Speech Synthesis


This article originally appeared in T&S Issue 40, Winter 1999/2000.

If male-to-female transsexuals are women in men’s bodies, why do they have to learn to ‘speak like a woman’? Debbie Cameron argues that these lessons, like those women working in call centres are subjected to, are strategies for producing patterns of gendered speech which maintain patterns of dominance and subservience.

When I taught an introduction to feminism many years ago, I would sometimes get students to examine the lyrics of popular songs for their ideological contradictions. One of my favourite texts for this purpose was ‘You make me feel like a natural woman’. Even when performed by the incomparable Aretha Franklin, there is no getting away from the imbecility of this song. If the attentions of a man are needed to turn the singer into a ‘natural woman’, what was she before? Similar puzzles arise with songs like ‘More than a woman’ (eh?) and ‘She’s always a woman to me’ (as opposed to what, a giraffe?). For the authors of these lyrics, evidently, there is more to being a woman than simply being female. Although being a woman is supposedly our ‘natural’ state, there is always room for improvement, and for anxiety about whether we’ve got it right yet.

The theoretical point I intended the song lyrics to illustrate — that women are made rather than born — has been pushed to its extreme by today’s postmodernists and queer theorists. They insist that biological femaleness is not just insufficient to make you a woman, it is more or less irrelevant to the enterprise. In support of this argument they are fond of invoking the drag queens, transvestites and transsexuals for whom gender ‘construction’ is literally a form of DIY. Radical feminists tend to dislike this move, particularly when it involves either treating gender as a lifestyle option akin to wallpapering your living room, or celebrating DIY women as feminist revolutionaries. I too find these tendencies irritating (and later on I will give some reasons why the second is way off target). But I do think there is something to be learned from looking more closely at how you make a DIY woman. If we are interested in current ideologies of femininity, there are few more revealing sources than step-by-step instruction manuals for turning men into women.

DIY women

My own interest in this subject springs from a broader concern with the relationship between gender and language, or more exactly speech. When the subject of transsexuals crops up in the media (as it seems to do incessantly, on everything from daytime talk shows to soaps to science documentaries), the main focus is usually on the chemical or surgical reshaping of the body. Wardrobes and sex lives also feature quite prominently. Far less attention is given to the changes many transsexuals make to the way they speak, though these are often just as dramatic as the changes in their physical appearance. (They may also be produced by the same drastic methods, i.e. hormone treatment and surgery.) The memoirs of ‘pioneering’ transsexuals like Renée Richards make clear that speech has always been a major concern for individuals seeking to pass, but over time a whole sub-industry has developed to deal with their linguistic needs. There are speech therapists, voice trainers and assorted consultants who specialise in advising transsexuals on how to talk, and innumerable books — ranging from abstruse clinical volumes to popular self-help texts — have been written on the subject. In all this literature it is constantly emphasised that sounding like a woman requires a great deal of work. One book written for the families and friends of transsexuals warns that ‘for most MTF transsexuals, considerable time and effort is needed to train their voices — even the way they cough and clear their throat — to become as gender appropriate as possible in register, pitch, inflection and intonation’. [1]

Interestingly, it is assumed that female to male (FTM) transsexuals do not have to work so hard at learning to speak in a ‘gender appropriate’ way. One explanation for the difference is that testosterone, taken by FTMs, thickens the vocal cords and lowers voice pitch, whereas oestrogen, taken by MTFs, does not have a parallel effect on the vocal tract. Thus MTFs need additional voice training. But on closer examination, this explanation is unconvincing. One article in a textbook on the clinical management of ‘gender dysphoria’ claims that even when MTFs undergo vocal tract surgery this ‘does not obviate the need for speech therapy in almost all cases’. Plainly, ‘gender appropriate’ language as specified in speech training manuals and self-help literature for transsexuals involves a great deal more than the pitch of the voice.

