Social constructionist claims are surprising and interesting when they entail that presumably natural kinds are in fact socially constructed. The claims are interesting because of their theoretical and political importance. Authors like Díaz-León argue that constitutive social construction is more relevant for achieving social justice than causal social construction. This paper challenges this claim. Assuming there are socially salient groups that are discriminated against, the paper presents a dilemma: if there were no constitutively constructed social kinds, the causes of the discrimination of existing social groups would have to be addressed, and understanding causal social construction would be relevant to achieve social justice. On the other hand, not all possible constitutively socially constructed kinds are actual social kinds. If an existing social group is constitutively constructed as a social kind K, the fact that it actually exists as a K has social causes. Again, causal social construction is relevant. The paper argues that (i) for any actual social kind X, if X is constitutively socially constructed as K, then it is also causally socially constructed; and (ii) causal social construction is at least as relevant as constitutive social construction for concerns of social justice. For illustration, I draw upon two phenomena that are presumed to contribute towards the discrimination of women: (i) the poor performance effects of stereotype threat, and (ii) the silencing effects of gendered language use.
The social sciences and humanities have claimed that various categories of things and people, events and properties, are socially constructed. For instance, it has been claimed that gender, race, sex, sexual orientation, and several psychological disorders are socially constructed and not natural categories. The claim that a given social kind is socially constructed has allowed people to be more aware that members of certain socially salient groups suffer from discrimination – i.e. unfair differential treatment qua members of the group –, and contributed to open avenues for social change. Social constructionist claims played an important part in demands for social change by supporting attempts to resist the idea that a category is natural.
Hacking (1999) summarizes the standard claims of constructionists thus:
X need not have existed, or need not be as it is: it is not determined by the nature of things, or is not inevitable.
X is bad as it is.
It would be better if X were transformed or done away with. (Hacking 1999, p. 6)
The three claims that according to Hacking characterize social construction are compatible with different senses of social construction. Authors like Haslanger (2003, 2006, 2012), Mallon (2007, 2009), Díaz-León (2015), and Sveinsdóttir (2013, 2015), among others, offer accounts of what it means for something to be socially constructed. A central distinction is, on the one hand, between the social construction of representations (ideas, concepts, predicates, etc.) and the social construction of objects (people, categories, events, properties, etc.) On the other hand, there’s a distinction between object construction in the causal sense, and object construction in the constitutive sense. For the purposes of the present paper, I will follow others and focus mostly on the objectual sense of social construction.
As Haslanger (2003) says, there are two sets of questions that need to be addressed and distinguished when we discuss social constructionist claims.
The first is whether employing a classification C… is theoretically or politically useful. The second is whether the theoretical understanding of C captures an ordinary social category, and so whether it is legitimate or warranted to claim that the proposed definitions reveal the commitments of our ordinary discourse. (Haslanger 2003, p. 319)
…the kind of social construction that is more closely relevant to the goal of achieving social justice by means of social action is the constitutive social construction of a certain human kind, because this claim will have clearer social and political significance concerning the feasibility of different social strategies for achieving social change. (Díaz-León 2015, p. 1148)
The claim that something is socially constructed, and that unfair discrimination of members of certain groups sometimes results from taking socially constructed categories as natural, has played a central role in efforts to effect social change. The question naturally arises to those interested in overcoming injustice: given the alternative ways of understanding constructionism, which of those ways (causal vs. constitutive) is more relevant for the political goals of achieving social change? If the argument of this paper holds, the answer to this question is “both”. This paper disputes Díaz-León (2015)’s contention, and argues that causal social construction is at least as relevant socially and politically as constitutive construction to achieve social change.
The aim of this paper is to address what I think is a misunderstanding about the relation between causally constructed kinds and constitutively constructed kinds, and to criticize the idea that a focus on constitutively constructed kinds should take priority over a focus on actual causal social relations. Part of the misconception is to think of social causes as mere causes of an individual’s instantiation of a (possibly intrinsic) property.  This ignores a fundamental metaphysical question concerning constitutive social kinds and their contingency on social reality: Why does a kind F exist instead of a kind K, given that there are indefinitely many possible constitutive kinds, and that not all have actual instances? Understanding why a constitutively constructed kind exists instead of another requires that we focus on the social causes of its existence. This is required by a proper development of the idea that socially constructed things are not inevitable. 
Section 2 briefly summarizes Díaz-León’s (2015) reasons for thinking otherwise. Section 3 draws from linguistic and psychological empirical research to illustrate the distinct causal effects of stereotypes and linguistic practices. It focuses on two (presumed) causes of discrimination on account of gender: (i) the effects of stereotype threat on girls’ mathematical performance, (ii) the social effects of silencing among professional women. If these are real discriminatory effects, they are unfair even if gender is not constitutively socially constructed. Any effort to overcome unfair discrimination would have to come to terms with its social causes. If there are actual constitutively constructed social kinds, on the other hand, they only exist because there are social relations and causes that keep them in existence. Section 4 makes the broader metaphysical point that any constitutively constructed kind is also causally constructed. Finally, Section 5 concludes that causal social construction is at least as relevant as constitutive social construction to address injustices through social action.
2 Kinds of Construction, Political Significance, and Feasibility
De Beauvoir (1952) in her The Second Sex, said “One is not born, but becomes a woman.” Feminists in the 1960s and -70s insisted that gender and sex are not the same. Sex would be a biological category and gender a socially constructed one. If gender is socially constructed, then, it was claimed, the norms and expectations concerning one’s gender could be changed. Feminists drew on anthropological research to reveal the variability of some gendered traits, expectations, and norms across societies and history. Distinguishing sex and gender had political significance, one that allowed activists to argue against the inevitability of social conventions and practices that are discriminatory towards women.
