Gossiping

 

Nicholas Emler Oxford University, UK

 

INTRODUCTION

 

Opinions differ on the subject of gossiping. On one side are views that it is at best a waste of any intelligent person's time and attention, and at worst so dangerous as to merit the severest punishment. On the other side, its benefits are alleged to be so numerous and certain it should be consumed more regularly than vitamin C, and should be more freely available. Views of the first kind accord more closely with popular opinion across numerous societies and periods of history (Schein, 1994; Spacks, 1985). The second permeates the writings of social scien­tists (ct. Gambetta, 1994), although it should be said that one of the more striking features of scholarly writing on the topic is that there is so little of it. In this chapter I consider the grounds for these contrasting views, as well as those for a third with which, as the reader will readily detect, I have more sympathy. This is the view that gossiping is the foundation and basis for the social life of humans as distinct from all other social animals so far evolved.

 

The chapter is organized as follows. The first part briefly reviews the public or popular image of gossip and considers some consequences of this image. The second part provides a brief survey of the writings of anthropologists and so­ciologists on the subject of gossip. Mindful that this Handbook as a whole is concerned with the social psychology of language, and not with its social history, anthropology or sociology, I confine observations to those most helpful to the construction of a context for a social psychological treatment of gossip. The third part seeks to separate fact from mythology, to establish what and how much we know about the phenomenon, leading finally to a consideration of the possible significance of gossip - why people do it and with what consequences. But first of all the vexed issue of definition must be confronted.

 

Although in works of this kind it is the convention to begin discussion of a phenomenon by setting out one's definition. in this particular case this first step is more than usually hazardous. A real hazard is that one will from the outset offend and alienate readers whose favourite conviction about this subject seems thereby excluded. Unfortunately, refuge behind the dictionary defini­tion is not especially helpful. The OED definition - "To talk idly, mostly about other people's affairs; to go about tattling" - merely reveals the difficulty: opinions about gossip are seldom neutral. In popular parlance to describe someone as gossiping is to make a judgment and typically a criticism. However, if we are better to understand the phenomenon, and not merely to reinforce prejudices. it is essential to begin with a behavioural definition that is neutral and non-judgmental. For this reason gossiping wil1 be defined as "informally exchanging information or opinion among two or more persons about named third parties".

 

For the moment I would point to the following feature of this definition. It makes no stipulations about the kinds of people likely to be involved in such exchanges or their motives, about the truth, validity or value of what is ex­changed. or about the consequences of these kinds of exchanges. These all figure prominently in popular images of gossiping, as we shall see in more detail shortly. but I propose they should be treated as questions to be decided em­pirically rather than features to be assumed a priori.

 

THE PUBLIC REPUTATION OF GOSSIPING

 

Among human activities, gossiping has had a dismal reputation. It is true that the degree of opprobrium attached to this particular activity has seen considerable fluctuation over the centuries and that there remain wide cultural variations in the strength of opinion against it. Nonetheless, the same accusations reappear in the literature and folklore of different societies and epochs with considerable consistency (Bergmann, 1993; Schein, 1994; Spacks, 1985). Thus the following are common features in popular images of gossip (d. Emler, 1994; in preparation):

  • participants: people who are intellectually shallow, idle, disloyal, indiscreet; women;
  • motivations: prurient interest in others. voyeurism, nosiness (listeners); treachery, malice, spite (tellers);
  • content: private matters; trivia;
  • truth value: unreliable, inaccurate, error-ridden; untrue (lies);
  • effects: invariably negative, potentially damaging, sometimes catastrophically so. to the reputations, livelihoods and lives of others; trust is betrayed; privacy is violated.

 

This representation of gossip has had two important consequences, one relating to the treatment of those involved, the other the behaviour of those upon whom suspicion of involvement is most likely to fall. In the first case we should expect that if the consensus within social groups holds that gossiping is a damaging activity then there should be sanctions against gossips that bear some relation to the degree of threat they pose. This is indeed the case. Sanctions have ranged from the implied or anticipated disapproval of others to more robust and direct remedies including scolds' bridles, ducking stools, torture, mutilation, and death by burning. In our more enlightened times we are inclined to forget, for example, that gossip was outlawed in medieval England (Oakley, 1972).

 

As to the chief suspects, they have always been women (Schein, 1994; Aebischer, 1985) and the climate of suspicion has had clear effects on women's behaviour, as a number of anthropological studies show. One well­ documented effect is extreme care about whom one is seen talking to (Hut­son, 1971), achieved in at least one community by avoidance of social con­tacts in public places altogether (Naish, 1978). In other words threats of informal and formal sanctions against gossiping have apparently caused those most under suspicion to take evasive action and thus to minimize their overt participation in gossip.

 

At this point we need to ask about the relationship between gossiping and its reputation. We might first consider the possibility that popular views of gossiping are no more than honest and broadly accurate opinions about a form of language use. This might be called the "naive realism" position. Do the various claims associated with this position fit the facts?

 

GOSSIPING: VIEWS FROM THE SOCIAL SCIENCES

 

Many of the theoretical arguments about gossiping seek to account for one or other of the features that form the popular image described above. This section will outline some of the more influential of these together with the idea that gossiping is significant only in a particular kind of society, the pre-industrial. Consider first of all the content of gossip.

 

The idea has been advanced in both sociology and social anthropology that informal conversations in which others are criticized can serve a social function. The argument that gossiping operates as an important means of informal social control in primary groups has a long history (see, for example, Ross, 1901) but it is most closely associated with the anthropologist Gluckman, who proposed that gossip maintains "the unity, morals and values of social groups". It does this by voicing criticism of specific actions of named individuals within the group, thereby reaffirming the boundaries of acceptable behaviour and reminding po­tential offenders of the costs of transgression (Gluckman, 1963, 1968; see also Boehm, 1993; Elias & Scotson, 1965; Haviland, 1977; Merry, 1984). A similar thesis has been advanced by social psychologists Sabini and Silver (1982): social norms are sustained, they propose, only by continually circulating and discussing concrete instances of their violation.

 

According to these views, gossip becomes a positive contribution to collective life rather than a destructive and disruptive activity. Can this be reconciled with another popularly claimed feature of gossip, namely its factual inaccuracy? How could criticism of others provide a useful service if this criticism is regularly mistaken? Two other interpretations of gossip can explain its inaccuracy but would not accord it any particular socially beneficial functions. According to the first of these, gossip will be inaccurate purely as a function of the nature of communication systems. All such systems are subject to a degree of information loss. The more links there are in a communication chain, the more the informa­tion will be degraded at each successive transmission. Allport and Postman (1947) adapted Bartlett's (1932) classic demonstration of the fallibility of mem­ory to demonstrate this progressive attenuation, selection and distortion of infor­mation as it passes along a chain of human communicators. This assumes, of course, that gossip is directly analogous to the game of Chinese whispers, an assumption I shall reconsider further on.

