptions of
language knowledge such as knowledge of phonology, grammar, semantics, pragmatics,
cohesion and schemata, to interpret these cues. The four major domains in the interpretation
of text, however, do not require or represent separate moves but these steps are
interdependent and supportive to one another in the sense, for example, that the
interpretation of the surface of utterance will facilitate the interpretation of coherence, which
in turns facilitates the interpretation of text structure. The interdependence in the opposite
direction is also possible when, for example, interpreters will start their interpretation from an
overall judgment of textual structure, which will assist and cue the interpretation at the more
localized levels.

While the interpretation of texts takes cues in the text as the starting point, the
interpretation of context bases on external elements from the situational context. These
external elements include both physical features of the communicative and social situation
and also linguistic information outside the text being interpreted. The interpretation of the
physical setting, according to Fairclough, must consider the social order at the societal levels
and institutional levels. That is because, these features of the situation will decide the
discourse type including content, subjects, relations and connections. Linguistic elements, on
the other hand, come from the inter-textual context made up from other texts having a
connection with the text in question. CDA sees discourse as a historical event, which means
there is no discourse existing alone by itself. Rather, discourse belongs to a historical series
so it is necessary to decide which series a text belongs to, and what can be taken as the
common ground for participants, or presupposed. In a nutshell, as stated by Fairclough
(2001), the stage of interpretation makes explicit what for participants is generally implicit: the
dependence of discourse practice on the unexplicated common sense assumptions of MR
and discourse type.

In the third stage of explanation, we see discourse as part of social process, as

social practice, showing how it is determined by social structures and what reproductive
effects discourses can have on those structures, sustaining them or changing them
(Fairclough, 2001: 135). These social determinations and effects are mediated by MR: that is
social structures shape MR, which in turn shape discourses; and discourse sustains or
changes MR, which in turn sustains or changes structures. Note that social structures are
subdivided into three levels of societal, institutional and situational levels, and on all these
three levels, social effects or determinations of discourse are seen. So, explanation is a

7
matter of seeing a discourse as part of processes of social struggle, within a matrix of
relations of power.

5. Research in CDA

CDA looks at different data sources ranging from media discourse, political discourse,
textbooks to everyday discourse and so on. The issues of concern are often racism, gender
inequality or any forms of unequal power relations enacted in discourse. In what follows, a
number of studies using CDA approach focusing on different topics are briefly presented for
illustrative purpose.

In Racism in the News: A Critical Discourse Analysis of News Reporting in Two

Australian Newspapers, Teo (2000) aims to look for ideological construction of racism
embedded within the structure of newspaper reporting. The study focuses on news reports
relating to a Vietnamese gang in Australia, whose violent and drug-dealing activities have
received publicity in two Sydney-based newspapers: The Sydney Morning Herald and The
Daily Telegraph. The analysis of these reports adheres to the analytic paradigm of CDA and
is undertaken in two stages. The first, a general characterization of the newspaper discourse,
reveals evidence of a systematic `othering' and stereotyping of the ethnic community by the
`white' majority. This is followed by a comparative analysis of two reports, which surfaces
evidence of a racist ideology manifest in an asymmetrical power discourse between the
(ethnic) law-breakers and the (white) law-enforcers. The study concludes with a discussion to
explain the evidence of `Racism in the News', which both reflects and reinforces the
marginalization of recent Vietnamese migrants into Australia.

Also interested in racism as a research topic, Kim (2014) examines how racism in the
global age has evolved into a subtler form. Using both frame analysis and critical discourse
analysis, he examines the Korean media’s discourse around migrants in the past two
decades (1990–2009). Findings show that the dominant attitude taken by the Korean media
toward migrants has been positive, which is counter-intuitive considering Korea’s notorious
ethnocentrism. However, a critical discursive approach reveals that these positive discourses
have a variety of effects: they ironically ‘victimize’ and ‘objectify’ migrants, overlook the
question of how to empower migrants, and reveal misunderstandings over what it means to
embrace diversity. He concludes that in Korea, where the tension between globalization and
nationalism is intense, racial prejudice becomes more disguised under the cliché of political
correctness, such as the rhetoric of multiculturalism and diversity.

