Toward a really temporalized theory of event: A Luhmannian critique and reconstruction of Sewell’s logics of history

The inner contradiction in Sewell

Based on an examination of the representational works of Immanuel Wallerstein, Charles Tilly and Theda Skocpol, Sewell (2005: Ch. 3) argued that historical sociology until now did not really think ‘historically’. Those historical studies were dominated, consciously or unconsciously, by teleological or experimental temporality. In other words, rather than taking eventful temporality (i.e. history as transformation) seriously, they either substituted teleology for history or froze history. Accordingly, events were conceptualized as nothing more than ‘markers on the road to the inevitable future’, ‘only effects, never causes of change’, or ‘artificially interchangeable units’ (Sewell, 2005: 88, 91, 95).

Against these ‘a-historical’ perspectives of analysis, Sewell appealed to and promoted an eventful sociology that assumes social relations characterized by path dependency, temporally heterogeneous causality and global contingency. According to Sewell (2005: 102),

… an eventful concept of temporality assumes that contingency is global, that it characterizes not only the surface but the core or the depths of social relations. Contingent, unexpected, and inherently unpredictable events, this view assumes, can undo or alter the most apparently durable trends of history.

Sewell explained, ‘[a]n assumption of global contingency means not that everything is constantly changing but that nothing in social life is ultimately immune to change’ (2005: 102). Events can change ‘not only the balance of causal forces operating but the very logic by which consequences follow from occurrences or circumstances’, because events bring about such historical changes ‘in part by transforming the very cultural categories that shape and constrain human action’ (Sewell, 2005: 101).

So far, so good. But in my view, a contradiction emerges when we scrutinize both Sewell’s idea of eventful temporality and his definition of events as a ‘relatively rare subclass of happenings that significantly transforms structures’ (Sewell, 2005: 100). If we assume global contingency, how can we classify only some rare happenings as events rather than treat every happening as a contingent event that could be otherwise? Even if one can accept that this is not a contradiction in a strict sense, one cannot but ask if a not fully temporalized conceptualization of event can fulfill the ideal opened and promised by the idea of eventful temporality. From my point of view, it is not radical enough to admit only (all?) social relations (this implies: structure!) as contingent ² and only some rare happenings with duration as events. It is also doubtful that every contingent event (or happening) must be unexpected, especially that the transformation of cultural schemas results necessarily and only from unexpected flows of resources (Sewell, 2005: 217).

Sewell’s (2005: Ch. 8) illustrative analysis of the case of the French Revolution, especially the taking of the Bastille, shows clearly that for him every event can be identified only in reference to a (existing and given) structure (see Sewell, 2005: 210–211, as well). In Sewell’s own words: ‘[t]o understand and explain an event, therefore, is to specify what structural change it brings about and to determine how the structural change was effectuated’, ³ and ‘the key to an adequate theory of the event is a robust theory of structure’ (2005: 218–219). This is not necessarily wrong or inadequate, just as I will gladly embrace Sewell’s idea that ‘[a] proper appreciation of synchrony is the secret ingredient of effective diachronic history’ and ‘no account of a historical transformation can be cogent unless it performs a dialectical oscillation between synchronic and diachronic thinking’ (Sewell, 2005: 184). But the citation above indicates that event in Sewell’s approach is only secondary and does not constitute an independent ‘theoretical category’. The other indubitable evidence is that the concept of event plays no role in Sewell’s theory of structure (2005: Ch. 4), even in his discussion about the transformation of structure. What he emphasized is always a ‘multiple, contingent, and fractured conception of society and of structure’ (Sewell, 2005: 140).

The problem, then, is if we treat event only as a secondary rather than the constitutive category of the social, how can we conceive of the ultimate element of the social that is both ontologically justifiable and compatible with our theory of structure and event (cf. Lipuma, 2006: 67)? Can Sewell’s conceptualization of event really match the basic idea that time is a constitutive dimension of social life rather than only a parameter, a context or an environment (Adam, 1990: 30; Griffin, 1992: 406–408; McLennan, 1990: 133–134)? Sewell’s (2005: Ch. 10) effort to refigure ‘the social’ shows clearly that he is neither shy of nor unfamiliar with ontological questions. But his suggested answer – the metaphor of language (game) and the metaphor of built environment – does not answer this question directly, not to mention solving the problem in relation to it satisfactorily.

Sewell (2005: 229) was aware of this problem and made a theoretical decision for the version of duration: ‘In spite of the punctualist connotations of the term, historical events are never instantaneous happenings: they always have a duration, a period that elapses between the initial rupture and the subsequent structural transformation’. ⁴ We should not be surprised to find a preference for ‘structuralism’ ⁵ in Sewell’s formulations, rhetoric and metaphors (such as small/big, micro/macro, cumulative, gradual, details, sedimented, accumulated; see e.g. Sewell, 2005: 110, 199, 219–220, 364–365), despite the fact that he emphasized both continuity and change, and reproduction and transformation (e.g. Sewell, 2005: 9, 102, 272–273).

Sewell (2005: 164) saw that ‘the important theoretical question is not whether culture should be conceptualized as practice or as a system of symbols and meanings, but how to conceptualize the articulation of system and practice’, and paid attention to the articulation between different structures and different semiotic practices. But he did not seek an analogous theoretical solution with regard to the relation or distinction of structure and event, which represent – according to the discipline’s tradition – the core category of sociology and history respectively, and make possible a further theorization of historical sociology.

The obstruction to radicalization

Sewell’s hesitation to qualify every happening or occurrence as event is not without good reason. ⁶ Sewell (2005: 183) warns that although the perspective of ‘history as transformation’ thinks in principle more historically than the perspective of ‘history as context’, ‘a history that recounted a series of changes over time but failed to indicate the distance of the lifeworld being described from the present would be dismissed out of hand as “anachronistic”’. By contrast, ‘a historical work that makes no effort (or only the most passing effort) to explicate or explain a historical transformation but portrays effectively the context of some past lifeworld can be hailed as a masterpiece’.

