Linguistics and Social Sciences

Abstract

Written with the suppression of the Tunisian students by their own government in view, Michel Foucault’s March 1968 ‘Linguistics and Social Sciences’ opens up a new horizon of historical inquiry and epitomises Foucault’s abiding interest in formulating new methods for studying the interaction of language and power. Translated into English for the first time by Jonathan D.S. Schroeder and Chantal Wright, this remarkable lecture constitutes Foucault’s most explicit and sustained statement of his project to revolutionise history by transposing the analysis of logical relations into the history of knowledge.

Introduction to Michel Foucault’s ‘Linguistics and Social Sciences’

Shortly after the publication of The Order of Things (1966), Michel Foucault obtained a leave of absence from the University of Clermont-Ferrand to teach philosophy at the University of Tunis. He moved to Sidi Bou Saïd in September 1966, and he stayed in post until October 1968. This brief period proved wildly transformative: Foucault shaved his head and drafted The Archaeology of Knowledge; at the same time, he was radicalised by a series of anti-colonial and anti-imperial student protests, which began as a response to the Six-Day War of 1967, were exacerbated by Hubert Humphrey’s visit to Tunisia in early 1968 and culminated on 15 March 1968 when thousands of Tunisian students gathered outside of the University of Tunis. They accused the Tunisian government of supporting US imperialism, called for the end of the Vietnam War, condemned Zionism and called on the Tunisian government to investigate torture and corruption. Authorities responded with violence, arresting over 200 students, torturing many and holding all without trial until September of that year (Medien, 2020: 495–6). During these months, Foucault allowed students to use his apartment as an organising space, hid a printing press in his garden so activists could print posters detailing the names of imprisoned Tunisians, gave sanctuary to student leader Ahmed ben Othmani, donated money for his legal defence and provided deposition testimony at Othmani’s September hearing (Hendrickson, 2013: 89–90). At the same time, he read texts by Leon Trotsky, Rosa Luxembourg, and the American Black Panthers. ‘It wasn’t May of ‘68 that changed me’, Foucault recalled. ‘It was March of ‘68, in a Third World country’ (1991 [1981]: 136).

The ‘Linguistics and Social Sciences’ talk was given at the Centre d’études et de recherches économiques et sociales of the University of Tunis during the fateful month of March 1968. Though Foucault does not connect the talk to his career, he considers various epistemological problems ‘that linguistics in its modern form raises for the human sciences’ that speak directly to Foucault’s own efforts to renovate the history of knowledge and, in 1968, to reconceive the relationship between his method and his politics (Davidson, 1997: 3–20). What structural linguistics shows, Foucault suggests, is that it is possible for an empirical field that studies humans to use formal discovery procedures to move from simple observation to the analytic discovery of relations that are ‘independent . . . in their form’, and, as a result, are ‘generalisable’ as types of relation. Because these relations are not determined by the objects they link together, he speculated that they ‘may be transposed to elements other than those of a linguistic nature’ – which is to say, to the objects of other disciplines. In the wake of this monumental event in the history of reason, he enjoins the human sciences to undertake three major tasks: 1) to purge themselves of the classical conception of causality and associated concepts (e.g. ‘predetermined syntheses’ such as the subject, author, history and the book), which had been used to rationalise empirical fields in the 19th century; 2) to formalise new procedures and concepts in order to discover to what degree ‘one can formalise this ensemble of relations . . . in terms of symbolic logic’; and 3) to design methods for other human sciences that successfully harness this ‘formidable instrument of the rationalisation of the real’ while accounting for differences between linguistics and other disciplines.

As a historian (and an ‘archaeologist’), Foucault proposes to retool linguistic method to detect the traces of material processes on language and, more specifically, knowledge. In contrast to linguistics, which reconstructs the grammaticality of language in order to predict what can be said, Foucault argues that the objective of history should be something different: deriving rules to predict what could have been said. By deriving rules of historical transformation, and by determining the statements that were not made, as well as irregularities in the statements that were made, one can make inferences about how excluded and modified statements have been shaped by invisible effects, effects which may be extra-linguistic in nature (economic, social, natural or other forces). With such a method, Foucault speculates, it would be possible to derive rules that would describe the laws of language in the sphere of its practical application. Written with the suppression of the Tunisian students by their own government in view, ‘Linguistics and Social Sciences’ opens up a new horizon of study, and epitomises Foucault’s abiding interest in formulating new methods for studying the interaction of language and power.

Translator’s note, Jonathan D.S. Schroeder

Linguistics and Social Sciences ¹

My topic will broadly be the following: what are the problems that linguistics, in its modern form, can introduce into thought in general, into philosophy if you will, and, more specifically, into the human sciences? ²

One frequently encounters the following thesis (as is the case with Lévi-Strauss in his Structural Anthropology): the analysis of language by Saussure and his successors – structural linguistics, in other words – has reached, over the course of the 20th century, what one might call a ‘threshold of scientificity’. ³ This threshold of scientificity can be seen, on the one hand, in the formalising methods towards which linguistics now tends and, on the other, in the relations that it enjoys with communication theory, and with information theory in general; thirdly, in its recent connections with biology, biochemistry, genetics, etc.; and, finally, in the existence of a technological domain of application, of which machine translation is, after all, merely one example. ⁴ Linguistics, then, has crossed a particular threshold, stepped out of the human sciences toward the natural sciences, out of the domain of interpretative knowledge into that of formal knowledge. ⁵ Linguistics, then, has aligned itself with genuine science, that is, true science or even exact science.

