Pennycook (1994) Incommensurable Discourses

Incommensurable Discourses ?
Downloaded from https://academic.oup.com/applij/article/15/2/115/184979 by Weil, Gotshal, & Manges user on 27 November 2023
ALASTAIR PENNYCOOK
University of Hong Kong
This paper is an attempt to come to terms with different understandings of the

term discourse. By comparing the common use of discourse analysis in
applied linguistics with its use both in critical discourse analysis and in a
Foucauldian use of the term, I try to show how these different approaches imply
profoundly different understandings of the relationship between language, the
individual, ideology, and society. Ultimately, I argue that there are limitations
to both the common applied linguistic and the critical approaches, and that it
would be useful to explore further the possibilities raised by a Foucauldian
understanding of discourse analysis.
INTRODUCTION: WHICH IS BIGGER-LANGUAGE OR DISCOURSE?

My desire to write this paper has its origins in a deep concern that at times I am
no longer able to communicate with my colleagues in applied linguistics, that at
times we are no longer engaged in the same discourse. The difference over
understandings of what we mean by 'discourse' came to the fore a few years ago
during a conversation over coffee with a colleague in Singapore, a conversation
that ended in frustration and misunderstanding. I was discussing some recent
research and had explained that my interest in the global spread of English was
not so much concerned with particular linguistic features of 'new Englishes' as
with cultural and political implications of this spread. One interesting way to
take this up, I suggested, was in the relationship between English and particular
discourses with which it is associated. Thus, I argued, because of the particular
connections between English and the global spread of capitalism, science and
technology, Western academic work, American popular culture, democracy,
environmentalism, and so on, to speak in English often implied taking up a
particular position in the discourses put into play by the spread of these forms of
culture and knowledge. Ultimately, I suggested, what could be said in English
was in part limited and produced by the deployment of these global discourses.
To my colleague, this argument made little sense, but in trying to clarify our
differences, he pointed to what seemed a key conceptual divergence in how the
two of us understood language and discourse. Was I suggesting, he asked, that
language use was, in a sense, determined by discourse, and that discourse,
therefore, operated at a higher level than language? Surely, he argued, it was the
other way round: language was the larger concept and discourses occurred
within a language. We appeared to be talking from utterly different positions: for
him, discourse was an instance of language use; for me, language use was an
instance of discourse. Or, to put it another way, on the one hand, there is a
position that emphasizes language as a system and then looks to discourse
analysis to explain how various contextual factors affect language in use, and, on
Applied Linguistics, Vol. 15, No. 2 © Oxford University Press 1994
116 INCOMMENSURABLE DISCOURSES?
the other, there is a position that looks at how meanings are a product of social
and cultural relationships and then turns to see how these may be realized in
language. From one point of view, we were participants in the same discourse
(the same conversation), while from the other we were each taking up positions
in different discourses (different ways of understanding).
The first view of discourse, which I take to be the predominant one in applied
linguistics, is captured by Brown's (1980:189) explanation of discourse analysis
as the exploration of how language is used beyond the sentence level: 'the
analysis of the functions of language can be referred to as discourse analysis to
capture the notion that language is more than a sentence-level phenomenon'.
The other position is exemplified by Ball's (1990a: 3) argument that 'the issue in
discourse analysis is why, at a given time, out of all possible things that could be
said, only certain things were said'. This second view of discourse analysis
immediately raises some rather different questions from those of the more
common applied linguistic understanding: first, the question asked is 'why', and
not 'what' or 'how', suggesting that there is a shift here from description to
explanation; second, the focus is not so much on how meanings are constructed
between sentences, but rather on how meanings come to be articulated at
particular moments; and third, there is an emphasis not so much on how
language works once it has been uttered, but rather on how utterances come to
be made, and how those choices are both produced and constrained. To put it
another way, as Luke, McHoul, and Mey (1990: 40) suggest, 'Paradoxical as it
may appear to the structural linguist, we can show that discourse is not
something that language does. Discourse is not a mere function of language.
Rather discourse is, to put it crudely, the condition by which language as a
structure or a system exists'.
This paper is an attempt to explore and account for these different positions.
Is there, despite the apparently deep divide, some way in which they can be
reconciled? Was the apparently diametric opposition that arose from the
conversation a real effect of profound difference or are they merely extremes at
either end of a general concern with language use and its contexts? Or do they
operate from such fundamentally different epistemological premisses that they
can never be reconciled? Are these two positions ultimately incommensurable?
In the next section, I shall give a brief account of the development of a notion of
discourse in what I see as mainstream applied linguistics.1 Following that, I shall
discuss the views of language, discourse, and ideology in the work of various
critical discourse analysts. From there, I shall move to a discussion of a
Foucauldian notion of discourse, a position which is close to the one from which
I was speaking in the above conversation. Finally, I shall illustrate the
implications of taking up one or other position for doing work in applied
linguistics.
DISCOURSE AS SUPRASENTENTIAL LANGUAGE USE

The term discourse has now entered the vocabulary of many language teachers
and applied linguists. There seems to be fairly broad agreement on its two core
ALASTA1R PENNYCOOK 117
meanings: first, language in use; and second, the relationships between

sentences. Thus, the Longman Dictionary of Applied Linguistics defines
discourse first as 'a general term for examples of language use, i.e. language
which has been produced as the result of an act of communication', and
second—in contrast to grammar, which deals with clauses, phrases, and
sentences—as referring to 'larger units of language such as paragraphs,
conversations, and interviews' (Richards, Platt, and Weber 1985: 83). Dis-
course analysis, then, is 'the study of how sentences in spoken and written
language form larger meaningful units such as paragraphs, conversations,
interviews, etc' (ibid.: 84). Discourse and discourse analysis are now widely
used in many domains of applied linguistics and with remarkable agreement on
what these terms mean. Brown and Yule (1983:1) define discourse analysis as
'the analysis of language in use'. For McCarthy (1991: 5), discourse analysis is
'concerned with the study of the relationship between language and the contexts
in which it is used'. Cook (1989:6) suggests that discourse is 'language in use, for
communication', and discourse analysis is 'the search for what gives discourse
coherence'. Finally, according to Hatch (1992: 1), discourse analysis is 'the
study of the language of communication—spoken or written'. What I want to
suggest in this very brief overview of discourse and discourse analysis in applied
linguistics is that:
1. this commonly-held view of discourse is a very particular one, and it is

important to account both for the adoption of this view of discourse and for
its rapid spread and acceptance; and
2. this view of discourse, while useful in helping to expand our thinking about
language and language teaching, also has a number of limitations.
Discourse analysis, as it grew up in applied linguistics, was a synthesis of a

