Volume 8, No. 2, Art. 1 – May 2007
Pêcheux's Contribution to Discourse Analysis
Niels Helsloot & Tony Hak
Abstract: Michel PÊCHEUX (1938-1983) was one of the main representatives of a critical and productive episode in French discourse analysis, from the late 1960s to the early 1980s. He shared with more famous contemporaries such as Michel FOUCAULT a background in BACHELARDian epistemology and ALTHUSSERian "post-structuralism" and an interest in theories of discourse, but his most important contribution to discourse analysis consisted in the development of tools for conducting empirical studies of discourses. In an attempt to break away from the "spontaneous ideology" of content analysis, PÊCHEUX developed a formal, potentially automatic instrument, which he called Automatic Discourse Analysis. This instrument could generate a structuralist description of a discourse by identifying and describing relations of selection and substitution of syntactic elements in a corpus of texts representing that discourse. When dealing with criticisms of this approach and attempting to overcome its limitations, PÊCHEUX moved away from structuralism and developed a more reflective theory of "interdiscourse" in which he tried to account for the ideological struggle and dynamic inequality between discourses. This article discusses the rationale of the different stages PÊCHEUX went through to develop an empirical instrument of discourse analysis.
Key words: Michel PÊCHEUX, French discourse analysis, automatic discourse analysis, epistemology, ideology, interdiscourse
Table of Contents
3. Discourse Theory and Linguistics
4. Automatic Discourse Analysis
5. Limitations and Adaptations
6. Deconstruction of Linguistic Theories
In the 1970s and the beginning of the 1980s, Michel PÊCHEUX (1938-1983) was a leading figure in French discourse analysis. Initially, his renown was based on his first major work, Analyse automatique du discours (1969a), which was received enthusiastically upon its publication. This work—henceforth ADA69—was only the beginning of a continuous process of revising and reformulating the principles of discourse theory and discourse analysis. Through this work, PÊCHEUX became a reference point for much work on discourse analysis published in the 1970s and 1980s. Gradually, ADA69 and PÊCHEUX's theorising became an integral part of textbooks on discourse analysis.1) The significance of his work as seen in the 1980s is also evident from the commemorative issues of the journals Mots and Langages,2) and from Denise MALDIDIER's informative and balanced anthology L'inquiétude du discourse.3) 
Outside of France, ADA69 found ready reception especially in Italy, Spain, Portugal and several Latin-American countries. As early as in 1972, an extensive summary of ADA69 was published in Italian (CIPOLLI, 1972). At the end of the seventies, a Spanish translation of ADA69 appeared (PÊCHEUX, 1978), which includes a critical elaboration written in 1975 (PÊCHEUX & FUCHS, 1975, referred to as ADA75 below). This addition is also part of a Portuguese edition, presented with introductions and revisions, which appeared in Brazil (GADET & HAK ,1990). Later on, a similar collection consisting of ADA69, ADA75 and a series of introductions was published in English (HAK & HELSLOOT, 1995). 
Until the English publication of ADA69 and ADA75 in 1995, PÊCHEUX's work on discourse analysis was not available to and, therefore, not known to the English speaking world. In some circles, however, PÊCHEUX was known as a discourse theorist or rather as a theorist of ideology. This was due to the impact of the English translation of his Les Vérités de la Palice (1975), which was published in 1982 as Language, Semantics and Ideology. This book was, wrongly we think, read as just another contribution to "post-structuralist" philosophical discourse as presented by such authors as Jacques LACAN, Michel FOUCAULT and Jacques DERRIDA. Although linguistic forms of discourse analysis constitute the empirical counterpart of this whole philosophical tradition, these approaches did not get the attention they deserve. The result was that PÊCHEUX's theoretical work used to be discussed irrespective of his contribution to discourse analysis.4) Thus, his significant contribution to the development of empirical strategies of discourse analysis was seriously underestimated, and is largely neglected even up to now. 
In this article, we highlight the intimate relationship between PÊCHEUX's theoretical work and his work on the development of methods of discourse analysis. Pivotal to the connection between theory and analysis in PÊCHEUX's work was the BACHELARDian notion of the instrument as materialised theory.5) This epistemological background will therefore be discussed before we turn to discourse analysis proper. 
Michel PÊCHEUX studied philosophy from 1959 to 1963 at the École Normale Supérieure, where Louis ALTHUSSER and Georges CANGUILHEM were among his teachers. Like Michel FOUCAULT and Jacques DERRIDA before him, he received training in BACHELARDian epistemology. The critique of current philosophies of science developed by Gaston BACHELARD confronted philosophy with the results of historical studies of the development of sciences (plural). BACHELARD's detailed historical analyses had shown that the constitution of sciences such as physics was the result of theoretical rather than empirical breakthroughs. Because of the importance he attributed to theory in the constitution of sciences, his epistemology was considered anti-empiricist. 
Epistemology formed the background for, among others, the work of ALTHUSSER and FOUCAULT in the 1960s, and provided those authors with their concepts, such as "épistémè" (FOUCAULT) and "epistemological break" (ALTHUSSER). BACHELARD coined the latter concept in reference to the distinction between "common" knowledge and scientific knowledge, i.e. between knowledge based on everyday experience and knowledge based on experimental technique (BACHELARD, 1949, p.102). In his terms, the distinction between these two modes of knowing is "philosophically decisive": "What is at stake is nothing less than the primacy of reflection over perception. [...] What mankind makes by means of a scientific technique does not exist in nature and is not even a natural result of natural phenomena."6) Experimental evidence is constituted or produced:
"[P]henomena must be selected, filtered, purified, shaped by instruments; indeed, it may well be the instruments that produce the phenomenon in the first place. And instruments are nothing but theories materialized. The phenomena they produce bear the stamp of theory throughout" (BACHELARD, 1984, p.13).7) 
As shown by CANGUILHEM, who took over BACHELARD's chair in epistemology at the École Normale Supérieure, the process of conceptualisation and "scientificisation", of breaking away from "error", is not confined to the transition from common sense to science, but goes on within science itself. His notion of a continuous breaking away from "error" became an important element of ALTHUSSERism. ALTHUSSER's point was that, in order to establish itself as the science of history and society, MARXism must liberate itself from "idealism" in a continuous struggle. This is the context in which ALTHUSSER and his students worked on a study of the "conditions of production" of MARX's Capital (cf. ALTHUSSER & BALIBAR, 1970). 
PÊCHEUX's scientific career began within this intellectual climate, which introduced a feel for the historical and practical context of scientific concepts and "discoveries". Among his first publications are a summary of the concepts of BACHELARDian epistemology (PÊCHEUX & BALIBAR, 1969), a lecture on the different effects of the "Galilean break" in the fields of physics and biology (PÊCHEUX, 1969b), and two articles on the theoretical conjuncture in the social sciences (PÊCHEUX, 1969c, 1969d)8). Together they give a clear impression of the drives behind PÊCHEUX's approach to discourse analysis. 
PÊCHEUX and BALIBAR (1969)9) describe an "epistemological break" as a point of no return from which a science begins to exist. They emphasise that
"the concept of break has nothing to do with the voluntaristic idea of a jump from ideology into science, with its inevitable implicit religious connotations. The name of Galileo in 'the Galilean break' is in fact a misnomer, because science is not the product of a single human being: Galileo represents the effect rather than the cause of the epistemological break that carries his name."10) 
PÊCHEUX and BALIBAR outline three effects of the epistemological break that constitutes a science. The first of them is that specific pre-existing ideological and philosophical discourses are explicitly dismissed: the new science breaks away from them. A second effect is that retroactively some philosophies are validated and others invalidated: the break redefines the values within a conflict-laden philosophical field. A third effect is the relative autonomy of the new science: its further development depends primarily on its experimental practices and on theoretical events rather than on influences from outside. This last effect implies that an epistemological break is a unique event in the history of a science. Subsequent theoretical displacements cannot be designated with the term "break" (PÊCHEUX & BALIBAR, 1969, p.12). 
