K. A. Cortes, W. P. Cockshott

1  Introduction

In the English speaking world today it is accepted that gender and genders exist. The term is used in news reports and official documents1. Law is passed that incorporates the notion of gender2. Old laws3which prohibited sex discrimination are reinterpreted as prohibiting gender discrimination.

This process has become controversial. Both conservatives and feminists have objected that laws, originally meant to protect women, are in danger of becoming ineffective; if men who claim to be women are treated as legally being women this not only goes beyond the intention of the original law, but it may place women at a disadvantage4. We have strong sympathies with these objections. In this article we will be arguing that genders, in the sense of sets of people, do not exist.

Prior to the late 1960s the word Gender was used in English to refer to sets of nouns and the rules for matching pronouns and adjectives to them in various, mainly foreign, languages. In this usage gender is a French loan, the Anglicization of genre meaning type or class, which the French grammarians had used to classify nouns. This in turn was a borrowing from the Latin grammarians who had used genus in the same sense, as a means of classifying nouns. If one looks at books that use the word gender before the 1970s they overwhelmingly use it in this grammatical sense.

From the end of the 1960s there is a dramatic change in usage. Genders were no longer categories of words but categories of people. Between 1968 and 2000 the usage of the word increased more than 30 fold(Figure 1). It was no longer a grammatical term but a psychological or sociological one. Gender, in its current usages, was a construct of specifically English academic literature. Where once the word sex had been used – the sexes, sex roles, sex differences – now authors wrote of genders, gender identities and gender differences.

The usefrequency of gender overtook sex in academic British English. French and Spanish culture though, seem to have remained more at ease with sex; genre and genero failed to increase relative to sexe and sexo (Figure 2).

If a new technology is invented, the word for it becomes much more common. It may take over an old word in the process. Think of the word computer, whose use grew 500 fold in 50 years. It originally meant a person whose job it was to calculate, and which came to mean a computing machine. The new meaning referred to something new in the world, something which now really existed.

The same thing happens with the words invented or adapted to refer to new scientific discoveries. Cultural fashions like Jazz or Punk, leave their own traces in word usage.

Sex is certainly not a recent discovery or invention, albeit that people did become less inhibited about mentioning it. But did the new use of gender designate a new invention, a scientific discovery, a cultural fashion?

So far we are just counting words. But it does tell us something. Because computers were used internationally, the corresponding word in French ordinateur also showed a similar increase. Similarly when the proton was discovered the word soon cropped up in books in many languages. Cultural fashions however are more nationally specific. This at least suggests that the specifically English use of gender says more about how Anglo-saxon culture interprets the world, than about underlying reality.

Figure 1: The relative frequency of occurrence of the words sex and gender in English language books in the 20th century. Data transcribed from a Google Ngram Viewer query and subsequently re-plotted. Note the phase change around 1970.


To go beyond suggestion we have to look at the actual usage of the word both in the research literature and in philosophy. If the new use of gender is a scientific concept designating something that exists in the world and is distinct from sex, then this would be apparent in research papers. They would show how gender is operationalised into something empirically measurable. In section 2 we examine how leading active researchers use the term gender. This section will show that genders distinct from sexes do not exist as operational concepts.

 Figure 2: French books show no increase in the popularity of genre, over the period the gender became popular in English. Data transcribed from a Google Ngram Viewer query and subsequently re-plotted.

Later we examine the work of Judith Butler who is arguably the leading philosophical advocate of the new usage of the term gender. We will demonstrate that she fails to establish a coherent scientific concept underlying her use of the term.

Finally we discuss the adverse social effect that is produced by a displacement of sex by gender.

Because of the amount of material we have prepared this is serialised over at least three blog posts.

2   Gender in empirical research

To see how the concepts of sex and gender are applied in practical research you have to look at the work of social scientists and psychologists. When do you this, it is problematic just what gender means. There is inconsistency in terminology used in published studies of differences between male and female participants as to whether these differences are called sex differences or gender differences. It does, however, become clear that empirical researchers do not talk of ‘genders’ or use ‘gender’ as a noun. In section 2.10 we will examine why this is the case.

2.1  Method

Search was performed on Google Scholar, for papers with an exact match in the title for the phrase “gender differences in” on the 27th May 2017. Out of the top 20 papers, which had been ranked by citations, downloads were made of all papers that had direct links to online copies. This yielded 13 papers: Blau and Kahn, [2000,Costa Jr et al., [2001,Gefen and Straub, [1997,Byrnes et al., [1999,Feingold, [1994,Croson and Gneezy, [2009,Hyde and Linn, [1988,Nolen-Hoeksema, [2001,Hankin et al., [1998,Piccinelli and Wilkinson, [2000,Venkatesh et al., [2000,Oliver and Hyde, [1993]. The most cited paper Croson and Gneezy, [2009] had at that date 2894 citations and leastVenkatesh et al., [2000] had 958.

Papers were read to determine if the studies they reported actually made any attempt to distinguish sex from gender in their experimental method and if so what approach they took; how they used the term gender; how their methodology and mode of argument compared with that of Butler examined in subsequent sections.

2.2  Gender differences in risk taking

Croson and Gneezy [2009] (2895 citations) present a meta-analysis or review study of 141 other papers. Nearly all of the papers they summarize, simply categorize the participants into into male female with no operational attempt to distinguish sex from gender. The focus of the whole papaer is on risk taking and examines the evolutionary hypothesis that males are less risk averse. If we take sex as being something biologically determined, then the basic hypothesis being tested is a biological one, so it is unclear why the authors chose to describe it in their title as a study of gender differences. Of the studies that Croson and Gneezy [2009] analyze, only a few a few, 3 out of 141, attempt to control for socialization by selecting children who hypothetically are less socialized. As such it is suggested that this method may make it easier to detect sex differences as opposed to socialized gender differences. Croson and Gneezy [2009] also mention cross cultural studies as another having some limited ability to control for nature versus nurture issues. That Croson and Gneezy [2009] explicitly point this out, indicates an awareness that there is an issue of separating out socialisation from biology. There is at least an awareness that some form of compensation should be carried out, but the language used to talk about the problem is not sex versus gender but nature versus nurture.