number of different approaches to the analysis of extended chunks of language
use: text linguistics, conversation analysis, and the ethnography of speaking.
Two key texts that drew links between discourse analysis and language teaching
were Coulthard's (1977) broad overview of discourse analysis and Widdow-
son's (1978) narrower application of a form of text linguistics to communicative
language teaching. This interest in discourse analysis can be partially explained
in terms of the pragmatic needs of language teaching and language teaching
theory for more comprehensive accounts of language use. As applied linguistics
started to emerge in the 1960s and 1970s from its earlier more literal sense of
being the application of linguistic theory to language teaching, there was a
growing awareness that not only did it need aricherview of psychology than that
afforded by behaviourism, but it also needed far broader views of language and
communication. Thus, when Coulthard and Widdowson introduced their
versions of discourse analysis, they were broadly welcomed by many people
involved in language teaching. Soon we had better tools for analysing how texts
were put together, how conversations worked in terms of turn-taking and
sequencing, how intonation was linked to the larger conversational structures in
which it occurred, and how patterns of language use might differ across different
cultural groups.
These shifts in applied linguistics can be seen not only in terms of pragmatic
needs, but also in terms of a general epistemological shift towards a more
empiricist and pragmatist view that stressed, for example, the need for data from
actual mother-child interactions, for research into the real happenings of
classrooms, or for the use of authentic materials in language teaching. As part of
this broader epistemological shift, applied linguists started stressing the import-
ance of 'real communication' in the classroom, of language acquisition as a
social process, and of communicative competence as a goal of language
teaching. In 1980, in their introductory article, 'Discourse analysis, what's
that?', to Larsen-Freeman's (1980) edited book, Discourse Analysis in Second
Language Research, Hatch and Long (1980: 1) pointed to the need 'to go
beyond sentence-level syntax if we wanted to understand how meaning is
attached to utterances'. Discourse analysis, they argued, opens new avenues for
understanding what language is learnt and how learners learn language. This led
to a much greater focus on language in use and interaction than on mere
linguistic input (see Hatch 1983; Long 1983). Meanwhile, work on the
definition of communicative competence was also starting to explore the
importance of discourse. Although left out of Canale and Swain's (1980) highly
influential formulation (linguistic, sociolinguistic, and strategic competence),
because they considered discourse to be insufficiently theorized at the time,
Canale (1983: 9) added 'discourse competence' to his later revision, defining it
as 'mastery of how to combine grammatical forms and meanings to achieve a
unified spoken or written text in different genres'. Thus, in a matter of a few
years, discourse spread rapidly through the field of applied linguistics, so that
today it is rare to find people involved in language teaching who are unaware of
the significance of discourse for teaching reading, writing, intonation, or spoken
language, and for the evaluation of students' communicative competence.
What I want to suggest, however, is that valuable though these developments
have been, the concepts of discourse and discourse analysis that are used in
applied linguistics are also limited. First, although there is an apparent dis-
juncture between the British tradition (e.g. Cook 1989; McCarthy 1991)—with
its focus on either text linguistics in the Widdowson mould, or spoken discourse
structures in the Birmingham style (Sinclair and Coulthard 1975)—and the
American tradition (e.g. Hatch 1992)—with its focus on the process of con-
versational negotiation (conversation analysis and ethnography of speaking)—
both share a view of discourse analysis that relates language as a system to
different but, as I shall argue, decontextualized contexts. Second, nearly all this
work takes as its principal focus the relationship between form and function-
how lexico-grammatical forms come to take on particular meanings in different
contexts—which, as will become clearer later, is only one particular way of
exploring how meaning is created in language use. And finally, although I argued
above that there were pragmatic reasons why discourse analysis had been so
rapidly adopted in applied linguistics, I think it is also crucial to understand that
ALASTAIR PENNYCOOK 119
pragmatism is itself an ideology (or discourse) and one that infuses a great deal
of applied linguistic thinking.
As applied linguistics sought to go beyond the dominant paradigms of
structuralism (in which meaning was either ignored or seen as held in place by
the language system itself), it focused predominantly on the relationship
between structures and their contexts. Thus, Widdowson (1978: 29), for
example, focuses on how it is that a conversation such as 'A: That's the
telephone. / B: I'm in the bath. / A: OX.' can be coherent. He does so by
producing a series of dichotomies that refer to the linguistic system on the one
hand and language use on the other: usage and use, correctness and appro-
priacy, signification and value, sentence and utterance, proposition and
illocutionary act, cohesion and coherence, and so on. Similarly, McCarthy
(1991: 7) asks how it is we can understand the comic interchange 'Ernie: Tell
'em about the show. / Eric (to the audience): Have we got a show for you tonight
folks! Have we got a show for you! (aside to Ernie) Have we got a show for
them?' and answers by appeal to the relationship between form and function.
Thus, what these approaches to discourse analysis have in common is a focus on
the reparation of the linguistic/semantic split occasioned by structuralist
linguistics. The issue, therefore, is tofindways in which the meaning of language
in use can be explained in terms of a relationship between the meanings of words
and sentences (forms) and the meanings supplied by the context (functions).
This is done by looking at:
1. the intentions or purposes of the language use (language functions or speech
act theory);
2. the adherence to or flouting of conversational rules (Grice's maxims or
Goffman's 'universalconstraints' on communication systems and rituals); or
3. general forms of inferencing from the context or background knowledge.
Thus, while discourse analysis has moved from a purely linguistic analysis of
suprasentential relations (which was how the term discourse analysis was
originally used by Zellig Harris 1952), and while applied linguistic use of
discourse analysis has broadened the options for considering language use, the
principal focus has been on the reparation of the structuralist linguistic/
semantic split, rather than on an exploration of the wider context of 'contexts',
the formation of background knowledge, or why and how a person comes to say
certain things. In this view of discourse, the basic concepts are contexts,
language, and discourse, and the basic issue is how the context affects the use of
language (discourse). The language-using subject is seen as a more-or-less
autonomous actor who establishes meanings by intention and inference. By
contrast, when we come to look at alternative understandings of discourse in the
following sections, it will become clear that they operate with a much more
politicized view of the subject who is, indeed, called into being—'interpellated'
in Althusser's (1971) terms—by discourse or ideology. Introducing some terms
that I will use again later, the common applied linguistic sense of discourse
analysis is basically a two-level view of discourse, with the 'micro' level being
language forms (lexical, grammatical, intonational/phonological) and the

'macro' the context of utterance (speaker intentions, background knowledge,
conversation/text structure).
There are two final reasons why a degree of caution is necessary towards the
applied linguistic adoption of discourse analysis. My first concern is with the
extent to which discourse analysis (see van Dijk 1985, and Stalpers 1988, for a
counterargument) and applied linguistics are now being considered as disci-
plines in their own right. As Grabe (1990: vii) says in his foreword to an edition
of the Annual Review of Applied Linguistics devoted entirely to discourse
analysis, 'Over the last ten years, discourse analysis itself has evolved to a point
at which it is emerging as a discipline in its own right, rather than being viewed
primarily as a set of disparate research techniques and approaches'. More
significant, however, is the extent to which applied linguistics is also now
conceived of as a discipline. For some this may be cause for celebration, but as
Foucault's work (e.g. 1970, 1972, 1979) on the foundation of disciplines has
shown, it is this process of discipline formation that is crucial in determining
which forms of knowledge are to be valued and upheld and which are to be
devalued and discarded. At a time, therefore, when applied linguistics is
solidifying its canon of disciplinary knowledge, with all the attendant consolida-
tions and discardings that such a process implies, there is a danger that
discourse analysis as commonly conceived in applied linguistics will increa-
singly come to define the questions that can be asked about language use. There
appears to be the feeling, furthermore, that by adding the notion of discourse to
the 'lower' levels of linguistic analysis (phonology, morphology, syntax), we
have now arrived at a more-or-less complete analysis of language. It is common,
therefore, to discuss the two traditions of discourse analysis—British and
American—and even to suggest that they may be irreconcilable (Cook 1989),
but rarely is it considered that there may be a vast range of other considerations
beyond the reconciliation of these two approaches. Thus, while discourse
analysis has clearly opened up a lot of important possibilities for language
teaching, it has also led to an over-confidence and, because of its narrow focus, a
concomitant closing down of broader possibilities.
Finally, I think there are serious problems with applied linguistics' political
quietism. I suggested earlier that there were pragmatic reasons for the adoption
of discourse analysis in applied linguistics, but it is also important to understand
pragmatism as an ideology (see Chua Beng-Huat 1983). Santos (1992) raises
the problem of the ideological stance of pragmatism when she points to the
almost complete absence of discussion of the political and ideological
dimensions of ESL writing and of ESL in general. The tendency to stress the
pragmatic over the ideological, she suggests, allows Swales (1990), for example,
to sweep aside the ideological implications of discourse communities and
genres. As Harland (1987) argues, this emphasis on realism, materialism, and
pragmatism is rooted in an Anglo-Saxon tradition of language study, from the
logical positivists and ordinary language philosophers to modern-day applied
linguists. Thus, the possibilities of dealing with broader social, cultural, or
political contexts of discourse are denied by appeal to an ideology of prag-