In his contribution to ALTHUSSER's "philosophy course for scientists",11) PÊCHEUX (1969b) analyses how a conflict between two theoretical ideologies develops differently within the fields of physics and biology. In physics the "Galilean break" led to a triumph of mechanistic explanations of magnetism and electricity over the animistic experience of wonder. PÊCHEUX draws special attention to the use of the word "dynamic" as opposed to "static", which inaugurates the new field of electrodynamics. Contrastively, in a "transversal analysis", he shows that in biology the opposition between "dynamic" and "static" became connected with the distinction between physiology and anatomy. Here the introduction of the term "dynamic" did not lead to mechanistic interpretations, but to the development of vitalistic perceptions of "forces", which are still conceived as animate. PÊCHEUX links this divergence between physics and biology to differences between the social practices connected with these sciences: physics is applied mainly to the field of the means of production (e.g. machines), whereas biology, through medicine, is applied to the field of labour (humans). 
In two other publications of the same year, PÊCHEUX does not make explicit use of epistemological concepts, but his approach is clearly BACHELARDian. In an article on the conjuncture of social psychology, he states that this discipline is haunted by (the struggle between) two kinds of ideology—a biopsychological one on the "biological" side of psychology and a set of political, religious and moral ones on the "social" side of psychology (PÊCHEUX, 1969c, p.291). These ideologies rely on an unexplicated, ideological concept of the "subject", which should be transformed theoretically by a critical application of psychoanalysis and Marxism. PÊCHEUX conceives this transformation not as something merely philosophical but rather as a difficult enterprise whose realisation implies profound transformations in research practice. It is a "theoretical intervention from the philosophical terrain into the practical work that is done by researchers in social psychology".12) 
In a contribution to a discussion among communists on how to influence the development of the human sciences, PÊCHEUX (1969d) emphasises the BACHELARDian theme of the conceptual difference between (pre-scientific) "experience" and (scientific) "experimentation": "We must take issue with the idea that there are primary "givens" from which theories could start: it must be stressed that a scientific problem can only exist within a conceptual and instrumental-experimental field".13) In the same article we find a quite different theme as well, namely a critique of the ideological uses of formalisation, such as in CHOMSKYan linguistics:
"We must make a distinction between those cases in which mathematics is applied technically and other cases in which it intervenes at a theoretical level between concepts and experimental devices. (An example of the latter would be generative and transformational grammars.)"14) 
In generative and transformational grammars, formalisation is not merely a technical application of mathematics. Formalisation intervenes at a theoretical level. It represents the ideological conception that language is formal. According to PÊCHEUX, the critique and abandonment of empiricism and formalism is not merely a philosophical task—though Marxist philosophy will play a part—, it is rather a practical matter of political intervention in the institutions of research and education.15) 
3. Discourse Theory and Linguistics
PÊCHEUX's work discussed so far clearly demonstrates his immersion in epistemology and his application of ALTHUSSERian notions to an analysis of the state-of-the-art of the social sciences, but it does not throw much light on how this BACHELARDian background relates to his interest in discourse analysis. At this point, some indications can be found in two articles that PÊCHEUX published under the pen name of Thomas HERBERT. In his very first publication (HERBERT, 1966), PÊCHEUX applies BACHELARDian and ALTHUSSERian concepts to the social sciences, in particular to social psychology. His diagnosis is that these sciences have not established themselves as proper sciences, because they have not established their own theoretical object. In other words, they have not accomplished the necessary epistemological break with ideology. These "sciences", therefore, do not produce scientific knowledge but reproduce the ideology of the social system. In line with the epistemology of BACHELARD, CANGUILHEM and ALTHUSSER, PÊCHEUX claims that, in order to become proper sciences, the social sciences must go through a theoretical transformation in which both their objects and their instruments should be redefined. Whereas the current practice of the "social sciences" consists of transforming (ideological) discourse into other (ideological) discourse, social science proper would transform that discourse into something else, which would be expressed in terms of a new theory. PÊCHEUX presents discourse analysis as a social scientific instrument that should be defined in terms of the social scientific theory that is yet to be established. For PÊCHEUX, the development of an instrument of discourse analysis thus was explicitly not a merely technical enterprise. On the contrary, it was part of the project of establishing a truly scientific social psychology. 
In his sequel to the HERBERT (1966) article (HERBERT, 1968), PÊCHEUX gives the following succinct summary of its main thesis: "any science is initially the science of the ideology with which it breaks".16) In this second HERBERT article, PÊCHEUX outlines the preliminaries of a "general theory of ideologies", which would make possible a scientific study of how ideologies function. Such a theory is necessary for an understanding both of how ideologies can function as obstacles for the establishment of a science and of how they can be superseded through an epistemological break. For this purpose, PÊCHEUX makes a distinction between two forms of ideologies: "empirical" ideologies (which have technical origins) and "speculative" ideologies (which have political origins). At this point, he introduces linguistic terminology to characterise the different ways in which these two forms of ideologies function:
"The empirical form refers to the relation between a signification and a reality, whereas the speculative form refers to the articulation of significations to each other, under the general form of discourse. When we use terms imported from linguistics, we can say that the empirical form of ideology puts forward a semantic function—the coincidence of the signifier and the signified—, whereas its speculative form puts forward a syntactic function—the connection of signifiers to each other."17) 
The ideological process must be understood as a combination of the semantic effect and the syntactic effect. The first effect produces the reality of the signified, whereas the second assigns it its proper place between all other things that can be present in discourse in the given ideological conjuncture. The "social sciences" are blind to the speculative functioning of ideologies because they approach discourses with only empirical means. This approach leads to a technical interpretation of speculation, which overlooks its effects within the empirical approach itself. 
The two HERBERT papers outline the questions to be addressed by PÊCHEUX's automatic discourse analysis. Discourse analysis is aimed at a description of the functioning of ideologies in general, and of how this functioning is an obstacle for the establishment of a real social science in particular. 
In this context, linguistics provides some essential concepts. PÊCHEUX uses the term "metaphor" for the semantic production of reality, and he calls the syntactic relations between signifiers "metonymical". These terms hint at PÊCHEUX's indebtedness to structuralism, which had become popular through the influence of Roman JAKOBSON on Jacques LACAN and Claude LÉVI-STRAUSS. The reputed founder of structuralist linguistics was Ferdinand de SAUSSURE. A preoccupation with SAUSSUREan linguistics was widespread in France in the 1960s (in particular in "post-structuralist" circles). However, SAUSSURE's Cours de linguistique générale was more quoted than actually studied. SAUSSURE was generally seen as the founder of structuralism, which was conceived primarily as a rigid and one-sided method.18) According to this view, the object of SAUSSUREan linguistics was langue (the language) as a system of formal relations—in isolation of its use, its history and its content, which were referred to the domain of parole (speech). "Post-structuralists" tried to adhere to the (supposedly SAUSSUREan) structuralist method while counteracting its one-sidedness.19) In the footsteps of the anthropologist LÉVI-STRAUSS (e.g. 1969), their work was directed to problems in other disciplines than linguistics: LACAN directed his analysis to the structure and workings of the unconscious; DERRIDA's work focused on philosophical and literary analysis; FOUCAULT’s observations on language and discourse were part of his historical analysis of the present. 
PÊCHEUX addressed the question of the theoretical place of "discourse" within the SAUSSUREan model. His problem was that such questions as "What does this text mean?" were systematically excluded from linguistic analysis. Their solution was presupposed and, thus, left to the self-evidences of everyday experience. According to PÊCHEUX, it was precisely this "leaving uncovered the territory" by linguistics, without its reoccupation by another science, that allowed ideologies to invade that terrain (again). In other words, though linguistics had established itself as a science through a "SAUSSUREean" epistemological break, it had "forgotten" to develop an adequate theory of meaning production in discourse. 
Within the "post-structuralism" of that time, PÊCHEUX's project was unique for two reasons that are intimately related: he wanted to establish a social scientific theory of "discourse", and he emphasised the need of developing an instrument for creating experimental (vs. experiential) results (cf. HENRY, 1995). His emphasis on the need of developing an empirical alternative to linguistic "speculation", with its implied stress on both theoretical and analytic rigor, induced PÊCHEUX to a much more intensive study of (SAUSSUREan) linguistics than was common among his fellow-philosophers. This study was necessary in order to find theoretical room within linguistics for the concepts of "meaning" and "discourse", and in order to construct a scientific instrument for the analysis of these objects. PÊCHEUX's ADA69 presents the results of this study. 
