THEORETICAL PROBLEMS IN INTERNATIONAL COMPARISONS:

TOWARD A RECIPROCAL IMPROVEMENT OF SOCIETAL APPROACH AND RÉGULATION THEORY BY METHODIC STRUCTURALISM

Bruno Théret

The context of that contribution is the recent reflections on the comparative methodology (Théret, 1997) to reach a synthesis between two methodological traditions in general sociology, i.e. Durkheim's and Weber's. This synthesis should combine inductive and in-depth historical analyses and, in order to study the degree of their generality, extensive geographical (international) confrontations of these analyses by means of theory and statistical methods. That perspective will allow to evaluate more precisely the specific contributions of SA and RA.

In this perspective, the first section of the paper deals with the specific contributions, as well as the present limits in this matter, of the societal and régulation approaches (from now on SA and RA respectively). The second section explains what could be drawn, on the methodological and theoretical levels, from their articulation, and in this perspective, goes back to the structural method of inter-societal comparisons used by Claude Lévi-Strauss in his research on the systems of kinship in the so-called "cold" societies, obviously adapting it to the context of contemporary societies.

I. The contributions of SA and RA to comparative methodology

In France, most of the comparatists who favour a synthesis of the statistical and historical methodologies recognise themselves either in SA or in RA. These paradigms aim, indeed, at keeping together both levels of universality and diversity, by using an empirical approach whose theoretical command is sought after. Moreover, compared to comparative international literature, they have an added value, and this is what is of particular interest here: SA and RA ask explicitly what the comparable objects of research should be, and give some theoretical and empirical answers to this question.

The societal effect approach

Marc Maurice introduced the "societal" or "inter-national" approach, as a "structural approach" which aims at going beyond what opposes the "cross national" universalist and "cross cultural" particularistic approaches. SA "does not consists in making an impossible integration of paradigms which oppose each other, but rather in shifting the logics of analyse which characterise them", only way to define a "comparability principle of the incomparable" (Maurice 1989: 183).

According to SA, the various objects (micro) or social sub-systems (meso) forming the social matrix are not comparable on a term to term basis, and from one country to another, because they only draw their meaning from the societal context in which they are embedded. On the other hand, one can compare "sets of phenomena which constitute, through their inter-dependence, national 'coherences' which are each country's own", i.e. "the (macro/micro) process of interdependence" and the "mediations they necessarily imply" (Maurice 1989: 182). "The pregnance of micro/macro interdependences which contribute to the construction of actors and which constitute each national coherence is such that it excludes all comparisons on a term to term basis, each element which constitute these coherences having some sociological significance only if it is related to the whole which they are part of. We are then in a case where a discontinuity (excluding a comparison on a term to term basis) and a continuity based on the comparability of the totalities which constitute each particular societal set coexist" (Maurice 1989: 183).

Thus, applying the SA to comparisons of family policies, Jean-Claude Barbier asks that one "refrains from making global comparisons on the basis of misleading lexical equivalences" and rather "look for a use of basic categories that should be detailed and pragmatical as well as inserted within the global logic and the context via a constant 'to and fro' move" (Barbier 1990: 345). For Barbier, indeed, the comparison of social policies of a given kind in various countries is "impossible without the comparison of the whole systems of social protection" of these countries. This task implies that these systems "be replaced in their current logic of operation, as well as in their historical genesis, their cultural context, and the explanation of the values and norms which rule the implementation of various policies within the national boundaries" (Barbier 1990: 355).

The SA is therefore based on the explicit definition of a comparability criterion: only systemic sets of relations defined by restricted interplays are comparable. In other words, only totalities which are definable by their internal context (if one may put it this way) and not by their external context - that is by systems of relations whose coherence does not depend on their environment but only on their law of composition - are comparable. Such a "principle of comparability", which solves the problem posed by the contextual effects by pushing as far as possible a "logic of endogenisation" (Maurice 1994: 657), obviously provides a first class justification to the comparative methodology.

However, the SA, at least in the way it has been developed until now, implies that the societal coherence comes under the national and therefore State level exclusively. This leads to two hypotheses which are debatable, especially nowadays:

- There are no internal discontinuities within the Nation-States;

- The international environment (the relations between States and the multinational scope of capital) does not affect the national coherences.

Indeed, for the supporters of the SA, it seems that the various "systems" (productive, educative, family, etc.) composing a national society must fit together in a holistic way. The society is considered as a real community constructed by social actors (which are, in the same time, organically defined by it) and not as resulting from a blind interplay and partial coherences. On the other hand, the culturalist point of view seems to be the right one, even if the national culture itself is considered as a historical product, a "constructed social entity", the social actors being created in a micro/macro dialectic at the national level.

