Structural-functionalism and the network idea: towards an integrated methodology

Structural-functionalism and the network idea: towards an integrated methodology

So&E Networks, 2 (1980) 371-383 @Elsevier Sequoia S.A., Lausanne - Printed 371 in the Netherlands Stru~ura~-~un~iona~ism and the Network Idea: Towar...

1MB Sizes 3 Downloads 70 Views

So&E Networks, 2 (1980) 371-383 @Elsevier Sequoia S.A., Lausanne - Printed

371 in the Netherlands

Stru~ura~-~un~iona~ism and the Network Idea: Towards an Integrated Methodology Edward McCord

In this puper it is argued that, for the sake of analytical precision, we ought to understund structurLEl-fUnCtiOnalism to be a social network theory, as apprised to the &omn~un practice ofdiss~~~iuting the two. The rise of ‘network theor.v ’ ugainst the ba~kgr~urld of traditi~~lu~ structural-fuF~ct~~~na~isrn is reviewed, and the uppreciation by some anthropologists that social structure is itself in fact an abstraction from social networks is noted. 1n a more originul vein, it is shown that some network theories have themselves possessed a f~n~ti~n~l form.

Anthropologists who have engaged in the analysis of sociaI networks agree that network analysis helps to account for some data which would not be captured if treated in a structural-functional manner. Some anthropologists suggest that the st~~tural-f~~n~tional approach is explanatorily unacceptabte for any social system (Schneider 1974: 196; see Boissevain 1973:vii-viii). Others believe that the approach illuminates small-scale social systems, but that it is inappropriate for large-scale, ‘complex’, societies (Epstein 1969:77; Mitchell 1969:9-10). Most are simply content to call attention to a few instances in which network analysis reveals social forces that are clearly outside the viewpoint of the stn~~tural-functionalist, and do not commit themselves to statements about the adequacy of structL~ral-functionalism in other areas. But until we understand the relationship between structuralfunctional groups and social networks, some controversies that arise from the one standpoint or the other are likely to remain unsettled, or even misunderstood. Indeed, we have already just clouded the matter. I shall argue below that it is not the ‘social network’ framework of analysis as such, but rather certain independent claims about what forms of explanation and what explanatory concepts are best, that are opposed to structuralfunctionalism. For the sake of analytical precision, I shall explain, we ought to understand structural-functionalism to be a social network theory, as against the common practice of dissociating the two. A historical review accompaIlies my argL~ment.

*Department

of Philosophy,

University

of Pittsburgh,

U.S.A.

Roughly characterized, functionalism in anthropology is the doctrine that one’s explanatory ll~~derstaildii~g of 3. social or cultural j~~~e~lot~lcnon is substantially enriched by an analysis of the role played by the phenomenon in the stable functioning of some relevant system. For Malinowski, the systems involved were usually sin& living persons. Social or cultural phenomena were understood by way of an exposition of how they facilitated ~ndivi~l~lals’ ~~sycl~oio~ic~~l or biological f~lnctionitlg. Ra~lcliffe-drown followed Durkheim in taking the systems concerned to be societies themselves. Social or cultural phenomena were understood by an analysis of their role in the functioning of whole societies. Radcliffe-Brown directed fuiictioi~alist scrutiny only upon those phenomena which fall within his definition of ‘social structure’, and “structural-f~ulctionalism’ was the result The structural-functional ~~j~pro~lcl~directs the antllroj~olog~st in the field to locate the patterns to which individuals’ behavior for the most part and somewhat continuously conforms, and then to provide and defend at1 analysis of the role of those patterns with respect to one another in the functioning social system comj)rised of them. Ra~icliffe-Br~~w~i ~l~aintaiI]e~l (1952: 192) that “in the study of social structure, the concrete reality with which we are concerned is the set of actually existing relations, at a given moment of time, which link tog&her certain 1~~1man beings”. Ironically, he too called this the ‘network of social relations’. But he argued that the etliffogral~l~er should not describe these actual changing relations between individuals. Instead, the cthnographcr should abstract data from them to achieve a clcscription of social structural form as it remains relatively stable or contirliioL]s over the years. tltat is, in .e[luilibri~~I~~‘. As a consequence. the patterns of behavior noted by structural-functionalists have been those associated with the rules and values of institutional groups. As Boissevain puts it, the structtlral-functiollai approach takes society to be “a system of enduring groups composed of statuses and roles supported by a set of values and related sanctions which maintain the system in ~(~Liilibri~l~~”(Boissevain 1974:9). Bruce Kapferer is a ‘network theorist’ who has criticized structuralfunctionalism. In his analysis of a dispute among mine employees in Zambia, Kapferer noted that employees refer to public values in ~lefelldii~g their positions in the dispute, and he agreed that public values were appropriately consistent with some of the events one might want them to explain. Thus he wrote that by referring to public norms “some ~l~derstanding . . . has been obtained as to why an accusation of fast work should come from Stand 11 to Stand IV in Unit 3, and whv Donald should suspect Abraham of invoking a witchcraft sanction against him” (Kapferer 1969:207). The accusation of working too fast should have come from Stand II to Stand IV because the norm about work-rate served mostly to protect the interests of the older workers, and the ages represented at Stand II and Stand lV, respectively, were more contrastable as old versus young than were the ages represented

