From: William A. Foley, 1997, Anthtropological Linguistics: An Introduction, Chapter 15, 286-306.

The sexual contrast in physique between male and female is an obvious one in all human societies; it is a biological given. (Or is it? For serious doubts over such a bald, perhaps Eurocentric statement, see Errington (1990), Foucault (1984), Moore (1988, 1994), and Yanagisako and Collier (1987); perhaps even the "biological given" of physical sex is a cultural construc-tion.) But what is made of this difference culturally, what significances are ascribed to it, is anything but an obvious given; indeed, if sex is a biological fact, then gender is a cultural construction. This contrast in sex typically provides a powerful basis for constructing opposing cultural categories of masculine versus feminine, but the content of these categories is not per-manently set, but rather is constructed in an ongoing fashion through the daily practices of social interaction. In line with the above discussion, the notions of gender daily inform cultural behavior in the understandings we bring to social relationships and in turn are constructed by our practices in these relationships; in short, our habitus is engendered.

Although the categories of gender are culturally constructed and hence their content is variable across cultures, one aspect of this opposition which does seem extremely widespread, if not universal (as claimed, for example, in Rosaldo ( 1974, 1980b) ) is the fact of the greater status or prestige granted to the masculine. Rosaldo (1974, 1980b) labels this as the universal fact of male dominance across cultures and describes it as follows (Rosaldo 1980b:394):

a collection of related facts which seem to argue that in all known human groups - and no matter the prerogatives that women may in fact enjoy - the vast majority of opportunities for public influence and prestige, the ability to forge relationships, determine enmities, speak up in public, use or forswear the use of force are all recognized as men's privilege and right.

Language and Gender 287

This thesis of the “universal asymmetry in cultural evaluation of the sexes” (Rosaldo 1974:17) has not gone unchallenged (see Errington and Gewertz 1987; Leacock 1978; Moore 1988; Sacks 1979; Sherzer 1987), but, regard- less of its universal status, such an asymmetry between the sexes is indeed very widespread. The question is: why should this be? Rosaldo (1974) argues that the basic delegation to women of childbearing and early rearing has resulted in a fundamental separation in all societies between a domestic, home sphere of influence and a public realm. Because of women's childbearing and rearing responsibilities, their roles are foregrounded in the domestic sphere of influence, but because men are to a large extent free of these responsibilities, especially in the early years of child rearing, they have a much wider scope to engage in the public realm, engaging in social inter-actions that forge political alliances, economic ties, religious sects, etc. But why should this lead to such a widespread asymmetric evaluation of these two realms, with those of women typically viewed as inferior, and why would this be found in otherwise egalitarian hunter-gatherer societies? It is well known that in most such societies it is the women, constrained by child rearing responsibilities, that provide the great bulk of food through gather- ing. But the men, who are free of these responsibilities, are able to engage in the potentially more risky activity of hunting, which in turn results in the more higher valued food, meat. Young men typically use this valued food, as well as other labor, to solicit the favor of older men in competing for their daughters as wives or lovers. The younger men's labor and especially the meat they provide is a medium of exchange in marital politics, but the veget- able food gathered by women, in spite of it usually providing the bulk of the

diet, lacks such a high valuation and plays no role in such politics. Women, then, are effectively excluded from the politics of marital exchanges, and, because these typically form the basis of wider political and economic rela-

tions in these small-scale societies, from the public realm generally. And it is through these relations in the public realm that the power to control and distribute resources in the society is determined; hence the positive evalua-

tion of the public realm over the domestic (although a question which really

needs to be asked here is whether the judgment is an effect of our own local Western ideology; is the public realm really evaluated as being more pres-

tigious than the domestic in all cultures?) .

Chodorow (1974) adds another important plank to this argument about the universal asymmetry of the sexes through a developmental perspective. As noted above it is women in their domestic sphere of responsibilities who raise children of both sexes. A girl has a clear role model present to follow;

femininity seems to be acquired through this model in a relatively easy and straightforward manner. The domestic sphere in which she is reared pro-vides the little girl with a clear model of what her lifetime responsibilities

and privileges will be like. Not so for the little boy. He must learn to be a

288 The Ethnography of Speaking

man. The characteristic public male activities of hunting, political debate, farming, etc. are largely unavailable to him as he grows up in the household. He finds he must sever these household ties and establish his maleness as separate from them. Hence, little boys seek play outside the household with other boys and jostle for position within these groups in order to estab- lish their identity. Because characteristic adult male activities are largely unavailable, boys typically know of masculinity only as abstract rights and duties and therefore see their identity in terms of the formal roles that these entail. Male childhood play typically concerns games in which boys com-pete for the prestige associated with these imagined formal roles (i.e. com-petition who will be the doctor and who will be the patient in familiar childhood games). The developmental differences discussed by Chodorow (1974) reinforce the sociological ones highlighted by Rosaldo ( 1974,1980b): women are from birth channelled to a domestic sphere of influence, but men are almost compelled to jockey for prestige in the political games of the public realm, and consequently, power.