This is the most highly cited paper on Google Scholar about gender differences in specific domains. Rather than testing a hypothesis about gender differences in the specialized sense used in gender studies it is actually a paper testing a hypothesis about sex differences. In turn it reviews a 141 other papers of which only a small minority made even a partial attempt to compensate for socialization.

Another very big meta analysis of risk taking in men and women is given by Byrnes et al. [1999] (2204 citations ). The procedure was an initial search for papers the query : ( ‘risk’ or ‘risk taking’) and ( ‘gender differences’ or ‘sex differences’). They read over 300 publications and systematically coded and analyzed 150 of them. Overall they found

At a general level, our results clearly support the idea that male participants are more likely to take risks than female participants. In nearly every case, the mean effect size for a given type of risk taking was significantly greater than zero,

Although some of the papers they used referred to sex differences Byrnes et al consistently use the term gender difference. Gender is only used as an adjective as in ‘gender gap’ or ‘gender difference’. Neither gender nor sex are used as nouns except for when sex is used to mean sexual intercourse as in ‘unprotected sex’ is one of the risky activities coded for.

2.3  Gender differences in verbal skills – do they exist

It is widely believed that one difference between males and females is that girls and women have better verbal skills than boys and men. Hyde and Linn [1988] (2064 citations) present a large meta-analysis addressing this topic. They considered 165 studies representing a total sampled population of 1,418,899.

They report, that up to the time of their publication, the superior verbal ability of females, had, for decades been accepted as a fact in psychology. Hyde et al say that previous reviews of the experimental literature were too unsystematic and small – they covered too few studies – to allow definite conclusions to be drawn.

Unweighted averaging the measured gender difference across all studies using the d statistic5 showed a small positive d that is to say a small superiority in female verbal ability. If the results are weighted by sample size, so that larger studies weigh more, the result was a small male superiority. The difference was accounted for by the single largest study – with a sample size of 977,361. If this single large study is excluded, reducing the sample population to 441,538, the measured effect is d=0.11, a female superiority in verbal ability equal to one-tenth of the within-sex variations.

They then broke down the studies by a number of categories to see if this influenced the measured gender differences. The first category used was the selectivity of the sample – did it, for example, use the general population, did it use college students, did it use students at elite colleges. They find that unselective studies generally showed no significant effect, whereas more selective studies did show an effect. Looking at the year of publication, they found that older studies tended to show more gender difference than recent ones. The sex of the first author of a study also influenced the reported gender difference, with women authors tending to report greater differences.

Overall they conclude that :

We are prepared to assert that there are no gender differences in verbal ability, at least at this time, in this culture, in the standard ways that verbal ability has been measured. … A gender difference of one-tenth of a standard deviation is scarcely one that deserves continued attention in theory, research, or ,textbooks.

They argue that such a small effect has no meaningful educational or psychological implications. The absence of a significant female superiority in verbal ability undermines theories of a female brain left lateralised for verbal tasks. Overall they conclude that gender differences in cognitive abilities are nonexistent and that other explanations must be sought for the real gender differences in earnings.

Within the paper, the term gender is used only in the context of ‘gender difference’. Genders are not used as explanatory variables, and the categorisation of the samples is into male and female rather than the gender theory terms masculine and feminine.

2.4  Gender differences in personality and across cultures

Costa Jr et al. [2001] (2054 citations) are concerned to investigate whether gender differences are consistent across cultures. They point out that there are consistent and measurable differences in terms of mean personality traits between samples of men and women and that two broad types of explanation have been given for these.

Two classes of theories, biological and social psychological, have tried to explain these gender differences in personality traits. The biological theories consider sex-related differences as arising from innate temperamental differences between the sexes, evolved by natural selection.


Social psychological theorists argue for more proximal and direct causes of gender differences. The social role model [Eagly, 1987] explains that most gender differences result from the adoption of gender roles, which define appropriate conduct for men and women. Gender roles are shared expectations of men’s and women’s attributes and social behavior, and are internalized early in development. There is considerable controversy over whether gender roles are purely cultural creations or whether they reflect preexisting and natural differences between the sexes in abilities and predispositions [Eagly and Wood, 1999,Geary, 1999].

The authors argue that one could in principle use data on cross-cultural gender differences to distinguish between the two hypotheses. Biological explanations would predict that gendered personality traits would be consistent across cultures. A finding that gendered personality differences showed no consistency across cultures would undermine the plausibility of a biological cause.

The study involved measuring 30 personality traits for samples of men and women from 26 different countries, with over 11,000 participants. The original dimensions were sub traits of Neuroticism, Extraversion, Openness to Experience, Agreeableness and Conscientiousness. Dimension reduction was performed from the 30 traits of the initial survey onto a 4-dimensional subspace, each dimension of which, when averaged across all cultures, polarized the sample between men and women.

It produced the striking finding that there was a indeed very large variation in gender differences between cultures. Among black Africans there was almost no measurable difference in mean personality between men and women in terms of these traits. For European and American populations there were strong differences between male and female personalities. For South and East Asian populations the differences, whilst detectable, were markedly less than those found in the West. Overall the authors report that their results are incompatible with gendered differences in personality being due to a mainly biological cause.