Structural-functionalism

and the network idea

373

at any other pairs of stands in the room (Kapferer 1969: 198, 199). Donald’s suspicion that Abraham might use witchcraft on him could be understood, according to Kapferer (1969:201), by referring to Donald’s violation of the norm that one should respect one’s elders (viz. Abraham) in working contexts, and by noting the commonly held belief among the Africans that older men, “largely because of their age and supposed familiarity with ‘traditional’ witchcraft practices, are the most dangerous and skillful exponents of the art” (Kapferer 1969:205). On the other hand, Kapferer maintained, such an appeal to institutional norms and beliefs is useless for answering such questions as “why specific individuals and not others were initially involved in the dispute, why certain issues and not others achieved prominence, and why this particular dispute should have resulted in a settlement in favor of one disputant and not the other” (Kapferer 1969: 183). Kapferer showed that often where institutional norms would seem by structural-functional accounts to be relevant to how a dispute ought to be resolved, those norms do not determine a unique outcome. Individuals can justifiably appeal to norms to defend conflicting views about what their role as participators should be. So it would evidently be useless to appeal only to institutional norms, as does the structural-functional approach, to explain individuals’ particular behavior. In his prefatory remarks, Kapferer wrote that: “structural-functional analysis often seeks out the regularities in behavior concentrating mainly on that social action which is consistent with the overall perceived morphology of the social system. Because of this, structural analysis often disregards behavior which is irregular and not consistent with the general structure of the society studied.” (Kapferer 1969: 183)

But this point rather confuses the issue. It is true that structural-functionalists sometimes ignore or explain away behavior that is inconsistent with theoretical expectations. Kapferer’s main proposal, however, was that structuralfunctional theories do not determine a unique class of expectations. Structural-functionalists “paint with a thick brush on a broad canvas” (Kapferer 1969: 183). But specific behavior occurs even when it is structuralfunctionally indeterminate. Unless such behavior is inexplicable, information that is not structural-functional is pertinent. Kapferer showed how ‘reticulum analysis’, a version of network analysis in which the networks are egocentric, seems to supply the additional explanatory data that are required. But now the question would arise why structural-functional theories, and not something along the line of these reticulum theories, should be thought to locate the important explanatory concepts, even for cases in which expectations are very narrowly determined by them. Kapferer concluded, for example, “that the general norms, values, attitudes, and beliefs which are overt in a situation of conflict are more the banners under which people act: they do not necessarily betray the underlying reasons for their action” (Kapferer 1969: 209). The idea was familiar from Edmund Leach’s pioneering departure from structural-functionalism, Politicul Systems of Highland Burmu (Leach 1964:264-278). There, the ‘underlying reasons’ for indivi-