Rosaldo and Chodorow's interpretation are not supported by all scholars in this field. Some (e.g. Leacock 1978; Sherzer 1987) wish to argue that in some cultures the categories of masculine and feminine are complement- ary: separate, but equal. Errington and Gewertz (1987) is an especially soph-isticated statement of this view. They argue that in societies like our own modern industrialized societies, in which immersion in the domestic sphere of influence entails exclusion from the public domain of paid work, separate can never mean equal. For in such societies it is work which confers status, prestige and interest, and, most importantly in these highly individualized societies, money, which confers independence and the power to appropriate resources. To the extent that commitment to the domestic sphere involves limited access to this highly valued public realm of work, women are both separate and unequal. However, Errington and Gewertz argue that this does not apply to the Chambri, a sedentary hunter-gatherer group of New Guinea. In this sociocentric culture, in which a person is defined by her social and kin position in the village, one's value is not determined individually, through a job or the patterns of personal consumption it permits. Rather, one's worth is directly linked to the network of one's social connections, in Chambri terms, largely the kinship categories one has with others. The positional identities of Chambri men and women within the society are completely different and not in competition. The strategies that Chambri men and women use to validate their sense of worth in their social position, generally through manipulation of kinship links, are also exclusive of each other: women through having babies and thus forging new kinship links and men in patriclans competing politically over power, secular and spiritual. Men do not dominate women in Chambri society, according to Errington and Gewertz, because they do not deprive them of the ability to make and enact

Language and Gender 289

decisions validating their socially defined self worth. Chambri men and women are by and large able to allow each other to pursue their separate strategies to validate self-worth; separate, but equal.

Whether this interpretation of the Chambri case is in fact warranted is at least debatable. Scholars sympathetic to Rosaldo and Chodorow's position would point to the fact of women's exclusion from most ritual life, sacred lore, and political debating as evidence of their secondary status within Chambri society, although the degree of economic independence enjoyed by Chambri women is certainly remarkable when compared with that of their Western sisters. Perhaps a safe conclusion to be drawn from this is that while male public activities do attract a more prestigious evaluation in all societies, this does not necessarily entail a devaluation of women's activities.

The ideology of gender categories is typically enacted in linguistic practices; indeed, it is through language that the individual cultural understandings of gender categories are learned (see chapter 17 and Ochs 1990) and the coordination of gender roles achieved. Commonly, highly valued styles of speaking are associated with men's activities in the public realm, and this skill is ideologically denied to women. A good example of this is Malagasy speech norms. As mentioned earlier in chapter 14, the Malagasy value in-directness in their speech, and this norm is most clearly articulated in the public ceremonial kabary speech, which only men are expected to cultivate. Kabary are ritualized dialogs between two men, which frequently involve mutual criticism, but, as direct confrontation is strongly negatively valued in this culture, the criticism is couched in very indirect terms, using prov-erbs, allusions, and innuendo. Further, before criticism begins, the speaker typically comments positively on his rival's talk, setting the stage for his own comments, as in this example (Keenan 1974:129)

Thank you very much, sir. The first part of your talk has already been received in peace and happiness. I am in accordance and agreement with you on this, sir. You were given permission to speak and what you said gave me courage and strength. You said things skillfully but not pretentiously. You originate words but also recognize what is traditional. But as for myself I am not an originator of words at all but a borrower. I am more comfortable carrying the spade and basket. You, on the other hand, have smoothed out all faults in the speech; you have woven the holes together. You have shown respect to the elders and respect to the young as well. This is finished. But . . . [criticism begins].

290 The Ethnography of Speaking

Kabary speech is associated most prominently with public male activities and is very highly valued culturally, conferring high status on those who use it with great skill.

Malagasy women are not permitted to engage in kabary displays. They are thus barred from garnering prestige through skillful use of this highly valued speech style. Further, women are expected, indeed, in some circum-stances, encouraged to violate the norm of indirectness, which is generally the hallmark of high status speech. Thus, their direct style of speaking identifies Malagasy women as being of secondary status and in fact is partly what constitutes them as such. Women perform all sorts of verbal activities which Malagasy cultural ideology negatively evaluate: berating others in public, expressing anger, arguing, haggling at the market. Men often find this directness a useful trait to exploit; when a man experiences a public injury to himself or his property, it will likely be his wife who makes accusations for him in public. Because of their latitude in speaking directly, it is they who sell produce in the market and buy necessities there. Fero- cious haggling is often required in this situation, one for which the female norms of directness are much more suited than the male ones of indirect- ness. In the markets, men typically sell things that have fixed prices, like meat, abrogating the need for haggling. It is important to note that regard- less of the useful ends that women's norms for speaking serve, their direct style of speech is devalued by both sexes. Thus, the respective linguistic behaviors of the two sexes is assigned differential prestige: the indirect male style, prototypically exemplified by the kabary language of male public encounters, is status bearing, but the everyday direct style (should we call it domestic language?) is not. Through these linguistic differences and ascribed status, cultural concepts of gender are constructed and transmitted.