Concerned about gender inequality, Harres (1993) investigates the representation of

women in three medical texts and finds a number of sexist uses of language that perpetuate
stereotypes of gender roles. The texts she studies are chapters from a gynecological
textbook (text one), a popular book on medical knowledge (text two), and a popular text book

8
by feminist women on the topic of health (text three). All the texts studied deal with the same
issue of fibroid tumors. The study offers some findings in terms of women’s role
representation in these medical documents. Specifically, in text one, women’s experience as
medical experts is marginalized and they are rendered invisible, for example, through the use
of the generic he to refer to physicians. This presupposes that the medical profession is still
dominated by men with few women acting as surgeons. Sexist language on the level of
lexicon and syntax is also found; for instance, the use of ‘good girls’ and ‘bad girls’ in the text
suggests that good girls are those who comply with the stereotypical portrayal of women as
chaste and passive (Harres, 1993: 45). In text two, the author finds that there is little scope
for women to participate in the decision-making process that affects their health; hence, text
two reinforces women’s dependence on gynecologists. The findings offered by this study on
women’s roles in medicine as a profession show that women’s roles in aspects of the public
sphere are still underrepresented, rendering women less visible; therefore, gender
differentiation and gender stereotypes are perpetuated.

Explicitly focusing on the positive representation of women in the public sphere, Lee
(2004) attempts to examine news discourse about female officials in Hong Kong. He focuses
on prominent themes that often appear together in newspaper coverage and demonstrate
how they are articulated into a portrayal of the female officials as perfect women – who excel
in everything they do, competent in every realm they participate in and serve as models for
other women in the society. The data of the study consist of selected articles about Hong
Kong female politicians from the newspapers Ming Pao and Apple Daily from 1998 to 2001.
The general finding of the study is that in the media discourse, female officials’ career
success does not prevent them from being good mothers, wives and daughters. Such a
representation suggests that a woman can take care of her different roles by ‘keeping the
balance’ and using time efficiently (Lee, 2004: 214). The author goes on to argue that Hong
Kong media coverage of the women politicians help the politicians’ careers, but it does not
question existing inequalities in the family institution. The construction of the perfect women
image renders the work-life balance entirely invisible; hence, gender inequalities can actually
be seen to be reproduced by such positive reports. The implication of such positive coverage
is that if some women can make it, then all women can, and if some of them cannot, then it is
their fault.

In an attempt to explore gender inequality in Vietnam context, Nguyen (2011) studies

gender representations in a number of Vietnamese media articles published on and related
to the International Women’s Day - March 8, 2010 to unearth the gender ideologies that
prevail. Adopting a feminist critical discourse analysis perspective (Lazar, 2005), the study
draws on a systemic functional framework for multimodal text analysis. The analytical
framework was devised based on that of Fairclough (2001) to include aspects of the
grammars of Halliday (1994), Kress and van Leeuwen (1996)’s visual grammar, van Dijk

9
(1988)’s guidelines for news structure and Labov and Waletzky’s narrative structure analysis
(1967). The study finds that besides the glorification of women’s roles in the public domain
and their equal status in the society in liberal terms, traditional Confucian expectations for
them in the domestic domain still persist. The ever-rising social responsibilities and the
‘monolithic’ traditional roles of Vietnamese women together pose an unfair challenge for
Vietnamese women. Despite a very positive representation of Vietnamese women, the
author argues, gender inequality is discursively reconstructed by the media.