For Sewell (2005: 100), as for most historians, not every happening deserves the title of ‘[historical] event’, ⁷ just as the fact that not every fact can automatically be a historical fact. This has been the dominant consensus among historians, at least since Carr (2001[1961]: Ch. 1) elaborated this point clearly and distinctly, and even since the postmodern challenge. According to this conventional wisdom, it sounds quite reasonable when Sewell (2005: 226–227) promotes the idea that the term ‘event’ can only signify ‘an occurrence that is remarkable in some way’ or even one that must ‘change the course of history’ in some sense, despite the fact that Sewell recognized clearly that not only surprising breaks with routine practice ‘actually occur every day’, but that ‘all social practices undergo constant revision even in the course of reproduction, and the accumulation of small revisions may eventually result in significant transformation’, so that ‘the difference between an act of reproduction and an event is always a difference in degree, not in kind. Distinguishing transformative events from ordinary implementations of structure is necessarily a matter of practical judgment’ (Sewell, 2005: 211).

Such a narrow concept of event does not necessarily lead to a serious deficiency in practice. But whether such a conceptualization of event is adequate on the level of theory deserves further consideration. ⁸ Can we really achieve in this way the following goal that Sewell (2005: 199) tried to reach when he praised Sahlins, who: ‘transform[ed] the unequal and radical opposition between structure and event, which makes the two categories hostile and mutually incomprehensible, into a more balanced relation, in which each category implies and requires the other’?

I analyze this problem step by step in the following sections. Sewell (2005: 8) rightly pointed out that ‘[t]he conceptual vehicle by means of which historians construct or analyze the contingency and temporal fatefulness of social life is the event’. But Sewell’s (2005: 7, 112–113) understanding of contingency seems too narrow: ‘the outcome of any action, event, or trend is likely to be contingent, that its effects will depend upon the particular complex temporal sequence of which it is a part’. In my view, Sewell understood contingency only from the perspective of ‘path dependency’ and equated ‘contingent’ with ‘dependent’, ⁹ as the citation shows.

This limited understanding of contingency and an accompanying underestimation of its role in history are reflected in the fact that Sewell treated contingency as a central concept that needs a serious theoretical interpretation only in Chapter 3. When he developed his theory of the event (Ch. 7), explored how the historical events could transform structures (Ch. 8) and answered the ontological question of what the social meant (Ch. 10), he was not as concerned about contingency.

Some would assume that this is only a minor problem. However, behind this seemingly superficial problem lies the central problem of temporality that correlates closely with the question of what ‘thinking historically’ and ‘logics of history’ mean. We must ask: Where is our theoretical starting point if we need to consider all those problems that the logics of history involve? What is the last unit of historical analysis? Should this unit itself be a temporalized concept? What is the ultimate constitutive element of the social? Should we take contingency or necessity as a departure? How do we conceive of time and understand temporality? Should we assert multiple and heterogeneous temporalities or claim for a single linear temporality? Should we focus on the recursive or the recurrent character of history? What kind of role does self-reference play in history? Are historical processes determined by homogeneous causal laws that transcend time and space or are they guided by heterogeneous or even circular causality?

Although Sewell provided many valuable threads to contemplate these problems, he failed to build a theory of history that satisfactorily solves them in a logically consistent manner. By contrast, Luhmann supplied us with many useful conceptual instruments that potentially solve these problems. My aim in the following is to push Sewell’s achievement a step further with the help of Luhmann’s theoretical insights.

Facing the ontological question

Sewell (2005: 319) mentioned that sociologists usually define the social in a tautological way, ‘with the terms “social structure” or “social relation” appearing somewhere in the definition’. In place of a tautological definition he suggested that the social be defined as:

… the complex and inescapable ontological ground of our common life as humans. It is best understood as, first, an articulated, evolving web of semiotic practices (this is the language metaphor) that, second, builds up and transforms a range of physical frameworks that both provide matrices for these practices and constrain their consequences (this is the built environment metaphor). (Sewell, 2005: 369)

Sewell overlooked that Luhmann (1995: Ch. 4) had previously provided an answer that was neither tautological nor metaphorical: communication as a synthesis of information, utterance and understanding. ¹⁰ Communication appears when and only when an observing system uses the difference between information and utterance to understand. This means that an observing system understands this difference by projecting it onto past events, and onto the observed system (Stichweh, 2000: 10). ¹¹ Doubtless, language usually plays an important role in communication. According to Luhmann (1995: 152–153), language enables the communication to differentiate from the context of perception. The reflexivity that is possible with the help of language can compensate for the risk of greater complexity and sharper selectivity. But communication is not limited to linguistic communication, not to mention that language itself is not communication. For Luhmann (1997b: 72), ‘[e]ach communication identifies itself by referring to past communication and by opening a limited space for further communication’.

There are some important agreements and disagreements between Sewell and Luhmann. Like Luhmann, Sewell (2005: 328–346) not only stressed the reflexive capacity of language but also objected to stretching the notion of language to cover all semiotic practices. However, according to Luhmann (1995: 139), Sewell was still accustomed to the use of the metaphor of transfer (of information or meaning) that implied too much ‘ontology’ (schemata of thing) and an asymmetrical direction. He did not and could not see the symmetrical, self-referential (and therefore recursive) character of communication as the ultimate constitutive element of the social because he supported an action theory with an imagination of Subject. ¹²

For Luhmann (1981a,b), the concept of action did not presume the concept of subject. The concept of subject served only as a theoretical refuge. The motive or the intention was actually a later product of attribution under social pressure. By attributing the actor and the definition of situation, the meaning of the action gained a stability of meaning which this action as event could not have. This created a time-binding effect that did not exist in the event itself and expanded the scope of connections beyond the immediate intention of the actor. The time horizons that could be integrated in the event of action were expanded exactly in this way.

