CAIRN-INT.INFO : International Edition

1 Anthropology and sociology are linked most notably by the approach they share: the ethnographic approach or field study. However, this article will not undertake to compare the use of the ethnographic approach in sociology and in anthropology. It will focus on field research in sociology—an approach inherited from anthropology—in the belief that the problems it poses and the solutions it brings are comparable overall to those which arise in the field of anthropology. What interests us here is how fieldwork is “interpreted.” The general argument is that empirical fieldwork and theoretical analysis (the moment of interpretation following the collection of data) should not be opposed, to the extent that any empirical approach, whether descriptive or explanatory, presupposes general or theoretical considerations. These general considerations themselves (this notion will be specified further on) have an empirical basis, but do not derive directly from the ethnographic study currently underway. Instead, they are drawn from a prior stock of knowledge to be used as part of the ethnographic process, without deriving directly from the field being studied.

2 First, we will reflect on the status of these commonly referred-to general propositions, and propose a rough classification of them in terms of how they are used to describe and explain field research. The aim of this study is to strengthen the reflexive character of the interpretation of data; that is, to insist on the fact that use of these general propositions should not be made “spontaneously” or intuitively, without verifying their degree of relevance, their relative validity, and the way they are incorporated into interpretive reasoning. This reserve of general propositions, constantly employed by those interpreting field research (in the same way, moreover, that any individual in the course of everyday life makes use of such general propositions) forms the object of this study on the modes of interpretation of fieldwork. It will be briefly noted that these general propositions are also utilized in quantitative studies, for explaining correlations between dependent and independent variables.

The Development of Fieldwork in Sociology

3 It is important to note that none of sociology’s founding fathers studied, strictly speaking, in the field (except perhaps for Tocqueville and certainly Le Play who, however, did not use the notion of a field), nor did they develop the basic principles of the burgeoning discipline of sociology in relation to this approach. Neither Comte nor Marx, Pareto, Simmel, Durkheim, nor Weber conducted fieldwork. They did not even try to theorize it. So, remarkably, the dominant practice in sociology today (in France at least) makes use of an approach that does not come from the classical heritage of the discipline, but rather from a method developed in and imported from another discipline, anthropology. In the French tradition, Marcel Mauss was the most vocal advocate of an ethnographic approach inspired by British anthropology, and he thereby made the link between the legacy of his uncle (considered one of the great inspirers of quantitative analysis in the social sciences) and the British heritage of anthropology. In reading Mauss’s ethnography manual, one is struck by the level of ambition assigned to field research, since it involves describing and explaining a “total” social phenomenon with a strong theoretical connotation even though Mauss himself described the ethnographic approach as relatively independent of a theoretical apprehension (Mauss [1947] 2002). Today, most sociological field research sets more modest goals of description and perhaps explanation of defined objects. Nonetheless, despite reducing the focus to relatively narrow objects, the interpretation of field studies constantly employs general principles that do not derive directly from the field in question; these general propositions now merit a closer look.

4 Malinowski ([1922] 1989) assigned three tasks to ethnographic studies: to record social organization, to describe “the imponderabilia of actual life,” and to constitute a corpus inscriptionum which serves as an account of native life. Fundamentally, the goal is to describe practices, the norms and institutions that govern these practices, and the ever-present gap between these practices and norms. This might still be considered the basic program of any ethnographic description today. However, the ordinary situation of a sociologist in the field today is very far removed from the one to which Malinowski was accustomed, and which he conceptualized for the first time in his introduction to the Argonauts. Six basic differences are of particular note (even if these are obviously not the only ones):


  • Malinowski studied an object described at the time as “primitive” (that is, an oral society), whereas sociologists of today deal with societies whose general characteristics are often close to their own;
  • the Trobriand Islands, not least for their insular character, formed a relatively well-defined, limited object of investigation, whereas the fields covered by sociologists today are always isolated from within wider wholes, which are nonetheless inevitably present to a large extent in the object of study. For example, studying a French suburb has meaning only in relation to the situation of the housing projects and their inhabitants within the wider setting of French society (and of Western society, in the context of globalization and migration);
  • thirdly, and this is an effect of the two previous points, the inaugural cases of field research involved the discovery of an essentially unknown social reality, the presentation of which was in itself a scientific event, whereas today it is extremely rare for the mere description of the field being studied to introduce the newness of a situation (compared to the one that most ordinary sociologists are more or less familiar with) comparable to the discovery of the potlatch or the kula ring, for example;
  • generally, contemporary societies and the fields isolated from within them are already thoroughly invested by practices (derived in part from the social sciences) that heavily transform them (for example, through economic policy or public policy in general);
  • today, a more or less advanced scholarly literature already exists on most of the objects that form fields of study, which the ethnographer has read. Researchers almost never tackle an object from a position of scientific virginity. In particular, there is often an existing quantitative literature on many of the subjects covered by ethnographic studies (for example, suburban youths are often the subject of statistical measurement on various aspects of their situation) and field studies must necessarily take into account this scientific data where they are available.
  • finally, the objects handled by the ethnographic approach are (at least partially) covered by other disciplines: for example, school choices are studied by economists, beliefs by psychologists, and languages by linguists. The field approach is therefore faced with complementary or rival scientific positions which deal with meaningful aspects of their object. The ethnographic approach is used in relation to objects that are (at least in part and under certain aspects) treated by other methods.