For a related but somewhat different case, consider gender differences in speech in Javanese (Smith-Hefner 1988). Javanese is a language with an extremely developed system of speech styles reflecting politeness and relative status of speaker and addressee, as well as that of the topic under discussion (more fully discussed in chapter 16). As expected, Javanese society is highly stratified along caste or class lines, but status distinctions along gender lines are not strongly in evidence. In fact, ethnographic studies emphasize the relatively high status of women in Javanese society (Keeler 1990). Javanese women are free to work in the rice fields, markets, or business. They do the shopping and bargaining and generally deal with money matters. They interact freely with men, have wide economic independence and participate extensively in religious, political, and social life. Since Indonesian independ-ence in 1949, boys and girls have been equally educated.

When we turn to linguistic practices, however, significant differences emerge between men and women's speech. Both sexes use the system of speech styles for politeness, but not in identical fashions. Within the family,

Language and Gender 291

wives typically use a more polite form of speech to their husbands than their husbands do to them, reflecting and declaring a somewhat higher status of the man. For example, she may call him mas “elder brother”, but he will call her by her first name, a nickname or dhik “younger sibling”, this difference in seniority emblematic of a difference in status.

Outside of the domestic sphere, there are even more significant differ-ences in linguistic usage between the sexes. While both may use the highly deferential polite speech styles, men use them with greater art and to more effect. In a pattern highly reminiscent of the previous Malagasy case, to which, of course, Javanese culture is historically related, Javanese women have a local reputation of being more talkative than men and doing so with less skill. They are said to speak off the top of their heads without reflection; this often results in a less appropriate or even incorrect choice of linguistic form. Typically, this involves using a polite form which is too polite, lead- ing to embarrassment on the part of the addressee, as it exalts them bey- ond what they feel is proper (humility, as we shall see, is positively valued in Javanese culture). Javanese men, again in common with their Malagasy brothers, are much more circumspect and controlled in their use of speech. They are more taciturn than women, and this increases with age. They are highly preoccupied with status distinctions and sensitively attuned to the proper use of polite speech styles. The artistic use of polite speech forms signals one's refinement and status. The proper use of these styles requires a fine attunement to the relative status of oneself and one's addressee. As humility is positively evaluated culturally, so actually higher status men may creatively use polite forms to their lower status addressees, but in so doing, subtly convey their great skill in the use of the extremely complex system of these forms and, hence, their superiority. He may express defer-ence, but actually hint at social superiority. As Javanese grow older, they may become increasingly engaged in these subtle games of linguistic eti-quette to enhance their status.

The linguistic differences in the use of polite speech forms of Javanese men and women reflect their differential engagement in the domestic and public realms of life. While possessing a strong local ideology of relative equality between the sexes, Javanese society is nonetheless highly stratified. This stratification is most prominently played out in the public political realm, and men excel in this realm through the artful use of polite speech form to enhance their status. Women, on the other hand, are focused on the home and the economic chores associated with its well being. The use of polite forms intrude a bit on this sphere, but not in any way parallel to their importance in the public male status arena. This is also reflected in women's less overall skill in using the polite speech styles. Again, the different cul-tural understandings of the two gender categories is articulated in and con-structed through language.

292 The Ethnography of Speaking

A final case is provided by the Kuna, a group of Native Amerindians in Panama. Sherzer (1987) describes Kuna society as basically egalitarian, both in its sociopolitical structure and its gender relations. Positions of leadership may, theoretically at least, be occupied by either men or women. The Kuna view the gender categories of masculine and feminine as complementary; separate, but equal. Men hunt and farm; women perform domestic chores. Economically, women have significant independence in that they produced the appliqued cloth blouses for which the Kuna are noted and sell these in markets as a source of cash for themselves and their families.

This complementarity in sociopolitical roles is carried over, according to Sherzer, into the types of linguistic behavior associated with each sex. Each has a characteristic set of speech genres, associated with both public and domestic contexts. Each genre has specific linguistic features. Public speech genres for men include political oratory and ritual incantations, especially curing chants. Political meetings commonly consist of debates among vil-lage leaders and ritualized chanting by chiefs; these speech genres are per-formed by men, because it is men who typically fill these roles. Public speech genres for women also include political speeches and debates, this time in the cooperatives for the marketing of their appliqued cloth blouses, and ritual chants. But the most important speech genres for women are those of the domestic sphere: lullabies and ritual wailing for dying and deceased relatives. Very significantly, however, Sherzer reports no speech genres distinctive to men for the domestic sphere, a fact no doubt which must buttress Rosaldo's (1974) claims. Kuna do not positively evaluate the speech of either men or women more highly; skilled speakers of either sex are equally lauded. Further, both men and women are equally loquacious and state their views directly and confidently. Thus, according to Sherzer's description, the egalitarian, but complementary ethos of gender roles in Kuna society generally finds expression in linguistic behavior. Styles appro-priate to the public political domain of men are equally available to women, although the genres might differ (note that, however, this does not apply the other way around: the domestic styles of women are not taken up by men). As men and women perform complementary economic and social roles in Kuna society, their characteristic speech genres are also complementary: separate, but equal.

