Beyond the antinomies of structure: Levi-Strauss, Giddens, Bourdieu, and Sewell
- 1.3k Downloads
In this article, I attempt to address some enduring problems in formulation and practical use of the notion of structure in contemporary social science. I begin by revisiting the question of the fidelity of Anthony Giddens’ appropriation of the idea of structure with respect to Levi-Strauss. This requires a reconsideration of Levi-Strauss’ original conceptualization of “social structure” which I argue is a sort of “methodological structuralism” that stands sharply opposed to Giddens’ ontological reconceptualization of the notion. I go on to show that Bourdieu’s contemporaneous critique of Levi-Strauss is best understood as an attempt to recover rather than reject the central implication of Levi-Strauss’ methodological structuralism, which puts Bourdieu and Giddens on clearly distinct camps in terms of their approach toward the idea of structure. To demonstrate the—insurmountable—conceptual difficulties inherent in the ontological approach, I proceed by critically examining what I consider to be the most influential attempt to resolve the ambiguities in Giddens structuration theory: Sewell’s argument for the “duality of structure.” I show that by retaining Giddens’ ontological focus, Sewell ends up with a notion of structure that is at its very core “anti-structuralist” or only structuralist in a weak sense. I close by considering the implications of the analysis for the possibility of developing the rather neglected “methodological structuralist” legacy in contemporary social analysis.
KeywordsStructure Structuralism Giddens Bourdieu Levi-Strauss Social structure Ontology Methodology
The notion of structure and the associated concept of “social structure” have been widely acknowledged since the origins of sociology as a scientific discipline to be an indispensable analytic tool. The problem of structure was salient in the minds of the early twentieth-century founders of American sociology who noted with despair how “the notion has been overworked, abused, distorted, misrepresented, and misunderstood.” Yet “everyone who has attempted to interpret men’s activities has been obliged to use the concept in some generic form” (Small 1902, p. 227). In spite of these initial attempts to clarify matters, four decades later one of the major figures of twentieth-century American anthropology was still justified in being wryly skeptical about the usefulness of the idea of structure in the social sciences: “[s]tructure sometimes has its common-sense meaning, as when we speak of the structure of a canoe. Sometimes it emphasizes form; sometimes organization; as in the term ‘social structure’ which is tending to replace ‘social organization’ without appearing to add either content or emphasis of meaning” (Kroeber 1943, p. 105).1
It was not until the immediate post-functionalist period that theories of structure came to form a seemingly integral part of the sociological toolkit. This happened at the same time as the notion of “structure” came to overshadow the idea of “function” in social theory (e.g., Giddens 1979, 1984). One reason for this rather belated ascendance of the concept of structure might have to do—as Sewell (2005b) has noted—with the notorious analytical slipperiness of the concept.2 The ascendance of the concept of structure in the social sciences can be traced to Levi-Strauss’s (1963e, 1966, 1969, 1985) pioneering importation of models from mathematics and linguistics (Cassedy 1988) for the understanding of kinship and myth in anthropology. Following Levi-Strauss’s intervention, the idea of structure went from being an idle terminological stand-in for the older organicist notion of “social organization” in the Durkheim-Schaffle-Spencer line to a productive source of formal representations of social reality (Giddens 1979). This opened the doors for structuralism, a cross-disciplinary intellectual movement that swept the social and human sciences during the 1950s and 1960s (Clarke 1978; Dosse 1997; Kurzweil 1980), to make its way into the attention-space of Anglo-American social theorists. As was noted by Clarke (1978, p. 405) towards the tail end of this period of creative ferment: “even though the concept of ‘structure’ is one of the oldest in the social sciences, it is only in the last 20 years or so that a school of thought known as ‘structuralism’ has grown up.”
One of the main reasons for the appeal of structuralism to social scientists was the belief that this theoretical framework opens the possibility finally to break with the “substantialism” of 19th century sociological organicism (Giddens 1984, p. 16). Structuralism does this by promising to provide the conceptual leverage to conceive of how the value and identity of particular elements in a given social and cultural context derives not from their inherent properties but from their position in an implicit (or “deep”) relational system (Benveniste 1971, p. 80). Mary Douglas (1968, p. 364) best summarizes the advantages of the structuralist approach: “[s]tructural analysis does not work by reducing all symbols to one or two of their number; rather, it requires an abstract statement of the patterned relations of all the symbols to one another.” This feature of structuralism was taken by sociologically oriented theorists as helping to transcend reductionist analyses from biology, psychology, and some versions of “materialist” ecology in the study of social and cultural phenomena.
In an Anglophone sociology dominated by structural-functionalism (and reactions against it) the Levi-Straussian appropriation of the linguistic-mathematical conceptualization of structure remained largely dormant until—primarily due to the work of Anthony Giddens (1979, 1984)—it was rehabilitated and deployed with the goal of helping to transcend some of the most difficult conceptual issues inherited from the mid-century structural-functionalist appropriation of the Durkheimian and Weberian legacies. Giddens’s swift success in convincing English speaking sociologists that a workable conception of structure was indispensable for effective sociological analysis is itself an unwritten chapter in the history of post-classical social theory. Three principal “internalist” reasons can be ascertained as having a lot to do with this development.
First, Giddens’s rhetorical tracing of a relatively (and deceivingly) straightforward line of intellectual descent of the (sociological) concept of structure from the classic nineteenth-century organicists such as Spencer and onward through Parsons and the “structural-functionalist school.” Second, his ambitious and ultimately successful bid to establish rhetorically the claim of being able to transcend the intellectual dead ends of this older line of thinking by turning to what at the time were relatively obscure conceptual resources in French anthropology and European linguistics (“structuralism”). Giddens’s final master stroke was his convincing demonstration of the relevance of this “new” (i.e., non-functionalist) form of structuralist theorizing to the perennial problems of classical sociological theory (i.e., those addressed by Marx, Weber and Durkheim): the mechanisms of social reproduction, the bases of order and power in contemporary industrial society, and the classic problematic of the sources of social integration in post-traditional societies. This was done in his classic monograph The Constitution of Society (1984) and his earlier study Central Problems in Social Theory (1979). In fact, it can be said without much danger of exaggeration that just like—as remarked by Whitehead—much of modern philosophy is but a footnote to Plato, the most widely accepted characterizations of the concept of structure in contemporary American sociology can all be seen as elaboration and commentary of Giddens’s influential synthesis of Levi-Straussian structuralism, Ericksonian ego-psychology and Garfinkel’s ethnomethodology developed in the late 1970s and early 1980s (Stones 2005).3 This is attested by the fact that most commentary on the notion of structure in England (e.g., Archer 1982) or the United States (e.g., Sewell 2005b) takes as its point of departure Giddens’s synthesis (if only to radically revise it or repudiate it in favor of an alternative formulation).
As a result of Giddens’s intervention the idea of structure went from being an unwieldy and unevenly used abstraction to a seemingly obligatory part of the vocabulary of contemporary sociologists. The validity of this latter claim is attested by the fact that empirical and theoretical works hailing from a remarkable range of methodological and substantive persuasions as well as subfields of study—e.g., criminology, social psychology, historical sociology, sociology of organizations, sociology of religion, social movements, political sociology—in contemporary sociology take Giddens’s conceptualization of structure—or the related notion of “structuration”—as the centerpiece of their theoretical orientation and empirical contribution (see for instance, Beisel 1993; Beisel and Kay 2004; Blair-Loy 1999; Desai 2002; Gerteis 2002; Heimer 1997; Logan 1996; Neuhouser 1998; Pedriana and Stryker 1994; Ridgeway et al. 1998; Sherkat 1998; Sherkat and Ellison 1997; Smilde 2005; Stryker 1994; Young 2002).
This established consensus on the centrality and indispensability of a theory of structure for contemporary sociological analysis can be thought of as a resounding success. In a field that is constantly threatened with constant disintegration and parcelization assent regarding the importance of some notion of structure for social-scientific analysis (along with the ability to interpret tables of multiple regression coefficients) is one of the few pieces of shared culture that serve to unite the field around a mutually recognizable stock of common knowledge.
This is not to say that there are no other sources of “sociological structuralism” that cannot be traced to Giddens. In fact, another set of contemporaneously developed—vis à vis Giddens—conceptualizations of structure in American sociology comes from network analysis (Emirbayer and Goodwin 1994). Network theorists trace their own conceptual debt to precisely the anthropological source that was rejected by Giddens and criticized by Levi-Strauss: Radcliffe-Brown and the structural-functional school of anthropology. This is the “British” strand of structuralism—Fararo and Butts (1999, p. 4) refer to it as “network structuralism”—that Maryanski and Turner (1991) distinguish from Levi-Strauss’s “French” structuralism.
Consistent with Radcliffe-Brown’s formulation network structuralists are more likely to consider structure as real, observable, and concrete patterns of social relations.4 These patterns are “tangible” as opposed to merely “ideal” (Bearman 1997, pp. 1384–1385). Social structure is the “whole network of social relations” (Levi-Strauss 1963c, p. 303; Blau 1974, p. 615). The network notion of structure however does not appear to have become as influential—as a high-level theoretical conceptualization—as that featured in structuration theory in contemporary sociology. This is due primarily to the apparent inability of the former to accommodate issues of change and agency along with the theoretically impoverished role of culture in these efforts (Emirbayer and Goodwin 1994). It was the network-structuralist notion that Giddens summarily dismissed for being unable to take into account agent-level concepts and the “discriminations employed by actors” (Giddens 1984, p. 211). This is the main reason why the Giddens-Sewell account of structuration dominates contemporary research and theory.5 This has led to a gradual attempt by network structuralists to shift to a conception of structure that is more “cultural” than concrete in recent formulations (e.g., White 1992).
This tension between the “concrete” (Radcliffe-Brownian) and the “abstract” (Levi-Straussian) notions of structure is certainly not new but has been a persistent feature of structuralist theorizing in the social sciences at least since Levi-Strauss’ fateful intervention in the 1940s and 1950s. Levi-Strauss was a harsh critic of the Radcliffe-Brown’s “concrete (“empiricist”) notion of structure”. He noted with despair that “[a]s a matter of fact, social structure appears in his work to be nothing else that the whole network of social relations” (Levi-Strauss 1963a: 303). But from a linguistic-mathematical structuralist perspective, what is concrete and observable “structure” may simply be the surface manifestation of a more fundamental (and “deeper”) non-concrete, not (directly) observable bundle of structural principles (Glucksmann 1974, pp. 38–40). Thus, contra-Radcliffe-Brown (and in a sense, contra contemporary network structuralism) for Levi-Strauss it is certainly questionable. “whether such dyadic relations are the materials out of which social structure is built.” Instead it is more reasonable to suppose that they “themselves result from a pre-existing structure which should be defined in more complex terms. Structural linguistics has a lot to teach in this respect” (304).6
Those theorists who have turned to approaches to the problem of structure in the Levi-Straussian lineage—such as Giddens (1979, 1984) and Sewell (2005b)—or toward formal methods of cultural interpretation inspired by structuralism (e.g., Mohr and Duquenne 1997) have largely ignored or glossed over the tension that exists between these two sets of “broad structural” perspectives in relation to the still influential “organicist” notion of structure in the narrow sense that contemporary network theory inherited from pre-war British anthropology. To his credit, Giddens (1979, pp. 22, 59–60, 62) was very sensitive to the conceptual tension that existed between these two dominant senses of the concept of structure even though his proposed terminological solution might have resulted in more confusion than enlightenment (Thompson 1989, p. 62, 68). In an attempt to deal with this issue, Giddens designates the realm of concrete structure understood as “...the ‘patterning’ of relationships between individuals or collectivities” as belonging to the category of system (Giddens 1981, p. 169). The term structure, on the other hand, is to be used exclusively to designate those ordered aspects of social systems that only have a “virtual” and not a concrete existence (1979, p. 3, 64; 1984, p. 25, italics added). This theoretical move allows for both competing notions—abstract and concrete; Levi-Straussian and Radcliffe-Brownian—to do explanatory and conceptual work in his structuration theory. This accounts more than anything else for the sense of incoherence that the reader is sometimes left with after grappling with Giddens’s treatment of these issues (Archer 1982; Thompson 1989, pp. 63–64; Sewell 2005b).
To make matters even more complicated recent network theory has attempted to shift from an objectivist, Radcliffe-Brownian conceptualization of structure as concrete patterns of social relations to a more Levi-Straussian (or “generative”) framework (e.g., Fararo and Butts 1999) without necessarily resolving the deep incompatibilities that exist between the “virtual” and “objectivist” models of structure. Recent reconsiderations of the network paradigm in term of “relationism” have also attempted to ease over these tensions (e.g., Emirbayer 1997) but they have only succeeded in underscoring the tension. The reason for this is that they have proceeded by attempting to graft a linguistic (derived from Saussure) structuralist notion of how elements attain their identity in relation to the entire system of other elements to the original empiricist notions of social structure as observable and concrete “relational patterns” inspired by Radcliffe-Brown and still adhered to by the majority of network theorists.
