From Text Grammar to Critical Discourse Analysis

Teun A. van Dijk, University of Amsterdam

In this article, I sketch some of the developments of my work from

'text grammar' in the early 1970s, to my present studies in 'Critical

Discourse Analysis' (CDA). This seems especially useful because most

of my work translated into Spanish is still about the first, text

grammatical, phase of my academic development, whereas only little is

known as yet about my later work on discourse and racism and on

critical discourse analysis.

Text Grammar

To understand my interest in text grammars it should be recalled that

my first academic love was literary theory. After a first degree in

French Language and Literature (with special interest in Surrealist

poetry) I also studied Literary Theory. In that study I especially

focused on literary language, and wanted to know whether literature

could be characterized specifically by its typical use of language.

Under the influence of Chomsky's Transformational-Generative Grammar,

such a question at the end of the 1960s was phrased in terms of a

special set of rules that would 'generate' (that is, structurally

describe) literary texts. However, TG-Grammar never was developed to

account for text structures, and thus the aim to develop a

'generative poetics' was soon replaced by the more important aim to

focus on a generative text grammar.

The point of such text grammars was to be able to provide an explicit

description of the (grammatical) structures of texts. The most

obvious task of such a description was to account for (semantic)

coherence relations between sentences (van Dijk, 1972). Although also

sentence grammars need to make explicit how clauses of complex

sentences are semantically related, there was no serious research in

that time that could be extended to the linear semantics of

discourse. Under the influence of French structural semantics

(Greimas), I therefore first assumed that meaning relations between

sentences had to be defined in terms of the identity of the 'lexemes'

or 'semes' of the words in such sentences. This assumption later

turned out to be totally misguided, although it remained popular in

French structural semantics for years. The point is that it is not

only meaning relations between sentences that define coherence, but

rather referential relations, that is, relations between the 'things'

the sentences in a text denote, as we shall see below.

New and interesting in this emerging theory of text grammar was the

introduction of 'macrostructures', a notion unknown in any form of

sentence grammar. The point of macrostructures was that texts not

only have local or microstructural relations between subsequent

sentences, but that they also have overall structures that define

their global coherence and organization. In my early work, such

macrostructures were of two different kinds, viz., global structures

of meaning, and global structures of form. Later I introduced the

notion of 'superstructure' to refer to the latter structures, that

is, the abstract, schematic structures that organize the overall form

of the text, as we know them from the theory of narrative or the

theory of argumentation (van Dijk, 1980).

The question after more than 25 years is whether these text grammars

were wrong or right? As I see it now, I would say that the basic

principles of text grammar are still sound today, as is obvious from

the large body of work still being done in many types of sometimes

highly sophisticated discourse grammars (as in the work of Talmy

Givon and Sandra Thompson in the USA). Indeed, in the same way as a

sentence grammar explains why arbitrary sequences of words do not

define sentences, a text grammar needs to account for the fact that

arbitrary sequences of sentences do not define a text. However, the

way we actually did text grammar then was still very primitive, and

largely speculative, imprecise, and partly misguided. What remained

though was the importance of the notion of coherence in any semantic

theory of discourse, and the obvious idea that texts also are

organized at more global, overall levels of description.

Later studies, also in psychology, about such local (intersentential)

and global (textual) coherence proved to be more sophisticated. Thus,

in my book Text and Context (van Dijk, 1977), I emphasized that local

coherence between sentences should be based on referential relations

between 'facts in a possible world', thereby using the then popular

notion of 'possible world' from formal semantics and philosophy. That

is, two subsequent propositions P1 and P2 are coherent if they denote

two facts F1 and F2 that are (for instance conditionally, or

causally) related. In my later work with Walter Kintsch on the

psychology of text processing, this referential relation was not

defined in terms of facts 'in some possible world', but in terms of

mental models (see below).

Another dimension of local coherence however showed up. Sentences (or

their meanings: propositions) not only cohere because of the

relations between the facts they denote, but also because of

relations between their meanings themselves. In more formal terms:

Coherence not only was 'extensional', but also 'intensional'.