6 In other words, even more so than in the early twentieth century, subjects are not addressed neutrally, but from a background stock of tacit knowledge, which involves both the relationship that the isolated field maintains with the other dimensions of social reality, and the accumulation of knowledge available to the interpreter, either within his or her discipline, or outside of it (and including all the general propositions which he or she feels legitimately able to use). “Meaning” is not addressed through a comprehensive immediacy, nor through the simple “distancing” from the object, but by the inevitable use of a set of propositions that are available to interpret the observed behaviors. Maintaining a distance necessarily involves the use of an interpretive language, and this interpretive language consists of a set of general propositions not directly supplied by the field. In the background of any interpretation of field research lies a dual operation of comparison: a comparison between broad social situations within which the more specific social situation being studied is characterized and identified; these relatively broad social situations themselves invoke comparisons between the various known social behaviors, in both their similarities and dissimilarities. These background comparisons produce a tacit theoretical knowledge that is brought to bear in the interpretation of field research.

7 As a sociologist, the author of this article would not like to make any immodest claims with regard to his perception of contemporary anthropology, which is not the subject of this paper. However, it seems safe to say that considering the present extension of anthropology (with its easily identifiable subdisciplines such as urban anthropology or economic anthropology), the fields it covers place anthropologists in a fairly similar situation to sociologists, which contrasts with the initial situation that Malinowski experienced. As for the type of subjects covered, while anthropology maintains an interest in oral societies, it now looks more frequently at societies that have gone through industrial development, as do sociologists. In general, with regard to the isolation of objects, the newness of findings, the interpenetration with contemporary societies and with their deliberate public policies, the existence of a considerable scholarly literature (especially quantitative) on the objects covered, and finally the presence of related knowledge from other disciplines (economic anthropology, for example, cannot ignore economic analysis), anthropologists seem to be in situation similar to sociologists vis-à-vis their field research, and therefore also fairly far removed from the situation experienced and symbolized by Malinowski. The result of these changes is that the stock of prior knowledge from which new fields are approached is increasingly consolidated.

8 So, unlike anthropology (if we consider Malinowski’s importance for that discipline’s development), classical sociology was not built around the notion of fieldwork. The current success of field studies in sociology (especially in French sociology) was arrived at gradually, and is linked essentially to four factors.

9 The first is the well-established (and inarguable) certainty that sociology is an “empirical” discipline: in the same way as the anthropological field research of the late nineteenth century was directed against “armchair anthropologists,” the promotion of fieldwork corresponds to a rejection of a priori speculation of sociologists who are not informed of their subjects by “direct” experience. From this perspective, sociology belatedly came to accept what was initially anthropology’s position, from the development of the first field studies. It was no doubt easier for sociologists to speak of things of which they had no direct experience, as they were generally more familiar, not as foreign. But it should be noted, for example, that the important writings of Weber on the sociology of religion are all based on the work of other scholars, and though far removed from any field experience, they are nonetheless extremely instructive.

10 The second is more specific to sociology: Durkheim introduced the founding principle of quantitative analysis in sociology, through the method of concomitant variations (later to become correlation analysis and multivariate analysis), and interpreted them as “causal” relationships on the basis of a hard metaphysics of social entities (even if Durkheim’s scientific practice allows a more relaxed reading of his principles than the most unambiguous accounts he gave of them). Where anthropology was organized primarily through fieldwork (although of course quantitative anthropology does exist), sociology, inspired by Durkheim, turned to quantitative studies, which revealed social realities just as unsuspected as the potlatch or the kula ring: the fact that a particular behavior (deviance, suicide) could be related to “determinant” variables of which the actor concerned did not have the slightest idea. Even today, it is one of the great advantages of sociology that it can highlight relationships between variables that are not necessarily obvious from an intuitive point of view. However, these quantitative methods can pose a number of difficulties and meet certain limitations in their very constitution, prior to their interpretation. This gives rise either to a need for complementary, nonquantitative empirical research, or sometimes to the rejection (less and less accepted in its more radical form) of any quantitative undertaking. Note here also that the directly causal interpretation of dependency relationships between variables is problematic, because the identification of these correlations, far from being explanatory in themselves, raises the problem of how to explain them. I shall come back to this point later, in order to underline that the interpretation of these correlations also employs explicit or implicit general propositions.

11 The third factor, which is much less important conceptually but important in practice, is that quantitative methods are more difficult to learn and master than the basic forms of ethnographic research, which are also generally less expensive to put in place than large quantitative surveys: field studies are therefore more likely to be adopted by students (here we are referring to an elementary psychological general proposition).

12 Finally—and this is not the same as the argument introduced in the first point—antitheoretical attitudes developed which have found more support in field research than in quantitative studies, as the latter are often marked by “theoretical hypotheses” to be empirically verified and validated. Field studies seemed closer to empirical data rich in immediate content than the abstractions of quantitative research, not to mention the now definitively suspect grand theory. Of course, experienced field research specialists never reject theory (Beaud and Weber [1997] 2010), but they seem to see it as something fairly secondary and remote compared to the richness of the empirical investigation of facts. In any case, this more or less pronounced refusal of theory is based on what is considered to be the fairly clear separation between empirical data and the theorization of this data at a later stage. I will try to show in what follows that this is not an obvious distinction, and that the collection and interpretation of empirical data presuppose “theoretical” considerations.