By forging this line of division along two competing notions of structure, contemporary Anglophone sociologists have managed to reconstruct spontaneously (but not dissolve) the same theoretical partition that separated Levi-Strauss’s structuralism from the variant of the approach advocated by Radcliffe-Brown in early twentieth-century anthropological theory (Levi-Strauss 1963c). The former approach uses the notion of structure in a highly generalized way, and as potentially applicable to various types of systems regardless of its ultimate constituents (social, material, symbolic). This standpoint is therefore agnostic (or “pluralistic”) as to what the actual “stuff” (ontology) that the theorist assumes social reality ultimately to consist of. The latter variant of structuralism is more strictly concerned with developing a workable notion of social structure in a more narrow sense with a specific ontological and empirical referent (e.g., “tangible” social relations). The fact that this division between structuralism in a broad sense and structuralism in a narrow sense lingers in contemporary sociological theorizing is an indication that the “problem of structure,” that is the problem of its proper conceptualization and its role in furthering sociological theory and research, has yet to be resolved in a satisfactory manner (Sewell 2005b).
Outline of the argument
In this article, I address and show possible pathways out of these enduring problems in the formulation and deployment of the notion of structure in contemporary sociological theory. The argument proceeds in four major parts. First, I revisit the question of the fidelity of Giddens’s own appropriation of the idea of structure with respect to the original Levi-Strauss original development of the structure concept. This necessitates a return to Levi-Strauss’s conceptualization of “social structure.” Second, I show how Giddens’s own highly selective exegesis of Levi-Strauss ended up denying the latter’s most valuable contribution to structural analysis, which I label his methodological structuralism. This approach is in many ways strictly antithetical (both substantively and epistemologically) to Giddens’s own “ontological” approach. Third, I review what I consider the most successful attempt to recover the Levi-Straussian legacy of methodological structuralism from within the “French” tradition: Bourdieu’s critique of Levi-Strauss’ “objectivism” in The Logic of Practice. Bourdieu’s critique of Levi-Strauss has been largely misunderstood by being interpreted as an undiscriminating broadside against “positivist social science,” as this term is usually construed when departing from the original Durkheimian problematic (e.g., Giddens 1976). Instead, Bourdieu’s critique of Levi-Straussian structuralism centers on a specific slippage in the latter’s original formulation of the status lay “cognitive models” in which Levi-Strauss breaks with his own methodological structuralism and falls into an ontological trap very similar to what mars Giddens’s structuration theory. Bourdieu thus attempts in this away to recover rather than reject the central implication of Levi-Strauss’s methodological structuralism, which puts him in a camp that is clearly distinct from Giddens, in spite of the current widespread beliefs that both Bourdieu and Giddens are “structuration” theorists. Fourth, to demonstrate the—what I believe to be insurmountable—conceptual difficulties inherent in the ontological approach, I go on to examine critically what I consider to be the most influential attempt to resolve the ambiguities in Giddens structuration theory (while incorporating insights from Bourdieu’s own conceptualization): Sewell’s (2005b) argument for the “duality of structure.” I argue that Sewell’s reconstruction of structuration theory remains one-sided in taking as given Giddens’s ontological appropriation of structuralism. I show that by retaining this ontological focus, Sewell ends up with a notion of structure that is at its very core “anti-structuralist” or only structuralist in a weak sense. I close by considering the implications of the analysis for the possibility of developing the rather neglected “methodological structuralist” legacy as a mature and viable analytical strategy in empirical social science.
Levi-Strauss’s “methodological structuralism”
One problem with ascertaining Levi-Strauss’s position regarding the notion of structure is that he never wrote a single theoretical magnum opus in which the essence of his approach is distilled. Instead his ideas regarding structural analysis in the social sciences are scattered through a variety of short articles and position pieces collected in the two volumes of Structural Anthropology (Levi--Strauss 1963g, 1976c) and the View from Afar (1985) as well as the theoretical introductions to his various books on totemism, myth, and kinship (1966, 1969). In the following I draw on several of these essays (i.e., Levi-Strauss 1963a, b, c, d, f, 1976a, b, 1985), but I concentrate on one-particular piece in which Levi-Strauss makes a concerted effort to delineate clearly the essence of the structural approach as he conceived of it: the essay on “Social Structure” in the first volume of Structural Anthropology (Levi-Strauss 1963c).
From the usual popular expositions of Levi-Straussian structuralism it is easy to get the impression that he subscribed to the view that “structures” (or “structure” in the singular) were the primary constituent of social reality or—as usually maintained by those who interpret Levi-Strauss through the (misleading) lens of the materialism versus idealism dichotomy—i.e., structures are the primary component of the “cultural system” as conceived in the Parsons-Geertz sense (Geertz 1973; Parsons 1973). From this perspective, the main contribution of the structuralist movement usually associated with his name was to highlight the crucial importance of structure as the ultimate “building block” of the social and cultural orders. What these interpretations of Levi-Strauss have in common is that they all presume that Levi-Strauss held an “ontological” view of the notion of structure whereby structures come to be conceived as the primary “substance” of the social in a manner similar to the older organicist conception of structure held by Radcliffe-Brown and the functionalist school but generalized to reality in its (ideationally conceived) totality. This received picture of Levi-Strauss’ theoretical views—summarily wrong on all counts—has probably been the source of the majority of conceptual confusions and theoretical misunderstandings regarding the Levi-Straussian project in particular and the notion of structure in general that continue to plague contemporary sociological theory. This includes the widespread misconception of Levi-Strauss is an “idealist” in spite of his own unabashed (monistic) materialism.
According to Levi-Strauss (1963c, p. 279) when considering the idea of social structure, “there is a point which should be cleared up immediately.” The notion of (social) structure “has nothing to do with empirical reality but with models which are built up after it.” In this respect, “[i]t should be kept in mind that what makes social structure studies valuable is that structures are models, the formal properties of which can be compared independently of their elements.” Glucksmann is correct when she notes that when “Levi-Strauss uses ‘model’ here he means structure, or representation of structure.... The structure accounts for the phenomena as observed by showing their connexion [sic] with hitherto unrevealed relationships” (Glucksmann 1974, p. 154). Levi-Strauss goes on to contrast the (Radcliffe-Brownian) idea of social relations as the “raw materials out of which the models making up the social structure are built” to his own conception of (social) structure as primarily of models constructed to account for the abstract patterning of this raw material. Thus, from his point of view structure “can by no means, be reduced to the ensemble of the social relations to be described in a given society.” For Levi-Strauss, structures are primarily methodological devices that have a purely pragmatic value in social scientific research; they are therefore ontologically neutral. The job of the structural analyst is thus “... to recognize and isolate levels of reality which have strategic value from his point of view, namely, which admit of representation as models, whatever their type” (284, italics added).
This is a fairly strong and clear statement of the Levi-Straussian approach. In contrast to the usual view, Levi-Strauss is staunchly against any “ontologization” of the notion of structure whether this ontological raw material is social relations (the Radcliffe-Brown approach) or anything else (cultural symbols, or “ideas”). Consequently, any interpretation of Levi-Strauss that departs from the post-Marxian coordinates of the materialism versus idealism problematic immediately finds itself going down a hopeless pathway. Levi-Strauss is also agnostic as to the particular “level” or slice of reality that may be represented using a structural model (invalidating any connection between Levi-Straussian structuralism and the ontology-inspired micro-macro debate). Because of this reticence to ontologize the notion of structure, the problematic of “duality,” so central in Giddens’s (1979, 1984) ontological approach, appear to be meaningless from the Levi-Straussian perspective. Structures are models of reality, and as such they cannot be “dual;” they are neither material nor ideal, neither objective nor subjective. Levi-Strauss concludes that, “Therefore, [the study of] social structure cannot claim a field of its own among others...” in the social sciences. “It is rather a method to be applied to any kind of social studies, similar to the structural analysis current in other disciplines” (284, italics added).
This methodological structuralist stance has a series of important repercussions, among which is radically dissociating questions of agency as a presumed capacity of the social actor (or questions of “materialism” versus “idealism”) from questions of structure.7 These two elements are therefore not mutually implicative in Levi-Strauss’s formulation as they are in Giddens’s appropriation, because structure refers not to some sort of substantive constraint (or liberating “medium of” action) existing in the world, but to a methodological device designed to represent the situated productions of the social agent. In Levi-Strauss’s formulation actors do not act through structures nor is structure (necessarily) “medium and outcome” of agentic action. In this respect, methodological structuralism is agnostic as to the question of the relationship between structure and agency since this is a question regarding the possible sources of a given social product (i.e., organizational discourses, patterns of affiliation across groups, mythological discourse) which may or may not have been generated by agentic action, since really-existing structural patterns—such as the spatial layout of an urban complex or an aboriginal village (Levi-Strauss 1963a)—could have been produced as the result of large-scale coercive efforts or even pure demographic or physiological imperatives (Collins 1981). For Levi-Strauss (1976a: p. 79, 1985) structures are meant to be primarily model-based reconstructions of some empirical domain whose primary purpose is to serve as an aid with which to engage in quasi-experimental manipulation.
The reason why Levi-Strauss thought of structures as important to the social sciences was due to his belief that structural representations (or “models”) provide the best way to simulate experimental setups in non-experimental (observational and historical) sciences. They are thus the non-experimental-social-scientist’s best bet in reproducing a semblance not only of theoretical representation but also of practical intervention in Ian Hacking’s (1983) sense. Structures are tools aimed at the simulation of practices and cognitive operations that could be thought to account for the patterns observed in carefully collected empirical materials. If through the manipulation of a (relatively simple) structural model the observed diversity of empirical facts could be generated (or some unobserved empirical facts could be predicted to exist) then the structure had served its explanatory purpose. It is in this way that Levi-Strauss differentiated observation that is the recording of empirical facts from the world and experimentation that consists in the manipulation of structural models “built-up” by the analyst with the aim of reconstructing the (presumably) unobservable mechanisms that presumably generated those observed social facts (Levi-Strauss 1963c, p. 280).
There is one way, however, in which structures could be manifested in reality (and this is the closest that Levi-Strauss comes to postulating structures as an ontological constituent of the social world): when structural models are manifested as regularly ordered set of cognitive schemes in a given group then the structural model that the analyst constructs (the “model of” reality) and the “model for” reality used by the members of the group in the production and reproduction of their everyday routines may coincide. For Levi-Strauss these “native” structural models may or may not be conscious (may be part of discursive consciousness or practical consciousness in Giddens’s terms). Furthermore, the conscious models that the lay actor may think are responsible for a given set of social practices may not be equivalent to the ones that the analyst constructs to account for those practices. For Levi-Strauss, it is both possible that the “true” structural model may be constructed by the analyst and the lay actors may not be aware of the model that is responsible for their social productions (myths, ritual, marriage practices, etc.).
Conversely, Levi-Strauss allows for the possibility that lay actors may have already developed accurate conscious representations of the models that serve to guide their behavior and which are more accurate than any that could be developed by an outside observer. Levi-Strauss notes that “... many ‘primitive’ cultures have built models of their marriage regulations which are much more to the point than models built by professional anthropologists.” For this reason, it is impossible to dispense with the study of a given culture’s “home-made models,” since these models “.... might prove to be accurate or, at least, provide some insight into the structure” of the phenomena to be explained. Furthermore, even if the local models are erroneous from the point of view of the observer, these biases and errors in the lay person’s perceptions may provide important clues to the actual features of the mental models responsible for a given set of social practices. This sounds like a direct ontologization of structures into the cognitive schemata of lay actors and indeed it is. As we will see below, this slippage in the original methodological structuralist project becomes the primary focus of Bourdieu’s (1990a) criticism of Levi-Strauss. Levi-Strauss (1963c, p. 282), to his credit, remained somewhat ambiguous on this point, but never quite resolved the issue.
The closest that Levi-Strauss came to examining this inconsistency in his deployment of the methodological structuralist strategy consisted in noting the importance of separating those “culturally produced models” characteristic of the members of a given social group or historical era from the model structures constructed by the observer. The analyst must not forget that the cultural norms and discursive justifications for practice “are not of themselves structures” (invalidating the usual misinterpretation of Levi-Strauss as equating structure with Weberian “ideas”). Instead, “... they furnish an important contribution to an understanding of the structures, either as factual documents of as theoretical contributions similar to those of the anthropologists himself.” In this way Levi-Strauss—like Piaget (1970)—analytically separates structures as convenient methodological reconstructions of the cognitive models of the social and natural worlds deployed by the social agent from the actually existing mental models held by lay agents. This distinction is important, since as we will see below, Sewell follows Giddens’s ontological path and comes to make those schemata actually responsible for practice—cognitively stored as “memory traces”—the main constituents of “structure.” A methodological structuralist approach in contrast would consider those schemas as the explanandum and only useful insofar as they aid the analyst in the generation of structural representations of those very same schemas.