matism.
CRITICAL DISCOURSE ANALYSIS
In the last section, I suggested that while discourse analysis as conceived in
applied linguistics had opened up a number of important dimensions for
language teaching, at the same time, by focusing on the relationship between
language forms and a limited sense of context (immediate surroundings,
speakers' intentions, background knowledge, or conversational rules), it tended
to be confined to a narrow understanding of the larger social, cultural, and
ideological forces that influence our lives. It does not seem sufficient to stop with
a version of discourse analysis that posits a completely free-willed subject and
language use free of ideological conditions.2 To explore these ideas further I
would hke to look at the work of the various people working under the rubric of
'critical discourse analysis' (CDA), since this work has gone much further
towards addressing the ideological dimensions of discourse. Although the
various approaches to critical discourse analysis (Fowler, Hodge, Kress, and
Trew 1979; Fairclough 1985, 1989; Kress 1985, 1990; Wodak 1989, 1990)
differ in a number of ways, they share a commitment to going beyond linguistic
description to attempt explanation, to showing how social inequalities are
reflected and created in language, and to finding ways through their work to
change the conditions of inequality that their work uncovers. Fairclough (1989)
describes his two principal goals as: first, helping to 'correct a widespread
underestimation of the significance of language in the production, maintenance,
and change of social relationships of power'; and, second, helping to 'increase
consciousness of how language contributes to the domination of some people by
others, because consciousness is the first step towards emancipation' (ibid.: 1).
Of significance here is these researchers' concern with, on the one hand, the
analysis of various forms of discourse as instances of language use, and, on the
other hand, with locating these discourses within wider questions of social
power. Discourse, for Fairclough, is 'language as social practice' (ibid.: 17), a
definition that has both similarities with and differences from the notion of
discourse as language use discussed in the previous section. It is similar in that
discourse is also used to mean chunks of language as it is actually used. It differs,
however, in at least two respects. First, language as social practice differs from
language use to the extent that it relates language to other social practices, rather
than leaving it in a separate domain. Second, such language practice is seen as
socially determined. Importantly, then, Fairclough is not looking for a relation-
ship between language and society in the dichotomizing fashion of post-
Saussurean linguistics, but rather is showing that language use is always a social
act in itself. Such acts, furthermore, are not the individualistic acts of language
users in cognitive isolation, but rather are determined by the larger social and
ideological conditions of society.
In his recent summary article, Kress (1990: 85) suggests that critical
discourse analysts share with other discourse analysts a focus on texts and their
contexts, but do so with a critical element: 'By denaturalizing the discursive

practices and the texts of a society, treated as a set of discursively linked
communities, and by making visible and apparent that which may previously
have been invisible and seemingly natural, they intend to show the imbrication
of linguistic-discursive practices with the wider socio-political structures of
power and domination'. He lists a number of shared critical assumptions,
amongst which are a view of language as social practice, texts as social products,
speakers as differentially located, and meanings as products of those socio-
political relationships. Such assumptions suggest that language is an opaque
rather than a transparent medium, that the concept of a homogeneous language
system is problematic because people always have differential access to that
system, and that discourse analysis must always look to social power, history,
and ideology to understand meaning.
Most critical linguists, then, view language somewhat differently from its
dominant conception in mainstream linguistic thought and discourse analysis.
Fowler et al. (1979: 1), for example, argue, first, that the language we use
'embodies specific views—or "theories"—of reality'. The key here, however, is
not to view this in terms of 'a language' embodying 'a world view', but rather to
accept that different uses of language within one language imply particular
understandings. Second, they suggest that 'variation in types of discourse is
inseparable from social and economic factors' and thus 'linguistic variations
reflect and, what is more, actively express the structured social differences that
give rise to them' (ibid.). Third, 'language usage is not merely an effect or reflex
of social organization and process, it is a part of social process' (ibid.). These
views, which locate language use as a social process and relate linguistic choice
and variation to social and economic difference, start to flesh out the concept
that language use is always embedded in its contexts. These contexts, further-
more, are not merely 'speech events', 'text genres', and so on, but rather are
always contexts based on an understanding of social, cultural, and political
difference.
If to some, this concentration on language in social contexts suggests socio-
linguistics, it is worth pointing out that these writers distance themselves very
carefully not only from mainstream linguistics, but also from mainstream
sociolinguistics. Sociolinguists, Fowler et al. suggest, are 'at best naive in accept-
ing the social structures they describe as neutral; while at worst they collude in a
view of existing social structures as unchangeable' (ibid.: 2). Indeed, as Fowler
and Kress (1979) suggest, the dualism between linguistics and sociolinguistics is
in itself harmful, since it implies that language and meaning can be studied in the
abstract through linguistics, leaving language and its social contexts as an
adjunct to be studied through sociolinguistics. This insistence on understanding
the social contexts of language use is not, then, the addition of sociolinguistic or
communicative competence to a version of linguistic competence, but rather an
attempt to understand language use within the social and ideological structures
of society. As Urwin (1984) points out, Saussure's universal competence, based
as it is on a notion of a unitary subject and a common core rationality, is
reproduced, not challenged, by a similar conception in sociolinguistic theory.
What emerges from the views being discussed here, by contrast, is an
understanding of humans as socialized; human subjectivities and language use
are produced within particular social and cultural contexts, contexts in which
ideological forms and social inequalities abound.
Fairclough (1989: 7), too, is critical of mainstream linguistics for being 'an
asocial way of studying language, which has nothing to say about relationships
between language and power and ideology'. Like the others, he is also critical of
other areas of language study such as sociolinguistics, pragmatics, conversation
analysis, and discourse analysis. Sociolinguistics he criticizes for remaining so
firmly located within the tradition of positivism: it tends to concentrate on the
'objective description' of social 'facts' and to correlate supposed social 'classes'
(which are often loosely defined social strata) with linguistic features. Prag-
matics he criticizes for its individualism: its concentration on actions, intentions,
and strategies of individual speakers who are assumed to be engaged in co-
operative interactions that ignore questions of social power. Similarly, he takes
conversation and discourse analysis to task for constructing an image of
conversation as 'a skilled social practice existing in a social vacuum, as if talk
were generally engaged in just for its own sake' (ibid.: 12).
A slightly different approach to CDA can be found in the work of the neo-
Hallidayan group of social semioticians, particularly Gunther Kress (1985;
Hodge and Kress 1988). Halliday's functional grammar is particularly useful for
pursuing some of these questions since, while in the mainstream tradition
described in the last section a principal focus is in matching form and function,
they are already combined in Halliday's model, allowing for analysis to move on
to broader concerns. From this point, Kress specifically relates his view of
discourse to Foucault's:
Discourses are systematically-organized sets of statements which give expression to the
meanings and values of an institution. Beyond that, they define, describe, and delimit
what it is possible to say and not possible to say (and by extension—what to do or not to
do) with respect to the area of concern of that institution, whether marginally or
centrally. A discourse provides a set of possible statements about a given area, and
organizes and gives structure to the manner in which a particular topic, object, process
is to be talked about. (Kress 1985: 7)
What is of immediate interest here, I think, is that when Kress suggests that
discourses 'define, describe, and delimit' what can and cannot be said, his view is
clearly fairly close to the one I suggested at the beginning of this paper, that the
issue in discourse analysis is how certain things come to be said at certain times
and in certain places.
Having pointed to some of the different approaches and shared assumptions
of CDA, I shall now try to show more clearly how their underlying visions of
language, discourse, ideology, and society operate. To return to the two
concepts that I introduced at the end of the last section, Fairclough's model of
discourse analysis operates, first, with a dialectical relationship between the
micro-structures of discourse (linguistic features) and the macro-structures of