One frequently comes across a second thesis: from the point at which linguistics is assumed to have abandoned its habitual association and earlier familiarity with the human sciences, its position in relation to the human sciences becomes that of a model worthy of emulation and application; by the same token, the human sciences will naturally try to rejoin linguistics within this new form of scientificity at which it has finally arrived. Thus a sort of high-speed chase is instituted, with linguistics aligning itself with the proper sciences, and all of the human sciences trying to rejoin the normative level of the exact sciences alongside linguistics. This is the case for sociology, mythology qua the analysis of myth, literary criticism, etc.

One can critique these generally accepted theses. One can point out that the social sciences appealed to the science of language for something like a form or a content of knowledge [connaissance] long before the present day. After all, ever since the 18th century the social sciences have sought help when analysing language, and I will mention just a few examples here. Take D’Alembert’s Discours préliminaire de l’Encyclopédie (1995 [1751]): he explains that his aim in creating a dictionary that takes the form of the analysis of a language is to erect a monument that will permit future generations to know what the customs, knowledges, methods of the 18th century were. In other words, L’Encyclopédie was given the form of a dictionary of words precisely to establish an image, a profile, a picture of and a monument to 18th-century civilisation and society. One can also cite the text that Schlegel (2001 [1808]) wrote around 1807 on the language and the wisdom of the Hindus, in which he analyses Hindu society, religion, philosophy and thought on the basis of the specificity of their language. One should also not forget how Dumézil, who is not a linguist but a philologist, has recently managed to reestablish the social and religious structure of certain Indo-European societies working from philological analyses. ⁶ Therefore, the sustained relationship between the social sciences and the science of languages does not date from the current moment. The epistemological dislocation [décalage] between the sciences of language and the other human sciences does not date from the present day. I am not questioning the fact that transformational linguistics or structural linguistics have achieved a high level of scientificity; rather it seems to me that, beginning in the 19th century, the sciences of language had arrived at a higher degree of precision and demonstrativeness than all of the other social or human sciences combined. ⁷ From the start of the 19th century, people like Rask, Schlegel, Grimm established a coherent domain of philological facts of which the general principles, the methods of analysis and many of the results remained unchallenged throughout the 19th century. ⁸

And neither Auguste Comte’s sociology, 40 years later, nor Durkheim’s, 80 years later, would be able to demonstrate advances like the laws of phonetic evolution or the kinship system of Indo-European languages. In other words, the fact that the sciences of language display a level of scientificity greater than that of the other human sciences is clearly a phenomenon dating back almost two centuries. This is why I do not believe that one can simply say that the sciences of language and the sciences of society are drawing closer because the science of language – for the first time – has ascended to a higher order and the social sciences want to reach the same level. It seems to me that things are a bit more complicated than that.

What has changed, rather, is that linguistics has just given the social sciences epistemological possibilities that differ from the ones it offered them to date. It is the reciprocal operation of linguistics and the social sciences, far more than the level of scientificity intrinsic to linguistics, that make it possible to analyse the current situation. We find ourselves therefore face-to-face with an abiding fact: the epistemological dislocation [décalage] between the sciences of language and the other human sciences is an old one. But what is specific to the current situation is that this epistemological dislocation [décalage] has assumed a new form. If, today, linguistics can serve as a model for the other social sciences, then it is in a different manner.

I would now like to list certain problems that linguistics in its modern form raises for the human sciences. Structural linguistics is not concerned with empirical series of individualisable atoms (roots, grammatical inflections, words), but with the systematic sets of relations between elements. What is remarkable about these relations is that they are independent within themselves, that is to say in their form, the elements on which they bear; in this respect, they are generalisable, with not a hint of a metaphor in sight, and can potentially be transposed to all manner of things except to elements of a linguistic nature.

It might therefore be that one finds the same kind of relation existing not only between phonemes, but also between elements of a story or even between individuals who coexist in the same society. ⁹ Because the form of the relation is not determined by the nature of the element on which it bears, this possible generalisation of the relation brings us to two sets of important questions:

• 1) To what extent can relations of a linguistic nature be applied to other domains and what are these other domains into which they can be transposed? We should try and establish whether a particular kind of relation can be discovered elsewhere, if one can, for example, move from analysis at a phonetic level to the analysis of stories, myths, kinship relations. This is an immense task of empirical clearing in which all researchers in the domain of the human sciences are called to participate.