However, this meaning relation was not defined in terms of the

meanings of isolated words (as in structuralist semantics) but in

terms of the relations between whole propositions. For instance, two

propositions P1 and P2 may also be coherent if P2 is a

Generalization, a Specification, an Explanation or an Example of P1.

That is, these notions define a functional relation between

subsequent propositions: P2 has the function of being a

Generalization of P1, etc. Later work in Mann & Thompson's Rhetorical

Structure Theory (RST) further develops this type of functional

relations between the sentences of texts.

At the same time, the notion of macrostructure was now specifically

defined in terms of rather precise semantic rules for the derivation

of macropropositions from sequences of micropropositions. In this

way, we have a formal account of the familiar phenomenon of

'summarizing' a text by its most important information. In the

psychology of text processing, these macrostructures later played a

fundamental role in accounting for the way language users understand,

store and recall texts. It is however strange to see that even today

there are discourse grammars that only operate at the 'linear' level

of subsequent sentences or propositions, and fully ignore the crucial

global structures (macrostructures, superstructures) that define the

overall meaning and form of texts. One major reason for this

ignorance is probably the fact that macrostructures are still strange

objects in grammatical theory, structures that need a different

account from the structures of the meaning of sentences or relations

between sentences.

The psychology of text processing

Precisely because my linguistic colleagues, even in text grammar, did

not feel very comfortable with strange notions such as

'macrostructures', I turned to psychology for inspiration and

support, and thus encountered Walter Kintsch. This American

psychologist of Austrian descent, had written a book in 1974 (The

Representation of Meaning in Memory) that for the first time in

psychology explicitly stated that the object of study for a cognitive

psychology of understanding no longer should be isolated sentences,

but whole texts. He thereby referred to my 1972 doctoral dissertation

on text grammar. We soon took up contact, and for more than 10 years

-- and while I was writing my Text and Context book, and studies on

the pragmatics of discourse (van Dijk, 1981) -- we thus worked

together on several articles, and finally a book (van Dijk & Kintsch,

1983).

Many of the original ideas on text grammar, including the elusive

macrostructures, found their way in the cognitive theory of text

comprehension. However, whereas in the beginning the mental processes

and representations involved in processing were still too close in

form with the structures and rules of text grammars, we later

discovered that actual language users are of course much more

flexible and at the same time more fallible: They make mistakes.

Thus, the notion of 'strategic' understanding was created, which

tried to account more realistically for what language users actually

do when they understand a text. For instance, whereas a grammar will

assign a structure to a sentence or sequences of sentences that is

already (abstractly) 'given', language users will already start with

the (tentative) interpretation of the first words a sentence before

it has been fully heard or read. That is, understanding is 'on line'

or linear and not 'post hoc'. Such strategic understanding is very

fast and effective, but it is hypothetical: mistakes may be repaired

later. Also unlike grammars, language users may use information from

both text and context at the same time, or operate at several text

levels (phonology, syntax, semantics, pragmatics) at the same time in

order to interpret the text. In sum, actual text processing is very

different from formal, structural text analysis.

Moreover, language users represent sentences and their meanings in

memory. That is, a psychological theory is a theory of mental

processing. This means for instance that Short Term Memory (STM) may

have a limited capacity, and needs to be emptied regularly, after

which its interpreted information is stored in Long Term Memory

(LTM). Thus, for all levels of discourse, Kintsch and I described the

strategies involved in their analysis, interpretation and storage in

memory. Instead of conditions or rules for local coherence or the

derivation of macrostructures, we now had effective strategies for

their manipulation in the minds of the language users. The result of

such a process of understanding is a Text Representation in the

Episodic part of LTM, that is, the part of LTM that represents

people's personal experiences. The notion of macrostructure played a

basic role in this process and representation: It was a structure

construed by the language user in order to organize a text

representation in memory. In other words, macrostructures in a

psychological theory are subjective: They explain how language users

understand what is most important in a text.