Giddens’s “ontological structuralism”
Most commentators view the theory of structuration developed by Anthony Giddens in a series of essays throughout the 1970s and 1980s (Giddens 1976, 1979, 1981, 1984) as an attempt to solve a difficult epistemological and methodological problem in the conceptualization of individual action in the face of social and institutional constraints: how can the analyst consider the social actor to be acting with agentic freedom (“the capacity to do otherwise”), while at the same time taking note of large scale regularities and constraints that impinge on social behavior? While it is undeniable that this problem was an important impetus behind the theory of structuration, for Anthony Giddens the primary problem that the theory of structuration was designed to solve was not methodological or epistemological but ontological. As Giddens (1984, p. xx) notes, while the former concerns are important, “concentration upon epistemological issues draws attention away from the more ‘ontological’ concerns of social theory, and it is these upon which structuration theory primarily concentrates.” For Giddens the main concern of social theory should be with “reworking conceptions of human being and human doing, social reproduction and social transformation.”
Thus rather than spending time thinking of better ways to study processes of transformation, reproduction, and everyday human action, social theory should instead concern itself with answering “what is” rather than “how to explain” questions. This is directly opposed to the Levi-Straussian approach, which was focused on structure as explanatory devices with a fundamental epistemological status, and is widely acknowledged even by sympathetic commentators to be one of the key weaknesses in Giddens’s project (e.g., Stones 2005, p. 120; Bryant 1992, pp. 143–144). It is not until the fundamental nature properties and capacities (such as agency) of the entities involved in such chronic and ever-present processes as social transformation, reproduction, and solidarity are established that we can then begin to develop more effective ways of studying them and explaining them. I refer to this tendency in Giddens’s overall theoretical program as his ontological structuralism. This theoretical strategy is fundamentally at odds with Levi-Strauss’s methodological structuralism. Thus, it is no surprise to find that while Giddens (1979, p. 63) is very well aware that for Levi-Strauss, “[s]tructures ... involve models posited by the anthropological observer” he immediately recoils from this view, and chooses to reject “this curious mixture of nominalism and rationalism” arguing instead that “structure has a ‘virtual existence’ as instantiations or moments; but this is not the same as identifying structure merely with models invented by sociological or anthropological observers” (italics added).
This is a crucial theoretical maneuver. By turning his back on methodological structuralism in order to concern with questions of “virtual existence” Giddens ends up adopting a problematic, quasi-referential notion of structure. This idiosyncratic concept now comes to “pick out” a set of problematically defined properties of putatively actually-existing entities in the social world. Structure is now characterized by an equally mystifying quasi-existence both in and out of “time-space” and only exhibits “paradigmatic” properties “as instances” (Archer 1982, 1995; Sewell 2005b; Thompson 1989). Ironically, what ends up ultimately being curious is not Levi-Strauss’s judicious deployment of methodological structuralism—mischaracterized by Giddens as “nominalism”—but Giddens’s hybrid ontology in which properties that make sense when applied to non-material structural models come to form part of actually existing entities in the real world (in analytical-philosophical terms, there is a confusion of intension and extension). Giddens (1979, pp. 64–65) concludes that [s]tructures exist paradigmatically, as an absent set of differences, temporally ‘present’ only in their instantiation, in the constituting moments of social systems. Thus, to think of structures as “....involving a ‘virtual order’ of differences ... does not necessitate accepting Levi-Strauss’s view that structures are simply models posited by the observer.” This view of structure would survive pretty much intact into Giddens’s later works (1984) and would constitute the shaky ontological foundations of the theory of structuration.8
The ontological strategy and the genesis of conceptual incoherence in the theory of structuration
Giddens’s ontologizing strategy and self-conscious abandonment of Levi-Straussian methodological structuralism in favor a denotative—or “realist” in his view—ontological conception irrevocably seeps over into his overall conceptualization of structure. Giddens (1984, p. 23) distinguishes “‘structure’ as a generic term from ‘structures’ in the plural and both from the ‘structural properties of social systems.’” After this dizzying array of semantic and terminological distinctions, one may ask: what then is structure? According to Giddens (1981, p. 170) the term structure, “‘... refers to rules and resources instantiated in social systems but having only a ‘virtual existence’” (italics added). By including resources (which refer to apparently material objects and modes of social organization) in his definition of structure, as noted by Sewell (2005b, p. 129), Giddens completely breaks with the Levi-Straussian legacy, in which structure could never be conceived as having any reference to real world entities, let alone material objects or actual human capacities (Giddens is ambiguous as to whether the term resource refers to actual things or the ability to use those things). For Giddens (1979, 1984), structure now refers to “rule-resource sets” implicated in the chronic temporal reproduction and historical transformations of social systems. However, Giddens also resorts to a Wittgenstenian notion of rule that appears even vaguer and harder to pin down than his previous definition of rules as comprising “rules of transformation” analogous to those used in generative linguistics. This time rules are defined as “generalizable procedures” deployed by knowledgeable actors in their everyday “enactment/reproduction of social life” (Giddens 1984, p. 21). As noted by Sewell (2005b), this conception of rules is “even more impoverished ... than it was in Central Problems in Social Theory, which at least implied an analogy with Levi-Strauss.”
It should be noticed however, that Giddens’s shift to a Wittgenstenian conception of rules had a lot to do with the incompatibility between Levi-Strauss’s ideas regarding the status of “rules”—a property of models—with Giddens’s ontologizing project, which required that they be a real property of human actors and institutionalized social arrangements. This ambiguity however, is not removed by Giddens’s move toward Wittgenstein; instead Wittgenstein comes to form the core of Giddens’s own practice theory, which in contrast to Bourdieu’s efforts in the same respect (departing from the French Maussian legacy rather than from the Anglophone analytical-philosophical one), comes to be irrevocably conflated with his own attempt to ontologize the notion of structure.
Contrary to some interpretations, Giddens by no means abandons his earlier Levi-Straussian conception of rules as a “virtual” and “paradigmatic” collection of transformational codes located “out of time and space” for the newer Wittgenstenian neo-pragmatist conception, but retains both emphases in spite of their obvious conceptual tensions and ultimate incongruity. As noted by Archer (1982: 478) Giddens never quite shakes off the linguistic analogy; “... the key concepts themselves come direct from linguistics: the ‘recursive character of language—and, by generalization, of social systems also’—is the source of the ‘duality of structure’: the notion that society, like language, should be regarded as a ‘virtual system’ with ‘recursive properties’ comes direct from Saussure.” Thus, Giddens’s early (1979) development of the notion of structure in analogy to “language and speech” is “the source ... of many of the difficulties of his account” (Thompson 1989, p. 75).
Under this definition, rules come to acquire their novel connotation as generalizable procedures chronically enacted in everyday action, but at the same time Giddens continues to insist that they are also virtual entities with no “extension” in space and time: “... [s]tructure, as recursively organized sets of rules and resources, is out of time and space, save in its instantiations and co-ordination as memory traces, and is marked by an ‘absence of the subject’” (1984, p. 25). The problems that Archer (1982), McLennan (1984); Thompson (1989), and, as we will see below, Sewell (2005b) identify with this theoretical move, in particular that of the “insubstantiality of structure” ultimately become problems not necessarily because Giddens retains remnants of the Saussurean and Levi-Straussian legacy in his version of structuralism but mainly due to the fact that—in direct contradiction to Levi-Strauss—Giddens comes to think of structure as denoting actual constituent elements of society of which structure is a “property” and not as a methodological tool. However, it is obvious that in this context Giddens elides his own analytic differentiation between “structure” and “system” and goes on to mix capacities associated with older “organicist” notions of structure in functionalism (such as organization or patterning) with properties associated with the “newer” linguistic and mathematical notions (such as virtuality and transformational capacities). In this manner, Giddens produces an ultimately incoherent ontology as has been noted by some commentators (Thompson 1989; King 2000).9
Once we ontologize the notion of structure by including real material objects and actor-level capacities in it (knowledgeability) the notion of a virtual existence (in “stasis”) loses any meaning, since material objects, embodied capacities—such as Mauss’s (1973) “techniques of the body”—or stark distributional realities cannot be considered to have a “virtual” existence until touched by the magic wand of human agency (since some of them are constitutive of this agency in a strong sense). As Archer notes (1982, p. 461) referring to the examples of famine and over-population, these do not require “instantiation” in order to be effective; “they are there and the problem is how to get rid of them or deal with them.” It is when structure is thus ontologized that its most intriguing (and useful) features such as its purely “differential” existence pose problems, since it is hard to accept that an entity with such ethereal “capacities” could be constitutive of actual social reality. These are not mere “meta-theoretical” matters, as these untenable ontological presuppositions have unfortunately come to influence surreptitiously most of the research that makes use of the theory of structuration in contemporary Anglophone sociological research. I will note one further hypostatization of properties of models into properties of reality below, which comes to the fore most clearly in Giddens’s account of the role of time in social life.
“Mechanical” models versus “statistical” models
In his discussion of temporality, Giddens (1984, p. 35) notes that “[t]he duree of daily life, it is not too fanciful to say, operates in something akin to what Levi-Strauss calls ‘reversible time.’” He goes on to equate this “reversible time” with that which characterizes “the events and routines of everyday life” contrasting it to the “irreversible time” of the individual lifespan and the (once again) reversible time of the longue duree of institutions. Notwithstanding the obvious arbitrariness with which Giddens bandies about the notion of reversibility (why is time spent in everyday activities necessarily “irreversible”? Why are long stretches of historical time “reversible”?), it is easy to show that this ontologization of Levi-Strauss’s notion of “reversible time” goes against the grain—and semantics—of Levi-Strauss’s original intentions, and thus represents an effective test-case to demonstrate the problems that arise when properties of models are illicitly “ontologized” by the analyst.
For Levi-Strauss (1963e), time-reversibility like “virtuality” was never meant to be a property of any object in the world, but of structural models constructed to account for empirical facts about the world. Real-world objects and experiences, whether conceived in their day-to-day duration or in the “long run” of historical time, are located in the irreversible “time’s arrow” described by the second law of thermodynamics (Barbosa de Almeida 1990). Levi-Strauss is thus careful to separate the characteristically reversible time of what he refers to as “mechanical models” from the irreversible time which is characteristic of “statistical models” (1963c, p. 286), the latter of which come closer to representing real-world ecological and historical rhythms. Structural models are distinct precisely because they postulate that a particular segment of empirical reality can be represented as a finite set of transformations located in a purely analytic synchronic or “reversible” time; other models such as those used to represent historical transformations as a set of ordered sequences of discrete events (Abbott 2001) can be used to represent what Levi-Strauss referred to as the statistical or irreversible time characteristic of actually-existing historical social systems.
In either case, the “time-reversibility” is a property of the model not of reality, since the observable world partakes of a single form of temporality (irreversible) that is the antithesis of what characterizes structural models. Speaking of reversible time in reference to the real world practices and institutions is therefore a category mistake in Whitehead’s sense. For Levi-Strauss, any “strip” of empirical social phenomena could be analyzed in a mechanical or statistical manner. For instance the rules (“invariants”) that govern the exchange of spouses in a given kinship system could be described by a mechanical (structural) model in “reversible” time (Levi-Strauss 1969, 1976a). Conversely, the historical succession of different kinship regimes and the temporally situated events that resulted in the transition from one to the other could be described using a historical narrative set in “statistical time” (Collins 1981). This is how Levi-Strauss understood the Saussurean distinction between langue and parole “... one being the structural side of language, the other the statistical aspect of it, langue belonging to a reversible time, parole being non-reversible” (1963f, p. 209).
Thus, the fact that social structures are located “outside of time and space,” is not for Levi-Strauss a curious (and ultimately hard to account for) property of “structure” as an ontological entity, but an unproblematic feature of those structural models used to represent empirical materials. In fact, Levi-Strauss thought that this was the best way to reconcile Radcliffe-Brown’s notions of structure as the total set of social relations (located in irreversible time and certainly not endowed with a “virtual” character) and his own notion of structure as what is represented by a given model of social reality. Giddens, on the other hand, refers to the “virtual” character of structure and the “structural properties of social systems”—in their status as actually existing (quasi)entities—as one of their primary defining features, while at the same time conceiving them as located in the reversible time of Levi-Straussian mechanical models. Archer (1982, p. 461) rightly criticizes this rather unorthodox ontological claim, asking “... why should we accept this peculiar ontological status for structural properties in the first place?”