This leads us directly to another crucial difference: time plays no role in Sewell’s discussion on the social – although Sewell as a historian highlighted temporality again and again. By contrast, Luhmann (1988: 33, 276–277, 282–284; 1995: 11, 47, 215, 286–292; 2004: 85) always underlined the fact that communication is, as seen from the perspective of time, an event that disappears as soon as it appears. It is not surprising that Sewell did not take the problem of time seriously in his ontological discussion, for time had never played a role in ontology, even though ‘[t]here is no theory of temporality without an ontology’ (Abbott, 2001: 232). ¹³ In order to correct the oversight of time, Luhmann (1981a, 1995) revived the temporal thinking of Luc de Clapiers, Marquis de Vauvenargues (1715–1747), appropriated the concept of event from Whitehead (1919, 1979, 1982) and emphasized that only this temporalized element, rather than object, is an adequate candidate for the basic ‘ontological’ constitutive element. ¹⁴ As he wrote, ‘[o]nly as event with a fixed time-point … can the individual action be elementary and sufficiently isolated, to be disposed of by the changing combinations, open expectations and ex post reinterpretation’ (Luhmann, 1981a: 117; my translation).

But Luhmann did not claim that events are ontologically given entities waiting for discovery or observation (cf. Mead, 1980: 20–21, 26). Rather, his ‘post-ontological’ ¹⁵ position only meant that whether an event is an event depends on which system reference (i.e. which distinction of system and environment) is chosen. ¹⁶ According to the concept of autopoiesis, ‘[t]he elements and structures of a system exist only as long as it manages to maintain its autopoiesis’ (Luhmann, 2004: 81). It is worth comparing Sewell’s (2005: 121–122) seemingly similar but actually different view: ‘the temporality of the theoretical category “event” is not self-evident but rather must be constructed theoretically in relation to the time-scale of the processes being studied’. ¹⁷ Sewell ‘ontologically’ denied the existence of an ‘eventful temporality’ and therefore emphasized the role of analytical construction in the issue of the eventful temporality. By contrast, Luhmann’s position is not based on ‘analytical realism’ but on ‘operative constructivism’. ¹⁸ There are systems (including their events) that really exist and are not imaginary or only ‘constructions’ of observers but constructed by those systems’ operations themselves. ¹⁹ In Luhmann’s (2004: 81) words, ‘the concept of autopoiesis shifts the idea of a self-referential make-up to the level of the elementary operations [i.e. events] of the system’. ²⁰ This means that ‘everything used as a unit by the system, whether its elements, its processes, or the system itself, has to be constituted by the system’ (Luhmann, 1990b: 115).

Some important corollaries would be derived from this post-ontological perspective. Multiple constructions and multiple belongings of events (keyword: poly-contexturality, see Günther, 1979) are not only possible but even normal, ²¹ especially in the modern functionally differentiated society. ²² Sewell (2005: 211) also gave an ambiguous presentation of this significant phenomenon when he discussed the determination of happening as event or reproduction according to the view of multiple structures – this seems contradictory to the traditional logic that is deeply influenced by ontological thinking. However, Sewell focused primarily on the structures and did not develop the possibility of multiple belongings into a complete theory. A more balanced view between structure and event would have considered the plurality of both structure and event more thoroughly (Sewell, 2005: 261). This would enable the plurality of temporalities that were not limited to trends, routines and events as he discussed (Sewell, 2005: Ch. 9, 273 esp.).

When Sewell (2005, Ch. 1: 10 esp.) disclosed historians’ implicit or working theories about social temporality, he rightly emphasized that ‘[t]emporal heterogeneity implies causal heterogeneity’. But it is inadequate to me that Sewell’s (2005: 280) ‘logics of history’ did not include – at least, did not emphasize – the temporal heterogeneity. Sewell highlighted instead the causal heterogeneity and subsumed temporal heterogeneity under it. This reverse made Sewell somewhat blind to the important problem of plural and heterogeneous temporalities outside trends, routines and events. ²³ But if Sewell (2005: 211) had admitted that a happening such as divorce could be ‘simultaneously an event from the point of view of the local family culture and an implementation of structure from the broader viewpoint’, ²⁴ I think he would have willingly admitted that every structure (and therefore, every system) has its own time and tempo (including trends, routines and events) too.

The fundamental question of time

Given the problematic of heterogeneous temporalities, we now approach the fundamental question: how to understand time (cf. Abbott, 2001: Ch. 7; Adam, 1990)? Based on the philosophical ideas of Edmund Husserl and Henri Bergson, Luhmann (1979: 11) pointed out that ‘[a]ll human experience of time has as its ultimate ground the experience of duration despite changing impressions. This ultimate ground, whatever it may be, lends itself to interpretation from two conflicting points of departure: duration and variation.’ That is to say, something can be identified either as an event that is fixed at a point in time or as state (in German, Bestand), which persists regardless of change over any points in time. ‘Neither the classical notion of time, related to the present, nor the modern notion, related to instants, is adequate’ (Luhmann, 1979: 12). We cannot sufficiently and adequately grasp time unless we combine these two mutually exclusive perspectives of time. For ‘to be able to conceive of events, change, continued existence, the present or even time, there has to be becoming and disappearing’ (Adam, 1990: 39; cf. Abbott, 2001: Ch. 7).

The two different perspectives of time, duration/variation (or in Sewell’s terms, continuity/change) hint at structure, because structure regulates the relation between reversibility and irreversibility (i.e. invariance and ‘change’ as rupture). ²⁵ According to Luhmann (1995: 345),

… one can speak of change only in relation to structures. Events cannot change, because there is no duration between their emergence and their passing away in which something ‘eventlike’ exists and can continue despite change … Only structures keep what can be continued (and therefore changed) relatively constant. Despite the irreversibility of events structures guarantee a certain reversibility of relationships.