374

E,‘.McCord

duals’ actions wcrc associated with self-interest. “Every individual of a society,” Leach argued, “each in his own interest, endeavors to exploit the situation as he perceives it and in doing so the collectivity of individuals alters the structure of society itself” (Leach 1964:8). Recent criticism of highly regarded structural-functional studies has somewhat bolstered misgivings about them. In a dramatic case, Meyer Fortes’ convincing analysis of Tallensi society from the standpoint of structuralfunctional equilibrium was substantially unsettled by Robert F. Stevenson’s historical approach to the same data (Stevenson 1968). Regarding EvansPritchard’s structural-functional model of the Nuer, ‘network theorist’ Jeremy Boissevain argued that, in spite of expectations encouraged by Evans-Pritchard’s study, Nuer who are related to one another patrilineally do not necessarily respond favorably to one another’s call for unity in fighting more distant relatives. Furthermore, the grouping of Nuer homesteads on the ground is not consistent with the importance Evans-Pritchard assigns to their patrilineal ideology; homesteads are sometimes related in genealogically diverse ways to the chief man around whose home they arc clustered (Boissevain 1974: 15- 16). Evans-Pritchard suggested that these ostensible deviations from his model are actually further evidence of its validity: “It is clear, consistent and deeply rooted lineage structure of the Nuer which pennits persons and farnilies to move about and attach themselves so freely, fol shorter or longer periods, to whatever community they choose by whatever . .. tie they find it convenient to enlphasize; .._ it is on account of the firm values of the structure that this flux does not cause confusion or bring about social disintegration. It would seen1 it may be partly just because the agnatic principle is unchallenged in Nuer society that the tracing of descent through wolnen is so prominent and matrilocality so prevalent. However much the actual configurations of kinship clusters may vary and change, the lineage structure is invariable and stable.” (Evans-Pritchard 195 1:X) But

in condemnation

of this maneuver,

Boissevain

pointed

out

that:

“people in Malta move about. as they do in Montreal and Amsterdam or any other place in the world without lineage structures, attaching themsevles freely to groups, big Inen, business houses, governnxnt departments, wonlen villages, guided tours and what have you by ‘whatever tie they find it convenient to emphasize’. The point is that they do this without a deeply rooted lineage structure.” (Boissevain 1974:17) Again, machinery

the presence is misleading,

of situations or does not

in which the lead anywhere

structural-functional at all, is reason

for being cautious in interpreting even the ‘successes’ of the approach. That there are deviations from structurally described Nuer behavior clearly deflates structural-functional accounts of even those situations in which deviations are absent. Or again, the proliferation which Blok (1973: 15 1) noted of talk about ‘loose’ or ‘informal social structure makes one wonder whether, or at least exactly why, structural-functionalism deserves acclaim when social structure is tight and formal.

Structural-functionalism

and the network idea

375

To put the point another way, many anthropologists have claimed that the success of structural-functional analysis is directly proportional to the complexity of the society or segment of society under study. But Mitchell rightly noted that: “the apparent complexity of social phenomena frequently bespeaks a lack of theoretical concepts available for their analysis. From this point of view segmentary societies were ‘complex’ until a set of analytical concepts incorporated in the idea of ‘lineage’became available with which to order the phenomena observed.” (Mitchell 1966:41) The ‘complexity’ of phenomena from the structural-functional point of view bespeaks a lack of structural-functional categories to achieve a convincing analysis. But if the behavior of individuals even within so-called ‘formal’ social settings is sometimes either indeterminate or downright anomalous according to structural-functional analysis, then perhaps it is presumptuous to equate structural-functional complexity with explanatory complexity. Explanatory complexity, of course, concerns more than manageability by means of analytical concepts; it also turns upon the strength and significance of conclusions obtained for a wide field of data by virtue of those concepts. We have needed an explanatory backdrop for structuralfunctional categories themselves, so that we can understand both the role of structural-functional institutions vis-d-vis particular instances of behavior, and the factors relevant to the presence or absence of such institutions. We have needed to undermine, for example, the explanatory distance between analyses of ‘formal’ societies and those of societies whose intractability from the structural-functional point of view has motivated the network approach. There were early clues that the network approach itself might ultimately help us in this.

