Skip to main content
Open Access

Amelioration and Inclusion: Gender Identity and the Concept of Woman*

Abstract

Feminist analyses of gender concepts must avoid the inclusion problem, the fault of marginalizing or excluding some prima facie women. Sally Haslanger’s ‘ameliorative’ analysis of gender concepts seeks to do so by defining woman by reference to subordination. I argue that Haslanger’s analysis problematically marginalizes trans women, thereby failing to avoid the inclusion problem. I propose an improved ameliorative analysis that ensures the inclusion of trans women. This analysis yields ‘twin’ target concepts of woman, one concerning gender as class and the other concerning gender as identity, both of which I hold to be equally necessary for feminist aims.

I. Introduction: Ameliorative Inquiry and the Inclusion Problem

Over the last few decades, feminist philosophers have sought to develop an analysis of gender concepts, and of the concept woman in particular. This task is rendered difficult by the fact that since there seems to be no single property that all women have in common, attempts to define woman risk excluding or marginalizing some women. Typically, it is women who are also members of other oppressed social groups, such as women of color and working-class women, who are at risk of exclusion or marginalization. Leaving the concept undefined, however, calls into question the project of feminism—supposedly a movement to end the oppression of ‘women’. Thus, the task is to develop a suitably inclusive concept of woman—one that avoids what we can call the inclusion problem.

Although many attempts have been made to develop such a concept, my focus here will be on one attempt in particular.1 In a series of papers published together in the collection Resisting Reality, Sally Haslanger develops a distinctive form of conceptual analysis that, she claims, can be applied to the concept of woman to offer a unique way of avoiding the inclusion problem.2 This method of analysis is termed ‘ameliorative inquiry’. According to Haslanger, an ameliorative inquiry into a concept F is the project of arriving at the concept of F-ness that a particular group should aim to get people to use, given a particular set of goals that the group holds.3 Ameliorative inquiries thus make use of normative inputs. The concept of F-ness that is generated by an ameliorative inquiry is the target concept of F. Ameliorative analysis is not bound to comply with our ordinary understanding or use of a concept: the target concept may be revisionary, provided that it furthers the goals guiding the analysis.

An ameliorative inquiry into the concept of woman invites feminists to consider what concept of woman would be most useful in combatting gender injustice. This opens the way for a revisionary analysis that can be tailored to avoid exclusion and marginalization. Thus, ameliorative inquiry seems a promising approach to adopt in the face of the inclusion problem. Embarking on this feminist ameliorative project, Haslanger proposes that the target concept of woman should be defined as (roughly) someone who is socially subordinated in some way on the basis of presumed female sex.4 Correspondingly, the target concept of man is defined as someone who is socially privileged on the basis of presumed male sex. She argues that this concept avoids the inclusion problem because although not all prima facie women are included, all prima facie women who are subject to oppression are included. Since these are the women about whom feminism should be concerned, Haslanger argues, the concept covers everyone who ought to be included from a feminist perspective.

My concern in this article is with the implications of Haslanger’s proposed target concepts for trans people.5 By ‘trans people’, I mean all people who identify as a gender other than the one to which they were assigned at birth, which includes trans women (people categorized as male at birth who later come to identify as women), trans men (people categorized as female at birth who later come to identify as men), and nonbinary trans people (people who identify neither as simply men nor as simply women). I am not making any distinction between ‘transsexual’ and ‘transgender’ identities. I will use the term ‘cis’ to describe anyone who is not trans. It will be relevant to my arguments that trans people in general are a severely disadvantaged and marginalized group in society, suffering oppression and injustice in multiple respects including discriminatory denial of goods such as employment, medical care, and housing; consistently negative portrayals in the media; and particularly high risks of violence. This oppression intersects with other axes of oppression such as race, socioeconomic status, and gender. The proposition that trans gender identities are entirely valid—that trans women are women and trans men are men—is a foundational premise of my argument, which I will not discuss further. Failure to respect the gender identifications of trans people is a serious harm and is conceptually linked to forms of transphobic oppression and even violence.6 It follows from this that an important desideratum of a feminist analysis of gender concepts is that it respect these identifications by including trans people within the gender categories with which they identify and not including them within any categories with which they do not identify.

In the first part of this article I argue that Haslanger’s proposal for target gender concepts does not in fact solve the inclusion problem because it does not include trans people within their identified genders. Focusing on trans women, I show that according to Haslanger’s definition of woman, some trans women would not count as women. Given that a target concept is a normative proposal for how feminists ought to use the concept woman, this is an unacceptable result: the adoption of the concept would exacerbate the existing (and illegitimate) marginalization of trans women within feminist discourse. Despite the unacceptability of Haslanger’s own proposed target concept of woman, however, the framework of an ameliorative project is, I argue, a useful tool for feminists seeking a definition of woman that avoids the inclusion problem. Accordingly, in the second part of this article, I carry out an ameliorative inquiry that explicitly aims to respect the gender identifications of all trans people. This requires adjusting the definition of ameliorative inquiry to allow for the possibility of arriving not at one single target concept but at multiple equally significant target concepts. I argue that an ameliorative inquiry that is attentive to the need to include trans people will yield two target concepts of gender. The first of these corresponds to Haslanger’s proposed concept and captures the sense of gender as an imposed social class. The second captures the sense of gender as a lived identity and draws on Haslanger’s (separate) account of racial identity. These two concepts, I argue, are equally crucial for feminist purposes—and, together, they form an account that succeeds in avoiding the inclusion problem. I conclude by briefly considering the implications of this result for the question of how gender terms should be used in everyday contexts.

II. Assessing Haslanger’s Proposed Target Concepts

A. Haslanger’s Account

Let us begin by taking a closer look at Haslanger’s proposed target concepts. For Haslanger, being gendered as a man or a woman consists of occupying a certain hierarchical social role on the basis of one’s presumed sexed anatomy. She seeks “to offer a focal analysis that defines gender, in the primary sense, as a social class,” where a focal analysis is one that “undertakes to explain a variety of connected phenomena in terms of the relations to one that is theorized as the central or core phenomenon.”7 Thus, although she acknowledges that gender terms are used in multiple ways, she takes the idea of a social class based on presumed sex to capture the sense of gender that is most important for a critical feminist analysis of gender.

For the purposes at hand, the target concept Haslanger proposes for this social class notion of gender can best be explained through first giving her definition of ‘functioning as a woman’:

S functions as a woman in context C iff:

(i) S is observed or imagined in C to have certain bodily features presumed to be evidence of a female’s biological role in reproduction;

(ii) that S has these features marks S within the background ideology of C as someone who ought to occupy certain kinds of social positions that are in fact subordinate (and so motivates and justifies S’s occupying such a position); and

(iii) the fact that S satisfies (i) and (ii) plays a role in S’s systematic subordination in C, that is, along some dimension, S’s social position in C is oppressive, and S’s satisfying (i) and (ii) plays a role in that dimension of subordination.8

The account then states that S is a woman if and only if (iff) S functions as a woman “regularly and for the most part.”9 A counterpart definition is given of men as individuals who are privileged on the basis of observed or imagined bodily features presumed to be evidence of a male role in biological reproduction.