An observer can ‘distinguish between constant structure and eventful (in German, ereignishaft) operations, just like between the immovable and the movable, but only when he makes this distinction can he register structural changes’ (Luhmann, 2004: 82, with my revision; see also 1995: Ch. 8). In other words, we cannot identify change, irreversibility or discontinuity unless we do so by referring to structures. ²⁶

This hints further at the problem of differentiation between system and environment that lies behind the multiple, heterogeneous temporalities (cf. Mead, 1980: 24–25, 38–40). It is precisely the case indicated by Luhmann (1979) when he said:

… as soon as systems differentiate from their environment by formulating boundaries, problems about time occur … For not all relationships between system and environment can be instantaneous one-to-one correlations; rather the maintenance of the difference, in more complex systems at least, necessitates detours, and these take time. (1979: 10–11; see also Luhmann, 1982: 230; 1995: 41–52)

In other words, time is the construct of the observer as system. ²⁷ The temporality varies with the reference of (observing) system. However, the multiple constructions of time by different systems deny neither simultaneity (in German, Gleichzeitigkeit) of the system and its environment nor the possibility of construction of an ‘objective time’. ²⁸ For ‘everything that is happening is happening in the present. This also means that everything that is happening is happening at the same time. Also past and future are always, and only, simultaneously relevant; they are time-horizons of the respectively present operations, and are discernible as such only in the present’ (Luhmann, 2004: 82, with my revision; see also 131). ²⁹ When rephrasing Alfred Schütz’s (1980[1932]: 102–107) emphasis on ‘growing old together’, we could say that all systems ‘live and grow old together in a common and ever-present continuity’. By relying on a process of ‘intersubjective’ construction, it is possible to construct an ‘objective time’ as a continuum of points which is not only the same for all systems, but keeps neutral the distinction between duration and variation (Luhmann, 1979: 11, 14). ³⁰

By introducing Luhmann’s more complex perspective about temporality into historical research, we could handle more complexities than Sewell’s original framework. Luhmann allows every system to have its own trends, routines and events. The methodological warning of Luhmann’s view is that one must be fully aware of which system reference is chosen when one observes and describes anything, including temporality of course. This leads us back to the problem of structure and to a point that must be discussed here. My reconsideration and reconceptualization of an event aim not only to solve the ontological problem of the basic element of the social but also the problem of the articulation between structure and event. In Luhmann’s above-mentioned discussion of duration/variation and state/event, we can see a strong affinity with the distinction of structure/event (or operation). Unfortunately Luhmann did not develop this theme systematically.

Before I unfold my discussion in the following sections, I want to highlight another important distinction that would be helpful for constructing a complete theory of event and reducing the doubt about the punctuated event as the elementary unit. This is the distinction between the point-like present and the enduring present. Luhmann (1995: 78; see also 1981b) explained that:

… the time span between past and future in which a change becomes irreversible is experienced as the present. The present lasts as long as it takes for something to become irreversible. On closer inspection one sees that two presents are always simultaneously given and that only the difference between them creates the impression of the flow of time. One present appears as punctual … the other present endures and thereby symbolized the reversibility that can be realized within all meaning systems. ³¹

I suggest the development of two complementary notions of event in analogy to the two notions of present. ³² We can combine event in Luhmann’s version and event in Sewell’s version into a congruous theory of event. On the one hand, as Luhmann emphasized, every event – no matter whether it is natural, psychic or social – is connected with a specific time-point and vanishes as it emerges, just like the present as punctuated instant. This point-like event is the basic ontological element. On the other hand, there is also the ‘Event’ that consists not only of many punctuated events, but is ongoing and has duration because the change is not yet finished, and it is still uncertain or reversible. In this case we are situated in the present that is still a reversible duration. ³³ Only with a reference to structure can the observer (including the observed system itself) ascertain whether the Event has generated an irreversible change and therefore is over. ³⁴ The usually called ‘historical event’, or event as transformation of structure according to Sewell’s definition, is an Event in the sense that is paradoxically simultaneously changing and enduring. If we understand this, we will not be surprised that some historians often like to say that the great Event, such as the French Revolution, lasts longer than normally recognized or even never ends – as seen from the perspective of its effects and consequences (cf. e.g. Furet, 1981; Suter & Hettling, 2001: 28). ³⁵ So, we could rephrase Sewell’s idea and say that the beginnings and endings of an Event are not self-evident but depend on the perspective of the observer as well as on the structure chosen as a reference.

It is nothing new to see the distinction of event/Event (see e.g. Morin, 1992[1977]; Wilden, 1980[1972]) or the macro-/micro-event (Rölli, 2004) as a theoretical desideratum. However, we can only give the twin concept of event/Event a thorough theoretical justification and interpretation based on Luhmann’s point of view. According to this view, any trivial and small happening must be qualified as event, but only a rarer subclass can be qualified as Event. Such a qualification depends on the perspective of the observer.

The central problem in social theory

In the above discussion there is a similarity between the ‘punctuated present/enduring present’ and the ‘event/state’, or ‘variation/duration’. There is also an affinity with the event/structure. One characteristic of structure is ‘duration’ or ‘relative invariance’. From the above discussion it is clear that the feeling or impression of duration is gained due to reversibility. Like the present, time itself is originally given in a fuzzy manner if it is given only as a (reversible and irreversible) change. ³⁶ But time leaves room for a transformation of irreversibilities to reversibilities of a higher order and vice versa. ‘The possibility of return or restoration does not contradict time, but rather superimposes itself on an “in itself” irreversible temporal course’ (Luhmann, 1995: 42–43).

Whether a change is reversible or irreversible depends on structure. As Luhmann (1995: 44, with my revision) said,

… structures hold the reversibility of time fast because they hold open a limited repertoire of possibilities for choice … Processes, by contrast, mark the irreversibility of time. They are composed of irreversible events. They cannot run backwards … The difference between structure and process reconstructs the original (= environmentally conditioned) difference between reversibility and irreversibility within a time that is ordered irreversibly.

It is self-reference that ‘enables one to return to earlier experiences or actions’. Thanks to the structures, we often think that ‘a mistake can be undone. The conclusive finality of an action can be forestalled or suspended by a presented intention, which still has yet to become irreversible’ (Luhmann, 1995: 79, with my revision). Why do we think so? Because we insist that things should go as we expect in spite of our recognition of the facts and the possibilities of disappointment.