society (social structures and ideology). He stresses that while macro-structures
of society may determine the micro-structures of discourse, these in turn
reproduce the larger social and ideological structures. In contrast to the view of
discourse discussed in the previous section, then, while the micro level
(linguistic features) remains much the same, the macro is expanded from a
notion of context and participants to a much broader conception of society.
Second, between these micro and macro levels, there are a number of different
levels, so that this view of discourse anaysis can be seen as a four-level model.
On the smallest level are text and discourse (these could be seen as separate
levels but I have kept them as one here): text refers to the product (spoken or
written) of the social process of discourse. Next, actual discourse is determined
by 'orders of discourse', which are sets of conventions associated with social
institutions. These in turn are determined by ideologies, which, in turn, are
determined by relations of power within the broader society. In this view,
coherence (to take one of the most common issues in applied linguistic
discourse analysis) is dependent on 'discoursal common sense' which in turn is
ideological (Fairclough 1989:107).
Kress's (1985) model is basically also a four-level one. In my discussion of
Fairclough's distinction between text and discourse, I counted these as
operating on the same level, as product and process. Although Fairclough also
draws on Halliday for this distinction, his conception of discourse as language as
social practice seems narrower than Kress's understanding. For Kress, by
contrast, the meanings of texts (the first level) derive from two particular factors,
discourses and genres (the second level). Thus, the forms and meanings of
texts—the actual use of language—'are determined by discourses—systems of
meanings arising out of the organization of social institutions—and by
genres—formal conventional categories whose meanings and forms arise out of
the meanings, forms, and functions of the conventionalized occasions of social
interactions' (Kress 1985: 31). While Kress is content to focus principally on
these two levels, at which the forms and meanings of texts are determined by
discourse and genre, he also suggests a third level (ideology): 'While discourse
and genre provide the systematically-organized linguistic categories which
make up a text, ideology determines the configuration of discourses that are
present together and their articulation in specific genres' (ibid.: 83). Discourses
for Kress, therefore, are systems of meanings embedded in certain institutions,
which in turn are determined by ideologies 'in response to larger social
structures' (ibid.). Thus, the micro level of this analysis is the text (determined by
discourse and genre, in turn determined by ideology) and the macro level is the
larger social structure. This larger social structure remains somewhat vague in
Kress (1985), though in Hodge and Kress (1988) the relationship between
ideology and society is given a more formally Marxist flavour: 'In order to
sustain these structures of domination the dominant groups attempt to
represent the world in forms that reflect their own interests, the interests of their
power' (Hodge and Kress 1988: 3).
ALASTAIR PENN YCOOK 125
While these different strands of CDA are extremely valuable in the
frameworks of analysis they provide for how the relationships between language
use and the social order operate, I nevertheless have a number of concerns
about some of their underlying assumptions. One of the key questions to be
asked here is, if we indeed understand social relations to be marked by deep
inequalities, how are such inequalities to be explained? For Fairclough, and
most of those working in CDA, this is done from a clearly neo-Marxist
perspective in which power is, in the final analysis, located in the relationship
between social classes and economic production, which is taken to be primary
(base or infrastructural), material, and causative of all other relations. Simply
put, this material reductionism leads to the position that 'the so-called base can
be separated from the so-called superstructure (which it can't); that the base is
"material" (which it isn't); and that it determines the rest (which it doesn't)'
(Worsley 1982: 113). While we should never lose sight of material and socio-
economic inequalities, there is a danger that by making economic class relations
primary, we lose sight of other sites of inequality (at the very least, race and
gender; see, for example, Black and Coward 1990 and Outlaw 1990),3
construct an over-simplified version of society whereby a 'dominant group' has
power while the 'oppressed' do not, and become too deterministic in ascribing
causality to socio-economic relations.
A concomitant result of this view is that it tends to posit a 'real' world that is
obfuscated by ideology. The estimable, though I believe problematic task, of the
critical linguist, then, is to help remove this veil of obscurity and help people to
see the 'truth'. This view emerges, for example, in Fairclough's (1989: 75)
distinction between 'inculcation' and 'communication', the former being 'the
mechanism of power-holders who wish to preserve their power' and the latter
'the mechanism of emancipation and the struggle against domination'. There are
echoes here of Habermas (1984) and his view that there can be a form of
'communicative action' that is devoid of ideology. Thus, one of the problems
that emerges from this approach to critical applied linguistics is that while all
language is seen as ideological, there is nevertheless a 'real world' beyond such
'misrepresentation'. In a recent book on the news, for example, Fowler (1991:
10) argues that 'because the institutions of news reporting and presentation are
socially, economically and politically situated, all news is always reported from
some particular angle'. Thus far, I would agree and indeed, like Fowler, would
extend this view not only to such domains as the news but also to all 'representa-
tional discourse': 'anything that is said or written about the world is articulated
from a particular ideological position' (ibid.). But what are we to make of the
following statement that 'The world of the press is not the real world, but a world
skewed and judged' (ibid.: 11)? This clearly suggests that the ideological
positions misrepresent a real world. This, I believe, is a problematic assumption.
It raises some awkward questions: How do we arrive at that real world without
the mediation of language? How can we deal with the real world without
ideological distortion? How is it possible to arrive at some Archimedean point
outside language and ideology? What this statement suggests is two problematic
divides: on the one hand, there is the real world as opposed to the unreal world
(the real world misrepresented through ideology); on the other hand, there is
ideological representation as opposed, presumably, to non-ideological repre-
sentation.
I also feel that most of this critical discourse analysis tends to operate with a
problematically static view of both language and society. What happens, by and
large, is that texts (the micro level) are read in order to reveal the workings of
social structures (the macro level). There are two dangers in this view: first, it is
subject to a representational fallacy, whereby a 'real world' of social relations is
represented in language; and second, in models whereby the micro is deter-
mined by the macro, which is in turn reproduced by the micro, there is little
space for an understanding of human agency, interpretation, or change: socio-
economic relations determine power, power determines ideology, ideology
determines orders of discourse, and orders of discourse determine discourse.
While I have been suggesting that an essential aspect of a critical approach to
discourse analysis lies in the understanding that our ability to act in the world is
constrained, it is nevertheless crucial here to allow for human agency rather
than constructing a model in which all is determined by socio-economic (or
other) relations. If in the last section I criticized mainstream applied linguistics
for its assumption of a completely free-willed subject, the criticism might be
made against critical discourse analysis that the pendulum has swung too far in
the other direction: we now have a subject determined by ideologies that can
simply be deduced from texts as they are read by critical analysts.
DISCOURSE AS POWER/KNOWLEDGE
So far, then, I have discussed both the need for and some of the pitfalls of
engaging in critical forms of discourse analysis. Clearly, however, although the
perspectives discussed in the last section relate discourse to broader social
structures, they do not suggest the type of reversal of language and discourse
that I raised in the opening of this paper. Such a reversal comes only if we take up
a more Foucauldian position on discourse. So why Foucault? To put it simply,
Foucault allows for critical analysis while avoiding the reductions and totaliza-
tions of more Marxist-based analysis. Indeed, Harland (1987: 166) suggests
that Foucault is 'avowedly anti-Marxist' in his rejection of 'the reduction of all
power-relations to class-relations' and of a Utopian social harmony towards
which history inevitably leads. As Henriques, Hollway, Irwin, Venn, and
Walkerdine (1984: 92) suggest, the vital contribution played by approaches
such as Foucault's is that they 'help deconstruct the monolithic, unitary
character of power and the social domain which has characterized Marxist
functionalist and structuralist social theory alike'.
A number of the critical discourse analysts discussed above explicitly draw on
Foucault's work; Kress states that his view of discourse is based on Foucault's,
and Fairclough's notion of orders of discourse is also supposed to be
Foucauldian. Although I do not want to appear to be arguing for some sort of
Foucauldian purity as if alternative readings of Foucault are not possible, I want
to suggest that there are, at the very least, some differences between their
readings of Foucault and mine.4 These differences lie in the tendency, first, to
see discourse still as a linguistic phenomenon, albeit socially embedded; second
to separate discourse from ideology and suggest that the latter determines the
former; third, to operate with a view of power only as something held by one
group and not by others; and finally to view discourse as only concerned with
the delimitation and regulation of what can be said, rather than also with the
production of what can be said. McHoul and Luke (1989) suggest that there are
two broad traditions here, one the Anglo-American, the other the European. In
the first, discourse analysis remains located in the realm of empirical linguistics
while drawing on various other social-scientific traditions. The second
embraces 'a socio-historical-political view of discourse' (McHoul and Luke
1989: 324). Thus, when in the first tradition discourse analysis does take on a
political element, there is a tendency to work with a dualistic framework
between linguistic analysis of discourse and an added-on political framework,
which interact dialectically with each other (see previous section). Anglo-
American approaches 'import political ideas, as it were, for grafting onto an
empirical-linguistic scaffold' (ibid.). A continental European (predominantly
French) approach, by contrast, tends to theorize discourses from the very
beginning as 'socio-historically specific systems of knowledge and thought'
(ibid.). A similar distinction is made more generally by Harland (1987) when he
suggests that Anglo-American pragmatism and materialism have always
opposed the inversion of superstructure and infrastructure (making the former,
and therefore language, culture, and discourse primary) by European struc-
turalism and post-structuralism.
A key part of Foucault's work (1970, 1972, 1979, 1980a, 1980b) is his
analysis of how various knowledges and disciplines—medical, psychological,
psychoanalytical, penitential, sexual, and so on—normalize social institutions
and practices in society. Central to this attempt to write histories of how human
beings are made subjects (the subject, 'individual', or 'man', are produced, not
pre-given, categories) is the notion of discourse, for 'it is in discourse', he
suggests, 'that power and knowledge are joined together' (Foucault 1980b:
100). Discourse, Weedon (1987: 41) explains, is a 'structuring principle of
society, in social institutions, modes of thought and individual subjectivity';
discourses are *ways of constituting knowledge' (ibid.: 108); they 'are more than
ways of thinking and producing meaning. They constitute the "nature" of the
body, unconscious and conscious mind and emotional life of the subjects which
they seek to govern' (ibid.).
Importantly, then, to the extent that discourses are organizations of know-
ledge, and are always linked to power, embedded in social institutions, and
produce ways of understanding, they are akin to the concept of ideology.
Foucault (1980a) explicitly rejects the notion of ideology, however, in favour of
discourse, since ideology is predominantly used in contrast to something that is
considered to be 'real' or 'the truth', and thus it is assumed that ideology
necessarily obfuscates, hides the truth and leads to 'false consciousness' (see
previous section). His interest, by contrast, is in the effects of claims to truth, in
'seeing historically how effects of truth are produced within discourses which in
themselves are neither true nor false' (Foucault 1980a: 118). This is a key point,
since it does not allow for some Archimedean point outside ideology or
discourse from which truth or falsity can be judged. A second criticism raised by
Foucault is that although ideology may help us to understand how individuals
are less than autonomous, free-willed subjects, it tends to imply a unitary rather
than a multiple subject. A final criticism is that even when ideology is taken as
materialist and thus more infrastructural than superstructural (as in the work of
Althusser and Pecheux),5 it always seems to be secondary and reducible to class
relationships and the means of production. Foucault's approach, by contrast,
avoids an ontological or teleological search for an ultimate determinant such as
class or relations of production and instead looks to a multiplicity of social,
cultural, political, economic, technical, or theoretical conditions of possibility
for the emergence of discourses. Thus, the constraints on human freedom of
thought are no longer reducible to the nature of 'man', to the sexual drives of the
subconscious, or to the relationship to the means of production, but rather are a
product of a multiplicity of relationships. As Foucault once put it, human
activity is not so much defined by labour as by 'pleasure, restlessness, merry-
making, rest, needs, accidents, desires, violent acts, robberies, etc' (quoted in
Harland 1987:166).
Discourse in this sense, therefore, does not refer to language or uses of
language, but to ways of organizing meaning that are often, though not
exclusively, realized through language. Discourses are about the creation and
limitation of possibilities, they are systems of power/knowledge (pouvoir/
savoir) within which we take up subject positions. To think in terms of discourse
in a Foucauldian sense is useful, I believe, because it allows us to understand
how meaning is produced not at the will of a unitary humanist subject, not as a
quality of a linguistic system, and not as determined by socio-economic
relations, but rather through a range of power/knowledge systems that organize
texts, create the conditions of possibility for different language acts, and are
embedded in social institutions. Thus we can speak of the discourses of
democracy, law, capitalism, socialism, education, linguistics, applied linguistics,
and so on. These discourses in effect 'map out' what can be said and thought
about what they define as their respective domains.6 Before discussing, in the
final section, various implications of taking up different discourses on discourse,
a brief illustration of some general applications of a Foucauldian notion of
discourse may help to clarify this concept.
Perhaps one of the best known is Edward Said's (1978) demonstration of
how 'The Orient' is a construction of the discourse of Orientalism, a discourse
which, in the writing of travellers, novelists, colonial authorities, academics, and
so on, is constructed around a series of we/they contrasts (East/West, Orient/
Occident, etc.) that produce an essentialized and homogenized Other ('The
Arabs', 'Muslims', 'The Oriental Mind', 'The Japanese', etc.). 'Without examin-
ing Orientalism as a discourse', Said suggests, 'one cannot possibly understand
the enormously systematic discipline by which European culture was able to

