JAPANESE DISCOURSE Vol.

1(1995), 17-27

A i m s of Critical D i s c o u r s e Analysis
Teun A. van Dijk

What is Critical Discourse Analysis?

"Critical Discourse Analysis" (CDA) has become the general label for a special
approach to the study of text and talk, emerging from critical linguistics, critical semiotics
and in general from a socio-politically conscious and oppositional way of investigating
language, discourse and communication. As is the case for many fields, approaches and
subdisciplines in language and discourse studies, however, it is not easy to precisely delimit
the special principles, practices, aims, theories or methods of CDA. Yet, work in CDA
is usually characterized by the following criteria (for references, see the Bibliographical
Note at the end of this paper):

It is problem- or issue-oriented, rather than paradigm-oriented. Any theoretical and

methodological approach is appropriate as long as it is able to effectively study
relevant social problems, such as those of sexism, racism, colonialism and other
forms of social inequality.
CDA does not characterize a school, a field or a subdiscipline of discourse
analysis, but rather an explicitly critical approach, position or stance of studying
text and talk.

In order to study social problems or issues adequately, C D A work is typically

inter- or multidisciplmary, and especially focuses on the relations between
discourse and society (including social cognition, politics and culture).

Historically and systematically, C D A is part of a broad spectrum of (usually

rather marginal or marginalized) critical studies in the humanities and the social

-18sciences, e.g., in sociology, psychology, mass communication research, lau

literature and political science.

CDA-studies (may) pay attention to all levels and dimensions of discourse, viz
those of grammar (phonology, syntax, semantics), style, rhetoric, schematic
organization, speech acts, pragmatic strategies, and those of interaction, amonj
others.

Many studies in C D A are however not limited to these purely "verbal"

approaches to discourse, but also pay attention to other semiotic dimensions
(pictures, film, sound, music, gestures, etc.) of communicative events.

When studying the role of discourse in society, CDA especially focuses on

(group) relations of power, dominance and inequality and the ways these are
reproduced or resisted by social group members through text and talk.

Much work in CDA deals with the discursively enacted or legitimated structures
and strategies of dominance and resistance in social relationships of class, gender,
ethnicity, race, sexual orientation, language, religion, age, nationality or world-region.

Much work in C D A is about the underlying ideologies that play a role in the
reproduction of or resistance against dominance or inequality.

Among the descriptive, explanatory and practical aims of CDA-studies is the

attempt to uncover, reveal or disclose what is implicit, hidden or otherwise not
immediately obvious in relations of discursively enacted dominance or their
underlying ideologies.

That is, C D A specifically focuses on the strategies of

manipulation, legitimation, the manufacture of consent and other discursive ways

to influence the minds (and indirectly the actions) of people in the interest of the
powerful.

This attempt to uncover the discursive means of mental control and social
influence implies a critical and oppositional stance against the powerful and the
elites, and especially those who abuse of their power.

On the other hand, studies in C D A try to formulate or sustain an overall

perspective of solidarity with dominated groups, e.g., by formulating strategic
proposals for the enactment and development of counter-power and counterideologies in practices of challenge and resistance.
It goes without saying that not all work in CDA is characterized by all these

criteria, which are not merely descriptive of critical studies but also more or less
programmatic: This is what C D A studies ideally should try to realize. However, these
criteria provide the major traits of an approach that distinguishes it fairly well from other