The literature on this subject, both expert and popular, is like a ‘best of’ compilation drawing on something like a century’s worth of stereotypes about women’s speech. One expert explains, for instance, that ‘when women talk they move their mouths more than men’, while another states that ‘women smile more and have many more encouraging nods’. This piece of ‘scientific’ advice clearly has nothing to do with pitch; it doesn’t even have anything to do with speech. Rather it has to do with a whole style of interpersonal communication, and the contribution that style can make to the perception of a person as ‘a natural woman’. A more popular work, Miss Vera’s finishing school for boys who want to be girls, makes clear what the experts are dancing around in their supposedly objective descriptions of ‘gender appropriate’ behaviour. Miss Vera advises readers that if they wish to pass as women they must develop ‘a greater willingness to listen’. Learning to speak like a woman does not just mean learning to speak differently, it also means training yourself to speak less.

Other common pieces of advice to MTF transsexuals are to speak more softly than they did when they identified as men and to ask more questions (especially ‘tag questions’ of the ‘isn’t it/don’t you’ variety). Less extensive guidance is offered to women on how to talk like men (an asymmetry which, once again, cannot be put down simply to the testosterone/oestrogen difference). But just as Miss Vera counsels DIY women to adopt the linguistic markers of subordinate status, advice for DIY men tends to focus on the linguistic markers of power. Writers suggest that the FTM should learn to ‘talk slow, take all the time in the world’, or to ‘tell people what you want rather than asking for it’. These are presented simply as ‘masculine’ habits, but it does not take a genius to work out that they are also the habits of a dominant social group. The speaker who feels able to ‘take all the time in the world’ has a strong sense of entitlement, which is typically produced by long experience of privilege. Quite similar advice has been handed out for years to women ‘high fliers’, the most notorious example being Margaret Thatcher. No one involved in Mrs Thatcher’s linguistic makeover ever suggested that the point was to make her sound like a man. The point, rather, was to give her more ‘authority’ by eliminating the telltale markers of subordinate status.

Gender stereotypes

An important source for many of the gender stereotypes in the literature I have been citing is a 1975 book titled Language and Woman’s Place, written by the US feminist linguist Robin Lakoff. Robin Lakoff set out to describe what she referred to as ‘women’s language’, a way of talking that women, she argued, are compelled to adopt by a society intent on depriving them of power. But the picture she painted has been modified considerably in the light of empirical research over the past 20 years: most language and gender researchers today regard ‘women’s language’ as a stereotype with only a tenuous relationship to the facts of women’s speech. It follows that if transsexuals model their behaviour on Robin Lakoff’s description, they are not imitating real women but reproducing a sexist stereotype, whose main characteristic is subservience or powerlessness. Robin Lakoff’s own intentions were clearly progressive — she was criticising the pressure put on women to ‘talk like ladies’ — but the use of her work as a sort of instruction manual for DIY women is entirely reactionary. (So much for the idea of transsexuals as the vanguard of feminist revolution.) Arguably it is also self-defeating, in the sense that someone who follows the recipe in its entirety will not be able to pass as a woman; rather they will sound like a parody of one.

Some transsexuals, to be fair, resist the ‘women’s language’ stereotype. Self-styled ‘gender outlaw’ Kate Bornstein, for instance, gives the following critical account of the voice lessons she attended:

I was taught to speak in a very high-pitched, very breathy, sing-song voice and to tag questions onto the end of each sentence. And I was supposed to smile all the time when I was talking. And I said, ‘oh, I don’t want to talk like that!’ The teachers assumed that you were going to be a heterosexual woman. No one was going to teach you to be a lesbian, because lesbian was as big an outlaw as transsexual.

Kate Bornstein is among those ‘transgendered’ individuals who deliberately set out to blur boundaries and create hybrid identities. Whereas the classic ‘transsexual’ wants to be ‘the opposite sex’, resembling the archetypal woman or man as closely as possible both physically and behaviourally, the ‘transgendered’ person does not set out to pass. S/he may identify as neither gender, or both at once; s/he may or may not go in for extensive chemical and surgical alteration; s/he often treats (homo)sexuality as a primary identity, as in the quote above where Kate Bornstein explains that she rejected the advice of her teachers because what she really wanted to sound like was not a ‘woman’ but a lesbian.

