It is truth universally accepted that men and women use language differently. Since the early 1990s their distinctive communication syles have been probed in a steady stream of best-selling books with titles like "Men are from Mars, Women are from Venus", "Why men don't listen and women can't read maps", and (my own particular favourite) "If men could talk".

Continually recycled in everything from women's magazine problem pages to stand-up comedy routines, the claims made in this literature (for instance, "men do report talk and women do rapport talk", or "women need to talk things through but men just want to go into their caves") have acquired the status of simple common sense.

But common sense can be too simple. Linguistic researchers don"t dispute that gender influences linguistic behaviour: any social division that affects the way people's lives are lived is bound to affect their use of language. But the way that works is complicated: it cannot be reduced to the simple generalizations which are endlessly repeated in popular sources.

One reason to be sceptical about these generalizations is that they vary so much depending where and when you look. The belief that men and women speak differently is found in most societies, but what people believe the differences actually are differs considerably from one to another. Among contemporary westerners, for instance, the prevailing belief is that women are less direct communicators than men, and that their preferred style of interaction is co-operative and supportive, whereas men are more competitive and assertive. Yet other cultures insist that the opposite is true: it is men who take pride in their diplomacy and verbal delicacy, while lamenting that women are so aggressively plain-spoken.

Another now-widely held western belief – that women are naturally better than men with words – is a surprisingly recent development. Less than a hundred years ago, the consensus among experts and laypeople alike was that men were superior on virtually every measure of verbal skill. This assessment followed logically from the general conviction that women were innately less intelligent than men. Even the observation that men were slower readers was interpreted as evidence of their greater intellectual prowess. "With the quick reader", the Danish linguist Otto

Jespersen explained in 1922, "it is as though every statement were admitted immediately and without inspection to fill the vacant chambers of the mind, while with the slow reader every statement undergoes an instinctive process of cross-examination". With hindsight it's easy to dismiss this as blind prejudice. But what we have replaced it with might seem just as biased in a hundred years" time.

What people believe about some aspect of human behaviour is not always a good guide to the reality of that behaviour. Observations about linguistic behaviour are notoriously unreliable: people's accounts of how they typically talk are rarely confirmed by analysis of their recorded speech. Part of the explanation for this inaccurate reporting is that no one has the capacity, or the time, to do statistical analyses while they are actually engaged in talking. But the gap between our perceptions and reality also reflects what scientists call "confirmation bias" – our tendency to notice and remember things that match our expectations, while overlooking or forgetting things that do not. We have strong expectations about the behaviour of men and women, and this probably explains why so many people are convinced that the claims made in popular books are borne out by their own experience. "My husband/boyfriend/father never listens", they say, or "it's definitely the women who do the talking in our house".

The idea that women talk more than men is a good illustration of the power of our perceptions to mislead us about the facts. No belief about gender differences in language is more widely or strongly held, yet none receives less support from the available research evidence. Research on this question has been going on for decades, and its conclusions can be summarized in two sentences. In informal exchanges among equals, there is usually little difference in how much men and women talk. In more formal situations – business meetings, seminars, public debates – the commonest finding (with some exceptions) is that men talk more than women.

On closer examination, however, this pattern is not primarily about gender: it has more to do with status. In formal contexts, higher-status people tend to talk more than lower-status ones. Gender enters into this indirectly, because in most institutions the highest-status positions are still far more likely to be occupied by men. It is these high-ranking men (rather than all men) whose behaviour accounts for most of the imbalance. In the few studies which have found the opposite pattern – women talking more than men – it has been either because the women outranked the men present, or else, interestingly, because the topic under discussion was something both sexes believed to be a female area of expertise.

To understand these findings, we have to bear in mind that talk is what conversation analysts call a "joint accomplishment". The patterns we observe are the collective product of group interaction: they cannot be explained by looking at the behaviour of male and female individuals in isolation. The reason high-status/male speakers get more talking time is not just that they have a more assertive way of speaking than low-status/female speakers. It is ultimately because others defer to them: their contributions are solicited, acknowledged and supported.

The importance of this dynamic, and the relevance of gender to it, has been underlined by research investigating the tendency for boys to monopolize discussion in mixed-sex classrooms. In a study carried out by Judith Baxter (2006), the dominant boys relied heavily on one or more "sidekicks". These were peers who echoed their observations, laughed at their jokes and generally helped them to take and then keep the floor. Girls who put themselves forward in a similar manner, by contrast, did not usually command the same support. Though they were just as verbally assertive, they could not establish themselves as leaders because their peers refused to act as followers.