13 It must be said that the notion of theory itself, in terms of the uses made of it, is unfortunately not very codified, and when spoken of, its meaning needs specifying. This matter would be worth discussing at length, but for the purposes of this article we shall deem “theoretical” all general considerations used in the interpretation of field studies, but which do not derive directly from these studies. In this sense, they are not empirical; that is, they are not directly derived from the description or explanation of the defined field, but are used to make sense of this field. However, these general theoretical considerations were not plucked out of thin air; they have an empirical origin and validity which come from a more general knowledge of behaviors and the constraints of action, derived from the analysis of broader objects and supported by wider comparisons of data.

14 Based on this initial observation, I would like to defend three related positions:


  • what matters in field research is not so much the distinction between theory and empirical data (which is quite relative), but the distinction between description and explanation, which each raise different issues. Together, these two functions form part of what might be called the “interpretation” of a field;
  • in terms of explanation, general propositions (which have a more or less explicit, and certainly always empirical, origin) are necessarily brought to bear, and their usage and status in the explanations proposed should be identified;
  • finally, this suggests that with regard to both description and explanation, there is no fundamental difference between field research and quantitative research, both of which make equal use of general propositions that are not directly derived from the subject being studied, but used to interpret it. This final point will be simply touched on.

Description, Explanation, and Normative Stances

16 The task of the social sciences can be considered to be descriptive, explicative, and normative. Description in principle simply describes what is, explanation attempts to clarify why that which is, is, and normative stances assess that which exists and suggest plans of action. Of course, this third task is fundamentally problematic (which does not mean necessarily unacceptable) from the perspective of a scientific undertaking; we will not go into it further here. We will simply note that, in practice, most social scientists, whether prestigious or modest, do not hesitate to take all sorts of normative stances in the course of their scientific work. A systematic analysis of this attitude’s epistemological foundations remains to be performed. It should be noted, for example, that describing an attitude as “ideological” is adopting a normative position (and therefore entails having some criteria to characterize an ideological attitude). The various social sciences have different traditions in this regard: economic analysis has a well-established “normative economics” which defines the conditions for a normative characterization of different situations, whereas sociology opposes theorists such as Durkheim who consider that the task of social science is to provide scientifically based normative assessments, and others who refuse normative assessments, such as Weber.

17 The principle of a contrast between description and explanation may itself be disputed, and may be disputable. Yet it is clear that there is quite often a fairly easy test to distinguish between these two approaches, as one replies to the question “what?” and the other to the question “why?” (Hedström 2005). Identifying that a thing exists, and describing it, is by no means the same as to say why it exists, especially in a given form. These are distinct orientations of reflection and interpretation. As such, demonstrating a statistical correlation between variables is not to explain this correlation.

18 The entire sociological tradition, in its analysis of the descriptive reconstruction of facts (particularly in the epistemological works of Simmel and Weber who, it must be said, based their reflection on historical data rather than on the quantitative or ethnographic data of sociology) has stressed (in opposition to Ranke) the fact that there is no “immediate” reconstruction of facts. Facts, necessarily approached by means of questions, are instead arrived at through the prism of selection, conceptual or analytical categorization, and ordering. Facts are not given directly to the analyst to gather like fruit from a tree and take home. First of all, he or she must select them according to certain questions, as absolute exhaustiveness is illusory (as in the natural sciences); he or she must then “interpret” them through the use of analytical categories, such as “norm,” “relationship,” “deviance,” and so forth. Finally, he or she must arrange them so that an intelligible and perhaps “coherent” result can be presented. All three of these operations (selection, analytical interpretation, ordering) are inevitably present in any ethnographic collection of data. They are obviously sources of difficulties.

19 Expanding on the classic sociology literature that stresses the role of the interpreter in constituting data, Geertz ([1973] 1998), in a famous and influential essay, criticized the positivist approach and insisted instead on the ethnographer’s and anthropologist’s position as writers (Geertz 1988). If we relate Geertz’s propositions to the epistemological reflections of classical sociology, many of his remarks seem largely relevant, to the extent that ethnographic accounts do indeed amount to a writing exercise that orders interpreted experiences.

20 Nonetheless, as Geertz himself points out, this is no reason to conclude that ethnographic descriptions are in any way arbitrary. The fact that they involve a “writer,” who selects, uses categories, and orders data, does not ultimately imply that the descriptions are unreliable with regard to their subject. To use the canonical example that Geertz borrowed from Ryle, one really must be able to tell in the end, when the ethnographic study is well done, whether an observed eye movement corresponds to a twitch, a wink, or a parody of a wink. In all cases, we need analytical categories that are not derived from the direct apprehension of the phenomenon, but relate to tacit knowledge formed elsewhere (as part of the accumulation of ordinary knowledge relayed and possibly transformed by scientific knowledge), and which make it possible to sort a phenomenon under the category “twitch,” “wink,” or “parody of a wink.” Therefore, it is clear that the introduction of these categories, at the most elementary level, already concerns a dimension that can be deemed theoretical, since the facts are registered within general orders that are not directly provided by the subject being studied, but instead make it possible to grasp that subject in a relevant and intelligible way. Once again, this does not mean that there is anything arbitrary in description. Similarly, the account that Geertz gives of an episode in Morocco of a conflict between a Jewish merchant, members of a Berber tribe, and some French soldiers refers to a set of general concepts (norms, rights, tribes, and so forth) that may be stabilized by their involvement in a relevant description. These are indeed descriptions, even if they are linked to a degree of interpretation. Description includes a dimension of “understanding,” to the extent that to describe a wink in a relevant way presupposes that one knows what a wink is, in the same way as to describe the trials of a Jewish merchant within a normative system presupposes that one “understands” what a normative system is.