This consideration of the national space as the unique place of societal closure makes the SA partly incoherent with its own epistemological presuppositions. The problem is not that the "national" cannot be the level of societal coherence where the contextual effects on the construction of social actors and of institutions close upon themselves. What is of concern here is that it must be established and located historically, and not stated as an apriorism. The nation itself is a category which does not necessarily have the same meaning in each country, and there is no certainty as to whether all the actors construct themselves and are definable in its framework. Besides, one cannot skip a theoretical elaboration on the State and the territory within which it fits, and act as if they were natural datas from which one could obviously start so as to carry out intersocietal comparisons.

Moreover, the notion of societal coherence thus restricted to the national frame creates risks of functionalism. If the social actors, in the process of their very construction, build the societal totality, the national space of coherence, is there any room for any disfunctionning outside the anomie of the individuals which cannot reach the status of social actors ? If the actor only exists through the coherence of the social system which he builds while building himself, can one reasonably postulate that such a coherence only exists at the national scale, and that, as a consequence, the national social space is not divided by contradictory logics of action?

Lastly, carrying out "inter-national" comparisons by choosing as a starting point the categories of the productive and educative systems (Maurice et al. 1979a) or labour markets or family structures (Benoit Guilbot 1987 and 1989) as it has been done, represents a real contradiction. Aren't these categories de-contextualised and therefore assimilated to invariants in proceeding that way? In order to create them as general categories, should not one consider them as coherences which - even if they can be of value at the national scale - are only infra-societal matters as they are only elements (among others) of the process of societalisation? This problem was recently illustrated by Barbier who tried to re-construct from the notion of "family impact" a concept of family policy which would make it possible to go beyond the incomparability of the administrative definitions of these policies and of the datas which go with them. Indeed, one wonders why this author, who himself highlights the socially created character - and by different means in each country - of the notion of family, eventually treats this notion as a "universal" by simply defining the field of family policies more widely than in the administrative way. The limits thus established to the "family impact" policies seem as arbitrary as ever. One could expect, on the contrary, either a construction of the concept of family as a system which would be general enough to incorporate all the national domestic structures found in the countries compared (Hilliard 1994), or a radical deconstruction of the notion of family policies (Hantrais & Letablier 1994) accompanied by a reconstruction, at the level of the social protection system where there is systemic coherence, of the concepts suitable for the empirical capture of "family policies" such as they are officially displayed.

In short, on the one hand, the SA has the advantage of explicitly theorising the question of international comparisons by defining a credible "comparability principle" and by emphasising the historical interplay of social actors and institutions as well as the micro/macro dialectic through which these actors and institutions are built. On the other hand, this approach is not completely satisfactory because it pre-supposes the exclusive pertinence of the national frame for the construction of coherences, and at the same time re-introduces universalist categories at an infra-societal level.

The régulation approach

The régulation approach (RA) has many features in common with the societal effect approach. Like SA, RA is a neo-structuralist theory which calls for inter-disciplinarity and which gives a primary role to history, to institutions and to strategies of social actors in the analysis of growth regimes and of national development modes. RA objects to the abstract normative universalism of the standard economic model and of some types of sociology, but no more endorses culturalist relativism. It concentrates on constructing theoretical categories which will make it possible to grasp simultaneously what is and what is not comparable between different historical periods and different politico-economic spaces. RA, from this point of view, considers that the reproduction of societies goes through the construction of macro-economic and macro-social coherences which can be caught thanks to the institutional forms taken by the basic structures of society. The whole set of these forms constitutes the mode of regulation which, consequently, "represents the connection" between the micro-economic level - the "industrial paradigm" of a given era - and the macro-economic level represented by the "regime of accumulation"; the interwoven set of the three levels constituting the "mode of development" (Delorme 1991). The mode of regulation thus defined, is obviously the equivalent of the societal effect insuring the coherence of a given society (of its mode of development).

Despite this kinship, RA is different from SA on several points. Its orientation is more economic than sociological; unlike the SA, it is less interested in societal coherences and social interactions than in macro-economic loops and interdependences between some institutional forms selected from an economic point of view. The conceptualised regularities - accumulation regimes for example - belong to the infra-societal level according to the SA criteria. Moreover, the basic categories put in relation are conceptually further constructed in RA than in SA; at least three of the "canonical" structural forms of RA - the wage-labor nexus, competition and money - which are characterised and articulated at the national level are not the outcome of an inductive partitioning of institutional systems, but a theoretical inheritance of the marxist approach to the capitalist system. What's more, the RA has positioned itself from the beginning in a perspective which is more historical than comparatist, more in the diachrony that in the synchrony, and its recent comparatist developments in terms of compared national trajectories have not been thought out on the theoretical side as much as those of SA. Until now the point for RA was more to juxtapose national studies by using the canonical categories of the approach, following a perspective whose aim was essentially to assess the diversity of institutional forms and social arrangements which could lead to similar economic results or to comparative advantages for some national configurations (Boyer & Saillard (ed.) 1995; Boyer et al. (ed.) forthcoming).