manage—and even produce—the Orient politically, sociologically, militarily,
ideologically, scientifically, and imaginatively during the post-Enlightenment
period.' (Said 1978: 3). Said's analysis shows how, while Orientalism does not
determine what is thought and said about the 'Orient', this discourse neverthe-
less defines, delimits, and produces it: 'it is the whole network of interests
inevitably brought to bear on (and therefore always involved in) any occasion
when that particular entity "the Orient" is in question' (ibid.).7
Foucault has also been taken up in various educational contexts (see
Henriques et al. 1984; Ball 1990b). Particularly interesting here is Valerie
Walkerdine's (1984) discussion of how the 'developing child' of child
psychology and the 'active learner' of learner-centred pedagogies are the
products of particular psycho-educational discourses rather than pre-existing
entities about which child psychologists and educationalists 'make discoveries'.
'By examining the power-knowledge relations which are made possible by the
regimes of truth of development psychology', Walkerdine (1984:163) suggests,
'it will be possible to demonstrate the conditions which have produced the
classification and monitoring of development as a science and as scientifically
validated pedagogy'. Given her observation that current pedagogical practices
'are totally saturated with the notion of a normalized sequence of child
development, so that these practices help produce children as the objects of
their gaze' (ibid.: 155), and in light of the belief in natural sequences of
development and the dominance of 'learner-centred' pedagogies within current
ESL orthodoxies, this view of discourse would suggest that the active, motivated
learner in a communicative ESL class is not so much a newly liberated category
of learner that has emerged through better teaching practices as he or she is a
product of some currently dominant psycho-educational discourses.
A Foucaldian notion of discourse has also been of use (though not
uncontentiously) in work on gender. Sawicki (1988: 176), arguing for the
importance of Foucault in feminist work, suggests that 'Foucauldian discourse is
radical not because it gets at the roots of domination, but inasmuch as it
introduces radically new questions and problems concerning prevailing ways of
understanding ourselves which continue to dominate our thinking about radical
social transformation'. Taking issue with Dale Spender's (1980) analysis of
'man-made language', for example, Black and Coward (1990) argue that we
need to understand words such as 'man' and 'he' not in some representationalist
fashion (language reflects reality) whereby they reference males, but rather as
located within a discourse in which men are represented as humans: 'there is a
discourse available to men which allows them to represent themselves as
people, humanity, mankind' (Black and Coward 1990: 132). Thus it is the
absence rather than the omnipresence of men as gendered subjects in language
that is represented by the use of 'he'. 'Our aim,' according to Black and Coward,
'is not just to validate the new meanings of women but to confront men with their
maleness. This is not just about masculine behaviour, but about discursive
practices' (ibid.).
This brief summary of Foucauldian approaches to discourse has highlighted a

number of differences from the notions of discourse discussed in previous
sections. Clearly, it shares a number of similarities with critical discourse
analysis in its concern with questions of power and social structures. On the
other hand, the actual notion of discourse is different both from the applied
linguistic and the critical version. To return to the concept of levels once again, I
think a Foucauldian analysis could best be described as a three-level model in
which the text (in a broad sense that may mean a linguistic text but could also
mean a body or a beach) is given meaning by discourses (which take up that
middle space of discourse and ideology in the critical approach), which in turn
derive from a multiplicity of non-discursive practices (which, perhaps most
importantly, are by no means only questions of economic or class relations).
Finally, whereas both mainstream and critical discourse analysis tend to locate
meaning in the relationship between linguistic form (discourse/text) and
function, context, or social structure/ideology, this view of discourse locates
meaning in discourse itself.
IMPLICATIONS AND APPLICATIONS