What about the inclusion problem? Since Haslanger’s proposed target concept of woman leaves open the kinds of subordinated social positions that women occupy, it does not center the experience of women of any particular culture, race, or class. It therefore avoids marginalizing women on the basis of any of these aspects of identity. Haslanger acknowledges, however, that her account does marginalize some people who are often thought to be women, for it will not categorize as a woman any (actual or possible) prima facie women who are not subordinated on the basis of being presumed to be female. She argues that this marginalization is not problematic, because it is not in tension with the goals of feminism: “For the purposes of a critical feminist inquiry, oppression is a significant fact around which we should organize our theoretical categories; it may be that non-oppressed females are marginalized within my account, but that is because for the broader purposes at hand—relative to the feminist and antiracist values guiding our project—they are not the ones who matter.”10 Haslanger, then, is perfectly explicit about the fact that her account marginalizes nonoppressed prima facie women. I will argue, however, that the account also gives rise to another marginalization, much more problematic, which goes unacknowledged.11

B. Implications for Trans Women

In this section and the next I show that Haslanger’s proposed target concept of woman does not include all trans women, thereby failing to respect trans women’s gender identifications. The arguments I give will also apply to the wrongful exclusion of other trans people from various gender categories. Related arguments could also be developed in regard to wrongful inclusion in a gender category with which one does not identify, both as it affects trans women and as it affects other trans people.12 Here, however, my focus will be solely on the wrongful exclusion of trans women from the category of women.

Since what it means to be a trans woman is open to multiple understandings, an analysis of the concept of woman that respects the gender identifications of trans women will need to provide space for a variety of articulations and interpretations of trans experiences. Two points of diversity will be particularly relevant to my argument. The first is that some trans women take steps to alter their body through medical intervention such as hormones or surgery, while others take no such steps. The second is that some trans women make their gender identity public through the use of feminine pronouns, names, or forms of presentation, while others choose to keep their gender identification private. The thought here is emphatically not that a feminist analysis of woman must “explain” this range of identities, for trans women’s identities do not stand in any special need of “explanation.” Rather, the point is simply that the analysis should allow for there being a range of different ways in which these identities are lived.

What does this mean for Haslanger’s proposed target concept of woman? In a footnote, Haslanger says that her definition entails that “a female functioning socially as a man or a male functioning socially as a woman” is a member of that gender.13 But under what circumstances would Haslanger’s definition count a trans woman as ‘functioning as a woman’? The complexity of the account means that the answer to this question is not immediately apparent. Accordingly, I will now clarify the account’s implications for trans women by distinguishing four possible scenarios a trans woman may be in and establishing how Haslanger’s account categorizes her social functioning in each case. These scenarios are intended as snapshots of circumstances in which a trans woman may find herself at a particular moment in time, not as situations that are necessarily stable or lasting. Moreover, they do not cover all possible scenarios in which a trans woman may find herself.

Scenario 1: A trans woman does not publicly present as a woman and is perceived as a man by people around her. It seems clear that, on Haslanger’s definition, a trans woman in this situation is not functioning as a woman, whether or not she is in any way subordinated.

Scenario 2: A trans woman publicly presents as a woman, but her gender presentation is not respected: she is seen by those around her as a man “pretending” to be a woman. Again, Haslanger’s definition entails that a trans woman in this situation is not functioning as a woman. This would be so even if she is experiencing subordination, because the subordination is not occurring on the basis of her being perceived as having bodily features associated with a female’s role in biological reproduction (because she is not being so perceived).14

Scenario 3: A trans woman publicly presents as a woman, and her gender presentation is respected by those around her (they use her correct pronouns, etc., and think of her as a woman). Specifically, though, her gender presentation is respected because she is perceived by those around her as having bodily features associated with a female’s role in reproduction (e.g., she may be perceived as having breasts or be presumed to have a vulva). There are two possible ways that this could be the case. First, the woman in question could be understood to be cis.15 Second, she could be understood to be trans and be perceived as a woman in virtue of having bodily features associated with a female’s role in biological reproduction. This would require, first, her being understood to have undergone some medical interventions that altered some of her bodily features and, second, these bodily features being understood as the same sorts of bodily features that cis women have.16 Haslanger’s definition entails that a trans woman in this situation is functioning as a woman, provided that she experiences some form of subordination on the basis of how she is perceived.

Scenario 4: A trans woman publicly presents as a woman, and her gender presentation is respected, but, unlike in scenario 3, this is not because she is perceived as having bodily features associated with a female’s role in biological reproduction. Although she may or may not be perceived as having such bodily features, her gender presentation is respected unconditionally, being taken as an indication of how she would like to be treated socially. Although this is often not the case in mainstream contexts, where bodily features are frequently taken to be evidence of gender, it is very commonly the case in limited contexts, such as trans communities, where resistant understandings of gender are in operation.17 It would seem intuitive to say that this woman is functioning socially as a woman. Let us further suppose that she experiences some subordination on the basis of being perceived as a woman. Even so, she does not count as functioning as a woman according to Haslanger’s definition. This is because being perceived as having bodily features associated with a female’s role in biological reproduction plays no role in how she is viewed and treated by others and therefore cannot be the basis for the subordination she experiences.

Recall now that being a woman, for Haslanger, requires that one function as a woman ‘regularly and for the most part’. So trans women will be categorized as women by Haslanger’s account only if they find themselves in scenario 3 most of the time. Some trans women will never find themselves in scenario 3, and many trans women will find themselves in that scenario only some of the time.18 Therefore, many trans women will not be categorized as women according to Haslanger’s definition.

Marginalizing trans women is importantly different from marginalizing nonoppressed (prima facie) cis women. This is partly because trans women are a severely oppressed group, and much of that oppression is closely tied to denials of the legitimacy of their genders. It follows from this that there is a particularly strong imperative to respect trans women’s gender identifications. Moreover, trans women in any of the scenarios are likely to experience oppression, and in the case of scenario 4 especially, this seems like oppression that is directed at them qua women.19 These factors mean that Haslanger’s justification for the exclusion of nonoppressed females cannot be deployed in relation to trans women: trans women most certainly do matter for feminist purposes, and any suggestion to the contrary is immediately and deeply troubling.20 Thus, the fact that Haslanger’s target concept of woman excludes some trans women renders her account vulnerable to the inclusion problem.

C. Possible Responses

There are three possible responses that could be made to the finding that Haslanger’s definition excludes many trans women. First, one might try to defend the account in its current form by showing that, despite initial appearances, it is not disrespectful to trans women. Second, one might acknowledge that the account as it stands is problematic but defend a modified version of the account. Finally, one could simply concede that the entire approach of giving an ameliorative definition of woman in terms of socially imposed subordination is unsuccessful and abandon the project altogether. In this section, I consider each of these responses in turn.

The most promising way to try to defend the account in its current form would be to draw attention to the fact that Haslanger is theorizing gender as an oppressive system of social class. Such an analysis is inevitably going to be a critical one—it describes without endorsing them a set of categories that contingently function in our social world. In talking about gender as an oppressive social class system based on perceived or imagined sex, then, the analysis simply says that not all trans women are positioned as women within this system. This claim in itself need not constitute a failure to respect trans women’s gender identifications, because it is a critical one. To describe an oppressive system is not to endorse it. Thus, it is open to Haslanger to say that part of what is oppressive about the gender class system that exists at present is precisely that it does not classify all trans women as women. And indeed, the claim that the current dominant gender order is cissexist/transphobic as well as sexist/patriarchal is a foundational premise of transgender studies.21

Following this line of thought would lead to the suggestion that the significance of the inclusion problem is overstated. If the concept of woman is defined as a subordinated social category, then (the thought would go) it may be no bad thing to be excluded from that category. The immediate problem with this defensive strategy is that it risks overlooking the intrinsic importance of respecting trans women’s gender identifications, as outlined above in Section I. In order to succeed in respecting trans women’s gender identifications, an analysis of gender concepts that took this approach would need at the same time to acknowledge trans women’s assertions that they are women as true for at least some sense of ‘woman’. This could perhaps be accomplished by leaving room for some other concept of gender according to which all trans women are women, despite not being categorized as such by dominant ideologies that govern the oppressive social classes of gender.