However, the RA's concern is more and more to go further than just juxtapose the institutional forms considered and to take into consideration the contextual effects linked to their interdependences. This can be seen in the various works of C. André, R. Delorme and Y. Saillard on the "State-economy" relations, indirect wages and social protection. These authors have adopted an essentially inductive and cautious approach which is very close to the way followed by supporters of SA who, like Barbier, work in the same field. Unfortunately, this explains also that these works register the same deficit of theorisation of the systems of social protection as systems of societal coherence as in SA.

Comparative advantages of the two approaches

It follows that the differences between SA and RA which are the source of their respective advantages and disadvantages should be better defined and evaluated. In order to do this, let us consider a few elements.

In favour of RA one can consider the usage it makes of categories resulting from a synthesis between the analytical concepts inherited from Marx and the teachings of the "total" and long-range history promoted by the École des Annales. On the one hand, these categories are the constructed totalities such as the regime of accumulation (the "mechanical mode" of closure of the productive system) and the mode of regulation (the coherent set of institutional forms surrounding and stabilising the regime of accumulation). On the other hand, they are the "sub-systems", interdependent but partially autoregulated by specific "regimes", which structure the canonical forms (wage-labor nexus, competition, money, government, modalities of international insertion). At this level, the advantage of RA as opposed to SA is not only that the universalism of the selected national "sub-systems" in interaction is more solidly founded on the theoretical side, which makes the structural forms comparable in a more legitimate way. It is also that the totalities and the coherences which are liable to being compared are not necessarily defined, whatever the historical period, at the national level. The RA is then capable of accounting for the fact that for some national actors, there can be infra-national and supra-national coherences. In the RA framework, there is a place for competition and even contradictions between different "societalisation" modes within the national space, a place for an heterogeneity between the various logics of collective action and for a plurality of forms of articulation of the institutions which regulate them. In this prospect, only the State logic, the institutional system which materialises it, and the social actors which embodies it, are immediately situated at the national level. To the contrary, the national definition of the reproduction regimes of the other structural forms must be seen as the result of struggles and "institutionalised compromises" (Delorme & André 1983) between actors constructed in spaces and driven by logics of action which are a priori different, heterogenous and hierarchised: local or regional space for the wage-labor nexus, supranational zones for the currency and world space for competition and international relations.

Nevertheless, this theoretical advantage of RA makes all the more unsatisfactory the little attention given until very recently to the analysis of the State as a proper system of rationalisation and therefore playing a role in the mode of regulation and societal coherence far beyond the only forms of its relations to economy. This lack of polity in RA has hindered, indeed, any genuine capture of the mode of regulation at a societal level (Boismenu & Drache (ed.) 1990; Théret 1990 and 1992). Another disadvantage of RA is the fact that it does not go further than a descriptive institutionalism, and that the interdependences between the various institutional forms are not theoretically based, but only stated. Indeed, even if the problem of complementarity and of the hierarchy of these forms is now put forward, the mode of regulation still appears somehow like a constellation of juxtaposed institutions rather than like a genuine configuration of interdependences which could be conceptually organised into a hierarchy.

However, one can think that this is a postulate stating the autonomy of each institutional form towards its environment constituted by all the others; this autonomy, in coherence with the principle of comparability of the SA, would assure the pertinence of their inter-national comparisons. If this is the case, we should then see in the mode of regulation a result which is essentially random, and not a "social construct" as is the societal coherence for the supporters of SA who consider this coherence according to a systemic model where the interactions are causal and fully built. On the contrary, RA would implicitly adopt a conception of the social cohesion founded on a communicational or informational model in which the interdependences between the structural forms go through communicational interferences treated in a strictly endogenous manner within each of these structures following their own autopoietical logics (Jessop 1990: 307-337). Nonetheless the absence of theorisation of such communicational forms of interdependence results in the fact that the mode of regulation appears to be either, at best, the result of a coevolution process with no principle of composition and determination of the trajectory of the societal whole, or, at worse, as a simple name given to an empirical gathering whose mode of composition is impossible to think of. In both cases, there is no statement of the mediations which are necessary to think, in the same movement, the diachronic autonomy and the synchronic heteronomy of the various structural forms, i.e. their own regimes and the regulation of the whole.

