Feminist Detachment from the Ordinary

Feminist thought has long been dealing with the problem of defining 'woman'. And, indeed, supposing for a moment - bold move - that women are the subject of feminism, it becomes necessary to understand who we are speaking about. In this article, I argue that such endeavour is still called for and that Wittgenstein's ordinary language philosophy can help us clarify the problems over which the debate around the meaning of 'woman' is founded. To do so, I begin by surveying Wittgensteinian “descriptive” analyses of the concept of 'woman'. These analyses, I later contend, must be at the basis of any “ameliorative” approach to the debate. Simultaneously, I will be arguing in favour of a Wittgensteinian view of philosophy as a critical exercise that should inform debates which must ultimately be settled politically, i.e., in pursuance of some political purpose.

While the first feminist wave did not problematise 'woman' as a place of conflict, the sex-gender distinction that established the movement's second wave, along with developments in social constructionist thought, actuated the debate about identity concepts.¹ From the 1960s to the 1980s, particularly in the North-American context, it was common among feminist theorists to endorse realist positions, in which they grounded their political work and practice: one was a 'woman' insofar as they shared particular features or lived under specific conditions. While it was generally assumed that these were largely socially or culturally constructed characteristics, some of the stances associated with philosophical realism took on a highly essentialist character that both homogenised the concept of 'woman' and naturalised certain gender norms.² Realist conceptions of 'woman' became subject to contestation from the 1980s onwards, as a growing concern with intersectional forms of oppression led some authors to assert that “the category of 'women' is normative and exclusionary and is invoked with the unmarked dimensions of class and racial privilege intact” (^{Butler 1990}, 14). This was the case with post-structuralist philosophers, whose emphasis on the oppressive normative dimension of language often resulted in a suspicious attitude towards any sort of classification that made it virtually impossible to posit any commonality between women. Some feminists have thus looked at this form of scepticism with disquiet, as it resulted in a fragmentation of the category of 'woman' found to be adverse to political action itself.³ The radical deconstruction of the concept as a fiction or as a discursive practice of gender was now regarded as negligent, as it would render inviable the very conceptualisation of oppression as a structural process, systematically directed toward specific groups. As Linda ^{Zerilli (2003}, 132) puts it, “How can feminism organize in the name of no one and without a sense of collective interests? […] The essentialist scare of the 1980s gave way to the post-structuralist scare of the 1990s, in which feminist critics of foundationalism were accused of having taken things too far.”

Former theoretical debates over the meaning of 'woman' have recently evolved into one of the most delicate issues of contemporary feminist practice, namely that of considering whether trans women ought to be included in the category of 'woman'.⁴ Current discussion of the matter contemplates more than mere metaphilosophical analysis over categories and kinds, assessing matters of great practical import such as the stipulation of criteria for legal gender reassignment, access to women-only spaces and resources, protection from violence for both natal and trans women, etc. Although I am not going to engage with the extensive practical consequences of such debate here, I find the terms of this contemporary dispute still reflect the same problems observed in the long-standing debate around the concept of 'woman'. More so, I find they reflect the very problems Wittgenstein has diagnosed in traditional philosophy.

If we look at the problem of defining 'woman' as a linguistic problem, it is useful to consider Wittgenstein's claim that “philosophical problems arise when language goes on holiday” (PI § 38), which suggests that the philosophical practice of abstracting concepts from their ordinary use often creates the very misunderstandings it seeks to resolve. Considering that an investigation of language provides important insights about the world we are to act on, Wittgenstein acidly blames “the general concept of the meaning of a word” for surrounding “the working of language with a haze which makes clear vision impossible” (PI § 5). Alternatively, he argues that for “a large class of cases of the employment of the word ‘meaning’ […] this word can be explained in this way: the meaning of a word is its use in the language” (PI § 43). And its use, far from being uncommitted to certain theoretical frameworks, is nevertheless more complex and flexible than theory makes it out to be.

Many of the conceptual problems we are faced with in feminist thought have often had the same origins as the philosophical problems Wittgenstein diagnoses: an excessive theorisation that leads us to abstract concepts and strips them of their actual use in language. In a reassessment of her own position within the debate over the meaning of ‘woman’, ^{Zerilli (2003}, 133) similarly states that “feminist theory has not been exactly immune to this theoretical practice of decontextualization and abstraction from the ordinary”. When examining the meaning of ‘woman’, both realist and post-structuralist theorists have delved into an abstraction of the concept either by pursuing an ultimate, neutrally attained definition, or by taking the irrevocably local character of our linguistic practices to vaticinate the abandonment of any claim to objectivity. On either account, the very notion of (objective) meaning has been too tightly associated with abstract modes of understanding which have kept us “riveted to - indeed obsessed with - searching out and destroying” the supposed fixity of the category ‘woman’ (^{Zerilli 2003}, 147).