If we leave aside the rather odd suggestion that someone could be taught to sound like a lesbian, Kate Bornstein is making an important point here which is often overlooked in discussions of ‘women’s language’. Put briefly and bluntly, the symbolic meaning of this kind of speech is not just powerlessness but more specifically eroticised powerlessness. The high pitched, breathy and ‘sing-song’ voice recommended to DIY women is also the voice of the archetypal sex siren — the voice for instance of Marilyn Monroe performing ‘I just want to be loved by you’. It is meant to arouse heterosexual men.

Sexualised speech

In support of this point, we might consider that MTF transsexuals are not the only recipients of advice on speech which is based on the ‘women’s language’ model. Very similar advice is offered to, and followed by, telephone sex workers catering to heterosexual male callers. The language used by telephone sex workers, and the instructions given to them on this subject by their employers, is the subject of an interesting article, ‘Lip service on the fantasy lines’, first published in 1995 by the anthropologist Kira Hall. The phone operators are advised to adopt high, breathy and swoopy voices, to ask the caller endless questions, and to use the kind of ’empty’ or ‘trivial’ vocabulary that Robin Lakoff identified as exclusive to women — words like ‘divine’ and ‘lovely’, or elaborate colour terms such as ‘ivory’ and ‘ecru’. The operators Kira Hall interviewed were clear that women’s language was an essential tool of their trade. Callers responded positively to it, which is to say that they stayed on the line longer when operators used it, thus enabling the operators — who are paid by the minute — to earn more money. Kira Hall did not record ‘live’ calls, but she did sample the pre-recorded fantasy narratives that are available on some sex lines, and she found that what operators said about their style of speech was borne out by these performances.

The women Kira Hall spoke to were essentially ‘passing’: they certainly did not use the same style of speaking in their leisure hours (or in the interview situation) as they did when they were taking calls. In many cases they were also engaged in other kinds of linguistic simulation. Some of them had two or three personas which were racially and ethnically distinct. Exploiting the fact that the telephone is not a visual medium, the same operator might interact with different callers as a Hispanic woman, an Asian woman and a white Anglo woman. Operators also maintained that the ‘best’ Black women were often, in reality, white. What they appeared to mean was that white women were better able (and perhaps less reluctant) than Black women to produce the kind of performance that callers wanted, which was essentially a racist stereotype. Kira Hall even encountered, on a line exclusively for heterosexual men, one gay male operator. Although he was untypical, in the fantasy world of the fantasy lines it would be hard to see him as any less ‘authentic’ than his female co-workers, and the customers apparently found his performances no less satisfactory.

Do the men who use these lines even care whether the person they are paying to arouse them is faking it or not? It is difficult to say, but when Kira Hall revealed to some of her male acquaintances that the ‘sexy schoolgirl’ at the end of the sex line might be a 45-year old washing dishes as she talked, or possibly even a man, they expressed surprise and disappointment.

Subordinated speech

It is hardly news that much of the sex industry depends on, among other things, the capacity of men to delude themselves. What are we to make, though, of the fact that increasingly, women in quite ‘ordinary’ workplaces are facing some of the same linguistic demands as telephone sex workers and MTF transsexuals? A good example is the ‘call centre’, a type of workplace where people are employed exclusively to process telephone calls: operators deal for instance with banking transactions and insurance claims, selling rail and airline tickets, taking catalogue orders and cable/satellite TV subscriptions. By the middle of 1998 Britain had more than 5,000 call centres, and most of their operators were/are women. There is a significant preference for employing young women in call centres, in part because employers perceive them as having the right kind of voice and communication style for the job. ‘They are naturally friendly and can smile down the phone’, as one manager told researchers. Another suggested that women are ‘better [than men] at creating rapport with people’ and ‘more tolerant’ — by which he meant, of rude and abusive callers. It seems then that the featherheaded subservience deplored by Robin Lakoff (and strategically exploited by Kira Hall’s informants) has now become a valuable commodity in the mainstream labour market.