In short, SA and RA do not differ in regards with the theoretical objectives which they assign to international comparisons. They also adopt similar systemic and relational principles of comparability. However, they are opposed as far as their way of considering the context of the institutional forms and the national cultural entities is concerned. SA is in favour of a global conception of the context where the national culture is considered as the analytical basic unity, whereas RA considers "the context as a set of variables" and the culture like a set of conditions for the elements submitted to analysis (Scheuch 1968: 200). In other words, SA is rather a durkheimian and holist approach which states a solidarity a priori of the social body assimilated to an "emerging totality" (Piaget 1968). RA, on the other hand, is rather a weberian and "holindividualist" approach; its starting point is a set of structures of social action orientated by various types of rationality, given that a priori the very existence of a society is the problem.

This is why RA has the advantage, compared to SA, of proposing forms of societal coherence liable to be compared from one country to another at a meso or infra-national level. However, it has the disadvantage not to take as much consideration to the interactions and mediations between the various social relationships as it does to the institutional forms and the dynamics of these relations. This can be interpreted as the hallmark of a representation of societal coherence which is communicational and symbolic rather than determinist, but this interpretation is not assumed. Indeed, RA has not gone further than to try and describe constellations of institutions without genuinely analysing the modes of correlation or compatibility between them (Peaucelle & Petit 1991; Amable et al. 1995). Therefore regulation, in its french meaning, is more the designation of a set of recurring observations historically situated than a well established concept. From there the paradoxical result that, contrary to SA, RA cannot indeed theoretically base international comparisons at the macro-social level, its processes being in fact meso-economic. Lastly RA and SA have nearly symmetrical advantages and disadvantages.

II. Towards a structural analysis of the institutional forms of societal regulation

It is interesting to try and use this symmetrical deficit in order to bring out what these approaches can contribute to one another, the deficiencies of one filling a gap in the other and vice versa. Don't they reveal two paradigms which respectively attest of two forms of societal coherence which are simultaneously in action? And even if each approach favours one paradigm, they can both have their relevance as long as different levels of analysis are distinguished and if they are then combined together.

Giving RA a societal dimension makes it possible to make better founded international comparisons

Indeed, by combining the functional model to the communicational model, by articulating societal and regulationist approaches, one can make out the existence of structural forms which make sense whatever their national context, and then reach a regulationist-societal conception where macro and meso comparisons are simultaneously relevant. To achieve this, let's consider that, because of the differentiation of the contemporary societies, there are discontinuities, internal to the national social space, between different orders of practices governed by heterogenous logics of action (see Théret 1992 and 1994): the economic order governed by the logic of capital accumulation, the political order governed by the logic of power accumulation, and the domestic order governed by the logic of "human resource" reproduction and which is the source of economic as well as of political accumulation.

The kind of functional coherence which is put forward by SA refers to these immanent logics of orders which, like language according to Saussure, "only know their own order", as they are logics of social action orientated by a specific rationality and channelled by institutions insuring the functional reproduction of each order. The coherence which is implicitly favoured by RA, on the contrary, refers to regulatory relations between the various orders in the national space. This space is the "order of orders" where the mode of regulation becomes meaningful, an order which is essentially communicational, due to the discontinuities which make impossible the interplay of functional relations of systemic integration between the various orders. This distinction between two kind of coherences, one functional, the other communicational, explains the paradox mentioned above. The conception of SA is not valid at the societal level where it has been carried out - at this level, the functionalism of SA is inadequate. It is only valid at the infra-societal level of the structural forms which are brought forward by RA. On the other hand, the conception of RA is also not valid at the level of the economic order where it has until now been carried out - at this level, the empirism of RA is liable to criticism. It is only valid at the societal level as it is envisaged by SA. In other words, there is no mode of regulation of the economic order which would be detectable in a purely economic way, but only regimes of regulation of the various orders and a mode of regulation (or social integration) of the society articulating all these regimes within a historically situated territorial space.

Naturally, such an articulation of the two paradigms of SA and RA has important general implications on the possible development of both approaches. Nevertheless we will only examine here what the consideration of the two kinds of coherences which we have just mentioned, implies as far as international comparisons are concerned: the functional coherence which is linked to the self-closure of orders of social practices and which can be captured "dynamically" through the notion of "regime" (of functioning); the communicational coherence assured by regulatory mediations between the various orders and which makes possible a coevolution of these orders so that the contradictions between their respective logics do not lead to the destruction of society.