We may ask: are socially constructed kinds, as gender arguably is, constitutively constructed or causally constructed? Haslanger (2003) offered two schematic definitions of causal and constitutive social construction:
X is socially constructed causally as an F iff social factors (i.e. X’s participation in a social matrix) play a significant role in X having those features by virtue of which it counts as an F.
X is socially constructed constitutively as an F iff X is of a kind or sort F such that in defining what it is to be F we must make reference to social factors (or: such that in order for X to be F, X must exist within a social matrix that constitutes F’s). (Haslanger 2003, p. 317–318)
But we should avoid conflating social kinds with things that have social causes. Sociobiologists claim that some social phenomena have biological causes; some feminists claim that some anatomical phenomena have social causes, for example, that height and strength differences between the sexes are caused by a long history of gender norms concerning food and exercise. It is an error to treat the conditions by virtue of which a social entity exists as causing the entity. (Haslanger 2003, p. 317)
Some social phenomena may have natural (biological) causes, and some biological phenomena (physical or anatomical traits) may have social causes. This would seem to disprove the claim I intend to make in this paper at the outset, which is: 
For any actual kind X, if X is socially constructed in the constitutive sense as a K, then X is socially constructed in the causal sense as a K.
In line with Haslanger’s point above, Díaz-León (2015) argues that constitutive social construction is “more closely relevant” to achieving social justice than causal social construction. Constitutive social construction is both more politically significant, and addressing it through social action is more feasible, she claims.
Díaz-León (2015) begins by drawing on examples from Mallon (2009) and Haslanger (2003) to the effect that some trait having social causes is neither necessary nor sufficient for it to be constitutively socially constructed. Mallon claims that although artifacts like watches actually have social causes, it is metaphysically possible that an item physically identical to a watch exists, although it was not created by anyone (Mallon 2009, p. 8). Hence, it is possible that an artifact actually has a social cause, but it could still be what it is if there were no social causes around, or so Mallon claims.
Haslanger (2003, 2006, 2012), on the other hand, holds that constitutively socially constructed kinds are so in virtue of their social role, and the social relations that define them. For instance, what makes someone a husband is being married to a woman, although being married to a woman does not cause one to be a husband.
I find both of these examples surprising. The first claim, that an artifact is an object that “merely” has social causes seems to be mistaken. Artifacts are paradigmatic constitutively socially constructed objects. The Oxford English Dictionary defines artifacts as objects made by human beings (i.e. that are causally constructed), and that typically have historical or cultural significance (which is not reducible to having social causes). It is the function or role of an object qua artifact, and not just its origins, that makes it what it is. It may be metaphysically possible that an item that is indistinguishable from actual artifacts exist without it being an artifact. In that case, that item would be deprived of the relevant cultural and historical significance – it would not be an artifact.  But there can be no artifacts – no products – unless there exist the producers and the action of production of the relevant product. No producers, no products; no artisans, no artifacts.
Yet, and even though in my view Mallon’s claim about artifacts is mistaken, it is not relevant for the main thesis I am making in this paper. The thesis I am arguing for is that if something is constitutively constructed socially, then it is also causally socially constructed. It could well be that there are things that are causally constructed but not constitutively constructed. In fact, I do not want to take a stand on whether some of the socially salient groups of people that have been claimed to be constitutively constructed social kinds are just causally constructed.  Haslanger appeared to refute the claim I intend to defend. It is true that what makes someone a husband is the kind of social role and relations that allow one to be a husband. These are not what cause someone to be a husband. But my claim (as I explain in more detail below, Section 4) is not that if S is among the social factors constituting the Fs, then S is also a causal factors for instances of F to instantiate S. My claim is rather that causal factors explain why there are instances of Fs at all.
Imagine a possible world where the number of male humans at least doubles the number of female humans, and where it is not the case that each man is guaranteed the possibility of marrying one female partner. In that world, societies implement a different sort of marriage institution, call it a shmarriage. In that world, men are willing to marry a woman and other men, accepting the possibility of indeterminate paternity of offspring. This is because the odds of having some offspring are higher under this arrangement than under other arrangements (on the imagined situation). The fact that there are shmhusbands in this world – a possible social kind that consists in being married to a woman and at least one other man, instead of husbands, is a result of the social causes that led to the existence of the schmarriage institution. Had the relevant social causes been absent (the arrival at an agreement, implementing the institution or custom, etc.), there would have been no shmhusbands, since the social institution that supports them would not exist.
Díaz-León’s (2015) point, however, is that there are important differences between social construction in the constitutive and in the causal senses with respect to which kind of social construction has greater political significance and allows for more feasible actions to achieve social justice. The argument for this is supported by three considerations. The first concerns “the project of arguing against the inevitability of a trait” (Díaz-León 2015, p. 1137). The second consideration concerns whether said traits are relational or intrinsic properties. The final consideration is that the difference affects the “project of arguing that a certain trait is not biologically real” (Díaz-León 2015, p. 1144).
Briefly, Díaz-León (2015) focuses on individual traits, which may be physical properties and their distribution – for instance weight, general health, or physical strength. These are traits that may have social causes. But it may not be feasible to take action to change someone’s instantiation of these properties, or their current distribution – there may be no straightforward way to act, or no possible retroactive reparation (Díaz-León 2015, p. 1145). On the other hand, a change in social practices can automatically change what constitutive kinds exist. For instance, if to be a husband is to be married to a woman, then if one bans the institution of marriage there would be no more husbands.
Secondly, socially caused traits may be intrinsic properties, and not relational properties, whereas constitutively constructed properties are necessarily relational. By analogy with Mallon’s watch example, Díaz-León argues that one could have the weight one actually has even though one is removed from all social interactions (Díaz-León 2015, p. 1147). Finally, Díaz-León argues that causal social construction is compatible with biological realism about a category or a trait, unlike constitutive social construction (Díaz-León 2015, p. 1148).