Recall that Haslanger acknowledges that the term ‘gender’ admits of a number of different interpretations, and her target concept is only intended to capture one of them. Another sense of gender that she mentions is ‘gender identity’, meaning “a broad psychological orientation to the world.”22 If gender identity is understood as a trans-inclusive concept (and I temporarily assume for the sake of argument that it is), so that all trans women are women so far as gender identity goes, then it could be argued that Haslanger’s proposed target concept does succeed in respecting trans women’s agency because it allows for another sense of gender according to which all trans women are women.23

This defense cannot, however, succeed. Recall that Haslanger’s proposed target concept is accorded theoretical primacy over other related concepts concerning gender—including, significantly for our purposes, gender identity.24 Although Haslanger’s focal analysis of gender leaves room for other concepts, then, it assigns these concepts a secondary or peripheral status. This means that in this feminist theory many trans women would only be counted as women in a secondary or peripheral sense—a paradigmatic case of marginalization.25 Thus, the analysis fails to respect the gender identifications of trans women, because it fails to fully include them within the category of women. Moreover, this failure of inclusion has serious practical consequences for feminism as a political movement. Presumably, Haslanger intends her target concept to be useful to feminists as a focal point of organizing against gender injustice. Adopting the current target concept for use in this context would therefore have the consequence of excluding and marginalizing trans women from feminist practice. Given trans women’s interest in organizing against the oppression they face as women and the many important contributions trans women have made to feminism, this would be both an injustice against trans women and a severe loss to feminism itself.

It follows that the need to include all trans women within the concept of woman is not obviated by the fact that Haslanger’s proposed target concept is a negative or critical one or by the fact that it allows for a secondary concept of gender identity. In Haslanger’s account, the concept of woman (a) is defined as a social class in a way that excludes many trans women and (b) is given primacy, within feminist analysis, over other senses of gender, including gender identity. These two features are jointly sufficient for the concept to entail that many trans women are, at best, only women in a secondary or peripheral sense (which is to say that neither would be sufficient without the other).

The second response that could be made to the finding of the previous section is to try to defend a modified form of Haslanger’s account. Given that the exploration of the first response revealed that it is the exclusionary contours of Haslanger’s definition of woman and the focal nature of her analysis that in conjunction give rise to the marginalization of trans women, removing either one of these features would suffice to render the account appropriately inclusive. In other words, the question that arises if we try to pursue the second response is this: is the problem with Haslanger’s account located in the detail of the definition or in the focal model within which that definition is situated? Which aspect should we seek to alter to render the account appropriately inclusive?

Locating the problem in the detail of the account would suggest that we should try to spell out a different set of features that play a role in subordination—a set of features that does include all trans women. This, however, will not work. Suppose, for example, that we were to alter the account of functioning as a woman so that clause (i) reads “S is observed or imagined to have a female gender identity.” Overall, then, S would function as a woman on this account if S was subordinated on the basis of (presumed) female gender identity. This seems to be as inclusive a feature as we could hope to offer while retaining the general approach of defining woman in terms of socially imposed subordination. Nevertheless, the trans woman in scenario 1 would not meet this definition, because no one around her is observing or imagining her to have a female gender identity. This demonstrates that it is impossible to define woman by reference to a set of features that function as a basis for socially imposed subordination in a way that includes all trans women: whatever features are selected, some trans women will always be excluded. Moving away from the idea of defining woman by reference to subordination at all would be simply to abandon the account wholesale (of which more in a moment). If we wish to try to modify the account to render it defensible, then, we are left with one option: to alter it so that the sense of gender as an oppressive social class is not privileged over gender identity. It is this option that I will be pursuing in the rest of the article.

First, though, let me say something about the third response we could make to the finding of the previous section: why should we not just abandon Haslanger’s ameliorative analysis? The tactic of giving a revisionary account of gender based around targeted social oppression may have seemed like a promising response to the inclusion problem, but we have now seen that the problems it evinces in relation to trans women run rather deep. One could be forgiven for thinking that the defect lies with the ameliorative framework itself.

And indeed, it is, in one sense, rather surprising that Haslanger’s proposed target concept of woman fails to meet the desideratum of respecting trans women’s gender identifications. This is because that desideratum finds a strong echo in the goals Haslanger herself sets for the feminist ameliorative inquiry into gender concepts. One of these goals is “the need for [an account] that … take[s] seriously the agency of women … and within which we can develop an understanding of agency that will aid feminist … efforts to empower critical social agents.”26 The concern to respect trans women’s gender identifications can easily be understood as one aspect of this more general aim: all that is required is that ‘women’ is interpreted so as to include trans women and that the idea of taking someone’s agency seriously is understood, quite naturally, to include respecting that person’s gender identification. However, if this was the way that Haslanger had understood this goal, then her inquiry would not have arrived at the target concepts that she proposes.

Does the problem then lie with the ameliorative approach itself? I believe not. Let us have closer look at the precise definition of an ameliorative inquiry. An ameliorative inquiry seeks to “elucidate ‘our’ legitimate purposes and what concept of F-ness (if any) would serve them best (the target concept).”27 In a footnote to the passage in which this quote appears, Haslanger explains: “I put ‘our’ and ‘we’ in scare quotes to indicate that there may be significant contextual variation, or at least there will be room for contestation.” In other words, different groups of social actors in different contexts will have different purposes in mind when they undertake an ameliorative project, meaning that different target concepts will be arrived at by these various groups in these various contexts. While it is not necessary to fix once and forever precisely who ‘we’ are in order to carry out an ameliorative inquiry, we do need to recognize that the results of ‘our’ inquiry will be relative to ‘our’ legitimate goals as ‘we’ conceive of them at the time. Call the referent of the ‘we’ in Haslanger’s description of ameliorative inquiry ‘the agents of ameliorative inquiry’.

It follows that an ameliorative inquiry can only be as good as the goals that guide it, and the way that these goals are formulated depends substantially on how the agents of ameliorative inquiry are understood. If I want to undertake a feminist ameliorative inquiry into a concept, my understanding of the goals guiding the inquiry will depend on what I think feminists need from that concept. Accordingly, if I only think about what white, middle-class, cis feminists need, then I will be liable to formulate goals that exclude things that are important to feminists who do not fall within this narrow privileged category. By tacitly starting with an illegitimately narrow conception of the agents of an ameliorative inquiry, then, important desiderata are ruled out of consideration even before the inquiry is properly under way. It follows that when embarking on ameliorative projects, we should be careful to reflect on how we are conceptualizing the agents of ameliorative inquiry.

In the case of a feminist ameliorative inquiry into woman, the ‘we’ whose goals are guiding the ameliorative inquiry must be conceptualized specifically as including feminists who are trans women. Once this is done, the desideratum of respecting trans women’s gender identifications will be included within the goals of the inquiry as part of the broader goal of respecting women’s agency. Thus understood, the framework of an ameliorative inquiry into gender concepts is in fact well placed to respond to the inclusion problem as it manifests in regard to trans women, provided due attention is paid to having an inclusive understanding of the identity of the agents of inquiry. This demonstrates that the problem does not lie with the ameliorative approach itself, which in turn gives grounds for further exploration of the potential of a modified version of Haslanger’s account. It is to this exploration that I now turn.

III. Two Senses of Gender

The findings of the previous section suggest that developing a defensible version of Haslanger’s proposed target concept of woman would require us to reject the focal nature of the account, an implication of which is that gender identity is assigned a secondary or marginal position. I will argue that once a focal account is rejected it is possible to pursue a trans-inclusive feminist ameliorative inquiry while maintaining many of Haslanger’s insights into gender as a hierarchical social system. Haslanger’s proposed target concept, I will suggest, captures one aspect of the phenomenon of gender, which I call ‘gender as class’; there is, however, a second aspect, which I call ‘gender as identity’, which is not captured by that concept. It will be my contention that feminism needs both senses of gender and that a truly inclusive ameliorative inquiry into the concept of woman is only possible when gender as class and gender as identity are given equal consideration. In this section, I first identify the difference between gender as class and gender as identity and then define each in turn. In Section IV, I show how a consciously trans-inclusive ameliorative inquiry can take both senses of gender into account and consider the implications this has for our use of the term woman.