-19work on discourse: CDA is essentially dealing with an oppositional study of the structures
and strategies of elite discourse and their cognitive and social conditions and consequences,
as well as with the discourses of resistance against such domination. In that respect, it goes
beyond the usual methodological criteria of observational, descriptive and explanatory
adequacy. Adding the criterion of critical adequacy presupposes social norms and values
and introduces a social or political ethics (what we find wrong or right) within the
scholarly enterprise as such. It is not surprising that such a view is often seen as "political"
(biased) and hence as "unscientific" ("subjective") by scholars who think that their
"objective" uncritical work does not imply a stance and hence a sociopolitical position,
viz., a conservative one that serves to sustain the status quo. Critical Discourse Analysis,
thus, emphasizes the fact that the scholarly enterprise is part and parcel of social and
political life, and that therefore also the theories, methods, issues and data-selection in
discourse studies are always political. Unlike other, implicitly political studies of
discourse, CDA explicitly formulates its (oppositional) stance.
How to do Critical Discourse Analysis?
As much of the scholarly aftermath of the student-revolution of 1968 has shown,
lofty ideals do not necessarily lead to good scholarship. This is also true for CDA, and
it is therefore crucial to examine in detail how critical studies of text and talk may
effectively realize the criteria and the tasks spelled out above. At a practical level, for
instance, students typically will want to know "how to do CDA," which requires us to
formulate proposals for successful research strategies. Theoretically and descriptively we
need to explore which structures and strategies of text and talk to attend to in order to discover patterns of elite dominance or manipulation "in" texts. Or conversely, focussing
on major social and political problems and issues such as sexism and racism, we need to
detail how such forms of inequality are expressed, enacted, legitimated and reproduced by
text and talk.
In sum, adequate CDA requires good theories of the role of discourse in the
enactment and reproduction of social dominance and resistance. More than theories that
merely claim descriptive or explanatory adequacy, however, successful CDA must be
effective: Its conclusions, recommendations and other practical interventions must work.
These are pretty tough criteria, and we should not have the illusion that they will be met
soon. In that respect, CDA is not only a scholarly practice, but also a scholarly program
of research.

-20Power, Access and Discourse

Since these few pages do not allow me to develop the many theoretical
frameworks involved in an account of the discursive reproduction of dominance and
resistance, I shall sketch a few prominent elements of such frameworks. A first major
problem to attend to is to examine the nature of social power and power abuse, and in
particular the ways dominance is expressed or enacted in text and talk. If social power is
(roughly) defined as a form of control, of one group by another, if such control may
extend to the actions and the minds of dominated group members, and if dominance or
power abuse further implies that such control is in the interest of the dominant group, this
means that dominant social group members may also exercise such control over text and
talk. That is, discursively implemented dominance involves preferential access to text and
context taken as a basis or resource of power, comparable to such social resources as
wealth, income, a good job, position, status, knowledge and education.
Patterns of discourse control and access are indeed closely associated with social
power. Thus, whereas ordinary people only have active access to, and control over such
discourse genres as everyday conversations with family members, friends or colleagues,
and more passive access to institutional (e.g. bureaucratic) discourse and media discourse,
the elites have access to and control over a vast array of both informal as well as public
and institutional forms of text and talk. Politicians have control over, e.g., governmental
and parliamentary discourse, and preferential access to the mass media. Journalists have
control over mass media discourse, and preferential access to a host of other forms of
official text and talk, such as press conferences, press releases, reports, and so on. Scholars
control academic discourse, such as lessons, textbooks, courses, and scholarly publications.
-Corporate managers control decision making discourse (e.g. boardroom meetings),
corporate reports and many other forms of text and talk in business contexts. And judges
not only control who can say what in the courtroom, but also have special access to such
discourse genres as verdicts.
In sum, the elites have control over, or more or less preferential access to the most
influential and important genres of discourse in society. Such access is defined in terms
of their (powerful) social or institutional position or function, and vice versa, their control
over or access to specific forms of institutional or public discourse sustain and reproduce
their power in specific communicative situations. Such control, however, may well be
legally or morally legitimate, e.g., when judges control verbal interaction in the
courtroom or have the privilege to pronounce a verdict, when police officers interrogate
a suspect or when a professor asks questions during an oral exam. Since CDA focuses on
abuse o/power, critical studies specifically focus on the (morally or legally) illegitimate