If we examine the variables used in Costa et al we see that sex and country are the independent variables, and personality traits the dependent ones. The authors consistently refer to the ‘two sexes’, not the two genders. Gender is only talked about in terms of gender differences in personality, which is operationally defined as the women’s average for a trait minus the men’s average for the same trait. Practically this makes a lot of sense. It is almost impossible to treat gender as an independent variable. To do so one would have to decide that a particular set of scores on personality traits would mark one of the experimental subjects as male, and another set of scores would mark them as female – presumably using Support Vector Machines or some other technique of linear discriminant analysis. But that would have put them in a position where they would have been unable to measure cross-cultural differences in these traits. The very attributes which may be specific to Western culture would then be used to attempt to discriminate male from female in Africa.

Costa et al not only provide strong evidence for the cultural production of gender differences, but also demonstrate the futility of assuming that gender itself can be used as an independently existing explanatory factor. Again researchers have to fall back on biological sex as the independent variable, even when demonstrating that gender differences cannot have a simple biological cause.

They do not use gender as a noun but consistently as an adjective: gender difference, gender role.

Feingold [1994](1954 citations) is also about gender differences in personality. Feingold goes deeper into theoretical explanations of gender differences in personality than Costa et al., distinguishing, biological, socio-cultural and bio-social theories. Among the socio-cultural theories discussed is the possibility that apparent gender differences may be an artefact of measurement techniques. Measurement is generally by self-reporting using standardised questionnaires.

The artifact model posits that sociocultural factors (e.g., gender stereotyping) result in men and women holding different values about the importance of possessing various traits and that these differences differentially bias self-reports of personality characteristics, engendering sex differences in scores on personality inventory traits that do not reflect corresponding sex differences in the personality constructs that the tests purport to measure.

One might argue that Feingold’s artefact theory since it attributes measure differences to bias in self-assessment, may fit in with Butler’s notion of gender being performative.

Feingold also discusses biosocial theories according to which current gender differences are historic relics of a pre-industrial economy in which the biological differences between the sexes had more of an impact:

The hypothesis that gender differences have both proximal and distal causes is plausible because social roles, based mainly on distribution of work tasks, may have evolved in preindustrial times as a consequence of physical differences between the sexes that were far more consequential then than in the current technological age.

Alternatively, another bio-social theory would be that there is an interaction between direct biological factors and social factors in shaping personality.

The method was a replication, using more sophisticated meta-analysis techniques of Maccoby and Jacklin, [1974], Hall, [1984] along with a meta-analysis of reported intercultural gender differences in personality traits using the same features as in Costa Jr et al., [2001]. The range of cultures studied was narrower than in the latter paper, with fewer Asian and no African studies reported. Unlike Costa et al. no conclusions are drawn about whether the results support biological or social origins of the differences shown. The word gender is only used as an adjective in conjunction with differences. Sex and gender are used as synonyms.

2.5  Gender differences in depression

Piccinelli and Wilkinson [2000] (1561 citations) and Nolen-Hoeksema [2001] (1201 citations)both address the issue of the greater prevalence of depression in women. The conclusions and material covered are similar in both cases. The explanation given is an integrative one in which a higher prevalence of stressors such as childhood sexual abuse and maternal responsibilities interact with dysregulation of the hypothalamic-pituitary-adrenal axis, which the paper claims may be physiologically more common in females. In Nolen-Hoeksema gender is only ever used as an adjective – for instance as ‘gender difference’ in this paper. It never appears as a noun. It uses the term gender consistently when referring to differences between women or men, it does not use the term sex differences, though it cites several papers with the term ‘sex difference’ in their titles. If gender difference is taken to refer to cultural effects, and sex differences to biological ones, then the paper, contra the title, is explaining differences in depression rates as being the result of both gender and sex. But there is no explicit theoretical distinction made between gender and sex. In contrast whilst the overall explanations given for gender differences in depression by Piccinelli and Wilkinson are the same, the latter do use the term gender as noun 12% of the time it occurs. They speak of ‘genders’, and once, use the term ‘gender identity’. There is thus some slight overlap with the language that Butler also uses. However, they are definite about sex being distinct gender identity.

  • Biological sex is an immutable socio-demographic variable not influenced by disease and thus is a useful starting point in the investigation of risk factors for depression.
  • The approach based on biological sex is rarely combined with the study of developmental processes underlying the acquisition of gender identity.

2.6  Gender differences in pay

Blau and Kahn [2000](1373 citations) reviews explanations of the gender difference in pay in the USA when compared to other countries. It has to explain why there are gender differences in pay, and why these differences declined over the period studied: from the late 1970s to the late 1990s. They also examine the narrowing of the gender pay gap in other countries and the extent to which it lower elsewhere. Overall they conclude

compared to women in the other countries, U.S. women are better qualified relative to men and/or encounter less discrimination. The mediocre ranking of the U.S. gender ratio in the face of these favorable gender-specific factors is a consequence of the higher level of wage inequality in the United States, which places a much higher penalty on being below average in the wage distribution.

Their justification for this is that the average female wage put them at the 37th percentile of male wages in the US as against the 32nd percentile for the group of other countries studied.

They examine human capital theory, unionization, having time off for children, and discrimination as explanatory factors. As they explain it is in practice difficult to fully separate out the weights to be assigned to these possible causal factors.

Whatever the adequacy of the explanation given by Blau and Kahn, we can surely agree that the topic is of great importance for the actual living conditions and social position of American women. They refer to these differences as a gender pay gap. This fits with a usage of gender as being anything to do with cultural differences between the sexes. There is no doubt that wages are a cultural institution. But within their argument, they never refer to genders. They do refer to ‘sex segregation of occupations’, ‘sex discrimination’, and ‘both sexes’. Gender itself is not used as an explanatory variable. Nor do the authors treat gender as meaning social classification into ‘masculine’ and ‘feminine’. Insofar as they classify it is by sex. Gender is just used as an adjective when qualifying a difference between the sexes in some measured attribute – pay in this case.