A similar deadlock is observable in some contemporary debates between trans advocates and gender critical feminists, where two competing views often fail to engage with each other. Even though these discussions are not circumscribed to a theoretical sphere - they have recently taken over social media and other media outlets -, they often seem to be endowed with the sort of abstract reasoning that has characterised philosophical disputes. Consider the question at the core of the debate to be: is ‘woman’ a gender concept or a sex concept?⁵ Thus formulated, the question is ambiguous enough - because intrinsically abstract - to jeopardise our answer at the outset. On a more careful examination, however, two questions seem to actually be implicit here: for one thing, we may want to pose a descriptive question: what does ‘woman’ mean? While this formulation might well be motivated by the desire to uncover the concept's presumed essential meaning, it may also be tracking its legal meaning, its dictionary definition, its mainstream uses, etc. Alternatively, we may formulate an ameliorative question: what do we want ‘woman’ to mean? As we will see, this question can be answered in different ways depending on the purposes according to which it is being considered, but it always presupposes some sort of a priori knowledge about the concept's descriptive meaning, i.e., about its use.⁶ In what follows, I will argue that an adequate descriptive analysis of ‘woman’ is key to any ameliorative approach that follows. I believe the distinction between these two types of approaches is crucial to guide us through the debate, and I suggest we look at the two questions through the lens of Wittgenstein's ordinary language philosophy.

The Descriptive Question: ‘Woman’ as a Family Resemblance Concept

Recently, some feminist philosophers have tried to approach the debate over the meaning of ‘woman’ from a new angle. Acknowledging the problematic character of traditional practices of abstraction and subsequent essentialisation of meaning, they have found an unlikely ally in Ludwig Wittgenstein. Indeed, Wittgenstein's later philosophy allowed them to elaborate an internal methodological critique and urge fellow feminists to go “back to the rough ground” of ordinary language (PI § 107). But the ordinary, it seemed, had been rejected as too flawed or too oppressive to be taken into consideration. What, then, might the role of ordinary language in accounting for meaning be?

In Philosophical Investigations, Wittgenstein rejects a naturalised vision of concepts thought to pre-exist language, a misunderstanding he believes had permeated the bulk of western thought. In particular, he is interested in challenging the dogmatic premises that grounded the scientific philosophy of his time and the metaphysical legacy upheld by twentieth century logicism. Underlying this critique is the recognition that we are dominated by a certain vision of the world, conveyed by Wittgenstein in passages such as the following: “A picture held us captive. And we couldn't get outside it, for it lay in our language, and language seemed only to repeat it to us inexorably” (PI § 115). The “picture” Wittgenstein refers to is, strictly speaking, the idea of a canonical language - a dominant philosophical model since at least Plato - whose perfection natural languages could only feebly search for. Calling this model into question, Wittgenstein's later ordinary language philosophy targets the dogmatic character of a conception of language which sustains the existence of a necessary logical structure of the world and fails to reflect our actual use of language (PI § 94).

In a Wittgensteinian account, then, a descriptive question about the meaning of ‘woman’ is a question about its use in ordinary language. Feminist philosophers such as Cressida Heyes, Marilyn Frye and Natalie Stoljar, among others, thoroughly engage with the idea that, in order to grasp the meaning of a word, we must “look and see” how it is used in language (PI § 66). Starting from the “rough ground” of ordinary language, they recognise the different uses the concept of ‘woman’ acquires in our discourse, examining the limits of its application. Their purpose is “to bring words back from their metaphysical to their everyday use” (PI § 116) and thus free us from the captivity of a dogmatic picture of language which has hampered the possibility of dialogue. Employing Wittgenstein's descriptive method to account for the meaning of ‘woman’, they take seriously his argument that what we find when we look at the world are mere resemblances which form assortments that language seeks to define, but which do not necessarily share a single common denominator. Not unlike the way members of a family are closely - and often genetically - related, so too different uses of some concepts share resemblances with each other. The Wittgensteinian notion of “family resemblances” has thus become a key element of feminist Wittgensteinian philosophies, enabling feminist authors to explain how the various senses of a concept such as ‘woman’ relate, without reducing it to a final static meaning.