Social networks ‘Social network analysis’, as commonly conceived, emerged largely because no useful structural-functional analyses of, for example, urbanized societies were imaginable. There was little concern for sketching networks in isolated, small-scale societies, since it was generally agreed that these could be handled well by the structural-functional approach. Naturally, then, a disparity between structural-functional and network concepts emerged as a result of some early definitions of ‘network’. For example, in her famous early study of the relevance of social networks to conjugal roles, ‘Elizabeth Bott wrote that: “in an organized group, the component individuals make up a larger social whole with common aims, interdependent roles, and a distinctive subculture. In network formation, on the other hand, only some, not all, of the component individuals have social relationships with one another. For example, supposing that a family, x, maintains relationships with friends, neighbors, and relatives who may be designated as A, B, C, D, E. F, .._ N, one will find that some, but not all, of these exter-

376

E. Mdord

nal persons know one another. They do not form an organized group in the sense defined above.” (Bott 1971:58)

But happily the trends have not favored the disengagement of network concepts from structural-functional arenas. Barnes ( 1969:6 1 -62), for example, criticized Bott for her suggestion, in a popular article, that networks do not exist outside urbanized societies. Acknowledging that he is right, she replied that “like Barnes and Srinivas, I also maintain in my more alert moments that networks of social relationship exist in all societies” (Bott 1971:317). Barnes’ views on this matter first appeared in a report on networks that preceded even Bott’s by a few years. In his study of social class and committees in Bremnes, a parish in Western Norway, Barnes adopted a broad understanding of ‘network’, although the particular kind of network he studied was narrowed to exclude groups. He characterized a ‘network’ as “a set of points some of which are joined by lines”, where the points “are people, or sometimes groups, and the lines indicate which people interact with each other” (Barnes 1954:43-44). He continued: “We can of course think of the whole of social life as generating a network of this kind. For our present purposes, however, 1 want to consider, roughly speaking, that part of the total network that is left behind when we remove the groupings and chains of interaction which belong strictly to the territorial and industrial systems. In Bremnes society, what is left is largely, though not exclusively, a network of ties of kinship, friendship, and neighborhood . . . . A network of this kind has no external boundary, nor has it any clear-cut internal divisions, for each person sees himself at the center of a collection of friends.” Without surrendering its power to detail particular dimensions of societies ~~ ties of kinship, friendship, and neighborhood in London or in Bremnes, for example - Barnes’ broad definition of ‘network’ enabled him to capture within the single framework of network methodology some differences between a/?_~ two social systems. Studies of urbanized societies and of societies that are not urbanized need not all proceed from different analytical standpoints. “We may note”, Barnes wrote, for example, “that one of the principal formal differences between simple, primitive, rural or small-scale societies as against modern, civilized, urban or mass societies is that in the former the mesh of the social network is small, in the latter it is large . . . This fact is of considerable practical importance for the study of societies by the traditional techniques of social anthropology, when we try to become acquainted with a limited number of persons whom we observe interacting one with another in a variety of roles. In a modern society, each individual tends to have a different audience for each of the roles he plays.” (Barnes 1954:44)

Barnes (1954:54) reported that through time the mesh of the social network in Bremnes was growing larger. Beteille and Srinivas came to the same conclusion about social structure in India. “Traditionally,” they wrote, “the villager lived in a narrow world where the ties of locality, caste, kinship and hereditary

service led back and forth between the same sets of persons ... . The situ-

Structural-functionalism

and the network idea

377

tion is changing in contemporary India. New interests tend to create relationships which cut right across the boundaries of the old established groups . . . . The network of social relations emanating from the individu~ does not as easily lead back to him. The closed circuit tends to become more and more open.” (Beteille and Srinivas 1964: 167) These reports indicated the capacity of network studies to relate smallscale and large-scale societies to one another within a single framework a capacity that goes hand in hand with their ability to describe formally at least some kinds of social change within a single society. The reports brought out a parallel between the degree to which a society is amenable to structural-functional description and the degree to which its networks possess certain characteristics, such as close-knittedness. BCteille and Srinivas, and others, suggested, in fact, that the network analyst has potentially at his disposal the constructs of s~~ctLlral-functional studies. As mentioned above, this is an idea which can even be traced to Radcliffe-Brown. Bt?teille and Srinivas ( 1964: 54) distinguished groups from networks by stressing the boundedness of the former, but they noted that groups appear when we compare each of the networks of relationships of a number of individuals to one another. The general idea was that if, and only if, individuals are members of the same group do they share certain multiplex relationships with one another. Mitchell (1973:3 1-3’2) was more specific. Members of a group, he maintained, will share certain aims and interests; a set of norms and rules will be accepted by all of them; they will be capable of acting in unison; their labor will differ in a way organized to further their common aims and interests. Furthermore, the presence of a group involves relationships extending beyond those actually existing between the group’s particular members. For example, membership relations will be durable in the midst of relplacements among the particular individuals who maintain them, and nonmembers will have access to membership criteria. After mentioning Barnes’ plea for a study of the partial networks within groups, Mitchell continued, in the ironically Radcliffe-Brownian vein:

“I would go further than this and say that the networks of relationships

are the starting point in the analysis of group behaviour and that they exist as analytical constructs which the observer erects partly by taking the participants’ perceptions into account and by fitting together observations not available to the participants themselves. The constructs may be at differing levels of abstraction, and when, at a fairly high level of abstraction, certain necessary conditions obtain which enable the observer to subsume a large part of the behavior in terms of a few compact formulations, he will be dealing with institutions and structured groups.”

It follows from this that any senses in which network and structuralfunctional methods are at odds with one another are not a matter of the presence or absence of certain constructs within the data yielded by network studies. Note also that criteria of the kind provided by MitcheIl will vary, depending upon the particular definition of ‘group’ that is adopted. Boissevain (1968:545), for example, pointed out that Mitchell’s definition would exclude clans whose members do not interact and, in fact, “are often dispersed and unknown to each other”.

378

E. McCord

If network methodology is unable to capture the kinds of analyses characteristic of str~~ctural-functioilalism, it is not because it lacks the relevant constructs. But there have been important extrinsic differences between the two approaches. In network analysis, groups appear when one compares a number of personal networks of various contents with one another. In other words, groups are associated explicitly with the superimposed configurations of the partial personal networks of which they consist. In st~ct~~ral-functional analysis, however, groups are generally not given substance by observations of personal networks. Instead, the investigator relies upon the fact that groups and institutions are interpersonal, public units. Religious or political institutions may publicize their own presence, for example. Frequently, groups have their own names and are a subject of public intercourse, so that the researcher learns of their existence straightforwardly, bypassing details about the personal networks of members. In so doing, of course, he runs the risk of misunderstanding exactly what he has discovered. For example, he may think, rightly or wrongly, that publicly professed allegiance governs private planning, or that publicly evident political organization is a self-sufficient influence on personal behavior. He will miss any dimensions of the relevant personal networks that resist such public expression. He will most likely miss noticing whether the multidimensional partial network that is an individual’s personal contribution to some group differs from the multidimensional partial network of others in the same group. There is only one group, of course, but the ties of individuals to it need not all be alike. Within a single group, different group values may be cherished by different members, and even among those group values that are unanimously cherished, differences in the intensity of members’ personal attachment to them may exist. These differences will be overlooked if the personal contribution of individuals to interpersonal units is neglected. Another difference between the network idea and structural-functionalism is that the former entails fewer conceptual and explanatory coIl~Initments than the latter. An exclusive adherence to st~ctL~ral-fu~lctionalism commits one to the view that social structural patterns exhaust the constructs available to anthropological investigation. The network idea, as we have seen, provides by itself a framework that accommodates both personal networks and groups, depending upon the level of abstraction, and it harbors no necessary presLlp~?ositio~is about whether one unit of analysis will turn out to be more potent in explaining phenomena than another. In other words, so far we have every reason to think of the network idea as the basis of a comprehensive methodology, of which structural-functionalism could be one special manifestation. Explanation We can further dramatize the simplicity and flexibility of the network idea through a discussion of explanatory form. Structural-functionalism, of

Structural-functionalisnz

and the network idea

379

course, is a version of functionalism, and it is philosophers, rather than anthropologists, who have shed the most light upon functional explanations. Consider, for example, the following stage of the classical treatment by Carl Hempel, in which our interest is in explaining the occurrence of a trait i in a system s at a certain time t (Hempel 1965:3 10): At t, s functions adequately in a setting of kind c (characterized by specific internal and external conditions). (b) s functions adequately in a setting of kind c only if a certain necessary condition, n, is satisfied. (c) If trait i were present in s, then, as an effect, condition n would be satisfied. (d) (Hence) at t, trait i is present in s.”