2.7  Gender differences in activity

Trost et al. [2002] (1324 citations) report on a study that used accelerometers worn by children to measure their levels of activity. It produced a robust finding that boys engaged in significantly more Vigorous Personal Activity (VPA) than girls for all school age groups studied. No operational steps are taken to distinguish sex from gender, and there is no mention in the discussion of socialisation as opposed to biological sex being an issue at all. With regard to the procedure and content, the paper could equally well have been titled : Age and sex differences in objectively measured physical activity in youth.

2.8  Gender differences in sexuality

This topic substantially overlaps with that of Butler’s book. The leading paper on it Oliver and Hyde [1993] (1187 citations) gives a particularly clear contrast with Butler’s roughly contemporary book. Butler’s book came out in 1990 and Oliver and Hyde’s meta analysis three years later. The importance of this is that the overwhelming majority of the 177 published papers used by the pair, were extant at the time that Butler wrote her own book.

Oliver and Hyde start out by presenting what they take to be the five main contending theories of sexuality in the literature:

  1. Psychoanalytic theories.
  2. Theories deriving from sociobiology.
  3. Social learning theory.
  4. Social role theory.
  5. Script theory.

This is a much wider range of theories than Butler examines. Even allowing for differences in terminology, Butler looked at the only (1) and (2). Oliver and Hyde are careful, when examining each of these theories, to bring to the fore any testable predictions they make. This is in striking contrast with Butler, who never seems to concern herself with whether the theories she discusses are true, testable, or even meaningful. They say their study is not designed to discriminate between the five theories. This is, in part, because the empirical predictions they all make, are rather similar. The theories

are all in agreement in predicting that females will have a smaller number of sexual partners than will males and that females will have more negative attitudes toward casual, pre-marital sex.

The authors examined papers reporting on 239 samples with 128,363 respondents in all. Data from the studies were transformed into a uniform coding scheme covering 21 different topics ranging from attitudes to premarital sex, homosexuality, the double standard to the frequencies of a range of actual sexual activities. As with many of the other papers we examine, Oliver and Hyde use the d metric in their coding, with, in their case, a positive value indicating a greater frequency of the activity or attitude among males.

Aggregation over all the studies indicated:

  • Negligible gender differences in attitudes to homosexuality.
  • Positive d for the incidence of homosexual activity – meaning it was more frequent in males.
  • More permissive attitudes to premarital and extra-marital sex by males.
  • Men claimed to have heterosexual intercourse more frequently than women. Although Oliver and Hyde do not say so, this must reflect male boasting or female reticence.
  • Men claimed to have lost their virginity earlier. This is possible if a portion of teenage boys has their first intercourse with somewhat older women.
  • Women tended to endorse the double standard more than men did. They were significantly more lenient in their attitudes to male infidelity than men were in their own attitudes to it.
  • The biggest single difference was in the frequency of masturbation, it was a whole standard deviation higher in males.

There was a sufficient number of studies of enough time to allow conclusions to be drawn about changes in gender differences over time. In terms of attitudes, the differences all declined with time. In terms of numbers of sexual partners and frequency of sex, d values also fell. This must reflect more accurate self-reporting with time since real average numbers of partners and frequency must be the same for both genders. Gender difference in masturbation remained high but showed some decline over time. Gender differences in attitudes and reported incidences of sexual activities were more marked in older than younger respondents.

Many of the gender differences in sexuality were quite small, but those with respect to permissive attitudes and masturbation frequency stood out as large. Large, that is, compared to gender differences of other traits obtained by meta-analyses. They are much bigger than differences in verbal or mathematical ability. Whilst the difference in permissiveness aligns with all of the theories reviewed, the big gender difference in masturbation was not predicted by any of them.

As with most of the research papers reviewed here, gender is used exclusively as an adjective, not a noun. Operationally a gender difference is defined by the authors as a  d value defined over biological sex as the independent variable.

2.9  Gender differences in technology use

Venkatesh et al. [2000] (958 citations) explicitly recognise that there may be a conceptual distinction between gender and sex :

it is important to recognize that there are at least two commonly understood definitions of gender in psychology—the first is consistent with biological sex while the second views gender as more of a psychological construct

but go on to say that, in this study, they are working on the basis that gender is defined as biological sex.

The hypothesis that they are testing, that men will be more driven by attitude than women are when adopting new technology. They link the possible importance of attitude to what they call ” “masculine” traits (e.g., assertive), as identified by different inventories including Bem’s Sex Role Inventory[Bem, 1981]”. So the hypothesis is arguably driven by evidence of what would often be called gender traits. Participants were divided into men and women but an attempt was made to remove confounding variables which correlate with gender : income, education and position in the organisational hierarchy. Men are likely to have more money, more education and a higher organizational position. The aim was to test the effects of gender alone by explicitly controlling these variables. They find that after explicitly controlling these variables, the salience of gender as a predictive factor in technology uptake was still strong.

The authors although aware of ambiguity in the term gender treated it to mean biological sex, but did compensate for what could be called gendered social position and their causal model included and compensated for what are sometimes described as gendered personality traits. Although they describe what they term gender as emerging as a strong causative factor, it is arguable that what they are measuring here is biological sex, both because they say that is what they consider gender to be, and also because they are trying to control for typical gender traits or sex-differentiated social position.

The authors use gender both as a noun and as an adjective.

2.10  Discussion

Discussing the use of the terms sex and gender in then recent social science Delphy wrote:

What they never ask is why sex should give rise to any sort of social classification. Even the neutral question “we have here two variables, two distributions, which coincide totally. How can we explain this covariance?” does not get considered.