Cressida ^{Heyes (2000)} offers the most thorough Wittgensteinian approach to the problem of defining ‘woman’ in feminist theory. By understanding similarities between women as family resemblances, she draws on this notion to present a feminist social ontology that seeks to sustain a cogent political alternative. According to Heyes, ordinary language philosophy provides feminist studies with a method that allows for important generalisations to feminist practice, while ensuring they do not incur in false or violent uniformization. It does so by means of a clarification exercise that registers the ambiguous and contested nature of many concepts, including that of ‘woman’. Heyes's version of Philosophical Investigations in “a feminist voice” thus reads:

Let us consider the construct we call “women.” I don't just mean white, middle-class, heterosexual, able-bodied, young, attractive, western women, but all women. What is common to them all? Don't say: “there must be something in common or they wouldn't be called ‘women.’” Likewise, don't say: “If women have nothing in common, then how can feminism form a political movement?” Look and see what the construct of women consists of, and what women might have in common. For, if you look, you will not see something that is common to all, but similarities, relationships, and a whole series of them at that. (^{Heyes 2000}, 77)

According to this view, this “complicated network of similarities overlapping and criss-crossing” (Wittgenstein, PI § 66) is the only thing connecting the - otherwise so distinct - members of the social category of ‘woman’. The various uses of the term ascribe it different, albeit related, senses which cannot objectively be reduced to any single characteristic. The case of bodily features is paradigmatic in a feminist context because, although they are central to the construction of the concept of ‘woman’, reality shows us not all women share every single primary and secondary sexual characteristic typically associated with female organisms. Some of us do not menstruate, have not developed breasts, others are born without a uterus. And what to make of genetic males born with female genitalia and raised like girls? If we look and see, as Wittgenstein recommends, we realise that ordinary language users do interchangeably apply the term ‘woman’ to refer to sex and gender and are not always referring to a defining characteristic that all women share. When we think of different women, instead of focusing on some picture of ‘woman’, we see that some of these features disappear while giving room for other similarities to crop up (PI § 66). This analysis leads to an understanding of the meaning of ‘woman’ that does no longer imply presenting necessary or sufficient conditions for membership in the category. But just as there is no finite set of defining features that all women share, Heyes is left with the bizarre consequence that two women may actually share nothing in common apart from the fact that they are both called ‘women’ (^{Heyes 2000}, 84). This paradox brings us to another Wittgensteinian approach that I believe complements Heyes's otherwise more comprehensive take on Wittgensteinian feminism.

In her examination of the concept of ‘woman’ as a category of feminist analysis, Marilyn ^{Frye (2011)} too concludes that ‘woman’ is more appropriately considered through the metaphor of “family resemblances” rather than the more traditional metaphor of “sets”. Asserting that ‘woman’ is a necessary analytic category of feminism, she argues that the observation that the concept bears no necessary or sufficient conditions for membership should not be taken as proof that the social category of ‘woman’ is somehow unreal, but simply that it is not a set (^{Frye 2011}, 86). Frye, however, reminds us that what we do when we approach meaning descriptively is to examine the principle of coherence of a given category. While this principle might not be as fixed as one initially made it out to be, it should nevertheless exist for it is a function of intelligibility of language use. In this sense, “it is not useful to be told that there is ‘nothing common’” to all members of the category of ‘woman’ (^{Frye 2011}, 91). Instead, we should expand on the Wittgensteinian idea of a “complicated network of similarities” connecting the different senses of a family resemblance concept to look for the particular kind of density that is common to all. In Frye's words (2011, 91), “There has to be something more here than overlapping attributes, even ‘complicated networks of them’ […]. For everything is similar to and different from everything in indefinitely many ways. […] ‘Similarity’ only has traction if something is at work to select some similarities as salient.”

What Frye seems to identify as the relevant commonality between members of a category is the very relation they maintain with typical and characteristic attributes observed in prototypical members (^{Frye 2011}, 91). Indeed, when applied to social categories, the metaphor of family resemblances resonates both with the idea of a “prototype structure” of categories and with the particular densities that are characteristic of living kinds, that is, the “webs of correlations which empirically support inferences from the presence of one feature to the presence of another” (^{Frye 2011}, 93). In this light, the principle of coherence guiding our application of the social category of ‘woman’ is given by a particular sort of clustering built upon observable patterns of correlations.