My own investigations of call centres suggest, however, that women’s allegedly ‘natural’ friendliness and tolerance is not taken for granted, but deliberately inculcated and systematically enforced. Operators are given detailed instructions about how to talk — smile, use expressive intonation, display empathy with the caller, ask them ‘open’ questions, sound excited about serving them, never get angry or impatient. New employees practise these ways of talking in role-play exercises during their training. Nor can they slack off as they become more experienced, for surveillance in the average call centre is intense. Supervisors can listen in covertly to calls, and some are also taped so that the operator’s performance can be appraised in more detail later (this is what companies mean by the small print in their advertisements, which announce blandly that ‘calls may be taped for training purposes’). Sometimes market researchers are employed to pose as customers and report back on individual operators. All this means that call centre operators must monitor their speech carefully to ensure it is in line with what the company demands. Failure to conform may result in poor appraisals and lost bonuses; in extreme cases it could even get you fired.

The kind of feminine speech that is valued by call centre managers is undoubtedly artificial, and in future many of the real women who are currently obliged to produce it may be replaced by something entirely artificial — speaking machines. Even now, we hear more and more synthesised speech: in lifts, on London Underground trains, on BT’s call-minding service. Some synthetic voices are more synthetic-sounding than others, but they are all quite clearly gendered, and the differences are instructive — once again, this is not just a question of voice pitch. If you listen to the female ones, you will often hear something eerily reminiscent of the advice to transsexuals I quoted earlier. Warm and smiley, breathy and swoopy, these man-made voices (I use the term advisedly) encapsulate every stereotype of what the ‘natural woman’ should sound like. The synthetic male voices heard in lifts and on trains are nothing like this. They are never breathy or smiley; on the other hand they are often significantly louder.

A friend of mine who works in the field of speech synthesis has a disparaging nickname for the preferred female voice: she calls it ‘sex in a box’. She once explained to me that when she plays her corporate clients a speech sample and asks them what they think, they tend to say things like ‘can you make it softer?’ and ‘maybe a bit less bossy?’ They will not go so far as to use words like ‘available’ and ‘sexy’, but that, my friend believes, is what they actually mean. Thus if she modifies the sample by adding extra breathiness, they respond enthusiastically.

Radical feminists are of course well aware that language is a political issue, but it seems to me that the politics of gendered speech and voice have been neglected by comparison with, say, the issue of sexist language. If I refer in a feminist context to ‘degrading linguistic representations of women’, I am likely to be understood as meaning the use of sexist epithets (‘bitch’, ‘slag’) rather than, for example, the use of smiley voice by call centre operators. Yet the latter is no less degrading than the former; also, and importantly, it is no less a representation. As I have tried to show using various examples, ‘speaking like a woman’ in many contexts is not just a question of doing what comes naturally, or of consciously designing your speech to fit your own ideas about women. Often it involves representing (and thus recycling) someone else’s idea of how a woman ought to speak. And this can be highly oppressive. For example, in one case that I know of — that of Safeway supermarkets in the USA — women workers who were forced to smile, make eye contact and greet customers ‘warmly’ by name complained that their employers’ expectations about their speech and body language were exposing them to continuous sexual harassment.

The coercion of women workers to adopt a subservient way of speaking is a growing trend, and I would like to hear more feminist voices raised in protest against it. Companies that insist on turning their female employees into smiley-voiced fembots in the interests of ‘customer care’ should be told in no uncertain terms that their feminist customers don’t care for it. And just as we deplore the presence of pornographic images in public space, we should also resist the sexualisation of female voices in real or synthesised public announcements. Linguistic representations, no less than visual images, affect the way women are perceived and perceive ourselves. We need to rewrite the script.

Note

[1] All quotations and facts about transsexual and trans-gendered speech in this piece are taken from a review article on ‘Transgender and Language’ in the journal GLQ (1999), written by the anthropologist Don Kulick. Though we don’t agree politically, I am grateful to him for letting me use his work, which so far as I know is the only detailed survey of the literature on this subject. I haven’t reproduced all the references, but if readers want more information they should consult Don Kulick’s bibliography.

References

Kira Hall, ‘Lip service on the fantasy lines’ in Debbie Cameron (ed) The Feminist Critique of Language, 2nd edn. (Routledge, 1998)

Robin Lakoff Language and Woman’s Place (Harper & Row, 1975)