What do the patterns of language usage and gender in these three societ-ies have to tell us about Rosaldo's (1974, 1980b) claim about the universal fact of male dominance, “the universal asymmetry of cultural evaluation of the sexes” in favor of the male and his public realm over the female and her domestic sphere. Malagasy and Kuna society are basically egalitarian, while Javanese is highly stratified: Malagasy and Javanese societies have highly salient differences in linguistic behavior between the sexes, while Kuna does not. But, nonetheless, the evidence from all three societies suggests that

Language and Gender 293

male dominance is expressed and constructed through language. Eckert and McConnell-Ginet (1992:483) see dominance as being 都ustained by privileg-ing in community practice a particular perspective on language, obscuring its status as one among many perspectives, and naturalizing it as ‘neutral’ or ‘unmarked’._ How language sustains male dominance in both Malagasy and Javanese society is clear, for the indirect norms of Malagasy males are canonized as 菟roperly_ speaking Malagasy, rather than just one way, and the obsessive concern of Javanese males with the artful use of polite speech to enhance status becomes the avowed aim of all in speaking Javanese “well” (Smith-Hefner 1988). The Kuna case is more subtle and potentially mis-leading. Note that men and women equally participate in characteristic speech genres for their sex in the public realm, but only women have distinct- ive domestic speech genres. The public realm is ungendered, available to persons of either sex, but the domestic sphere is stamped linguistically to be engendered as feminine, the domain of women. Masculine is, thus, the unmarked gender; feminine, the marked. The masculine is the unmarked perspective and thereby a privileged one, while the feminine with its spe- cific domestic responsibilities (and speech genres) in Kuna society is a less privileged one. Thus, while Kuna society is much more egalitarian in its engendered speech norms than Malagasy or Javanese, there is nonetheless a weak form of male dominance.

These case studies now lead us to ask how is male dominance in the norms for speaking reproduced in modern Western societies? A number of stud- ies of cross-sex conversational interactions in American English indicate that generally men do indeed assume a more dominant role; in McConnell-Ginet's (1988:98) words, conversation turns out not to be “an equal-oppor-tunity activity”. For example, the usual stereotype is that women talk more than men in this culture, but actual quantitative studies show the opposite to be the case. Swacker (1975) asked cross-sex dyads of college students to describe some pictures as thoroughly as they liked with no time constraints on their statements. The mean time in this task for males was 12.0 minutes, while it was only 3.17 for females (the males out-talked the females by a ratio of 4 to 1!). Edelsky (1981) observed conversations in university faculty meetings and again discovered that the speaking turns of the men ranged between 25 percent and 400 percent longer than those of the women. Fur-ther, the more formal the topic, the greater the difference.

Interruption between interlocutors in cross-sex conversation is one area that has attracted a great deal of study. Zimmerman and West (1975) recorded cross-sex conversations in different public places between acquaintances.

294 The Ethnography of Speaking

They tallied the number of interruptions, which they defined as “an incursion into the current speaker's talk prior to the last lexical constituent that could define a possible terminal boundary” (Zimmerman and West 1975:114), contrasting this crucially with overlaps, “simultaneous speech where a speaker other than the current speaker begins to speak at or very close to a possible transition place in a current speaker's utterance (i.e. within the boundaries of the last word)” (Zimmerman and West 1975:114). Zimmerman and West viewed overlaps as mistakes in timing to the next speaker's turn, but inter-ruptions as deliberate challenges to the speaker's right to hold the floor. They found a strong difference in patterns of interruptions and overlaps in the same-sex versus cross-sex dyads. In same-sex conversations, overlaps and interruptions were about evenly distributed between the two speakers, regardless of the sex of the dyad, but in cross-sex dyads men produced 96 percent of the interruptions to the women's 4 percent and also 100 percent of the overlaps! Further in same-sex dyads most of the breaks into talk were overlaps (22 overlaps as opposed to 7 interruptions), while in cross-sex dyads they were interruptions (48 interruptions as opposed to 7 overlaps), with all but 2 of these interruptions coming from men. Zimmerman and West's work demonstrates a clear asymmetry in speakers' rights to hold the floor in conversations; men are far more likely to usurp women's rights than the other way around. By the sheer control they exercise in the develop- ment of a conversation, male domination must be seen as basic to conver-sational norms in American society.

Most significantly, the greater conversational power of males emerges even in contexts in which the female conversation partner has stronger claims to higher status and power in a wider domain, demonstrating that this is indeed an asymmetry in engendered norms for .speaking, inculcated tacitly in the habitus. Woods (1988) recorded triadic conversations in work settings, involving both male and female supervisors and subordinates. She found that gender was the most significant factor determining speech behavior, more important than status. With respect to patterns of interrup-tion, higher status people were more likely to interrupt successfully (i.e. gain the floor, switch the topic, etc.) than lower status conversational part-ners, but men of lower status were still usually successful in interrupting a higher status female speaker. For example, a subordinate male was success-ful in interrupting nine times and only unsuccessful once, while his female supervisor successfully interrupted six times and was unsuccessful three times. When status and masculine gender go together, the dominance of men in these conversations is pronounced indeed: a male supervisor inter-rupted successfully four times and never unsuccessfully, while his female subordinate interrupted successfully once and unsuccessfully four times. Not surprisingly, a male subordinate to this woman still fared better than she did in this interaction; he interrupted successfully four times and only

Language and Gender 295

twice unsuccessfully. Woods' data again demonstrate male dominance in conversational interaction: men hold and snatch the floor more successfully than women, regardless of their relative status.