A. Identifying the Gap

As explained in Section I, Haslanger sees gender as a social class system based on presumed sex, within which, on the whole, males are privileged and females are subordinated. The treatment one receives within this system, however, is open to contestation and therefore does not fully determine an individual’s experience. In Haslanger’s words, “there are dominant ideologies and dominant social structures that work together to bias … micro-level interactions, however varied and complex they may be, so that for the most part males are privileged and females are disadvantaged.”28 On this view, the hierarchically defined genders of man and woman are products of a certain social state of affairs: it is one’s positioning within a complex social matrix of practices, norms, institutions, material structures, rationales, and so forth, based on particular understandings of the sexed body, that makes one a woman or a man. Haslanger also emphasizes that the dominant ideologies that shape this matrix do not completely determine our experiences of gender, for they can be contested, resisted, or temporarily elided in various ways.29

Employing language from genealogy, a method Haslanger uses elsewhere, we can say that genders are subject positions that emerge from this matrix.30 Although gendered subject positions emerge only within the matrix of gender practices, and so forth, they can be understood at a local level in ways that differ from the dominant ideology governing that matrix.31 If this is so, then there is a potential gap between the subject position as it is defined by the dominant ideology and the subject position as it is inhabited by an individual. A subject position such as ‘woman’ comes to exist within the context of a set of practices governed by a dominant ideology, but once it exists it can be lived from within in ways that depart from, and may even run counter to, the logic of the system within which it developed. There are thus two aspects of the matrix of practices that constitutes gender to which we need to be able to refer. I use the term ‘gender as class’ to refer to the way that gendered subject positions are defined by dominant ideology and the term ‘gender as identity’ to refer to the way that gendered subject positions are taken up by individuals.32

B. Gender as Class

I will base my definition of gender as class on Haslanger’s proposed target concepts of woman and man. Thus, being classed as a woman is defined as follows:

S is classed as a woman within a context C iff S is marked in C as a target for subordination on the basis of actual or imagined bodily features presumed to be evidence of a female’s role in biological reproduction.

Correspondingly:

S is classed as a man within context C iff S is marked in C as a recipient of privilege on the basis of actual or imagined bodily features presumed to be evidence of a male’s role in biological reproduction.

These definitions are condensed versions of Haslanger’s proposed target concepts of gender. Like those concepts, these definitions are intended to capture what it is to be classed as a woman or as a man by the ideology of gender that is dominant at present, since this is the context in which the current feminist struggle is taking place; other (actual or possible) nonhierarchical practices of gender would require different concepts. Thus, when I refer to ‘being classed as a woman’ in the following discussion, I mean ‘being classed as a woman within the context of current dominant ideology’.

C. Gender as Identity

Having identified the distinction between gender as class and gender as identity and having defined gender as class, the next task is to define gender as identity. Although Haslanger’s account of gender yields a good definition of gender as class, it does not yield a similarly promising definition of gender as identity. According to Haslanger’s brief remarks on the topic, having a ‘feminine gender identity’ is (at least in part) a matter of having internalized norms of appropriate feminine behavior.33 However, this understanding of gender identity does not include people who consider themselves to be women (whether cis or trans) but who do not follow, or feel inclined to follow, norms of feminine behavior. It seems, however, quite possible to have a sense of oneself as occupying the subject position of ‘woman’ without having internalized norms of feminine behavior. This understanding of gender identity is therefore unsuitable for describing the phenomenon I have in mind.

In a different paper, Haslanger offers an account of racial identity that is very different from, as well as much more extensive than, her brief mention of gender identity.34 In the rest of this section, I will develop a definition of gender as identity that draws on this extremely interesting analysis of racial identity and demonstrate the advantages of this definition.

Haslanger is committed to understanding racial identity as an embodied phenomenon rather than a purely intellectual one: “There are important components of racial identity … that are somatic, largely habitual, regularly unconscious, and often ritualised.”35 In order to capture this idea, she uses the image of a map, citing William E. Cross: “In a generic sense, one’s identity is a maze or map that functions in a multitude of ways to guide and direct exchanges with one’s social and material realities.”36 To have a White racial identity, then, is to have a ‘map’ of this kind (which may be “sometimes tacit and unconscious, sometimes more explicit and conscious”) that is “formed to guide someone marked as White through the social and material realities that are (in that context) characteristic of Whites as a group.”37 This does not require that one in fact be marked as White. Indeed, one of Haslanger’s aims in developing the account is to capture the sense in which her own racial identity was changed by the experience of mothering (adopted) Black children: although she is not socially marked as Black, and does not want to say that she is Black, she also feels that her racial identity has been influenced by her close relationship with her children such that it is no longer straightforwardly White.

Applying this model to gender suggests the following definition of gender identity in general:

S has a gender identity of X iff S’s internal ‘map’ is formed to guide someone classed as a member of X gender through the social or material realities that are, in that context, characteristic of Xs as a class.38

This means that having a female gender identity works as follows:39

S has a female gender identity iff S’s internal ‘map’ is formed to guide someone classed as a woman through the social or material realities that are, in that context, characteristic of women as a class.

Gender identity is thus linked to how gender as class operates in the context in which S exists. In the context of current dominant ideology, having a female gender identity means having an internal ‘map’ that is formed to guide someone who is subordinated on the basis of having actual or imagined bodily features that are presumed to be evidence of a female’s role in biological reproduction through the social or material realities characteristic of a person who is so subordinated. In another context in which the class of ‘woman’ was defined differently, what it meant to have a female gender identity would be correspondingly different. Therefore, the account does not hold the meaning of gender identity to be fixed everywhere and forever.40 Since our overall experience of gender is a product of the interaction of dominant systems of meaning and our own subjective experiences, gender as we live it is a function of both gender as class and gender as identity.

This definition of gender identity is importantly different from Haslanger’s account of ‘feminine gender identity’ as a matter of having internalized norms of femininity. On my definition, having a female gender identity does not necessarily involve having internalized norms of femininity in the sense of accepting them on some level. Rather, what is important is that one takes those norms to be relevant to oneself; whether one feels at all moved to actually comply with the relevant norms is a distinct question. This is in line with Haslanger’s understanding of racial identity, as she makes clear: “Although I don’t uncritically accept the norms of the local Black community in deciding on the appearance or behaviour of my children (isn’t everybody’s relationship to their local norms complex and negotiated?), those norms are the ones I daily consider and respond to.”41

To illustrate how this works for a female gender identity, take the example of body hair. Consider a woman who feels that having visible body hair on her legs is unattractive, embarrassing, and unacceptable. In a visceral way, having hairy legs feels wrong for her. This feeling—this instinctive sense of how her body ‘ought to be’—is part of her gender identity. It is in line with a dominant norm of feminine appearance and will therefore enable her to navigate the social and material reality of someone classed as a woman in a way that avoids receiving social censure for violating that norm. Contrast this with the experience of another woman who does not remove hair from her legs. Her awareness of her body includes the awareness that in having hairy legs she is contravening dominant norms of feminine appearance—on some level she knows that people like her are not meant to look like that, according to dominant ideology. This may be so despite the fact that she is perfectly content to have hairy legs and for them to be seen by others. Her experience of social and material reality includes navigating the norm that women should have hairless legs, even though she is not complying with it.