I will discuss the argument concerning the feasibility of social change for each kind of social construction from the perspective suggested here in Section 5. In the next section, I present two possible causes of discrimination, stereotype threat and silencing through gendered language use.
3 Social Causes
This section illustrates how different social causes – stereotypical thinking and communicative practices – can contribute to the discrimination of women. Stereotype threat shows how (the implicit) association with a group can cause one to confirm the stereotype, and as such it is a form of explicit (self-)discrimination. The use of gendered language is a social practice that has silencing effects. These effects take place even if the users of gendered language have no explicit negative attitudes about their targets, and the people affected do not explicitly associate with the target group.
I assume, as is standard, that ‘discrimination’ designates the unfair differential treatment of socially salient groups. The differential treatment of a socially salient group by itself does not entail that the group in question amounts to a kind or that the group is discriminated against.  Not all kinds of differential treatment are discriminatory – differential treatment that is justified, like excluding cheaters from a contest, is not discriminatory. Also, not all kinds of differential treatment that are unfair are discriminatory – it is unfair to pick on a student and fail him because I do not like his haircut, for instance, but it does not amount to discrimination as such, because there is no social salient group of people with that haircut that are typically the target of discriminatory treatment.
Although some authors have claimed that sex is socially constructed,  most people tend to treat it as a biological category – a natural human kind – constituted by the possession of certain physical traits. If sex is a biological category, there are (at least) two identifiable sexes, male and female, distinguished by the possession of XY vs. XX chromosomes (although some people have XXX or XXY chromosomes); usually chromosomes lead to the development of gonads into testes or ovaries; usually sexual gonads lead to hormonal differences in the sexes: androgens (including testosterone), or progesterone and estrogen; usually these lead to varying physical traits. It is debatable whether these lead to varying cognitive abilities, and to behavioral and dispositional differences. Some people believe that there are indeed stronger dispositions towards certain cognitive skills and dispositions in human males, and other stronger dispositions towards other skills and dispositions in human females.
Boyd (1988) argued that something belongs to a biological species if it has enough of a set of clustered properties, where (among other things): (1) certain properties are co-occurrent; (2) and that is not accidental (there are underlying homeostatic mechanisms or processes); (3) the clustering of properties has causal effects; (4) the set of things with those properties forms a kind; (5) a thing may display some, but not all, of those properties; (6) some cases may inconclusively belong to a kind (there will be borderline cases) (Boyd 1988, p. 196–199). Following Boyd, we can say that human sexes are like two (natural) sets of clustered properties.
It is often claimed that some cognitive and behavioral differences between the sexes are part of the set of co-occurring properties that distinguish males from females. If that were true for a significant set of such behavioral and cognitive differences, then the alleged difference between sexually determined differences, on the one side, and socially determined gender differences, on the other, could be undermined.  This alleged naturalness of behavioral and cognitive differences is also often taken as a good justification for different treatment and social norms for men and women. In other words, differential treatment and social norms are seen as consequences and not causes of gender differences. The alleged natural differences are often assumed to justify differences in social treatment and norms without further argument.
There are numerous studies that support the plasticity of the mind, and confirm culture’s influence on people’s behavior. For instance, stereotype threat is claimed to cause behavior. In one study about the effects of stereotype threat on cognitive performance more than 100 university students enrolled in a calculus class (Good et al. 2008). Students in the threat condition were told that the test was designed to measure their mathematical ability. Students in the nonthreat condition were told that despite testing on thousands of students no gender difference had been found.
The men and women in the two groups had, on average, all received much the same course grades. You’d expect then…that males and females in the threat and nonthreat condition would perform at about the same level on the test. Instead,…females performed better in the nonthreat condition…Among these participants, men and women in the threat condition, as well as men in the nonthreat condition, all scored about 19 percent…women in the nonthreat group scored an average of 30 percent correct, thus outperforming every other group. (Fine 2010, p. 30–31)
In an earlier notorious study, Steele and Aronson (1995) showed that performance in academic contexts is harmed by awareness of one’s own racial identification. In the Good et al. (2008) study, it was shown that performance on calculus tests is harmed by awareness of one’s own gender identification. Being thus harmed seems unfair. Students do not perform to their best abilities, their real capacities are not being tested, fewer opportunities will be open to them, and they will have fewer possible future career choices for themselves. Moreover, the negative stereotypical associations are based on false beliefs about what members of that (social) kind are really like. If these studies’ results are correct, the negative associations with a group that are pervasive in society, when primed, are a self-fulfilling prophecy: they cause actual differences in cognitive performance.
However, the results of a recent study by Finnigan and Corker (2016) go against the thesis that stereotype threat affects performance, as earlier studies reported. Finnigan and Corker’s study shows that effects of stereotype threat might be smaller than earlier reported, and that stereotype threat might not be a major part of the explanation for the gender gap in math performance. Now, even if this study is right, the appeal to the example of stereotype threat I make here is conditional: if it does have the effects earlier studies predicted, then it counts as a kind of direct discrimination. If it does not have those effects, then it may not count as a kind of direct discrimination.
Lippert-Rasmussen (2013) defines discrimination as the differential treatment (including acts, policies, and practices) on the basis of real or imputed membership in socially salient groups (i.e. groups whose perceived membership structures interaction in many social contexts), which either shows a biased attitude towards those discriminated against (as in the case of direct discrimination), or which does or would result in the discriminatees being made worse off relative to others (as with some instances of direct discrimination, and as with indirect discrimination). If the poor performance effects of stereotype threat are real (which at the moment we do not know is the case), then stereotype threat amounts to a form of direct discrimination. It is a case where agents have biased attitudes towards themselves, and as a result of that are (typically) made worse off relative to others.