Another crucial element, lacking in text grammar, needed to be

introduced, viz., knowledge. In order to understand a text, vast

amounts of social-cultural 'world' knowledge needs to be presupposed.

Indeed, it will often be impossible to define coherence relations

between sentences, or indeed to construct macrostructures, without

such knowledge. About the same time (in 1977), Schank and Abelson

published their famous book about 'scripts', taken as the abstract

ways people organize their knowledge about stereotypical events such

as shopping or eating in a restaurant. In other words, in order to

understand a text, language users will normally activate one or more

scripts, and use the relevant information in the construction of a

Text Representation in Episodic Memory.

In our later work, Kintsch and I introduced another crucial notion,

viz., that of a (situation) model, a notion that was also used,

though in a different way, by the psycholinguist Johnson-Laird in his

books Mental Models (1983). The point of that notion was that

language users do not merely construct a (semantic) representation of

the text in their episodic memory, but also a representation of the

event or situation the text is about, for instance a car accident.

This notion of model proved to be very successful. It explained many

things that hitherto were obscure or ignored. First of all, it

beautifully 'grounded' the theory of textual coherence: Sentences (or

their propositions) were simply defined to be coherent relative to a

model. That is, if people are able to construe a possible or

plausible model for a sequence or a whole text, then the text is (at

least subjectively) coherent. Similarly, macrostructures of texts

could be related to the higher level 'macrostructures' of models.

Secondly, models also provided an elegant explanation for the fact

that when people recall a text, they will usually 'falsely' recall

information that never was expressed in the original text at all.

However, if we assume that people during understanding also construct

a model of an event, and if much of the information in such a model

may be derived from more general, sociocultural knowledge, then these

'false' recalls can be explained in terms of the contents of the

model constructed for a text. That is, what people remember of a text

is not so much its meaning, as rather the model they build about the

event the text is about. This is of course trivial when we realize

that most readers are interested not so much in the abstract meaning

of a text, but in information about 'reality'. In sum: Understanding

a text means that people are able to construct a mental model for the

text. And conversely, in text production, the model is the

starting-point for all processing: People know something about an

event, and this knowledge is represented in their model of the event,

and this model will serve as the basis for e.g. telling a story about

the event.

Finally, models also explain how general knowledge is related to text

processing: Whereas models are personal, subjective and ad hoc (tied

to the present context of understanding), knowledge may be seen as a

generalization and abstraction from such models.

Learning-from-one's-experiences, thus, is typically an operation on

models. Conversely, general knowledge is used by 'instantiating'

fragments of such knowledge in specific models. Many later

experiments in cognitive psychology confirmed that models indeed play

a crucial role in understanding and recall.

Besides models of events talked or written about, language users also

build models of the communicative event in which they participant.

These so-called 'context models' feature subjective representations

of Self, the other speech participants, the Setting (Time and Place),

social relations between the participants and overall aims, purposes

and goals. Such context models therefore also form the mental basis

for the account of context-dependent speech acts, style and rhetoric.

That is, they control the ways information from event models is

selected and eventually expressed in discourse.

Whereas in this cognitive work on discourse the main focus was on

individual processing, and only limited attention was paid to

general, abstract and socially shared cognitive representations (such

as knowledge) my later work on ideology (see below) further assumed

that models and therefore the discourse based on them also feature

evaluative beliefs, that is, opinions about social and communicative

events. These opinions are partly purely personal, and partly based

on socially shared opinion-structures, viz., attitudes and

ideologies. Much of my work during the 1980s, including the work on

prejudices, focused on these social 'social cognitions' underlying

text processing.

Discourse pragmatics

Unlike most other researchers, I have a rather restricted and precise

conception of (formal) pragmatics, viz., as the study of speech acts

and speech act sequences. Whereas syntax has to do with forms,

semantics with meanings or references, pragmatics has to do with

action. And whereas syntax provides rules of well-formedness, and

semantics the conditions of meaningfulness, pragmatics formulates the

conditions of appropriateness of utterances defined as (speech) acts.