The response is always: sex comes first, chronologically and hence logically – although it is never explained why this should be so. [Delphy, 1993]

Examination of the actual research procedures used in studies of gender differences provides the answer as to why sex is a logically prior category.

In all of these studies, in order to discover gender differences, the experimental subjects first have to be categorized by sex in order to discover the gender differences in pay, personality etc. Sociologists, economists and psychologists would have been unable to investigate these differences any other way. They do not have any prior objective way of categorizing experimental populations by gender.

Suppose that experimentalists had some objective gender test which they could apply to people they enrol in an experiment or survey. How would this test be performed?

It would be no good asking the subjects if their gender was masculine or feminine. We have no guarantee that the subjects would not just interpret that the same way as asking them if their sex was male or female. Pryzgoda and Chrisler [2000] experimented with such questionnaires, using US college students and other educated Americans, populations exposed to modern gender theory. Whilst the definitions given by the students for sex and gender correlated reasonably well with the current academic use of the terms, there were sufficient inconsistencies in their use of language to indicate that, even with this highly educated group, many would interpret a question about their gender as being the same as a question about their sex.

Since simply asking would not differentiate between sex and gender, the only other way would be to use personality traits, pay rates etc, which we know to be gender differentiated to categorise subjects by gender. Besides being unreliable – given the overlap of personality traits, pay etc, this would not avoid the logical priority of sex. The statistical criteria that would have to be used to categorise by gender, have been established by earlier studies of gender difference that used sex as the independent variable.

It follows from the logic of investigation, that sex has to be prior. We could not even systematically study gender differences without a prior ability to categorise people by sex. Categorisation by gender is non-operational in the experimental sense, in consequence, the empirical studies fall back on categorisation by sex. It is still worth distinguishing differences in traits between male and female populations that are socially rather than biologically determined. In some cases this is easy – nobody doubts that differences in the incidence of dress wearing by women and men are social and not biological, or that height differences are biological. It may be convenient to label the former as a gender difference and the latter as a sex difference. In other cases, it is not so clear, for example, different rates of depression. Here, only extensive research can tease out the relative importance of social circumstance and biology. Until the research has reached a definitive result, calling different depression rates a gender not a sex difference, is more a matter of fashion and caution than rigour.

3  Butler: sex a social construct?

In section 2 have looked at a selection of the most highly cited research literature on gender differences to see how empirical social scientists and psychologists use the concept gender. But the contemporary debate on gender legislation is probably more influenced by philosophy than it is by these empirical studies of measurable differences between men and women. In this section we move on to the ideas of an influential philosopher who has written about gender: Judith Butler. One of the novel and ambitious hypotheses put forward by Butler is that not only gender, but sex itself is a social construct.

Does being female constitute a natural fact or a cultural performance, or is naturalness constituted through discursively constrained performative acts that produce the body through and within the categories of sex? Butler, [1990,page xxviii]

She poses this as a question, with an implied answer that: no being female is not a natural fact. But it is unclear what it could mean to deny that femaleness is a natural fact. Are there not male and female animals?

Sexual dimorphism into male and female bodies is a conserved mammalian trait. So in a temporal sense, it is obviously prior to human society. Female animals existed before there were any humans. If Butler is just saying that being female acquires varying social representation over human history, that would be an uncontroversial point. The same could be said of childhood.

All primates have childhoods during with they grow, play and learn. In our species this is unusually long in purely physical terms – the time it takes to stop growing. But the social category of childhood has a variable duration based on the form of economy. The modern industrial economy is unique in having children as non-productive members of society and extends the duration of childhood as a social or legal category to the very limit of the physical growth period. Clearly, the social definition of child impinges on the social definition of man and woman, changes in the definition vary the number legally constituted women and men, affect age of marriage etc. But we don’t, on the basis of these social variations, decide that childhood is just a social construct. No change in the laws regarding child labour or the laws against paedophilia would be sufficient to make children actually the same as adults in strength or capacities.

Reading on, though, we find that Butler really does doubt that sex is, in both a temporal and ontological sense, pre-social. She asks:

what is sexanyway? Is it natural, anatomical, chromosomal, or hormonal, and how is a feminist critic to assess the scientific discourses which purport to establish such facts for us? Does sex have a history? Does each sex have a different history, or histories? Is there a history of how the duality of sex was established, a genealogy that might expose the binary options as a variable construction? Are the ostensibly natural facts of sex discursively produced by various scientific discourses in the service of other political and social interests?Butler, [1990,Page 10]

This comes close to conspiracy theory: the suggestion that sexes do not exist in nature, and that observations in the biological literature, or for that matter by farmers, that other animals reproduce sexually are just a conspiracy by political and social interests.

The rhetorical technique, asking a set of ‘bold’ questions without producing any evidence or argument to show that the questions are reasonable is certainly familiar. One encounters it among the more outlandish conspiracies propagated on the web, things like: Is Queen Elisabeth really a shape-changing alien? Do the royal family regularly drink the blood of small children sacrificed for them by jewish priests? Was this all foretold in secret scrolls buried beneath the Great Pyramid?

The rhetorical purpose of this technique is to suggest something absurd to the reader without an author having to be pinned down as having actually written it. It is the sort of stuff that fringe websites promote.