4. Automatic Discourse Analysis
ADA69 starts from a criticism of traditional forms of content analysis and text analysis (PÊCHEUX, 1995a, pp.63-71 [PÊCHEUX, 1969a, pp.1-8]). Such analyses presuppose a subject (the analyst or his coders) capable of "reading" the meaning of a text. PÊCHEUX wanted precisely to avoid this reliance on the reading subject, because it would inevitably produce an ideological reading. However, it must be acknowledged that the role of "intuitive reading" was a matter of concern for content analysts as well. In the same year (1969), for instance, KRIPPENDORFF stated that
"... traditional content analysis may be said to have presented a technique for reliably intuiting content rather than analyzing it. However, when modern content analysts take the word 'analysis' in its literal sense and use, for example, computers for compatible interpretation of text, no part of the procedure can be delegated to the inexplicable process of intuition. [...] The explication of intuitively obvious semantic interpretations and judgments constitutes the most formidable obstacle in computer applications" (KRIPPENDORFF, 1969, p.6). 
Though this problematisation of intuitive reading was quite similar to PÊCHEUX's criticism, the solutions proposed were very different. One solution in content analysis was the construction of dictionaries in which entry words were defined with one or more "tags" representing categories in the investigator's theory. The prime example of this approach is The General Inquirer (STONE, DUNPHY, SMITH & OGILVIE, 1966). A different solution was to restrict the analysis to formal aspects of the studied texts only, i.e. to those kinds of analysis that do not presuppose interpretation. This approach resulted in a proliferation of lexicometric studies. These different solutions have in common that they are blind to the question that is primary to PÊCHEUX: the question of how "meaning" and the "subject" are produced in discourse. 
Central to PÊCHEUX's approach is the concept of conditions of production of discourse. Taking JAKOBSON’s model of communication as a starting point, he "sociologises" this model by requiring that the two subject positions in the model—the position of speaker/writer and that of listener/reader—be interpreted as locally and temporally specific imaginary positions. What matters is the place that each of them attributes to itself, to the other and to the "referent" (the object of which they speak). Such positions are imaginary, not in the sense of being "unreal" but of being related to images which produce material effects. The protagonists are not "free" in the choice of these images, which depend on structural relations (such as between worker and boss) and on what is said earlier and/or elsewhere. Such restrictions account for the relative stability of discourse through different occasions. This implies that "meaning" is more or less stable through such occasions, but that it changes when the conditions of production change, which is the case, e.g., when the same speaker speaks to another person or to the same person on a different subject. 
Whereas PÊCHEUX's theory of the conditions of production of discourse may appear as a more or less straightforward application of ALTHUSSERian concepts to JAKOBSON's model, his next step is more daring. This step consists of reintroducing the SAUSSUREan theory of value, which explains the meaning of words by their relations to all other words of the language, and applying it to the conditions of production of discourse. The meaning of words in a discourse (i.e. in a text or utterance) is explained by their relations to other words that are not said: words that could have been said but were not, words that were said previously (either on the same occasion or on other occasions), and words that could not be said. This interrelatedness between words is what PÊCHEUX calls "metaphoric" relations, and their meaning-effect is called a metaphoric effect (see PÊCHEUX, 1995, pp.96-100 [PÊCHEUX, 1969a, pp.29-33]). He outlines a theory of meaning as an effect of metaphoric relations (of selection and substitution) which are specific for (the conditions of production of) an utterance or a text. This theory of meaning is the bedrock of his instrument for (automatic) discourse analysis. 
Before outlining the main features of PÊCHEUX's automatic discourse analysis, it is to be recalled that it was his aim to develop an instrument that would produce experimental results (i.e. results which are the product of a theory-driven practice) as opposed to experiential results (i.e. results which are based on everyday experience). In order to study the meaning of discourses, he had to construct an instrument that required discourses (not meanings) as its input, and that would have information about the meaning of those discourses as its output. In other words, the instrument must construct metaphoric relations without the analyst "feeding" it with information about the experiential meaning of the words that build up these discourses. 
A method of discourse analysis that was outlined by the American linguist Zellig S. HARRIS (1952) provided PÊCHEUX with an instrument capable of doing exactly this: "We were fascinated by Harris, because we felt that he could offer something that would allow to escape both merely intuitionist hermeneutic positions and positivistic 'lexicometric' positions".20) It was HARRIS's aim to extend descriptive linguistics beyond the limits of the sentence. He therefore proposed to focus on a sequence of sentences or a text, and to conduct a formal analysis in terms of recurring patterns among its constituent elements. All assumptions about pre-given meanings of the elements, as well as all references to discursive sequences beyond the text in question, were excluded (cf. MARANDIN, 1979, pp.34-45; THOMPSON, 1983, p.217). Precisely because of this exclusion of pre-given meanings, HARRIS's method of discourse analysis fitted PÊCHEUX's need of a formal instrument. But, contrary to HARRIS, PÊCHEUX did not exclude discursive sequences occurring in other texts. Because it was PÊCHEUX's aim to construct a field of metaphors, i.e. of words that could have been in the discursive sequence under analysis but were not, he used the formal discourse analysis of HARRIS to relate sequences within a corpus of texts instead of sequences within one text. 
In order to guarantee the formal character of HARRIS's method, i.e. to prevent the tacit (re)introduction of "self-evident" meanings, it had to be made "automatic". But in what sense is ADA69 "automatic"? In fact, it is not "automatic" at all. Or, rather, it is as "automatic" as, say, instruments in astronomy and in physics. Like such instruments, automatic discourse analysis requires a very specific kind of input (which must be constructed as an input for the instrument) and it produces results that have meaning only within a very specific theoretical frame. The input of ADA69 is constructed in two subsequent steps, which are called the phase of corpus construction and the phase of linguistic analysis respectively. The third step then is automatic discourse analysis proper. But the output of this third phase does not consist of findings. The "findings" of an ADA69 procedure become available only through an interpretation of the results of this third phase. It is precisely this requirement of theoretical intervention, before and after the use of the "automatic" instrument (which itself is defined in terms of the theory), which allows for a scientific (re)occupation of the field of semantics. 
The aim of automatic discourse analysis is to provide the analyst with a metaphoric matrix that gives information about the production of meanings under theoretically specified conditions of production. The first phase of ADA69, that of corpus construction, consists of delineating the object of study ("Which conditions of production are to be studied in this analysis?"), and of selecting the set of texts or utterances that, according to the theory, represent those conditions. This set of texts is called the corpus. The metaphoric matrix must be constructed from the elements (words) that constitute this corpus. The second phase of ADA69, the phase of linguistic analysis, consists of rewriting all sentences of the corpus in a standard format, which is required for their being used as input for the phase of discourse analysis. This rewriting is called "linguistic analysis" because it consists of a form of syntactic parsing and because this parsing is done according to a linguistic theory. The internal cohesion of the separate texts of the corpus is preserved in this phase by attaching symbols to pairs of sentences which represent the connections between these sentences. 
The "linguistic analysis" phase of ADA69 is an application of the "SAUSSUREan" idea that the language is a system that is shared by a community (a nation or a culture) and that the theory of such a language, linguistic theory, can be treated as a neutral, formal device in an instrument of discourse analysis. This instrument produces metaphoric matrices from a standardised input, irrespective of the specific kind of linguistic analysis that has produced that input. There is no theoretical preference for a specific type of linguistic analysis. The theory of ADA69 requires that the input to the third—discourse analytic—phase has a standard format, which forms the basis for the construction of metaphoric matrices in this phase. But there is no discourse theoretical reason for the choice of a specific format. This choice is a linguistic one. In practical terms, however, the technical details of the phase of discourse analysis proper are dependent on the format of its input. And, more important, the resulting metaphoric matrices, called semantic domains, are constructed according to substitutions of words in places (within that format) that are defined by the linguistic theory underlying this format. The choice of linguistic theory, thus, has practical consequences. In other words, although different linguistic formats (resulting from different linguistic theories) require technically different treatments in the phase of discourse analysis which can result in partially different outcomes, ADA69 does not provide criteria that justify the choice of a specific linguistic format.21) 
The term "automatic", in sum, refers to the formalised character of PÊCHEUX's method. There are no "automatic" findings. ADA69 produces outcomes in a formal way, but they remain simply outcomes that must be interpreted. In his 1969 book, PÊCHEUX only mentions this final phase of interpretation of the results (see PÊCHEUX, 1995a, p.118 [PÊCHEUX, 1969a, p.110], where he introduces the concept of "reading"). Later, the difficulties that PÊCHEUX encountered in this phase inspired him to make revisions of his theory (which are documented in PÊCHEUX, 1995b, pp.175-183 [PÊCHEUX, 1969a, pp.70-77]). The revisions occasioned by difficulties in interpreting the results of ADA69 are an excellent example of the BACHELARDian theory regarding the role that theoretically constructed instruments play in the development of a science. 