That is, pragmatics for me is not the study of the 'use' of language

(or --as Charles Morris had put it nearly 60 years ago-- as the study

of the relations between 'signs' and their users). If that would be

the case, psycholinguistics, sociolinguistics, ethnolinguistics and

discourse analysis would all be part of 'pragmatics'. This would be a

rather useless extension of the scope of pragmatics. The same is true

for the study of specific interactional strategies, such as those of

politeness of impression management. These are part of a theory of

(conversational) interaction, and not, in my view, a theory of

pragmatics. But as I said, the notion of pragmatics is rather

generally and vaguely used to denote studies on the action,

interaction and the relations between speech participants.

My few studies on pragmatics naturally focused on the pragmatics of

discourse, and not on the pragmatics of isolated sentences (van Dijk,

1981). Interestingly, the theory of semantic coherence could be used

as an example for a theory of pragmatic coherence of sequences of

speech acts: the speech acts A1 and A2 are coherent if A1 is a

possible condition for the appropriate accomplishment of A2.

Similarly, in the same way we may map sequences of propositions on

macropropositions, we may map sequences of speech acts on overall,

macro speech acts. That is, also as a whole a news report is an

assertion, and a ransom note a global threat. This may also be what

language users recall of a conversation: Not so much the detailed,

local speech acts, but rather the pragmatic upshot or 'point' of a

discourse, that is, its overall speech act. At the same time, the

notion of macro speech act was systematically related to semantic

macrostructures: The propositional 'content' of a macro speech act is

typically a macroproposition.

Discourse and racism

In 1980 my work took a rather different orientation. Also because of

my first longer stay in a "Third World" country, viz., during a

course I taught at the Colegio de Mexico, I finally decided it was

time to do something serious. Text grammars, and psychological

theories had very little to do with the real problems in this world,

and I thought that the time was ripe to work on more social and

political issues. One of these fundamental issues, especially in

Europe, was racism. I thus became interested in the ways racism is

expressed, reproduced or legitimated through text and talk.

Conversations

In several extensive projects, I thus systematically studied the ways

white majorities think, speak and write about ethnic minorities,

immigrants, refugees and about people from the South in general. One

major project, for instance, focused on how members of the majority

group in the Netherlands and in California speak about the Others in

everyday conversations. After recording, transcribing and analyzing

hundreds of spontaneous interviews in various neighborhoods in

Amsterdam and San Diego, my students and I soon found that at all

levels of structure, such conversations are rather typical (van Dijk,

1984, 1987).

For instance, at the level of topics, we found that, unlike in other

conversations, only a very limited number of topics tend to come up

when people talk about 'foreigners'. Typically, such topics are about

Cultural Differences, about Deviance (crime, violence, etc.), and

about Threats (economic, social, cultural), thereby expressing and

reproducing prevailing stereotypes and prejudices. At the local level

of semantic relations between sentences, we found that people

typically make use of specific semantic 'moves', such as the

disclaimers of Apparent Denial ("I have nothing against Blacks,

but...") and Apparent Concession ("Not all Blacks are criminal,

but..."). These moves seem to locally implement the overall

conversational strategies of Positive Self-Presentation (We are not

racist, we are tolerant, etc.), and Negative Other-Presentation (the

negative part following the but). That the 'positive' part are

largely forms of face-keeping, may be inferred from the fact that by

far the largest part of the conversations are negative about 'them'.

In an analysis of storytelling, we further found that the obligatory

narrative category of the Resolution is often lacking in stories

about immigrants. This seems to suggest that in their mental models

of ethnic events, white people indeed do not actually see a

'solution' for the 'foreigner-problem'. Stories thus focus on the

(usually negative) Complication, and therefore in fact become

complaint-stories that have a function in an argument, in which the

personal experiences of the story serve as the credible premises of

negative conclusions such as "They do not want to adapt" or "They

only come here to live of our pocket", etc.