-21forms of control and access, e.g., when politicians, journalists, professors, managers or
judges prevent others from legitimate forms of text or talk (censorship), or when they
engage in talk or text that otherwise limits the freedom or rights of other participants, as
when police officers threaten immigrant defendants with expulsion, or when sexist
professors threaten female students with a bad grade if they do not consent in sexual
favors.
Such patterns of control also extend to the more detailed structures of text and
context. Thus context control may consist of control of "calling" a communicative event,
making an appointment or setting the agenda. It may involve decisions about time and
location, and who may participate in the event and in what role. Similarly, discursive
control may apply to all levels and dimensions of text and talk, such as language variants,
genres, topics, grammar, lexical style, rhetorical figures, overall organization, local and
global coherence, speech acts, turn taking, politeness forms, and so on. Thus, in a
parliamentary hearing, in the courtroom, in a police interrogation, during an oral exam
or a tax audit, the institutional power holders may oblige participants to use a specific
language variant, to answer questions, to provide information, and to use a polite style.
In all these institutional communicative events such power may be abused to censor,
intimidate or otherwise limit the freedom of less powerful participants. Detailed
contextual and textual analysis is necessary to pinpoint the sometimes subtle strategies of
such forms of discursive dominance.
Mind Control
Preferential access to, and control over discourse and its properties are forms of
the direct enactment of social or institutional power. They allow specific social actors to
engage in (verbal) action that is prohibited to others, or they may force or oblige others
to engage in discourse or to use discourse properties as desired by the powerful actor,
thereby limiting the freedom (and hence the power) of the less powerful.
Discourse, however, is not limited to verbal action, but also involves meaning,
interpretation and understanding. This means that preferential access to public discourse
or control over its properties (e.g., specific, preferred topics) may also affect the minds of
others. That is, powerful social actors not only control communicative action, but
indirectly also the minds of recipients. We know that these processes of "influence" are
exceedingly complex: The fate of "effect research" in mass communication, or the
fuzziness of critical notions such as "preferred meanings" or "manufacturing consent", all
show that controlling the mind through text and talk is not a straightforward process.
Research on attention, interpretation, comprehension, memory storage, and other aspects

-22of information processing defining "reception", has shown that these are a function of
properties of the text as well as of properties of the context, and especially of the previous
knowledge, attitudes or ideologies of recipients.
Yet, we all k n o w that news reports, political propaganda, advertising, religious
sermons, corporate directives or scholarly articles somehow influence the "minds" of
readers and hearers: They convey knowledge, affect opinions or change attitudes. We also
know that recipients, in a specific context and given their extant knowledge and beliefs,
may disregard, reject, dis-believe, or otherwise mentally act in opposition to the intentions
of powerful speakers or writers. They may have relative freedom to interpret and use
discourses as they please and in their own best interests, as is also the case for media
messages.
Despite such freedom, however, there are many constraints. These may be a
function of the power, as well as the status or credibility, of the speaker/writer, as well
as a function of the properties of text or talk: People may be lied to, manipulated,
persuaded or otherwise influenced against their best interests, or in the interests of the
powerful speaker/writer. People may lack alternative sources of information, or they
may lack the knowledge of rules and strategies of grammar or discourse, they may not
have sufficient knowledge to detect lies and manipulation, or strong counter-opinions or
counter-ideologies to argue against and reject influential text and talk.
In sum, powerful speakers may control at least some parts of the minds of
recipients. C D A studies the ways in which such influence and control of the mind is
socially or morally illegitimate, e.g., when powerful speakers self-servingly control the
minds of others in a way that is in the interest of the powerful. Since action is based on
mental models of actors, models which in turn embody social knowledge and attitudes,
influencing such models or the beliefs on which they are based may be an effective way
to (indirectly) control the actions of other people.
For instance, if European politicians and media attribute major social problems
(such as unemployment or a housing shortage) to immigration and immigrants, they may
thus influence the beliefs of large segments of the majority, and thus indirectly the models
that underlie racist stories or discriminatory action of dominant group members, as well
as other expressions of prejudice and resentment that may again be used by the politicians
to legitimate political decisions on immigration restrictions.
Again, all this is more or less known in the cognitive and social psychology of
language, discourse and communication.