5. Limitations and Adaptations
PÊCHEUX's 1969 book was a theoretical book. It presented a theory of discourse and meaning production, and outlined an instrument based on this theory. But ADA69 did not present results of the proposed procedure, because the instrument was not "built" yet. 
In the following years, PÊCHEUX constructed such an instrument. This required further work in two respects. On the one hand, the algorithms that could produce the metaphoric matrix in the phase of discourse analysis were implemented in computer programs (see GADET et al., 1995). On the other hand, the linguistic phase was refined in a "manual" (HAROCHE & PÊCHEUX, 1972). After the instrument ADA69 proposed in PÊCHEUX (1969a) had been actually built, the first empirical work with it was done (GAYOT & PÊCHEUX, 1971; PÊCHEUX & WESSELIUS, 1973; PÊCHEUX et al., 1979). The actual analyses confronted PÊCHEUX with new problems. At the same time, ADA69 was discussed critically in the linguistic literature (PROVOST-CHAUVEAU, 1970; TROGNON, 1972; FISHER & VERON, 1973). Both these criticisms and his own experiences in using the instrument resulted in a revision of the theory, and in changes in the procedures of automatic discourse analysis. 
In this period (1969-1975), PÊCHEUX was working on Les Vérités de La Palice (PÊCHEUX, 1975 [1982a]). This book was published in the same year as the ADA75 article in which the revisions of ADA69 are discussed (PÊCHEUX & FUCHS, 1975 [PÊCHEUX, 1995b]). In both publications, PÊCHEUX reconsidered the relationship between ideology and language, and—as a consequence—the relationship between discourse analysis and linguistics. Those publications, however, were directed to different audiences. The book can be seen as a sequel to the two HERBERT articles from the 1960s. It discussed problems of semantics, ideology and Marxist philosophy for a philosophically oriented audience, whereas the ADA75 article (PÊCHEUX & FUCHS, 1975), which discussed the same theory in relation to technical matters of discourse analysis, was written for an audience of empirical researchers. Yet, the two publications presuppose one another: results of automatic discourse analysis are used as examples in the book, which functions in its turn as a theoretical background for the discussion in the ADA75 article (PÊCHEUX & FUCHS, 1975). 
In this period of "thematization of uneven intrication" (cf. PÊCHEUX, 1995c, pp.237-238 [1990, pp.297-298]), PÊCHEUX did not radically change the actual procedures of automatic discourse analysis. What changed were the theoretical underpinnings of the whole enterprise and, correspondingly, the interpretation of its results: PÊCHEUX now explicitly introduced the ALTHUSSERian theory of ideology and LACANian psychoanalysis into his theory, which allowed him to theorise asymmetries between discourses. Moreover, he explicitly addressed the problem of the interpretation of the results of automatic discourse analysis (i.e. of metaphoric matrices) by distinguishing between different effects of metaphoric relations: they may be either relations of synonymy or "oriented" relations (PÊCHEUX, 1995b, p.165 [PÊCHEUX & FUCHS, 1975, p.61]). Because ADA69 was based on the assumption of stable conditions of production, it could not theoretically account for "ideological struggle" in terms of discourse. In 1969, PÊCHEUX's conception of the metaphoric matrix had included only relations of synonymy (e.g. the president/De Gaulle) and opposition (e.g. De Gaulle/the workers), but the produced matrices contained relations between elements that could not be interpreted that way (e.g. strike/the workers). The latter kind of relation is now defined as being oriented: one term is the origin, the source, or the argument of the other (e.g. the workers begin a strike). This relation is not an explicit part of the discourse studied, but tacitly presupposed by it. In other words, we see here the influence of another discourse within the one studied. By theorising this influence, PÊCHEUX introduced a true innovation into the theory of discourse: his theory of interdiscourse, defined as the "complex whole in dominance" of conditions of production of discourse.22) 
Neither ADA69 nor ADA75 were designed for studying the workings of interdiscourse, which was conceived as external to the (stable) conditions of production that were assumed. After 1975 it became increasingly clear to PÊCHEUX that this was a serious drawback of the theory underpinning ADA69. This brought him back to the study of linguistics (GADET & PÊCHEUX, 1981) and of the analytical tradition of Gottlob FREGE and Ludwig WITTGENSTEIN (PÊCHEUX, 1984).23) These studies resulted in a serious criticism of extant linguistic theories of grammar and semantics. In his last, unfinished article (GADET, LÉON & PÊCHEUX, 1984), for instance, PÊCHEUX discusses differences in meaning-effect between sentences that, according to transformational-generative grammar, have the same deep structure. It appeared to be impossible to adapt ADA69 to these new theoretical requirements. At the beginning of the 1980s, PÊCHEUX virtually abandoned it.24) This abandonment, however, does not imply a rejection of the possibility of designing other instruments of automatic discourse analysis corresponding to new theoretical standards. PÊCHEUX stuck to the BACHELARDian idea that the construction of instruments and the production of experimental results are necessary to the development of a science. He remained, therefore, very much interested in experimenting with recently constructed software (DEREDEC) as a tool for parsing sentences (through a recognition grammar) and in ways of using the results of this linguistic analysis as an input to other (discourse analytic) procedures.25) 
6. Deconstruction of Linguistic Theories
ADA69 started from the assumption that the production of meaning-effects can be explained by metaphoric relations within a discourse conceived as an isolated whole in which substitutions are systematically regulated. At the time of ADA75, the limitations and adaptations of this approach gave cause to conceiving the substitution of discursive elements as an effect of heterogeneity rather than coherence and systematicity. A discourse was now conceived as affected by several, mutually different, discursive formations involved in unequal power relations. The irremediable heterogeneity and equivocality brought about by relations, including struggle, between different discourses within a dominant whole tend to be obscured in linguistics by appeals to "evident" presuppositions or innate ideas. In order to develop a method for describing such relations, an "interdiscursive" domain had to be taken into account which was conceived of as the linguistic "outside" of single discourses. 
The exploration of the theoretical primacy of alterity and difference over identity, of "the other" over "the same", did not only disqualify appeals to linguistic axioms; it also made problematic a mere revision of ADA69, or an appeal to other formalised procedures of discourse analysis. Thus, the thematisation of the heterogeneity of discourse resulted in a lack of new method. This may have been due to the theoretical and practical problems inherent in finding homogeneous and stable procedures for describing an object conceived of as heterogeneous and unstable. But it can also be explained by a new style of working. From 1976 to 1983, PÊCHEUX did not search for a new, monolithic method, but tried to practice linguistics as a way of challenging existing self-evidences. 
A first step towards this new style was made in PÊCHEUX (1975), in which he repeated his earlier conclusion that the self-evidence of "objective" meanings and the self-evidence of "subjective" positions of those who deal with meanings are equally problematic. Whereas, however, in ADA69, he limited himself to the development of an instrument allowing to circumvent the ideological self-evidences of the analysing subject, this time, he reflected on the possibility of a non-subjective position for the analyst himself or herself. Subjects by definition identify with positions within interdiscoursive struggles between "dominant" discourses and counterdiscourses. Because analyses nourished by ideological self-evidences cannot be considered scientific, a scientific analysis is to be produced independently of subjects. In other words, the break with ideology requires a process without a subject. This requirement implies a disidentification of the subject, which is a political task implicitly added to the epistemological need of an automatic instrument. 