The idea of ‘woman’ as a cluster concept whose coherence is dependent on webs of correlations is also present in Natalie Stoljar's (1995) understanding of ‘woman’ as a type that applies to a "resemblance class". In the same vein as Heyes and Frye, Stoljar concedes that there is no disjunctive property or set of features that delimits membership in the category of ‘woman’. Instead, she suggests that membership is defined by participation in the "relevant resemblance structure" (^{Stoljar 1995}, 264). Expanding on Wittgenstein's idea of family resemblance concepts, Stoljar presents a definition of cluster concepts as types whose application to individuals is dependent on resemblance relations to certain general features or paradigms. She goes on to identify four general features that cluster together to form the ordinary concept of ‘woman’: primary and secondary female sex features; phenomenological features such as female sexual experience or menstrual cramps; gender roles such as wearing typical female clothes or being oppressed on the basis of sex; self-attributions and attributions of others which result from the previously identified features (^{Stoljar 1995}, 283-284). On Stoljar's account, then, an individual is part of the resemblance class in as much as they resemble any one of these paradigms sufficiently closely (^{Stoljar 1995}, 284). Though perhaps not faultless, Stoljar’s explanatory model constitutes a noteworthy effort to examine the limits of the application of the ordinary concept of ‘woman’. Indeed, its inclusion of both biological and social features in the extension of ‘woman’ accounts for the roots of contemporary disputes over the concept by revealing its meaning to involve two primary, distinct, while often interchangeable, uses: the sex concept and the gender concept.⁷

All three feminist Wittgensteinian approaches I have surveyed insist on the “contingent and revisable”, albeit “observationally supported” (^{Frye 2011}, 93), character of their analyses. Notwithstanding, they also share a concern toward the post-modern legacy of category-scepticism and about misled understandings of the workings of language. With Wittgenstein, they establish that inquiring about the meaning of a concept entails an exercise of clarification of its ordinary uses. The relevance of these approaches lies, then, on having found a middle ground between purportedly descriptive, essentialist accounts of ‘woman’ and anti-essentialist, “anything goes” conceptions of meaning. By doing so, they begin to outline an alternative to both identity politics founded on linguistic essentialism and the post-modern legacy of political paralysis. Such correspondence between conceptual analysis and practice is particularly apparent in Heyes’s approach, which recommends we take seriously Wittgenstein's claim that it is possible to delimit a concept's boundaries for a “special purpose” (PI § 69). While admitting that our ontological concerns might sometimes offer good reasons for leaving a concept open, she argues that conceptual delimiting is a strategy we should engage in our political practice (^{Heyes 2000}, 96). But while Heyes illustrates the need for conceptual delimiting with specific examples of contextual decisions about inclusion and exclusion in feminist practice, she evades the question of whether feminists can legitimately advocate in favour of a substantial revision of the concept of ‘woman’. Is a Wittgensteinian account of meaning as use reconcilable with a revisionary approach of this kind?

The Ameliorative Question: Rethinking Normativity

Feminist Wittgensteinian philosophers have expanded on the idea of family resemblances to develop descriptive approaches to the meaning of ‘woman’ which reveal the multiplicity of uses and senses it acquires in different ordinary contexts. This should not be taken to mean, however, that all senses falling under the concept of ‘woman’ are acceptable. Indeed, feminists have long denounced the way some uses of ‘woman’ are profoundly constrained by the existence of power relations that act normatively to keep them in place. In this section, I will not attempt to articulate a new ameliorative concept of ‘woman’. Instead, I will argue against a traditional view of Wittgenstein's later philosophy that sees it as incompatible with this sort of project by indicating that a descriptive account of ‘woman’ as a family resemblance concept is reconcilable with an ameliorative approach. I will also suggest that Wittgenstein's conception of normativity might help us navigate this project without falling prey to the threat of category-scepticism.

Wittgenstein introduces the notion of “language-game” to consider a “whole, consisting of language and the activities into which it is woven" (PI § 7), which refers to the very use of language. Furthermore, he places the application of a natural language in the context of a “form of life” (PI § 23), i.e., a particular linguistic community where we learn to use language. As learning the meaning of a word always occurs through participation in language-games, understanding a word is not tantamount to possessing a sort of unshakable knowledge about its meaning, but being able to grasp the rules that establish its use, just as we learn the rules in a game. The language-game is, thus, an activity that integrates individuals who share specific forms of life and defines the meaning of a concept over its participants' mutually intelligible use. To understand a word is, therefore, to be able to use it in a certain language-game effectively.