Bourdieu’s recovery of methodological structuralism
Pierre Bourdieu (1990a), in contrast to Giddens, was a direct intellectual descendant of the Levi-Straussian legacy in anthropology, and as such stands in a very different relationship to this intellectual heritage—as an initial qualified advocate and later critic of it—than Giddens. As such, we should not expect Bourdieu to have made the same theoretical choices as Giddens in regards to the issue of the ontological versus the methodological status of structure. While Giddens can be said to have selectively incorporated various generalized notions associated with structuralism in linguistics and mathematics (e.g., virtuality, the “absence of the subject,” “transformational capacities”) into his ontological project, Bourdieu—following Levi-Strauss—critically accepts structuralism as a primarily methodological program. Bourdieu comes to criticize Levi-Strauss at the point where he abandons his own methodological structuralist position and “short circuits” directly from the “model of reality” to the (ontological and phenomenological) “reality of the model.”
Following this line of argument, in this section I show that the theoretical projects of Giddens and Bourdieu—in spite of some very superficial similarities (such as their common self-proclaimed attempts to “transcend” the agency/structure problematic)—stand as radically opposed charters. Giddens (and as we will see below Sewell) follows the ontological path and collapses Levi-Straussian structuralism as a methodological tool into structuration as an ontological theory of the ultimate constituents of the social. Bourdieu, however, remains committed to the methodological structuralist principle of keeping a clear conceptual and practical barrier between the structural models (and their principles of functioning and presumed properties) constructed by the analyst and the practical models deployed by the social agent in routine activities and ultimately responsible for the production of social practices (which can then be represented in toto by some structural model).
Structuralism as method versus the structuralist reduction
First, Bourdieu is careful to separate the virtues of structuralism as a method from any philosophical commitment to the presumed reality or ontological features of “structure” as a stand-in for organized patterns of social interaction. This is clearly noted in the preface to The Logic of Practice, where it is emphasized that it was above all “... the ambiguities and contradictions which the very effort to push the application of the structural method to its furthermost conclusions constantly raised, that led me to question not so much the method itself as the anthropological theses tacitly posited in the very fact of consistently applying it to practices” (Bourdieu 1990a, p.4, italics added). Rather than rejecting Levi-Strauss’s methodological structuralism, Bourdieu accepts it as a practical procedure of scientific representation since “so long as they are taken for what they are, logical models giving an account of the observed facts in the most coherent and most economical way” (italics added). However, Bourdieu refuses to accept the structuralist account (as Giddens does in his account of rules) as a credible ontological or empirical description of the practical models and cognitive structures (schemas) deployed by the social agent. Bourdieu instead holds fast to the belief that logical models “... become false as soon as they are treated as the real principles of practices.”
... as the model of a practice which does not have this model as its principle, the diagram and all of the oppositions, equivalences and analogies that it displays at a glance are only valid so long as they are taken for what they are, logical models giving an account of the observed facts in the most coherent and most economical way; and that they become false and dangerous as soon as they are treated as the real principles of practices, which amounts to simultaneously overestimating the logic of practices and losing sight of what constitutes their real principle. One of the practical contradictions of scientific analysis of a practical logic lies in the paradoxical fact that the most coherent and also the most economical model, giving the simplest...account of the whole set of facts observed, is not the principle of practices which it explains better than any other construct ... practice does not imply—or rather excludes—mastery of the logic that is expressed within it (Bourdieu 1990a, p. 11, italics added; see also the discussion of the calendar and the “synoptic illusion.” pp. 203–209).
The “practical contradiction” of the analysis of practices (and other forms of cultural phenomena) consists precisely of the fact that as a rule, the principles underlying the best representative and explanatory tools available are antithetical to the practical logic that routinely produce (in natural contexts) the very correspondences among diverse empirical facts to be explained. Bourdieu acknowledges the pragmatic and epistemological virtues of structural models while rejecting their propriety as providing access to “the real principle of practices.” In Bourdieu’s reconstruction of Levi-Straussian methodological structuralism, it is therefore illicit for the analyst to “slip from regularity, i.e. from what recurs with a certain statistically measurable frequency and from the formula which describes it, to a consciously laid down and consciously respected ruling ...or to unconsciously regulating by a mysterious cerebral or social mechanism” (Bourdieu 1990a, p. 39, italics added).
Bourdieu concludes that Levi-Strauss is ultimately guilty of failing to distinguish between the logic underlying the structural model and the embodied practical logic deployed by social agents in their everyday dealings. This is because the latter thought that the fundamental structures of the human brain would be ultimately shown to be essentially the same in their overall functioning and patterning as the logic of functioning of the structural models.10 This set of inconsistencies in the Levi-Straussian project come out most clearly in Bourdieu’s (1990a) discussion of Levi-Strauss’s (1987) critique of Mauss’s theory of the gift (an earlier  version of which was cited approvingly by Giddens [1979, pp. 25–26]). Levi-Strauss suggests that while Mauss’s analysis of the gift economy was path-breaking, it did not quite reach a satisfying level of explanatory accuracy, precisely because it remained within the ambit of the “‘phenomenology’ of gift exchange,” Levi-Strauss in contrast, “... holds that science must break with native experience and the native theory of that experience and postulate that ‘the primary, fundamental phenomenon is exchange itself, which gets split up into discrete operations in social life’ (Levi-Strauss 1987: 47, quoted in Bourdieu 1990a: 98).” The “unconscious principle” at work here is, according to Levi-Strauss, that what inheres in the “... the ‘automatic laws’ of the cycle of reciprocity are the unconscious principle of the obligation to give, the obligation to return a gift and the obligation to receive” (1987, p. 43, quoted in Bourdieu 1990a, p. 98).
Levi-Strauss’s mistake here lies precisely in “.... postulating that the objective model, obtained by reducing the polythetic to the nomothetic, the detotalized, irreversible succession to the perfectly reversible totality, is the immanent law of practices, the invisible principle of the movements observed” (italics added). However, following this pathway is the quickest road to a denial of a need of a theory of practice and the postulation of thoroughgoing structuralist determinism, since “... the analyst reduces the agents to the status of automata or inert bodies moved by obscure mechanisms” (Bourdieu 1990a, p. 98, italics added). Following this path requires us to “... ignore the fact that the agents practice as irreversible a sequence of actions that the observer constitutes as reversible” (Bourdieu 1990a, p. 104, italics added). This equation of the irreversible rules of practical action with the reversible transformation rules that allow for experimental manipulation of the “mechanical” structural models constructed to account for the entire system of practices is “one of the most disastrous fallacies in the human sciences ... that which consists of passing off, in Marx’s well-known phrase, ‘the things of logic for the logic of things’” (Bourdieu 1990b, p. 61).
In structural anthropology this confusion originates from the fact that the word rule has at least three central connotations: first, it could refer to the fact that something occurs regularly; that is to “a set of objective regularities imposed on all those who join a game.” Second, it could refer to the fact that there exists a clearly objectified and consciously held precept that is, “a principle of the juridical or quasi-juridical kind.” But most importantly, we could “also be thinking of a third meaning, that of the model or principle constructed by the scientist to explain the game” (Bourdieu 1990b, pp. 60–61, italics added). Thinking of the social agent’s behavior as being directed by the same “mechanical” rules that govern the structural model constructed by the analyst results in a structuralist reduction—a charge that King (2000) shows is applicable to the theory of structuration but errs in thinking it also applicable to Bourdieu (2000b)—of really existing social practice, with its fuzziness, open-ended character, and location in “statistical” time (Bourdieu 1990a, pp. 98–99), to the airtight logic and “reversible” synchrony of the structural model. Instead, the practical models responsible for the practices observed by the analyst stand to the structural models constructed by the analyst to reconstruct the entire system of practices as “... as old houses, with their successive annexes and all of the objects, partially discordant but fundamentally in harmony with them ... [stand to] to apartments designed from end to end in accordance with an aesthetic concept imposed all at once and from the outside by an interior designer” (Bourdieu 1990a, p. 13).
It is clear that the structuralist reduction is generated from an unwarranted hypostatization of the properties of models into properties of the real bodily and cognitive mechanisms responsible for the production of practices: “[t]he generative formula which enables one to reproduce the essential features of the practices treated as an opus operatum is not the generative principle of the practices, the modus operandi.” If this were the case, “... and if practices had as their principle the generative principle which has to be constructed [by the scientific observer] in order to account for them,” those practices “... would be stripped of everything that defines them distinctively as practices, that is, the uncertainty and ‘fuzziness’ resulting from the fact that they have as their principle not a set of conscious, constant rules, but practical schemes, opaque to their possessors, varying according to the logic of the situation” (Bourdieu 1990a, p. 12, italics added). Structuralism can in this way be rejected as an (objectivist) description of the “deep” (ontological) structure of the social world, while at the same time the construction of the structuralist model by the analyst—which “has to be constructed” to account for the larger patterning of practices—can be accepted as a non-negotiable requirement of effective social-scientific analysis.
Given this, any argument for a “convergence” between Giddens’s and Bourdieu’s approach to structural analysis can clearly be shown to depend on superficial generalities. For instance, it can be argued that Giddens’s discussion of rules conforms to the Levi-Straussian idea of a “logical model” and as such is open to Bourdieu’s criticism of the “structuralist reduction.” Giddens (1984, pp. 19–20) thinks that the best representation of a rule is not an empirical regularity nor a formally written prescription, but a mathematical formula that provides the key to a particular empirical pattern of occurrences such as a sequence of numbers, and which allow us to predict the next event in the sequence: “[a] formula is a generalizable procedure—generalizable because it applies over a range of contexts and occasions, a procedure because it allows for the methodical continuation of an established sequence.” It is in “... the nature of formulae that we can best discover what is the most analytically effective sense of ‘rule’ in social theory.” Giddens never quite adequately addresses the issue, so central to Bourdieu’s critique and reappropriation of Levi-Strauss’ methodological structuralism, of the presumed correspondence between structural models of reality and the embodied practical models ultimately responsible for the practices of the social agent.
This question is not meaningful from within the strictures of the Giddens’s conceptual framework since he rejects the methodological status of structures from the very beginning in favor of a problematic ontological conceptualization. However, it is clear that we are left to assume that if structure only exists as “memory traces” then these practical models must share the same set of abstract properties as those displayed by such a “generalizable procedure” as a mathematical formula. It is possible to defend Giddens from the charge of “ontologizing” structure by pointing out that he never meant structure to refer to real constituents (or “building blocks”) of social systems, practices, and objects (e.g., Bryant 1992, p. 142), but the abstract principles that pattern those practices a reading that is belied by Giddens’s own stronger pronouncements (e.g., Giddens 1979, p. 63). As we have seen, even if the claim that Giddens was not committed to the ontological reality of structure could be defended, by making the ultimate site of structure the “memory traces” that “store” the discursive and practical knowledge responsible for practice, Giddens retraces the same misstep that led to Bourdieu’s critique of Levi-Strauss from the point of view of the latter’s own methodological structuralist perspective. From this point of view, a consistent methodological structuralism can only be sustained by refusing to locate structure as a “really existing” entity (even if it is only in “memory traces”).
We have seen that Giddens (1984) makes a similar mistake in his theory of temporality by suggesting that the durée of everyday life operates in Levi-Straussian reversible time. This is an error according to Bourdieu (1990a, p. 81), since practices are always located in the irreversible flow of what Levi-Strauss referred to as statistical time. Postulating otherwise, represents a clear-cut case of “the theoretical error that consists in presenting the theoretical view of practice as the practical relation to practice, and more precisely in setting up the model that has to be constructed to give an account of practice as the principle of practice” (italics added). For Bourdieu, radically different accounts of temporality—strictly analogous to the problematic account of time given by Giddens, which ends up eliding temporality by collapsing agency into structure (Archer 1982)—lie at the “root of this error.” In particular, the analyst fails to take into the account properly “... the antinomy between the time of science and the time of action.” This ends up damaging the theorist’s conception of practice, “... by imposing on it the intemporal [sic] time of science.” This was Levi-Strauss’s error, a mistake that Giddens reproduces for entirely different reasons. While the theoretical model is indeed necessary, it is important to realize that “[t]he shift from the practical scheme to the theoretical schema, constructed [by the analyst] after the event, from practical scheme to theoretical model ... lets slip everything that makes the temporal reality of practice in process.” Because of this it must be kept in mind that “[p]ractice unfolds in time and it has all the correlative properties, such as irreversibility, that synchronization destroys. Its temporal structure ... is constitutive of its meaning” (Bourdieu 1990a, p. 81, italics added).
Sewell’s reconstruction of structuration theory
In what is without a doubt the most influential statement regarding the idea of structure in contemporary American sociology, Sewell (2005b) attempts to produce a workable notion of structure for sociological research.11 He does this primarily by clearing up and fixing the various conceptual and linguistic muddles that come from Giddens’s ontologization of the Levi-Straussian notion of structure, but without renouncing the latter’s ontological project. We will see that Sewell pays dearly for this abandonment methodological structuralism, by becoming embroiled in his own—ultimately irresolvable—set of meta-theoretical dilemmas.