Later on, however, PÊCHEUX (1982b) distanced himself from this hope for desubjectification, largely for the same reasons for which he distanced himself from ADA69. Like he had to reject unproblematic, homogeneous conceptions of meaning, he also had to abandon the all too enthusiastic hope for a pure, desubjectified, subject position. PÊCHEUX's new style, provoked by his eye for the complexities of both interdiscourse and being an analyst, is manifest in an illuminating way in his introduction to the proceedings of the conference Matérialités discursives, which took place in 1980. The discourse analyst doesn't know anything anymore, he writes, not even what "reading" is. One has to "act silly, that is to say: decide to know nothing about what one reads, to remain a stranger to one's own reading, to exaggerate systematically the spontaneous disintegration of the sequences, so as to rescue the verbal material from even the last remains of meaning that still adhere to it".26) On first sight, PÊCHEUX thus seems to embrace a merely disidentificatory conception of meaning, realising a total dispersion of the initial univocality of discourse and of the subject. However his analytic drive is still manifest in continuous safeguards against absolute relativisations of meaning. His aim to smooth the way for an adequate understanding of ideological struggle remains in full force.
"We are not, therefore, talking about a multiple or plural way of reading in which a subject playfully increases the number of possible points of view to recognize itself more easily, but about a way of reading that is subordinate to a multi-layered and heterogeneous corpus, a way of reading the structure of which changes dependently upon this reading itself. This kind of reading implies that the reading subject accounts for the meaning it deciphers while, at the same time, loosing itself through it. After all, the interpretation follows the interdiscursive traces, which are preconstructed and transversal themselves."27) 
In discourse analysis, readings are not arbitrarily "invented". They are aimed at tracing the interdiscourse, which is socially given (preconstructed): "What is discussed time and again is the discursive analysis of a sequence with regard to an interdiscursive corpus of socio-historical traces".28) The analytic possibilities which are left seem to come down to continuing such traces by engaging in interdiscourse.29) 
This consideration of the problems of discourse analysis raises more questions than it provides answers (see especially PÊCHEUX, 1995c, pp.240-241 [1990, pp.300-301]). PÊCHEUX tries to develop an approach which, on the one hand, does not fall into the (sociologistic) assumption that the unity of the language is a mere illusion produced by political and social institutions, and which, on the other hand, does not "forget" the political, social and even poetical dimensions of language (as is done in logicism). GADET and PÊCHEUX (1981) elaborate on the question of whether such a two-sided approach is possible at all. They elaborate on a reading of SAUSSURE in which the language is conceived as a system of differences "without positive terms".30) Within this scope, the work of CHOMSKY is critisised because of its logicism. This criticism particularly touches on the way CHOMSKY deals with ungrammatical sentences—which occur quite frequently in everyday language use. What is the status of grammatical rules if they can be violated with impunity every now and then? (GADET & PÊCHEUX, 1981, pp.152-173). Neither logicism nor sociologism leave enough play for heterogeneity and "ungrammaticality". Either approach is reigned by "the order of the serious". In this respect, CHOMSKYan linguistics can be put on a par with Stalin's unificatory language politics (GADET & PÊCHEUX, 1981, pp.101-104; cf. GADET, 1977). 
In his contribution to the conference Matérialités discursives, PÊCHEUX (1981b) links the way of writing of two literary authors (BORGES and JOYCE) to the inclusive and exclusive effects that can be ascribed to compound sentences.31) In the analysis of complex sentences, in some cases, an utterance (e.g. a "restrictive" clause) is seen as included within another utterance. In other cases, the inserted utterance is interpreted as an apposition, which remains separate. Along these lines, PÊCHEUX distinguishes between a "writing of interpolation" and a "writing of disengagement". The first, "narcissistic" way of writing (which he finds in BORGES) maintains the reign of "the same". Grammar remains fully intact and applicable; discursive equivocalities and contradictions are presented as interpretative puzzles within the logical space of grammar. In the second, more differential and "split" way of writing, discursive breaks make themselves directly felt in the expression itself through "juxtaposed utterances with implicit connections, nominal sentences, interrupted or partially swallowed sentences, and grotesques accumulations and enumerations bristled with "ill-matched spouses".32) 
In accordance with this distinction, PÊCHEUX (1982c) shows that linguistic attempts to get rid of interdiscourse present themselves in many forms.33) Again and again, one attempts to analyse language as a simple "play within the rules", which leads inescapably to hypostatising certain discursive rules, presented as "natural" and "logical". Instead, PÊCHEUX formulates the necessity of a linguistics theorising language as a "play with the rules":
"The attempt to think of the language as a sphere of rules intrinsically capable of play, as a play with the rules, implies assuming within the language an order of rules that is neither logical nor social; it implies hypothesizing that syntax as a specifically linguistic sphere is neither a logical machine (an autonomous formal system, detached from the lexical, the semantic, the pragmatic and the enunciative) nor a fictitious metalinguistic construction (reducible to power effects inscribed in a mastery that would control written discourse)." 34) 
Such a conception of language may contribute to finding forced displacements in the possible interpretations of grammatical structures that are logically stabilised by linguistics. So, PÊCHEUX's discourse analytic partiality for silliness may get the challenging bearing
"that one would have to put pressure on constructions so as to enlarge the gaps, create new points of coagulation and run upon points where zones of resistance reveal themselves: to experience what we have called 'the impossibility within language'" (GADET et al., 1984, p.33). 
The epistemological problematisations presented by PÊCHEUX in his work on discourse analysis stay far from giving a "hard" basis to the social sciences. The analyst keeps getting stuck in formulations of contradictions one would rather like to avoid: logicism versus sociologism, seriousness versus play, linguistics versus poetry, heterogeneity versus homogeneity, interpolation versus disengagement. Yet, PÊCHEUX convincingly defends such contradictions against the compulsiveness towards unity displayed by the main trends in the social sciences (including linguistics). 
We hope to have demonstrated that also in the problems he raises, PÊCHEUX does more than draw impracticable consequences from merely theoretical premises. His theorising is accompanied by the construction of an instrument and the production of experimental results. This makes his work not only relevant as a contribution to the history of discourse theory and discourse analysis, but shows, in an exemplary manner, the nature of the obstacles, both in terms of theory and of method, encountered in any possible form of discourse analysis. The transition towards a sophisticated theory of discourse and a scientific method of discourse analysis is a matter of both forcing an "epistemological break" with empiricist conceptions of meaning and of producing the corresponding experimental results. This requires a renewed confrontation with the problem of how to develop a scientific practice of discourse analysis—a problem acutely defined by PÊCHEUX and not satisfactorily addressed to date. 