Butler then goes on to suggest slightly more strongly that sex is just a social construct:

If the immutable character of sex is contested, perhaps this construct called sex is as culturally constructed as gender; indeed, perhaps it was always already gender Butler, [1990,Page 10]

Scientifically speaking, we know that sex is not culturally constructed. Beyond reproduction, we know that disease presents differently in women and men. This speaks to something other than construction. The thesis that sex is just a social construct is now expressed as a conditional, with the implication that the condition is true, without any attempt to assess the probability or plausibility of the claim. The logical form of her argument is no better than arguing: Since the theory of Darwin is contested, perhaps the whole theory of evolution is a Masonic project, indeed perhaps we should just accept the book of Genesis as literally true

Having thrown doubt on the existence of sex as a natural category Butler then goes on:

It would make no sense, then, to define gender as the cultural interpretation of sex, if sex itself is a gendered category. Gender ought not to be conceived merely as the cultural inscription of meaning on a pregiven sex (a juridical conception); gender must also designate the very apparatus of production whereby the sexes themselves are established. As a result, gender is not to culture as sex is to nature; gender is also the discursive/cultural means by which sexed nature or a natural sex is produced and established as prediscursive,prior to culture, Butler, [1990,Page 11] 

She goes from a conditional formulation “ if sex itself is a gendered category” to an unqualified conclusion that states that sex is the result of gender. Just what is meant by this is not spelled out.

The only way to interpret her text that is compatible with what we know from science would be as follows. Butler writes “gender must also designate the very apparatus of production whereby the sexes themselves are established”. The ‘apparatus of production’ of the sexes is the set of genes whose differential activation produces sexual dimorphism during the growth of the individual, along with some triggering mechanism. The cascade may be triggered by among other things: the presence or absence of Y chromosomes in mammals; by temperature in crocodiles; being homozygous at the sex determining locus in bees[Gempe et al., 2009]. In this literal interpretation of the text, gender becomes the biological processes which give rise to adult males and females. Gender becomes nest temperature plus suchian genetics, Y chromosomes plus mammalian genetics. Gender = genotype, sex = phenotype. But from context, we can be sure that she did not mean this.

Was it cultural gender differences that have led biologists to say that all placental mammals share a set of defining characteristics which include reproduction by internal fertilization, gestation and lactation?

Was it gender ideology that led farmers to believe that cattle reproduce sexually?

Would cows, left to themselves without bulls, reproduce by parthenogenesis like aphids?

All of these are absurd questions, but they do no more than cast her own questions in a more concrete form. Butler comes close to suggesting that the entire corpus of the biological sciences insofar as they relate to sexual reproduction are a fabrication.

This is an extraordinary claim. Extraordinary claims, if they are to be plausible, need correspondingly strong evidence. If she has observations of, or experimental evidence for, the prevalence of asexual reproduction in mammals she should have cited it. Failing that, her hypothesis must be treated as a particularly outlandish conspiracy theory.

Insofar as she produces any support for sex being a social construct it is by citing not biologists, but by imputing it as the conclusion of the views of another philosopher:

Beauvoir is clear that one becomes a woman, but always under a cultural compulsion to become one. And clearly, the compulsion does not come from sex. There is nothing in her account that guarantees that the one who becomes a woman is necessarily female. If the body is a situation, as she claims, there is no recourse to a body that has not always already been interpreted by cultural meanings; hence, sex could not qualify as a prediscursive anatomical facticity. Indeed, sex, by definition, will be shown to have been gender all along Butler, [1990,Page 12]

or earlier in Butler [1986] she wrote

if the distinction (sex to gender WPC & KDC) is consistently applied, it becomes unclear whether being a given sex has any necessary consequence for becoming a given gender. The presumption of a causal or mimetic relation between sex and gender is undermined. If being a woman is one cultural interpretation of being female, and if that interpretation is in no way necessitated by being female, then it appears that the female body is the arbitrary locus of the gender ‘woman’, and there is no reason to preclude the possibility of that body becoming the locus of other constructions of gender

Butler directly deduces female erasure in a few lines going from Beauvoir’s point about not being born a woman. But just taken by itself the sentence from Beauvoir would not logically allow this extension.

In that sequence, you certainly can not conclude that the set of women is disconnected from the set of female adults.

Butler grants that Beauvoir says that women become women under cultural compulsion. But de Beauvoir is assuming, reasonably enough that it is female children who are under this social compulsion. The fact that de Beauvoir does not provide a guarantee of this is irrelevant. De Beauvoir should not be obliged to prove obvious and uncontroversial points. Suppose, counterfactually, that social customs in late 19th century France were unchanged in all respects other than one thing, that gender was unconnected with sex. The chances of a baby boy being socialised, and legally categorised as a female would then be the same as that of a baby girl. Given that marriage was the predominant state of French adults, and that marriages would still have to be between legal men and legal women, half of all marriages would end up being same sex, even if legally they were classified as being between opposite genders. In consequence, the birthrate would have fallen by almost half, followed, in a couple of generations, by the extinction of the French nation at the hands of an Eastern neighbour. This is so obvious that de Beauvoir hardly needed to point it out. De Beauvoir was describing forms of cultural compulsion that had existed for centuries.

On ne naı̂t pas femme : on le devient. Aucun destin biologique, psychique,
économique ne définit la figure que revêt au sein de la société la femelle humaine
; c’est l’ensemble de la civilisation qui élabore ce produit intermédiaire entre le
mâle et le castrat qu’on qualifie de féminin. .6[Beauvoir, 1949]

Any long-standing cultural institution let alone, ” l’ensemble de la civilisation “7, has to be compatible with the survival and reproduction of that culture.

But even if we take Butler’s logic explicitly :

If being a woman is one cultural interpretation of being female, and if that interpretation is in no way necessitated by being female, then it appears that the female body is the arbitrary locus of the gender ‘woman’,

In formal logic we can write “If being a woman is one cultural interpretation of being female” as

(x ∈ F) & A → x ∈ W

In words If Judith is a female and some other fact A, the cultural interpretation, is true then Judith is a woman We  can formalise the second proposition  that interpretation is in no way necessitated by being female as

∃y 😦 y ∈ F) & (y ∉ W)

In words  there exists at least one female person, let us say Sheila, who is not a woman.