Style, rhetoric and other formal properties of these conversations

complete this overall image. For instance, pronouns and

demonstratives may be selectively used to enhance social distance,

e.g., when speakers rather refer to their Turkish neighbors with the

pronoun "them" or "those people" than referring to them, as would be

normal, with the descriptive phrase "my (Turkish) neighbors". In

conversations we also found that people tend to hesitate, make errors

or repairs when they have to name the Others, a breach of fluency

that might be explained in terms of the (cognitive and social)

face-keeping and impression management strategies at work in speaking

about a 'delicate' topic such as minorities.

The Press

The other studies on the expression of ethnic prejudice and the

reproduction of racism in discourse focused on institutional, elite

text and talk. One major project focused on the press. We thus

analyzed many thousands of news reports in the British and the Dutch

press (van Dijk, 1991). What we wanted to know, first of all, is how

mainstream newspapers write about the Others, and what role the press

plays in ethnic relations, the propagation of stereotypes, and the

reproduction of white dominance, that is, racism, in general.

Interestingly, though not unexpectedly, many of the features of

everyday conversations can also be observed in the press, and we may

therefore assume that there are mutual relations between what the

public at large says about 'foreigners' and what their newspapers

say. For instance, also in the press, the selection of main topics

about minorities is restricted and stereotypical, if not negative.

Again, we find the special focus on Difference, Deviance and Threat.

Ethnic crime, also in the respectable and liberal press, is a major

topic, as are the many problems associated with immigration. This

means that the positive side of immigration (contributions to the

economy, cultural variation, etc.) will seldom be topical in the

press. Minorities are always portrayed as Problem People, whereas the

problems 'we' cause for 'them', such as, lack of hospitality, harsh

immigration laws, discrimination and racism, are seldom major topics.

Quotation patterns are similarly predictable. By its own rules of

balance, one would expect the press to always quote also competent

and credible minority spokespersons about ethnic events. Nothing is

less true, however: Especially white (majority) institutions and

elites are quoted. And when minorities are quoted, they can never

speak alone. This is especially the case when difficult topics such

as discrimination or racism come up: If the Others are allowed to

speak about that, it is always marked as an (unwarranted) accusation

(and hence 'racism' mostly appears between quotation marks), and not

as a fact.

These biased structures, which may also be observed in disclaimers,

descriptions of minority actors, the structure of headlines, style

and rhetoric, may be expected when we realize that the newsroom of

most newspapers in Europe is still virtually 'white': Very few

minority journalists work for major newspapers, at never at the

higher editorial levels. Similarly, minority organizations and

spokespersons are found less credible, less 'objective', and

therefore have less access to the press.

The conclusion from this large-scale research was therefore that

although in some respects the press merely reflects what the

politicians or the general public are saying about minorities, they

also have their own role and responsibility in ethnic affairs.

Especially because of their immense scope and power: Unlike a biased

speaker in a conversation, a biased news report or editorial may have

hundreds of thousands, and --as is the case for the British

tabloids-- sometimes millions of readers, and therefore have a

tremendous influence. In our research on everyday conversations, we

frequently were able to observe this influence of the press. This is

why we concluded that the press in Europe plays a central role in

maintaining (and sometimes aggravating) the 'ethnic status quo', if

not in the reproduction of racism.

These studies of the role of the press in the reproduction of racism

run parallel with another project in the 1980s, viz., a systematic

study into the structures, production and reception of news in the

press (van Dijk, 1988a, 1998b). Strikingly, very little discourse

analytical work had been done on this probably most pervasive form of

written discourse in our everyday lives. In several theoretical and

empirical studies, I thus tried to extend discourse analysis to one

of its most obvious domains of application: mass communication

research. I assumed that news discourse had a canonical structure or

'news schema' that organizes news reports, and emphasized the fact

that also news production is largely a form of text processing, viz.,

of the many source texts (written or spoken) the journalists uses

when writing of a news report. One of the empirical studies examined

how in the world press one event (viz., the assassination of

presentident-elect Bechir Gemayel of Lebanon in September 1982) was

covered. Thus hundreds of stories in a large number of newspapers in

many languages were systematically studied to see whether there were

'universals' of news reports, and/or where news reports in different

countries, languages, cultures and political systems would typically

provide a different 'picture' of the event. One of our conclusions of

this research was that news reports across the world, possibly under

the influence of the format of the reports of international news

agencies, were rather similar. Differences exist rather between the

quality press and the tabloid, popular press.