What the idea of meaning as use tells us is that the criteria for the correct application of language are actually laid out by rules arising out of usage itself, which suggests that meaning is normative in at least one relevant sense. Indeed, it is from the notion of “rule-following” that Wittgenstein's famous argument against the possibility of a private language derives (PI § 243-315). Surely, if intelligibility by other participants in the language-game sets the criteria for the correct application of a term, then this implies 1) that the criteria of correctness cannot be found in some a priori truth, and 2) that speakers do not have private domain over signification, i.e., we use a given concept in situations in which our application of it will generally agree with our interlocutors' (^{McManus 2003}, 67). These two consequences have earned ordinary language philosophy significant criticism over the years, as Wittgenstein's account of meaning as use may now appear to legitimise the status quo.⁸ In light of this, one might certainly feel discouraged to ally feminist purposes to an author that seems to urge us to bring back words to their ordinary uses and leave “everything as it is” (PI § 124). If our goal, as language users, is to reach a certain degree of intelligibility when applying concepts, are we not trapped with prevailing rules of usage?

Let us recall what we said was the primary purpose of Wittgenstein's later philosophy: to get rid of a philosophical picture that has held us captive and is to blame for most philosophical misunderstandings (PI § 115). David Owen translates this purpose into political philosophy as the goal to dissolve “a certain class of nonphysical constraints on our capacity for self-government”, that is, on our capacity to judge and act according to those judgements (^{Owen 2003}, 82). Owen refers to such type of constraints as “aspectival captivity” and describes it as the phenomenon in which a picture functions either as the implicit background of our thought and action or as an explicitly acknowledged limit taken to be utterly necessary (^{Owen 2003}, 82-83). While some versions of post-modern thought would have us believe captivity comes about due to the normative character of such a picture, the fact that all meaning is itself normative in at least one relevant sense should make us suspicious of this claim.

To be sure, pictures do set forth sets of rules that regulate our language and action, and some of these are certainly profoundly oppressive. But we should be clear at this point: according to Wittgenstein, not all world pictures are bad pictures, and their existence is, moreover, a necessary condition for thinking and acting in the world. This is where the Wittgensteinian critique goes a step further than post-modern analyses: trust in a system of judgements is never entirely epistemically grounded, but rather translates our progressive adherence - perhaps through learning - to a picture of the world, and that is not a bad thing per se. Therefore, Owen suggests that “the value of a given picture is given by its capacity to orient our practical judgements such that we can go on in the world, that is, experience ourselves as agents” (^{Owen 2003}, 83). Accordingly, we can identify an illegitimate picture to be one whose coercion power is so strong that it imposes even when it is prejudicial to our lives - in the form of, say, a philosophical, social, or cultural model. This means that the world pictures we inherit, along with the rules we learn to follow, can easily stop making sense to us and compromise our agency (^{Owen 2003}, 84).

Over the years, feminists have staunchly denounced the many ways in which pictures of ‘woman’ have compromised women's agency by associating the concept with certain imposed social roles. In a Wittgensteinian account of normativity, however, such pictures are illegitimate not because they are regulated by rules, but inasmuch as those rules are naturalised in language in a way that allows for no questioning. Accordingly, whether a conception of ‘woman’ is oppressive - and therefore illegitimate - turns not on its normative nature, but on whether it integrates systematically imposed practices that do not reflect the necessities or interests of its participants. A misguided equivalence between the normative character of our language and its oppressive force has, however, led some authors to engage in a form of anti-essentialism that has been rightfully accused of a negligent and irresponsible attitude towards politics. In view of this, feminists who believe in the political relevance of preserving the category of ‘woman’ might look at the Wittgensteinian practice of conceptual delimitation as an alternative to both linguistic essentialism and scepticism. Instead of dismissing categories as normative and thus “exclusionary”, this approach entails an examination and revision of prevailing rules of usage. This solution has the advantage of valuing the importance of meaning and language in our political lives and analyses and may motivate the impending ameliorative question: what do we want ‘woman’ to mean?

Contrary to common interpretations, Wittgenstein does suggest we engage in such a project by acknowledging that language is flexible enough to allow for the rejection of particular sets of rules and enter new language-games (PI § 23). While it is certainly true that his language critique does not seek to replace a specific set of rules with another, it does urge us to awake for a permanent state of alert and to revaluate the pictures that guide our conduct. Even if that is not the role he attributes to philosophy, Wittgenstein believes “a [language] reform for particular practical purposes, an improvement in our terminology designed to prevent misunderstandings in practice, may well be possible” (PI § 132). Inciting us to go forward and make “new discoveries and inventions” (PI § 126), it is precisely in its purely critical character that lies Wittgenstein's contribution to political philosophy. The implication here is that the capacity to experience ourselves as agents is one which must be secured before any sort of political action takes place - an insight which additionally meets the interpretation of Wittgenstein's later thought as a sort of therapy destined, not to solve philosophical problems, but to dissolve them (PI § 133-133d). Accordingly, it is not for philosophy to ground any sort of conduct, but to dissolve linguistic confusions that precede problems of a practical nature by surveying the state of affairs from which they emerge.