Interested in medical discourse, Wodak (2006) closely examines a complete corpus

of 102 recorded interviews between doctors and patients. She particularly studies the
occurrence of patient initiatives (asking for information, complaining about pains, etc.) and
the ways doctors deal with initiatives (answering, interrupting, ignoring and so on), which
demonstrate forms of exercising power on the part of the doctors as well as the voices of the
patients. She also pays attention to how doctors deal with problems perceived by the
patients and how doctors and patients create and maintain social relationship, for example,
by means of social turns. Some of the important findings of her study includes ‘difficult’
patients are accused of causing the disturbances, and so the scapegoat is outside the clinic
rather than anchored in the structure. The clients are made into external enemy; the
institution defends itself and functionalizes its contradictions both consciously and
subconsciously. She also finds some differences in the ways men and women communicate
their pains, for example, women often downgrade their pains while men tend to describe their
pains very concretely and men display more strongly than women a desire for explanation of
causes.

References

Chouliaraki, L. & Fairclough, N. (1999). Discourse in late modernity – Rethinking critical

discourse analysis. Edinburgh: Edinburgh University Press.

Coates, J. (1998). Language and gender – A reader. Oxford: Blackwell Publishers Ltd.

Eckert, P. & Mc Connell-Ginet, S. (2003). Language and gender. Cambridge: Cambridge

University Press.

Fairclough, N. & Wodak, R. (1997). Critical Discourse Analysis. In T. Van Dijk (ed.)
Discourse studies:A multidisciplinary introduction (pp 258-284). London: Sage Publications.

Fairclough, N. (1995). Media discourse. New York: Edward Arnold.

Fairclough, N. (2001). Language and Power (2nd ed). London: Longman.

Foucault, M. (1972). The archaeology of knowledge. London: Tavistock Publications.

10
Goddard, A. & Patterson, L. M. (2000). Language and gender. London; New York :
Routledge.

Halliday, M. A. K. (1994). An introduction to functional grammar (2nd ed.). London: Edward

Arnold.

Kim, S. (2014). Racism in the global era: Analysis of Korean media discourse around
migrants, 1990-2009. Discourse & Society, November 2012 (23): 657-678.

Kress, G. & van Leeuwen, T. (1996). Reading images (2nd ed.). London: Routledge.

Labov, W. and Waletzky, J. (1967). Narrative analysis: oral versions of personal experience.
In J. Helms (Ed.), Essays on the verbal and visual arts. Seattle: University of Washington
Press.

Lazar, M. M. (2005). Feminist Critical Discourse Analysis. New York: Palgrave Macmillan.

Lee, F. L. F. (2004). Constructing perfect women: the portrayal of female officials in Hong
Kong newspapers. Media, Culture & Society, 26(2), 207 – 225.

Litosseliti, L. (2006). Gender and language: Theory and practice. New York: Oxford
University Press.

Mills, S. (1997). Discourse. London: Routledge.

Nguyen Thi Thu Ha (2011). Gender ideologies in Vietnamese printed media. In D.Majstorovic
& I. Lassen (Eds.), Living with Patriarchy (pp. 195-218). Amsterdam: John Benjamins.

Phillips, L. & Jorgenson, M. (2002). Discourse analysis as theory and method. London: Sage
Publications.

Simpson, P. (1993). Language, ideology, and point of view. London: Routledge.

Teo, P. (2000). Racism in the News: A Critical Discourse Analysis of News Reporting in Two
Australian Newspapers. Discourse & Society, January 2000 (11): 7-49.

Thompson, J.B. (1990). Ideology and Modern Culture: Critical Social Theory in the Era of
Mass Communication. Cambridge: Standford University Press.

Van Dijk, T. A. (1988). News as discourse. New Jersey: Lawrence Erlbaum Associates.

Van Dijk,T A. (1986) News as discourse. New York: Longman.

Wodak, R. & Meyer, M. (2001). Methods of Critical Discourse Analysis. London: Sage.

Wodak, R. (1997). Gender and discourse. London: Sage Publication.

Wodak, R. (2006), Medical discourse: Doctor–patient communication. In: Keith Brown,

(Editor-in-Chief) Encyclopedia of Language & Linguistics, Second Edition, volume 7, pp. 681-
687. Oxford: Elsevier.

11