Consider how in a physical sense the hairy-legged woman is doing (or not doing) the same as most men—like most men, she has hair on her legs that she does not remove. But her experience of having hairy legs is not the same as it would be if she identified as a man: if she identified as a man she would not be conscious of violating a norm of feminine appearance, since she would not see those norms as applying to her. This embodied and largely subconscious sense of what it means for her to have hairy legs is therefore part of her female gender identity.

It is perhaps worth noting that gender identity, as I am conceiving of it, has both a subjective and an objective element. It is subjective in that it concerns people’s own sense of which norms are relevant to them and not the fact of which norms other people are most likely to apply to them. On the other hand, it is objective in that there must be some genuine correspondence between the norms people take to be relevant to themselves and the norms associated with the relevant gender class in at least some context, although this correspondence need not be perfect. As a rather artificial limit case, imagine a person who thinks of herself as a woman and sets great store by ‘being feminine’ but who has an entirely idiosyncratic conception of the norms of femininity. This person is not attuned to or aware of any of the norms associated with people who are classed as women in any society (including the very minimal norm that the pronouns she/her/hers, or equivalents in other languages, should be used for these people). Nevertheless, she takes great care to engage in certain behaviors that she thinks of as being ‘feminine’—say, making sure she always wears green socks. According to my account, this person could not be said to have a female gender identity (although her sense of self would of course deserve respect). For an identity to count as a gender identity on my account, it must, as a matter of fact, have at least some contact with the actual norms that are applied to people within some system of gender as class.

As this objective aspect of the account highlights, I am conceiving of gender identity as a response to the social norms that are associated with the social positions that constitute gender as class. A consequence of this is that the nature of those social positions will have implications for gender as identity. The account therefore acknowledges that the oppressive nature of the social position of ‘woman’ plays a role in shaping female gender identity. Nevertheless, the account does not entail that female gender identity is inherently oppressive, for the idea of a map being ‘formed to guide someone marked as a woman through social and material reality’ is a broad one. One way in which a map could guide someone classed as a woman through material and social reality is by guiding her toward behaviors that are prescribed as ‘feminine’ by dominant ideology. In this case, it may be a form of internalized oppression, because many norms of feminine behavior encode and express women’s subordination.42 Equally, though, a map could guide someone classed as a woman through material and social reality by guiding her to resist norms of acceptably feminine behavior. In this case, it may be enabling her to reject subordinating ways of being, thereby forming part of an emancipatory project. For example, if the hairy-legged woman described above experiences her violation of dominant norms of feminine appearance as something positive that contributes to her sense of herself as a feminist, then this aspect of her gender identity will have an emancipatory function for her. On my account, then, a female gender identity is neither inherently oppressive nor inherently emancipatory (nor even inherently neutral) but has the potential to function in any of these ways depending on how it guides the person through material and social reality.

Crucially for present purposes, this definition of gender identity entails that all trans women have a female gender identity. The definition allows that identifying as a woman can mean many different things for different people (and indeed for the same person at different times). This is because the phrase ‘social or material reality’ applies to a broad range of aspects of one’s embodied existence, so that having an internal map that is formed to guide someone marked as a woman can mean different things depending on which aspects of existence the map is picking up on. This includes how one’s body looks and feels and how other people relate to one. The definition therefore respects all trans people’s gender identifications in a way that is compatible with different understandings of trans identities. For instance, for one trans woman, having a female gender identity may be primarily a matter of feeling that she ought to be treated in a certain way by others, for example, that people should refer to her using feminine pronouns and a particular name. For another trans woman, having a female gender identity may be primarily a matter of having the sense that her bodily features ought to be a certain way, for example, that she ought to have a vulva and not a penis and testes.43 Both of these cases concern aspects of social/material reality that are characteristic of those classed as women and thus fall under the definition of a female gender identity. The definition therefore entails that all trans women have a female gender identity, without relying on the false assumption that all trans women experience their female gender identities in the same way. Moreover, the definition does not require that one is aware of one’s gender identity; it therefore allows that a trans woman can have had a female gender identity before consciously coming to identify as a woman.44

IV. Inclusive Amelioration

So far, I have defined two senses of gender: a sense of gender as class based on Haslanger’s proposed target concept of gender and a sense of gender as identity that draws on her account of racial identity. I will now argue that both of these two senses of gender are required for a successful ameliorative inquiry into the concept of woman, such that they should be considered ‘twin’ target concepts.

A. Class, Identity, and Oppression

The aim of a feminist ameliorative inquiry into gender concepts, as will be recalled from Section I, is to locate the versions of these concepts that will be most useful in opposing gender-based injustice. From this point of view, there is no reason to privilege gender as class over gender as identity, because both gender as class and gender as identity are relevant to understanding the oppression of women. On the one hand, understanding how systems of domination function to oppress certain people will require us to be aware of how subject positions are defined according to dominant ideology. On the other hand, understanding how individuals experience oppression will require us to be sensitive to the ways in which subject positions are taken up by those who inhabit them. In the case of gender, then, some forms of sexist oppression affect those who are classed as women, regardless of how they identify, while others affect those who identify as women, regardless of how they are classed.45

Taking employment as an example, one form of oppression is outright discrimination, such as not giving a job to a woman candidate, or offering it to her with a lower salary, simply because she is a woman. This form of oppression will potentially affect those people who are classed as women within the relevant context, because it is a matter of how one is perceived by others. Another form of oppression is internalized negative stereotypes of women’s capabilities that might lead a woman to, for instance, choose a lower-paid and stereotypically ‘feminine’ career over a higher-paid and stereotypically ‘masculine’ one, or not to apply for a leadership role within her current career. This form of oppression will potentially affect those people who have a female gender identity, because it relates to how one perceives oneself.46

This means that if feminists overlook gender as class, we will not be able to explain why some people are targeted for certain forms of gender-based oppression and others are not. We will also find it difficult to understand why some gender-based oppressions take the form that they do—for example, the sexist belief that women of a particular age should not be hired for demanding jobs because they may become pregnant. Conversely, if we overlook gender as identity, we will not be able to explain how oppression can operate through self-policing behavior even in the absence of external coercion. We will also find it difficult to understand how people can experience certain forms of gender-based oppression even if they are not classed as women. Thus, the concepts of gender as class and of gender as identity are equally important for fulfilling feminist aims. I therefore contend that, contra Haslanger, the concept of gender as identity should not be assigned a secondary or peripheral status within a critical feminist analysis of gender but should have equal status with the concept of gender as class.

It is also important to recognize that gender as class and gender as identity should not be thought of as sharply dichotomous phenomena best discussed separately. In the main we need to be aware of both together, although it may be useful, at specific moments, to foreground one or the other for the purposes of understanding a specific experience, as I have been doing in this section. This is not a point of tension, for it is perfectly possible to discuss different aspects of a system while at the same time remaining aware that it is one system.

The claim that feminism requires two senses of gender with equal theoretical status seems to sit oddly with the understanding of ameliorative inquiries outlined at the start of this article. If there are two equally important senses in which one can “be” a woman, that is, being classed as a woman and having a female gender identity, what exactly is the target concept of woman? If we still have two concepts on our hands, does this mean that the ameliorative inquiry simply has not yet concluded? In response to these questions, I propose to maintain the basic idea of an ameliorative inquiry while making it more pluralist. This means revising the definition so that the inquiry is understood as asking: “what concept or combination of concepts of F-ness (if any) best serves our legitimate purposes?” This revised definition allows for the possibility that it may turn out to be the case that no single concept can, by itself, meet all of the legitimate purposes. An ameliorative inquiry may, on this understanding, arrive at two (or more) distinct but equally important concepts. Such an inquiry can be thought of as taking a ‘branching’ route, starting with one set of goals but arriving at multiple target concepts. In saying that feminism needs both senses of gender, then, I am saying that the ameliorative inquiry in fact branches so as to deliver the twin target concepts being classed as a woman and having a female gender identity, both of which deserve equal status within feminist theory.