Textbooks

Another important source for ethnic stereotypes and prejudices, of

which hundreds of thousands of children are the victims, are

textbooks at school. We therefore analyzed a large number of social

science textbooks from secondary schools, and posed the same

questions as in the other projects: What do they say about

minorities, and what is their role in the reproduction of prejudice

and racism.

Although, especially in the USA, the situation is slowly improving

with the introduction of more 'multicultural' learning materials,

most textbooks, especially in Europe, continue either to ignore

minorities altogether (thus implying that Europe --and the classrooms

-- are still homogeneously 'white'), or tend to confirm simple

stereotypes or even racist prejudices. Minorities as well as people

of the South in general, are thus represented not only as 'poor',

'backward', or 'primitive', but also as criminal and aggressive, as

also is the case in the media and everyday conversations. Especially

cultural 'deviance', viz., other habits, another language or another

religion is focused upon and problematized. As is elsewhere the case

in institutional and elite discourse on ethnic affairs,

discrimination and racism are seldom topicalized, or even denied.

Parliamentary debates and other 'elite discourse'

Another major domain involved in the public discourse of ethnic

affairs, is politics. We therefore analyzed the parliamentary debates

about immigration, minorities or affirmative action, in the

Netherlands, France, Germany, Great-Britain and the USA. Obviously,

such public and official discourse is seldom openly racist, with the

exception of the statements of members of extremist right-wing

parties. However, in a more indirect and subtle way, we tend to find

many of the typical features of 'foreigner-talk' we also found in the

media and textbooks.

Especially interesting are the many strategic moves used to limit

immigration or the rights of minorities. Blaming the victim is a

major move: Minorities are blamed for their own marginal position,

their lack of work and housing, and so on. It is suggested that it is

'better for them' if they stay in their own country so that they can

'build that up'. Rather cynically it may be added in such discourse

that it would be better for 'them' if they would not be confronted

with the racism in the poor neighborhoods where they would have to

live. And of course, immigration and immigrants will primarily be

associated with financial, employment and housing problems, if not

with crime drugs, and so on.

Corporate discourse

Given their role in employment and the labor market, also the

discourse of corporate managers was studied, viz., on the basis of

interviews with personnel managers. As may be expected, corporate

managers, like other white elites, will of course deny that in their

company discrimination or racism takes place. At the same time, they

are adamantly opposed to any form of Affirmative Action (which they

will call 'Reverse discrimination'). They may be as concerned about

minority unemployment (in Holland three or more times as high as

majority unemployment), but they will always blame the Others: They

don't speak our language, they have a different culture, they have

insufficient education, they lack motivation, and so on. That other

research shows that more than 60% of employers rather hire a white

man, than women or minority men, is obviously not part of their

dominant explanations of minority unemployment. Neither is that the

case in debates in politics and the media: If minorities have

problems, they will somehow always be caused by themselves.

Elite discourse

As finally also was shown for academic discourse such as contemporary

sociology handbooks, all these forms of dominant, majority

discourses, and especially the various genres of elite discourse,

show many resemblances. Besides the ideological prejudices and

stereotypes, we thus find 'textual' stereotypes in the ways

minorities and ethnic relations tend to be described. The major

strategy in such text and talk, is that of positive self-presentation

and negative other-presentation. 'Our' racism is systematically

denied or at least mitigated, whereas their negative characteristics

are focused upon and emphasized. If racism exists in 'our' society,

then it should be sought for in the inner city ghettos, that is,

among the poor whites, and never among the elites of the boardrooms,

classrooms, newsrooms, or courtrooms. Elites thus always present

themselves as tolerant and modern, while blaming the poor social

victims. At the same time, populist politics will precisely (and

'democratically') refer to the resentment among the people against

further immigration. Also because of their role in decision making,

teaching, research, employment, the bureaucracy, information and

communication, the elites and their ethnic ideologies and practices

have a tremendous impact on society. Although maybe seldom very overt

and harsh, the elites often merely preformulate what will be soon

accepted in (white) society at large. In other words, elites play a

central role in the reproduction of racism.