What we hardly know, however, is what

properties of speakers and especially of discourses are the preferred and most effective
means of such forms of mind control. To be sure, political propagandists, advertisers,

-23journalists or professors have enough practical experiences to have some idea about what
kind of messages will have what kinds of effects. They know how to effectively change
the knowledge and opinions of recipients, and what kind of social actions will typically
result from such mind control. However, such intuitive, practical knowledge is largely
implicit, and few theories of communicative influence or discourse processing spell out
in detail under which precise textual or contextual conditions, and for what recipients,
specific forms of mind control will occur.
Besides spelling out the detailed forms of discourse control and access as described
above, CDA needs to focus on the morally illegitimate forms of discursive mind control
by the powerful. It needs to specify how powerful speakers control properties of text and
talk in such a way that they are able to monitor the mind of recipients in their own
interests. For instance, by emphasizing specific topics at the expense of others (e.g.,
through headlines or summaries) or by preventing others to address other topics, they
may influence the overall (mental) model structures that are involved in discourse
comprehension. In the example mentioned above, defining immigration as a problem or
as a threat (e.g. as an "invasion of refugees") instead of as an economic and cultural
contribution to the country or city, is one of the persuasive ways dominant speakers are
able to mold the models of (white) citizens, if these have no alternative representations of
immigration. In the same way, immigrants and their actions may be described by negative
lexical terms or rhetorical figures. Conversely, negative actions of the elites (or of "our
own group") may be de-emphasized topically, stylistically or rhetorically, whereas "our"
positive actions and properties may be textually emphasized in interactional strategies of
face keeping and positive self-presentation.
At all levels of text and talk we may thus postulate structures that preferentially
affect the structure and content of mental models, as well as the more general and abstract
forms of knowledge, beliefs, opinions, attitudes or ideologies that are shared by groups
of recipients. For instance, strategic generalizations in discourse ("this always happens like
that"; "they are all the same," etc.) may influence the generalization of ad hoc situation
models to more abstract group attitudes and prejudices. Specific forms of local coherence
may facilitate attributing negative properties to THEM (e.g. "lacking motivation") and
positive ones to US (e.g., employment programs) in explanations of minority
unemployment. Generally, rhetorical figures such as hyperboles and metaphors may
similarly be used to emphasize OUR good properties and THEIR bad ones, whereas
figures of mitigation may de-emphasize OUR negative properties and THEIR good ones.
Narrative structures may be deployed to detail and structure models of events in specific
ways, and generally to enhance credibility, for instance as forms of "personal experience

-24evidence" in argumentations that emphasize THEIR negative properties. Similar remarks

hold for argumentative structures, specific speech acts (e.g., accusations vs. excuses) and
many face-keeping strategies in talk. In sum, all other things being equal, we must assume
that the vast array of different discourse structures not only function to strategically enact,
express, signal, disguise, emphasize or legitimate, social position, and hence power, of
speakers, but also to control the minds of recipients in desired ways. As soon as such
control is in the interest of the powerful and against the interests of the less powerful
(either as recipients or as people talked about), CDA has a task to examine the textual and
mental structures and strategies involved in much more detail.
Conclusion
Critical Discourse Analysis, as described above, is a special approach in discourse
analysis which focuses on the discursive conditions, components and consequences of
power abuse by dominant (elite) groups and institutions. It examines patterns of access
and control over contexts, genres, text and talk, their properties, as well as the discursive
strategies of mind control. It studies discourse and its functions in society and the ways
society, and especially forms of inequality, are expressed, represented, legitimated or
reproduced in text and talk. CDA does so in opposition against those groups and
institutions who abuse their power, and in solidarity with dominated groups, e.g., by
discovering and denouncing discursive dominance, and by cooperating in the empowennent of the dominated.
Such lofty aims and tasks can be effectively realized only when CDA is a
thorough theoretical and descriptive enterprise which eschews fuzziness, impressionism
or superficiality. CDA therefore presupposes detailed theories of the structures and
strategies of text and talk, and especially explicit theories about the roles and functions of
such properties in social contexts and cognitive processing. As long as we lack such
insights, also CDA will be merely a program and not an adequate social and scholarly
practice. On the other hand, CDA has its own aims, research programs, theoretical
frameworks and methods. It may focus on properties of language and discourse ignored
elsewhere, as has been shown in feminist studies. It may emphasize the relevance of
power and dominance in studying text and talk, or of social arrangements and the social
order in general in their relation to language use and communication. It hardly needs to
be stressed that such a complex and vast program is as yet only in its infancy. Most work
in CDA still lies ahead of us.