21 Certainly, for anthropology in particular, the question of “translation” is raised by the existence of cultural worlds that are very different from the ones anthropologists or their readers come from. This remains a possible distinction with sociology, which deals with more familiar environments. Yet the gap is relative. Cultural difference exists within contemporary industrial societies (such as the difference between varied lifestyles associated with very different norms and standards). Conversely, societies which are culturally very distant from Western societies have much in common with them, for example, the practical handling of certain aspects of action. The great achievement of anthropology is precisely to have provided intelligible descriptions (based on analytical categories such as exchange, gift, hierarchy, and so forth) of phenomena as important as the kula ring, the potlatch, or caste to members of societies that do not have any direct experience of them. These objects are approached and discussed through these categories of more immediate intelligibility which enable them to be constituted. For example, Malinowski ([1985] 1926) himself reinterpreted the descriptions he had given of the kula ring at a later date. Certainly, as these facts are necessarily grasped in a conceptual and analytical way, using categories that go beyond the specific subject being interpreted, they give rise to incessant debates that seem always to go unfinished. However, at the same time, these debates deepen the description of objects, their nuances, their internal differences; they do not indicate any arbitrariness in the description.

22 The use of translation with regard to distant cultures (such as for the translation of a book) does show that there is a certain indeterminacy in translation, not an absolute “necessity,” since several different translations are possible (Affergan 1997), and that therefore there is some “construction” of the object. However, if there are several translations possible, this also means that not all translations are possible. While several similar formulations might compete without there being any necessity to choose one over the other, these possible translations thereby dismiss all the (infinitely more numerous) impossible translations.

23 Description can assume different levels of generality: beyond singular events and singular actions, there are sets of rules and institutions that are shown, making up general situations described in their different forms (structuring of behaviors, rules, and institutions). These can be understood through reference to ideal types, which have no explanatory power in themselves, but are rather theoretical constructs with an analytical purpose, which highlight objects with a certain number of common properties. Translations, meanwhile, are based on elements of similarity between different cultures, which allow them to be compared in connection with relatively similar analytical frameworks.

24 We might add here that with regard to its objects, the quantitative approach is fundamentally in the same situation as the ethnographic approach. The construction of variables and the demonstration of dependency relationships between them have a descriptive purpose. These variables arise from the selection of certain features (which could be selected differently), which refer to abstract analytical categories (such as age, income level, education level, and so forth) and represent a (partial) ordering of data.

25 In short, in line with a well-established tradition in anthropology and sociology, we may never talk about raw facts—they always arise through a process of selection, categorization, and ordering. This is something that Malinowski stated very clearly:


The main achievement of fieldwork consists not in a passive registering of facts, but in the constructive drafting of what might be called the charters of native institutions. While making his observations, the field worker must constantly construct: he must place isolated data in relation to one another and study the manner in which they integrate. To put it paradoxically, one could say that “facts” do not exist in sociological any more than in physical reality; that is, they do not dwell in the spatial and temporal continuum open to the untutored eye. The principles of social organization, of legal constitution, of economics and religion have to be constructed by the observer out of a multitude of manifestations of varying significance and relevance. It is these invisible realities, only to be discovered by inductive computation, which are scientifically important for the study of culture.
(Malinowski 1935, 317)

27 These operations do not mean, however, even when the selection, categorization, and ordering correspond to processes of translation between very different cultures, that an essentially accurate, relevant, and useful result cannot be obtained. Admittedly, there are many areas where the border between explanation and description is porous, particularly when the description depends on highly controversial interpretative frameworks (especially when they involve normative or “ideological” dimensions). For example, should a situation of pronounced power imbalance between two people within a company be described as a situation of “domination?” It is clear that this type of subject can turn into an interpretative battle which, beyond the descriptive elements, involves significant normative aspects. Moreover, these are already present in the subject, since the interpretative controversy can also reflect a controversy between actors who themselves interpret their common situation differently. However, in any event, this type of argument only makes sense because the existence of an asymmetrical or unequal distribution of power is recognized descriptively, and can then be characterized with varying normative connotations (but still within a relatively small range of possible interpretations).