• 2) Which links exist between the relations that are discoverable in language or in societies in general and what we call ‘logical relations’? What might be the link between these relations and logical analysis? Can one formalise this set of relations entirely in terms of symbolic logic? The problem that then arises, which is philosophical on the one hand and purely empirical on the other, is at bottom that of the insertion of logic into the very heart of the real. This problem is philosophically and epistemologically very important. It used to be the case that rationalisation of the empirical mostly came about via and thanks to the discovery of a particular relation, the relation of causality. It was thought that an empirical domain had been rationalised when a causal relation had been established between one phenomenon and another. But now, thanks to linguistics, we discover that the rationalisation of an empirical field consists not only of discovering and being able to assign this precise causal relation, but in bringing to light an entire field of relations which are probably of the order of logical relations. Yet these relations know nothing of the relation of causality. We also find ourselves in the presence of a formidable instrument of the rationalisation of the real: the analysis of relations, one that is probably formalisable, and we realise that this very fertile rationalisation of the real no longer occurs through the assignment of determinism and causality. I believe that this problem of the presence of a logic which is not the logic of causal determination is currently at the heart of philosophical and theoretical debates. The recovery, the reactivation, the transformation of Marxist themes in contemporary thought turns on this. Indeed, the return to Marx or research on Marx of the Althusserian type show that Marxist analysis is not tied to an assignment of causality; this research tries to free Marxism from a kind of positivism within which certain people would like to imprison it and consequently to detach it from an elementary causalism in order to recover in it something like a logic of the real. But it is still necessary that this logic not be dialectical in the Hegelian sense of the term. Through a return to the texts, to the concepts of alienation, to the Hegelian period, in short, by a certain rapprochement with the Phenomenology of Spirit, there were solid attempts to emancipate Marx from the elementary positivism in which he had been imprisoned. Yet this Hegelian dialectics has nothing to do with all these logical relations that the aforementioned sciences are in the process of empirically discovering. What one is trying to find in Marx is something that is neither the deterministic assignation of causality nor logic of the Hegelian type, but rather a logical analysis of the real.

We come now to the problem of communication. The philology of the 19th century worked on particular languages; linguistics, after Saussure, works on language in general, like the grammarians of the 17th and 18th centuries. But the difference between structural linguistics and the old Cartesian analysis of language and of the General Grammar is that Saussurean linguistics does not consider language to be a translation of thought and of representation; it considers it to be a form of communication. ¹⁰ Thus considered, language and its function posit:

• - a sending pole on one side and a receiving pole on the other;

• - messages, as a series of distinct events;

• - codes or rules of construction for these messages which make it possible to individualise them. ¹¹

As a result, the analysis of language, instead of being conceived of as a theory of representation or a psychological analysis of the mentality of subjects, now finds itself placed on the same footing as all the other analyses that are able to study senders and receivers, coding and decoding, the structure of codes and the transmission of a message. The theory of language thus finds itself tied to the analysis of all the phenomena relating to information. This is important firstly because of the possibility of formalising and mathematising linguistic analyses to a significant degree, and subsequently because a new definition of what one might call the collective then comes into view. The collective of this new perspective is no longer universality of thought, in other words, a kind of grand subject who functions as a kind of social conscience or fundamental personality, or a ‘spirit of the age’. The collective, today, is a unit constituted by the poles of communication, by codes which are used effectively and by the frequency and the structure of the messages that are sent. As a result, linguistics colludes with analyses pertaining to the codes and messages exchanged between the molecules that constitute the nuclei of living cells. Nowadays biologists know approximately which code and which form of message implicate the phenomena of heredity inscribed in the nuclei of genetic cells. You see too that, by the same stroke, linguistics finds itself tied to the social sciences in a new mode, insofar as now the social can be defined or described as a set of codes and pieces of information that characterise a given group of senders and receivers. Phenomena like fashion, tradition, influence, imitation, which after Tarde came into existence as phenomena that were analysed exclusively in psycho-sociological terms, can now be read on the basis of the linguistic model. ¹² It is in this vein that we have to address the problem of historical analysis. We are accustomed to say that linguistics turned away from philology by adopting the synchronous point of view and abandoning the old diachronic point of view. Linguistics was allegedly to study the present and the simultaneity of language, while philology was to study the linear phenomena of evolution from one stage to another.

In fact, it is true that structural linguistics’ point of view is synchronic, but the synchronic point of view is not ahistorical and, what is more, it is not an anti-historical point of view. To choose the synchronic is not to choose the present over the past and the stationary over the evolutionary. The synchronic point of view linked to structural linguistics does not deny history, for several reasons in particular:

• 1) The sequential is only one dimension of history; after all, the simultaneity of two events is no less a historical fact than their sequentiality. We should not identify history with the sequential, as we naively do. We have to concede that history is every bit as much simultaneous as it is sequential.

• 2) The synchronic analysis that linguists carry out is not at all an analysis of the immobile and the static, but is in reality an analysis of conditions of change. Essentially, the question posed is this: what are the modifications that any language must go through for one of its elements to undergo change? Which correlations across a language are both requisite and adequate to bring about a single modification? In other words, the synchronic point of view is not a static cross-section that would deny evolution, but is on the contrary the analysis of the conditions under which evolution can occur. The old sequential analysis posed the question: where there is a change, what might have caused it? By contrast, synchronic analysis poses the question: in order to bring about a change, which other changes also have to be present in the contemporaneous field? It is a matter then, really, of a different way of analysing change, and not a way of denying this change to the advantage of the immobile.

If, in this kind of analysis, the assignment of causality no longer features as a directing theme, we can at least make the point that only synchronic analysis makes it possible to locate something like a causal assignation. So that the investigation of causality doesn’t disappear in something of a magical fog, we first have to define the conditions that will enable change.

This analysis of the conditions that are necessary and sufficient for a local change to occur is also necessary and almost indispensable if we are to transform this analysis into a practical and effective intervention, because the problem posited is knowing what I need to change if I want to change something in the overall field of relations. Far from synchronic analysis being anti-historical, I would say that it appears more profoundly historical since it integrates the present and the past, makes it possible to define the precise domain where a causal relation might be located and makes it possible, finally, to move towards practice.