The conclusion of this work on gender asymmetries in the control of conversational interactions through interruptions has recently been chal-lenged (James and Clarke 1993; James and Drakich 1993; Tannen 1989, 1990). One argument assumes the "two cultures" model of engendered speech styles to be discussed below, but the basic point is that it is difficult to distinguish true interruptions, bidding for the floor, from statements trying to support the speaker's turn and her contribution to the conversa- tion. In other words, what counts as an interruption is really a matter of an analyst's interpretation of the interactions, requiring a great deal of back-ground knowledge about the interlocutors. Given this background, could we view the interruption as a reinforcement of what is being said or as a bid for control of the floor? Such questions also imply there may be great difficulty in imposing a sharp methodological distinction in the functions of overlaps and interruptions, especially across different types of conversa-tional contexts. For example, Edelsky (1981), the study of speech in univer-sity faculty meetings discussed earlier, found that if the one speaker at a time rule, which usually applies in formal meetings, was relaxed, women spoke just as much as men, using overlapping to reinforce the contributions of others. Finally, the view of interruptions as a bid to control the floor may not be a universally correct interpretation of its function, even in white middle-class American society. Tannen (1990) argues that speakers from different sub-cultures may have "high-involvement" styles of speaking, which display short pauses, fast pacing, and what are read as interrup- tions, but these are often used to reinforce the speaker's point, rather than contradict him or snatch the floor. Interruptions in such sub-cultures are not necessarily claims to dominance, but rather assertions of interest and solidarity.

Indeed, quite the opposite, the refusal to speak very much at all, can often be an effective claim to dominance. Pamela Fishman (1983) tape-recorded many hours of conversations between the husband and wife of three white middle-class graduate student couples. In interpreting her material she hit upon the idea of conversational work: a conversation is produced by the active participation of at least two participants, paying attention and responding to the remarks of the other. The conversation is successfully maintained as long as this goes on; failure to do so subverts the conversation. Fishman's work revealed a number of strategies to initiate, maintain, or subvert con-versation. Maintenance strategies include: (1) attention getters, like this is interesting, which attempt to legitimize interest in the following topic; (2) questions, which set up a required answer, thereby insuring some response from their interlocutor; and (3) ritualized rhetorical questions, like do you

296 The Ethnography of Speaking

know what?, which requires a further question mhat? in response, and then an answer, setting in train the development of a topic. Fishman found that all three strategies were used overwhelmingly more frequently by women than men: attention getters twice as often, asking questions two-and-a-half times more frequently, and ritualized rhetorical questions again twice as often. Fishman claimed that this asymmetry demonstrates that women shoulder much more of the burden than men in conversational work. Indeed, she argues further that women actually aid men in developing their topics in conversation, especially through the strategy of asking questions, but also through generally encouraging responses. Men, by contrast, often behave in quite the opposite fashion, commonly providing minimal responses like hmm or yeah following the woman's speaking turn. These act to discourage the woman in her development of the topic, displaying little interest and no explicit picking up of the woman's topic, so that her contribution to the conversation quickly fades out. In this case male dominance of the con-versation, in determining where it will and will not go, is realized through asymmetrical speech norms typified by the male's lack of speech, rather than his hogging the floor.

What could be the source of these strikingly asymmetrical norms for speak-ing between men and women? Remember the theories of Chodorow (1974) discussed above. Unlike girls, who have clear role models in their mother and the other women of the domestic sphere, boys have to learn how to be men. He must sever his ties with the feminine sphere of the home and establish his identity in the world outside, jostling with other boys for position and status to establish their claims of masculine identity. Male childhood play consists of games in which players compete for status- bearing roles. The play of girls, on the other hand, tends to model the co-operative connections among women in the household. Chodorow's ideas have found their way into work on language and gender in what is some-times called the “two cultures” model (Maltz and Borker 1982; Tannen 1990). This model argues that men and women inhabit two different cul- tural worlds as far as the understandings they bring to conversational inter-actions; men see conversations as status contests in which they can be one up or one down, while women see them as a means of forging interpersonal connections. These different views of the ends of conversation lead to com-municative misunderstandings not too dissimilar to those highlighted by Gumperz (1993) in his work on crosscultural communication, hence the name, “two cultures” model.