28 However, it does not seem accurate to state definitively that any description depends on an arbitrary theoretical language in a strict sense, in which case there would always be a plurality of possible descriptive languages, based on incommensurable “metaphysical” paradigms or languages. This is what Passeron suggested, based on a proposition by Popper:


The basic statements closest to natural perception contain theory, in the sense that their assertoric meaning is never self-sufficient, as it is always beholden to a language for describing the world, whether that language is constructed by a scientific discourse or practiced spontaneously within the preconstructed framework of a culture and a customary language. It was the empiricist illusion of the first logical theory of Carnap and Neurath to think it possible to define as “protocol statements” the basis statements of empirical science, inasmuch as they were supposed to be able, by minimizing theoretical interpretation and coming closer to the description of singular, dated, and localized experiences, to tend asymptomatically towards the coincidence between a statement and a state of affairs.
(Passeron [1991] 2006, 545)

30 This complex quotation would be worth developing in detail. Here, we will defend the (no doubt insufficiently justified) position that descriptions are indeed not neutral in their relation to the subject, and that they include an analytical and therefore theoretical dimension through their use of general categories (particularly when a complex institution is being presented); but we do not conclude that there is an infinite number of possible competing discourses on an subject, or even that, in relation to current social events, there are “several” possible languages overdetermining the description. To return to the two examples from Geertz mentioned above—the wink and the conflict in the Moroccan desert, retold from an old narrative—it is hard to see how varied, competing “metaphysical” languages (in Popper’s sense) could lead us to descriptions highly divergent from those given. As we see it, two positions should be maintained at once: to refuse the naivety that facts are transparently readable, and can be simply reconstructed “empirically” through experience in the field; but, at the same time, to accept the idea that some descriptions are more pertinent than others, grasp their subject more cleverly, and are not dependent only on definitively rival and incommensurable “paradigms” or languages. It is certain, however, that several explanations of a situation are often possible, and that the data make it difficult to decide between the various explanatory scenarios conceivable.

31 It is also true that any description refers back to a scientific community seeking analytically accurate and relevant descriptive languages. It cannot be said, however, that this is simply a specific community, to be seen in comparison with other communities with their own descriptions of the situation and their own languages. What Geertz’s Moroccan example presupposes is the evolution towards an accurate descriptive language which, while it is upheld by a scholarly community, allows to the greatest extent possible the limits of partial languages that are mutually incomprehensible in their interpretation of a situation to be overcome, in order to arrive at a fuller, more accurate, and more relevant description of the events in question through the very project of a scientific description.

From Description to Explanation

32 Because it is description, ethnographic description is not explanation, even though it reveals complex, richly meaningful, and highly theorized worlds (the kula ring, potlatch, the caste system, and so forth), no more than the demonstration of strong dependency relationships between variables is explanatory (even if standard usage refers to “explanatory” variables). As shown above, however, questions regarding the relation to theory arise at the descriptive level, even before any explanatory enterprise gets underway. This procedure of description can be associated with a procedure of “understanding,” since description presupposes the intelligibility of the things being described (such as a wink).

33 For its part, explanation involves showing “why” the phenomena described are constituted in a given way compared to other possible situations, and why a given dependency relationship exists between given variables. It must be stressed that to reveal, for example, the configuration of variable and complex kinship systems is in itself a remarkable scientific achievement, and that even if the social scientist fails to explain why the kinship systems are variable, the fact remains that the description and conceptualization of these kinship systems in themselves already form a very significant achievement for scientific knowledge.

34 It seems to me that the notion of understanding, which needs to be briefly touched on here, is not a kind of third way that is neither description nor explanation. We understand something first of all because it is associated with an intelligible description (which includes aspects of translation when different cultures are involved), which reconstructs what exists and what happens in its organization. For instance, we understand the potlatch system through a description of how it works (which presupposes intelligible categories, such as exchange). Next, we understand something because it is explained when it is deduced from a system of rules that make it intelligible (for example, a deep bow is deduced from a social norm corresponding to a form of courtesy; it is the same when a given behavior is related to the principle of rationality). The thing is explained by appealing to the norm of reference. Finally, we understand a social phenomenon when we demonstrate the factors that produced it. For example, Merton’s Matthew effect shows how the more often an author is cited, the more he tends to be cited further. This social phenomenon—the concentration of citations from a small number of researchers—is understood because it is explained by a number of factors that cause it. This demonstration of causal factors itself presupposes the recognition of general relations of causality, which allow the connection between different phenomena to be described precisely in a causal manner.

35 It is rare for field research to stop at the merely descriptive phase; it usually adopts an explanatory approach. We would like to try and identify the characteristics typical of such an explanatory approach, which essentially consists in referring to general propositions, on the basis of which specific situations are deduced. These general propositions come from an interpretative stock that is more or less “obviously” available to the analyst. Most of the time, however, the use of these general propositions in itself is not analyzed, which can be a source of inaccuracy in the explanations proposed; hence the possibility of significant oversights within a line of reasoning (Boudon 1990).

36 I will rely here on a few examples chosen, somewhat arbitrarily, from among countless other possible examples. In general, what characterizes sociological work based on field studies is that it is at once rigorous in its processing of data, and not very formalized in terms of how its various propositions (especially its general statements) are linked and prioritized, mixing ethnographic data and quantitative references (both explicit and implicit), and descriptive propositions and explanatory propositions in turn (not to mention the frequent adoption of normative stances, which we will not go into here). In sociology, ethnographic analyses generally have a hybrid bent: they incorporate quantitative dimensions into ethnographic description.

37 In the following brief discussion, based on a few quotations, we will entirely disregard the truth value of the explanation given, which may be completely wrong. We will focus instead on its form, considering that this form is characteristic of the explanatory enterprise undertaken on the basis of descriptions in field studies.