1) E.g. GHIGLIONE and BLANCHET (1991), and MAINGUENEAU (1991). <back>
2) Mots 9 (1984) and Langages 81 (1986); see also MALDIDIER (1984). The latter text opens the volume Histoire et linguistique (ACHARD, GRUENAIS & JAULIN, 1984), which is dedicated to the memory of Michel PÊCHEUX. <back>
3) MALDIDIER (1990) offers an extensive overview of PÊCHEUX's work, including a complete bibliography (cf. ACHARD, 1991, and HELSLOOT & HAK, 1992 for short reviews). We ourselves edited an anthology in Dutch translation which focuses on PÊCHEUX's later works (HAK & HELSLOOT, 1991). <back>
4) Examples are MacCABE (1979), COUSINS (1985), MACDONELL (1986), and WILLIAMS (1992). Some of the few publications in English that discuss PÊCHEUX's work in relation to empirical discourse analysis are WOODS (1977) and THOMPSON (1983). <back>
5) Cf. HENRY (1995), and also PÊCHEUX (1995c) in which he gives an overview of his own development. <back>
6) Translations without a reference to a published translation, such as this one, are ours. The original text is: "Il ne s'agit rien moins que de la primauté de la réflexion sur l'aperception, [...]. Nous aurons à montrer que ce que l'homme fait dans une technique scientifique de [l'époque contemporaine] n'existe pas dans la nature et n'est même pas une suite naturelle des phénomènes naturels" (BACHELARD, 1949, p.103). <back>
7) "Alors il faut que le phénomène soit trié, filtré, épuré, coulé dans le moule des instruments, produits sur le plan des instruments. Or les instruments ne sont que des théories matérialisées. Il en sort des phénomènes qui portent de toutes parts la marque théorique" (BACHELARD, 1937, p.12). <back>
8) BACHELARD developed his epistemology by studying the history of natural sciences such as physics. CANGUILHEM widened this approach by studying life sciences such as biology. Subsequently, ALTHUSSER and FOUCAULT applied epistemological concepts to the history of the social sciences. PÊCHEUX was the first in this tradition who studied the present conjuncture in the social sciences. He had been working in this field since 1966, when he obtained, with the support of CANGUILHEM, a research position in the laboratory for social psychology of the Centre National de la Recherche Scientifique (C.N.R.S.). <back>
9) In fact, PÊCHEUX and BALIBAR do not claim to be the "authors" of this text. It is presented as a summary of a lecture presented by François REGNAULT, which, for unclear reasons, could not be published (PÊCHEUX & BALIBAR, 1969, p.7). REGNAULT's lecture was part of a "philosophy course for scientists" organised by ALTHUSSER at the École Normale Supérieure in the winter of 1967-1968. This course consisted of the following contributions (ALTHUSSER, 1990, p.71; WOLF, 1985, pp.155-156): ALTHUSSER, Introduction (5 lectures); MACHEREY, The empiricist ideology of the "scientific object" (3); BALIBAR, From the "experimental method" to the practice of scientific experimentation (3); REGNAULT, What is an epistemological break? (1); PÊCHEUX, Ideology and the history of the sciences (2); FICHANT, The idea of a history of the sciences (2); BADIOU, The concept of the model (2). More lectures had been announced but the series ended when the "events" of May 1968 occasioned BADIOU to not continue his lectures. Part of the lectures have been published: ALTHUSSER (1990, pp.69-165), BADIOU (1969), FICHANT and PÊCHEUX (1969). <back>
10) "[...] le concept de rupture n'a rien à voir avec le projet volontariste d'effectuer un 'saut' hors de l'idéologie dans la science, avec la connotation religieuse qui s'attache inévitablement à ce projet, et les impossibles 'héros de la science' qu'il implique. Le nom de Galilée [...] est une unité mal choisie, car une science n'est pas le produit d'un seul homme: Galilée est l'effet, et non la cause, de la coupure épistémologique que l'on désigne sous le terme de 'galiléisme'" (PÊCHEUX & BALIBAR, 1969, pp.10-11). <back>
11) See Note 9. <back>
12) "l'intervention […] des points théoriques […] hors du champ philosophique, dans le travail effectif des chercheurs en Psychologie Sociale" (PÊCHEUX, 1969c, p.297). <back>
13) "Nous pensons qu'il importe [...] de prendre parti contre l'idée qu'il y a des 'données' premières à partir desquelles on imaginerait des théories: il importe de souligner qu'un problème scientifique se pose toujours dans un champ conceptuel et instrumental-expérimental" (PÊCHEUX, 1969d, p.75). <back>
14) "Une tâche urgente est donc de parvenir à distinguer les cas où les mathématiques sont l'objet d'une application technique et ceux où elles interviennent au niveau théorique, entre concepts et dispositifs expérimentaux. (On pourrait avancer comme exemple de ce dernier cas l'apparition des grammaires génératives et transformationnelles ...)" (PÊCHEUX, 1969d, p.76). <back>
15) The political orientation of PÊCHEUX's work is in many respects similar to the work of Ernesto LACLAU (cf. LACLAU, 1981, LACLAU & MOUFFE, 1985), who likewise attempts to theorise "the social" in the light of the irremediable loss of past certainties—though without PÊCHEUX's stress on linguistics. For a comparison of PÊCHEUX with other "marxist linguists" (GRAMSCI and VOLOŠINOV), see HELSLOOT (1995a). <back>
16) "… tout science est principiellement science de l'idéologie dont elle se détache" (HERBERT, 1968, p.74). <back>
17) "[L]a forme empirique concerne la relation d'une signification et d'une réalité, cependant que la forme spéculative concerne l'articulation de significations entre elles, sous la forme générale du discours. Pour user de termes importés de la linguistique, on dira que la forme empirique de l'idéologie met en jeu une fonction sémantique—la coïncidence du signifiant avec le signifié—, cependant que sa forme spéculative met en jeu une fonction syntaxique—la connexion de signifiants entre eux" (HERBERT, 1968, p.79). <back>
18) This received interpretation, which was inaugurated by Charles BALLY and Albert SECHEHAYE, the editors of the Course, and which still prevails, is disputed in re-readings of SAUSSURE's work by GODEL (1957), ENGLER (1967-1974), DE MAURO (1972), STAROBINSKI (1979), MARINETTI and MELI (1986), and HELSLOOT (2003). <back>
19) Especially DERRIDA, in much of his work, comes close to a reinvention and radicalisation of tendencies that can already be found in SAUSSURE's courses. <back>
20) "Man war fasziniert von Harris, weil man fühlte, da war etwas, das es erlaubte, aus einer sowohl rein intuitionistischen hermeneutischen wie auch aus einer positivistischen 'lexikometrischen' Position [...] herauszukommen" (PÊCHEUX, interviewed in WOETZEL & GEIER, 1982, p.389). <back>
21) Yet, PÊCHEUX justifies his choice for a specific standard format extensively (PÊCHEUX, 1969a, pp.39-86). The details of this discussion are superseded by later developments within linguistics, among them developments towards automatic parsing that are treated by LECOMTE, LÉON and MARANDIN (1984). <back>
22) Cf. PÊCHEUX (1982a, pp.113-118 and 184-193 [1975, pp.146-153, 240-250]). See on this book HOUDEBINE (1976); see also WOODS (1977), MACCABE (1979), COUSINS (1985), and MACDONELL (1986, pp.43-59). <back>
23) He also started to reread Friedrich NIETZSCHE (cf. HENRY, 1995, p.34, note). <back>
24) This process is documented in PÊCHEUX et al. (1982) and PÊCHEUX (1995c). <back>
25) LECOMTE, LÉON and MARANDIN (1984) gives a glimpse of DEREDEC's possible uses for discourse analysis. <back>
26) "Faire l'imbécile: c'est-à-dire décider de ne rien savoir de ce qu'on lit, de rester étranger à sa propre lecture, d'en rajouter systématiquement sur le morcellement spontané des séquences, pour achever de libérer la matière verbale des restes de sens qui y adhèrent encore" (PÊCHEUX, 1981a, p.16). <back>
27) "Es geht also nicht um ein mehrfaches Lesen, um eine plurale Lektüre, in der ein Subjekt spielerisch die möglichen Standpunkte vervielfacht, um sich darin besser zu erkennen, sondern um ein Lesen, das einem mehrschichtigen und heterogenen Korpus untergeordnet ist und dessen Struktur sich in Abhängigkeit von diesem Lesen selbst verändert. Das ist eine Art der Lektüre, in der das lesende Subjekt den Sinn, den es entziffert, zugleich verantwortet und von ihm enteignet ist. Denn die Interpretation folgt den interdiskursiven Spuren, die als solche vorkonstruiert und querlaufend sind" (PÊCHEUX, 1983, p.54). <back>
28) "In Frage steht jeweils die diskursive Analyse einer Sequenz hinsichtlich eines interdiskursiven Körpers soziohistorischer Spuren" (PÊCHEUX, 1983, p.55). <back>
29) Cf. LACLAU (1996, p.56): "[T]he impossibility of a free, substantial subject, of a consciousness identical to itself which is causa sui, does not eliminate its need, but just relocates the chooser in the aporetical situation of having to act as if he were a subject, without being endowed with any of the means of a fully fledged subjectivity." <back>
30) SAUSSURE (1983, p.118; 1972, p.166); GADET and PÊCHEUX (1981, pp.51-59) relate this theme explicitly to SAUSSURE's research of anagrams (STAROBINSKI, 1979). Cf. GADET's later work on SAUSSURE (GADET, 1987). On the anagrams, see also HELSLOOT (1995b). <back>
31) The distinction between the preconstructive and sustaining effects (effet de soutien) of language is already present in PÊCHEUX (1975). It is borrowed from HENRY (1975), who shows that the linguistic decision whether a subordinate clause is restrictive or nonrestrictive depends upon a prior non-linguistic interpretation of the meaning of the sentence. In PÊCHEUX's terms, the distinction can only be made if one of these readings is "evident" within the interdiscourse. <back>
32) "[...] d'énoncés juxtaposés aux connections implicites, de phrases nominales, de phrases interrompues ou partiellement effacées, d'accumulations et d'énumérations grotesques où pullulent les 'conjoints mal assortis'" (PÊCHEUX, 1981b, p.147). <back>
33) CHOMSKY is one of the outspoken examples of this exclusion; he speaks of the "system of knowledge and beliefs" rather than interdiscourse, see GADET and PÊCHEUX (1981, p.165). <back>
34) "Tenter de penser la langue comme espace de règles intrinsèquement capables de jeu, comme jeu sur les règles, c'est supposer dans la langue un ordre de règle qui n'est ni logique, ni social: c'est faire l'hypothèse que la syntaxe comme espace spécifiquement linguistique n'est ni une machine logique (un système formal autonome, extérieur au lexical, au sémantique, au pragmatique et à l'énonciatif), ni une construction fictive de nature métalinguistique (réductible à des effets de pouvoir inscrits dans une maîtrise supposée gouverner le discours écrit)" (PÊCHEUX, 1982c, p.23). <back>
Achard, Pierre (1991). Compte rendu de Denise Maldidier, Ed., L'inquiétude du discours. Langage et société, 56, 101-104.