 

A different focus of explanation for the unreliability of gossip invokes not the inherent imperfections of communication systems but the motivations of the information transmitter. Barkow (1992), taking the perspective of evolu­tionary psychology, argues that as genetic competitors we are programmed to disseminate misleading information about sexual rivals. This explanation also makes sense of the negative, indeed slanderous content of gossip. Barkow draws support for his conclusion from research by Buss and Oedden (1990) apparently showing that young adults are strongly disposed to derogate sexual competitors.

 

As regards the destructive effects of gossip, in so far as theory touches on these at all it does so by implication. Barkow's view, for example, is consistent with the idea that gossip is a form of covert and indirect aggression, a verbal assault on others' reputations. Some of the explanations for women's greater propensity for gossip also emphasize its aggressive nature and purpose. Thus, Scheler (1921) argued that because women are physically weaker than men they are both more vindictive and must use non-physical means to attack others.

 

Other writers, while equally disposed to accept that there is an association between women and gossip, interpret the link in a more positive light. For Jones (1980) women gossip to sustain solidarity in their identity as women; she does not make it clear, however, why men should not gossip among themselves for the same purpose. Tannen (1990) regards women's gossip as an essential basis for friendship; it allows them to form and sustain relationships with one another that are capable of providing strong mutual support. This argument is based partly on the claim that women are better able than men to chat about nothing in particu­lar, a facility that allows them to keep a relationship going when there is no specific task or problem to discuss. A slightly different kind of explanation starts from the limited power of women and argues that gossip is a weapon of subver­sion (e.g., Schein, 1994). Thus we are back with the position that gossip is dangerous but how sanguine one feels about this presumably depends on one's degree of sympathy with those who might be endangered.

 

Finally, a different kind of theme within the social sciences concerns the declining significance of gossip. The argument goes that if it is a mechanism of social control it is best suited to the small-scale face-to-face communities of the pre-industrial age. If an instrument of aggression and competition again it is most likely to be effective in closed communities in which all members are mutually acquainted. And gossip can only threaten reputations so long as indi­viduals inhabit social worlds in which reputations can exist and matter, namely worlds in which they are personally known to most of the other people they encounter and with whom they do business. The kind of society created by the Industrial Revolution supposedly swept away these particular social conditions. According to the authors of a sociology text published in 1950 "In the large community of the modern city, contacts in secondary groups tend to be imperso­nal and escape into anonymity is possible. Under these circumstances, gossip and ridicule are less effective instruments and their place is taken by the police and the courts" (Ogburn and Nimkoff, 1950, p. 115; see also Locke, 1998).

 

Thus far we have seen there are theories to explain one or another feature of gossiping but no theory which appears successfully to account for all of them (and there is little if any theorizing about the reasons why gossip should hold a special appeal for those who are intellectually shallow, gullible or idle). A pos­sibility we should now consider, therefore, is that these theories may be at fault because they have misrepresented the phenomenon and too often taken naive realism at face value. Or to put it another way, they have sought to explain properties of gossiping without any strong evidence that it does indeed have these properties. So let us now look at what evidence as distinct from opinion there is about gossip.

 

SEPARATING FACT FROM THE IMAGE: ISSUES OF METHOD

 

It would be wrong to suggest that the theories identified above have without exception been completely speculative, unsupported by any empirical obser­vation. At the same time there has been very little systematic hypothesis testing in this area. The reason we have so little secure evidence about the phenomenon of gossiping is not hard to find. The first step in scientific inquiry is, or should be, a sound and objective description of the phenomenon of interest. One needs, in other words, first to establish what is happening, before moving on to ask how it happens and why it happens. Unfortunately, social science's two preferred methods for answering what questions, direct observation and self-report, are not ideally suited to this particular phenomenon.

 

Social anthropology has provided a rich stock of reflections on gossip; indeed this literature has been the principal source of ideas for social scientists interested in the phenomenon. Moreover, most of the material has been derived from direct observation. This is not really so surprising because research methods described technically as "participant observation" or "ethnography" are in prac­tice attempts to participate in the gossip of a community (ct. Malinowski, 1922). However, Bergmann (1993) takes care to point out that as a source of evidence to test hypotheses about gossip these observations have been limited; their primary purpose was not to document and understand gossip as a phenomenon in its own right but to employ gossip as a means of ethnographic inquiry.

 

Researchers studying gossip in its own right have more often observed without participation. An example is provided by Levin and Arluke's (1985) study of sex differences in gossip (see also Kipers, 1987), which involved eavesdropping on the conversations of students in a residence. An alternative which would seem to offer various obvious advantages is to set up a conversation under laboratory conditions and record it with the full knowledge of the participants (e.g., Leaper and Holliday, 1995). For example, the nature of the relationship between the participants can be determined more accurately, more information can be col­lected about the content of the conversation and recorded more accurately, and potential ethical problems are avoided. The disadvantages are equally obvious and rather more weighty, such that resemblance to gossip in its natural state may be rather tenuous (cf. Weick, 1985).

 

But direct if covert observation of spontaneously occurring conversations also has some significant limitations. First, it is expensive and though cost is not an inherent limitation its practical consequences need to be acknowledged; sample sizes for such research have so far been small. Second, it cannot by itself tell us a number of important things about those involved, in particular the relationship between them. Third, direct observation is limited to certain kinds of settings, namely public places in which speech can be overhead by people (the observers) for whom it was not intended. Conversations in such public places are not necessarily representative of conversations in other, less accessible locations. It is entirely possible that people are more guarded about what they will say concern­ing third parties when they know they can be overheard.

 

Perhaps for reasons of cost, some researchers have made use of the major alternative to direct observation, self-report. This method has been used so far to answer three questions about gossiping: what kinds of people are most likely to do it (Nevo, Nevo & Derech-Zehavi, 1993a, 1993b, 1994; Jaeger et aI., 1994), what kinds of people are most likely to be the targets of gossip (Jaeger et aI., 1994; Jeager, Skleder & Rosnow, 1997), and what attitudes people have to gossiping (Wilson et aI., 2000).

 

In so far as self-reports are employed to provide behavioural descriptions ­answering the kinds of questions posed in the Nevo et al. and Jaeger et al. studies - we are confronted with some quite severe problems of validity. There are at least two reasons why people's self-reports of their own gossiping behaviour may lack validity. The first is that self-reports may be consciously or unconsciously distorted by self-presentational motives. Thus anyone sensitive to the disreputa­ble image of gossip might not wish to claim they do it very much. For example, Jaeger et al's (1994) finding that people who self-reported gossiping least also had significantly higher scores on a need for social approval measure may have reflected differences in the tendency to distort self-reports in a socially desirable direction.

 

However real the risk of distortion here, it is probably less serious than a second threat to the validity of such self-reports, the sheer difficulty of giving factually accurate answers about one's own gossiping. Reis and Wheeler (1991) identify three sources of inaccuracy in self-reports. As they note, questionnaires typically require of respondents that they give a global assessment of some activity, for example frequency of interaction with close friends. This requires that the respondent (a) makes an appropriately representative selection of oc­currences of the activity in question, (b) accurately recalls the relevant features of these occurrences, and (c) aggregates or combines the recalled events to provide the requested global assessment. Each of these requirements is liable to introduce inaccuracies, all the more so to the extent that the activity in question­occurs often, is routine, and is mundane, which is to say unmemorable. There is now ample evidence that self-reports of activities and events with these qualities - such as exercise, diet, television viewing, and mood - can be very inaccurate.