In this theoretical framework, international comparisons of systems of infra-societal relations (meso) are possible as soon as their functional independence vis-à-vis all the others is founded within the boundaries of national societies. Macro-systems can also be compared, once remembered that the structural forms which coompose them are heterogenous: several of them are structures of order which develop in the diachrony, others are specific synchronic relations between these structures of orders and articulate their heterogenous logics. These relations of mediation which are structured in autonomous fields insure the conversion-translation of the values from one order to another, and thus allow contradictory orders of practices to hold together in a societal coherence. Even if they adjust from day to day the dynamic evolutions of the various orders, these regulatory structural forms cannot be analysed only through the logics of the orders linked by them. They must be analysed through their proper way of operation corresponding to their communicational function. This strategic "function of regulation" (held, among others, by the professional mediators like jurists, bankers and intellectual of various kinds) consists in reducing the noise which each order produces in the environment of all the others by translating this noise into assimilated information, or by fixing limits and rules for the practices of interference between the orders by the construction of specifically adapted organisations.

In short, according to the proposed synthesis of SA and RA in terms of societal regulation approach, in order to compare societies or sub-systems composing them, we have first to establish that they have reached a similar degree of differentiation and that the same structures of order can be met within them. Only then, can one look for the regimes of order and the regimes of mediation between the orders whose configuration form the mode of societal regulation. These various regimes are internationally comparable and they can even be grouped together in typologies.

Finally, on the methodological side, SA and RA lead to formulation of three hierarchised principles:

- One must not compare elements but relations between these elements and the autonomous systems of these relations;

- One must not carry out the comparison of these systems at the surface of the institutional forms but at a level of abstraction which makes it possible to clarify underlying structures common to these multiple forms;

- One must not only compare these structures according to the modalities of their own historical development but also their synchronic assembly in communicational systems producing societal coherences.

Structuralism as a comparative method

On this basis, we now would like to try and finally demonstrate that the analyses made previously tend to revive structuralism, from the theoretical point of view as well as in terms of comparative methodology, however unfashionable it has become. Indeed, structuralism originally and above all (that is before it became the basis for various philosophies) is "nothing but a comparative method" (Descombes 1979: 105). It is this method, developed first by linguists and then by historians like Georges Dumézil and an anthropologist like Claude Lévi-Strauss, which constitutes "the most innovative and essential contribution" of structuralism (Marc-Lipiansky 1973: 78). The fact that such a tool has been practically completely forgotten in the comparative literature is striking; this points out the hegemony of the Anglo-Saxon positivism in this literature. However, structuralism has had some implicit influence at least on authors whose concern was to carry out methodological criticism, as well as with those who looked for general typologies. Besides, a kind of structuralism is claimed as their own by SA and by RA through its relation to Marxism. But what has been remembered of structuralism by these two approaches is not its methodological character (Piaget 1968), but what would be its holism in order to oppose it to the individualism of orthodox political economy and of behaviourism.

In fact, contrary to this commonly held unfortunate idea, structuralism is not necessarily a holism. It considers of course that "what is preserved from one whole to another within a same cultural era, are the relations between the elements, the structures" and that, as a result, it is necessary to compare the "wholes rather than the elements, and the relations rather than the terms" (Descombes 1979: 105). But deciding that one must "favour the whole as opposed to parties" (Marc-Lipiansky 1973: 89) does not imply in any way a holism as long as such a principle is only applied to partial wholes and not, as structuro-functionalism does, to the society as a whole. For Lévi-Strauss, for example, the totality called "French society, English society or any other" does not have a structure (in singular) but a "certain number of structures" that "one must try and find where they are and not anywhere else" and which are only partial expressions of the social whole (Lévi-Strauss 1958-1974: 103). More generally, for a methodic structuralism free from any functionalism, the society-totality does not have a general structure a priori, because it can only be approached as an "insecure" assembly of partial and various structures. The interpretation of this assembly comes under a communication theory, the structures (closed totalities which reproduce on the basis of a principle of immanence according to self-ruled transformations) being of different nature, and only forming systems through a communicational mode of regulated coevolution. Thus, there are only few differences between the so-called holism of the structural method and the methodological individualism of historico-comparative sociology like that of Weber (see Kalberg 1994: chap.1). In both cases, we have a "holindividualism" which is close to the societal regulation approach which we have put forward above.

