It has been demonstrated by sociolinguists that women and men do not speak alike. In fact, even in the analysis of dialect variation, where the focus is on women (or men) as innovators, different linguists have radically different opinions, buttressed, it would appear, by incontrovertible evidence. The present paper will suggest that the naivete of linguists has influenced the research designs which permit the collection of data on which these analyses of language variation were made; it will propose methods for collecting data which will reveal more clearly the language-gender connection, without confusion induced by the Observer's Paradox (Labov 1972), the tendency for the interviewees' speech habits and reports of speech habits to be influenced by the mere fact that they are being observed by an outsider.
The paper will suggest that controlling for sex composition of interviewer-interviewee dyads in studies of linguistic change is relevant to all levels of communicative behavior. Until we design our studies to compare linguistic practices by the sex composition of the interacting dyad - especially among age similar pairs - it is premature to speculate about sex differences in language change. The paper provides examples of studies that have confounded the cross-gender influence on language choice with the influence of interviewee sex alone, and discusses the implications for language change studies, and for our understanding of convergence and divergence. The paper suggests three methods which can be used to control for this problem: The researcher can use homogeneous pairs, can use naturalistic observation, or systematic comparisons by dyadic sex differences. Hopefully this discussion will not only sensitize researchers to the potential shortcomings of the research designs in common use [and the dangers of drawing conclusions from studies based on such research], but will provide more adequate possible designs.
Review of the literature
Principle I: Women as rearguard guerrillas
Many British linguists portray women as providing a conservative/retrograde influence in changes which are assumed to be conscious, or 'from above [the level of awareness]' (Trudgill 1972, 1983; Wardhaugh 1991:[section]8).(1) In fact, women are reported to be fighting this rearguard action, using standard forms, even when linguistic changes are completed. Labov (1990) refers to this as Principle I of the sexual differentiation of sound change.
* The linguistic variable -ing - or (ing), with parentheses denoting it as a sociolinguistic variable - is known to be fairly stable in English, since the variation between -ing and -in' has been attested for centuries, without the change having become categorical. The 'blame' for retention of - ing is placed on women and the Upper Middle Class (Trudgill 1972, Chambers 1995).
* Wolfram's (1969) interviews with Afro-Americans in Detroit also revealed that women favored Standard forms more than men: men used more negative concord [can't get no satisfaction], [f] for (q) as in [wIf] for with, and copula deletion [He . . . ready.] than women.
* Trudgill (1972) and Chambers and Trudgill (1980) found that in both England and Norway women report themselves as speaking even more Standard than they actually do, while men report themselves as using more vernacular pronunciation.
Different explanations have been given for this contrasting behavior of men and women. Trudgill (1983:[section]9) theorized that men have economic sources of prestige, while women tap into the prestige of a higher economic group, using language as a status display. Alternatively, he suggests that women adapt themselves to more standard usage for interacting with children. Either of these hypotheses addresses the apparent fact that women limit vernacular change by using older - 'Standard' - forms.
However, other hypotheses have been proposed which permit other conclusions:
* Like Trudgill, Cheshire (1982) found that Reading lower working class preteens and teenaged girls also use much more Standard syntax than the boys. She drew the conclusion that women's use of Standard syntax is not a 'rearguard action', but is a response to economic needs: Boys expect to gain their livelihood in blue collar occupations (truck drivers, masons), while girls look forward to white collar occupations (hair dressers, secretaries). Cheshire concludes that the linguistic marketplace supports the sex differences observed.(2)
* Nichols (1983, 1984) found that South Carolina Creole features are much stronger in the speech of men than of women. Like Cheshire, Nichols showed that the men in her study had blue collar jobs which neither required, nor permitted them to interact with Standard speakers, while women worked in service jobs which required them to interact with and accommodate to Standard speakers.
* Milroy (1980) found that Belfast vernacular-speaking women used the more Standard form more consistently in three of four variables studied; moreover, since younger men use the vernacular forms more consistently …