 

Communicative activity would appear to fall into this category. In a series of studies Bernard, Killworth and Sailer (e.g., Bernard & Killworth, 1977; Bernard, Killworth & Sailer, 1982; Killworth & Bernard, 1977) compared people's re­collections, or their predictions, of whom they talked to and how often, with independent records of their communications. There were major inaccuracies of every kind; individuals communicated with were frequently forgotten, others were recalled when they had not been in contact during the specified period, and about half the time respondents were wrong about the person they communi­cated with most frequently. Bernard et a!. (1982) concluded that "what people say about their communications bears no resemblance to their behaviour" (pp. 30-31), a fairly damning judgment on the value of self-reports in this area (but see Freeman, Romney & Freeman, 1987; Kashy & Kenny, 1990). What we do not know is how accurate recall of the content of routine conversations is likely to be, and this is crucial to testing hypotheses about gossip.

 

The limitations of direct observation and self-report have attracted some re­searchers to a third option: event and experience recording methods in which the actor becomes the observer/recorder. These methods therefore combine some of the features and advantages of both self-report questionnaires and direct, sys­tematic behavioural observation. They also, it should be acknowledged, contain some of the weaknesses of each (Stone, Kessler & Haythornthwaite, 1991). Nonetheless, these methods have become increasingly popular with researchers studying social interaction because they can provide quite large samples at rela­tively low cost and because they can provide data that would be very difficult to collect by other methods.

 

In what follows I shall draw upon evidence provided by these newer methods, but recognizing their imperfections I shall not ignore evidence generated from self-reports and direct observation. If quite different research methods give us similar answers to our questions about gossip, we can have some confidence in the answers. If they do not we shall need strong grounds for preferring the answer provided by one of them.

 

MORE RADICAL STEPS

 

We now need to consider what questions could usefully be answered if we are better to understand the what, how and why of gossip. My list of such questions includes the following more general parameters: What do humans spend their time doing, and more particularly what proportion of that time is spent in the company of others, interacting with others, and talking to others? In so far as time is spent talking to others, who are these others? Where do these conversa­tions take place, and how do they arise? What are the topics of these conversa­tions? Evidently, only the last of these is directly about gossip. But if the answer to the last question is that a high proportion of the content fits my definition of gossip - information or opinion about named third parties - then answers to the others will have told us a great deal about its nature and significance. To antici­pate, I think the best evidence available justifies the following conclusions: most people naturally spend much of their waking time - between 60% and 80% - in the physical presence of others. Where circumstances allow, much of this time is spent in conversation; indeed, conversation is the most common form of social interaction. Most conversations occur between people who know one another and most are unscheduled - they occur without prior arrangement. Most occur within private or institutional settings. The most freqent topics of conversations are the doings of named third parties and relationships with and among third parties. These conclusions of course require some significant qualifications and they still leave unanswered the why questions about gossip - what is gossip for­but they do give us some powerful clues. To explore these, however, we need first look in more detail at the evidence which leads to these conclusions.

 

The Place of Talk in Human Time-Energy Budgets

 

Our scientific knowledge of other species of animals is now likely to include details about the proportions of time they typically spend sleeping or resting, feeding, foraging and procreating. These kinds of details have, for example, supported arguments about the degree to which different species are naturally social or solitary. Thus there is evidence on the proportion of time macaques, as compared to baboons or gibbons, spend feeding, at play, in social observation, etc. Equivalent data about humans is both harder to come by and more difficult to interpret with confidence. A massive cross-national study on the use of time, based on both interviews and self-recording (Szalai, 1972), did reveal something of the characteristic daily pattern for a number of cultures and for different categories of people within these cultures. For example, the data showed that on weekdays adult male Belgians spent an average of 8 hours in sleep, 6.3 hours working, 1.5 hours watching television, 1 hour travelling, 2.9 waking hours alone, and 1 hour reading (the categories here are not all mutually exclusive). Typically, across cultures people spent about 80% of their waking time in the company of others, the daily average ranging from 13-plus hours for the Yugoslav sample to 10 hours for an American sample.

 

Interpreting these figures is a problem because the societies in the survey do not correspond to any natural pattern. All are products of historical and not merely biological evolution, and the social scientist might have reason to suspect that such conditions as the nature of the economy, level of industrialization, and the diffusion of communication technologies will influence the pattern. If some commentators are to be believed then the slightly less social lifestyles of the Americans in the survey is an accelerating trend to which other nations and cultures will also succumb, if they have not done so already (Locke, 1998). There is some evidence to support this pessimism, at least for the United States. Robin­son and Godbey (1997) report a 16% drop in time spent by Americans in social activities over the 20 years to 1985.

 

It does not follow of course that because or to the extent that people spend a lot of time in one another's presence they are therefore spending this time conversing. Nor does it follow that this is mainly what humans do with their language-using ability. They might, for example, spend more time using this ability to meditate, read, write or pray. But it is still a fair bet that conversation is a more natural and commonplace employment for this ability than the these other four candidates. Reading and writing are very recent innovations in the natural history of the species and the Szalai surveys confirm that they are spe­cialized and occasional activities; the average member of a highly literate society at the time of the surveys was spending little time reading - under one hour a day - and even less writing. Moreover, research in formalized bureaucratic organiza­tions (e.g. Davies, 1953; Mintzberg, 1973; Rogers, 1983) indicates that oral, face-­to-face communication has not been displaced by written communication. Lan­guage undoubtedly does play an important part in human thinking (d. Macphail, 1987), and the psychological study of language has placed as much emphasis on its role as a tool of thought as it has upon its manifestation as an aspect of social interaction. But it is not clear how much time is spent in silent reflections or solitary mental calculations associated with language.

 

If we allow that spoken interaction is a significant form of human social behaviour, just how much of it is then:? First it needs to be recognized that people may more often be in the role of listeners than speakers in such interac­tions. Even if participation is equally distributed it is necessarily the case that in a group of more than two each participant will spend more time listening than talking. We therefore need some indication of the sizes of the groups in which individuals are participants, as well as of their formality and focus. Is it for example the case that "in the company of others" most often means in a work group with no words exchanged beyond those needed for the task at hand, or common-interest groups pursuing their joint interests - amateur dramatics, foot­ball games, white-water rafting - or in committees with formal agendas and formalized contributions, or in large meetings addressed by "platform speakers"? It may then be that "the company of others" seldom means or allows informal chat.

 

Some indication of how much time people devote to conversations has emer­ged from the use of event recording methods. Wheeler and Nezlek (1977) asked a group of college students to keep a structured record of their own interactions over two separate two-week periods. Their data indicate that from 5 to 6V2 hours per day were spent by their sample in social interaction of some form. Seventy per cent of the recorded interaction time was classified as conversation and at least two of the other four categories of interaction involved conversational exchange. For various reasons, the Wheeler and Nezlek results probably under­estimate the frequency and cumulative duration of conversational interactions, at least for this population. Their method required participants to record only interactions of at least 10 minutes duration. Data we have collected from student samples (Emler, 1990; Emler & Grady, 1987; Emler & McNamara, 1996), also using event self-recording, indicate they engage in many conversations shorter than 10 minutes.