Critical Discourse Analysis

After this vast research on discourse and racism, the early 1990s

required extension of this work to the more general study of

discourse, power and ideology. Thus, in one study I examined the ways

'access to (public) discourse', e.g., that of the media, is

distributed over various groups of people. I found that access to

discourse is a major (scarce) social resource for people, and that in

general the elites may also be defined in terms of their preferential

access to, if not control over public discourse. Such control may

extend to the features of the context (Time, Place, Participants), as

well as to the various features of the text (topics, style, and so

on).

Against this background, and together with other researchers in

discourse analysis and related disciplines, it was proposed that

discourse analysis should also have a 'critical' dimension. That is,

in the choice of its orientation, topics, problems and issues,

discourse analysis should actively participate, in its own academic

way, in social debates, and do research that would serve those who

need it most, rather than those who can pay most. CDA thus is

becoming a movement of scholars who focus rather on social issues

than on academic paradigms, and typically study the many forms of the

(abuse of) power in relations of gender, ethnicity and class, such as

sexism and racism. They want to know how discourse enacts, expresses,

condones or contributes to the reproduction of inequality. At the

same time, such scholars will listen to the experiences and the

opinions of dominated groups, and study the most effective ways of

resistance and dissent.

Critical scholars in many countries are now forming an international

network, CRITICS (Centers for Research Into Texts, Information and

Communication in Society), which may soon also be a 'list' on

Internet. Already in 1990 the international journal Discourse &

Society was founded as a major forum for this more critical,

social-political work.

After my own critical work on racism, one of the ways I intend to

further contribute to the foundations of CDA, is to study the

relations between discourse and ideology (using e.g., editorials in

the Washington Post and the New York Times as my main data). The aim

of this project is first to redefine ideologies in a very specific

and precise way, viz., as basic systems of 'social cognition', that

is, as socially shared mental representations that control other

mental representations, such as social group attitudes (including

prejudices) and mental models. It is postulated that ideologies also

have a fixed internal schema, viz., the same schema as that of the

self-representation of groups. Such ideologies also control discourse

and other social practices (and vice versa: text and talk often is

used to persuasively convey ideologies).

Secondly, we want to systematically investigate by what discourse

structures, such as those of semantic (topics, coherence), syntax

)word order, etc.), the lexicon, speech acts, etc., ideological

opinions become manifest in text and talk. Finally this combined

discourse-cognition dimension will be embedded in a socio-political

framework, which explains part of the rationale, the functions, and

the forms of ideologies and discourses in their social context, e.g.,

relative to the interests of social groups. This complex project,

started in 1994, will certainly take several years to complete.

Conclusion

The academic itinerary sketched above, like all stories and accounts,

also needs a conclusion, if not a moral. After more than 25 years of

doing discourse analysis, one should have learned something about the

discipline and its practitioners. One important point to emphasize is

that despite the variety of the topics I studied, and the broad

orientation of my work as a scholar, I have only a very limited grasp

of what goes on nowadays, in many countries, in the now very vast

field of discourse analysis. There are several domains and directions

of research I barely know. However, as a journal editor (of both TEXT

and Discourse & Society) and an editor of the Handbook of Discourse

Analysis and another two-volume introduction (Discourse, An

multidisciplinary contribution), I have always tried to integrate,

unify and further develop the many different domains of studying text

and talk, as one new cross-discipline of 'discourse studies'.