Where does this leave us with regard to respecting trans women’s gender identifications? As argued above, my definition of gender identity entails that all trans women have a female gender identity. By contrast, however, my definition of gender as class, since it is based on Haslanger’s target concept of gender, does not categorize all trans women as women. A trans woman will only count as being classed as a woman if she is perceived to have bodily features presumed to be evidence of a female’s role in biological reproduction and is subordinated on this basis—and, as discussed in Section II.A, this applies to some trans women and not to others.47 However, as I argued in Section II.B, the reason that Haslanger’s proposed target concept of gender marginalizes trans women is that it both (1) does not categorize all trans women as women and (2) purports to capture the primary or central sense of gender so far as feminism is concerned. In the branching version of the inquiry, both the concept of being classed as a woman and the concept having a female gender identity are considered to be core or central to feminism. It follows that on my account trans women are categorized as women in a core sense of the term, not in a secondary or peripheral sense. Therefore, they are not marginalized within the account.48 This means that Haslanger’s original response to the issue of inclusion (i.e., her claim that the account includes everyone who ought to be included for the purposes of feminism) is now viable.

B. Reaching Terms

A branching ameliorative inquiry must have something to say about how terms should be allocated between the various concepts. The question therefore remains of how the two target concepts of woman should be matched with linguistic terms. This question is extremely important because the terms ‘woman’ and ‘man’ carry great significance for many people. So which sense of gender, if either, can most profitably be associated with the term ‘woman’? Or is ‘woman’ best used for both senses, clarifying as needed?

According to Haslanger, the question of how terms should be matched with concepts in an ameliorative inquiry has semantic, pragmatic, and political aspects. Regarding the political aspect, she states that “the politics of such appropriation will depend on the acceptability of the goals being served, the intended and unintended effects of the change, the politics of the speech context, and whether the underlying values are justified.”49 A full analysis of the semantic, pragmatic, and political implications of various options for deciding how the term ‘woman’ can most productively be used by feminists is beyond the scope of this discussion. Nevertheless, I believe that there are two highly salient political considerations that weigh heavily in favor of using the term to refer to people with a female gender identity and not, in general, using it to refer to people classed as women. First, I think that an unintended effect of using ‘woman’ to refer to gender as class (even if it were also used to refer to gender as identity) would be the reinforcement and perpetuation of the existing marginalization of trans women within feminism.50 Second, it seems to me that the politics of the speech context are such that cis women have more access to power within feminist spaces than trans women, so that any use of terms that further marginalizes trans women is to be strenuously avoided. Both of these considerations support using ‘woman’ exclusively for the concept of having a female gender identity.

Furthermore, I am not convinced by the reasons Haslanger gives for using the term ‘woman’ to refer to those classed as women by dominant ideology. She intends her definition to “articulate a negative ideal that challenges male dominance.”51 Using ‘woman’ in this way invites people to give up their attachment to that gendered subject position. However, the invitation to give up identifying with the term ‘woman’ will tend to have a different meaning for cis women, who are pushed toward this label, than for trans women, who typically face a struggle to claim it. Giving up one’s attachment to thinking of oneself as a woman may feel very different if one’s right to identify in that way has received at best a partial or patchy recognition. Thus, even if ceasing to identify with the label ‘woman’ may be a constructive option for some women, it will very probably be undesirable or unworkable for many others, including at least some trans women. Therefore, conflating an invitation to oppose hierarchical gender positions with an invitation to move away from identifying with the term ‘woman’ may actually hinder feminist aims by creating division: all women have reason to do the former, but not all women will be in a position to do the latter. Moreover, the disaffiliation Haslanger seeks to promote is not necessary, given that gender identity is not determined by the dominant system of meaning and that there is room for resistant interpretations of gender. Ceasing to identify with the label ‘woman’ is not the only way of resisting being positioned as a woman by the dominant system of meaning: another option is to take up the gender identity of ‘woman’ in a resistant way. This second option does not have the same exclusionary implications as ceasing to identify as a woman altogether and would be furthered by using the term ‘woman’ to refer to people with a female gender identity.

I do not wish to overstate my disagreement for Haslanger on this point, for she explicitly states that she is willing to be flexible about the use of terms: “I think there are rhetorical advantages to using the terms ‘gender’, ‘man’ and ‘woman’ … for the concepts I’ve defined, but if someone else is determined to have those terms, I’ll use different ones.”52 What I am suggesting, then is that feminists ourselves should be determined to have the terms ‘man’ and ‘woman’ for the concept of gender identity and not for the concept of gender as class, for the reasons given above. I do not have a clear proposal for what terms should be used to talk about gender as class. Perhaps using the phrase ‘being classed as a woman’ would work well, or it could even be defined more explicitly as ‘someone who is subordinated on the basis of being presumed to be female’. Different locutions may even work best in different contexts, and a certain amount of stipulation will inevitably be required. The issue is not, I think, one that can be settled a priori but will be best explored through engaged and inclusive feminist activism.

Allocating the term ‘woman’ to gender identity may, at first blush, sound like a move away from the twin-concept model I have been advocating. In order to understand why this is not the case, it must be recognized that the considerations that motivate this allocation of terms, although immensely important, are entirely contingent. It is an unfortunate fact that the recent history of feminism includes a significant degree of exclusion and even hostility toward trans women.53 Moreover, society at large displays great amounts of transphobia and cissexism. Given that this is so, I find it hard to imagine a situation in which ‘woman’ could be used to refer to people classed as women without detriment to trans women. If these two factors weren’t in play, however, it seems to me that it would be possible to use the term ‘woman’ in that way. Imagine a context in which trans people’s identities were respected, such that it was understood that our relations with individual people should always be informed by gender as identity rather than gender as class (with the upshot, e.g., that correct pronouns were always used). Against this background, within a specific discussion focused on gender as class, the term ‘women’ could, in my view, safely be used to refer to people classed as women. Thus, the case for using ‘woman’ exclusively for gender as identity rests on the fact that this background context is never present, given the way things are now and the history from which we are moving forward. This is an observation about words, about power dynamics, and about speech contexts: it is not about what concepts we need. As far as concepts are concerned, then, feminists need both the concept of gender as identity and the concept of gender as class, and our need for each is equally great. Thus, there is no contradiction in advocating twin concepts while prescribing an asymmetric allocation of terms.

C. Inclusive Amelioration in Action

In order to better illustrate the overall approach that I am advocating, and to show how it might work in practice, let me close by offering an example from my own experience. Three years ago I was involved in organizing a Reclaim the Night march. Reclaim the Night is a protest against ‘violence against women’: the premise of the march is that violence and the threat of violence, especially sexual violence, are among the most significant ways in which women are subordinated. As an organizing committee, we agreed that we wanted to make the march women-only due to the symbolic value of conspicuously violating the social norm that a woman ought to be accompanied by a man when walking after dark—a norm that substantially limits women’s freedom and is often invoked in the context of victim blaming. But who counts as a woman for this purpose? In other words, what did we really mean when we said that we wanted the march to be ‘women-only’? This was the subject of a lot of discussion. There was unanimous agreement that the sense of ‘woman’ we had in mind included all trans women. We decided to use the term ‘self-defining women’ to highlight explicitly that this was the case. However, this didn’t capture everything that we wanted it to: we recognized that there might be some people who did not identify as women but who were, in a very real sense, targets of the kind of violence and threat of violence against which our protest was directed. We felt both that these people could legitimately expect to be included in our protest and that our protest could only be strengthened by their presence. The kind of people we had in mind were primarily nonbinary people who had been assigned female at birth and trans men who felt that they were regularly misgendered as women, thereby becoming targets for violence directed at women. We tried to find a broader concept that would capture what we meant but could find none that was sufficiently specific. For example, we considered making the march open to ‘those who consider themselves to be affected by violence against women’ but rejected this on the grounds that many cis men would (rightly) consider themselves to be ‘affected’ by violence against women in virtue of the way it harmed those for whom they cared and more generally insofar as it is as a grave injustice taking place in a society to which they belong. Positive as their intention might be, having these people participate in the march would undercut the symbolic power of our action, which we all agreed was very important.