 

College students do, of course, form an unusual population and it would be unsafe to generalize from this source alone. We know, for example, that they 1ave more frequent conversational encounters and with more contacts than do similarly aged people who have full-time jobs, are in vocational training or are unemployed (Emler, 2000). College students also inhabit temporary social systems, have few significant responsibilities, and a lot of uncommitted time that could be devoted to socializing. It would be helpful to know more about the conversational activity of other significant groups, such as housewives, blue- and white-collar workers, workers in different occupational sectors, teenagers, re­tired people and unemployed adults.

 

The "event-contingent" self-recording methods employed in the studies described above provide information about the frequency of events but are imperfect estimates of the time involved. Better estimates of this are provided by signal-contingent procedures. In response to 42 signals transmitted in normal faking hours at random intervals over seven days, a sample of American teenagers recorded conversation as either the primary or a secondary activity on 41 % f occasions (Csikszentmihalyi, Larson & Prescott, 1977). This gives a figure of 6 ours a day involved in conversation. No other category of activity recorded, [eluding TV viewing, games and sports, eating, walking, reading and working, approached this level; the nearest was watching TV, recorded as the primary or  secondary activity on 12.5% of occasions.

 

Let us now return briefly to the question of conversational group size. The Wheeler and Nezlek (1977) study found that the majority of recorded interactions involved only two participants, just as we have found for conversational interactions (e.g., Emler & Grady, 1987). Observations reported by Dunbar, uncan and Nettle (1995) of informal conversations in a variety of semi-public settings - a college refectory at lunchtime, people waiting outside buildings during fire drills, participants at a reception - also indicate two people as the most common conversational group size (54% of those observed). To conclude therefore, conversations, most often between just two people, constitute a very common form of human social interaction, if not the most common. Moreover, more time may be devoted to this single activity than to any other except sleep.

 

Who Talks to Whom

 

As I have suggested elsewhere (Emler, 1994), much hangs on how this question is answered, nothing less than an entire model of social life. The social con­sequences of the Industrial Revolution was one of the great preoccupations of nineteenth-century social theorists and many of them were convinced that indus­trialization and the mass urbanization it promoted had initiated a fundamental change in the character of social life (e.g., Tonnies, 1887/1957). The contrast emphasized by Tonnies and later by Durkheim, Weber, Simmel, Parsons and Wirth among others was between the rural community and the city. It was assumed that in the former almost the only people individuals encountered in the course of their daily lives were kin or other acquaintances. In the latter, encounters were typically "impersonal, transitory and segmental" (Wirth, 1938). In other words, the inhabitants of pre-industrial villages interacted exclusively with people they knew whereas the inhabitants of modern cities would be likely to interact not as personal acquaintances but as the impersonal and frequently anonymous occupants of broad social categories or formally defined roles.

 

The implication of this "community lost" argument (Wellman, 1978) is that the conversations we have with others, unless these others are members of our immediate families, will occur outside any relationship of personal acquaintance. Our transactions with them will be predicated on our organizational roles and not on our respective personal identities or on any established relationship be­tween us as unique individuals. The further implication is that the substance of our talk will relate to the performance of these roles and their associated func­tions. As shop assistants, teachers, out-patients or building site foremen our talk will relate to the task of discussing a potential purchase, evaluating an essay, describing symptoms or directing work activities.

 

Although some social scientists argue the trend towards impersonal, deper­sonalized social relations has merely accelerated since the end of the nineteenth century (d. the earlier quotation from Ogburn & Nimkoff), others have argued equally strongly and with rather better evidence - namely detailed studies of social life in cities - that the "community lost" model both exaggerates and misrepresents social change in the last 200 years (e.g., Boissevain, 1974; Fischer, 1981; Gans, 1962; Litwak & Szelenyi, 1969; Mitchell, 1969; Wellman, 1978; Young & Willmott, 1957). They assert instead that even those who live in large cities continue "to dwell among friends" (Fischer, 1981), and that most of their interactions occur with people they know personally. Given potential concerns about the validity of self-report evidence, it is reassuring that these conclusions are also consistent with evidence using self-recording methods. For example, we have found over a variety of samples of young people - in higher education or vocational training, in employment or unemployed - that the majority of conver­sational encounters recorded were with people known to the young person, namely family members, friends or acquaintances (Emler & McNamara, 1996). Among the samples in higher education, 3% of all recorded encounters involved strangers and a similar percentage were classified as purely business or service; the averages in the former category were lower 'for the non-university samples. Data we have collected from people in middle-level management occupations using this method (Emler, 1990) only superficially suggest managers inhabit the impersonal, anonymous social world imagined by "community lost" theorists. Although they defined their relations with a majority of their contacts as "for­mal" these were encounters with individuals the managers knew well; on aver­age they had known each contact for two years.

 

One important finding to emerge from the event recording research concerns sex differences. No differences of this kind have yet been reported either in the frequency of conversational interactions or in the numbers of different people with whom such interactions occur. So, if women really do gossip much more than men, this will reflect the course their conversations take rather than the kinds or numbers of people they talk to or the numbers of conversational en­counters they have.

 

On the other hand, compared to whether one is male or female, differences in institutional status - whether employed or unemployed, in full-time education, or a member of a family household - have far more impact. On the earlier point concerning the likelihood of differences in communicative activity across social groups. our own research on the circumstances of conversational contacts (Emler & McNamara, 1996; Emler, 2000) indicates why group differences should be expected. We examined both where such contacts occurred and how they arose. The most important settings. in terms of numbers of conversational en­counters. were homes - one's own and other people's - places of education or training and places of work. In contrast, more public settings such as bars, shops, the street and public transport - the kinds of setting most often sampled in non­participant observational studies of conversation - accounted for a minority of these encounters. In other words, having a formal role in a setting and thus regular legitimate opportunities to be there enables informal social contact. This interpretation is reinforced by evidence on how contacts occur. People seem to rely little on arranging meetings in advance, rather more on knowing others' movements and thus where to find them when wanted, and to a similar degree on chance encounters with acquaintances, but most of all upon their own and oth­ers' routines, a strategy supported by the tendency for institutions, both formal and informal, to give a temporal structure to activities.

 

It is likely therefore that the nature of occupations will influence opportunities for informal conversational contact. Consider, for example, the occupations Mars (1981) called "donkey" jobs which commit people to a single setting for long periods, and which isolate them physically from others or symbolically through status differentials. But we should also expect an influence of the variety of institutional statuses a person has, and thus the settings with their associated routines to which they have legitimate access. We might therefore anticipate that housewives, retired people, the unemployed and the occupants of "total institu­tions" (d. Goffman, 1961) will have more limited ranges of conversational con­tacts than other social categories.