If stereotype threat is a cause of poor cognitive performance among members of certain groups, it meets two of Hacking’s broad criteria stated in the beginning of this paper. There are many cases of poor cognitive performance among black people/women that need not have existed: (i) such cases are not determined by people’s nature and are not inevitable; (ii) the many cases of poor cognitive performance that result from stereotype threat are bad as they are. It seems unfair, to people who suffer it, that their professional or academic options are hampered. It is unfair because it is discriminatory. It has effects on people’s performance and it deprives them of future possibilities that would have been otherwise available.
Silencing works as a social obstacle to the felicitous performance of run-of-the-mill illocutionary acts, like orders, recommendations, requests, and so on. I will illustrate how “gendered language” silences professional women. In their well-known paper, Hornsby and Langton (1998) develop an account of silencing:
We say that people are silenced when they are prevented from doing certain illocutionary things with words. People who utter words but fail to perform the illocution they intend may be silenced. The silenced person encounters illocutionary disablement: his or her speech misfires; what she does is unhappy (Hornsby and Langton 1998, p. 21). 
Recently, Grünberg (2014) has argued that silencing might occur when one’s ability to exploit the inherent dynamic of language is ‘blocked’ by one’s gender or social status. Kukla (2014) also distinguishes between silencing and what she designates as discursive injustice, thus offering a theoretical background for the kind of discourse interactions I discuss below.
Thanks to recent campaigns, the use of words like “bossy” to describe women has come to public attention. This and other adjectives are standardly used to describe women, and are known as “gendered language”. A salient recent example involves the firing of Jill Abramson, the first female executive editor of the New York Times, whose competence and skills were not questioned, but who was described by her superiors as “pushy”, “brusque”, “stubborn”, and “condescending”. Another case is Hillary Clinton’s representation in the media during the Democratic primaries (both in 2008 and in 2016). In a recent article in Cosmopolitan, Gupta (2016) wrote
The criticism Gipple waged against Clinton – that she is “dishonest” – is not new. Though perhaps spurred by the email scandal or by accusations that she enabled and obscured Bill Clinton’s womanizing, it is an old one that took roots more than 20 years ago, when Clinton was jockeying for power in a man’s political sphere…Publicly, she dealt with sexist attacks from the media that criticized her as “shrill” and “bitchy,” and compared her to a nagging mother. She was simultaneously not tough enough and yet too tough. She was too feminine and yet not feminine enough. She was a woman with ambition and therefore could not be trusted. In 2006, the Onion parodied the double standard against Clinton in a piece called, “Hillary Clinton Is Too Ambitious to be the First Female President.” In her 2008 campaign for president, a male advisor told her to tone down her feminism because it would alienate voters. Even the criticism from Democrats who supported Obama, whose White House remained a boys’ club, took on an overly harsh, sexist tone.
In an analysis that confirms the attacks against female politicians, Nic Subtirelu used the Corpus of Contemporary American English (COCA) to research the occurrence of the standard gendered descriptions applied to women. His first research focused on “bossy” (see Figure 1).
Here is how he describes his methods,
I searched for bossy used as an adjective (to try to screen out, for example, Bossy the last name). I then took a random sample of 101 occurrences (the final number that resulted after several were eliminated as errors or as not clearly specifying one or the other gender) from the resulting 400 hits. I proceeded to read through and determine the gender and the relative age (young person or adult) of the person being referred to as bossy in each example. My results are presented in the plot below. (Subtirelu 2014b)
Alon Lischinsky, at Oxford Brookes University, used Google books Ngram viewer to analyze the most frequently used nouns after the word “bossy” (see Figure 2). The resulting graph of the 10 most common occurrences is below, with “woman”, “wife”, “women” and “mother” among the more used nouns associated with “bossy”.
After Abramson’s dismissal from the New York Times, Subtirelu (2014a) analyzed COCA again, this time to discover the frequency of the gendered association of the adjectives used by Abramson’s co-workers and superiors. He found that “brusque” and “stubborn” were gender-neutral. He also found that “condescending” occurs more frequently to describe men in positions of authority. But “pushy”, which was used to describe Abramson after she asked for a pay raise and better pension benefits, behaves similarly to “bossy” as a gendered term for women (see Figure 3). 
How does gendered language contribute to silencing professional women? My conjecture is that gendered descriptions like “bossy” or “pushy” are tools for silencing, i.e. means of preventing women from doing certain illocutionary acts with words. This, in turn, undermines their social status and authority (one is either liked or competent, but not both). The high-frequency of the descriptions of women, and of working women in particular, as “bossy” and “pushy” accompanies straightforward forms of illocutionary disablement, as illustrated in the schematic situations described here:
A woman with years of experience in a field of her expertise tries to give advice to an inexperienced male colleague. The colleague describes her publicly as bossy. She is a colleague and not in a position of authority. By describing her as bossy, the conditions for her to give felicitous advice are undermined, and her act is mischaracterized as an order, which she is not in a position to give. She is out of line; she is overreaching her authority.
A woman is promoted to a position of responsibility in a company that is at risk. She requests the same rights as others who have previously held the same position. She is deemed pushy and bossy for making demands. This presupposes that she does not have the right to make such requests. She is also out of line; she is overreaching her authority.
A competent woman with years of public service and expertise is described as dishonest. Her statements are not perceived as authoritative and credible, because her voice is shrill, it lacks “elegance and grace”. Her pledges are not taken seriously, her promises are not believed, her assertions are presumed to be lies. She is out of place; she is not trustworthy as an epistemic agent.