Eventually we settled on the following wording for our publicity: “The march is open to all self-defining women. If you do not define as a woman but experience discrimination because you are perceived as female, you are also welcome to attend.” Here, the phrase ‘self-defining women’ captures gender as identity, while the rest of the wording captures gender as class.54 Neither concept of gender by itself could have expressed the sense in which we wanted the march to be ‘women-only’, nor would any single broader concept do the job: we had to appeal to a disjunctive description. To reword our sentiments in terms of the analysis offered in this article, we were of the view that ‘violence against women’ is a form of oppression that operates both through gender as identity and through gender as class, affecting both those who are classed as women and those who have a female gender identity. Accordingly, we needed to refer disjunctively to both gender as identity and gender as class in our explanation of who was invited to participate in the march. Incidentally, as soon as this description was proposed (not by myself), it commanded universal agreement. The experience of participating in this discussion has helped to shape the arguments presented in this article.

V. Conclusion

I have objected to Haslanger’s proposed target concept of woman on the grounds that it fails to respect trans women’s gender identifications, and I have proposed a pair of target concepts that do not suffer from this problem. Although I take Haslanger’s account of social identity categories as a valuable starting point, the analysis of woman I propose differs from hers in that it moves away from a focal analysis toward a more pluralist understanding of ameliorative inquiries. On this model, ameliorative inquiries can follow a branching route that delivers multiple target concepts that enjoy equal theoretical status. In the case of woman, this means that the task of discovering which senses of gender are ‘central’ and which ‘secondary’ becomes part of the ameliorative project itself rather than something that is fixed at the outset.

Working within this model, I have identified two senses of gender, gender as class and gender as identity, that give rise to twin target concepts. These twin concepts are the concept of being classed as a woman, which is defined as ‘being targeted for subordination on the basis of actual or imagined bodily features presumed to be evidence of a female’s role in biological reproduction’, and the concept of having a female gender identity, which is defined as ‘having an inner map that is formed to guide someone classed as a woman through the social and material realities of someone who is so classed’. For pragmatic reasons, I advocate using the term ‘woman’ to express the concept of having a female gender identity and not using it to express the concept of being classed as a woman.

My account of gender includes within the category of women everyone who needs to be included for the purposes of feminism. This analysis of the concept of woman allows for a feminism that engages critically with the social reality of gender as class while at the same time taking seriously the agency of trans women, and other trans people, by respecting their gender identifications.

Notes

*. I would like to thank the following people for helpful comments on earlier versions of this article: Beatrice Balfour, Elizabeth Barnes, Talia Bettcher, Shannon Dea, Miranda Fricker, Sally Haslanger, Kathryn Maude, and Jennifer Saul. I am very grateful to Rachel McKinnon, a then-anonymous reviewer for this journal, to one other anonymous reviewer, and to the editors, for their insightful and constructive comments. I am also grateful to participants in the ARCHE Workshop on Haslanger on Ameliorative Projects and to members of the University of Nottingham Feminism Reading Group, where earlier versions of this article were presented.

1. For an overview of this extensive debate, see Mari Mikkola, “Feminist Perspectives on Sex and Gender,” in The Stanford Encyclopedia of Philosophy, ed. Edward N. Zalta (Stanford, CA: Stanford University, 2012).

2. Sally Haslanger, Resisting Reality (Oxford: Oxford University Press, 2012).

3. Thus, different groups with different aims could undertake ameliorative inquiries into the same concepts and arrive at different results.

4. Sally Haslanger, “Gender and Race: (What) Are They? (What) Do We Want Them to Be?” in Resisting Reality, 221–47.

5. My focus in this article will be narrowly on the problem of trans marginalization. For a broader discussion of some problems facing Haslanger’s proposed target concepts, see Mari Mikkola, “Gender Concepts and Intuitions,” Canadian Journal of Philosophy 39 (2009): 559–84.

6. Talia Mae Bettcher, “Evil Deceivers and Make-Believers: On Transphobic Violence and the Politics of Illusion,” Hypatia 22 (2007): 43–65.

7. Haslanger, “Gender and Race,” 228.

8. Ibid., 235.

9. Ibid., 234. Haslanger herself first explains being a woman and then explains functioning as a woman; I have reversed the order of exposition here for the sake of concision and because I will be talking in more detail about her account of functioning as a woman, for reasons that will become clear.

10. Ibid., 240.

11. I reserve judgment on whether the marginalization of nonoppressed females is a problem. However, my revised analysis aimed at ensuring trans inclusion will also have the consequence of including nonoppressed females as women. For more on this issue, see Mikkola, “Gender Concepts and Intuitions.”

12. I am grateful to Talia Bettcher for bringing the issue of wrongful inclusion to my attention.

13. Haslanger, “Gender and Race,” 237. As this quote highlights, Haslanger uses a sex/gender distinction and corresponding female/woman terminology. Although I reproduce some of this language when discussing Haslanger, invoking this infamously troubled distinction in the context of a discussion of trans identities strikes me as particularly problematic, and I move away from it in my own account below.

14. For example, even if the woman in this scenario is being subordinated in virtue of expressing femininity, she would still not satisfy Haslanger’s definition. See Julia Serano, Whipping Girl: A Transsexual Woman on Sexism and the Scapegoating of Femininity (Emeryville, CA: Seal, 2007).

15. That is, not trans.

16. For an explanation of why these conditions raise problems concerning intersectionality, see n. 18 below.

17. Talia Mae Bettcher, “Trans Identities and First-Person Authority,” in “You’ve Changed”: Sex Reassignment and Personal Identity, ed. Laurie J. Shrage (New York: Oxford University Press, 2009), 98–120, and “Trans Women and Interpretative Intimacy: Some Initial Reflections,” in The Essential Handbook of Women’s Sexuality, ed. D. Castenada (Santa Barbara, CA: Praeger, 2013), 51–68.

18. For example, a trans woman might find herself in scenario 3 in all of her social interactions but in scenario 4 in the context of medical treatment she is receiving. The ability to be perceived by others in the relevant way for being in scenario 3 will moreover be dependent on various social factors including economic status, since there are financial costs involved in accessing things that tend to support this perception, such as hormones, surgery, and certain kinds of clothing; see Emi Koyama, “The Transfeminist Manifesto,” in Catching a Wave: Reclaiming Feminism for the 21st Century, ed. Rory Dicker and Alison Piepmeier (Boston: Northeastern University Press, 2003), 244–59; Rachel McKinnon, “Stereotype Threat and Attributional Ambiguity for Trans Women,” Hypatia 29 (2014): 857–72. This means that there are class and other intersectional issues raised by the exclusion of some trans women from Haslanger’s definition of woman, which is particularly troubling.

19. The account of gender identity and oppression that I will go on to give also entails that trans women in scenarios 1 or 2 can experience oppression qua women. See n. 45.

20. This is not to deny that it may be troubling to deny that some prima facie cis women are women; see n. 11 above.

21. Sandy Stone, “The Empire Strikes Back,” in The Transgender Studies Reader, ed. Susan Stryker and Stephen Whittle (New York: Routledge, 2006), 221–35; Serano, Whipping Girl; Koyama, “The Transfeminist Manifesto”; C. Jacob Hale, “Tracing a Ghostly Memory in My Throat: Reflections on Ftm Voice and Agency,” in Shrage, You’ve Changed, 43–65; Bettcher, “Trans Identities and First-Person Authority.”