“(a)

This characterization of functional explanatory form involves faulty reasoning, Hempel explains. The conclusion (d) derives from the premises by way of the fallacy of affirming the consequent. If we avoid this by rephrasing (c) so that i is a necessary condition for the satisfaction of tz, then the truth of (c) becomes very doubtful. But some anthropologists have made such claims. Here Hempel cites Malinowski’s highly dubious opinion that magic satisfies a need in society that no other conceivable element of society could satisfy. Other sociologists’ remarks, however, suggest replacing the single trait i in the analysis by the set I consisting of i and its functional equivalents. This would mean that (c) and (d) would become (c’) and (d’), as follows (Hempel 1965:313): “(4.3) (c’) I is the class of all empirically sufficient conditions for the fulfillment requirement n in the context determined by system S in setting c. (d’) Some one of the items included in I is present in S at t.”

of

Note that by virtue of the definition of I we avoid the fallacy of affirming the consequent. Hempel believes that this characterization of structural-functional explanation trivializes the conclusion (d’), for we generally do not have much information about the items contained in class I. There is also the problem that this characterization does not enable conclusions to be drawn about the items present in I at any other time than that at which (a) is established. But this is remedied by simply dropping (a) from the characterization and replacing (d’) by (d”), yielding the following full schema (Hempel 1965:316): “(5.1) (b) System S functions adequately in a setting of kind c only if condition n is satisfied. (c’) Z is the class of empirically sufficient conditions for n in the context determined by S and c; and I is not empty. (d”) If S functions adequately in a setting of kind c at time t, then some one of the items in class I is present in S at t.”

This characterization does justice to the way some structural-functionalists formulate their predictions. Those anthropologists are mistaken, on the other hand, whose predictive statements occur within a framework like (4.3). Of course, (5.1) still suggests that functional explanations generally

380

E. McCord

only warrant rather trivial conclusions. Hempel (196.5:329) infers that the job of functionalists is mostly heuristic ~~ “to search for specific selfregulatory aspects of social and other systems and to examine the ways in which various traits of a system might contribute to its particular mode 01 self-regulation”. This latter idea that functionalists seek to show how traits of a system is the cornerstone of Robert Cummins contribute to its functioning of functional explanation, although Cummins recent interpretation mistakenly thinks that Hempel overlooked the idea. Cummins (1975 :74 1 ) claims that classic treatments of functionalism like Hempel’s proceed bander tlic following misconceptions: “(A) The point of functional characterization in science is to explain the presence of the itern (organ, mechanism, process, or whatever) that is functionally characterized. (B) For something to perform its function is for it to have certain effects on a containing system, which effects contribute to the performance of some activity of, or the maintenance of some condition in, that containing system.” Regarding

(A).

Cummins

believes

that

what

one

is explaining

is not

the

presence of the item functionally characterized. but rather the capacities ot systems of which the item is a part. Clearly, argues Cummins ( 1975 :745 -746), we cannot deduce hearts from circulation. and even if we coulcl that would no more espluitl tlic presence of hearts than “deducing the presence and height of a building from the existence anct length of its shactow woultl explain why the building is there and has the height it does”. Cummins move is to discard (A) and propose that something like the following exemplifies the typical kind of functional explanation (Cummins 1975 :749): “(5) a. Vertebrates incorporatin g a beating heart in the usual way (in the way s does) exhibit circulation. b. Vertebrate s incorporates a beating heart in the usual way. c. Hence, s exhibits circulation. Cummins’ second move is to undermine (B) hy stressing

(5a). He writes (Cummins

a difficulty

with

1975:753):

“The problem is that, whereas the production of certain effects is essential to the heart’s perfortning its function, there are sonic effects (e.g. heart sounds) production of which is irrelevant to the functioning of the heart. This problem is bound to infect any ‘selected effects theory: i.e., any theory built on (B).” Rather than embrace proposals that salvage (A) or (B) at some expense,

Cummins (1975:762) suggests that we revamp our conception of functionalism. We attribute functions to entities, he believes, as a result of efforts to explain capacities of systems that contain them not, in accordance with (A). as a result of efforts to explain the presence of the elements in the systems. In overcoming the problem with (B), Cummins emphasizes the role of the background concerns that guide our inquiries about the capacities of systems. He writes that: “it is appropriate to say that the heart functions as a putup against the background of an analysis of the circulatory system’s capacity to transport food, oxygen.