Achard, Pierre; Gruenais, Max-Peter & Jaulin, Dolores (Eds.) (1984). Histoire et linguistique. Paris: Maison des Sciences de l'Homme.
Althusser, Louis (1990). Philosophy and the spontaneous philosophy of the scientists. London: Verso. [1967, Philosophie et philosophie spontanée des savants. Paris: Maspero]
Althusser, Louis & Balibar, Étienne (1970). Reading capital. London: New Left Books. [1968, Lire Le Capital. Paris: Maspero]
Bachelard, Gaston (1949). Le rationalisme appliqué. Paris: PUF.
Bachelard, Gaston (1984). The new scientific spirit. Boston: Beacon Press. [1937, Le nouvel esprit scientifique. Paris: Alcan]
Badiou, Alain (1969). Le concept de modèle. Introduction à une épistémologie matérialiste des mathématiques. Paris: Maspero.
Cipolli, Carlo (1972). Considerazioni teorico-metodologiche sull'analisi del discorso a proposito del methodo AAD di M. Pêcheux. Lingua e stile, 1, 199-229.
Cousins, Mark (1985). Jokes and their relation to the mode of production. Economy & Society, 14, 94-112.
Engler, Rudolf (1967-1974). Édition critique du Cours de linguistique générale de Ferdinand de Saussure. Wiesbaden: Harrassowitz.
Fichant, Michel & Pêcheux, Michel (Eds.) (1969). Sur l'histoire des sciences. Paris: Maspero.
Fisher, Sophie & Veron, Eliseo (1973). Baranne est une crème. Communications, 20, 162-181.
Gadet, Françoise (1977). Théorie linguistique ou réalité langagière? Langages, 46, 59-89.
Gadet, Françoise (1989). Saussure and contemporary culture. London: Hutchinson. [1987, Saussure. Une science de la langue. Paris: Presses Universitaires de France]
Gadet, Françoise & Hak,Tony (Eds.) (1990). Por uma análise automática do discurso. Uma introdução à obra de Michel Pêcheux. Campinas: Unicamp.
Gadet, Françoise & Pêcheux, Michel (1981). La langue introuvable. Paris: Maspero.
Gadet, Françoise; Léon, Jacquelin & Pêcheux, Michel (1984). Remarques sur la stabilité d'une construction linguistique: la complétive. LINX, 10, 23-50.
Gadet, Françoise; Léon, Jacqueline; Maldidier, Denise & Plon, Michel (1995). Pêcheux's approach to formalization. In Tony Hak & Niels Helsloot (Eds.), Michel Pêcheux. Automatic discourse analysis (pp.57-59). Amsterdam: Rodopi.
Gayot, Georges & Pêcheux, Michel (1971). Recherches sur le discours illuministe au XVIIIe siècle. Louis-Clause de Saint-Martin et les circonstances. Annales, 3-4, 681-704.
Ghiglione, Rodolphe & Blanchet, Alain (1991). Analyse de contenu et contenus d'analyses. Paris: Dunod.
Godel, Robert (1957). Les sources manuscrites du Cours de linguistique générale de Ferdinand de Saussure. Geneva: Droz.
Hak, Tony & Helsloot, Niels (Eds.) (1991). De taal kan barsten. Spanning tussen taalkunde en maatschappijwetenschap. Amsterdam: Krisis.
Hak, Tony & Helsloot, Niels (Eds.) (1995). Michel Pêcheux. Automatic discourse analysis. Amsterdam: Rodopi.
Haroche, Claudine & Pêcheux, Michel (1972). Manuel pour l'utilisation de la méthode d'analyse automatique de discours. T.A. Informations, 13(1), 13-55.
Harris, Zellig S. (1952). Discourse analysis. Language, 28, 1-30. [Reprinted in: Zellig S. Harris (1970), Papers in structural and transformational linguistics (pp.390-457). Dordrecht: Reidel]
Helsloot, Niels (1995a). Marxist linguistics. In Jef Verschueren, Jan-Ola Östman & Jan Blommaert (Eds.), Handbook of pragmatics (pp.361-366). Amsterdam: Benjamins.
Helsloot, Niels (1995b). Anagrammar. In defense of Ferdinand de Saussure's Philology. Beiträge zur Geschichte der Sprachwissenschaft, 5, 231-252.
Helsloot, Niels (2003). Divine Rock. Ferdinand de Saussure's poetics. Beiträge zur Geschichte der Sprachwissenschaft, 13, 187-231.
Helsloot, Niels & Hak, Tony (1992). Review of Denise Maldidier (Ed.), L'inquiétude du discours. Journal of Pragmatics, 16, 372-376.
Henry, Paul (1995). Theoretical issues in Pêcheux's Automatic discourse analysis (1969). In Tony Hak & Niels Helsloot (Eds.), Michel Pêcheux. Automatic discourse analysis (pp.21-40). Amsterdam: Rodopi.
Herbert, Thomas (1966). Réflexions sur la situation théorique des sciences sociales et, spécialement, de la psychologie sociale. Cahiers pour l'analyse, 2, 174-203.
Herbert, Thomas (1968). Remarques pour une théorie générale des idéologies. Cahiers pour l'analyse, 9, 74-92.
Houdebine, Jean-Louis (1976). Les Vérités de La Palice ou les erreurs de la police (d'une question obstinément forclose). Tel Quel, 67, 87-97.
Krippendorff, Klaus J. (1969). Introduction to part I. In George Gerbner (Ed.), The analysis of communication content. Developments in scientific theories and computer techniques (pp.3-16). New York: John Wiley and Sons.
Laclau, Ernesto (1981). La politique comme construction de l'impossible. In Bernard Conein, Jean-Jacques Courtine, Françoise Gadet, Jean-Marie Marandin & Michel Pêcheux (Eds.), Matérialités discursives (pp.65-74). Lille: Presses Universitaires de Lille.
Laclau, Ernesto (1996). Deconstruction, pragmatism, hegemony. In Chantal Mouffe (Ed.), Deconstruction and pragmatism (pp.47-67). London: Routledge.
Laclau, Ernesto & Mouffe, Chantal (1985). Hegemony and socialist strategy. London: Verso.
Lecomte, Alain; Léon, Jacqueline & Jean-Marie Marandin (1984). Analyse du discours: stratégie de description textuelle. Mots, 9, 143-165.
Lévi-Strauss, Claude (1969). The raw and the cooked. New York: Harper and Row. [1964, Le cru et le cuit. Paris: Plon]
MacCabe, Colin (1979). On discourse. Economy and Society, 8, 279-307.
Macdonell, Diane (1986). Theories of discourse. Oxford: Blackwell.
Maingueneau, Dominique (1991). L'analyse du discours. Paris: Hachette.
Maldidier, Denise (1984). Michel Pêcheux. Une tension passionnée entre la langue et l'histoire. In Pierre Achard, Max-Peter Gruenais & Dolores Jaulin (Eds.), Histoire et linguistique (pp.xi-xiv). Paris: Maison des Sciences de l'Homme.
Maldidier, Denise (Ed.) (1990). L'inquiétude du discours. Textes de Michel Pêcheux. Paris: Cendres.
Marandin, Jean-Marie (1979). Problèmes d'analyse du discours. Essai de description du discours français sur la Chine. Langages, 55, 17-88.