 

The Content of Conversational Interactions

 

If it is true that we not only spend a great deal of our waking time in face-to-face, one-to-one conversations but that the people we talk to are people we know personally and perhaps know very well, what are we talking about? Producing a clear answer to this question presents a number of technical difficulties. Conver­sation may be ubiquitous but it is not easily studied, at least not if one wishes to document with any precision what happens naturally and spontaneously.

 

Consider first direct observation. A recent example is provided by Bischoping (1993), who attempted to replicate the findings of much earlier studies by Moore (1922) on sex differences in conversational content. Moore had reported that male-­male conversations overheard in the street or public bars were mainly about busi­ness (50%) whereas the most common topic of female-female conversations was men, followed by clothes and other women. Bischoping found similar but smaller sex differences. Her study also confirms one of the problems of this method of direct observation: sample size. Hers included 27 females but only eight males.

 

Dunbar, Duncan and Marriott (1997) achieved a slightly larger sample of conversations with this method (n = 44), initially attempting to tape the observed conversations. Unobtrusive tape-recording in public and semi-public places pro­ved also to be poor-quality recording and was abandonned in favour of "direct auditory monitoring" (Dunbar et aI., 1997, p. 234). In other words, the observers listened, and made a judgment as to the topic, doing this at 30-second intervals, using 14 topic categories. Few significant differences emerged between the sexes. The most prominent of these was a tendency for males to devote more conversa­tion time to intellectual or work-related topics, particularly when females were present. The most striking features of their findings, however, were not these differences but the similarities. In terms of speaking time, the topic to which most attention was given was "personal relationships" in all but one of six groups of subjects observed. Across the six groups the percentage of speaking time for this topic varied from 15.3 to 49.5.

 

Some problems with purely observational data of this kind have already been noted, among which potentially the most serious concerns representativeness of the conversations that are available for such observation. Evidence based on event self-recording methods indicates that the majority of conversations do not take place in public settings and many occur in places where observers could not enter unnoticed (Emler & McNamara, 1996; Emler, 2000). Gossip, particularly if it involves betrayal of confidences or bad-mouthing acquaintances, both re­garded as violations of the informal rules of relationships (Argyle & Henderson, 1984), would surely flourish more readily in private places.

 

In an attempt to achieve a more representative sample of conversational content we adapted an event-contingent self-recording method (e.g., Emler, 1989). Trial studies with several versions indicated that (a) recorders have diffi­culty coping with more than a small number of simple categories, and (b) far too much information is exchanged in routine daily conversations for all of it to be recorded in this way with any degree of thoroughness over much more than a single day. However, these trials did indicate that, for a student population, personal topics (about the speakers or people known to them) were far more prevalent than impersonal topics. On the basis of these trials a larger study was undertaken to sample six topics of theoretical interest: own doings, others' doings, own emotional states, others' emotional states, practical information and politics. Participants recorded details of every conversation in which they parti­cipated over a seven-day period. However, they were asked to track only one of I these six topics and to note whether or not the topic had figured in each conver­sation. There were approximately 60 participants tracking each topic and for each topic a total of around 22,000 conversations were sampled.

 

Only two sex differences emerged; males were more likely than females to discuss politics though even in male-male exchanges this topic occurred on only 9% of occasions. The other difference was that females were more likely to discuss the feelings or emotional states of third parties. However, by far the most commonly occurring topic, in more than 40% of conversations, was "others' doings" and there was no difference between males and females in this respect. This contradicts Nevo et al.'s (1994) questionnaire-based evidence in which females report gossiping more (in so far as gossiping is the discussion of others' doings), and Levin and Arluke's (1985) observational data which similarly sug­gests this sex difference.

 

The picture of conversational content emerging thus far therefore is that people, or at least those people who are students, frequently discuss the activities of others they know and their relationships with others. It can still be argued that matters may be very different in more task-oriented settings, for example places of work, business, or training. What we know about the conversational topics of people who are not students remains perilously small as a basis for any strong conclusions.

 

SPECULATIONS ABOUT FUNCTION

 

An understanding of what motivates gossip - why people do it - needs to be consistent with various observations, particularly that we all seem to do it, that we do it so often and with so many different established partners, that it concerns people we know and that it is information-rich. For this last reason I suspect that the recently promoted hypothesis that the significance of gossip is to perform a social grooming function (cf. Dunbar, 1993, 1996) cannot be the whole story.

 

I propose the basic reasons why we gossip, why we so regularly exchange observations about other people, are quite straightforward. We are inhabitants of social as well as physical environments. Successful adaptation to the former kind has the same fundamental requirements as adaptation to the latter, namely the achievement of some degree of prediction and influence. If we are to predict the behaviour of our social environment we need to know things about its particular inhabitants, and not just about people in general or in the abstract. Specifically, we need to know what they are like - their personalities, character, abilities - and what their relations are with one another. Other social animals, notably the apes, do this by social observation. Research on attribution indicates how humans might use similar evidence, namely direct observations of others' actions and the effects of their actions, to make inferences about their attitudes, temperament and relationships. But the huge advantage conferred by language is that unlike other apes we are freed from exclusive dependence upon direct observation. Verbal exchange gives us rapid access to a larger sample of the relevant social information than we could ever achieve through our own direct observations. Equally as important as prediction is influence, and gossip is par excellence an instrument for subtle social influence. Gossip, I would submit, is therefore a fundamental tool of social adaptation.

 

If a goal of gossip is prediction then the quality of the data provided is an issue. We have already seen that gossip has been characterized both popularly and in some scientific treatments as unreliable (in the colloquial rather than the psycho­metric sense). Is this characterization justified? A priori, it would seem odd for people to devote so much time and effort to the collection of tainted evidence. In an interesting study, Wilson et al. (2000) show that when accuracy is important people do pay careful attention to the quality of information obtained through gossiping (see also Harrington & Bielby, 1995), while others have noted that spreading baseless gossip does rebound on the standing of the source (Schein, 1994). This is clearly a matter that deserves more study but it would be reason­able to expect that competent social actors would strive to make accurate judg­ments about their social environments. Psychometrics tells us the basis for valid measurement is aggregation, combining different observations. Thus a sensible strategy is to gossip with several sources about the same matters so as to com­pensate for their different biases (technically, to control for method variance). Is this what is reflected in high rates of conversational interaction involving gossip? This question brings us to ,two related and important issues: whether multiple sources could in principle be used in this way, and whether the validity of verbally transmitted social information is compromised by long communication chains.

 

By way of introduction, consider the more general goals of adaptation. Suc­cessful adaptation to a social environment involves securing its support for our own aspirations and welfare. Crudely, we want other people to aid us rather than hurt us. Prediction helps us to avoid the fools, scoundrels and carriers of other liabilities and gravitate towards the virtuous and talented. And even scoundrels will want to know the difference. Schumann and Laumann (1994) offer an inter­esting example of this process: choosing a sexual partner linked to one's social network allows more accurate assessment of the risk of contracting a sexually transmitted disease from that partner. Prediction, then, is important but we can do even better if we can persuade others to act in ways which benefit us.       