Structural-jimctionalism

and the rletwork idea

381

wastes, and so on, which appeals to the fact that the heart is capable of pumping. Since this is the usual background, it goes without saying, and this accounts for the fact that ‘The heart functions as a pump’ sounds right, and ‘The heart functions as a noise-maker’ sounds wrong, in some context-free sense.”

The following formulation, therefore, captures Cummins’ proposal: x functions as a 4 in s (or: the function of x in s is to r+4)relative to an analytical account A of s’s capacity to $, just in case x is capable of @ing in s and A appropriately and adequately accounts for s’s capacity to $ by, in part, appealing to the capacity of x to $ in s.

Note that neither Cummins’ nor Hempel’s analysis mentions groups or institutions. That is not just because Cummins and Hempel are interested in functionalism in biology. Hempel’s treatment, in fact, seems primarily motivated by sociological concerns. Rather, the reason is that neither a particular unit of analysis nor even a particular subject matter plays a role in characterizing explanatory form. As far as explanatory form is concerned, individuals, networks, groups, or any other constituents of society may appear in functional explanations in sociology. Assuming that we have some rigorous characterization of social equilibrium, network studies would provide a Hempelian functional explanation if they were to supply evidence that at least one social phenomenon constitutes empirically sufficient conditions for the equilibrium of a social system in a particular setting. For example, there is empirical support for viewing a set of three individuals who all like one another, or of which two like each other and do not like the third, as ‘stable’ in a way in which a group of three individuals, two of whom are friendly with a third but not with one another, is not. One’s friends are suspect if their friends are one’s enemies. Thus, a clear exposition of one kind of social equilibrium appears in the form of the definition of ‘balance’ which Cartwright and Harary (1956:277-293), under Heider’s inspiration, provide. Adapting their formulations to our purposes, we might get the following scheme, commensurate with Hempel’s (5.1) above: (1) For any individuals A, B, C, . .. E, and F, define a circle as the set containing the relationship between A and B, the relationship between B and C, . . . . the relationship between E and F, and the relationship between F and A. Then the system s is balanced in a setting of kind c (in which A, B, C, ..,, E, and F remain alive and socially functional, for example) only if the number of generally unfriendly relationships in any circle in s is even. (2) I is the class of empirically sufficient conditions for there being an even number of generally unfriendly relationships in any circle in s, given setting c. (3) If s is balanced in a setting of kind c at time t, then some one of the items in class I is present in s at t.

The class / might itself include features of networks but, in any evidence that social systems of a particular kind fit the scheme, and or another social phenomenon is a member of I, would obviously network analyses, ‘since a circle is a network. Here, then, is Hempelian functional explanation that arises from a mainstream idea.

case, the that one include a simple network

382

E. lwcCord

It is also easy to associate model. The following example

typical network is sufficient:

information

with Cummins

Relative to an analysis A of political faction F’s capacity to raise a great deal of tnoney, distribute propaganda widely, infiltrate opposition ranks, etc., the function of person P is to cultivate political contacts. For P is capable of cultivating political contacts, and A appropriately and adequately accounts for F’s capacity to raise trioney, distribute propaganda, infjltrate opposition ranks, etc., by, in part, appealing to the capacity of P in F to cultivate political contacts. For our purposes, it is intended that an explanation and substantiation of P’s capacity to cultivate political contacts would include a display of the social network. The more central P’s position in the social network, the more effective will be his or her ‘function’ in the sense concerned.