Maltz and Borker ( 1982) trace these differences to the styles boys and

Language and Gender 297

girls develop in childhood during play. Boys typically play in larger groups than girls and these groups are structured hierarchically according to status. Status is relative and ever changing so the main point of boys' play interac-tions is to manipulate their peers to enhance their status. In these compet- itive attempts to enhance status, speech has three roles: (1) to assert one's dominance, (2) to attract and hold an audience, and (3) to assert oneself when others have the floor. The styles of speech used will reflect these functions: orders, threats, and ridicule (asserting one's dominance), boast- ing or other types of display of verbal skills (attracting and holding an audience), and refusals to listen to others or interruptions of their turns (asserting oneself when others have the floor). The world of girls is utterly different. Their play groups are smaller, often just pairs, and their games are cooperative and organized in non-competitive ways. Differentiation among girls is not established in terms of status differentials of power, but in terms of relative closeness. Friendship is construed in terms of intimacy, commit-ment, and loyalty. The goal of interaction is to reinforce these feelings of connection. To put it baldly and oversimplistically in terms of Brown and Levinson's ( 1987) theory, boys are more focused on their negative face, in that status gives them independence and freedom to act, including coercing others to act as they want (i.e. power), while girls are more concerned with positive face, the positive evaluation they give each other through mutual closeness. Of course, conflict necessarily arises sometimes in all human groups, even those exclusively of girls. Boys typically resolve conflict through displays of dominance, often involving physical force, but in the more egalitarian cooperative groups of girls this option is not easily available. Girls must learn to use interactive strategies which attend to the positive face needs of the other, but at the same time effectively criticize them for their behavior. Girls, then, are necessarily more "polite" than boys. Speech for girls must serve both these ends: (1) to create and maintain relationships of equality and closeness and (2) to criticize others effectively without rup-turing a desired relationship. Their typical styles of speech will emphasize hortatives rather than commands and exhibit a high proportion of modal constructions and emotional terms (marking close, equal relationships), and be relatively polite and indirect (acceptable criticism).

Goodwin (1980) is a study of play activities among black children in an urban neighborhood of Philadelphia that is often cited as confirming Maltz and Borker's proposal. In a task of making slingshots in preparation for a fight, the boys consistently gave each other orders, often in a bald form: Gimme the pliers!; Man, don't come in here where I am!; Get off my steps. Or they refused to perform as directed, posing a direct challenge to the status of the person who ordered them: No. I'm not going in there. I don't feel like it; I'm not getting 登ut_ of nowhere. Or they ridiculed the person who gave the order: You shut up you big lips. The whole point of the verbal

298 The Ethnography of Speaking

interchanges in these tasks is to negotiate for relative status: each boy post-ures, challenges, and counterchallenges until a hierarchy for this activity is settled. Of course, hierarchies in boys' groups are fluid; what emerged in this task may have been quite different in another task domain. The girls studied engaged in a task of making glass rings out of bottle necks. Instead of commands, the girls typically used hortatives or modalized expressions: Let's go around Subs and Suds: come on; Let's go find some; Maybe we can slice them like that; We gonna paint em and stuff: Also, rather than refusing the suggestions of others, the girls typically agreed to them: Hey, let's go in there and ask do they have some cases; Yep Okay? Yep let's go and ask them. Further, if a conflict emerged, rather than regarding it as a challenge to

status, girls typically reacted with requests for further information (Good- win 1980:169)

(on reaching a city creek)

Pam: Y'll gonna walk in it?

Nettie: Walk in it, You know where that water come from? The toilet.

Pam: So I'm a walk in it in my dirty feet. I'm a walk in it and I don't

care if it do come.

Nettie (overlaps): You could easy wash your feet (to investigator). Gonna

walk us across? Yeah. I'll show y'all where you can come.

Note that at the end of this interaction each of the participants has given a directive and countered the other's action; neither is one up or one down, as would be the typical end goal of an equivalent conversation involving point scoring.

Tannen (1990) elaborates Maltz and Borker's proposal, developing a full model of male and female conversational practice along the lines of the 鍍wo cultures_ model. She interprets the various findings of research in language and gender, such as the above asymmetries in interruptions and conversa-tional maintenance, in the light of this model. She argues that women emphasize connection and intimacy with others and use language to estab-lish and maintain these connections. They avoid direct confrontation and use language to suggest actions, rather than ordering them, hence what is commonly noted as the greater politeness of women, both positive and neg-ative. Males, on the other hand, continue the status games learned in boy-hood into adulthood. Language is a weapon for asserting themselves and their status and power. It is a major arena in which men can compete with each other for the much desired one-up position. They use language to display their power and skills and to defend themselves from the kind of verbal attacks they typically launch on others' claims to status and power. In short, the type of status games that Javanese men engage in with the artful use of the system of speech levels is just one exemplar of a much

Language and Gender 299

more pervasive male perspective on language. If, for women, language is the means to forge connection, for men, it is the way to establish position, and ultimately, power.