Another important conclusion is that my work represents only one of

many orientations, methods, theories, and directions of research.

Emerging from French Structuralism in poetics and semiotics, it soon

focused on modern linguistics, then cognitive psychology and then the

social sciences. My aim is always to be clear, pedagogical, and to

avoid esoteric writing: The crucial criterion must always be that

also our students, and not only the initiated, can read and

understand our work. This does not mean, however, that as to domain

of research, methods and style of writing, the other forms of

discourse analysis are less interesting for me. The problem is that

even over so many years, one must necessarily restrict oneself.

Much to the regret of some of many readers, I have avoided to remain

trapped in one problem or paradigm, and always like to change fields

and to explore new ways and problems of doing discourse analysis. I

may only hope that more people in discourse analysis would more often

be 'foolish' enough to leave their current field in which they feel

so well at home, and start to explore neighboring fields. It is

precisely at the boundaries of fields and disciplines that new

phenomena are observed and new theories developed.

As may be obvious from the account above, discourse analysis for me

is essentially multidisciplinary, and involves linguistics, poetics,

semiotics, psychology, sociology, anthropology, history, and

communication research. What I find crucial though is that precisely

because of its multi-faceted nature, this multidisciplinary research

should be integrated. We should devise theories that are complex and

account both for the textual, the cognitive, the social, the

political and the historical dimension of discourse. Indeed, a

problem such as racism cannot be fully understood in light of only

one discipline, or in terms of simple theories.

With the discipline as a whole, I have learned much about discourse

during the last 25 years. And yet, at the same time I know that much

of what we know is incomplete and misguided. I am not afraid to make

mistakes, and see this as the inevitable problem of all new

disciplines and original explorations of uncharted territories. If

only we are willing to admit such errors later, when other research

shows that and where we were wrong. Compared to the primitive 'text

grammars' of the early 1970s, contemporary formal work on discourse

structures is of course much more sophisticated. And compared to the

simplistic cognitive, social and interactional models of text and

talk of 20 years ago, new work on text processing, socio-political

discourse studies and conversational analysis also has much advanced.

Many different discourse genres in many social domains have been

studied: those in politics, the media, education, the law, and so on.

Levels and dimensions, as well as analytical categories, have been

multiplied, so that contemporary discourse analysis is incomparably

more complex and empirically more precise that two decades ago.

Yet, there is still a lot to do. There are still fields that are

underdeveloped (as is the case for the political science of

discourse). And more importantly, we only now have begun to study

discourse in the much more relevant framework of serious social

issues, such as racism. In my view, the real value of discourse

analysis as a discipline in society depends on its contributions to

the solution of such problems.

Short Bibliography

Major books in English:

-Some Aspects of Text Grammars (The Hague: Mouton, 1972)

-Text and Context (London: Longman, 1977)

-Macrostructures (Hillsdale, N.J.: Erlbaum, 1980)

-Studies in the Pragmatics of Discourse (The Hague: Mouton, 1981)

-Strategies of Discourse Comprehension (with W. Kintsch; New York:

Academic Press, 1983)

-Prejudice in Discourse (Amsterdam: Benjamins, 1984)

-Discourse and Communication (Ed.)(Berlin: de Gruyter, 1985)

-Handbook of Discourse Analysis (Ed.)(4 vols., London: Academic

Press, 1985)

-Communicating Racism (Newbury Park, CA: Sage, 1987)

-News as Discourse (Hillsdale, NJ: Erlbaum, 1988)

-News Analysis (Hillsdale, NJ: Erlbaum, 1988)

-Discourse and Discrimination (Detroit: Wayne State U.P, 1988)(with

Geneva Smitherman, Eds.).

-Racism and the Press (London: Routledge, 1991)

-Elite Discourse and Racism (Newbury Park, CA: Sage, 1993).

-Discourse. A Multidisciplinary introduction. 2 vols. (Ed.). (London:

Sage, 1997).

-Discourse and ideology (London: Sage, 1998, in press)