A Foucauldian notion of discourse has significant implications for applied
linguistics and language teaching. As a number of people have commented (e.g.
Hoskin 1990), a key element of Foucauldian analysis is rendering the familiar
unfamiliar. As I commented earlier, at a time when both discourse analysis and
applied linguistics are being consolidated into disciplines, this may be a crucial
juncture for such a rendering, for an attempt to question the 'givens' of applied
linguistics. While there are many domains of applied linguistics that could
usefully be subjected to such analysis—the discourses of communicative
language teaching, the native speaker, and learner variables suggest them-
selves—my own work has focused particularly on the discourse of English as an
International Language (EEL).81 have been trying to show how this discourse
grew out of colonial views on English education, was constructed in a particular
way because of its connections to the structuralist and positivist orientations of
linguistics and applied linguistics, and has become intertwined with discourses
on global development and international capitalism. The discourse of EEL in
effect maps out the possible ways of thinking, speaking (writing, etc.), and acting
with respect to the global spread of English, producing a view of this spread as
generally natural, neutral, and beneficial, and focusing attention on questions of
linguistic variety in English rather than cultural and political implications of its
spread. Thus, it both limits the possibilities for language teachers to think about
our work differently and produces texts and subject positions in accordance
with this view. Rather than following Phillipson's (1992) interesting, but I think
over-deterministic, structuralist-Marxist analysis of 'linguistic imperialism', the
notion of discourse here has allowed me to explore both historically and in the
present how the idea of 'English as an International Language' has been
produced.
This view of discourse also has considerable implications for language
ALASTA1R PENN YCOOK 131
teaching. It is easier to compare a standard applied linguistic and a CDA

approach, since both consider discourse to be essentially linguistic. The
difference lies in what they do with their texts/discourses. Thus we could
compare, say, how McCarthy (1991) and Fairclough (1989) deal with the
questions of coherence. Analysing a text about a boy who had been attacked by
a python, McCarthy (1991: 27) argues that, in order to be coherent, the text
'requires us to activate our knowledge of pythons as dangerous creatures which
may threaten human life ...'. Meanwhile Fairclough (1989: 103) analyses how
an advertisement for an English course is made coherent by appeal to 'common
sense' knowledge—that the reader 'is looking for success, the capacity to
dominate and influence others', and so on—which in turn is ideological: 'the
coherence of discourse is dependent on discoursal common sense' which 'is
ideological to the extent that it contributes to sustaining unequal power
relations, directly or indirectly' (ibid.: 107). McCarthy looks to background
knowledge to explain coherence, while Fairclough looks to the ideological
construction of common sense. Alternatively, we could take a text from the
exercises supplied by Cook (1989), and show that while his focus is on how the
sentences in a passage are put together (Task 90, ibid.: 149), a critical analyst
would probably focus on how this passage appears to be arguing for a biological
basis for human warfare.
A Foucauldian analysis presents a different possibility. It is not concerned
with how discourses (texts) reflect social reality, but how discourses produce
social realities; it does not look for relationships between discourse and society/
politics, but rather theorizes discourse as always/already political; it does not
seek out an ultimate cause or basis for power and inequality, but rather focuses
on the multiplicity of sites through which power operates; and it does not posit a
reality outside discourse, but rather looks to the discursive production of truth.
Rather than the version of conscientization offered by a Freirean pedagogy (see
Weiler 1991) or CDA, a pedagogy based on a Foucauldian notion of discourse
does not allow for easy talk of 'oppressed' and 'oppressors', or of 'truth', or
'reality'. In her discussion of the difference between a Freirean-style pedagogy
of 'talking back' and a pedagogy of 'talking story', Schenke (1991: 53) suggests
that it is important to be 'more wary of resolution, unconvinced by the binary
language of oppression and liberation, more responsive to the capillary effects
of power, and more attuned to smaller, less "heroic" acts of resistance and
personal/social change'. This is not about trying to get students to understand
how ideologies have obscured the 'truth', but how discourses construct our lives.
According to Kearney (1988: 386), 'it is certainly unlikely that any amount of
"knowledge" about the falsehood of our experience is going to help us think or
act in a more effective or liberating way. A form of pedagogy, however accurate
and scientific, which does no more than explain the intricate mechanisms of our
enslavement offers little consolation'.
Recalling the reversal of priority of language and discourse with which I
opened this paper, it becomes clear that the approach I am suggesting here is
based on an understanding of how discourses map out our different worlds.
Language teaching becomes a process of making the familiar unfamiliar, of

linking the process of learning a second language to a pedagogy that seeks to
question how we come to understand ourselves as we do. One immediate
implication of this is in terms of language teaching curricula. The search for
'content' in language teaching has always been a contentious one, whether in
terms of formal study of language as content, or the functionalist or pragmatist
orientations of communicative language learning or language learning for
specific purposes. More critical work has sought to deal with 'social issues'
(crime, abortion, etc.), or, in the case of CDA, with the ideological under-
pinnings of texts, but these approaches, by dealing with 'serious issues' rather
than dealing with issues seriously, and by starting with a dichotomy between
language and social structure, have often failed to link the focus of these classes
either to the language being learnt or to the lives of the students.
The understanding of language and discourse that I am suggesting here can, I
believe, alleviate some of these difficulties by exploring the specific relation-
ships between particular discourses and the particular language being taught.
Thus, in the context in which I teach, it is in the connections between English
and particular discourses (what I have elsewhere (Pennycook, in press) called
the 'worldliness' of English), discourses that are of importance to the particular
context of language use, that a curriculum of relevance, language learning
potential, and social change can be forged. Thus, I am exploring with my
students how English in Hong Kong intersects with discourses of popular cul-
ture, national culture, capitalism, colonialism, and education. Another way for-
ward is through what Arleen Schenke (1991: 47) has called a 'genealogical
practice in memory work', a practice that can start to attend to the discursive
formation of student (and teacher) subjectivity and memory. Arguing that there
is an important difference between student autobiography per se (student
experiences, stories, and histories) and the 'autobiographical "I"', the voice
through which such stories are narrated and heard, she suggests that it is by
attending to the latter 'that we touch upon the discursive formations of
subjectivity and memory, and that we can work towards a more historicized and
engaged practice of feminist/ESL teaching' (ibid.: 48).
Finally, we need to be very aware both of the discourses into which we are
asking our students to move and of the discourses in which we as teachers are
engaged. The point here is that, as Foucault argued with respect to the mode of
confession from the Catholic confessional to the psychiatrist's couch (Foucault
1980b; Tambling 1990), to be summoned to speak is not to give expression to
some true inner self, but to be called into a certain discourse, a regime of truth:
'To speak is to assume a subject position within discourse and to become
subjected to the power and regulation of the discourse' (Weedon 1987: 119).
Thus, as we urge our students to speak or write, and as we listen or read, we need
also to consider what discourses are constructing those moments of speaking
and understanding. To quote Schenke (1991) once again: 'Because auto-
biographical work in teaching is a practice in "breaking the silence" of personal
and social histories, and because these histories, in ESL teaching in particular,
ALASTA1R PENNYCOOK 133
are traversed by legacies of colonialism, it matters fundamentally who speaks

and who listens, under what conditions of possibility, and along the lines of
which political and pedagogical agendas' (Schenke 1991:48).
This notion of discourse, then, offers a number of possibilities for engaging
critically with language and meaning without falling into some of the materialist
or deterministic pitfalls that CDA can present. Like CDA, it locates the context
of language use, the speakers and their intentions in a wider social, cultural, and
political context than the view common to discourse analysis in applied
linguistics. This, I believe, is a crucial step in opposing the discourse of
pragmatism in applied linguistics and in acknowledging the political in second
language education. Unlike much of CDA, however, a Foucauldian conception
of discourse does not posit a reality to be unmasked, a truth that is represented
or misrepresented in texts, or an ultimate location of power and inequality in
socio-economic relations. Rather, it permits us to see how meaning is produced
by discursive regimes that are related to a 'will to know' in diverse areas of social
life. Such a view, I think, presents a useful way of pursuing questions of how we
come to speak and to mean as we do.
CONCLUSION
I hope that this critical view of discourse analysis has not implied that either
standard applied linguistic discourse analysis or CDA are not worthwhile
activities. Both approaches are useful for applied linguistics: the first can help
focus on discourse types, the relationship between discourse and background
knowledge, awareness of interaction in the classroom, how texts are con-
structed, how turn-taking occurs in different cultures, whether teaching
materials have been designed with an awareness of discourse, and so on; the
second can help us develop an awareness of how language use is always
connected to issues of power, how background knowledge is never an innocent
way of knowing, how larger conditions of social reality are never absent from
texts. And yet, as I have argued, the decontextualized contexts and the political
quietism of applied linguistics, and the often reductionist and deterministic
frameworks of CDA, suggest that these approaches should be used with
caution.
One question that I raised in the introduction has been left hanging: are the
applied linguistic (and possibly CDA) approaches incommensurable with the
Foucauldian position I have been outlining? Certainly one response would be
that they are not: could we not, for example arrange these positions along a
continuum? At one end, a very linguistically based view of discourse analysis,
where discourse merely refers to the way sentences are connected together,
moving gradually through positions increasing the importance of context, until
we arrive at a position whereby language is subsumed within discourse. There
are two reasons why I feel such a notion is untenable: first, on epistemological
grounds, it seems that different versions of discourse imply very different
understandings of the world. CDA is clearly based, to differing degrees, on a
neo-Marxist critical tradition that takes social inequalities to be its prime focus
134 INCOMMENS URABLE DISCOURSES?
and sees ideologies as a fundamental way in which those inequalities are