The “two cultures” model is not without its critics. Some argue that it greatly exaggerates the real differences in the engendered norms for speak-ing. Indeed, in one of the very studies commonly cited to support the “two cultures” model discussed above (Goodwin 1980; 1991 ), the author actually argues against such an interpretation of her findings. She claims that girls are just as skillful in countering the status claims of others as boys, rather they simply choose to use these resources less frequently. But if need be, girls can maintain their position just as effectively as boys (and probably, if restricted to just verbal means, more effectively than boys). The “two cultures” model's emphasis on the separateness of the two cultures as learned in childhood does not square with these findings, nor does it really account very well for the great deal of clear and understood communication that does occur across the sexual divide. After all, men and women do spend a great deal of their time together and often do communicate effectively, interpreting the other's utterances in the way they intended (or sometimes deliberately in a way they did not intend, due to an understanding of the other's speech norms, as when a man correctly interprets a woman's indir- ect wh-imperative like why don't you leave your shoes outside as a command

But feels he can safely choose to ignore it since it is cast in the form of a question. He would be unlikely to do so if a male buddy said leave your shoes on the doorstep). The “two-culture” model claims that misunderstand-ings arise when shared norms for speaking between the sexes are mistakenly assumed. But a close look at the ethnographic and sociological literature demonstrates that much more typically some differences in the engendered norms are in fact taken for granted (see Malagasy and our own folk culture beliefs about "speaking like a lady" and the above wh-imperative example). Finally, the nature of the separation of men and women is not equivalent to that between, say, Australian and Yimas culture; but the “two cultures” model has not come to terms with this.

Another side to the relationship between gender and language might be called gender deixis, in which some actual linguistic elements are index- icals of some fact about gender, maybe that of the speaker or that of the addressee, or both. Such gender deixis is well attested in languages and has been so since at least the sixteenth century, during which the Spanish conquistodores commented that the Native inhabitants of the Caribbean islands had distinct languages for men and women. This was somewhat

300 The Ethnography of Speaking

of an exaggeration; in actuality there were some differences in grammat- ical forms in the language depending on whether the speaker was male or female. Since this time, such differences indicative of gender deixis have been reported for many languages, especially the Native languages of North America (e.g. Bogoras 1922; Ekka 1972; Flannery 1946; Sapir 1949:206-12; Taylor 1982). I will consider here the case of Koasati, a Native language of the south-eastern United States, as described by Haas ( 1964). In this lan-guage the speech of men and women differ in certain verbal forms, so the choice of one set of forms as opposed to the other can be said to point directly to the speaker's sex. Here are some of the basic differences in the verb, summarized by Haas (1964:228) in three basic rules:

1 If the women's form ends in a nasalized vowel, the men's form substitutes

an s for the nasalization:

W M

lakaww?· lakaww ?·s “he will lift it”

k?· k?·s “he is saying”

W M

lakaww?l lakawwís “I am lifting it”

molh?l molhís “we are peeling it”

3 If the women's form has final falling pitch and ends in a short vowel followed by n, the men's form retains the falling pitch, but substitutes an s for the n and lengthens the preceding vowel:

W M

lakaw??n lakaw??·s “don't lift it”

ta?ilwan ta?ilwa·s “don't sing”

Haas thus analyzes the women's forms as basic and the male speakers' forms as derived from them. This entails a claim that the historically older, con-servative forms are those of the women, while the male forms have arisen by a series of phonological changes to these over time.

Koasati exhibits gender deixis in which the forms involved specific- ally signal the gender of the speakers, regardless of the addressee. In other such cases, the male forms may be restricted to only the situation when the addressee(s) is exclusively male, with the female form used elsewhere (in Yana; Sapir 1949), or more complexly, as in Kurux (Ekka 1972) in which the forms basically signal same-sex dyads, male to male and female to female, but the male forms are used with cross-sex dyads, except for the third-person singular in which a specific form exists for males speaking to females.

Language and Gender 301

Cases like Koasati, with specific language forms which signal gender deixis, are relatively rare across the world's languages. Much more common, indeed

segmentable morphemes or allomorphs, but more importantly, they are not presupposing of fixed contextual backgrounds; rather it is their very usage which frames the meaning of context, as, for example, when my actual

York City English, especially middle-class women.

Peter Trudgill (1974) in his study of sociolinguistic variation in Norwich in England discovered a similar pattern. Considering the pronunciation of the final /?/ in words ending with -ing (it is well known this varies from [?]

302 The Ethnography of Speaking

15.1

to a more casual, “substandard” [n] ), he found the distribution of the [n] pronunciation across the classes according to sex shown in 15.1 (Trudgill 1983:86). Note, again, that in all classes, women use the more standard [?] pronunciation more frequently; indeed, among the middle-class women, the [n] pronunciation hardly seems to exist at all.

The findings from Koasati, New York City, and Norwich English all demonstrate women to be more conservative than men linguistically, stick-ing to speech forms less innovative than those of men. Because older, less innovative forms tend to be more prestigious and more “standard,” women are typically described as speaking in more statusful styles than their male counterparts. Indeed, women seem to be at least subliminally aware of this tendency. In a brilliant study, Trudgill (1972) investigated the normative judgments men and women made toward their own speech. He considered a number of variables, and I will focus on the variable pronunciation of the final vowel in ear, idea, here, which varies from the standard pronunciation [I« ] to the local [E ·] (making ear homophonous with air, and here, with hair). Trudgill tape-recorded interviews and noted the actual distribution of pro-nunciations for each interviewee. He then asked each of them how they thought they pronounced these words, noting that some self-reports were accurate, some overestimated their use of the standard varieties (over-reporting) and some underestimated it (under-reporting). The results, which are very instructive are set out in 15.2 (Trudgill 1983:92). There is a marked gender asymmetry in the figures. Well over half the women (68 percent) claimed to be using the prestige standard pronunciation [I« ], when in fact they were not, but half of the men claimed to be using the local, less prestigious and more working-class variant [E ·], when, in fact, they were using the standard! What could be the cause of this asymmetry? Follow- ing Labov (1972a), Trudgill called the source of the male behavior, covert prestige, in that the speech forms typical of working-class speech have an appeal to middle-class men, associated, as it seems to be, with masculinity