All this is tied, I believe, to the renewal of the historical disciplines. We are accustomed to say that the historical disciplines now lag behind and that they have not reached the epistemological level of disciplines such as linguistics. Yet it seems to me that in all the disciplines that study change an important renewal has recently taken place: we have introduced the notions of discontinuity and transformation. Notions such as that of the analysis of the correlative conditions of change are well-known among historians and economists. The problem facing the specialists of the human sciences is how to use the example of linguistics, history and economics to introduce a rigorous analysis of change and transformation within the human and social sciences. They shouldn’t, in any case, turn away from linguistic analyses as though they were incompatible with a historical perspective.

Linguistics has ultimately made it possible not only to analyse language, but discourse too; in other words it has made it possible to study what one can do with language. Hence the analysis of literary works, myths, popular narratives, fairy tales, historico-religious texts, etc. One can now perform all of these analyses in the full realisation of what the description of language itself has brought about. The postulate that has not yet been fully accepted and remains to be revised is therefore this: since literary works, myths, popular narratives etc. are made of language, since it is really language which serves as the material in all this, can we not discover, in all of these works, structures which are similar, analogous, or, in any case, which may be described on the basis of the structures that we have been able to find in the material itself, namely in language?

To summarise all of this, I will say that linguistics presently speaks to the human and social sciences by means of an epistemological structure that is specific to it, but which allows it to reveal the character of logical relations at the very heart of the real; to reveal the – if not universal then at least extraordinarily extensive – character of communication phenomena that extend from microbiology to sociology; to reveal the conditions of change thanks to which one can analyse historical phenomena; and, finally, to begin at least the analysis of what we might call discursive productions.

Discussion

Salah Garmadi: Michel Foucault has emphasised the fact that synchrony is in no way opposed to diachrony. I tried, in my talk, to show how the methods of synchronic structural analysis have benefited linguistic studies of the diachronic type. Diachronic explanation of linguistic facts is effectively no longer the study of individual elements turning into other individual elements and has become the study of collectives in synchronic correlation and their transformation into other collectives in synchronic correlation, and this at each stage of the transformation.

But I would like to ask Michel Foucault a question in relation to the definition of synchrony. Michel Foucault says that synchrony is the explication of the conditions of change. In other words, for a change to occur, for an element to change, what kind of synchronic relations must there be in a given state of the language?

Well, for linguists, a synchronic description of the state of a given language is not a description of the conditions under which change is possible so much as a description of the conditions under which this state of the language functions at a given moment in its development. In linguistics, it is more properly diachronic study that deals with defining the conditions of change, by seeking to know, on the basis of an element that changes, how the entire structure of the language changes and by trying to establish the repercussions that a change to an element has on all the other elements in the linguistic structure, both on those elements which are comparable to it as well as those which are not.

Michel Foucault: Your point is this: what linguists analyse is the fact that when a change has come about, it brings about a certain number of other changes in the language. Well, I do not believe that this is precisely what linguists do. Linguists say: here is state A of the language, a state characterised by a certain number of traits. Here, now, is state B in which one establishes that a change has taken place and that in particular element a’ has been transformed into a’’. This is the point at which linguists establish that this change is always correlative to other changes (b’ to b’’, c’ to c’’, etc.).

Structural analysis does not then consist in saying: the change a’ to a’’ has brought about the series of changes b’ to b’’, c’ to c’’, but rather: there is no change from a’ to a’’ unless there is also the change from b’ to b’’, from c’ to c’’, etc.

Hichem Skik: I would like to say first of all that structural linguistics in no way excludes diachrony. In France, the first great work of structural linguistics and phonology is after all a work of diachrony and not of synchrony: it is the work of the current master of French phonology, André Martinet, a work entitled Économie des changements phonétiques (1955), which takes up the entire history of phonetic change studied in the 19th century from the viewpoint of philology and historical grammar. ¹³ So, from this point of view, linguistics is rooted in history.

To respond to M. Garmadi, I will say that the synchronic description of the language that one speaks at a given moment not only makes it possible to define the conditions of phonetic change, but also to see changes which are in the process of or on the cusp of taking place. This can appear paradoxical since, in principle, if we talk about analysing change, we are talking about analysing a point of departure and a point of arrival, which doesn’t seem to belong to synchrony. To understand this, we must call upon two very important ideas in structural linguistics:

• 1) The idea of neutralisation. In Parisian French, there is a distinction between two e’s: the closed é (in the article ‘les’) and the open è (in ‘lait’); we can say that é and è are opposed. But analysis of the language shows us that this opposition only occurs in a single position: in the final syllable. In any other position, the speaker has no choice between é and è: he has to use one or the other (é in an open syllable ending in a vowel: été; è in a closed syllable ending in a consonant: cette). We can say that the é–è opposition is neutralised, in Parisian French, apart from in the final syllable. One is entitled to think that an opposition which tends to be neutralised in most positions is quite a fragile opposition and hence threatened by extinction in the relatively long term.

• 2) The idea of functional load. But to appreciate in a more precise manner the solidity of an opposition and its chances of survival, linguistics calls upon another very important idea, the usage of which is sadly not yet very common, for we have not succeeded in making it operational: the notion of the functional load of an opposition. This involves analysing the load of an opposition in the language, that is to say the number of times where one needs this opposition to distinguish between words and to be understood. For example, if one takes in French the opposition ‘an – on’, one has no trouble amassing the words which are distinguishable only by one of these two sounds (bon and banc, blond and blanc, son and sang. . .). However, only with difficulty would we find more than three or four pairs distinguished only by the opposition ‘in – un’ (brin, brun, Ain, un. . .). One can say that the functional load of the opposition ‘an – on’ is very strong, while that of the opposition ‘in – un’ is very weak. Based on these observations (largely, though not exclusively, because in reality things are more complex), one might say that this opposition is very fragile and that it is at risk of being eliminated, given its infrequent usage (and indeed, we observe that in France the majority of people no longer make this ‘in – un’ distinction and say ‘inélève’ and no longer ‘unélève’).