Sewell first proceeds to disentangle Giddens’s account of rules from the various inconsistencies that come from conceiving of this concept in both the traditional sense of rules as procedural cognitive representations of collectively shared notions of how to go about our everyday routines and from the methodological structuralist sense of rules as the “principles” that govern the functioning of a given mechanical model constructed to represent some empirical segment of social reality. Giddens attempted to deal with this problem by (mis)interpreting the notion of a mathematical formula as a “general” representation of a possible empirical “series” as comparable—and even more strongly, somehow synonymous—to the Wittgenstenian idea of shared practical competences conceived as generalizable (mental and bodily instantiated) procedures. This analytical equation is as Thompson (1989, p. 64) argues implausible.
Sewell notes the inconsistency and correctly notes that if a “rule” is just a cognitive procedure that can be used in a variety of situations then there is already a perfectly adequate term in cognitive anthropology available: the notion of a schema. Sewell abandons Giddens’s problematic quasi-mathematical notion of rule, for a simpler cognitive version. However, this is not the only thing that Sewell does away with by resorting to this conceptual move: whatever connection structuration theory might have still retained to Levi-Straussian methodological structuralism, with rules as a property of manipulatable structural models, goes by the wayside. This analytical move precludes from consideration the key issue of the relationship between the “really existing” schemas instantiated by the agent and the reconstruction of those schemas in the structural models produced by the analyst. In this way, both Giddens and Sewell in their own way appear guilty of the violation of the methodological structuralist principle of hypostatizing what is essentially a possible model of social practices (such as a formula) into the real model of practices physically instantiated (embodied) in the social agent’s motor and cognitive system.
Sewell must still deal with the infelicitous ontological consequences that carry over from Giddens’s unwarranted ontologization of the properties of (non-material) structural models to rules as physically instantiated in the social agent’s mind-brain (as “memory traces”). These cannot be considered to be “virtual” in the same sense that a mechanical model of social reality in the Levi-Straussian sense is virtual; since the most obvious sense of “virtuality” in this context is insubstantiality or non-materiality (Archer 1982). But if schemas are embodied in the agent as both Giddens and Bourdieu agree, they cannot be immaterial in this latter sense. Sewell attempts to deal with this by proposing that the “... generalizability or transposability of schemas is the reason [why] they must be understood as virtual. To say that schemas are virtual is to say that they cannot be reduced to their existence in any particular practice or any particular location in space and time: they can be actualized in a potentially broad and unpredetermined range of situations.” It is straightforward to see however, that the ability to be used across contexts is not a sufficient warrant to conceive of virtuality as an alleged quality of a (presumably physically instantiated) cognitive scheme. The reason for this is that virtuality is an ontological claim that a given entity stands “outside of time and space,” that is, that it lives in the “synchronic,” reversible time of structural models and not in the reversible time where real social agents and real social practices are located.
The fact that a given cognitive scheme can be used in different situations does not mean that it stands “outside of time and space” or that qua mental and bodily scheme it partakes of the “reversible” and “virtual” properties that a structural model of that scheme may partake of (for instance, a cognitive scheme can degrade with age or atrophy as a consequence of disuse). For instance, Bourdieu’s (1990a, p. 69) use of the notion of transposability—which Sewell acknowledges to be preferable to Giddens’s idea of generalizability—is free of Giddens’s vestigial ontological fallacy of misplaced concreteness; in Bourdieu’s formulation, the bodily schemes constitutive of habitus are both transposable and real (embodied) with a clear location in time and space and with an empirically verifiable and definable (irreversible) ontogenetic historical legacy (Bourdieu 1996b, p. 286), related to their own conditions of production (such as the “diffuse” pedagogy of the family environment or the more explicit pedagogy of the scholastic situation). Transposability is therefore not a necessary warrant for virtuality. By considering it so, Sewell remains trapped in the inconsistencies of Giddens’s ontological project.
After attempting to clear up Giddens’s problematic use of the term “rules,” Sewell proceeds to deal with the even more problematic status of the other side of the “duality” of structure: the idea of structure as being “composed” of resources. As we have seen, this is the aspect of the theory that initially drew the most critical fire precisely because it is where Giddens’s illicit transfer of the properties of structural models most conflicts with the obvious materiality of the usual things that are referred by the word resources. While structural models and structural representations could be conceived to be virtual it is much harder to think of material and social resources as possessing this rather puzzling feature. However, due to the rather strange postulation that a putatively material element was part structure as defined in mathematics and linguistics, Giddens’s concept of resources ends up being “... even less adequately theorized than his concept of rules.”
The simplest way of conceptualizing structure would be to return to Giddens’ starting point in structuralism and to assert that structure refers only to rules or schemas, not to resources, and that resources should be thought of as an effect of structures. In this way, structures would retain their virtual quality, and concrete distributions of resources would be seen not as structures but as media animated and shaped by structures, that is, by cultural schemas (Sewell 2005b, p. 134).
In proceeding in this manner, Sewell succeeds in clearing up Giddens’s conceptual muddle, but at what cost? First it is important to note that it is a mistake to think that a “return to Giddens’ starting point in [Levi-Straussian] structuralism” would actually entail adopting the idea that “structures” refer to “rules or schemas” because these seem intuitively “less concrete” (and thus less “real”) than resources. A more consistent interpretation of Levi-Strauss’s methodological structuralism would draw a sharp analytic differentiation between the actually instantiated practical schemes “inside the head”—or inscribed in the body (Mauss 1973; Wacquant 2004)—of the social agent and the structural models designed to represent the totality of these schemes as a differentially organized “cultural system” (in the Parsons-Geertz sense) which is not located in anyone’s head or inscribed in anyone’s body (Bourdieu 1990a).
Sewell errs in thinking that the only two options available are either a direct ontologization of structure as thoroughly “composed” exclusively of virtual schemes or Giddens’s seven less desirable hybrid ontology of structure as quasi-immaterial (rules) and quasi-material (resources). I suggest that there is a third alternative: a return to the methodological structuralist stance of conceiving of structure as “made up” of neither rules nor resources but as an ontologically neutral and non-referential set of analytic procedures or models used to represent the totality of a given set of social practices or classification schemes. Sewell’s reconceptualization of Giddens in contrast, follows the first route of directly ontologizing structures as “virtual” set schemas, which retains the problematic mind/matter dualism at the level of ontology, and the inconsistent hypostatization of properties of structural models—virtuality—into properties of actually existing entities.
Another unfortunate consequence of Sewell’s conceptualization of structure as solely “composed of” virtual schemas and of resources as therefore “not structure” (which are then thought of as being the result of the operation of these “structural” schemes) concerns his inevitable entry into thorny metatheoretical issues related to the casual interplay of putatively immaterial schemes and seemingly material entities (an old Cartesian problematic). Thus, following Sewell we are now led to think of structure (as schemas) as having “effects” on resources by virtue of their capacity to “activate” them in everyday social action. Sewell is thus forced to conclude—if he is to retain the logical virtues of his reconstruction of structuration theory—that “...resources can plausibly be thought of as effects of cultural schemas” (2005b, p. 135, italics added). The reader may notice that the putative virtuality of schemas begins to feel suspect here, as one may ask how exactly is it that virtual schemes that stand outside of time and space and are characterized “by an absence of the subject” come to have real world effects on material entities that are presumably firmly located in some spatial and temporal context.
Notice that by abandoning the ontological neutrality of the methodological structuralist stance, Sewell is now on the horns of the perennial issue of materialism versus idealism (uncomfortably pinned on the side of the latter). Sewell immediately realizes this and is quick to point out that this privileging of schemas as the driving force of social reproduction and change in his version of structuration threatens to render “[s]tocks of material goods and people’s knowledge and emotional commitments,” in short everything that is not “structure” under the new definition, “... inert, mere media for an outcome of the determinative operations of cultural schemas,” an outcome that Sewell finds unsatisfactory. He tries to resolve this thorny issue by attempting to rescue Giddens’s highly problematic—as we saw above—notion of duality. Instead of Giddens’s ontological duality, however, Sewell (Sewell 2005b, pp. 136–137) now speaks of a chronically instantiated duality of cause and effect, in which resources are seen as simultaneously being the effects of schemas, and schemas as the effects of resources.
For Sewell therefore, the initial idealist connotations of his conception of schemas as the “structure” and resources as the “material” that is in-formed (given shape) by structure, are apparently “solved” by an equally unsatisfying “casual interactionism” in which the ideal causes the material and the material causes the ideal in an endless self-supporting positive feedback loop (Spillman 1995, p. 147). Even if we decide to sidestep the thorny problem of how is it that these two heterogeneous domains can interact casually if they are located in separate ontological registers, one may ask whether the casual interactionism proposed by Sewell is an effective solution in the first place. Suffice it to mention that this very same underspecified casual picture was precisely one of the untoward analytical consequences of Giddens’s structuration account critically isolated by Archer (1982). Archer argued against this “loop” view of structural causality, which by introducing a circular picture of time reproduces a “reversible” representation of temporal order in place of what should be—for analytic purposes if any type of “structural” explanation of events is desired—an “irreversible”—casually “staggered” image (Coleman 1986). In this latter formulation, structural conditioning at “time one” should be analytically distinguishable from “structural elaboration” at a subsequent point in time. The Sewell-Giddens position on the other hand, in effect takes time out of the explanatory picture (Archer 1995). This is a rather ironic end point for an analyst as theoretically sensitive to the issue of temporality as Sewell (2005c).
The point of all of this is not to evaluate Sewell’s own rather ordinary—in comparison to the analytical brilliance of his reconstruction of the structuration problematic in the same piece—attempt to escape from the horns of the materialism versus idealism dilemma in suggesting an “it goes both ways” interactionist picture. The key point is that from a truly methodological structuralist stance there is no reason to have to fall into this meta-theoretical trap in the first place, since one of the virtues of methodological structuralism is precisely its agnosticism as to what are to be considered the ultimate “constituents” of structure with the term structure now being used in its unremarkable neo-organicist sense. By directly ontologizing structures as mental schemes however, Sewell is unable to avoid this metatheoretical dilemma. I am skeptical that any ontological formulation, Sewell’s, Giddens’s, or anybody else’s, can quite escape these inconsistencies without collapsing into the trivial usage of the idea of structure to imply general patterns of “organization” that Kroeber criticized more than half a century ago.
However, Sewell worsens the situation in his belief that the only way to make his theory of structure versatile enough is to take advantage of the semantic vagueness of such terms as “schemas” and “resources” in order to generalize the idea of structure to various possible real-world cases, and to extend an assortment of other problematic properties to this already overloaded notion of structure. Thus, Sewell postulates that structures (as real entities) are “multiple,” and that resources are “polysemic,” and that structures can “intersect.” However, because Sewell retains an ontological (and thus referential) interpretation, this generalization does not entail the usage of the same structural model to explain the co-occurrence of various empirical patterns, but the ontological proliferation of possible real world referents of the notion of structure. 12 This ontological proliferation of possible empirical referents for the notion of structure has several undesirable consequences, the most important of which is the return of the same semantic vagueness and conceptual incoherence to the notion of structure that Sewell so laboriously endeavored to undo.
It quickly becomes clear that if we follow the logical implications of Sewell’s formulations to their full extent, there is hardly any form of social organization, set of social conventions or collectively shared procedures that cannot be referred as “structures” since they all involve some form of deployment of cognitive models and some sort of manipulation of human or non-human resources by way of those models. But if structures refer to everything, they also refer to nothing at all. Unfortunately, this is the logical end point of strictly following Sewell’s deployment of the ontological strategy: a collapse back into the trivial—non-methodological—notion of structure as historical modes of “social organization” or recurrent “pattern,” or at best, mutually associated sets of cognitive schemes and material resources at a given historical juncture.13 It is unclear whether such a wide-ranging notion can be useful for actual research beyond providing certain broad orienting statements to demonstrate to contemporary academic gatekeepers that the research in question is sufficiently “theoretical.” Sewell’s “theory of structure” thus ends up being an eclectic theory of historical forms of social cognition and how these collectively shared schemas are implicated in historically specific forms of social organization and material exchange. However, there is nothing “structural” about this theory, since it is hard to think (without falling into Giddens’s ontological contradictions) how these structures have anything to do with structural models as a representation of the systemic organization of a set of elements, whereby these elements gain meaning from their position in the complete relational system.