These are schematic descriptions of possible situations faced by women. (2) is a broad description that is applicable to Abramson’s case, where her requests and proposals are represented as out of line demands – and hence she’s bossy or pushy. (3) is based on various comments made about Hillary Clinton, where her statements, and the way she makes them, are held to a higher standard of truthfulness than those of male politicians – and hence she is dishonest and a liar.  Some speech acts have felicity conditions that are related to one’s legitimate status as a source of authority – someone whose requests, orders, or suggestions, deserve to be heard and considered seriously. Other speech acts have felicity conditions that are related to one’s legitimate status as a source of information – to be a reliable source of information. Silencing in the latter case illustrates a form of testimonial injustice, which, according to Fricker (2007), occurs when a person’s testimony is subjected to more doubt and criticism just because a person is a member of an oppressed group.
Forms of silencing through gendered language use are forms of discriminatory treatment of women. The higher regularity of application of certain descriptions to women is an indication that there is a regular practice of silencing women as authoritative or knowledgeable. The practice may not include any explicit awareness of negative attitudes towards women (it is hence unclear whether it should count as explicit, implicit, or structural discrimination).
Not all silencing is discriminatory, however. There may be people who deserve to be silenced, i.e. who deserve to suffer illocutionary disablement. If it were proven that there is a causal relation between the rise of inflammatory hate speech and mass violence, for instance, silencing inflammatory hate speech would be a justifiable way to prevent escalation into mass violence.  Silencing professional women through gendered language use also meets Hacking’s three criteria: it is not inevitable, it is bad as it is, and it would be better if it could be done away with.
4 Inevitability, and Causal vs. Constitutive Construction
The previous section indicated two possible sources of women’s discrimination. Even if being a woman is not a constitutively socially constructed kind, if women are regularly subject to for instance silencing, then they are regularly discriminated against as speakers. Women are harmed or disadvantaged as speakers because of their membership in a socially salient group. I am assuming that socially salient groups are not automatically socially constructed kinds.
Alphabetism illustrates how discrimination may afflict a socially salient group, even if the majority of people do not have explicit attitudes about that group, and the group does not correspond to a constitutively constructed social kind. Social reality may not create a social position or a social role for members of such a group. If sociological research backs up the existence of practices that discriminate negatively against people whose names fall towards the end of the alphabet, then there are social practices that have a negative causal impact on some people – what Carlson and Conard (2011) call the last name effect. There could be constitutively alphabetist social kinds, the AI’s and the RZ’s (allowing for the intermediate neutral JP ’s), after Noah (2011)’s dubbing:
RZ: S is an RZ iff S is systematically subordinated along some dimension (economic, political, legal, social, etc.) and S is ‘marked’ as a target for this treatment by having a family name whose first letter falls towards the end of the alphabet.
AI: S is a AI iff S is systematically privileged along some dimension (economic, political, legal, social, etc.) and S is ‘marked’ as a target for this treatment by having a first letter whose first letter falls towards the beginning of the alphabet.
JP: S is a JP iff S is not marked to be either systematically privileged or subordinated along some dimension by having a family name whose first letter falls towards the middle of the alphabet.
Although these kinds could actually exist, there is no reason to think that there are such kinds, and the strategies for addressing alphabetism as a form of discrimination do not require conceiving of the discriminated people as members of a kind.
If gender were not constitutively socially constructed, there could still be social causes for the harms and disadvantages that women suffer through discrimination. An effort to rectify injustice would require addressing those social causes, finding other means to rectify their effects, or adopting preventative measures to avoid further discrimination in the future.
If, on the other hand, gender is a constitutively socially constructed kind, how should it be characterized? Haslanger (2003, 2006) offers analyses of two gender concepts, woman and man, as examples of constitutively social constructed kinds, while being aware that her proposed analyses are highly counterintuitive.
Woman: S is a woman iff S is systematically subordinated along some dimension (economic, political, legal, social, etc.) and S is ‘marked’ as a target for this treatment by observed or imagined bodily features presumed to be evidence of a female’s biological role in reproduction.
Man: S is a man iff S is systematically privileged along some dimension (economic, political, legal, social, etc.) and S is ‘marked’ as a target for this treatment by observed or imagined bodily features presumed to be evidence of a male’s biological role in reproduction. (Haslanger 2006, p. 6–7)
Haslanger’s definitions of men and women make it part of their nature that they occupy relative positions in a social hierarchy.
On some political views, a society where hierarchies exist is an unjust society. I assume, however, that it is possible that social hierarchies and social justice coexist – political power may be fairly delegated on social representatives; factories may require some people to take leadership positions, whereas others take on jobs with less responsibility; in hospitals, only some people are in a position to decide what kind of treatment is affordable, and to what extent treatment must be continued. So, I assume that ‘hierarchical subordination and privilege’ in the gender definitions should be understood as unfair or unjust subordination and privilege, which involve discrimination.
The forms of differential treatment discussed in the previous section – the effects of stereotype threat, and the silencing effects of gendered language use – are discriminatory ways of treating women. If the definition captures the nature of women, it is only normal that women are silenced because silencing should subordinate by means of the illocutionary disablement it produces. It reminds women of their rightful position.
However, the definitions are counterintuitive. Saul (2006) points to some of the difficulties that a constitutive account like Haslanger’s must overcome. As Saul (2006) points out, a man who decides to become a woman does not automatically decide to become subordinated; many people who are sexually females consider themselves to be women without considering themselves to be subordinated; and it is not logically impossible to be a woman who is not subordinated. Moreover, as Saul stresses, it is unclear whether this constitutive account is what best serves our interests of achieving social change, in part because its counter-intuitiveness may be an obstacle in persuading people that change is needed. 