These examples were intended to demonstrate the following: that even synchronic description, in defining the structure of the language in question, in showing, if you like, its weak points, can be a dynamic description and open onto the past and future history of this language.

Fredj Stambouli: I am a little embarrassed to speak at a gathering of linguists, but I will play along, and since the theme of our discussion is ‘Linguistics and Social Sciences’, I will try to make some remarks from my perspective as a sociologist.

I was extremely interested by Professor Foucault’s remarks, particularly those that relate to the position of structuralism in respect to history. Structuralism, we have just been told, this time with system and vigour, far from being opposed to change, that is to say to history, is simply one modality in the analysis of change, one modality of analysis that ‘precipitates’ change in some way and allows us to account for it.

Having said this, the question that I ask myself relates to the degree of operational utility of the concept of structure, but in sociology this time rather than linguistics. Though the structural approach has made itself positively felt in linguistics and in ethnology, because the focus of these two sciences is relatively autonomous vis-à-vis the subject and where the conscious praxis of individuals and groups is concerned, in sociology, the science of contemporary society and its future, by contrast, the structural approach is not an obvious one and encounters numerous difficulties.

If it is true that the structural approach focuses on the unconscious dimension of phenomena in order to sketch the logic that underlies them, and if it is true that it can also successfully privilege this level in linguistics and in ethnology, it can no longer do this as easily in sociology, nor in history. Indeed, if social facts lend themselves to scientific treatment, they cannot do so fully, for they remain marred by a degree of vagueness that the structuralist problematic does not yet seem to have lessened.

It is as if structuralism experiences difficulties whenever it operates on the present. However, whenever it operates on things that have settled and are thus relatively disconnected from individuals and collectives living out social facts, historical or otherwise, structural analysis is insightful. We can see this with Lévi-Strauss in ethnology, where, this time, one successfully operates on societies that have been described as ‘cold’ and have vanished; one operates on what remains of these, notably to myths. We can also see it with Foucault’s ‘archaeology of knowledge’ which analyses and gives an account of Western thought on the basis of three moments in the historical past, namely: the Renaissance, the classical period and the modern period, while being very careful, for the moment at least, not to make clear pronouncements about the contemporary period. This difficulty that structuralism experiences in giving an account of the present constitutes, in my opinion, its limits in sociology.

In conclusion, structuralism as a method of scientific knowledge remains limited in its applications and partial in its treatment of the human History, the individual and liberty, notions of duration, of rupture, of change, new versions of human and new configurations of socio-economic structures – there are many questions left hanging for structuralism and it does not yet seem to have provided satisfactory responses.

Mahmoud Seklani: I am happy to hear linguists affirm once again that linguistics, whatever its field of investigation and its methods may be, is not philology and also not literary analysis. Being in possession of languages is not enough to make you a linguist. Through its preoccupations, its problems, its methods and its content, linguistics is a social science par excellence, or at least this is the case in its current stage. This helps, at least I hope it does, to remove certain doubts. I would nonetheless like to contribute to this debate by clarifying the relationship between linguistics and certain other social sciences like sociology and demography.

On the occasion of this conference at which we find ourselves, and which we hope to repeat as often as possible in the future, we can speak of two main themes: we can ask ourselves if linguistics has attained a degree of scientificity superior to that of the other social sciences, to the extent that it can offer them its own methodology and concepts, because linguistics is now able to articulate its own methods thanks to structuralism. Can linguistics and the social sciences mutually enrich each other where their respective methods are concerned?

The history of scientific thought certainly shows that whenever a discipline has shown even a partial commitment to the scientific path, sooner or later it succeeds in committing to it entirely. Linguistic models that try to specify the forms of exchange between signs or of possible communications between individuals (language), between elements of machines (analogical systems, automatic translations) or between other entities draw on information theory and are dominated by ideas borrowed from thermodynamics, meaning that the notions of quantitative measures (entropy. . .) that have been introduced commit the discipline more and more to the scientific path and thus increase its degree of scientificity.

Statistical linguistics, which delivers laws similar to those found in biology or ecology (Zipf’s law), might perhaps reveal the identity of the internal structures of forms, which is probably due to the nature of those things whose classification indicates the possible existence of ‘functions’ akin to what is communicated by signs and symbols. From this perspective, biology and psychology have a terrain of convergence with linguistics.

There is no question that general linguistics and furthermore transformational linguistics, as we have just heard, seek to establish laws, and that the deductive and experimental method underpins their every step. But they do not appear to have subjugated all of their research to the scientific methods borrowed from the exact sciences, although the mathematical aspects of linguistics are currently developing at a vertiginous pace. Does the degree of superior scientificity where linguistics finds itself perhaps stem from this approach?

Foucault: There are two reasons why I will respond to the second question in the affirmative. First of all, because the social sciences are currently at different stages of development; those best served by circumstances, or by chance, can probably furnish others with an assortment of more or less varied tools likely to help them develop, and to better master their core data. Next, I believe that the majority of the social sciences follow the same scientific steps. In this current stage of their development, which some people consider prehistoric compared with the natural sciences, they seek to explain and interpret social phenomena, complex phenomena in fact, without necessarily establishing causal relations. At this stage, they need to affirm themselves and to lay claim to the scientific spirit which, unfortunately, is not always a given.