22. Haslanger, “Gender and Race,” 228.

23. I here grant for the sake of argument the assumption that gender identity is a trans-inclusive concept, but in Sec. III below I show that Haslanger’s definition of feminine gender identity would in fact exclude many women, trans and cis alike. I imagine here that gender identity is defined in an inclusive way in order to show that even if this were so, the account as it stands would still be problematic.

24. See Sec. I.

25. See Naomi Scheman, “Queering the Center by Centering the Queer: Reflections on Transsexuals and Secular Jews,” in Feminists Rethink the Self, ed. Diana Tietjens Meyers (Boulder, CO: Westview, 1997), 124–62.

26. Haslanger, “Gender and Race,” 227.

27. Sally Haslanger, “What Are We Talking About? The Semantics and Politics of Social Kinds,” in Resisting Reality, 365–80, 376.

28. Haslanger, “Gender and Race,” 233.

29. Ibid., n. 14. Haslanger uses ‘dominant ideologies’ as a placeholder for a more specific account ‘to be decided on later’; I also follow this strategy. Readers are therefore invited to insert their preferred account of ideology, e.g., in terms of “background,” “hegemony,” or “habitus.”

30. Haslanger, “What Are We Talking About?”

31. There also is a ‘looping effect’ here, in that different ways of understanding a subject position can prompt subjects to act differently, which may in turn affect the matrix that generates the subject position in the first place; see Sally Haslanger, “Ideology, Generics and Common Ground,” in Resisting Reality, 446–77, 465–67; Ian Hacking, The Social Construction of What? (Cambridge, MA: Harvard University Press, 1999).

32. The notion of being ‘classed as’ a certain gender is related to the notion of ‘gender attribution’ as considered by Suzanne J. Kessler and Wendy McKenna, Gender: An Ethnomethodological Approach (New York: Wiley, 1978). However, the two are distinct. ‘Gender attribution’ refers simply to being categorized as male/man or as female/woman by others. The notion of being ‘classed as’ a certain gender, on the other hand, refers to the fact of being categories as a male/man or as female/woman by others together with the social implications of being so categorized in a particular context. Thus, the notion of ‘gender as class’ incorporates a critical stance, whereas ‘gender attribution’ is primarily a descriptive notion (although it may well be used in critical ways).

33. Haslanger, “Gender and Race,” 228. Haslanger’s use of ‘feminine’ is intended to indicate that she is referring to gender rather than to sex. In my own account I instead use the phrase ‘female gender identity’ to refer to a gender identity of ‘woman’, because the connotations of the term ‘feminine’ do not sit well with the definition I propose, as will become clear.

34. Sally Haslanger, “You Mixed? Racial Identity without Racial Biology,” in Resisting Reality, 273–97.

35. Ibid., 284–85.

36. William E. Cross Jr., Shades of Black: Diversity in African-American Identity (Philadelphia: Temple University Press, 1991). Quoted in Haslanger, “You Mixed?” 290.

37. Haslanger, “You Mixed?” 291.

38. The phrase ‘formed to guide’ should not be read as implying that the map has been intentionally fashioned to fulfill this purpose. In the sense I have in mind, a map is formed to guide someone through something if it is formed in such a way that it actually functions to so guide that person.

39. I am talking about ‘having a female gender identity’ rather than ‘identifying as a woman’ in order to avoid implying that this is necessarily a conscious or active process. It could also be useful to be able to talk about ‘identifying as a woman’ as an action that a person can take on top of the fact of having a female gender identity, something like ‘consciously or actively affirming one’s female gender identity’. For instance, a trans woman who had recently come to think of herself as a woman might want to say that she had always had a female gender identity but that she had only recently begun to identify as a woman. The way I am using terms here allows for this claim to be made. My usage is also in line with McKinnon’s suggestion that differentiating between ‘woman’ (gender-term) and ‘female’ (sex-term) is unhelpful for trans people and for intersex people (“Stereotype Threat and Attributional Ambiguity for Trans Women”).

40. My definition of gender as identity allows that someone’s gender identity may be mixed, fluid, or nonbinary. Briefly, a mixed identity might be characterized thus: “S has a mixed gender identity iff S’s internal ‘map’ is partly formed to guide someone classed as a woman through the social or material realities that are, in that context, characteristic of women as a class and partly formed to guide someone classed as a man through the social or material realities that are, in that context, characteristic of men as a class.” A fluid gender identity might be characterized thus: “S has a fluid gender identity iff S’s internal ‘map’ is at times formed to guide someone classed as a woman through the social or material realities that are, in that context, characteristic of women as a class and at other times is formed to guide someone classed as a man through the social or material realities that are, in that context, characteristic of men as a class.” A nonbinary identity might be characterized thus: “S has a nonbinary gender identity iff S’s internal ‘map’ is neither formed so as to guide someone marked as a woman through the social or material realities that are, in that context, characteristic of women as a class nor formed to guide someone classed as a man through the social or material realities that are, in that context, characteristic of men as a class.” Each of these possibilities, of course, requires more elaboration and exploration than I can give them here. Haslanger considers nonbinary and mixed identities in some detail in relation to racial identity; Haslanger, “You Mixed?”

41. Ibid., 289.

42. Sandra Bartky, Femininity and Domination: Studies in the Phenomenology of Oppression (New York: Routledge, 1990).

43. Of course, these are oversimplified examples; in reality, everyone’s gender identity is likely to be extremely complex and include elements that concern one’s body and elements that concern social interactions.

44. See n. 39 above.

45. Note that a woman in scenarios 1 or 2 as discussed in Sec. II.B may well experience identity-directed oppression as a woman even if she does not experience class-directed oppression as a woman.

46. This is, of course, not to say that a certain form of oppression is liable to affect all people with a female gender identity in the same way. For an excellent and highly relevant discussion of the specific ways in which stereotype threat affects trans women, see McKinnon, “Stereotype Threat and Attributional Ambiguity for Trans Women.”

47. The trans woman in scenario 4 presents an interesting case for my account. As I noted in my initial discussion of the case, it seems counter-intuitive to say that this woman is not functioning as a woman, for she is being perceived by all those around her as a woman. At first glance, my account seems to entail that this woman is not classed as a woman, although she has a female gender identity—which is still counter-intuitive. This is technically true; however, recall that my account of gender as class is an account of the dominant ideology of gender only. I suggest that the members of the community in scenario 4 hold an understanding of gender that is different from the dominant one. The trans woman in this scenario, then, is classed as a woman by the ideology of gender operating in that limited context but is not classed as a woman by the dominant ideology of gender and therefore does not count as being classed as a woman on my definition.

48. Note that having a female gender identity does not require one to be in fact subordinated. This means that a woman who is not subordinated at all and therefore does not count as a woman in the class sense may still count as a woman in the gender identity sense. This ensures that if there are any prima facie women, trans or cis, who are not subordinated at all and who are not classed as women for this reason, their gender identities will still be respected by the account.

49. Haslanger, “Gender and Race,” 225.

50. In calling this effect ‘unintended’, I mean that it would occur whether or not it was intended; deplorably, it might also in some cases be an intended effect.

51. Haslanger, “Gender and Race,” 240.

52. Ibid., 246.

53. For the most egregious example, see Janice Raymond, The Transsexual Empire: The Making of the She-Male (Boston: Beacon, 1979).

54. Strictly speaking, the phrase ‘those with a female gender identity’ may have been preferable to the phrase ‘self-defining women’ (see n. 39 above).