As Saul (2006)’s concerns highlight, what people mean by ‘women’ is ambiguous – some people consider themselves women without considering themselves to be subordinated, for instance. To avoid confusion, let us separate the current use of the word ‘woman’ from its use in Haslanger’s definition. For the latter acception, we can use instead ‘woman1’. Let us assume, then, that there are women1 and that most people who are considered to be women are actually also women1. It is an open question whether all women (as currently understood) are women1, and whether all women1 are women.
Once we frame the debate in this way, we open the possibility of considering other social positions that human females could have occupied. Let us call the socially constructed kind that could have existed in one such situation ‘women2’.
Woman1 To be a woman is to be someone with observed or imagined bodily features presumed to be evidence of a female’s biological role in reproduction, and to be targeted for systematic subordination because of it.
Woman2 To be a woman is to be someone with observed or imagined bodily features presumed to be evidence of a female’s biological role in reproduction, and to be targeted for systematic privilege because of it.
‘Woman2’ designates a constitutively constructed social kind as much as ‘woman1’. But woman1 is an actual social kind, and woman2 has presumably no actual instances. The same people (people whose observed or imagined bodily features are presumed to be evidence of a female’s biological role in reproduction) are actually women1, but could have been women2. The privilege of women2 would occur in situations where another social position is created, that of men2. Men2 do not actually exist, but if they did, they would be unfairly discriminated against.
Women1 and women2 define constitutively socially constructed kinds, and as such they license inferences to necessity claims. What this means is that, given the definitions of constitutively socially constructed kinds as capturing the nature of the kind, it is not possible to be a woman1 and not be targeted for subordination along some dimension. In other words,
Necessarily, to be a woman1 is to be targeted for subordination.
Necessarily, to be a woman2 is to be targeted for privilege.
Prima facie, it is an odd consequence of social constructivism that the claim that a certain category is socially constructed should entail a necessity claim about that category. After all, one of the central features of social constructivist claims about a given kind X is that X is not inevitable, need not be as it is, and is not determined by the nature of things. Otherwise there would be no question of addressing injustice through social action.
Capturing the sense in which social construction goes against the inevitability of a trait is a main motivation for Díaz-León (2015), when she talks of the three claims that Hacking summarizes as characteristic of social constructionism (see introduction, Section 1, above):
This is what I call the project of arguing against the inevitability of a trait. The project is especially appealing when (2) and (3) [in Hacking’s list of general features of social constructionist claims] are held about a certain category: in those cases, claim (1) [X is not inevitable, need not exist or need not be as it is] opens a path for change and transformation regarding what is seen as a harmful state of affairs. (Díaz-León 2015, p. 1138)
If (1), the claim that X is not inevitable, is read as “X need not be as it is”, then there is a problem for the claim that constitutive social kinds are “theoretically or politically useful” (Haslanger 2003, p. 319). As we just saw, a constitutively socially constructed kind like woman1 needs to be as it is – that is what its social nature entails, and what follows from statements like “if we change those social practices of privilege and subordination, there will no longer be any women” (Díaz-León 2015, p. 1145).
The idea that X is not inevitable, and can be changed through social action, must be understood as “X need not exist”, if X is to be theoretically or politically useful. The constitutive social kind X need not exist – i.e. Women1 need not exist. But on the assumption that there are women1, what explains that this is the case? If there is a difference between a world where there are women1 and one where there are women2, there must be a difference in the causal social history of the two worlds.
The fact that some specific gender roles exist in a society is not independent of the causal history that led to the creation of those “social positions” and that keeps them in place. It is contingent that any social roles exist because it is a contingent fact that societies developed as they did. The cross-cultural and historical variability of gender norms is some indication of the different ways societies may develop. We can plausibly claim that gender is socially constructed constitutively because there are social causes that have led to the systematic discrimination and subordination of women, and the relative privilege of men. There are many social causes that converge towards this end. Stereotype threat is one presumable cause, and silencing professional women through gendered language use is another. 
If this is right, any existing kind X that is constitutively socially constructed as an F is also causally socially constructed as an F. In other words, if F is an existing constitutively social constructed kind, then there are social factors that have caused the kind F to exist and to persist – social norms, conventions, practices, institutions, ideologies, and so on. This establishes one of the central claims of this paper:
For any actual kind X, if X is socially constructed in the constitutive sense as a K, then X is socially constructed in the causal sense as K.
Recall that Haslanger warned us that “it is an error to treat the conditions by virtue of which a social entity exists as causing the entity” and that “we should avoid conflating social kinds with things that have social causes” (Haslanger 2003, p. 317). However, as the husband example illustrates, this depends on what we mean by “the conditions by virtue of which an entity exist”.
There are social (and other) causes that make it the case that if one man is married to a woman then he is a husband, without it being the case that being married to a woman causes one to be a husband. As the husband example illustrated, there could have been other kinds of marriage institutions. The reason why there are not other marriage institutions is not simply because there is no one who is a man married to a woman and at least another man. This is because our causal social history did not establish a marriage institution with such a legally recognized social status.
If there are constitutively constructed social kinds, their existence is not separable from the social causal history that made each an actual social kind. The next and final section of this paper argues that causal social construction is at least as relevant as constitutive social construction to achieve social justice by means of social action.
5 Political Relevance and Feasibility, Again
Díaz-León (2015) claimed that social change is more feasible under an understanding of social construction as constitutive rather than causal, because constitutive construction “immediately opens a way for creating feasible strategies for social change” (Díaz-León 2015, p. 1145). For example, if being a woman is to occupy a position of subordination in a hierarchical system, then if we change those social practices of privilege and subordination, there will no longer be any women. When constitutive constructivists say “there will no longer be any women”, they do not mean that the people who are women will disappear from the face of the Earth. What they mean is that the social position that these people occupy will no longer exist.