Summary and conclusions We have seen that differences between network analysis and structuralfunctionalism are neither a matter of there being analytical constructs available to the latter approach which are not available to the former one, nor of the inadmissability of functional explanatory form within the yield of network studies. We ought to think of network analysis as a versatile and conceptually fundamental enterprise of which structural-functionalism is one potential fragment. Structural-functionalism, in other words, is a network theory. This integration of the approaches would help bring structuralfunctionalism into perspective. On the other hand, we have seen how structural-functionalists’ basis of awareness of groups could mislead them about the nature of those groups, and how an inability to handle many social phenomena other than groups jeopardizes their ability to compare large-scale and small-scale societies. or to account for change within any one society. We have seen how problems arising from a narrow commitment to structuralfLlnctionalism raise doubts about the source of any explanatory success Even the widely made distinction claimed by the theory’s adherents. between ‘complex’ and ‘simple’ societies may be misstated. It would be a mistake to think that network theorists could not fall under similar kinds of deception. In their move away from structuralfunctionalism one can imagine network theorists assuming unnecessarily that the explanations they seek are all egocentrically based. It tnay turn out, however. that constructs of higher abstraction than personal networks will play very fundamental roles within the best social theories. Boissevain (1968: 543.L556) and Blok (1973: 15 1~~165) are among those who have already warned against anthropologists’ restricting thetnselves narrowly to the analytical level of personal networks. Similarly, the upshot of our discussion is that the network approach and the structural-functional idea are by no means incompatible, and that investigators should discover what kinds of explanations and what kinds of explanatory units are fundamental in the cultural theory that properly integrates tne two approaches.

Structural-jimctionalism and the nehvork idea

383

References Barnes, J. A. “Class and committees in a Norwegian island parish”. Human Relations 7:38-58. 1954 “Networks and political process”. In J. C. Mitchell (ed.), Social Networks in Urban Situa1969 tions. Manchester: Manchester University Press. Beteille, A. and M. N. Srinivas “Networks in Indian social Structure”. Man 64: 165-168. 1964 Blok, A. Coalitions in Sicilian peasant society. In J. Boissevain and J. C. Mitchell (eds.), Network 1973 Analysis: Studies in Human Interaction. The Hague: Mouton. Boissevain, J. 1968 “The place of non-groups in the social sciences”. Man (New Ser.) 3: 542-556. 1973 “Preface”. In J. Boissevain and J. C. Mitchell (eds.), Network Analysis: Studies in Human Interaction. Manchester: Manchester University Press. 1974 Friends of Friends. Oxford: Blackwell. Bott, E. 1971 Cartwright, 1956

Family and Social Network. New York: Free Press. D. and F. Harary “Structural balance: a generalization of Heider’s theory”. Psychological Review 63:277293. Cummins, R. “Functional analysis”. Journal OfPhilosophy 72:741-765. 1975 Epstein, A. L. “The network and urban social organization”. In J. C. Mitchell (ed.), Social Networks in 1969 Urban Situations. Manchester: Manchester University Press. Evans-Pritchard, E. E. Kinship and Marriage Among the Nuer. Oxford: Oxford University Press. 1951 Hempel, C. G. 1965 Aspects of Scientific Explanation. New York: Free Press. Kapferer, B. 1969 “Norms and the manipulations of relationships in a work context”. In J. C. Mitchell (ed.), Social Networks in Urban Situations. Manchester: Manchester University Press. Leach, E. 1964 Political Systems of Highland Burma. Boston: Beacon Press. Mitchell, J. C. 1966 “Theoretical orientations in African urban studies”. In M. Banton (Ed.), The Soctil Anthropology of Complex Societies. London: Tavistock. 1969 “The concept and use of social networks”. In J. C. Mitchell (ed.), Social Networks in Urban Situations. Manchester: Manchester University Press. In J. Boissevain and J. C. Mitchell (eds.), Network 1973 “Networks, norms, and institutions”. Analysis: Studies in Human Interaction. The Hague: Mouton. Radcliffe-Brown, A. R. 1952 Structure and Function in Primitive Society. New York: Free Press. Schneider, l-1. K. 1974 Economic Man. New York: Free Press. Stevenson, R. F. 1968 Population and Political Systems in Tropical Africa. New York: Columbia University Press.