They are therefore on the lookout for analytical tools, or rather methodologies that will generate them, since the majority of their insights are dependent on the methodology pursued, for we shouldn’t forget that they were birthed, with difficulty, from literary, philosophical and historical disciplines. How old is sociology, the daughter of philology or of grammar, compared with astronomy, or demography even, compared with arithmetic and medicine? But I also believe that those sciences that can exchange concepts and methods, and I must stress methods and not information, with the social sciences, are the ones that present themselves as sciences of laws. None of the social sciences are defined in this manner.

What structural linguistics might have to offer, according to what we have just heard expressed most succinctly, is that there might be a convergence between linguistic analyses and sociological analyses. In the same manner that one might contrast, for example, langue and parole in linguistics, one might contrast the collective and the individual or some of their respective traits in sociology. It is possible that concepts belonging to one could serve as analytical tools for others, probably after a degree of adaptation.

I will mention for example concepts that are common to linguistics and demography, without knowing which of these two disciplines loaned them to the other: synchrony and diachrony. Demography often utilises them in its analyses and calls them ‘cross-sectional and longitudinal studies’. The generations can be studied through the events of their eras across a period of time (diachronically) or at a precise moment in their lives (synchronically)... These comparisons would doubtless be more fruitful if we could extend them even further. Could the concept of analysing population structures, essential in demography, one day find a field of application in sociology and linguistics?

Generally speaking, statistical demography has brought into focus a collection of methods that could be utilised advantageously by the majority of the other social sciences such as economics, sociology, geography and perhaps even linguistics. Allow me to borrow the following statement from Lévi-Strauss:

From the point of view of the absolute generality and immanence in all other aspects of social life, the object of demography, which is the number, is located on the same level as language. For this reason, perhaps, demography and linguistics are the two human sciences that have managed to go the furthest in terms of rigor and universality. (Lévi-Strauss, 1976 [1964]: 296–7)

But what distorts the scientific pretensions of the social sciences, or at least disrupts them, is that they all proceed from an ambiguous content immanent in the human: his nature and his behaviour.

And it is here that I am in agreement with what M. Stambouli says with respect to sociology. Up to this point, we can manage more or less happily to analyse what stems from human nature, or at least everything that is invariable, but when we come to the analysis of human behaviour, all of the social sciences’ methods and concepts are unreliable and sometimes inapplicable. I do not think that structuralism, put to the test, will succeed better than the others in addressing this challenge.

M’hamed Fantar: Professor Foucault has emphasised the difference between the relations found by linguists in language and the relations that scholars seek to find in other disciplines. He has particularly emphasised the negation of the notion of causality in structural linguistic research. He tells us, for example, that we have a linguistic state which is composed of different elements a, b, c. When a change affects a which becomes a’, b automatically becomes b’ and c becomes c’. But in order for a change to affect a which becomes a’, something has to happen. Also, it seems to me that one cannot deny causality completely. In fact, there is always something which acts, but instead of acting on a single element, it acts on an ensemble of elements between which there are organic relations.

When one speaks of linguistics, the discussion is about information, coding. I wonder whether linguistics has managed to find a system which makes it possible to decipher languages that have remained indecipherable until now, like Etruscan. Up until now, in order to decipher a language, linguistics has relied on translations. If it does not find bilingual inscriptions, it is not able to decipher a language. So we haven’t moved on from [Jean-François] Champollion. A language is always deciphered in relation to another language: Phoenician is based on the bilingual Maltese inscription, cuneiform is based on the bilingual inscription of Persepolis, etc. ¹⁴

Ahmed El-Ayed: I would like to begin by recalling a phrase put forth by Professor Devoto during the Tenth International Congress of Linguists in Bucharest. Seeking to put an end to the famous conflict between diachrony and synchrony, Professor Devoto used the following image: diachrony would be a river that traverses a lake, the lake of synchrony. ¹⁵

This said, I would like to return to semiology, also known as ‘semiotics’. This young science of signs heralded by Saussure is becoming more and more independent and seems to have immense fields of application. This is why an American scholar, Thomas Sebeok, was able to talk, at the same congress, about a zoosemiotics, a pan-semiotics, even predicting the possibility of extraterrestrial communication. ¹⁶

Human and animal communications make up so many of the manifestations of a genuine code of life. This is why modern linguistics, with its different branches, and in particular its overlaps with semiotics and genetics, seems to have become the definitive science, extending tentacles of interest to the other disciplines known as human sciences.

Abdelkader Zghal: I would like to highlight one or two ways in which sociologists and linguists might collaborate. While studying cooperative farms, I became aware of the existence, among the farmers [paysans], of a very particular kind of language that does not exist in the cities, neither among the administrative elite nor among the rest of the urban population.

I am not a linguist, but I think that, in order to study the extent to which ideology and the message of the administrative elite has reached this farming [paysanne] population, linguists would have to examine this particular language, recently created on a regional level and gradually taken up by the farmers [paysans]. Collaboration on this subject between linguists and sociologists would help the latter to know the extent to which the projects dreamt up by administrative authorities have been disseminated by this language, how many farmers [paysans] were reached by the message and what, finally, the respective percentages of words adopted or rejected by the farmers [paysans] were. I will give one example here: the notion of amortisation [amortissement]. The farmer [paysan] receives this idea under the Arabic form نقصٌ ‎مُعتبر (naqs’un mu’tabar), literally ‘considered diminution’ [diminution considerée], and it would appear that he does not understand it, since a tractor for example is still new and has 10 per cent of its value deducted all the same.