-25Bibliographical Note
There as yet no general, book-length introductions to Critical Discourse Analysis.
Against the combined background of Marxist and Neo-Marxist approaches to the links
between language and social class, of Frankfurt School Critical Theory, of critical
sociolinguistics, British Cultural Studies and functional linguistics, among other directions
of research too vast to be bibliographically detailed here, one of the first collections of
critical linguistics is that of British and Australian linguists Towler, Hodge, Kress &
Trew (1979)(see also *Fowler, 1991; *Hodge & Kress, 1988; *Kress & Hodge, 1993), most
of whom work in the functional framework of Halliday's systemic grammar and
semiotics. Similarly in Britain, and within a more neo-marxist tradition, *Fairclough
(1989, 1992) explores the relations between language and power. *Chilton (1985, 1988)
specifically focused on critical studies of political discourse, e.g., about so-called "Nukespeak" (for other more or less critical studies of political discourse, see also *Geis, 1987;
*Seidel, 1988; *Wilson, 1990). More recently, such critical linguistics tends to be
subsumed under the broader labels of Critical Discourse Analysis or Social Semiotics. In
a continental tradition of critical sociolinguistics and text linguistics, Wodak and her
associates focus on institutional power, male dominance, racism and anti-Semitism, among
many other critical topics (among much other work, see, e.g., *Wodak, 1989; *Wodak,
et al., 1987; *Wodak, et al., 1990). My own critical work in Amsterdam focuses on racism
and on the media, and more recently on the relations between discourse and ideology (see,
e.g., ""van Dijk, 1987, 1988, 1991, 1993). A recent issue of Discourse & Society publishes
several papers of these critical linguists and discourse analysts. Both in North America
and in Europe, many papers and some book collections have focused on the study of
language and power (*Kedar, 1987; *Kramarae, Schulz, & O'Barr, 1984; *Lakoff, 1990;
*Ng & Bradac, 1993), but not always within a discourse-analystic framework. Similarly,
also work on professional language and discourse often takes a critical position (*Fisher
& Todd, 1986). Mainstream conversation analysis has as yet avoided taking critical
positions, but new developments point to an increasing interest in linking the study of
talk with social structure (*Boden & Zimmerman, 1991; *Drew & Heritage, 1992).
Critical language and discourse studies would have been unthinkable without the
powerful example of feminist work in the study of gender, language use and discourse
(among many other studies, see, e.g., "'Bull & Swan, 1992; *Lakoff, 1975; T h o m e ,
Kramarae & Henley, 1983; *Fisher & Todd, 1988). For further critical studies on
discourse, see also the contribitions in Discourse & Society. "Whereas the references given
here are all in English, there is of course a rich tradition of critical studies in Germany,

-26France, Italy, Spain and Latin America, among other countries, which however cannot
be detailed here, except with reference to the work of Techeux (1975; 1982) who inspired
much work on language and ideology; of Foucault (e.g., Toucault, 1980) who is often
cited in work on discourse and power, and the work of Habermas (e.g., *Habermas 1991)
on communication and its influence on critical pragmatics, among many others.