From Luhmann’s (1985: Ch. 2; 1995: Ch. 8) perspective social structures can only be structures of expectations. ³⁷ However, social structures consist not only of normative, but also of cognitive expectations. In the case of normative expectations we insist that we are right and stick to our original expectation. But we tend to modify our earlier expectation when we adopt an attitude of cognitive expectation. No matter the structure involves normative or cognitive expectation, it can serve as an orientation for the communicative event because it makes possible a determination of the next event by constraining other available possibilities. The so-called stability of structure is at the level of expectation, or more often at the reflexive level of expectation, rather than at the level of fact or explicit behavior.

This touches on an important point and a critical difference between my view and Sewell’s. I agree with Sewell that the key to an adequate theory of the event is a robust theory of structure. But I want to add another point: either the theory of event or the theory of structure must be able to explain not only both the reproduction and the transformation of structure (cf. Clemens & Cook, 1999; Orren & Skowronek, 1994) but also be able to explain both the slowly accumulated transformation and the sudden transformation (cf. Abbott, 2001: 234–235). Sewell’s theory of event is in a sense irrelevant to or even determined by his theory of structure. ³⁸ Similarly, Sewell’s standpoint seems to oscillate between accumulated and sudden transformation. In his explicit argument, Sewell emphasized ruptures or transformations caused by Events and hence portrayed an image of sudden change. But in his formulations, he usually showed a perspective of change by accumulation, or by series or cascades of events. Appropriating Sahlins’s (2005) title, I would say that Sewell highlighted only one type of transformation, one type of ‘structural work’: ‘how microhistories become macrohistories’ (cf. Steinmetz, 2008: 540).

For example, Sewell (2005: 228) emphasized that:

… a single, isolated rupture rarely has the effect of transforming structures because standard procedures and sanctions can usually repair the torn fabric of social practice. Ruptures spiral into transformative historical events ³⁹ when a sequence of interrelated ruptures disarticulated the previous structural network, makes repair difficult, and makes a novel rearticulation possible.

Therefore, Sewell (2005: 226) thought that ‘[e]ven the accumulation of incremental changes often results in a buildup of pressures and a dramatic crisis of existing practices rather than a gradual transition from one state of affairs to another’. In principle I am not against Sewell’s argument if he does in fact mean ‘rarely’ and ‘often’. These arguments reflect on the one hand that the distinction of event and Event is theoretically necessary ⁴⁰ and on the other hand that Sewell neither explained nor believed sudden structural transformation by way of event (not Event).

This weakness results from Sewell’s oversight regarding the self-referential character of history and social reality. With regard to epistemology, this oversight is a fundamental deficiency because:

… the concepts of observation and self-reference imply each other. For, on the one hand, observers can only observe when they can distinguish between themselves and their observation instruments, their distinctions and designations, and so manage not to keep confusing themselves with their objects. On the other hand, this is precisely what self-reference is needed for. (Luhmann, 2004: 87)

With regard to history, as Arthur C Danto’s (2007[1968]) analysis of the ‘narrative sentence’ showed, it is a normal phenomenon that ‘a subsequent event transforms a prior one into a cause, therefore that a sufficient condition for the earlier event is produced later than the event’ (Ricœur, 1984: 143–149). For the meanings of past events change with the emergence of every present, although the past events themselves are irrevocable (Mead, 1980: Ch. 1). This elementary logic of history means that, if it is translated into the terminologies of systems theory, self-reference and circular causality are ubiquitous in history and in the operations of society (cf. Abbott, 2001: Ch. 1, 226–227). ⁴¹

Sewell (2005: 236) has a professional sensitivity about this but did not attempt to build a theory compatible with it as he said that:

… making sense of the taking of the Bastille requires us to reconstruct the sequence of action and interpretation that led from the rupture … to the new articulation … While this process began at the Bastille and in the surrounding streets on July 14, it was not until some days later, in the meeting hall of the National Assembly in Versailles, that it can be said to have been definitely achieved. ⁴²

As this case shows, it is very often that we can only identify or confirm a structural change or transformation ex post facto, just as we can only ‘retrospectively attribute’ the character of a ‘world-historical event’ to the French Revolution or Columbus’s arrival in the Americas (Stichweh, 2007: 143). This has nothing to do with the ‘post hoc fallacy’. In addition, structural change can happen by way of a single event whose transformative effect is doubtlessly based on earlier events. Nevertheless, this structural change doesn’t need to presuppose a continuity of similar or interrelated events without any break in the point-liked, ongoing present.

There is still one nuance to elaborate. Sewell saw the emergence of new articulation as a necessary component of Event as transformation of structure. I will not debate this, but will point out that there would be a sudden rupture or the change could suddenly become irreversible if the actor as an observer thought or realized that a change had occurred and was irreversible. ⁴³ This means that the Event might still be ongoing and not finished, but the structure had definitely been changed by an event, although a consolidated new structure is not clear yet. ⁴⁴ If we consider that social change often leads to an overlay of new and old structures, rather than a total substitution of new structures for old structures (Orren & Skowronek, 1994: 323–325; Stichweh, 2007: 135), an ambiguity may be a normality for contemporaries. By contrast, the historian with the advantage of hindsight might be less bothered by this. ⁴⁵ But at any rate, a structural change could result logically either from a longtime accumulation or from a sudden accident. The change could be very sudden because the structure lies at the level of expectation rather than behavior.