Therefore, structuralism must not be reduced to a functionalist systemism. Rather, as a comparative method it allows not only to go beyond the limits of inductivism thanks to its theoretical aim, but also, by bringing forward the idea of a structure to discover, to give relevance to generalisations whose scope goes further than that of the weberian modelisation of ideal types. Indeed, in the method of structural analysis, theorisation goes the same way as the conceptual elaboration of the structure of the analysed objects. This conceptualisation starts by locating "what is preserved in an isomorphism between two wholes" whose disparity is first the most striking feature (Descombes 1979: 105). The comparison, "essential process to structural research", must "make it possible to discover the invariable structure common to objects whose concrete modes of realization can be so different that one would not even think at first of bringing them together. The comparison of such objects can only be carried out after the reduction of their specific character to 'the only stable elements - and always partial - which make it possible to compare and classify', that is to the relations of correlation and opposition expressing their common characteristics and differences" (Marc-Lipiansky 1973: 140). In other words, for the methodic structuralism, "the comparison is not the base for generalisation, it is the other way round" (Lévi-Strauss 1958-1974).

Thus, with its method "which makes it possible to demonstrate that the elements which are apparently disparate are not necessarily so, and that beyond the confusing diversity of the facts which are presented to empirical observation there can be invariable properties which can be differently combined" (Lévi-Strauss 1958-1974: 98), structuralism gives means to construct generalities which found the comparison because they establish the comparability. The method of generalisation consists in going up to the theoretical level of abstraction where the incomparable becomes comparable, where the objects which are being compared are constructed in such a way that there exists a general theory which, by fixing what is common to each of them, makes it possible at the same time to define precisely what their differences are. A structural analysis then allows to clarify systematically these differences, because the forms or structures lying under the multiple institutional contents at the phenomenal level are far fewer than these contents.

But the structural method is more than just a method of construction of the comparability of the objects which it investigates, because it also carries out deductive processes whose heuristic scope goes beyond the Weber-like theorisation by construction of ideal types. More precisely, what is important here is the distinction made by the structural method between three levels of analysis which are articulated and defined permanently, thanks to a round trip movement between the empirical and the theoretical: the level of the structure, that of the models, and that of the social relations as such. Indeed, as far as structuralism is concerned, even though structures are considered as real as the phenomenal forms through which they appears empirically, the "notion of social structure does not relate to empirical reality but to the models constructed from this reality, (…) the social relations 'being nothing but the raw material used for the construction of these models which highlight the social structure itself' (Lévi-Strauss 1958-1974: 331-332). The notion of structure is thus linked to the notion of model, the model constituting a necessary intermediary between empirical reality and the discovery of the structure" (Marc-Lipiansky 1973: 38-39).

However, the model of the methodic structuralism, as an intermediary concept, that is as a symbolic system built from the empirical reality so as to explain its main aspects (Marc-Lipiansky 1973: 42), is not yet different at this level from the weberian ideal type. Isn't this model also a "mental construct for the measurement and characterisation of individual - that is, significant in their uniqueness - contexts", a conceptualisation "patterned orientation of meaningful action" (Kalberg 1994: 84)? It is only once the first stages of the "structural reduction" from which "one reaches the structure" through the mediation of the model that the latter proceeds in a deductive manner from the structure. From then on, "structuralism goes from the structure to the model" (Descombes 1979: 106). Considering now that what is structured is not the thing itself, but the whole from which this thing can be considered as a representation, compared to other wholes", "it reconstructs or reproduces the given phenomena which it offers to analyse" (Descombes 1979: 106).

The structure is now, therefore, the language, the common code which not only founds the comparability of objects put into relation, but it also fixes the rules (vocabulary and grammar) to which the models must conform in order to represent in a comprehensive and explicative manner the observed social relations and thus "to restitute their empirical content" (Marc-Lipiansky 1973: 132). The structure has become the starting point of analysis and the models are now deducted from it by logical operations of transformation. From now on, they are logically constructed tools which must make it possible to capture by isomorphism the compared phenomena. Two logics are mobilised for this purpose: first the static binary logic which uses polar oppositions between some elements of the structure, and the dialectical logic in dynamics, which aims essentially at deducting the terms of integration or mediation within itself (Marc-Lipiansky 1973: 109-125).

What separates the model of methodic structuralism from the weberian ideal type, then, is the process of their respective genesis which, in one case, gives importance to logical deduction whereas, in the other, an empirical anchoring remains privileged. The heuristical superiority of structuralism over the historico-comparative sociology of weberian type comes from the fact that one includes the other by adding to it an "economical" method in order to derive logically from a simple structure different types of models, making it possible to explain the variety of historical and spatial forms of social relations in a same domain. This superiority is also due to the fact that by researching isomorphisms, the structural method helps find simple structures in various related fields, making easier in this way the modelisation of these fields. One last important point to make about structuralism (important because it predisposes it to be a useful method for the societal regulation approach) is that it concentrates to a great extent on communicational rationality starting with the idea that society is a set of symbolic systems. The methodic structuralism, in order to be a semiology of the "social", emphasizes a kind of non-deterministic causality in a way which is particularly well adapted, in our opinion, to capture the regulatory mediations between the "economic" and the "political" through which the societal regulation passes.