This is the methodologically productive notion of structure initially imported by Levi-Strauss from mathematics and linguistics, and which gave birth to the structuralist movement in the social and human sciences. It is therefore somewhat ironic that in spite of his initial rejection of the theoretical position—mistakenly—attributed to Levi-Strauss by most commentators, Sewell (2005b, p. 151) ends up with a not too dissimilar formulation of structure as ultimately “profoundly cultural” in nature. We can thus say of Sewell something similar to what Levi-Strauss (1976b, p. 17) said of Radcliffe-Brown’s usage of the notion of structure, who “affirms it, but in seeing it where it is not, he takes away from the notion of structure its strength and its significance.”
Structure and method in contemporary social science
I close by noting certain necessary limitations of the analysis offered above as well as responding to possible ways in which the argument could be misunderstood. Finally, I briefly outline some of the main implications that I see it having for contemporary theory and research in sociology that takes the ideas of structure as a point of departure.
First, it is important to underscore that my characterization of the intellectual career and vicissitudes of the concept of structure in recent social theory is necessarily partial. Not all authors or intellectual traditions of thought have received equal treatment. My perspective on the concept has been limited to those lines of conceptual development that have come to influence contemporary empirical research in Anglophone (primarily American) sociology, and as such it is admittedly “presentist” in its considerations. I believe that my sustained critical attention to Giddens and Sewell can be justified precisely in terms of the contemporary relevance and diffuse influence of his formulation—and Sewell’s subsequent revision—as a simple citation analysis would attest. In this respect, this influence is arguably more sweeping than those alternative intellectual lineages that have dedicated themselves to bringing a more tangible empirical specificity to the idea of social structure—mainly associated with American network analysis (Marsden and Laumann 1984, p. 54). Network-structuralism, while undoubtedly extensive in its intellectual reach has been less forthcoming in terms of explicitly articulating its own conceptual underpinnings and its relationship to other important lines of thinking in the sociological tradition (Emirbayer and Goodwin 1994; Fararo and Butts 1999).
It is also important to point out that my argument for the existence of a current of thought associated with “methodological structuralism” could have easily have been extended both (backwards) chronologically towards the classical theory tradition and in terms of scope towards cognate arguments on the methodology of the social sciences that have become influential in other disciplines. In the classical theory tradition, it is undeniable that the first figure rigorously to formulate a program akin to Levi-Strauss’s methodological structuralism as a way to build a “charter” for the social sciences was Georg Simmel. According to Wolf (1959), rather than being characterized by having its own sphere comprised of “social” phenomena and processes as its unique subject-matter, the discipline of sociology is instead, “... in its relationships to the existing sciences ... a new method, an instrument of investigation, a new way of getting at the subject matters of all those sciences ... it contains no subject matter not already treated in one of the existing sciences. It only proposes a new way for all of them, a method which, because of its very applicability to all problems, does not constitute a science on its own.”
Not surprisingly, Simmel’s “formal” sociology represents the primary classical intellectual influence on contemporary network analysis (Emirbayer and Goodwin 1994, p. 1415; White et al. 1976, p. 730), in spite of the fact that it clashes with the equally influential Radcliffe-Brownian definition of structure as “concrete.” Simmel’s influence is also the classical source of the “anti-categorical” imperative that has produced so much difficulty in terms of forging a non-reductive relationship between network theory and cultural theory (Emirbayer and Goodwin 1994), also an ironic development given Simmel’s status as one of the foremost theorists of culture in the classical tradition (Frisby 2002).
In a similar manner, key founding figures in other disciplines have produced proposals aimed at disengaging the manner in which phenomena are investigated from the actual “stuff” to be investigated ,which bear a Wittgenstenian “family resemblance” to the methodological structuralist approach to the structure problem. In this respect, the locus classicus of structuralism, Saussure’s (1966) Course in General Linguistics can be seen as such a “formalist” manifesto, in which the discipline of linguistics is defined by the largely methodological separation between synchrony and diachrony and thus between the “form” and the “substance” of language. We may even think of Friedman’s (1953) influential tract on “The Methodology of Pure Economics” as belonging to a comparable tradition.14 Friedman attempts to eliminate by methodological fiat any consideration regarding the psychological makeup or the social embedding of the agent as a criterion with which to judge the adequacy of a given model of economic behavior (and thus any relation of verisimilitude in relation to the underlying ontological reality) in order to focus on purely “pragmatic” qualities related to the predictive ability of said model.
However, it is important to underscore that the above argument is not to be read as an endorsement of such pragmatic irrealism at the level of social-scientific methodology. The adoption of a methodological structuralist strategy does not in my view mean that concerns for fidelity to the underlying nature of the social world, and thus the “realisticness” (Maki 2002) at the level of the ontology of sociological theories are completely renounced. Rather than being an argument for the replacement of ontological considerations regarding the notion of structure (i.e., ontological theories of structure) in favor of purely methodological heuristics, I believe that the key implication of the above argument is that the two endeavors should be partially decoupled. That is, the practical use of structuralist methods and structural models should be kept separate from allegiance to any specific account as to the underlying nature of social reality. This does not mean that such accounts are not to be relied on (or developed) by the sociological investigator, but only that the relationship between structuralist methods and substantive theories of structures should be one-to-many rather than one-to-one, as the same structural models designed to objectify social reality may be compatible with competing ontological accounts of the “substance of the social” (Schatzki 1996).
Ontology and epistemology
This puts into question what I believe is one of the most seductive (but misleading) arguments regarding the relationship between theory and method in contemporary social science. The argument goes as follows: from the reasonable premise that some ontological conceptualization of the social world is an inescapable part of any research endeavor, it is concluded that the choices of specific methodologies are highly constrained by certain ontological commitments, or worse, that the underlying ontological structure of the social world dictates the type of method that is to be applied to a given set of empirical materials (Abbott 1988; Sewell 2005a). This argument (implicitly) follows Giddens’s (1984) ontological project in supposing that questions of ontology are inescapable and have primacy over (and in fact strongly limit the range of answers to) epistemological questions. This implies (even though Giddens’s writing were ambiguous in this respect) that ontological questions have to be settled before the practical activity of research begins. The researcher is advised to assume consciously an ontology of “the social” (Schatzki 1996) to prevent an implicit one from being assumed for her (usually by the methods that the analyst puts to use).
The question of the social is an ontological question, a question about the nature of the world. Social scientists tend to be rather shy about making explicit ontological statements. What I have said about sociologists above is true of most social scientists: they are much more comfortable theorizing about “science”—that is, how we may gain knowledge about the world—than about the social. Or, to put the same point in different words, the social sciences have far more highly developed methodologies than ontologies, whether stated or assumed.... Adopting this or that methodological stance .... implies a conceptualization of the nature of the social world. What we are able to study in social science and how we are able to study are inseparable from our social ontologies.
This argument appears to be bolstered by the fact that even Pierre Bourdieu, who as we have seen is somewhat exceptional among contemporary analysts in having appreciated the primarily heuristic role of structuralist models in social inquiry, was not always completely unambiguous in his deployment of the term “structure.” A simple inspection of his writings reveals that he was committed to a fairly strong (and “realist” in Giddens’s sense) theory of the ontological efficacy of both “subjective” (cognitive) and “objective” structural orders.15 According to Bourdieu, these tended to become interpenetrated to produce a state of “ontological correspondence” (Bourdieu 1996a, p. 38) or “ontological complicity” (Bourdieu 1990b, p. 108). This accounted for the origins of the Schutzian “natural attitude” in which certain arbitrary classificatory and resource-distribution arrangements come to be taken for granted by the social agent. In this respect, it is undeniable that Bourdieu does his share to contribute to the conceptual cacophony associated with the term “structure”—and to the ambiguity surrounding the ontological and methodological usages of the concept—by using it both to refer to really-existing social and cognitive orders and not just to models constructed by the analyst to bring order to empirical materials. It is thus fair to say that Bourdieu would certainly agree with Sewell in thinking that the “question of the social is an ontological question.”
However, it is not correct to presume that Bourdieu endorsed the use of structuralist methods because he thought that this was the best way to represent a social reality that somehow partook of the properties associated with structural models. As we have seen, both Sewell and Giddens make this mistake. While Bourdieu is certainly guilty of terminological slippage between the “dual” usages of the term structure (the denotative and the methodological), his argument regarding the practical use of structural representations on the part of the scientist is of a different order. For Bourdieu, structuralist methods are to be resorted to precisely because they have the heuristic effect of bringing to light global patterns in empirical data that are obscured by our habitual experience of the social world as engaged agents constrained to a particular (interested and practical) point of view (Bourdieu 1996b, pp. 193–205). The scientific justification for structural models is thus primarily epistemological. Structural methods are not “demanded” by the underlying nature (ontology) of social reality; that is they are not to be used because they are the only ones that “fit” some analytically pre-constructed “deep structure” of the social world. Thus, Bourdieu would not agree with Sewell’s suggestion that “adopting this or that methodological stance ... implies a conceptualization of the nature of the social world.”
I would thus argue that (in line with Levi-Strauss’s original formulation) the analyst’s decision to use a given structural model is quite separable from the analyst’s conceptualization of the fundamental nature of social reality (the realm dedicated to ontological theories of structure). In fact, as we saw above, we should expect that as a rule the properties of models should be antithetical to, and not isomorphic with, the actual properties of social and cognitive structures that they are meant to give a “picture” (or representation) of. As Bourdieu (1990b, p. 67) notes, the inability to integrate this distinction properly into theoretical and research practice “haunts the entire history of linguistics, which, from Saussure to Chomsky, tends to confuse the generative schemes functioning in the practical state with the explicit model—the grammar—constructed [by the analyst] in order to explain the statements made” (italics added). The representative and explanatory functions of models are thus distinguishable from their fidelity to the underlying structure of the social. Slippage between these two levels is the main culprit for the ontological incoherence of the Giddens’s theory of structuration, and by extension necessitates Sewell’s “weak-structuralist” solution.
Thus, it is clear than for both Bourdieu and Levi-Strauss (and it is evident in the Kuhnian “exemplars” that they left behind of how to put their precepts into practice), it was precisely the ability to construct the scientific object in a manner that liberated that construction from strong pre-experiential ontological presuppositions that was the primary reason to resort to structuralist methods. This allowed the practical separation of our “lived” ontology from that which was (re)created by the analyst in the process of social scientific observation. For instance in his influential paper on “the field of cultural production” Bourdieu (1983, p. 311) notes that “[f]ew areas more clearly demonstrate the heuristic efficacy of relational thinking than that of art and literature. Constructing an object such as the literary field requires and enables us to make a radical break with the substantialist mode of thought” (italics added).
In this manner, questions of ontology can be thought of as partially separable from questions of methodology, when the analyst is clearly aware that he or she is enacting that separation in practice. The examples that Sewell (2005a, p. 320) notes of “bad practice”—where the taking of a “methodological” standpoint comes to imply the surreptitious smuggling of ontological presuppositions (see also Abbott 1988 for a similar argument)—are to be disallowed not because they couple ontology to methodology (or because they demonstrate the supposed inevitability of this coupling), but because the analyst in these examples is clearly presumed to be using these methodologies—“... multivariate positivism, rational choice, network analysis, post-structuralism, or hermeneutics” (Sewell 2005a, p. 320)—in a non-reflexive manner. This predictably reproduces the “scholastic fallacy” whereby the model of reality is mistaken for the (ontological) reality of the model (Bourdieu 1998, pp. 127–140; 2000, pp. 49–84; Gartman 2007, pp. 394–395).
[t]he problem is not that social scientists engage in arguments of the mechanical type, but that they reify such arguments, imagining that the social world is constituted by a complex mechanism. The proper procedure, as I see it, is not to reify mechanisms, but to use mechanistic arguments as means toward the proper goal of the social sciences: de-reifying what appear to be mechanical causes, identifying the articulated semiotic practices that actually construct the unintended consequences and unrecognized conditions of human social action.
The methodological structuralist strategy outlined above can thus be read as extending the same attitude of agnostic pluralism to the ontological picture implied by linguistic, semiotic, or otherwise “relational” theories of structure (i.e., the view of the social world as, at its very core, “composed” of articulated semiotic practices) that Sewell and other agency-structure theorists prefer. That is, to paraphrase Sewell, from the methodological-structuralist point of view the problem with “theories of structure” is not that social scientists engage in arguments of the “structuralist type” (or that they construct models of society using structuralist methods), but that the reify such arguments, imagining that the social world is constituted by a complex of “structural” networks of semiotic objects and processes, or, in the usual network-structuralist rendering, that such analytic constructions as “networks of social relations” have an objective ontological existence and exert efficient causality on social actors (Gould 2003). The proper procedure is to engage in the manipulation of models of structure while keeping in mind that the actual logic or practice (for instance, the practical logics of meaning-making or forming and breaking friendships) may bear little or no relationship to the “semiotic” (or mathematical) logic of the model. Methodological structuralism, in this manner, partially brackets (but does not dismiss as unimportant) questions of the ontological composition of “the social” (which as we saw above only invite quite irresolvable meta-theoretical disputes).