38 The first case considered addresses the situation of African families in contemporary French suburbs. The quotation given is fairly long, as it is to be used as a document to illustrate the form of reasoning.


In Europe, African husbands, the first to arrive, established rules which isolate and sometimes trap their wives. Most women who have come from the Sahel recognize the eminent value of traditional authority and the importance of age hierarchies as they have been handed down by tradition. They do not evade the need to comply with customs based on respecting the choices of elders and parents, even when what they say seems arbitrary or unfair. Yet, in Europe, this is not enough to preserve them from male resentment. More often than not, men from the African continent have tried to cure themselves of the contempt they feel subject to here by transferring their frustrations onto their own wives. These women are subjected to contradictory commands: they should exercise authority over their boys while the fathers are working or absent, but, as the father has disqualified them, they do not have the means to do so. . . .
On the one hand, the fathers cannot hide from their sons that they have spent part of their lives keeping their heads down. On the other hand, they have destroyed what would have been the only strength they could have had here: an unbending solidarity with their wife and parental cohesion. Tensions between husbands and wives deriving from men’s authoritarianism undermine the statutory authority of the older generation. When the man does not play his part and, in addition, demeans the mother in the eyes of the boys, she cannot really help these boys become autonomous individuals, capable of integrating into society. Blurred generational positions, combined with early school failure, provokes in these adolescents an attitude of mistrust towards authorities and institutions. In this manner, unlike the customs of Western families where generational asymmetry—the subordination of children to parents—unequivocally prevails over the asymmetry between the sexes, these patriarchal customs altered by immigration tend to insist on the sexual hierarchy—women's subordination to men—above the generational hierarchy. They confuse educational benchmarks. The radicalization of women’s inferiority divides the parental camp and sets sons above their mothers.
(Lagrange 2010, 189–190)

40 This extract is characterized by a mixture of descriptive and explanatory considerations. It may be noted at the outset that the descriptive propositions implicitly contain a quantitative dimension, since the sentences describing the attitudes of African fathers and mothers in Europe, on the basis of field research, indicate general behaviors characteristic of members of the group, without, however, there being any detailed quantitative proposition. This is often the case in field studies, as the description tends to reconstruct the behaviors of members of a group. The explanatory dimension corresponds to the fact that given factors are said to cause a certain number of attitudes. These are therefore causal propositions, which implicitly refer to general relationships that enable the causality in question to be described. In order to follow such reasoning, it is necessary to make implicit use of general causal propositions that help explain what is observed.

41 In my view, there are fundamentally three types of general propositions corresponding to causal constants that are put forward (Demeulenaere 2011):


  • psychological or cognitive propositions which correspond to abiding features of behavior apparent in certain situations (these might include considerations on a “rational” or “irrational” behavior);
  • practical propositions concerning the expected effects of certain actions, and the general constraints affecting how the actions take place, which correspond to the empirical knowledge available about the way that actions can take place or not in a given context;
  • institutional, normative, or “cultural” propositions that indicate the existence of norms and their consequences: individuals behave in a certain way because certain norms are in force and are accepted. The fact that certain norms or the existence of certain institutions are accepted has consequences in terms of predictable behaviors.

43 So, on the basis of a field study, supported by quantitative studies (let use reiterate that we will not express any opinion here on the validity of the findings; we are only interested in the form of reasoning), it is demonstrated that young men from African immigrant families have a greater propensity for crime, on an equivalent social level, than children from other ethnic origins, including French.

44 The reasoning used to “explain” this fact identified through fieldwork is that:


  • the members of these families refer to norms of subordination to male authority;
  • these norms have been undermined by the situation of African men within French society, because of unemployment and their low social status;
  • they “compensate” for the disparagement they suffer, by an increased disparagement of women;
  • the women’s authority is consequently greatly weakened;
  • because their authority is weakened, they do not have the practical possibility of exercising effective authority over their male children;
  • consequently, these boys do not adopt the norms of meritocracy that would allow them to succeed at school;
  • their early failure at school causes them to resent institutions and society;
  • this resentment in turn leads to a propensity to crime (rather than to conventional social achievement).

46 This causal reasoning, which aims to explain the phenomenon studied, is based on general propositions that are utilized implicitly:

47 psychological propositions that describe typical behaviors: when people experience social disparagement, they tend to become resentful; this resentment tends in turn to provoke an attitude of increased disparagement towards other people whom they consider to be in an inferior position to their own;

48 practical propositions (of an instrumental kind): for parents to successfully exercise authority over their children, this authority must be respected within the family, and not be diminished by one spouse undermining the authority of the other. In order to succeed at school, children must adopt the norms of work and effort assumed in the running of this institution;

49 institutional propositions and prevailing norms: women’s attitudes towards their husbands (and sons’ towards their mothers) are explained by reference to the existence of tacitly accepted cultural norms which establish the subordination of women to men.