In fact, all these properties of the structural method fit perfectly well with the theoretical and methodological requirements which we have discussed in the first two parts of this paper. Unfortunately we do not have enough space here to demonstrate in a concrete way the productivity of this method, especially to grasp the various forms of the welfare-State and its various places in the societal regulation in different countries. Indeed, the value-added of the structuralist comparative method is obvious when applied to the national systems of social protection. It does not only make it possible to get the results of the "classical" typological approaches, it also leads to two important new outcomes. First of all, it allows to locate an ideal type which is absent from the literature and which can be exemplified by the Japanese case. Second, it leads to construct a normed space of measurement in which it is possible to draw the historical trajectories of the various national configurations of the welfare-State by relocating them, one against another. Moreover, the structural method seems to be applicable in the same way as for the welfare-State to every social sub-system which incorporates the domestic sphere in a closed structure of institutional mediations between the political and theeconomic orders, using this incorporation as the basis of its auto-development. But as far as this last point is concerned the Anglo-Saxon reader will have to give us some credit.

REFERENCES

Amable, B., R. Barré, and R. Boyer (1997), Les systèmes d'innovation à l'ère de la globalisation, Paris: Economica.

André, C., and R. Delorme (1989), Les relations État-économie en France et en République Fédérale d'Allemagne: une comparaison, Couverture orange n° 8914, Paris: CEPREMAP.

André, C., and R. Delorme (1990), Les interventions publiques au Royaume Uni et en France, Paris: CEPREMAP.

André, C., and Y. Saillard (1992), La transformation des systèmes de retraite dans les douze pays de la Communauté européenne, Solidarité-Santé. Études statistiques (3): 53-66.

Armer, M., and A.D. Grimshaw (ed.) (1973), Comparative Social Research: Methodological Problems and Strategies, New York, Londres: Wiley.

Ashford, D. (1978), The Structural Analysis of Policy or Institutions Really Do Matter, in: D. Ashford (ed.), Comparing Public Policies. New Concepts and Methods, 81-98. London: Sage.

Barbier, J.-C. (1990), Comment comparer les politiques familiales en Europe : quelques problèmes de méthode, International Social Security Review (3): 342-357.

Benoît Guilbot, O. (1987), Les structures sociales du chômage en France et en Grande Bretagne, influences sociétales, Sociologie du Travail XXIX(2): 219-236.

Benoît Guilbot, O. (1989), Quelques réflexions sur l'analyse sociétale: l'exemple des régulations du marché du travail en France et en Grande Bretagne, Sociologie du Travail XXXI(2): 217-225.

Boismenu, G., and D. Drache (ed.) (1990), Politique et Régulation. Modèle de développement et trajectoire canadienne. Montréal, Paris: Méridien and L'Harmattan.

Boyer, R., and E. Caroli (1993), Changement de paradigme productif et rapport éducatif: performances de croissance comparées France-Allemagne. Paris: CEPREMAP.

Boyer, R., and E. Caroli (1994), Production Regimes, Education and Training Systems: from complementarity to mismatch?, in: C. Buechtemann (ed.), Human Capital and Economic Performance, New York: RAND Corporation.

Boyer, R., and Y. Saillard (ed.) (1995), La théorie de la régulation: l'état des savoirs, Paris: La Découverte.

Boyer, R., L. Miotti, and C. Quenan (ed.) (forthcoming), Trajectoires nationales et régulation en Amérique latine, Paris: Éditions de la Maison des Sciences de l'Homme.

Caroli, E. (1993), Les interactions entre système éducatif et système productif : analyse historique, théorie et formalisation dans le cas de la France et de l'Allemagne, Report CEPREMAP for the Direction de l'évaluation et de la prospective, Ministère de l'Education Nationale. Paris.

Delorme, R. (1991), État et hétérogénéité. ERIC et le MPEE, Cahiers de recherche sociologique (17): 153-184.

Delorme, R., and C. André (1983), L'État et l'économie, Paris: Seuil.

Descombes, V. (1979), Le même et l'autre. Quarante-cinq ans de philosophie française, Paris: Minuit.

Hantrais, L., and M.-T. Letablier (1994), Construction et déconstruction de la famille en Europe : une analyse comparative, Recherches et Prévisions (37): 1-10.