These conditions would be true if Sheila was aged 2. You certainly can not conclude that “the female body is the arbitrary locus of the gender ‘woman'”

∃z 😦 z ∈ W) & (z ∉ F)

Or in words, Butler implies: “There exists a person, let us call them Zebedee, who is a woman but who is not female.”

It remains an invalid argument. A formally equivalent argument would be to say that:”if being a whale is one evolutionary outcome of being a mammal, and if being a whale is in no way necessitated by being a mammal , then it appears that being a mammal is entirely arbitrary to being a whale, so sharks can also be whales.”

All that you can logically conclude from the premises she puts forward is the trivial point that women are a subset of females. This can be taken both in terms of ontogenesis and historical genesis. Women are a subset of females since baby girls are not women, and women in their current cultural role are historically a subset of all women who have existed since women in other societies had different cultural social roles.

Note the reference Butler makes to the impossibility of sex existing “pre-discursively”. This is part of a tendency she has to attribute considerable causal powers to language, and which involves something of a disregard for our other senses, particularly sight. It is, after all, mainly by looking at people that we tell what sex they are. It is only in contexts where sight does not apply, e.g. filling in official forms, that it is necessary to use language and ask someone what sex they are.

The claim, by Butler, that sex can not be known about or, even exist, prediscursively is testable. An elegant study [Lederberg et al., 1986] proves conclusively that Butler is wrong. Were the Butler hypothesis true, deaf children would be unable to distinguish boys from girls. The researchers monitored deaf and hearing children aged 3 to 5 in terms of their preferences for same sex versus opposite sex playmates. The children were observed at playgroups and the time spent playing with same and opposite gender children were recorded. Both hearing and deaf children showed a clear preference for same-sex play partners.

It may be objected that some of the deaf children may have had some sign language ability and this may have been a means by which information about sex or gender categories was communicated to the infants, in which case Butler’s theory that the perception of sex depends of language might still have some credibility, but no. First very few of the deaf children or their parents had any signing ability, and when this was tested for, it was found that sign language ability of the deaf children and their parents had no detectable influence on their preference for same-sex playmates.

So the ability to perceive sex differences and to identify one’s own sex must indeed be prediscursive and mediated by visual clues rather than by language.

There is a certain lack of caution in Butler’s claims. It is very risky indeed to make claims about the world, the only support for which are your own inferences from the work of another philosopher. It is somewhat safer to follow through the implications of your, hypotheses to see whether empirical data support or contradict them.

Other examples of incautiously stating speculative hypotheses as established facts:

It would be wrong to think that the discussion of identity ought to proceed prior to a discussion of gender identity for the simple reason that persons only become intelligible through becoming gendered in conformity with recognizable standards of gender intelligibility.

It is unclear how one could even formulate the claim in a way that was empirically testable. Perhaps one could run a variant of the Lederberg et al study in which one controlled for clothing and hairstyle as a possible “recognizable standard of gender intelligibility”. If Butler is right then having children play naked with short hair would blind the children to the sex of their playmates. Unless such an experiment, or a similarly controlled one has been done, we can not know if Butler’ claim is true or false, but it is rash to state it as a fact.

The body as a natural fact never really exists within human experience, but only has meaning as a state which has been overcome.

Tell that to the hungry! Has this woman never broken a bone?

Butler makes repeated reference to the premise that gender or sex are only recognisable as an effect of language, or discursive practice as she calls it. On page 24 we have:

If identity is an effect of discursive practices, to what extent is gender identity, construed as a relationship among sex, gender, sexual practice, and desire, the effect of a regulatory practice that can be identified as compulsory heterosexuality?

Here we see the typical technique of asking a question in order to suggest a conclusion for which no supporting evidence is adduced with the whole shrouded in ambiguity and qualification. What is the purpose of the quotation marks?

They appear to be there to allow a carefully tailored ambiguity. But let us trace down the ambiguous meanings and see if they make sense.

Proposition :   Identity is an effect of language

(a) Identity as in a person’s name.

Well yes, language is necessary for naming someone.

(b) Identity as in gender identity.

No, Lederberg et al show that it is not dependent on language. The toddlers were able to match their own genders with those of their playmates even though they had no access to language.

Answers to the whole question  :

Taking meaning (a) the premise is true, no sensible answer can be given since there is no connection between the true proposition and the consequence allegedly drawn from it.

Taking meaning (b) since the premise is false, we can again draw no conclusion from it, though the fact that very young children with no language can identify their own gender we can rephrase it as: Given that gender identification including self-identification can occur a very young age, and without the use of language, is gender identification the effect of social regulations relating heterosexuality?

And the answer is that it would be very rash to conclude that it is.

The whole book is so shot through with these sorts of ambiguity, rhetoric, and logical errors that it would be tedious to reviewer and reader to list them all. In our next blog post we will go on to some implications of Butler’s philosophy.


[Beauvoir 1949]
Simone de Beauvoir. Le deuxième sexe ii, paris. Gallimard (coll.Folio Essais), page 13, 1949.

[Bem 1981]
SL Bem. A manual for the bem sex role inventory. Palo Alto, CA: Mind Garden. Blake, RR, & Mouton, JS (1978). The new managerial grid Houston: Gulf. Cann, A., & Siegfried, W D.(1987). Sex stereotypes and the leadership role. Sex Roles, 17: 401-408, 1981.

[Blau and Kahn 2000]
Francine D Blau and Lawrence M Kahn. Gender differences in pay. Technical report, National bureau of economic research, 2000.