50 Let us cite another text, by different authors but of a similar tone, which also focuses on urban violence:


Because they are permanently excluded from the unskilled labor market, these “young immigrants” flee the region . . . or fall back on a few employment niches (“ethnic” businesses, local youth-employment centers). Others “hang out,” or live off small-scale illicit trade. These youths, the new “useless of the world,” form—to use Norbert Elias’s expression—the “minority of the worst,” to which the “French” increasingly assimilate any and every child of immigrant background. In return, the feeling of having almost no hope of finding work and the dramatic narrowing of the future quickens resentments and explains the radicalization of young immigrants’ attitudes and behavior. This radicalization is reflected in an uncontrollable spiral that turns the violence constantly suffered (economic violence, violence of material poverty, violence of racism) into violence redirected sometimes towards themselves and often towards others—towards these “French” who hold them up to public obloquy.
(Beaud and Pialoux 2003, 65)

52 Once again in this text, we find a series of implicit propositions, of a general and causal nature, that explain the phenomenon in question:


  • violence suffered tends to generate violence in return, directed against those who inflict it. Resentment leads to radicalization and so forth. A psychological “mechanism” (resentment, which breeds to violence towards those who incite it) translates into a social mechanism, where violence generates an increase in violence;
  • practical propositions: in a position of drastically limited opportunities, individuals will be brought to choose those that are available, such as “hanging out;”
  • finally, all the reasoning and mechanisms described are obviously based on the existence of norms that lead to the formation of groups and of a sense of belonging to them, which means that we speak of the “French” or “immigrants.”

The Deductive Nature of Explanation

54 Using these two examples, we can suggest a generalization (based on a reading of the field studies available): ethnographic studies tend to employ implicit general propositions that make explanatory causal scenarios possible. Without the use of such general propositions, there would be no explanatory approach possible (as opposed to a description, which itself is based on general categorizations).

55 These general propositions fall into three categories: general behavioral propositions, propositions about the actions and interactions possible in given circumstances and, finally, propositions about the impact that the prevalence of norms and institutions has on people in certain contexts.

56 These general propositions are usually used without being made explicit in themselves. They form a stock considered to be relatively obvious, which helps interpret the situations being observed.

57 The use of such propositions therefore amounts to a deductive model. As part of explanatory procedures, which emphasize the “meaning” of the actions in question, use is made of underlying general propositions on which basis the information from the study is “explained.” This deductive model is fundamental to the analysis of social phenomena, and should not be contrasted with hermeneutics which, for its part, might ignore these procedures of deduction. We must instead stress the fact that procedures to elucidate meaning are deductive in nature (Føllesdal 1979; Mantzavinos 2005). This is true as much from the descriptive perspective, when a set of data is related to a single principle (like, for example, the role of the pure and impure in the formation of hierarchy described by Dumont [1966]), as from the explanatory perspective, when reference to a general principle is used to grasp a specific situation. For example, in a field study on New York doormen, Bearman referred to the role of the pure and impure in the hierarchy of professions: he explained doormen’s lesser social prestige compared to professions with an equivalent income by the fact that they are in contact with the disruptions of the street and the dirty side of life (dealing with trash), indicating that in all societies such situations are considered “impure,” and therefore inferior (Bearman 1995, 14). This general proposition may itself refer either to a psychological constant, or to a widespread social norm, but which would by no means be essential (and which would itself need to be explained).

58 Nonetheless, unlike the deductive model put forward by Hempel (1942) to characterize the form of explanation in general, including for social sciences, the general propositions on which explanations are based are not necessarily “laws” (Demeulenaere 2011). There is certainly no need to exclude, a priori, the existence of laws of behavior that may account for typical situations, like, for example, the fact that social disparagement in a meritocratic society tends to create resentment. Yet the general propositions on which explanations are based are not all laws.

59 As mentioned above, general propositions may correspond either to patterns of behavior, or to patterns in the constraints and possibilities for action and interaction, or finally to the presence of norms and institutions in certain situations that affect people’s behavior. This is how, in a different type of situation, the legal syllogism is formed which makes it possible to some extent, taking into account certain accepted norms, to anticipate court decisions based on deduction from general principles. It must be stressed here that in social life, these dimensions are constantly interacting, and are distinguished only from an “analytical” viewpoint to help clarify the procedures of description and explanation.

60 Where do these general propositions come from? As stated earlier, they belong to the interpretative stock which is available to ethnographers. If these general propositions are to have some validity, they are necessarily of empirical origin, formed elsewhere, prior to the ethnographic study underway. Behavioral knowledge such as “in a meritocratic society, individuals who experience social disparagement tend to feel resentful,” cannot be derived from the analysis of a particular field. In order to be built up and to enable an interpretation of the field being studied, it requires a large number of corroborating results through which the general proposition can be reached, on the basis of the comparison between multiple situations. The fact that it is general, and as such “theoretical” in relation to a singular event which it will help to clarify, does not mean that the proposition is not empirical in origin, insofar as it is itself descriptive of repeated past (or future) behavior. It is information on which the interpreter relies, and will employ in his or her explanatory reasoning, treating it as an explanans in his or her ethnographic approach.

61 At the same time, as always, this explanans can from another point of view and for other reasons be treated as an explanandum. The degree of generality of these propositions is a problem in itself; there is probably no reason to strictly and definitively oppose general theory and middle-range theory, as do Hedström and Udehn (2009) following on from Merton, to the extent that general propositions are always more or less general. Moreover, the same could be said with regard to the natural sciences (Kincaid 1990). There is therefore a need for reflection on the degree of generality of these propositions, their sphere of application, the way they are justified, their relevance in interpreting specific situations and the possibility of their assimilation to the status of law, and so forth.