Along similar lines, some researchers have questioned the theory that women lose out to men in the workplace because their speech is insufficiently authoritative and direct. One recent study of workplace meetings (McRae 2009) found that the men and the women did not differ in their styles of speaking – both favoured a relatively co-operative style – but overall it was still the men who were more likely to get their points acknowledged and ultimately acted upon.

Who has, and who does not have, power and influence in interaction is not simply a question of how they speak: how they are heard and responded to by others is at least as important. And gender seems to affect that whether or not there are measurable differences in men's and women's linguistic behaviour. In the classroom, girls may be as assertive as boys, but they do not get the same support from their peers. In the workplace, men may be as co-operative as women, but what they say still carries more weight. These are certainly gender issues, but the problem isn"t that men and women communicate in different ways. By making such an issue of our supposed linguistic differences, popular folklore just obscures the real problem, and reinforces the prejudices which help to cause it.

The debate on which sex talks more, and why, illustrates two of the most important insights of recent research on gender and language-use – insights which have led most current researchers to reject the popular obsession with cataloguing differences between men and women. Thirty years ago, researchers did believe that it was possible to make general statements about gendered linguistic behaviour. But as more and more evidence has accumulated over time, we have come to realize that such statements are at best partial truths, while at worst they are seriously misleading.

Why is this? First, because linguistic behaviour is extremely context-dependent, and this makes it difficult to generalize about the influence of gender from one situation to another. Many features of speech-style that are often attributed to the influence of gender may in fact be more directly linked to contextual factors, such as the roles participants are playing, the setting in which talk is occurring and the activity of which it is part.

A simple example concerns the distinctive style of speech used to talk to infants. This used to be considered a "female" style, and referred to as "motherese". Today it is more often labelled "caretaker speech", because we now know it is used in exactly the same way by men who look after children. Its association with women reflected the fact that, until recently, the care of the very young was an almost exclusively female preserve.

When the 1997 general election brought a record number of women MPs to Westminster, many commentators predicted that women's preference for consensus and co-operation would make the ultra-adversarial (and previously very male-dominated) House of Commons a kinder, gentler place. But when the linguist Sylvia Shaw put this theory to the test by comparing the behaviour of male and female MPs, she found that the women were no more co-operative than the men. They had adjusted to the norms of the activity they were engaged in – which was, after all, the inherently non-consensual activity of debating with political opponents (Shaw 2006).

Similarly, the anthropologist Bonnie McElhinny (1995) observed that recently-recruited women police officers in Pittsburgh, USA, quickly adopted the same unemotional way of interacting as their male colleagues. Their intonation was monotonous and they very rarely smiled. When she asked if they had felt pressure to shift to a more masculine style, they said it had nothing to do with gender, it was simply the most appropriate style for the job. It is not hard to see their point: a deliberately unemotional demeanour may help officers defuse the highly-charged and potentially dangerous situations they are often called to deal with. Conversely, a warmly empathetic and supportive style might be helpful in a job like nursing, where the aim is to make patients feel cared for. And sure enough, when Joanne McDowell (2008) investigated male nurses" talk, she found they were just as empathetic as their female colleagues.

The commentators who predicted that a large influx of women would change the linguistic culture of Parliament were implicitly assuming that the speech style a person uses is like their eye colour or their blood type, a permanent attribute which they carry with them into every situation. But in reality, speech style is something that changes to meet the demands of varying situations. Male and female speakers have often been studied doing different things in different social settings – not surprisingly, given that there is still a strong cultural tendency towards gender separation. Often men and women choose, or get channelled into, different occupations, leisure activities, friendship groups and domestic roles. But when we look at groups of men and women doing the same thing in the same setting – be that politics, police work, nursing or parenting – we do not tend to see marked differences in their speech. This suggests that what people are doing may have more influence on their way of speaking than gender in and of itself.

The other main reason why gender generalizations are misleading is that "men" and "women" are not internally undifferentiated categories. The first question that needs to be asked about any statement that "men do so-and-so and women do such-and-such" is "which men and women?" Are we talking about David Beckham or the Archbishop of Canterbury? Ann Widdecombe or Madonna? Time and again, researchers who set out to compare men's and women's ways of speaking have discovered that the differences among the men or the women are as large as, or even larger than, any difference between the two. Glossing over the diversity that exists within each gender group is like looking through a telescope at something that needs to be examined under a microscope.

The diversity of gendered linguistic behaviour can be related to one of the most basic functions of language, which is not just a tool for exchanging information, but also a resource for expressing our identities. Whatever else we may be communicating when we speak, we are always communicating something about who we are, who we feel we are like and who we feel we are different from. And though a person's status as either a man or a woman is obviously important for their sense of who they are, there is far more to gender identity than just being male or female. After all, in reality no one would ever answer the question "what sort of person are you?" or "what sort of person is X?" by saying simply "I"m a woman" or "he's a man". We define ourselves not as generic men and women, but as specific kinds of men and women. Often, therefore, what we care most about are the small details that distinguish one kind of man or woman from another.