perpetuated. Such a view of ideology is not easily reconciled, for example, with a
more liberal conception whereby ideology is taken to be either an apolitical
system of thought or a series of beliefs held by people with a shared political
orientation. And a Foucauldian notion of discourse, implying as it does a
rejection of many of the staples of critical and non-critical thought (oppressed
and oppressors, truth, reality, and so on), seems even further removed. Second,
on moral grounds, I find it hard to accept the implicit relativity of the notion of
continua, whereby different views are simply seen as different and are not
engaged with. So are we then inevitably caught in a trap whereby even when we
appear to be engaged in the same discourse (from one point of view), we are
nevertheless constantly engaged in different discourses (from another point of
view)?
And yet, neither do I wish to end with the defeatist and equally relativist view
of incommensurability. If there are good reasons to abandon, or at least
question, the Enlightenment belief in a rational subject in control of language
and meaning, then rather than looking for alternatives in other modernist
frameworks, and thus continuing a search for teleological explanations of how
we come to think as we do (located, for example, in a humanist belief in a 'true',
'real', or 'natural' self, in the Freudian subconscious, or Marxian socio-
economic relations), I believe we need to take up a more postmodern stance and
thus to abandon such teleologies. In spite of Lyotard's (1984: xxv) assertion that
postmodern knowledge 'refines our sensitivity to differences and reinforces our
ability to tolerate the incommensurable', however, I have been trying to point to
the need, in Hoy's (1988: 22) terms, of thinking 'the great unthought', of
acknowledging that, while we cannot know ourselves or the world around us in
any objective fashion, we nevertheless need to ask how it is we come to think as
we do. My hope is that,first,even if these views cannot be reconciled or merged,
they can at least be mutually understood. By comparing and discussing these
different understandings of discourse, my intention at the very least has been to
make the familiar terrain of applied linguistic discourse analysis slightly less
familiar and to render the unfamiliar terrain of Foucauldian discourse analysis
slightly more familiar. And, second, I hope that others will take up some of the
challenges that I have posed here so that the chasms that sometimes exist
between us can be more easily traversed.
(Revised version received September 1993)
ACKNOWLEDGEMENTS
I should like to thank Roger Simon, whose careful reading of an earlier version of this
paper has helped me clarify some of my ideas. Two Applied Linguistics reviewers also
deserve my thanks for careful and constructive criticism of my work. Comments by
colleagues from the English Centre at Hong Kong University, where I also presented an
earlier version of these ideas, have also been helpful.
NOTES
1
There is an inevitable degree of over-generalization in my view of mainstream
applied linguistics here (a problem that is also true of my description of critical discourse
analysis). It is important to point out that what I am referring to as applied linguistics here
is, in the narrower sense of the term, concerned principally with language teaching and
thus with understandings of language, language use, and language learning, rather than
the broader sense that refers to a wider range of applications of language theory, including
stylistics, translation, neurolinguistics, clinical linguistics, and so forth (see Crystal 1987).
I am only concerned, therefore, with how this narrower sense of applied linguistics has
borrowed forms of discourse analysis from the much broader field of discourse analysis
itself (see, for example, van Dijk's (1985) four-volume series).
2
This is not only the case for language use, but also for domains such as language
acquisition (see Urwin 1984) and communicative competence (see Peirce 1989).
3
I do not wish to suggest here that work in CDA has not addressed questions of race or
gender. My argument is that the neo-Manrist base of much of this work operates with a
problematic assumption about causes and roots of inequality.
4
Another issue that I have not dealt with here is the shift in Foucault's thinking from
his earlier archaeological period to his later genealogical period. Although there is debate
over how much should be made of this distinction (see, for example: Dreyfus and
Rabinow 1982; Hoy 1988; Hoskin 1990), part of the difference in these readings can be
attributed to some critical discourse analysts' favouring an archaeological approach to
discourse and my tendency towards a more genealogical approach.
5
See, for example, Althusser (1971) and Pecheux's (1982) important work that
relates Althusser's concept of ideology explicitly to a notion of discourse. For a discussion
of this work, see Macdonell (1986).
6
I owe this image of 'mapping out' reality to Roger Simon.
7
I am, of course, aware that there has been a great deal of discussion about Said's
notion of Orientalism. This does not, however, seem to be the place to engage in an
extended exploration of the problems with parts of his analysis. For an interesting
discussion, see Clifford (1988).
8
I have explored these ideas in much greater depth elsewhere (Pennycook, in press).
REFERENCES
Althusser, L. 1971. Lenin and Philosophy and Other Essays. London: New Left Books.
Ball, S. J. 1990a. 'Introducing Monsieur Foucault' in S. J. Ball (ed.) 1990b: Foucault and
Education: Disciplines and Knowledge. London: Routledge.
Ball, S. J. (ed.) 1990b: Foucault and Education: Disciplines and Knowledge. London:
Routledge.
Black, M. and R. Coward. 1990. 'Linguistic, social and sexual relations: A review of Dale
Spender's Man Made Language' in D. Cameron (ed.) 1990: The Feminist Critique of
Language: A Reader. London: Routledge.
Brown, H. D. 1980. Principles of Language Learning and Teaching. Englewood Cliffs,
NJ: Prentice Hall.
Brown, G. and G. Yule. 1983. Discourse Analysis. Cambridge: Cambridge University
Press.
Canale, M. 1983. 'From communicative competence to communicative performance' in
J. Richards and R. Schmidt (eds.) 1983: Language and Communication. New York:
Longman.
Canale, M. and M. Swain. 1980. 'Theoretical bases of communicative approaches to