 

Biologists have pointed out that the possibility of being exploited or cheated is not just a threat to individual welfare but to the survival of social life altogether. All social species have somehow solved this "prisoners' dilemma": the pos­sibility that the individual member of the group could do better by exploiting the cooperation of others than by supporting it (Axelrod, 1984). However, the solu­tion which Axelrod and others have argued underpins social life, the so-called tit-for-tat behavioural strategy of instant retaliation against cheats, has many imperfections not the least of which is that, as Enquist and Leimar (1993) point out, in large social groups cheats can still do rather well while incurring costs for many of their fellows. But suppose that the normal mechanism of cheater detection, which basically involves learning by being the cheater's victim, was supplemented by gossip - all members of the group talking to each other about the conduct of every other group member. It turns out, at least in a computer simulation - now the preferred method of biologists studying behavioural strat­egies - that the cheat's effective reign is drastically reduced by gossip (Enquist &.Leimar, 1993). Nonetheless, this still seems to involve influence through im­proved prediction.        

 

A more direct influence mechanism, I would argue, depends upon the actor anticipating such predictions. If I know that you can learn about my treatment of others from others and if furthermore I hope for your cooperation, I would be wise to treat others well, and for similar reasons to treat you well. In other words", influence operates through the concern people have with their reputations and' gossip creates a pressure to keep those reputations honest. Gluckman's (1963, 1968) argument for the social control function of gossip has been strongly crit­icized (Paine, 1967; Bergmann, 1993), partly on the grounds that individuals gossip for their own benefit rather than the common good. But these criticisms lose some of their force if we see social control not as the reason for gossip but as a by-product of individuals pursuing their own interests, namely to be well informed about the conduct of others.

 

This also relates to the negative tone attributed to gossip by various writers (e.g., Rosnow & Fine, 1976) and indeed assumed in Gluckman's original argu­ments about its social control function: control is exercised by criticizing others' behaviour. Some researchers have even claimed to identify the predominance of this negative and critical quality in women's if not men's gossip (Eder & Enke, 1991; Leaper & Holliday, 1995). Dunbar et al. (1997), however, found that only a tiny proportion of the conversations observed involved malicious gossip or indeed negative comment of any kind. This may reflect the discretion of speakers in public settings, but the control effect does not require negative or evaluative comment, only factual observations about what a person has or has not done.

 

Enquist and Leimar's (1993) modelling of the effects of gossiping does assume that information about the same actor can be secured from more than one source. Though this seems a reasonable assumption to make about small-scale and relatively "closed" village communities, is it true of contemporary social conditions? To put it another way, to what extent do the people I talk to also talk to each other? Or are the links of personal acquaintance created through conver­sational contacts more accurately represented as chains which only ever intersect with each other at one individual? This question is defined by network analysts in terms of the density of an individual's network (cL Boissevain, 1974; Gra­novetter, 1973): what percentage of the links (of acquaintance, regular conversa­tional contact, etc.) that could exist among an individual's set of acquaintances do actually exist? Both practically and technically the question is difficult to answer with precision because, for example, there are several options for defi­ning both a link and the set to be analysed. Nonetheless, different methods have produced similar values for samples of adults; among an individual's circle of regular acquaintances, between 30% and 40% of the links that could exist do exist (e.g., Cubbitt, 1973; Emler, 1990; Friedkin, 1980). In other words, when any two acquaintances talk the chances are that they will have a large number of shared acquaintances about whom they could talk.

 

The possibility of this kind of triangulation - I can check your account of your dealings with a mutual friend against his or her account, and indeed against accounts provided by other mutual acquaintances - is, I think, the key to the prediction and influence/control functions of gossip (Emler, 1990, 1994). Cole­man's (1988) much discussed argument about social capital provides a comple­mentary analysis. According to Coleman, social capital is a product of a particular structure of social relations, specifically a structure in which A and B not only know one another but both also know C. These conditions, says Cole­man, promote the circulation of information, the effective enforcement of social norms, and the creation of trust. I would argue that the most basic of these three is the ready circulation of reliable social information; the other two are deriva­tive features (Emler, 2000; see also Burt & Knez, 1996).

 

If access to multiple sources of social information supports more accurate judgments about others, is accuracy nonetheless likely to be lost in long communication chains? There is little direct evidence to answer this, but strong reasons to suspect that most gossip chains are actually very short. In one of the studies by Wilson et al. (2000) subjects expected to give less weight to informa­tion about another that had been mediated by more than one link. On the basis of his studies of friendship networks, Boissevain (1974) argued that the effective extent of any person's influence seldom reaches beyond the friends of his or her friends, in other words beyond one intermediate link (see also Granovetter; 1973). Combining what we know about interaction rates and the density of social networks, individuals should seldom need to seek information mediated by more than one link from its origin and would anyway have difficulty coping with the volume if they did so on a regular basis; they would simply be exposed to too much information about too many people (Emler, 1990, in preparation).

 

Finally. the foregoing might seem to imply the happy conclusion that nice guys, do finish first (d. Dawkins, 1989), that the spoils of social life go to those who are cautiously if not enthusiastically virtuous. But this neither accords with the common experience that very unpleasant people can do quite well, nor does it recognize the power of gossip as a subtle instrument of social influence. Part of the reason why gossip provides imperfect protection against ambitious villains is to be found in the operation of top-down mechanisms for the allocation of organizational power. These mechanisms - of selection and promotion - seem to be systematically insensitive to moral flaws (d. Cook & Emler, 1999). This may in turn be because gossip, which potentially contains the basis for accurate character appraisals, naturally flows less easily along the vertical than the horizontal axes of social organizations, and so less readily reaches decision makers controlling top-down selection. But another part of the reason is surely that both the flow and the content of social information can and will be manipulated by those with sufficient guile, an ability that may be uncorrelated with moral virtue.  

By careful choice of what one says and does not say and to whom, one can simultaneously promote one's own cause and damage rivals without telling any lies. Gossip is undoubtedly a powerful instrument in the politics of everyday life (Bailey, 1971).

 

CONCLUSIONS

 

I have argued that gossip serves functions basic to social life and does so because it provides the inhabitants of human communities with invaluable information about their other members, incidentally a conclusion reached many years ago by the French psychologist Janet (1929). In emphasizing this quality of gossip I would not rule out other significant functions and effects. For example, gossip surely plays an important role in social comparison processes (d. Suls, 1977). A further use is to manage the boundaries of social groups (Gluckman, 1963; Elias & Scotson, 1965). But the choice of emphasis reflects my judgment that, just as with the enforcement of social norms and trust creation, so these other consequences derive from our dependence on gossip to provide us with social information.

 

A number of interesting questions have not been addressed here about which much more could be said - why gossip occurs primarily in face-to-face encoun­ters which are also informal, unscheduled, one-to-one and between acquain­tances, why an inclination to gossip is unlikely to be associated with intellectual superficiality, or why historically it has been associated with women in public consciousness despite so little evidence that this is the case (see Emler, 1994, in preparation, for further discussion of these issues). But to explore these in the current state of our knowledge would take us even further into the realms of speculation than we have already come. I hope to have made the case for taking gossip much more seriously - seriously enough to give it the research attention already directed to other topics on language, for example the closely linked topic of self-disclosure. In the earlier edition of this Handbook, gossip had no chapter. Should there be a subsequent edition I hope that much more of the speculation will have been replaced by hard evidence and thoroughly tested hypotheses, to the point that gossip will not appear as an application of language but one of its principle and most significant applications.