The second area where sociologists and linguists might collaborate is the organisation of the kinship system in Bedouin society. I was personally struck when I observed, among the semi-nomadic populations of Tunisia and across the whole of the Maghreb, the importance of the rule of marriage with the daughter of the paternal uncle, together with the solidity of the attachment to the ethnic and familial group, but at the same time I observed the absence of precise words expressing that reality. The words قبيلة (qabīleh), عائلة (‘ā’īleh) or عرش (‘arsh) are very imprecise and express very different things. Though the structure of the kinship system is restrictive, the words that exist are imprecise. There is a genuine contradiction between the degree of submission to the structure of the kinship system and the weakness of the words expressing the agnatic groupings. The word قبيلة = qabīle expresses ‘various things that the French translate according to a hierarchised structure as confederation, tribe, faction, sub-fraction’. ... But this is a feudal logic, whereas tribal organisation is a restrictive segmentary organisation. I believe that linguists, on this point, might help sociologists understand the relations between language and tribal organisation. Here too I would like to give an example: we know that there are tribes that were very important in the 15th and 16th centuries and whose names have disappeared. The population still exists, but the names have gone. What is the history of those names? Some have disappeared, different ones have come into being and yet the population has remained the same and the organisation of the kinship system is still restrictive.

Gilbert Naccache: First I would like to mention my concern regarding the conception of sociology as defined by M. Stambouli. I do not see how there can be a science in which one thinks that the action of the individual is fundamental. I do not see how an approach like this can be scientific. It seems to me that, all things considered, the action of the individual on language is much more constraining than the action of the individual on the social milieu. I believe that where sociology and linguistics are concerned, causality must be sought elsewhere than within these two disciplines.

In reality, what is decisive both in sociology and in linguistics, it seems to me, are the economic conditions, conditions that are extra-sociological and extra-linguistic. M. Zghal has just spoken about a linguistic change that has taken place in cooperative farms under the manifest influence of changing economic conditions.

The recent history of Tunisia shows that there are three linguistic registers, four including the language of the cooperatives, five with that of other economic sectors, etc. These different registers are tied to particular social classes. These are not the same social classes that speak French, classical Arabic, or Arabic dialect, and Tunisians who use these three languages do not use these languages at random with random interlocutors. The transformation of language does not result from its own dynamic and its own structure, but from the influence of external economic and social conditions. Perhaps, in our next multidisciplinary debate, we could interrogate these mutual relations more forcefully – these interdependences between linguistics, ethnology, sociology, on the one hand, and economics, on the other.

Noureddine Bouarroudj: It seems to me that Michel Foucault has emphasised the analogies between the structural approach in linguistics and methods of structural analysis in other sciences, notably information science. But I believe that we have forgotten an important subject, one that lies at the centre of problems of linguistics and information, namely the study of the human brain or, in other words, of the centres of language.

The second point where I would like to hear what the linguists have to say is whether it is possible for human language to be communicated to animals. It seems to me that certain animals can respond in an intelligent fashion to certain elements of language. I wonder therefore if linguists have carried out research on this dimension of the question, and especially, in bio-psychology, on the domain of conditioned reflexes.

Foucault: I would like to respond to some of the questions I have been asked. First I would like to let you in on something that appears to still be a secret in Paris, which is that I am not a structuralist. With the exception of a few pages that I regret, I have never used the word ‘structure’. When I talk about structuralism, I talk about it as an epistemological object that is contemporaneous to me. That said, there is a method that interests me in linguistics, which M. Maamouri presented to you a short while ago and which has been given the name ‘generative or transformational grammar’. ¹⁷ It is something of this method that I am trying to introduce into the history of ideas, of science and of thought in general.

M. Stambouli said just now that the notion of structure is not directly usable in sociology. I am entirely in agreement with him on this issue and I don’t think there is a single linguist or structuralist who would tell you that it is fully usable. I believe that the problem is this: given the extraordinary provision of concepts, methods and forms of analysis that linguistics and a certain number of related disciplines such as semiology have delivered recently, it seems to me that the analysis of certain social phenomena could be facilitated and enriched by the very transformation of these analytic methods. I think that the sociologist could even enrich linguistics, on condition that he abandons his attitude of total rejection or total acceptance and asks himself the question: what would I have to change in the concepts, methods and forms of analysis of linguistics so that they become usable for me in a particular field?

As for M. Fantar, he has raised two problems: causality and translation. Where causality is concerned, I do not see how his opinion differs from what I have argued up to now, that is to say the fact that structural analysis tries to define the field within which causal relations might be assigned. When we say: there is no change from a’ to a’’ without a change from b’ to b’’, c’ to c’’, etc., we are not assigning causality, of course, but any individual who wishes to discover the cause of phenomena has to begin by giving an account. In other words, what defines structuralism is the field of effectuation of a causal explanation.

Khalil Zamiti: M. Foucault has just raised the problem of applying methods used by structural linguistics to sociology. This approach would only be legitimate to the extent that authentic discoveries could ensue from inexact hypotheses, a process for which the history of knowledge offers multiple examples.