Hilliard, B. (1994), Sociological Conceptions of the Family, in: Hantrais, L., and M. T. Letablier (ed.), Conceptualising the Family, 58-60. Loughborough, Paris: Loughborough University Press - CNAF.

Jessop, R. (1990), State Theory. Putting Capitalist States in their Place, Cambridge, UK: Polity Press.

Kalberg, S. (1994), Max Weber's Comparative-Historical Sociology, Chicago: University of Chicago Press.

Köbben, A.J.F. (1968), The logic of cross-cultural analysis: why exceptions?, in: Rokkan, S. (ed.), Comparative Research across Cultures and Nations, 17-53. Paris, The Hague: Mouton.

Leach, E. (1970), Lévi-Strauss, Londres: Fontana.

Lévi-Strauss, C. (1958-1974), Anthropologie structurale. Paris: Plon - Agora.

Marc-Lipiansky, M. (1973), Le structuralisme de Lévi-Strauss, Paris: Payot.

Maurice, M. (1989), Méthode comparative et analyse sociétale. Les implications théoriques des comparaisons internationales, Sociologie du Travail XXXI(2): 175-192.

Maurice, M. (1994), Acteurs, règles et contextes. A propos des formes de la régulation sociale et de leur mode de généralisation, Revue française de sociologie XXXV: 645-658

Maurice, M., F. Sellier, and J.-J. Silvestre (1979a), La production de la hiérarchie dans l'entreprise: recherche d'un effet sociétal, Revue française de sociologie XX(2): 331-365.

Maurice, M., F. Sellier, and J.-J. Silvestre (1979b), Priorité à la régulation conjointe ou aux rapports sociaux ?, Revue française de sociologie XX(2): 377-380.

Mishra, R. (1986), Social Analysis and the Welfare State: Retrospect and Prospect, in: Øyen, E. (ed.), Comparing Welfare States and Their Futures, 20-31. Aldershot: Gower.

Øyen, E. (1986), Identifying the Future of Welfare State, in: Øyen, E. (ed.), Comparing Welfare States and Their Futures, 1-19. Aldershot: Gower.

Peaucelle, I., and P. Petit (1991), La gestion du rapport salarial. Une modélisation du rôle des incitations et des institutions, Revue Économique 42 (3): 493-519.

Piaget, J. (1968), Le structuralisme. Paris: PUF.

Reynaud, J.-D. (1979), Conflit et régulation sociale. Esquisse d'une théorie de la régulation conjointe, Revue française de sociologie XX(2): 367-376.

Saillard, Y. (1993), La transformation des systèmes nationaux de protection contre le risque maladie dans leur contexte économique: quelques enseignements d'une comparaison internationale, in: CEBS (ed.), Économie sociale et espace européen, 233-252. Nantes: CEBS - Association d'Économie sociale.

Scheuch, E.K. (1968), The cross-cultural use of sample surveys: problems of comparability, in: Rokkan, S. (ed.), Comparative Research across Cultures and Nations, 176-209. Paris, The Hague: Mouton.

Scheuch, E.K. (1990), The Development of Comparative Research: Towards Causal Explanations, in: Øyen, E. (ed.), Comparative Methodology. Theory and Practice in International Social Research, 19-37. London: Sage.

Swanson, G.E. (1971), Framework for Comparative Research: Structural Anthropology and The Theory of Action, in: Vallier, I. (ed.), Comparative Methods in Sociology. Essays on Trends and Applications, 141-202. Berkeley, Los Angeles: University of California Press.

Teune, H. (1990), Comparing Countries: Lessons Learned, in: Øyen, E. (ed.), Comparative Methodology. Theory and Practice in International Social Research, 38-62. London: Sage.

Théret, B. (1990), La place de l'État dans les théories économiques françaises de la régulation: éléments critiques et nouvelle position à la lumière de l'histoire, Économie Appliquée XLIII(2): 43-81.

Théret, B. (1992), Régimes économiques de l'ordre politique. Esquisse d'une théorie régulationniste des limites de l'État., Paris: PUF.

Théret, B. (1994), To have or to be. On the problem of the interaction between State and economy and its solidarist mode of regulation, Economy and Society 23(1): 1-45.

Théret, B. (1997), Méthodologie des comparaisons internationales, approches de l'effet sociétal et de la régulation: fondements pour une lecture structuraliste des systèmes nationaux de protection sociale, L'Année de la régulation, I: 163-228.

Théret, B. (1998), Politica economica e regulação : uma abordagem topologica e autopoietica, in: B. Théret and J.C. Braga (ed.), Regulação Economica e globalização. São Paulo: UNICAMP.

Retour en haut