 

REFERENCES

 

·         Aeibischer, V. (1985). Les femmes et Ie langage: Representations sociales d'une difference. Paris: Presses Universitaires de France.

·         Allport, G. & Postman, L. (1947). The psychology of rumor. New York: Holt, Rinehart & Winston.

·         Argyle, M. & Henderson, M. (1984). The rules of friendship. Journal of Social and Personal Relationships, 1,211-237.

·         Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books.

·         Bailey F.G. (Ed.) (1971). Gifts and poison: The politics of reputation. Oxford: Blackwell. Bales, RF. (1958). Task roles and social roles in problem-solving groups. In E.E. Mac­coby, T.M. Newcomb & E.L. Hartley (Eds), Readings in social psychology (3rd edn). New York: Holt.

·         Barkow, J.B. (1992). Beneath new culture is old psychology: Gossip and social stratifica­tion. In J.H. Barkow, L. Cosmides & J. Tooby (Eds), The adapted mind: Evolutionary psychology and the generation of culture (pp. 627-637). New York: Oxford University       Press.

·         Bartlett, F. (1932). Remembering. Cambridge: Cambridge University Press.

·         Bergmann, J.R (1993). Discreet indiscretions: The social organization of gossip. New York: de Gruyter.

·         Bernard, H.R. & Kilworth, P.D. (1977). Informant accuracy in social network data II. Human Communication Research, 4, 3-18.

·         Bernard, H.R, Kilworth, P.D. & Sailer, L. (1982). Informant accuracy in social network data V. An experimental attempt to predict actual communication from recall data. Social Science Research, 11, 30-66.

·         Bischoping, K. (1993). Gender differences in conversational topics, 1922-1990. Sex Roles, 28, 1-17.

·         Boehm, C. (1993). Egalitarian society and reverse dominance hierarchy. Current An­thropology, 34, 227-254.

·         Boissevain, J. (1974). Friends of friends: Networks, manipulators and coalitions. Oxford: Blackwell.

·         Burt, RS. & Knez, M. (1996). Trust and third party gossip. In RM. Kramer & T.R Tyler (Eds), Trust in organisations: Frontiers of theory and research (pp. 68-89). Thousand Oaks, CA: Sage.

·         Buss, D.M. & Dedden, L. (1990). Derogation of competitors. Journal of Social and Personal Relationships, 7, 395-422.

·         Coleman, J.S. (1988). Social capital in the creation of human capital. American Journal of Sociology, 94 (Supplement), 95-120.

·         Cook, T. & Emler, N. (1999). Bottom-up versus top-down evaluations of managerial potential: An experimental study. Journal of Occupational and Organizational Psychol­ogy, 72, 423-439.

·         Cubbitt, T. (1973). Network density among urban families. In J. Boissevain & J.e. Mitchell (Eds), Network analysis: Studies in human interaction. The Hague: Mouton.

·         Csikszentmihalyi, M., Larson R & Prescott, S. (1977). The ecology of adolescent activity and experience. Journal of Youth and Adolescence, 6, 281-294.

·         Davies, K. (1953). Management communication and the grapevine. Harvard Business Review, 31, 43-49.

·         Dawkins, R. (1989). The selfish gene (2nd edn). Oxford: Oxford University Press.        

·         Dunbar, R.l.M. (1993). Coevolution of neocortical size, group size and language in hu­mans. Behavioral and Brain Sciences, 16, 681-735.

·         Dunbar, R.I.M. (1996). Grooming, gossip and the evolution of language. London: Faber &.Faber.

·         Dunbar, R.I.M., Duncan, N.D.e. & Nettle, D. (1995). Size and structure of freely forming conversational groups. Human Nature, 6, 67-78.

·         Dunbar, R.I.M., Duncan, ND.e. & Marriott, A (1997). Human conversational be­haviour. Human Nature, 8, 231-246. Eder, D. & Enke, J.L (1991). The structure of gossip: Opportunities and constraints on collective expression among adolescents. American Sociological Review, 56, 494-508.

·         Elias, N. & Scotson, J.L (1965). The established and the outsiders. London: Frank Casso Emler, N. (1989). Social information exchange. Paper presented at EAESP East-West Meeting, Jablonna, Poland.

·         Emler, N. (1990). A social psychology ofreputation. In W. Stroebe & M. Hewstone (Eds) European review of social psychology (Vol. 1, pp. 171-193). Chichester: Wiley.

·         Emler, N. (1994). Gossip, reputation and social adaptation. In RF. Goodman & A Ben­ Ze'ev (Eds.). Good gossip (pp. 117-138). Lawrence, KS: University Press of Kansas. Ii

·         Emler, N. (2000). Social structures and individual lives: Effects of participation in the social institutions of family, education and work. In J. Bynner & RK. Silbereisen (Eds), I Adversity and challenge in life course in the new Germany and in England (pp. 62-84).  London: Macmillan.

·         Emler, N. (In preparation) Serpent's tongue: The psychology of gossip. Oxford: Oxford University Press.

·         Emler, N. & Grady, K. (1987). The university as a social environment. Paper presented at British Psychological Society, Social Psychology Section Annual Conference, Brighton.

·         Emler, N. & McNamara, S. (1996). The social contact patterns of young people: Effects of participation in the social institutions of family, education and work. In H. Helve & J. Bynner (Eds), Youth and life management: Research perspectives. Helsinki: Helsinki University Press.

·         Enquist, M. & Leimar, O. (1993). The evolution of cooperation in mobile organisms. Animal Behaviour, 45, 747-757.

·         Fischer, C. (1981). To dwell among friends: Personal netWorks in town and city. Chicago: University of Chicago Press.

·         Freeman, Le., Romney, AK. & Freeman, S.e. (1987). Cognitive structure and informant  accuracy. American Anthropologist, 89, 310-325.

·         Friedkin, N.E. (1980). A test of the structural features of Granovetter's "Strength of  Weak Ties" theory. Social Neworks, 2, 411-422.

·         Gambetta, D. (1994). Godfather's gossip. Archives Europeennes de Sociologie, 35, 199­

·         Gans, R.J. (1962). The urban village. New York: Free Press.

·         Gluckman, M. (1963). Gossip and scandal. Current Anthropology, 6, 281-293.           

·         Gluckman, M. (1968). Psychological, sociological and anthropological explanations of witchcraft and gossip: A clarification. Man, 3, 20-34.

·         Goffman, E. (1961). Asylums: Essays on the social situation of mental patients and other inmates. Chicago: Aldine.       

·         Granovetter, M. (1973). The strength of weak ties. American Journal of Sociology, 78, 1360-1380.

·         Harrington, C.L. & Bielby, D.D. (1995). Where did you hear that? Technology and the social organisation of gossip. Sociological Quarterly, 36, 607-628.

·         Haviland, J.B. (1977). Gossip as competition in Zinacantan. Journal of Communication, 27,186-191.