By contrast, socially caused traits may be intrinsic, they may be biological traits, and, more importantly, there may be no feasible way to take action to change the instantiations of such properties, or their distribution. By comparison, “changing social practices of privilege” seems to be a more feasible possible course of action.
The political significance of social constructionist claims derives from the possible social impact of those claims, not just theoretically, but for projects of social action and change, in particular addressing the discrimination of socially salient groups. I have just argued that any actual constitutively constructed social kind is also causally constructed. This seems support the other main claim of this paper:
Causal construction is at least as relevant as constitutive construction to the goal of achieving social justice through social action.
Díaz-León could resist this move. Even if she were to concede that if X is constitutively socially constructed as kind K, then it is causally constructed as K, she could claim that is less feasible to act on social causes than it is to act on the “social practices of privilege” that keep social kinds in existence.
I think this is disputable, for two reasons. In the first place, social causes may be addressed in the right way. It is obviously true that one cannot act on past causes of present injustices. In this sense, it is not possible to prevent certain effects of social causes. At best, institutions may atone for their past errors (the Catholic Church may apologize for the Inquisition, for instance), or reparations may be sought. But Díaz-León (2015) says:
In particular, the fact that a certain individual has a trait that has been causally socially constructed in this way, does not automatically pave the way for creating feasible social strategies for changing those traits in those individuals. Perhaps uncovering the social origins of these traits will give rise to strategies for preventing the instantiation of those traits in future generations, but this will not necessarily help to change the facts that we take to be unjust today. (Díaz-León 2015, p. 1144)
As I have insisted earlier, our concern is not merely with facts about certain individuals having certain traits, but whether certain kinds of treatment of certain groups of individuals, because of their (perceived) membership in the group, are unfair. More important, however, is the legitimate question of what may be done to address social causes of injustice or harm.
In the discussion of silencing in Section 3, I mentioned research by Maynard and Benesch (2016) on the correlation between inflammatory speech, mass violence, and genocide. Their research project focuses on investigating ways to monitor dangerous speech and to prevent its effects. Their research seeks to develop feasible models of prevention of the harmful effects of specific social causes – in this case, inflammatory speech against a population, like an ethnic or a religious minority. It does not proceed by proposing a change in the identity of the ethnic or religious groups in questions.
In the second place, Díaz-León (2015) claims that achieving social change through social action on institutions and practices is more feasible. I strongly doubt that it is any easier to act on the “social practices of privilege” than it is to try to monitor and prevent unfair consequences of certain social practices. Furthermore, if my claim is correct (that causal social factors explain the existence of instances of a social kind), then there is no straightforward way to eliminate a social role or kind without addressing the causal relations that contribute to their existence.
Many acknowledge the difficulty of achieving political change in society. In The Prince, Machiavelli said
And it ought to be remembered that there is nothing more difficult to take in hand, more perilous to conduct, or more uncertain in its success, than to take the lead in the introduction of a new order of things, because the innovator has for enemies all those who have done well under the old conditions, and lukewarm defenders in those who may do well under the new. (Machiavelli 2006, ch. 6)
Given reasons to doubt the claim of the greater feasibility of achieving social change by changing social relations and institutions, and the fact that actual processes to address social justice seek to monitor and prevent certain social practices from effecting their consequences, I conclude that causal construction is at least as relevant as constitutive construction for achieving social justice by means of social action. This is further supported by the fact that any actual constitutively socially constructed kind is also causally socially constructed.
6 Concluding Remarks
The aim of this paper was to address what I think is a misconception about the relation between causally constructed kinds and constitutively constructed kinds, and to criticize the idea that a focus on constitutively constructed kinds should take priority over a focus on actual causal social relations.
Part of the misconception is to think of social causes as mere causes of an individual’s instantiation of a (possibly intrinsic) property. This ignores a fundamental metaphysical question concerning constitutive social kinds and their contingency on social reality: Why does a kind F exist instead of a kind K, given that there are indefinitely many possible constitutively constructed kinds, and that not all have actual instances? Understanding why a constitutively constructed kind exists instead of another requires that we focus on the social causes of the existence of a social kind. This is required by a proper development of the idea that socially constructed things are not inevitable.
Furthermore, to properly address the discrimination of socially salient groups, we should look at the persistent causes that perpetuate the direct or indirect discrimination of such groups. It may be indeterminate whether or not a socially salient group amounts to a constitutively constructed social kind, but various forms of differential treatment might anyway systematically and unfairly harm that group.
So we are faced with a dilemma. On the one hand, if group X is not constitutively socially constructed as a kind K, but is discriminated, then the causes of X’s discrimination must be addressed. We should understand the social causes of injustice towards socially salient groups, and seek measures that can be pursued to offer reparations where possible, to monitor the reoccurrence of those causes, and to prevent future unfair treatment. On the other hand, if a group X is constitutively socially constructed as a kind K, the social causes of X’s instantiating K need to be understood and addressed, particularly if the kind in question perpetuates unjust practices and structures. If race, gender, disability, sexual orientation, etc., are constitutively constructed kinds, then (as I have argued) we need to understand not just what they are constitutively, but how social causes contribute to their continued existence. Stereotype threat presumably contributes to perpetuating unfair differences (depending on the final verdict about its effects on poor performance). If it does affect performance, we may seek ways to counteract or neutralize its consequences. The use of gendered language appears to have silencing effects. The occurrence of silencing requires measures to either monitor or prevent the use of gendered language, or alternative forms of sidestepping the resulting illocutionary disablement. The proposal to monitor and prevent inflammatory speech, probably a precursor or contributing cause to mass violence, illustrates one way of addressing social causes of harm or injustice.