References
Boden, Deirdre and Don H. Zimmerman, (eds.) 1991. Talk and Social Structure: Studies
in Ethnomethodology and Conversation Analysis. Cambridge: Polity Press.
Bull, Tove and Toril Swan, (eds.) 1992. Language, sex and society. Special issue of
International Journal of the Sociology of language, 94.
Chilton, Paul, (ed.) 1985. Language and the Nuclear Arms Debate: Nukespeak Today.
London: Pinter.
.1988. Orwellian Language and the Media. London: Pluto Press.
Drew, Paul and John Heritage, (eds.) 1992. Talk at Work: Interaction in Institutional
Settings. Cambridge: Cambridge University Press.
Fairclough, Norman L. 1989. Language and Power. London: Longman.
. 1992. Discourse and Social Change. Oxford: Polity Press.
Fisher, Sue and Alexandra Dundas Todd, (eds.) 1986. Discourse and Institutional Authority:
Medicine, Education, and Law. Norwood, NJ: Ablex.
. (eds.) 1988. Gender and Discourse: The Power of Talk. Norwood, NJ: Ablex
Publishing Company.
Foucault, Michel. 1980. Power/Knowledge: Selected Interviews and Other Writings, 19721977. New York: Pantheon.
Fowler, Roger. 1991. Language in the News: Discourse and Ideology in the Press. London:
Routledge.
, Bob Hodg, Gunther Kress, and T. Trew. 1979. Language and Control. London:
Routledge & Kegan Paul.
Geis, Michael L. 1987. The Language of Politics. New York: Springer.
Hodge, Robert and Gunther Kress. 1988. Social Semiotics. London: Polity Press.
Kedar, Leah, (ed.) 1987. Power Through Discourse. Norwood, NJ: Ablex.
Kramarae, Cheris, Muriel Schulz and William M. O'Barr. (eds.) 1984. Language and
Power. Beverly Hills, CA: Sage.
Kress, Gunther and Bob Hodge. 1993. Language as Ideology. Second Edition. London:

-27Routledge and Kegan Paul.

jLakoff, Robin T. 1975. Language and Woman's Place. New York: Harper and Row.
_
. 1990. Talking Power. The Politics of Language. New York: Basic Books.
jsfg, Sik Hung and James J. Bradac. 1993. Power in Language. Verbal Communication
and Social Influence. Newbury Park: Sage.
pecheux, Michel, (ed.) 1975. Analyse du discourse. Langue et ideologies. Langages 37.
Paris: Didier-Larousse.
Seidel, Gill, (ed.) 1988. The nature of the right: A feminist analysis of order patterns.
Amsterdam: Benjamins. Thome, B., Kramarae, C, & Henley, N. (eds.). (1983).
Language, Gender, and Society. Rowley, MA: Newbury House.
Van Dijk, Teun A. 1987. Communicating Racism. Newbury Park, CA: Sage.
.
. 1988. News Analysis: Case Studies of International and National News in the Press.
Hillsdale, NJ: Erlbaum, 1988.
. 1991. Racism and the Press. London: Routiedge.
.1993. Elite Discourse and Racism. Newbury Park, CA: Sage.
Wilson, John. 1990. Politically Speaking. Cambridge: Blackwell.
Wodak, Ruth, (ed.) 1989. Language, Power and Ideology. Amsterdam: Benjamins.
Wodak, Ruth, R. De Cillia, K. Bliiml and E. Andraschko. 1987. Sprache und Macht.
(Language and Power). Wien: Deuticke.
. p. Nowak, J. Pelikan, H. Gruber, R. De Cillia and R. Mitten. 1990. "Wirsind
unschuldige Tdter. " Studien zum antisemitischen Diskurs im Nachkriegssterreich.
("We are innocent offenders." Studies into antisemitic discourse in postwar
Austria). Frankfurt/Main: Suhrkamp.