Following a similar line of analysis, we can provide a different interpretation of the epistemological status of one approach that has recently emerged as a strong contender to replace the structure-agency framework first formulated by Giddens: field theory (Martin 2003). The most popular interpretation of field theory, as its name implies, is as a substantive theory of how society is “constituted” by various fields of striving with the term constitution understood in Giddens’s (1984) ontological sense; fields are “the parts” of society. However, this was not the way that one of the pioneers of field theory understood its primary contribution. As Lewin (1997, p. 201) remarks, “[f]ield theory is probably best characterized as a method; namely a method of analyzing causal relations and of building scientific constructs” (italics in the original). Thus, Lewin’s version—and therefore more recent reinterpretations—of field theory is thoroughly compatible with the Levi-Strauss-inspired methodological structuralist approach outlined above.
It is clear that Bourdieu followed (in his empirically oriented analyses for the most part) an epistemological strategy that was faithful to Lewin’s recommendation. Although Bourdieu’s work was also notoriously ambiguous on this account, sometimes referring to fields as substantive “worlds” that constituted the “parts” of differentiated societies (1990b, p. 73), in his empirical writings, he was clear that a “field” rather than being a “part” of a substantively conceived social totality was best thought of as an analyst’s construct. This construct, the product of the practical activity of the scientist (Bourdieu 1996b, p. 180), was ultimately to be judged according to its pragmatic utility in breaking with the naïve first person substantialism characteristic of the natural attitude. This is stated most clearly in a key chapter of The Rules of Art, appropriately entitled “Questions of Method” (Bourdieu 1996b, pp. 177–208).17 For Bourdieu, the notion of field rather than being used in the traditional “realist” manner to designate some pre-existing component of the social totality, instead initially “served to designate a theoretical posture, generative of choices of method (negative just as much as positive) in the construction of objects” (181).
As an example, we may take Bourdieu’s reconstruction of Weber’s discussion of the interrelationships between the figures of the Prophet, the Priest and the Magician (see Bourdieu 1987). Bourdieu’s analytic strategy here was to move away from Weber’s interactionist (e.g., realist) conceptualization of the relations between these agents, proposing instead “a construction of the religious field as a structure of objective relations” which allowing him to account for the interaction among the various positions in a way that sidestepped the problems of Weber’s “realist typology” (182). What Weber failed to see, according to Bourdieu, was that the “universes of [religious] specialists function like relatively autonomous microcosms, structured spaces ... hence spaces amenable to structural analysis ...” (204). Bourdieu’s pragmatic deployment of relationism as a self-conscious heuristic and his reference to the literary (1983) and religious (1987) fields as a constructed scientific objects are thus clearly at odds with any exclusivist ontologization of “relations” or “semiotic practices” as the true underlying “substance” of the social (e.g. Emirbayer 1997; Sewell 2005a).
The mere fact that the social space described here can be presented as a diagram indicates that it is an abstract representation, deliberately constructed, like a map, to give a birds-eye view, a point of view on the whole set of points from which ordinary agents (including the sociologist and his reader, in their ordinary behavior) see the social world. Bringing together in simultaneity, in the scope of a single glance—this is its heuristic value—positions which the agents can never apprehend in their totality and in their multiple relationships, social space is to the practical space of everyday life, with its distances which are kept or signaled, and neighbours who may be more remote than strangers, what geometrical space is to the ... ‘hodological space’ [life-space] ... of ordinary experience, with its gaps and discontinuities (emphasis mine).18
In summary, structural models and related representational tools are heuristically useful precisely because they allow the analyst to see what no actor enmeshed in the everyday flow of events and experiences could possibly conceive of: the “higher order” correspondences between the socially distributed totality of (conscious or unconscious) categorical systems used by a variety of different agents to navigate and partition their slice of the social world as well as the sets of practices applied to those classified objects and persons by all relevant actors (Mische 2003; Mohr and Duquenne 1997). In addition structural representations may allow the analyst to connect these last with the concrete settings during which those acts of category-practice coupling and decoupling take place (Mische and Pattison 2000). In this manner, the deterministic, complete, and “closed” (in a mathematical sense) “determinism” of the model must not be confused with any type of pronouncement regarding the actual possibility that such simple patterns of determinism or “completeness” take place in the actual world (Bourdieu 1990a, pp. 98–111), as has been the usual Sewell-inspired critique of Bourdieu.
Conclusion: theories versus models
This brings us to a final observation. While is customary to make the theory-observation distinction in sociology (e.g., Kiser and Hechter 1991) even if this is usually done to establish the argument that observation is “theory-laden,” a related differentiation is seldom made: that between theories and models. This is true in spite of the fact that the “partial autonomy” of models from both theory and observation, and their status as a “third” partially independent source of knowledge generation and theoretical development in science, has been one of the primary insights of recent philosophical and historical studies of scientific practice (Hughes 1997; Morrison and Morgan 1999; Morrison 1999). According to Morrison and Morgan (1999, p. 10), “... models occupy an autonomous role in scientific work” and they are best conceived of as “autonomous agents” that function as “... instruments of investigation.” Therefore “... it is precisely because models are partially independent of both theories and the world that they ... can be used as instruments of explorations of both domains” (italics in the original). I propose that while theories of structure partake of ontological and referential characteristics, models of structures do not. Second, the practical act of the construction of theories of structure in scientific practice is analytically and practically separable from the deployment of structural models to organize empirical materials.
Thus, taking this separation between theories and models as a point of departure serves to underscore the renewed need to heed the original Levi-Straussian warnings (1963a, f, 1976b) to keep a clear line of distinction between structural models and sociological theories of action and even ontological “theories of structure” in the Giddens-Sewell sense. This leads to the rather counterintuitive conclusion that structuralism as a method and “the theory of structure” in a metatheoretical sense may sometimes be radically different (and sometimes antithetical) projects. Giddens proceeded by eliding the Levi-Straussian distinction between structural models and structural theories, thus collapsing the language used to talk about the former into a general (quasi)ontological formulation of the latter. This implicitly denies the partial autonomy of models from theories. Sewell (2005a, b) follows Giddens in this respect, fixing some conceptual muddles left behind by Giddens but keeping his structuralist analysis at a purely ontological level. As a result of this, most researchers who engage in informal uses of “theories of structure” in contemporary sociological research are seldom careful enough to distinguish the self-conscious practical deployment of structural models as representational tools for the construction of analytic objects from general ontological pronouncements regarding structure in the Giddens-Sewell lineage, or either of the latter from empirical observations that may be brought to order by way of structural representations.
Exceptions to this pattern in recent work can certainly be noted. It is no surprise that the distinction between substantive theories and non-denotational models has been partially recovered by empirically oriented structuralists working in the sociology of culture today (e.g., Breiger 2000; Mohr and Duquenne 1997; Mische and Pattison 2000). For instance, Mohr and Duquenne (1997, p. 310) state that the “mutually constitutive” relationship between culture and practice—as an empirically observable phenomenon can be conceived in “structuralist” terms using lattice analysis, since the latter “provides an appropriate tool for formally representing the structural properties of such a linkage”; furthermore, and consistent with Morrison and Morgan (1999), they differentiate both observation and their structural model from their theory of practice. Here, in stark contrast to Giddens, who ontologizes the notion of structural properties as an inherent characteristic of real systems, the authors conceive of them more soundly as properties pertaining to a structural model in its status as a (relatively) autonomous tool for scientific representation of social reality.
Notice that by following this route the analyst can accept that mathematical models of structure provide structural representations of social reality that highlight an otherwise hard to perceive dual co-constitution of the practical and symbolic orders (or a macrolevel abstract organization of social ties), while remaining agnostic as to whether such a duality is an actual property of the actual world, which is a question relevant for institutional and practice theorists. In a similar way in the case of the network analysis of role structures, the analyst does not have to settle ex ante the issue or whether the “theoretical groups” uncovered by the model are actual groups in the phenomenological sense, which is also an empirical and not a theoretical issue (Bourdieu 1985).
As noted in the foregoing, any ontologization of the notion of structure—in particular those that remain close to the analytical divisions between “cultural structures” and “social structures” (Parsons 1973)—is bound to trap the theorist in ultimately irresolvable issues related to the “casual effect” (or “autonomy”) of one of these ontological levels on the other one (e.g., Alexander 2003). To escape the charge of diffuse idealism or materialist reductionism, most analysts are quick to adopt an equally unsatisfying casual egalitarianism, in which casual influences are seen as interacting mutually. As has been pointed out by critics of this approach (Archer 1982) this results in an elision of temporality from the picture, and a related under-specification of linkages from the micro to the macro realm. Yet I have argued that the there is no necessary connection between methodological structuralism and any of these metatheoretical positions.
This last is a claim that runs counter to most of the received wisdom on the subject starting with Giddens’s own interventions, where the theory of structure is usually seen as a “cure” for the idealism versus materialism problem. I have argued instead that methodological structuralism is concerned primarily with social-scientific representations of social reality, regardless of the “ultimate” constituents of that reality (symbolic, material, social, practical, etc.). This stance is also agnostically “pluralistic” in regard to issues regarding causation and determination (most structural models are by construction not “causal models” in the statistical sense) and leaves these as ultimately empirical and theoretical issues partially autonomous from practical modeling decisions. In this manner it is possible to make use of the idea of structure as a productive representational tool, without having to take a position such as Sewell’s (2005b) schema-resource duality, which is still situated in a culture/matter dualist ontological framework.
Furthermore, repeating Giddens’s mistake of embroiling questions of agency with questions of “structure” has cemented the almost knee-jerk tendency to ontologize structure as a quasi-material “thing” (seemingly standing in the way of agency) or conceived in a thoroughly uninformative “constraining and enabling” manner as analytically inseparable from agency. This has resulted in much undue energy being placed on first finding the most “agentic” ontological account of structure possible—with Sewell’s effort having done the lion’s share of the work in recent sociological research—rather than exploring the virtues and vices of different forms of structural representations of empirical phenomena (Breiger 2000). In consequence, the exploration of the properties and uses of structural models—outside of small communities of scholars involved in formal network analysis or formal approaches to the study of culture—for the representation of empirical sociological materials has been unduly hindered as a result.
The main upshot of methodological structuralism is that the question of agency—ultimately an empirical question—does not have to be settled before resorting to the use of structural models, especially when these models are seen as partially autonomous from both theories of structure and empirical materials, “mediating” between these usually opposed terms. In this manner, it is possible to combine methodological structuralism with a microtheory attentive to the agentic nature of practical action (Mische 2003; Mohr and Duquenne 1997) or with a theoretical outlook in which questions of agency and “reflexivity” only figure as a secondary concern and in which the materiality of the human body is paramount and action remains in a practical state (Bourdieu 2000; Wacquant 2004). This stance has much to recommend if we are to move beyond the usual antinomies and confusions generated by the concept of structure in contemporary social science, and if anything resembling a successful “(post)structuralism” can become a viable project in sociology.
This history appears to be in accord with Thompson’s (1989, p. 62) assertion that “[f]ew concepts in the social sciences are more basic and essential, and yet more ambiguous and contested, than that of structure.”
Bourdieu’s own attempt to go beyond Sartrean subjectivism and Levi-Straussian objectivism in Outline of a Theory of Practice (1977; French original: 1972) and in the later Logic of Practice (1990a; French original: 1980) during the same period (1972–1980), while having acquired equally influential status of late was not as immediately influential in Anglophone sociology as Giddens.
Giddens devoted an entire section of The Constitution of Society (1984, pp. 205–213) to a criticism is this “objectivist” idea of structure as it manifested itself in the work of some prominent American network structuralists of the time such as Bruce Mayhew and Peter Blau.
In contrast to network structuralism, structuralists in sociology who depart from Giddens’ structuration theory tend to emphasize the inherently cultural nature of structure as embodied in “virtual” cognitive schemas (Sewell 2005b, p. 151).
This division between the concrete and the abstract versions of structure itself reappears in mid-twentieth century linguistics, separating the “Anglosaxon” (like Radcliffe-Brown, empiricist) conceptualization of linguistic structure as a corpus of observable and “recordable utterances” proposed by Zelig Harris and the “American” structuralist school from language as an unobservable “system” proposed by members of the Prague school who themselves drew on the formative influence of Saussure. The problem with the empiricist definition of structure is, as Benveniste notes, that “this language realized in recordable utterances could be considered the manifestation of a hidden substructure” (Benveniste 1971, p. 14).
The term “methodological structuralism” was first used—to my knowledge—by Boudon in 1971 in reference to Levi-Strauss, in order to distinguish his approach from what he referred to as “philosophical structuralism,” which is closer to what I refer to as ontological structuralism below (Boudon quoted in Kurzweil [1980: 2]).