·         Hutson, S. (1971). Social ranking in a French Alpine community. In F.G. Bailey (Ed.), Gifts and poison: The politics of reputation (pp. 41-68). Oxford: Blackwell.

·         Jaeger, ME., Skelder, A. A. , Rind, B. & Rosnow, R.L. (1994). Gossip, gossipers, gos­sipees. In RF. Goodman & A. Ben-Ze'ev (Eds), Good gossip (pp. 154-168). Lawrence, KS: University Press of Kansas.

·         Jaeger, M.E., Skelder, A.A. & Rosnow, RL. (1997). Who's up on the low down: Gossip in interpersonal relations. In B.H. Spitzberger & W.R. Cupach (Eds), The dark side of close relationships (pp. 103-117). Mahwah, NJ: Erlbaum. Janet, P. (1929). L'evolution psychologique de la personalite. Paris: Chahine. Jones, D. (1980). Gossip: Notes on women's oral culture. Women's Studies International Quarterly, 3, 193-198.

·         Kashy, D. & Kenny, D. (1990). Do you know whom you were with a week ago Friday? A reanalysis of the Bernard, Kilworth and Sailer studies. Social Psychology Quarterly, 53, 55-61.

·         Killworth, P.D. & Bernard, H.R (1976). Informant accuracy in social network data. Human Organization, 35, 269-296. Kipers, P.S. (1987). Gender and topic. Language in Society, 16, 543-557. Landis, M.H. & Burtt, H.E. (1924). A study of conversations. Journal of Comparative Psychology, 4, 81-89.

·         Leaper, e. & Holliday, H. (1995). Gossip in same gender and cross-gender friends' con­versations. Personal Relationships, 2, 237-246.

·         Levin, J. & Arluke, A. (1985). An exploratory analysis of sex differences in gossip. Sex Roles, 12, 281-286.

·         Litwak, E. & Szelenyi, 1. (1969). Primary group structures and their functions. American Sociological Review, 35, 465-481.

·         Locke, J.L. (1998). The de-voicing of society: Why we don't talk to each other anymore. New York: Simon & Schuster.

·         Macphail, E.M. (1987). The comparative psychology of intelligence. Behavioral and Brain Sciences, 10, 645-696.

·         Malinowksi, B. (1922). Argonauts of the Western Pacific. New York: Dutton.

·         Mars, G. (1981). Cheats at work: An anthropology of work-place crime. London: Unwin. Merry, S.E. (1984). Rethinking gossip and scandal. In D. Black (Ed.), Toward a general theory of social control: Fundamentals (pp. 271-302). Orlando, FL: Academic Press.

·         Mintzberg, , H. (1973). The nature of managerial work. Englewood Cliffs, NJ: Prentice­ Hall.

·         Mitchell, J.e. (1969). Social networks in urban situations. Mancheser: Manchester Univer­sity Press.

·         Moore, H.T. (1922). Further data concerning sex differences. Journal of Abnormal and Social Psychology, 4, 81-89.

·         Naish, J. (1978). Desirade: A negative case. In P. Caplan & J.M. Burja (Eds), Women united, women divided: Cross cultural perspectives on female solidarity (pp. 238-258). London: Tavistock.

·         Nevo, 0., Nevo, B. & Zahavi, A.D. (1993a). The development of the tendency to gossip questionnaire: Construct and concurrent validation for a sample of Israeli college stu­dents. Education and Psychological Measurement, 53, 973-981.

·         Nevo, 0., Nevo, B. & Derech-Zehavi, A. (1993b). Gossip and counselling: The tendency to gossip and its relation to vocational interests. Counselling Psychology Quarterly, 6, 229-238.

·         Nevo, 0., Nevo, B. & Derech-Zehavi, A. (1994). The tendency to gossip as a psychologi­cal disposition: Constructing a measure and validating it. In R.F. Goodman & A. Ben Ze'ev (Eds), Good gossip (pp. 180-192). Lawrence, KS: University of Kansas Press.

·         Oakley, A. (1972). Sex, gender and society. London: Temple Smith.

·         Ogburn, W.F. & Nimkoff, M.F. (1950). Sociology. Boston: Houghon Mifflin.

·         Rogers, E.M. (1983). Diffusion of innovations (3rd edn). New York: Free Press.

·         Ross, E.A. (1901). Social control: A survey of the foundations of order. New York: Macmillan.  

·         Paine, R (1967). What is gossip about? An alternative hypothesis. Man, 2, 278-285.

·         Reis, H.T. & Wheeler, L. (1991). Studying social interaction with the RochesterllInterac­tion Record. Advances in Experimental Social Psychology, 24, 269-318.           

·         Robinson, J.P. & Godbey, G. (1997). Time for life: The surprising ways Amerfcans use their time. University Park, FA: Pennsylvania State University Press.        

·         Sabini, J. & Silver, M. (1982). Moralities of everyday life. Oxford: Oxford Univers ty Press. Schein, S. (1994). Used and abused: Gossip in medieval society. In RF. Goodmfln & A. Ben Ze'ev (Eds), Good gossip (pp. 139-153). Lawrence, KS: University o  Kansas Press. Scheler, M. (1921). Ressentiment. New York: Free Press.

·         Schumann, P.L. & Laumann, E.O. (1994). The importance of social capital for sexual relations. Paper presented at International Sociological Association meeting.

·         Spacks, P.M. (1985). Gossip. New York: Knopf. Stone, A.A., Kessler, Rc. & Haythornthwaite, J.A. (1991). Measuring daily life events and experiences: Decisions for the researcher. Journal of Personality, 59, 575- 7.

·         Suls, J. (1977). Gossip and social comparison. Journal of Communication, 26,16 168. Sultzberger, c.F. (1953). Why it is hard to keep secrets. Psychoanalysis, 2, 37-43.

·         Szalai, A. (1972). The use of time: Daily activities of urban and suburban populations in twelve countries. The Hague: Mouton.     

·         Tannen, D. (1990). You just don't understand: Women and men in conversation. New York: Ballentine Books.

·         Tonnies, F. (1957). Community and society. New York: Harper (first German edition, 1887).

·         Weick, K. (1985). Systematic observational methods. In G. Lindzey & E. Aronsorj (Eds), The handbook of social psychology (3rd edn). Vol. 1, pp. 567-634. New York: R)andom House.

·         Wellman, B. (1978). The community question: The intimate networks of East Yorkers. American Journal of Sociology, 84,1201-1231.         

·         Wheeler, L. & Nezlek, J. (1977). Sex differences in social participation. Journal of Person­ality and Social Psychology, 35, 742-754.

·         Wilson, D.S., Wilzynski, c., Wells, A. & Weiser, L. (2000). Gossip and other components of language as group-level adaptations. In C. Heyes & L. Huber (Eds), The evolution of cognition (Vienna series in theoretical biology) (pp. 347-365). Cambridge, MA: MIT Press.

·         Wirth, L. (1938). Urbanism as a way of life. American Journal of Sociology, 44, 3-24.

·         Young, M. & Willmott, P. (1957). Family and kinship in East London. Harmondsworth: Penguin.