The ubiquity of self-reference

Sewell gave his attention to heterogeneous causality rather than circular causality resulting from self-reference. His underestimation of self-reference leads to the weakness of a relatively naïve notion of path dependency (see e.g. Sewell, 2005: 7, 100–102), ⁴⁶ since he still understood structure as regularity or recurrent pattern (see e.g. Sewell, 2005: 129, 199, 221, 280, 350) and couldn’t grasp the recursive character of social process. ⁴⁷

Self-reference can run in two directions, forward and backward. As Luhmann (1995: 79) said,

… history is constituted in the specific meaning dimension of time. By history we do not simply mean the factual sequence of events, according to which what is present is understood as the effect of past causes or as the cause of future effects. What is specific to the history of meaning is that it enables optional access to the meaning of past or future events, and thus leaps within the sequence. History originates in the release from sequence. ⁴⁸

Although Sewell emphasized heterogeneous temporality and causality, he still stuck to a unilinear notion of temporality and causality, and did not take the importance of the ‘leap’ into account. ⁴⁹

Thanks to the time-binding effect of attribution, the pure successivity of event chains could be overcome by overarching or continued relevance. Therefore, as selection, an action event became indeed irreversible, but it could continue as selectivity. Due to the structure of the duplication of present that came necessarily into being, we could take the double possibilities of reversibility and irreversibility into account, as well as make leaps and retroactive reinterpretations possible (Luhman, 1981b: 133, 137, 142).

History and society are a self-substitutive order. They operate recursively. This means that they operate simultaneously as operator on themselves as operand (cf. Mead, 1980: 30). Therefore their developments are certainly ‘path-dependent’, contingent on their earlier traces, but not necessarily linear, not to mention unilinear. The possibility and capacity to leap forward or backward could be attributed to the temporalization of basic ontological elements, and to punctuated events, as Luhmann (1995: 287) said:

The temporal punctualization of elements as events is possible only in time and only thanks to time, but through passing away and through total modification it realizes a maximum freedom vis-à-vis time. This freedom is acquired at the expense of structural formation, because it becomes necessary to regulate the reproduction of events by events.

This leads us back to the theme of contingency. Contrary to the common understanding of contingency, Luhmann (1976: 508–509; 1988: 282; 1995: 296–297; 1998: 44–46) worked on the revival of its original connotation in modal theory. Contingency is defined by a negation of necessity and impossibility. It can take two different forms. Before the decision is made, contingency presents itself as a situation of choice. Even after the decision is made, contingency does not disappear, but indicates that it could have been otherwise. In fact, action is often seen as a decision only retrospectively. However, it is important not to confuse contingency with ‘everything possible’ or ‘anything goes’. The concept of contingency ‘does not describe the possible in general, but what is otherwise possible from the viewpoint of reality … The reality of this world is presupposed by the concept of contingency as its first and irreplaceable condition of possibility’ (Luhmann, 1995: 106). Lacking a contingent, temporalized event as a basic ontological element and basic unit of analysis, neither Sahlins’s ‘possible theory of history’ nor Sewell’s theory of event is as open-ended as Sewell thought.

Sewell (2005: 321–328) discussed nevertheless a ‘meta’ self-reference. He observed that a problem always arose about ‘the ontological frame of our human existence’. He further said that both society and history are ‘our God’ because they signified ‘the really real’ in modern times. Luhmann went a step further, however. He explained why this was the case instead of appealing to God as metaphor, and elaborated the necessity and unavoidability of the self-foundation of the modern society by analyzing its structure of functional differentiation (e.g. Luhmann, 2004: 403–412). With regard to history, this had to do with the ‘historicization of time’ (Luhmann, 1981b: 143–146) on the one hand and with the ‘temporalization of history’ (Koselleck, 1985) on the other.

The decisive question of our current discussion is what is our concept for the self-reference as the ultimate reference? Historians or philosophers using hermeneutics, such as Ricœur (1981: 293–296), would say that ‘[w]e belong to history before telling stories or writing history. The game of telling is included in the reality told.’ By contrast, Luhmann insisted that society must be our concept for the self-reference as the ultimate reference, although he recognized the fact that all writings of history take place in a history that runs constantly forward (1997a: 576–577). But history takes place also in a continually evolving society. As sociologists we can observe this fact again, observe how the society observes and describes itself. I think this may be the ultimate divergence between history and sociology, although in principle ‘there is no logical or methodological schism’, but only ‘substantive divisions of labour’ between them, as Anthony Giddens (1984: 358) emphasized. Which side shall we as historical sociologists choose, then? Or is this not a problem at all?

The articulation between structure and event

Last, I must discuss the concept of structure and the articulation between structure and event. ⁵⁰ Sewell concentrated his efforts on modifying the structuralist version for his concept of structure and emphasized that structures are plural and not limited to cultural or symbolic structures. I endorse Sewell’s (2005: 140–143, 204–213, 219–223) idea of multiple structures, as well as his insight that the structure of conjuncture usually results from the conjuncture of structures. My question is this: Must we deny the theoretical possibility of change in cases of one single structure or one system (e.g. the contemporary world society ⁵¹ or an interaction system)? ⁵² Sewell might say yes. Such an attitude is reflected in his critique of Sahlins and his different valuations of Wallerstein’s and Michael Mann’s approaches (Sewell, 2005: 85–88, 113–123, 205–213). But most sociologists will distance themselves from such a standpoint, which is theoretically untenable, or at least incomplete, no matter whether Sewell’s view of plural structures is insightful or not.

In light of the question of how a singular structure can change, there is some inconsistency and deficiency in Sewell’s theory or formulation. First, Sewell saw the relation of structure and event as the relation of mutual implication, if not the relation of mutual constitution, since he said: ‘if structures define and shape events, it is also true that events (re)define and (re)shape structures’ (Sewell, 2005: 199–200). But at the same time he adopted Sahlins’s view and claimed that events ‘are transformations of structure, and structure is the cumulative outcome of past events’. ⁵³ The use of ‘are’ and the cumulative image suggest that Sewell saw structure and event as belonging to one and the same level of reality. It is not surprising that Sewell (2005: 127–128, 205) ‘heartily endorse[s]’ Giddens’s idea of the ‘duality’ of structure. Margaret S Archer’s (1982, 1995) critique rightly pointed out that such a view of inseparability would lead to a conflation and the confusion of structure and agency.

It is here worth comparing Luhmann’s (1995: 289) view that

… the concept of structure complements the conceptualization of elements as events. It indicates a condition of possibility for basal self-reference and the system’s self-referential reproduction. Therefore, structure can – as the verb ‘complement’ indicates – never be conceived as a sum or mere collection of elements. The concept of structure indicates a level of order in reality different from the concept of event.