The fundamental difficulty of the methodological transposition that is being interrogated boils down to the following fact: if language is a product of social life, although discourse determines, to a certain extent, the field of the real, which prompts Lévi-Strauss (1987 [1950]: 21) to seek ‘a symbolic origin for society’, the subjects of collective existence are at once products and producers of language [langue].

Perceiving the multiple relations of complementarity, opposition, mutual implication woven between the different material and ideological aspects of real collective units, ethnologists and sociologists have adopted the concepts of totality and system: modelled on language, all culture is systematic or tends toward systematisation. The temptation is therefore strong for the sociologist: confident of this general homology, he hopes to establish an advance bond to his discipline by using the more rigorous methods of the linguist. Yet each time he runs up against the same obstacle, because the facts that emerge from his domain are not at the same level of empiricity as the much more elaborated materials with which his colleague is accustomed to working. When he tries to contain moving social reality within schemas and interpretive models borrowed from the linguist, relentlessly unpredictable factors intervene to upset his scaffolding.

Let’s take an example. Institutionalised in Great Britain after the First World War, vocational training was developed in that country during the 1930s. In France, it dates back to 1939, when [Raoul] Dautry, the Armaments Minister, decided to promptly place key workers in factories to help the war effort. A critical means of transmitting technical knowledge, this training is part of the suite of institutional relations that individuals organised among themselves as their means of adapting to the situation and milieu. In Western Europe in the 20th century, vocational training and industrialisation have been two complementary aspects of the same reality, the same discourse. By contrast, the wholesale introduction of vocational training in Tunisia via a decree promulgated on 12 January 1956, when industrialisation was still in its infancy, created challenging employment issues and hence a disjunction, in society at large, just as serious as the semantic scandal provoked by the introduction, in deterministic language, of an alien term. The analogy can be pushed even further, since, in the two cases, that of language and of society, the analysis tends to be expressed in terms of the adoption or rejection of an exogenous code inserted into a receiving system that is either in agreement or disagreement.

But the divergence quickly declares itself, for, if the injection of a linguistic element into another system exhausts itself in non-sense and radical rejection, the training received by a subject is one of the factors that is likely to allow that subject to contribute to the creation of a job and thus to reestablish the coherence that was initially lost. The ambivalence of the subject thwarts all efforts to assimilate it to the language. Several debates, which are now legendary, such as the one between Malthusians and populationists, were only made possible thanks to the contradictory and unilateral interpretations, of one or the other, of the two indissociable dimensions of this original dialectical drama of human liberty. In this respect, [Alfred] Sauvy liked to recall a Far Eastern proverb: ‘Give a man a fish and you feed him for a day, teach a man to fish and you feed him for a lifetime.’ This is the reason why we can probe, with desire but without hope, the eventual extension to sociology of the prestigious methods of linguistic analysis of the structural type.

S. Garmadi: I would like to return to M. Stambouli’s contribution with a view to determining more precisely what seems to me to constitute common ground in the problematic of our two disciplines, sociology and linguistics. What M. Stambouli has said might lead us to believe that the distance between these two social sciences is great. Sociology would be the study of human phenomena, present and yet to come, insofar as these are connected to the transmitter subject. And as structural linguistics tends more and more to dissociate the study of discourse and of human speech from that of the subject, of their intentions, of the sociological conditions in which they live, etc., one might believe that there is a kind of profound antagonism between methods of analysis in linguistics and in sociology. In fact, this seems to me to be secondary compared to what I believe profoundly binds the methodology of these two disciplines together.

It seems to me in effect that what is primary in sociology is the study of tensions. I have the impression that, when sociologists speak, they speak above all about what some social collective or other is resisting. In other words, contemporary sociology seems to privilege a domain, that of the study of the space between the norm and the real behaviour of people, whether this norm is current or whether it exercises its power and its reality in the past.

Well, this is exactly what modern linguists do. Linguistics, by dissociating langue from actual parole, paradigm from syntagm that is to say, by dissociating the norm of real linguistic behaviour from speakers, takes exactly the same approach. A Tunisian who says ‘donne-moi la parapluie’ makes an error of gender for the grammarian. But for the linguist, this ‘error’ is significant and he will bestow his full attention on it. For him, the Tunisian speaker in question is torn [en tension], for when forming his syntagm [donne-moi la parapluie] he has at his disposal two norms, two codes: the French code that demands ‘le parapluie’ and the Arabic code that suggests ‘la parapluie’ (the equivalent Arabic sh’āba = سحابة being feminine). In this particular case, it is the Arabic code (the ancestral code, sociologists would say) which has won him over. Sociologists, by studying for example the degree of acceptance or rejection of new norms, anti-conceptual practices, by a society with ancient contrasting norms, therefore follow the same methodological path as linguists do. To finish on this point, I will say that, in these domains of tension, of distortion between the formal, that is to say the norm, and the real, that is to say behaviour, of the gap between social conduct and ancient and modern linguistics, an extremely fruitful collaboration can and should be established between Tunisian sociologists and linguists.

And I will conclude by thanking M. Zghal for the two areas of sociolinguistic collaboration that he has suggested and that have allowed us to consider new linguistic forms in actual use by the Tunisian population, and by wishing that C.E.R.E.S. makes a habit of organising, once a year perhaps, interdisciplinary debates such as these, in which there is certainly great interest.

Translated by Jonathan D.S. Schroeder and Chantal Wright