Marinetti, Anna & Meli, Marcello (Eds.) (1986). Ferdinand de Saussure. Le leggende germaniche. Este: Zielo.
Mauro, Tullio de (1972). Notes. In Ferdinand de Saussure, Cours de linguistique générale (pp.405-477). Paris: Payot.
Pêcheux, Michel (1969a). Analyse automatique du discours. Paris: Dunod.
Pêcheux, Michel (1969b). Idéologie et histoire des sciences. In Michel Fichant & Pêcheux, Michel (Eds.), Sur l'histoire des sciences (pp.13-57). Paris: Maspero.
Pêcheux, Michel (1969c). Sur la conjoncture théorique de la psychologie sociale. Bulletin de psychologie, 23(4-5), 290-297.
Pêcheux, Michel (1969d). Les sciences humaines et le "moment actuel". La Pensée, 143, 62-79.
Pêcheux, Michel (1975). Les Vérités de La Palice. Paris: Maspero.
Pêcheux, Michel (1978). Hacia el análisis automático del discurso. Madrid: Gredos.
Pêcheux, Michel (1981a). Ouverture du colloque. In Bernard Conein, Jean-Jacques Courtine, Françoise Gadet, Jean-Marie Marandin & Michel Pêcheux (Eds.), Matérialités discursives (pp.15-18). Lille: Presses Universitaires de Lille.
Pêcheux, Michel (1981b). L'énoncé: enchâssement, articulation et dé-liaison. In Bernard Conein, Jean-Jacques Courtine, Françoise Gadet, Jean-Marie Marandin & Michel Pêcheux (Eds.), Matérialités discursives (pp.143-148). Lille: Presses Universitaires de Lille.
Pêcheux, Michel (1982a). Language, semantics and ideology. London: MacMillan. [1975, Les Vérités de La Palice. Paris: Maspero]
Pêcheux, Michel (1982b). The French political winter. Beginning of a rectification (Postscript for English readers). In Michel Pêcheux, Language, Semantics and Ideology (pp.211-220). London: MacMillan.
Pêcheux, Michel (1982c). Sur la (dé)construction des théories linguistiques. DRLAV, 27, 1-24.
Pêcheux, Michel (1983). Über die Rolle des Gedächtnisses als interdiskursives Material. Das Argument, Sonderband 95, 50-58.
Pêcheux, Michel (1984). Matériel en vue de l'article "Complétives/Infinitifs/Infinitives". LINX, 10, 7-22.
Pêcheux, Michel (1995a). Automatic discourse analysis. In Tony Hak & Niels Helsloot (Eds.), Michel Pêcheux. Automatic discourse analysis (pp.63-121). Amsterdam, Rodopi. [1969a, Analyse automatique du discours. Paris: Dunod]
Pêcheux, Michel (1995b). Overview and prospects. In Tony Hak & Niels Helsloot (Eds.), Michel Pêcheux. Automatic discourse analysis (pp.123-186). Amsterdam, Rodopi. [1975, Mises au point et perspectives à propos de l'analyse du discours. Langages, 37, 7-80]
Pêcheux, Michel (1995c). Three stages of discourse analysis. In Tony Hak & Niels Helsloot (Eds.), Michel Pêcheux. Automatic discourse analysis (pp.235-241). Amsterdam, Rodopi. [1990, Analyse de discours: trois époques (1983). In Denise Maldidier (Ed.), L'inquiétude du discours. Textes de Michel Pêcheux (pp.295-302). Paris: Cendres]
Pêcheux, Michel & Balibar, Étienne (1969). Définitions. In Michel Fichant & Michel Pêcheux (Eds.), Sur l'histoire des sciences (pp.8-12). Paris: Maspero.
Pêcheux, Michel & Fuchs, Catherine (1975). Mises au point et perspectives à propos de l'analyse du discours. Langages, 37, 7-80.
Pêcheux, Michel & Wesselius, Jacqueline (1973). À propos du mouvement étudiant et des luttes de la classe ouvrière: 3 organisations étudiantes en 1968. In Régine Robin (Ed.), Histoire et linguistique (pp.245-260). Paris: Colin.
Pêcheux, Michel; Haroche, Claudine; Henry, Paul & Poitou, Jean-Pierre (1979). Le rapport Mansholt: un cas d'ambiguité idéologique. Technologies, Idéologies, Pratiques, 2, 1-83.
Pêcheux, Michel; Léon, Jacqueline; Bonnafous, Simone & Marandin, Jean-Marie (1982). Présentation de l'analyse automatique du discours (ADA69). Mots, 4, 95-123.
Provost-Chauveau, Geneviève (1970). Compte rendu de Michel Pêcheux, Analyse automatique du discours. La Pensée, 151, 135-138.
Saussure, Ferdinand de (1983). Course in general linguistics. London: Duckworth. [1972, Cours de linguistique générale. Paris: Payot]
Starobinski, Jean (1979). Words upon words. The anagrams of Ferdinand de Saussure. New Haven, CT: Yale University Press. [1971, Les mots sous les mots. Les anagrammes de Ferdinand de Saussure. Paris: Gallimard]
Stone, Philip J.C.; Dunphy, Dexter; Smith, Marshall S. & Ogilvie, Daniel M. (1966). The General Inquirer. A computer approach to content analysis. Cambridge, MA: MIT Press.
Thompson, John B. (1983). Ideology and the analysis of discourse: a critical introduction to the work of Michel Pêcheux. Sociological Review, 31, 212-236. [1984, reprinted in John B. Thompson, Studies in the theory of ideology (pp.232-254). Cambridge: Polity Press]
Trognon, Alain (1972). Analyse de contenu et théorie de la signification. Université Paris VII (dissertation).
Williams, Glyn (1992). Sociolinguistics. A sociological critique. London: Routledge.
Woetzel, Harold & Geier, Manfred (1982). Sprachtheorie und Diskursanalyse in Frankreich. Interview. Das Argument, 133, 386-399.
Wolf, Frieder O. (1985). Nachwort. In Louis Althusser, Philosophie und spontane Philosophie der Wissenschaftler (pp.151-171). Berlin: Argument.
Woods, Roger (1977). Discourse analysis: The work of Michel Pêcheux. Ideology and Consciousness, 2, 57-79.
Tony HAK, b. 1950, PhD (Political and Social Sciences, University of Amsterdam, 1988). 1967-1973, MSc in Psychiatric Epidemiology and Medical Sociology, Rotterdam Medical School. 1975-1995, Assistant Professor of Medical Sociology, Erasmus University Rotterdam. 1996-2000, Lecturer in Medical Sociology, University of Liverpool. 2000-Present, Associate Professor of Methodology, RSM Erasmus University. Tony HAK edited (with Françoise GADET) a volume of core texts of Michel PÊCHEUX in Portuguese, Por uma análise automática do discurso, Editora da Unicamp (1990) and (with Niels HELSLOOT) a similar volume in English, Michel Pêcheux. Automatic Discourse Analysis, Rodopi (1995). His next book (with Jan DUL) is Case study methodology in business research, Butterworth-Heinemann (2007).
Tony Hak, Associate Professor of Methodology
RSM Erasmus University
P.O. Box 1738
3000 DR Rotterdam
Tel. +31 10 4089594/4082006
Fax +31 10 4089014
Niels HELSLOOT, b. 1959, PhD (Erasmus University, Rotterdam, 1999), 1978-1985 Dutch linguistics and literature (Free University, Amsterdam) and 1981-1987 Philosophy (University of Amsterdam); freelance university teacher and translator, independent researcher in the history and philosophy of theories of language and communication. Niels HELSLOOT published, among other things, on the philological approaches of Antonio GRAMSCI, Ferdinand de SAUSSURE and Friedrich NIETZSCHE, and edited (with Tony HAK) the volume Michel Pêcheux. Automatic Discourse Analysis, Rodopi (1995).
NL-1094 JA Amsterdam
Tel.: +31 20 6656308
Helsloot, Niels & Hak, Tony (2007). Pêcheux's Contribution to Discourse Analysis [47 paragraphs]. Forum Qualitative Sozialforschung / Forum: Qualitative Social Research, 8(2), Art. 1, http://nbn-resolving.de/urn:nbn:de:0114-fqs070218.