Sewell (2005b, p. 129) errs in thinking that Giddens’s notion of structure as rules fundamentally changes from a “Levi-Straussian” notion of structure as cultural rules analogous to the Saussurean langue in Central Problems in Social Theory to a more Wittgenstenian notion in the Constitution of Society. The Wittgenstenian critique of classic structuralism is already present in the earlier work in its entirety (see for instance Giddens 1979, pp. 37–38). In fact it would be easy—but outside of the scope of this article to demonstrate rigorously—to show that there is little that is fundamentally new at a conceptual level—outside of extending the structuration critique to other sociological thinkers (Goffman, Blau, et al.) and developing the linkages to Ericksonian psychoanalytic theory (e.g., the notion of “ontological security”)—in the 1984 exposition of the theory of structuration in comparison to that offered in the earlier book. In fact the 1979 work is in many ways a paragon of clarity of argumentation and analytical sharpness in comparison to the conceptual mélange offered in the Constitution of Society, and as such represents a much more useful source of theoretical exegesis regarding the conceptual origins of the theory of structuration than the latter tome.
For instance, the idea that structures do not have a “real” existence but only a “virtual” existence until they are “instantiated” by actors (a position that Giddens never abandons) is a clear-cut extension of the paradigmatic linguistic situation to the case of social action in general; it is straightforward to show that this analogy is not warranted, and furthermore leads to fundamental (and predictable) distortions of the nature of social action (see Bourdieu 1973: 58–59). However, what has not been adequately realized is that the illicit move is not due to some inherent flaw in the idea of structure—as argued by King (2000)—but to Giddens ’s illicit construal of properties of models (virtuality, lack of extension in time and space, etc.) as properties of real-world objects and processes that the new notion of structure comes to “denote.”
Bourdieu isolates Levi-Strauss as one of the most influential examples of the aforementioned fallacy; but as we have seen Giddens is responsible for a similar mistake in his ontologization of structure as rules (and resources) productive of the chronic instantiation of practices.
The original article was published in 1992 in the American Journal of Sociology.
For instance, among the “structures” considered by Sewell (2005b, pp. 140–151) are: modes of production based on private property in capitalist societies, institutionalized forms of labor organization, the “theological modes” that constrain religion in Christian societies, language, the nation state, military structures, financial structures, and a whole host of other heterogeneous entities and competences.
This important set of problems in Sewell’s account has already been noted by Ann Swidler (2001, p. 79), who remarks that thinking of “... everything from capitalism to handshakes as structured practice can be liberating, but it can also lead to trouble.” The reason for this has to with the tension that is introduced between the usual idea of a structure as implying a simple ordered organization and the newer idea of structure as containing contradictory elements characterized by contingency, unpredictability, and heterogeneity. Swidler notes that “... we can say a great deal about what any one part of such system implies about the other parts ...” when holding on the former view of structure. “But if there is multiplicity, multivocality... and contradiction between structures ... then the reinterpretation of structures may lead to its own dead ends.”
I acknowledge an anonymous reader for bringing this commonality to my attention.
For instance, in the prologue to The State Nobility, entitled “Social Structures and Mental Structures”—to take the clearest out of many possible examples—Bourdieu (1996a, p. 1) notes that it is “the goal of sociology to uncover the most deeply buried structures of the different social worlds that make up the social universe, as well as the ‘mechanisms’ that ensure their reproduction and transformation.”
I am skeptical of this argument for the necessary coupling between so-called quantitative methods and a “mechanistic” ontology. As Fararo (1989) has noted, the underlying mathematical model objects that form of the core of traditional quantitative methods can also be thought of as implying a non-substantialist “process-ontology” strictly antithetical to a mechanistic one.
A chapter that appropriately opens with an epigraphic quote from Lewin.
Richard Nice translates the French phrase espace hodologique as “traveling space” (but is judicious enough to leave the French original in brackets in the English translation). This translation is somewhat unfortunate, since it is clear that in this context Bourdieu is referring to Kurt Lewin’s term “hodological space,” which the latter used as synonymous to the individual “life-space” and “psychological space.” Rendering this term “traveling space” breaks the “family resemblance” that Bourdieu’s methodological structuralism bears to Lewin’s field theory (the only other major theorist in the social sciences who seriously dealt with the question of the relationship between formal models constructed by the analyst and real psycho-social processes that those models are designed to shed light on).
I would like to thank Ron Breiger, John Levi Martin, Michael Strand, and in particular the Theory and Society Ediros and reviewers for their very helpful comments and suggestions. This article benefited in more ways than I can count or remember from their feedback. All remaining errors and omissions are my responsibility.
- Abbott, A. (2001). Time matters: On theory and method. Chicago: University of Chicago Press.Google Scholar
- Alexander, J. (2003). The meanings of social life. Oxford: Oxford University Press.Google Scholar
- Archer, M. S. (1995). Culture and agency. Cambridge: Cambridge University Press.Google Scholar
- Benveniste, E. (1971). Problems in general linguistics. Miami: University of Miami Press.Google Scholar
- Bourdieu, P. (1973). The three forms of theoretical knowledge. Social Science Information, 12, 53--80.Google Scholar
- Bourdieu, P. (1977). Outline of a theory of practice. Translated by R. Nice. New York: Cambridge University Press.Google Scholar
- Bourdieu, P. (1984). Distinction: A social critique of the judgment of taste. Translated by R. Nice. Cambridge: Harvard University Press.Google Scholar
- Bourdieu, P. (1985). The social space and the genesis of groups. Theory and Society, 14, 723--744.Google Scholar
- Bourdieu, P. (1987). Legitimation and structured interests in Weber’s sociology of religion. In S. Lash & S. Whimster (Eds.), Max Weber, rationality, and modernity (pp. 119–136). London: Unwin Hyman.Google Scholar
- Bourdieu, P. (1990a). The logic of practice. Cambridge: Polity.Google Scholar
- Bourdieu, P. (1990b). In other words. Stanford: Stanford University Press.Google Scholar
- Bourdieu, P. (1996a). The state nobility. Stanford: Stanford University Press.Google Scholar
- Bourdieu, P. (1996b). The rules of art. Stanford: Stanford University Press.Google Scholar
- Bourdieu, P. (1998). Practical reason: On the theory of action. Stanford: Stanford University Press.Google Scholar
- Bourdieu, P. (2000). Pascalian meditations. Stanford: Stanford University Press.Google Scholar
- Collins, R. (1981). Levi-Strauss’ structural history. In Sociology since midcentury: Essays in theory cumulation (pp. 109–132). New York: Academic Press.Google Scholar
- Dosse, F. (1997). History of structuralism: Volume I, the rising sign, 1945–1966. Minneapolis: University of Minnesota Press.Google Scholar
- Douglas, M. (1968). The social control of cognition: some factors in joke perception. Man, 3(3), 361--376.Google Scholar
- Friedman, M. (1953). The methodology of positive economics essays in positive economics (pp. 3--43). Chicago: University of Chicago Press.Google Scholar
- Frisby, D. (2002). Georg Simmel. New York: Routledge.Google Scholar
- Gartman, D. (2007). The strength of weak programs in cultural sociology: a critique of Alexander’s critique of Bourdieu. Theory and Society, 36(5), 381--413.Google Scholar
- Giddens, A. (1976). New rules of the sociological method. New York: Basic Books.Google Scholar
- Giddens, A. (1979). Central problems in social theory. Cambridge: University of Cambridge Press.Google Scholar
- Giddens, A. (1981). Agency, institution, and time-space analysis. In K. Knorr-Cetina & A. V. Cicourel (Eds.), Advances in social theory and methodology: Toward an integration of micro-and macro-analyses (pp. 161–174). Boston: Routledge and Kegan Paul.Google Scholar
- Giddens, A. (1984). The constitution of society. Berkeley: University of California Press.Google Scholar
- Geertz, C. (1973). The interpretation of cultures. New York: Basic Books.Google Scholar
- Glucksmann, M. (1974). Structuralist analysis in contemporary social thought: A comparison of the theories of Claude Lévi-Strauss and Louis Althusser. Routledge/Thoemms Press.Google Scholar
- Hacking, I. (1983). Representing and intervening: Introductory topics in the philosophy of social science. Cambridge: Cambridge University Press.Google Scholar
- Kiser, E. & Hechter, M. (1991). The role of general theory in comparative-historical sociology. American Journal of Sociology, 97(1), 1--30.Google Scholar
- Kroeber, A. L. (1943). Structure, function and pattern in biology and anthropology. The Scientific Monthly, 56, 105–113.Google Scholar
- Kurzweil, E. (1980). The age of structuralism. New York: Columbia University Press.Google Scholar
- Levi-Strauss, C. (1963a). Social structure. In Structural anthropology (pp. 277–323). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1963b). Structural analysis in linguistics and anthropology. In Structural anthropology (pp. 31–54). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1963c). Language and the analysis of social laws. In Structural anthropology (pp. 55–66). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1963d). The structural study of myth. In Structural anthropology (pp. 206–231). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1963e). Do dual organizations exist? In Structural anthropology (pp. 132–163). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1963f). Postscript to chapter XV. In Structural anthropology (pp. 324–345). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1963g). Structural anthropology. New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1966). The savage mind. Chicago: University of Chicago Press.Google Scholar
- Levi-Strauss, C. (1969). The elementary structures of kinship. Boston: Beacon Press.Google Scholar
- Levi-Strauss, C. (1976a). The scope of anthropology. In Structural anthropology: Volume II. New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1976b). The meaning and use of the notion of model. In Structural anthropology: Volume II (pp. 71–81). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1976c). Structural anthropology: Volume II. New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1985). The lessons of linguistics. In The view from afar (pp. 138–147). New York: Basic Books.Google Scholar
- Levi-Strauss, C. (1987). Introduction to the work of Marcel Mauss. Taylor & Francis.Google Scholar
- Maki, U. (2002). Fact and fiction in economics: models, realism and social construction. New York: Cambridge University Press.Google Scholar
- Marsden, P. & Laumann, E. (1984). Mathematical ideas in social structural analysis. The Journal of Mathematical Sociology, 10(3), 271--294.Google Scholar
- Mische, A. (2003). Cross-talk in movements: Reconceiving the culture-network link. In M. Diani & D. McAdam (Eds.), Social movements and networks: Relational approaches to collective action (pp. 258–280). London: Oxford University Press.Google Scholar
- Morrison, M. (1999). Models as autonomous agents. In M. S. Morgan & M. Morrison (Eds.), Models as mediators: Perspectives on natural and social science (pp. 28–65). Cambridge: Cambridge University Press.Google Scholar
- Parsons, T. (1973). Culture and social structure revisited. In L. Schneider & C. Bonjean (Eds.), The idea of culture in the social sciences (pp. 33–46). Cambridge: Cambridge University Press.Google Scholar
- Piaget, J. (1970). Structuralism. New York: Basic Books.Google Scholar
- Ridgeway, C., Boyle, E., Kuipers, K., & Robinson, D. (1998). How do status beliefs develop? The role of resources and interactional experience. American Sociological Review, 331--350.Google Scholar
- Saussure, F. (1966). Course in general linguistics. New York: McGraw-Hill.Google Scholar
- Schatzki, T. (1996). Social practices: A wittgensteinian approach to human activity and the social. Cambridge: Cambridge University Press.Google Scholar
- Sewell, W. H. (2005a). Refiguring the ‘social’ in social science: An interpretivist manifesto. In Logics of history (pp. 318–372). Chicago: University of Chicago Press.Google Scholar
- Sewell, W. H. (2005b). A theory of structure: Duality, agency and transformation. In Logics of History (pp. 124–151). Chicago: University of Chicago Press.Google Scholar
- Sewell, W. H. (2005c). Three temporalities: Toward an eventful sociology. In Logics of history (pp. 81–123). Chicago: University of Chicago Press.Google Scholar
- Spillman, L. (1995). Culture, social structure, and discursive fields. Current Perspectives in Social Theory, 15, 129–154.Google Scholar
- Stones, R. (2005). Structuration theory. Palgrave McMillan.Google Scholar
- Swidler, A. (2001). What anchors cultural practices. In T. R. Schatzki, K. K. Cetina, & E. Von Savigny (Eds.), The practice turn in contemporary theory (pp. 74–92). New York: Routledge.Google Scholar
- Wacquant, L. (2004). Body and soul: Notes of an apprentice boxer. Oxford: Oxford University Press.Google Scholar
- White, H. P. (1992). Identity and control. Princeton: Princeton University Press.Google Scholar
- Wolf, K. H. (1959). (Ed.), Georg Simmel, 1858–1918: A collection of essays, with translations and a bibliography. Columbus: Ohio St. Univ. Press.Google Scholar