I want to argue that a descriptive approach of the kind Wittgensteinian feminists have engaged in is a desideratum for the elaboration of any ameliorative definition of ‘woman’. Failing to distinguish between descriptive concepts (what use reveals the concept to mean) and ameliorative concepts (what we want a concept to mean, according to some special purpose) contributes to hinder any constructive debate over the meaning of ‘woman’. Hence, the importance of the Wittgensteinian feminist projects put forward by Heyes, Frye and Stoljar is that, by clarifying the rules that govern the concept's ordinary application, they may act as a steppingstone to any ameliorative project that follows. Ordinary language philosophy has demonstrated how two distinct sets of rules have come to govern the application of the concept of ‘woman’: a biological set of rules that defines ‘woman’ in virtue of certain biological features and a social set of rules that picks out particular social roles, identities, or status. By revealing the ordinary meanings of ‘woman’, it presents the very objects of ameliorative revision.

Far from committing us to ordinary conceptions of ‘woman’, understanding how such a broad concept effectively circulates in ordinary speech is rather an opportunity to argue in favour of some of its senses and condemn others. This is the goal of feminist ameliorative projects: to revise the rules that guide our usage of concepts and delimit them for some special purpose.⁹ While meaning is descriptively determined by its ordinary uses, in an ameliorative project it is in principle legitimate to question such uses and to offer reasons why these should be revised. An answer to the ameliorative question is thus dependent on what one defines as the relevant purposes guiding the development of a revised concept of ‘woman’. Indeed, as our practical commitments might differ in many legitimate and relevant ways, there is no shortcutting the need to thoroughly define what motivates our revision. Ultimately, the very recognition that ‘woman’ bears an oppressive normative charge is precisely the reason why feminism must assume the responsibility to articulate the concept so that it expands what Seyla Benhabib described as the “feminist commitment to women's agency and sense of selfhood” (^{Benhabib 1995}, 29).

I believe current disputes over the meaning of ‘woman’ should be located in this context. Feminists who engage in ameliorative projects, within or outside academia, can plausibly agree that a relatively well-defined concept of ‘woman’ is fundamental to feminist politics if it is to understand women's problems as collective ones and avoid succumbing to the liberal atomisation of the individual. For this to be realised, however, the ameliorative claim must not pretend to pass as a descriptive one, and the argument for which it is the conclusion must be clearly laid out. When presented as an ameliorated concept, the claim can be seen “as what it is, as an object of comparison - as a sort of yardstick; not as a preconception to which reality must correspond” (Wittgenstein, PI § 131). By presenting both ordinary and ameliorated concepts as what they are, the method offered by ordinary language philosophy allows for a clarification of what is at stake in feminist disputes over the meaning of ‘woman’. Meanwhile, whether ‘woman’ will ultimately amount to a biological or a social category - or whether a compromise is made between these two - is a matter of how ordinary speakers are going to employ it. Feminists might well have an important part to play in persuading ordinary speakers to revise their ordinary use of ‘woman’. But if they are committed to an ameliorative project, they must not bypass the need to present reasons and engage in practices of deliberation guided by a vision of their political commitments. In this context, Wittgenstein's recommendation that we attend to the various uses of a concept, the purposes it serves, its historical load, etc., may only provide us with the necessary tools to strengthen our own claims to meaning.

While philosophy might not supply the criteria to settle political disagreements, a Wittgensteinian understanding of the workings of language shows us the way forward, presenting the (often hidden) possibilities for breaking the rules that guide our conduct and holding us accountable for the creation of new ones. Thus, if Wittgenstein allies with feminist theory in a critical exercise of clarification, he also assigns it a reconstruction task which is intimately articulated with practice. What follows is a struggle for the meaning of ‘woman’ - one feminism has long been fighting, even when it got entangled in abstract discussions over it. For is that not largely what politics is about? In this context, the most important lesson feminism can draw from Wittgenstein's ordinary language philosophy is this: philosophy may help clarify the state of affairs before the problem is solved (PI § 125), but it is on us to take responsibility for what we want ‘woman’ to mean and, I would add, what we want women to be.