62 The task of a theoretical and epistemological reflection on field research consists in:


  • identifying the general propositions involved in interpretation, both from a descriptive and an explanatory perspective;
  • distinguishing these propositions from each other in order to clarify the references to these general propositions in relation to the specific data collected during the particular ethnographic study;
  • examining the level of validity of these general propositions, their origin, and their justification, as well as the possible relationships between them;
  • determining their causal dimension and the way in which it can be established.

64 This task cannot be considered external to the ethnographic enterprise, but should be integrated, through an analytical approach, into the very process of writing. As has been mentioned several times, this process necessarily concerns quantitative research as well, in its procedures of explaining correlations between independent and dependent variables, which themselves make use of general propositions so that the explanation can be elaborated. Beyond this, it is a matter of creating a grammatical framework for the meaning and intelligibility of descriptions and explanations in the social sciences, in order to overcome the intuitive nature of procedures of “understanding” the data provided by ethnographic and quantitative studies, a task which also concerns the findings of real or simulated experiments used in the social sciences.


In this paper, we assert that we should not divide ethnography between empirical fieldwork and theoretical interpretation that would be a separate moment. The idea is that any ethnographic enterprise, whether descriptive or explanatory, is based on general or theoretical propositions that are not deduced from the direct fieldwork it aims to interpret. These general propositions are necessary to the interpretation of the fieldwork. The paper proposes a typology of these general propositions and an assessment of their localization in explanatory reasoning, which is mainly deductive.


  • ethnography
  • description
  • explanation
  • theoretical propositions

Bibliographic References

  • Affergan, Francis. 1997. La Pluralité des mondes: Vers une autre anthropologie. Paris: Albin Michel.
  • Bearman, Peter. 1995. Doormen. Chicago: The University of Chicago Press.
  • Beaud, Stéphane and Florence Weber. 1997 (2010). Guide de l’enquête de terrain. Paris: La Découverte.
  • Beaud, Stéphane and Michel Pialoux. 2003. Violences urbaines, violences sociales: Genèse des nouvelles classes dangereuses. Paris: Fayard.
  • Boudon, Raymond. 1990. L’Art de se persuader. Paris: Fayard.
  • Demeulenaere, Pierre. 2011. “Causal Regularities, Action and Explanation.” In Analytical Sociology and Social Mechanisms, edited by Pierre Demeulenaere, 173–197. Cambridge: Cambridge University Press.
  • Dumont, Louis. 1966. Homo hierarchicus : Le système des castes et ses implications. Paris: Gallimard.
  • OnlineFøllesdal, Dagfinn. 1979. “Hermeneutics and the Hypothetico-Deductive Method.” Dialectica 33: 319–336.
  • OnlineGeertz, Clifford. 1973. “Thick Description: Toward an Interpretive Theory of Culture.” In The Interpretation of Cultures: Selected Essays, Clifford Geertz, 3–30. New York: Basic Books. (“La description dense. Vers une théorie interprétative de la culture.” Enquête 6: 73–105, 1998.)
  • ———. Works and Lives: The Anthropologist as Author. Stanford: Stanford University Press.
  • OnlineHedström, Peter. 2005. Dissecting the Social : On the Principles of Analytical Sociology. Cambridge: Cambridge University Press.
  • Hedström, Peter and Lars Udehn. 2009. “Analytical Sociology and Theories of the Middle Range.” In The Oxford Handbook of Analytical Sociology, edited by Peter Hedström and Peter Bearman, 25–47. Oxford: Oxford University Press.
  • OnlineHempel, Carl. 1942. “The Function of General Laws in History.” Journal of Philosophy 39: 35–48.
  • OnlineKincaid, Harold. 1990. “Defending Laws in the Social Sciences.” Philosophy of Social Science 20: 50–83.
  • Lagrange, Hugues. 2010. Le Déni des cultures. Paris: Le Seuil.
  • Malinowski, Bronislaw. 1922. Argonauts of the Western Pacific. London: London University Press. [Les Argonautes du Pacifique occidental. Paris: Gallimard, 1989].
  • ———. (1926) 1985. Crime and Custom in Savage Society. New Jersey: Rowman and Allanheld.
  • ———. (1935) 1965. Coral Gardens and Their Magic. London: Routledge.([1965] 2002. Les Jardins de corail. Paris: La Découverte.)
  • OnlineMantzavinos, Chris. 2005. Naturalistic Hermeneutics, Cambridge: Cambridge University Press.
  • OnlineMauss, Marcel. (1947) 2002, Manuel d’ethnographie. Paris: Payot.
  • Passeron, Jean-Claude. (1991) 2006. Le Raisonnement sociologique. Paris: Albin Michel.
Pierre Demeulenaere
Pierre Demeulenaere is a professor of sociology at Paris-Sorbonne University. He specializes in the study of social norms and the epistemology of the social sciences. His most recent book is Analytical Sociology and Social Mechanisms (Cambridge: Cambridge University Press, 2011).
Uploaded on on 19/05/2014
Distribution électronique pour P.U.F. © P.U.F.. Tous droits réservés pour tous pays. Il est interdit, sauf accord préalable et écrit de l’éditeur, de reproduire (notamment par photocopie) partiellement ou totalement le présent article, de le stocker dans une banque de données ou de le communiquer au public sous quelque forme et de quelque manière que ce soit.
Loading... Please wait