[Butler 1986]
Judith Butler. Sex and gender in Simone de Beauvoir’s Second Sex. Yale French Studies, (72): 35-49, 1986.

[Butler 1990]
Judith Butler. Gender Trouble: Feminism and the Subversion of Identity. Routledge, 1990.

[Byrnes et al. 1999]
James P Byrnes, David C Miller, and William D Schafer. Gender differences in risk taking: A meta-analysis., 1999.

[Costa Jr et al. 2001]
Paul Costa Jr, Antonio Terracciano, and Robert R McCrae. Gender differences in personality traits across cultures: robust and surprising findings. Journai of Personality and Social Psychology, 81 (2): 322-331, 2001.

[Croson and Gneezy 2009]
Rachel Croson and Uri Gneezy. Gender differences in preferences. Journal of Economic literature, 47 (2): 448-474, 2009.

[Delphy 1993]
Christine Delphy. Rethinking sex and gender. In Women’s Studies International Forum, volume 16, pages 1-9. Elsevier, 1993.

[Eagly 1987]
Alice H Eagly. Sex differences in social behavior: A social-role interpretation. Lawrence Earlbaum, 1987.

[Eagly and Wood 1999]
Alice H Eagly and Wendy Wood. The origins of sex differences in human behavior: Evolved dispositions versus social roles. American psychologist, 54 (6): 408, 1999.

[Feingold 1994]
Alan Feingold. Gender differences in personality: a meta-analysis. Psychological bulletin, 116 (3): 429, 1994.

[Geary 1999]
David C Geary. Evolution and developmental sex differences. Current Directions in Psychological Science, 8 (4): 115-120, 1999.

[Gefen and Straub 1997]
David Gefen and Detmar W Straub. Gender differences in the perception and use of e-mail: An extension to the technology acceptance model. MIS quarterly, pages 389-400, 1997.

[Gempe et al. 2009]
Tanja Gempe, Martin Hasselmann, Morten Schiøtt, Gerd Hause, Marianne Otte, and Martin Beye. Sex determination in honeybees: two separate mechanisms induce and maintain the female pathway.PLoS biology, 7 (10): e1000222, 2009.

[Hall 1984]
JA Hall. Nonverbal sex differences: Communication accuracy and expressive style. Johns Hopkins Univ Press, Baltimore, 1984.

[Hankin et al. 1998]
Benjamin L Hankin, Lyn Y Abramson, Terrie E Moffitt, Phil A Silva, Rob McGee, and Kathryn E Angell. Development of depression from preadolescence to young adulthood: emerging gender differences in a 10-year longitudinal study. Journal of abnormal psychology, 107 (1): 128, 1998.

[Hyde and Linn 1988]
Janet S Hyde and Marcia C Linn. Gender differences in verbal ability: A meta-analysis., 1988.

[Lederberg et al. 1986]
Amy R Lederberg, Steven L Chapin, Victor Rosenblatt, and Deborah Lowe Vandell. Ethnic, gender, and age preferences among deaf and hearing preschool peers. Child Development, pages 375-386, 1986.

[Maccoby and Jacklin 1974]
Eleanor E Maccoby and Carol Nagy Jacklin. The psychology of sex differences, volume 1. Stanford University Press, 1974.

[Nolen-Hoeksema 2001]
Susan Nolen-Hoeksema. Gender differences in depression. Current directions in psychological science, 10 (5): 173-176, 2001.

[Oliver and Hyde 1993]
Mary B Oliver and Janet S Hyde. Gender differences in sexuality: A meta-analysis., 1993.

[Piccinelli and Wilkinson 2000]
Marco Piccinelli and Greg Wilkinson. Gender differences in depression. The British Journal of Psychiatry, 177 (6): 486-492, 2000.

[Pryzgoda and Chrisler 2000]
Jayde Pryzgoda and Joan C Chrisler. Definitions of gender and sex: The subtleties of meaning. Sex Roles, 43 (7-8): 553-569, 2000.

[Restaut MDCCXLI]
Pierre Restaut. Principes généraux et raisonnés de la grammaire françoise. MDCCXLI.

[Trost et al. 2002]
Stewart G Trost, Russell R Pate, James F Sallis, Patty S Freedson, Wendell C Taylor, Marsha Dowda, and John Sirard. Age and gender differences in objectively measured physical activity in youth.Medicine and science in sports and exercise, 34 (2): 350-355, 2002.

[Venkatesh et al. 2000]
Viswanath Venkatesh, Michael G Morris, and Phillip L Ackerman. A longitudinal field investigation of gender differences in individual technology adoption decision-making processes. Organizational behavior and human decision processes, 83 (1): 33-60, 2000.


1For example on the World Health Organisation website, we find a definition of gender as “the socially constructed characteristics of women and men – such as norms, roles and relationships of and between groups of women and men”. Attempts to get the equivalent pages in French or Spanish return the messages “Cette page n’est pas disponible en francais” and Ësta pagina no esta disponible en castellano”. we will return to this point later.

2 For example Statues of Canada, Chapter 13, An Act to amend the Canadian Human Rights Act and the Criminal Code, Bill C-16, June 19, 2017

3For example Title VII of the US Civil Rights Act of 1964.

4Examples cited are the risk to women in prison if male sex-offenders are able to be reassigned to such prisons after taking on a female persona or unfair competition in women’s sports.

5 This is widely used in gender difference studies it is given by

d= ef −em

σf σm

where ef, em denote measured effect in females and males and σf, σm denote the within sex standard deviations of the effect.

6Öne is not born, but rather becomes, woman. No biological, psychic, or economic destiny defines the figure that the human female takes on in society; it is civilization as a whole that elaborates this intermediary product between the male and the eunuch that is called feminine.”

7“civilization as a whole”

File translated from TEX by TTH, version 4.08.