Sewell did not maintain consistency with the perspective of mutual constitution but leaned toward a cumulative view (Sewell, 2005: 164, 332). Ricœur’s (1974: 92–95) idea might be helpful here. He tried to overcome the antinomy of structure/event and system/act by introducing the concept of discourse as operation into structuralism and indicated the word as a point of crystallization of all the exchange between structure and event, as ‘intersection of language and speech, of synchrony and diachrony, of system and process’.

How then should we conceive the relation between structure and event? The ‘Möbius strip’ would be a heuristic metaphor here. As Luhmann (2004: 84) said, ‘there is no “difference in essence” or any “material difference” between operation [i.e. event] and structure’ with regard to social system, the system of communication. But this is to insist at the same time that structure and event belong definitely to different levels of reality. ⁵⁴ The so-called cumulation of events could indeed change the structure, ⁵⁵ just as an ant would suddenly come into ‘the other plane’ in a Möbius strip, ⁵⁶ even though it kept crawling unilinearly; but neither ‘is’ structure a cumulation of events, nor ‘is’ Event a transformation of structure. The relationship between structure and event is nothing more than the relationship between expectation and action, as seen from the perspective of action, and this relationship is ‘one of reciprocal enabling’ (Luhmann, 1995: 293).

Sewell’s overestimation of causality led him to believe that the event was the cause of structural transformation. There is no doubt that ‘all structural change presupposes events because systems are composed of events and can transform themselves only through them’ (Luhmann, 1995: 352–353). But it ‘never makes sense to say “the” event is “the cause” of structural change; an event only identifies change’. Similarly, structure is ‘no productive factor, no underlying cause, but merely the constraint on the quality and connectability of the element [i.e. event]’ (Luhmann, 1995: 283).

Structure ‘concerns the connections or relations (Zusammenhänge) which do not enter into the strict sequence of events’, and the structural givens ‘enter into a momentary event, but they are preexistent in a sense different from that contained in a chronological precedent’, as Koselleck (1985: 107, with my revision) put it. Or, in Luhmann’s (1995: 45) words, ‘[t]he preselection of what can be chosen is experienced as validity in the case of structure, but as the sequence of concrete events in the case of processes’.

This brings us to the second question of whether Sewell’s conceptualization of structure is compatible with the perspective of mutual constitution of event and structure. At first glance, Sewell’s (2005: 214) definition of structure as ‘made up of both cultural schemas and material resources’ seems compatible, but it is inconsistent with the cumulative image that tends to see structure and event as belonging to the same level of reality. Upon closer inspection, Sewell’s dualistic perspective of structure is at odds with the idea of mutual constitution. The problem is not only that the ‘epistemic metaphor’ structure is ‘reified’ as a real resource, or becomes deterministic when structure is defined as ‘constituted by mutually sustaining cultural schemas and sets of resources’ (cf. Riley, 2008), ⁵⁷ but also that it is not easy to conceive of an articulation as well as a mutual constitution of event and structure as resources.

This brings us back to the question of the basic, ultimate element of the social – the one single operation that is sufficient to define a social system. Sewell noticed the material dimension, whereas Luhmann overlooked it. ⁵⁸ But Sewell’s metaphor of ‘built environment’ also clearly revealed that the material dimension, strictly speaking, is not the social system itself but belongs to its environment. ⁵⁹ If we agree that the social only consists of communications or various semiotic practices, it is questionable, if not improbable, to construct a convincing argument about the mutual constitution of semiotic practices and material resources.

Borrowing from Giddens, Sewell (2005: 129) also talked about ‘virtual’ existence of structure. ⁶⁰ Although this view is incompatible with the idea that structure includes resource, it makes sense if we take Luhmann’s view seriously that structures are at the level of implicit expectation rather than at the level of explicit behavior. ‘Structures have a value as reality (Realitätswert) only when they are used for linking communicative events’ (Luhmann, 2004: 82, with my revision). Although structures are preexistent in respect of their effect, validity and function as orientation of action, they do not directly enter into the sequence of events.

Third, the character of structure as potentiality implies that structure will be either confirmed, modified or changed at all times and places as soon as it is activated. Event and structure are complementary concepts because ‘temporalized elements … are determined from the outset to connect to something different. They can only actualize “current” connections, and therefore from moment to moment they create new situations, in which the system must choose between repetition and change’ (Luhmann, 1995: 47).

But even ‘repetition is a rather complicated process and one that is clearly different from mere copying’ (Luhmann, 2004: 314). Repetition is always accompanied and enabled by the double effects of ‘condensation and confirmation of meaning’. ⁶¹ On the one hand,

… the repeated uses of meaning must condense the used description in order to make sure that the meaning is recognized as the same, even in a new context. This leads to invariance … On the other hand, such repeated uses of meaning must confirm the reused meaning and demonstrate that the meaning can also apply in a different context. This leads to the surplus of references … which render[s] any concrete fixed definition of meaning impossible. (Luhmann, 2004: 144–145)

Sewell did not take into account poststructuralists’ insights about difference (or différance), repetition or iteration (cf. Deleuze, 1994; Derrida, 1982, 1988; Pedersen, 2008: 572–574). ⁶² In fact, the term ‘reproduction’ (and: ‘practice’) is already a modification of the paradigm of structuralism that emphasizes synchrony and invariance. It is therefore not surprising that Sewell’s position could be more rigid than Sahlins’s (cf. Riley, 2008: 559). For example, Sahlins (1981: 68; 1985: 144) claimed that ‘all structural transformation involves structural reproduction’ and

… every practical change is also a cultural reproduction … the more things remained the same the more they changed, since every such reproduction of the categories is not the same. Every reproduction of culture is an alteration, insofar as in action, the categories by which a present world is orchestrated pick up some novel empirical content.