Language and Gender 303

and toughness, reflecting clear articulations of gender understandings in British culture. If covert prestige is really operative in the speech of men, then we would expect male speech to be more innovative, to articulate

claims of class and gender affiliation.

But what about the women? Why are they sticking to the more prestig-ious forms, indeed, over-reporting their use of them? The work of Milroy (1987a) points to an explanation. In her study of Ballymacarrett, a Belfast

neighborhood, she found that men had a large network of social connections within the neighborhood, while women did not (remember Rosaldo's (1974) contrast of the male public realm versus the female domestic sphere). Men live their lives mostly in this neighborhood network and use the local lin-guistic features to signal this affiliation (covert prestige). Women, on the other hand, either stay within the home, a circumscribed domestic sphere, or work outside the immediate neighborhood. In order to claim connection with this wider world and to function effectively within it, prestige standard linguistic features are more valued.

That something like this explanation is necessary is shown by cases like that described by Gal (1978), in which women are the vanguard in language ehange, rather than the conservative custodians they are usually portrayed as. In a bilingual Hungarian-German speaking part of Austria, women are choosing to give up speaking Hungarian in favor of German, while their male compatriots are clinging to Hungarian. This reflects an ambit claim by women in favor of the wider industrial world in which German is spoken, with its greater possibilities for careers and life choices, especially as they increasingly choose German-speaking husbands from this world. Hungar- ian is the language associated with the farming life of the local community. Most of the young men in the community are still selecting this farming life-style as their preferred life choice, so they maintain Hungarian as their primary language. In this case, women are the linguistic innovators, al-though in favor of the more prestigious language in terms of the wider world.

Similar findings are reported by Nichols (1980, 1983), concerning a tran-sition, on a small island in rural south-eastern United States, from an English-based creole called Gullah to Standard American English. The original

304 The Ethnography of Speaking

language of the island was Gullah, but as industrialization has come to the adjoining mainland, islanders have migrated out and have adjusted their speech accordingly. Women migrants have typically taken up service jobs, such as sales clerks or even professional jobs like teachers, while the men usually have jobs in the construction industry, often several islanders work-ing together. The jobs held by the women clearly require some facility in varieties of spoken English close to standard, but this is not the case with those of the men. Not surprisingly, the women are switching from Gullah to Standard English much more quickly than the men. The island forms an interesting contrast with a nearby coastal community with far more limited employment opportunities, the men holding laboring jobs in plantations or unskilled factory work and the women at most domestic laborers or seasonal farm workers. The much more restricted access to the wider world for women in this community results in their being linguistically conser-vative, exhibiting a higher rate of creole English linguistic features in their speech than do men. The case is probably an analog of the Ballymacarrett one discussed above, but in this situation the linguistic norms conserved are substandard ones, rather than prestigious standard forms. Another similar case is that of Yimas village, in which younger males are rapidly switching to Tok Pisin, indicative of communication and action in the wider world, while the women continue to favor the local vernacular, with its local dome-stic connotations. Thus, women tend to be conservative linguistically when their worlds of action are circumscribed and restricted, especially to the domestic sphere, but to the extent that women may make a greater claim to a wider public world of action than men, their linguistic usage will be innovative. How this last point squares with Rosaldo's ( 1974) claims re-mains to be clarified.

Finally, one stereotype of women's speech is that it is more polite than men's. Studies at least since Lakoff (1975) have made this claim, but is it true? Lakoff (1975) mentions a number of linguistic features which she believes is associated with women's greater politeness. One of these is tag questions, in which the subject and verb of the preceding statement is repeated in a question of typically reversed polarity, such as Bill took Luke to the party last night, didn't he? or Louise and Lucille didn't leave together last night, did they? Lakoff claims that female speakers tend to make use of tag questions as a consequence of their reluctance to make direct assertions, to avoid potential conflicts with addressees, a type of negative politeness in Brown and Levinson's ( 1987) terms. Tag questions have come under inten-sive study since Lakoff, and most of this work contradicts her claim of a

Language and Gender 305

gender asymmetry in their use (Dubois and Crouch 1975). An especially interesting study was done by Holmes (1984). She argues first of all, that tag

questions serve two types of functions: modal tags, which request informa-tion from an addressee or indicate that the addressee confirms the truth of a statement (Luke and Bill went to the party, didn't they?) and affective tags, which display the speaker's concern for the addressee (You didn't go there,