When I was a teenager, the girls at my school were divided (by ourselves rather than the authorities) into two distinct categories, "slags" and "swots". Whichever one you were, the cardinal rule was to avoid any way of behaving that was generally associated with the other. We paid far less attention to not behaving like boys: it wasn"t the boys that we ever really compared ourselves to. And this, it turns out, is typical.

Linguistic studies conducted in schools have repeatedly found that girls who belong to different subgroups or cliques tend to be more different from each other than they are from the boys in their group. And while this tendency to emphasize small distinctions is at its height during adolescence, it's fair to say that it doesn"t completely disappear in later life.

How femininity or masculinity is expressed varies with all the other social attributes which make up a person – their age, ethnicity, nationality, social class – and also with a person's attitude towards the various alternatives on offer. Two objectively very similar women (both, say, white middle-class professionals who combine paid work with raising a family) may define themselves quite differently: one may think of herself as a "career woman" who also has children, while the other describes herself as a "mum" who also works. A young man may embrace the identity of a "lad" while emphatically rejecting that of a "new man" or a "metrosexual". If there were only one way to be a woman or a man, these distinctions, and the labels we use to make them, would be meaningless. But if there isn"t just one way for men or women to be, why would we imagine there is only one way for them to speak?

The theory behind the lore of Mars and Venus is that men and women speak differently because their minds work differently. In the words of the old adage, "language is the dress of thought". But it would be more accurate to liken language to dress in the literal sense. The way we talk, like the clothes we wear, is an aspect of our personal style. And just as the rules for gender-appropriate dress leave plenty of scope for individual self-expression, so too do the conventions of gender-appropriate speech.

Some postmodernist theorists conceive of identity, including gender identity, not simply as something people have, but as something they perform. This goes along with the idea that in contemporary societies, identity has become more fluid than it was for past generations: who we are, or will become, is no longer determined once and for all by birth and tradition. We are not just permitted but expected to be active in defining ourselves through the choices we make about all kinds of things, from what we do for a living to what we buy at the supermarket – and also, of course, the way we express ourselves linguistically.

Am I suggesting that we are as free to choose our speech styles as we are to choose our clothes or our brand of detergent? Not quite. For one thing, as I have already said, our linguistic behaviour is always shaped by the behaviour of those we interact with. And there are certainly aspects of our speech which are not matters of individual choice. I didn"t choose to be a native speaker of English, or to speak it with a northern accent (which I can modify, but not eradicate). Gender might seem to belong in the same category of things we don"t control; but while most of us do consider our basic status as men or women to be fixed and unalterable, the existence of so many different styles of masculinity and femininity means we still have choices about the way we perform it.

In summary, gender influences language-use for three main reasons. First, it affects what people habitually do, and the way people talk is always shaped by the demands of the activity they are engaged in. Second, gender is linked to power and status, and these play a significant part in the dynamics of verbal interaction. Third, gender is an element of personal identity, and communicating identity is one of the key functions of language – though since gender identities are varied rather than uniform, the language used to express them will be equally varied. Together, these three factors explain why there are gender differences in language, but also why it's so difficult to generalize about them. Stories about the two sexes coming from different planets, or having different kinds of brains, cannot account for the complexity and the diversity of men's and women's language in the 21st century world.

References

Baxter, J (2006), "Do we have to agree with her?" How high school girls negotiate leadership in public contexts, in Baxter, J. (ed.) Speaking Out: The Female Voice in Public Contexts. Houndmills, Basingstoke: Palgrave Macmillan.

Jespersen, O (1922), "The woman", reprinted in Cameron, D. (ed.) (1998) The Feminist Critique of Language. London: Routledge.

McDowell, J (2008), "Gender, language and occupational roles: Exploring men's use of language within the female dominated environment of nursing", Ph.D thesis, University of Ulster, Jordanstown.

McElhinny, B (1995), "Challenging hegemonic masculinities: female and male police officers handling domestic violence", in Hall, K. and Bucholtz, M. (eds.) Gender Articulated: Language and the Socially Constructed Self, pp. New York: Routledge.

McRae, S (2009), "It's a blokes" thing: gender, occupational roles and talk in the workplace", in P. Pichler and E. Eppler (eds.), Gender and Spoken Interaction. Houndmills, Basingstoke: Palgrave Macmillan.

Shaw, S (2006), "Governed by the rules? The female voice in Parliamentary debates", in Baxter, J. (ed.) Speaking Out: The Female Voice in Public Contexts. Houndmills, Basingstoke: Palgrave Macmillan.