second language teaching and testing.' Applied Linguistics 1/1:1-47.
Chua Beng-Huat. 1983. 'Re-opening ideological discussion in Singapore: A new
theoretical direction.' Southeast Asian Journal of Social Science 11/2: 31-45.
Clifford, J. 1988. The Predicament of Culture: Twentieth Century Ethnography,
Literature and Art. Cambridge, MA: Harvard University Press.
Cook, G. 1989. Discourse. Oxford: Oxford University Press.
Coulthard, M. 1977. An Introduction to Discourse Analysis. London: Longman.
Crystal, D. 1987. The Cambridge Encyclopedia of Language. Cambridge: Cambridge
University Press.
Dreyfus, H. L. and P. Rabinow. 1982. Michel Foucault: Beyond Structuralism and
Hermeneutics. Chicago: University of Chicago Press.
Fairclough, N. 1985. 'Critical and descriptive goals in discourse analysis.' Journal of
Pragmatics 9: 739-63.
Fairclough, N. 1989. Language and Power. London: Longman.
Foucault, M. 1970. The Order of Things. New York: Random House.
Foucault, M. 1972. The Archaeology of Knowledge. London: Tavistock.
Foucault, M. 1979. Discipline and Punish: The Birth of the Prison. New York: Vintage
Books.
Foucault, M. 1980a. Power/Knowledge: Selected Interviews and Other Writings, 1972-
1977. (edited by C. Gordon). New York: Pantheon Books.
Foucault, M. 1980b. The History of Sexuality. (Volume 1). New York: Vintage Books.
Fowler, R. 1991. Language in the News: Discourse and Ideology in the Press. London:
Routledge.
Fowler, R., B. Hodge, G. Kress, and T. Trew. 1979. Language and Control. London:
Routledge and Kegan Paul.
Fowler, R. and G. Kress. 1979. 'Critical linguistics' in R. Fowler, B. Hodge, G. Kress, and
T. Trew 1979: Language and Control. London: Routledge and Kegan Paul.
Grabe, W. 1990. 'Foreword' in W. Grabe (ed.) 1990: Annual Review of Applied
Linguistics 11: vii-xi.
Habermas, J. 1984. Theory of Communicative Action. Vol. 1: Reason and the
Rationalization of Society. (Translated by T. McCarthy). Boston: Beacon Press.
Harland, R. 1987. Superstructuralism: The Philosophy of Structuralism and Post-
Structuralism. London: Methuen.
Harris, Z. 1952. 'Discourse analysis.' Language 28:1-30.
Hatch, E. M. 1983. Psycholinguistics: A Second Language Perspective. Rowley, MA:
Newbury House.
Hatch, E. M. 1992. Discourse and Language Education. Cambridge: Cambridge
University Press.
Hatch, E. M. and M. H. Long. 1980. 'Discourse analysis, what's that?' in D. Larsen-
Freeman (ed.) 1980: Discourse Analysis in Second Language Research. Rowley, MA:
Newbury House.
Henriques, J., W. Hollway, C, Irwin, C. Venn, and V. Walkerdine. 1984. Changing the
Subject: Psychology, Social Regulation and Subjectivity. London: Methuen.
Hodge, R. and G. Kress. 1988. Social Semiotics. Oxford: Polity Press.
Hoskin, K. 1990. 'Foucault under examination: The crypto-educationalist unmasked' in
S. J. Ball (ed.) 1990b: Foucault and Education: Disciplines and Knowledge. London:
Routledge.
Hoy, D. C. 1988. 'Foucault: Modern or postmodern?' in J. Arac (ed.) 1988: After
Foucault: Humanistic Knowledge, Postmodern Challenges. New Brunswick: Rutgers
University Press.
Kearney, R. 1988. The Wake of Imagination. Minneapolis: University of Minnesota
Press.
Kress, G. 1985. Linguistic Processes in Sociocultural Practice. Oxford: Oxford
University Press.
Kress, G. 1990. 'Critical discourse analysis' in W. Grabe (ed.) 1990: Annual Review of
Applied Linguistics 11: 84-99.
Long, M. H. 1983. 'Native speaker/non-native speaker conversation and the negotiation
of comprehensible input.' Applied Linguistics All: 126-41.
Luke, A., A. W. McHoul, and J. L. Mey. 1990. 'On the limits of language planning: Class,
state and power' in R. B. Baldauf Jr. and A. Luke (eds.) 1990: Language Planning and
Education in Australasia and the South Pacific. Clevedon: Multilingual Matters.
Lyotard, J-F. 1984. The Postmodern Condition: A Report on Knowledge. (Translated by
G. Bennington and B. Massumi.) Minneapolis: University of Minnesota Press.
Macdonell, D. 1986. Theories ofDiscourse: An Introduction. Oxford: Blackwell.
McCarthy, M. 1991. Discourse Analysis for Language Teachers. Cambridge: Cambridge
University Press.
McHoul, A. and A. Luke. 1989. 'Discourse as language and politics: An introduction to
the philology of political culture in Australia.' Journal of Pragmatics 13: 323-32.
Outlaw, L. 1990. 'Towards a critical theory of "Race"' in D. Goldberg (ed.) 1990:
Anatomy of Racism. Minneapolis: University of Minnesota Press.
Pecheux, M. 1982. Language, Semantics and Ideology: Stating the Obvious. (Translated
by Harbans Nagpal). London: MacMillan.
Peirce, B. N. 1989. 'Toward a pedagogy of possibility in the teaching of English
internationally.' TESOL Quarterly 23/3:401-20.
Pennycook, A. (in press). The Cultural Politics of English as an International Language.
London: Longman.
Phillipson, R. 1992. Linguistic Imperialism. Oxford: Oxford University Press.
Richards, J., J. Platt, and H. Weber. 1985. Longman Dictionary ofApplied Linguistics.
Harlow: Longman.
Said, E. 1978. Orientalism. New York: Random House.
Santos, T. 1992. 'Ideology in composition: LI and ESL.' Journal of Second Language
Writing 1/1:1-15.
Sawicki, J. 1988. 'Feminism and the power of Foucauldian discourse' in J. Arac (ed.)
1988: After Foucault: Humanistic Knowlege, Postmodern Challenges. New Bruns-
wick: Rutgers University Press.
Schenke, A. 1991. 'The "will to reciprocity" and the work of memory: Fictioning
speaking out of silence in E.S.L. and feminist pedagogy.' Resources for Feminist
Research/Documentation sur la Recherche Feministe 20(3/4): 47-55.
Sinclair, J. McH. and R.M. Coulthard. 1975. Towards an Analysis of Discourse.
Oxford: Oxford University Press.
Spender, D. 1980. Man Made Language. London: Routledge and Kegan Paul.
Stalpers, J. 1988. The maturity of discourse analysis.' (Review article) Language in
Society 17:87-97.
Swales, J. M. 1990. Genre Analysis: English in Academic and Research Settings.
Cambridge: Cambridge University Press.
Tambling, J. 1990. Confession: Sexuality, Sin, the Subject. Manchester: Manchester
University Press.
Urwin, C. 1984. 'Power relations and the emergence of language' in J. Henriques, W.

Hollway, C. Irwin, C. Venn, and V. Walkerdine (eds.) 1984: Changing the Subject:
Psychology, Social Regulation and Subjectivity. London: Methuen.
van Dijk, T. (ed.) 1985. Handbook of Discourse Analysis (Four volumes). New York:
Academic Press.
Walkerdine, V. 1984. 'Developmental psychology and the child-centred pedagogy: the
insertion of Piaget into early education' in J. Henriques, W. Hollway, C. Irwin, C. Venn,
and V. Walkerdine 1984: Changing the Subject: Psychology, Social Regulation and
Subjectivity. London: Methuen.
Weedon, C. 1987. Feminist Practice and Poststructuralist Theory. Oxford: Blackwell.
Weiler, K. 1991. 'Freire and a Feminist Pedagogy of Difference.' Harvard Educational
Review 6 1 / 4 : 4 4 9 - 7 4 .
Widdowson, H. 1978. Teaching Language as Communication. Oxford: Oxford
University Press.
Wodak, R. (ed.) 1989. Language, Power and Ideology: Studies in Political Discourse.
Amsterdam: John Benjamins.
Wodak, R. 1990. 'Discourse analysis: Problems, findings, perspectives.' Text 10(1/2):
125-32.
Worsley, P. 1982. Marx and Marxism. London: Tavistock.

Pennycook (1994) Incommensurable Discourses

Uploaded by

Copyright:

Available Formats

Pennycook (1994) Incommensurable Discourses

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pennycook (1994) Incommensurable Discourses

Uploaded by

Copyright:

Available Formats

Incommensurable Discourses ?

This paper is an attempt to come to terms with different understandings of the

INTRODUCTION: WHICH IS BIGGER-LANGUAGE OR DISCOURSE?

DISCOURSE AS SUPRASENTENTIAL LANGUAGE USE

meanings: first, language in use; and second, the relationships between

1. this commonly-held view of discourse is a very particular one, and it is

Discourse analysis, as it grew up in applied linguistics, was a synthesis of a

language forms (lexical, grammatical, intonational/phonological) and the

political contexts of discourse are denied by appeal to an ideology of prag-

contexts, but do so with a critical element: 'By denaturalizing the discursive

micro-structures of discourse (linguistic features) and the macro-structures of

the enormously systematic discipline by which European culture was able to

This brief summary of Foucauldian approaches to discourse has highlighted a

IMPLICATIONS AND APPLICATIONS

teaching. It is easier to compare a standard applied linguistic and a CDA

Language teaching becomes a process of making the familiar unfamiliar, of

are traversed by legacies of colonialism, it matters fundamentally who speaks

and sees ideologies as a fundamental way in which those inequalities are

Canale, M. and M. Swain. 1980. 'Theoretical bases of communicative approaches to

Urwin, C. 1984. 'Power relations and the emergence of language' in J. Henriques, W.

You might also like