I conclude, thus, that the above is reason enough for causal social construction to be at least as relevant as constitutive social construction. They are both useful theoretically and politically, and neither seems to offer a more feasible path for achieving social change through social or political action.
Boyd, R. (1988): “How to be a Moral Realist”. In: Moral Realism. pp. 181–228. Ithaca: Cornell University Press.Search in Google Scholar
Butler, J. (1990): Gender Trouble: Feminism and the Subversion of Identity. Oxford: Routledge.Search in Google Scholar
De Beauvoir, S. (1952): The Second Sex. New York: Vintage Books.Search in Google Scholar
Dretske, F. (1988): Explaining behavior: Reasons in a world of causes. Cambridge, MA: MIT Press.Search in Google Scholar
Fausto-Sterling, A. (2000): Sexing the Body. New York: Basic Books.Search in Google Scholar
Fine, C. (2010): Delusions of Gender: How our Minds, Society and Neurosexism Create Difference. New York: W. W. Norton.Search in Google Scholar
Fine, C. (2013): “New Insights into Gendered Brain Wiring, or a Perfect Case Study in Neurosexism?” The Conversation. http://theconversation.com/new-insights-into-gendered-brain-wiring-or-a-perfect-case-study-in-neurosexism-21083, Accessed 4 December 2013.Search in Google Scholar
Finnigan, K. M. and K. S. Corker (2016): “Do Performance Avoidance Goals Moderate the Effect of Different types of Stereotype Threat on Women’s Math Performance?” In: Journal of Research in Personality 63, p. 36–43.10.1016/j.jrp.2016.05.009Search in Google Scholar
Good, C., J. Aronson and J. Harder (2008): “Problems in the Pipeline: Stereotype Threat and Women’s Achievement in High-Level Math Courses”. In: Journal of Applied Developmental Psychology 29. No. 1, p. 17–28.10.1016/j.appdev.2007.10.004Search in Google Scholar
Grünberg, A. (2014): “Saying and Doing: Speech Actions, Speech Acts and Related Events”. In: European Journal of Philosophy 22. No. 2, p. 173–199.10.1111/j.1468-0378.2011.00481.xSearch in Google Scholar
Gupta, P. (2016): “Do you Really not like Hillary Clinton, or are you just Sexist?” Cosmopolitan. http://www.cosmopolitan.com/politics/news/a52707/why-dont-young-people-like-hillary-clinton/, Accessed 10-February-2016.Search in Google Scholar
Hacking, I. (1999): The Social Construction of What? Cambridge, MA: Harvard University Press.Search in Google Scholar
Haslanger, S. (2003): “Social Construction: The “Debunking” Project”. In: F. Schmitt (Ed.): Socializing Metaphysics, Oxford: Rowman and Littlefield, pp. 301–325.Search in Google Scholar
Haslanger, S. (2006): “What Good are our Intuitions? Philosophical Analysis and Social Kinds”. In: Aristotelian Society Supplementary 80. No. 1, p. 89–118.10.1111/j.1467-8349.2006.00139.xSearch in Google Scholar
Hornsby, J. (1994): “Illocution and its Significance”. In: S. L. Tsohatzidis (Ed.): Foundations of Speech Act Theory: Philosophical and Linguistic Perspectives, London: Routledge, p. 187–207.Search in Google Scholar
Ingalhalikar, M., A. Smith, D. Parker, T. Satterthwaite, M. Elliott, K. Ruparel, H. Hakonarson, R. Gur, R. Gur and R. Verma (2014): “Sex Differences in the Structural Connectome of the Human Brain”. Proceedings of the National Academy of Science 111. No. 2, p. 823–828.10.1073/pnas.1316909110Search in Google Scholar
Lippert-Rasmussen, K. (2013): Born Free and Equal? A Philosophical Inquiry Into the Nature of Discrimination. Oxford: Oxford University Press.10.1093/acprof:oso/9780199796113.001.0001Search in Google Scholar
Machiavelli, N. (2006): The Prince (EBook #1232 ed.). The Project Gutenberg. https://www.gutenberg.org/files/1232/1232-h/1232-h.htm, Accessed May 2016.Search in Google Scholar
Mallon, R. (2009): “Naturalistic Approaches to Social Construction”. In: Stanford Encyclopedia of Philosophy.Search in Google Scholar
Maynard, J. and S. Benesch (2016): “Dangerous Speech and Dangerous Ideology: An Integrated Model for Monitoring and Prevention”. Genocide Studies and Prevention: an International Journal 9. No. 3, p. 1–27.Search in Google Scholar
Noah, T. (2011): “Tyranny of the Alphabet”, Slate.http://www.slate.com/articles/business/the_customer/2011/01/tyranny_of_the_alphabet.html, Accessed 1 February 2016.Search in Google Scholar
Steele, C. M. and J. Aronson (1995): “Stereotype Threat and the Intellectual Test Performance of African Americans”. Journal of Personality and Social Psychology 69. No. 5, p. 797–811.10.1037/0022-3518.104.22.1687Search in Google Scholar
Subtirelu, N. (2014a): “Beyond Bossy on our Gendered Characterizations of Leadership and Authority”. Linguistic Pulse. https://linguisticpulse.com/2014/05/19/beyond-bossy-more-on-our-gendered-characterizations-of-leadership-and-authority/, Accessed June 2014.Search in Google Scholar
Subtirelu, N. (2014b): “No Really, Bossy is Gendered”. Linguistic Pulse. https://linguisticpulse.com/2014/03/28/no-really-bossy-is-gendered/, Accessed June 2014.Search in Google Scholar
©2016, Teresa Marques, published by De Gruyter.
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.