Correspondingly, Sewell still conceived of structure as a recurrence or regularity, and did not see that structure is actually a temporarily stable product of recursive operations that gives one the impression of ‘recurrent’, ‘regular’ or ‘invariant’.

Fourth, although the idea of multiple structures is praiseworthy, Sewell’s (2005: 206) critique of singular structure implies that structural effect is almost the same. I argue that people can have a different understanding, decoding and appropriation of the same structure. The one-sided stress on multiple structures makes Sewell see the categories as given, although he accepted Sahlins’s view and admitted that action constantly ‘puts cultural categories at risk’. ‘Events transform the meanings and relations of cultural categories not only because the world fails to conform to categorical expectations, but because actors bend categories to their own ends in the course of action’ (Sewell, 2005: 203–204). From Luhmann’s (1989: Ch. 5) perspective, Sewell did not see the ‘self-abstraction of the object’ (Luhmann, 1995: 2) and stays in principle still on the level of observation of the first order, namely, only observes things and actors’ interests or intentions as such, and lacks a reflection from the level of observations of the second order, to observe how the actor as observer observed. It is not surprising that Sewell was not fully aware of the importance of contingency because ‘everything becomes contingent whenever what is observed depends on who is being observed’ (Luhmann, 1998: 48). The decisive weakness results from the fact that Sewell did not take into account self-reference. It is therefore improbable to go a step further from self-reference to autology, from the level of operation (autopoiesis) to the level of observation, to acknowledge the unavoidable condition that ‘the knowing system is itself one of the objects it has to know: when it describes its objects it then also describes itself, and the description modifies the object to be described’ (Esposito, 1996: 271).

Furthermore, according to Sewell (2005: 140–141, 212–213), the overlapping of structures would lead to the transposability of categories and senses of self. This view is based on the assumption that society consists of individuals as human beings. It is questionable from the view of systems theory because this view did not differentiate social system from psychic system seriously enough.

Fifth, Sewell (2005: 221–222) is aware that in a situation of high uncertainty ‘mutual redefinitions of the situation that significantly restructure practice are likely’. But he restricted such a situation to the case of conjuncture of structures. By contrast, for Luhmann (1995: 324), ‘order emerges out of chance events in the course of time’. It is almost normal and unavoidable to use any accessible chance to build structure in the beginning of system formation because in the situation of double contingency any initial chance is productive. Due to the autocatalysis of double contingency as a problem, the chances can be transformed into probabilities of structure building. This is so because every social structure is a structure of expectation, and even chance events ‘form expectations by their very occurrence’ (Luhmann, 1995: 293), although any connections become less subject to chance.

A reconstruction of the logics of history

We cannot move toward an adequate theory of event unless we take up Luhmann’s concept of event and structure. It is possible to reconstruct Sewell’s ‘logics of history’ (i.e. fatefulness, contingency, complexity, eventfulness and causal heterogeneity) from the thoroughly temporalized event. First, history is ‘fateful’, because ‘[e]vents disappear as they emerge: they are no longer available to react in the following instant’ (Luhmann, 1995: 449). Despite this, ‘every event brings about a total change in past, present, and future – simply because it gives up the quality of being present to the next event and becomes a past for it (i.e. for its future). This minimal displacement can change the perspective of relevance that structures and bounds the horizon of past and future. In this sense, every event brings about a total modification of time’ (Luhmann, 1995: 287) and is itself irreversible.

Second, the event that vanishes is contingent. It is not only the fact that appearance of an event excludes the realization of other events, but also the fact that an event presents itself as a choice before its appearance and indicates that it could be otherwise after its disappearance. There is no event in itself, but only event embedded in a system, as seen from the perspective of a specific distinction of system and environment. Contingency matches better with the punctuated event than with the durable Event. Contingency means not only a dependence on particular temporal sequences but a wide space of probabilities minus necessity and impossibility, where leaps forward or backward are both possible.

Third, complexity is closely related with contingency due to the common problem of selection. Luhmann said that ‘every operation [i.e. event] of the system that establishes a relation has to choose one among many – complexity enforces selection’ (1990b: 81). Meaning, understood functionally as the difference of actuality and potentiality, ‘is nothing but a way to experience and to handle enforced selectivity’ (Luhmann, 1990b: 82; see also 1995: Ch. 2). This has to do with the self-referential character of meaning. The instability of actuality is synonymous with the constant replacement of punctualized events. Event and time vary with the system reference. By choosing event rather than Event as the departure, we can really have an ‘eventful’ conception of temporality in a serious sense, i.e. multiple, heterogeneous, complex temporalities.

Fourth, eventfulness is almost self-evident if we take the event that vanishes as soon as it takes place as the basic ontological element and the last unit of analysis. For ‘[e]very event, every action appears with a minimal feature of surprise, namely, as different from what preceded it. To this extent, novelty is constitutive of the emergence of action’ (Luhmann, 1995: 288). To understand this, the concept of information that ‘is definable as a difference which makes a difference’ (Luhmann, 1995: 40) is helpful here. Like event, ‘[i]nformation is always information for a system’ and means ‘an event that selects system states. This is possible only with the help of structures that delimit and presort possibilities.’ An event could be dramatic or eventful because it includes an information that can change ‘the state of the system and has thereby left behind a structural effect’. But an event could also be a routine affair because ‘a piece of information that is repeated is no longer information. It retains its meaning in the repetition but loses its value as information’ (Luhmann, 1995: 67, with my revision).

Fifth, causal heterogeneity is in a sense a corollary of temporal heterogeneity as mentioned above. If events vary with the context or system in which they are embedded, they follow different logics, temporalities and causalities in different contexts. The causal heterogeneity not only has to do with time and place, as Sewell emphasized, but also has to do with system, with the differentiation of system and environment.

Sewell’s excellent and creative synthesis has indicated a possibility to construct a theory of history by building a theory of event. The problem, however, is that Sewell’s possible theory of Event is neither consistent nor radical enough. We cannot speak of an actual theory of event unless we re-found Sewell’s theory on the concept of event as a temporalized ontological element.