Is Haslanger’s ameliorative project a successful conceptual engineering project?

Supporters of conceptual engineering often use Haslanger’s ameliorative project as a key example of their methodology. However, at face value, Haslanger’s project is no cause for optimism about conceptual engineering. If we interpret Haslanger as seeking to revise how people in general use and understand words such as ‘woman’, ‘man’, etc., then her project has been unsuccessful. And if we interpret her as seeking to reveal the meaning of those words, then her project does not involve conceptual engineering. I develop and defend an alternative interpretation of Haslanger’s project and argue that, so interpreted, it is a successful conceptual engineering project after all. In so doing, I develop what I call a particularist account of the success conditions for conceptual engineering.

There are two standard interpretations of Haslanger's project within the conceptual engineering literature. On the first, Haslanger is seeking to revise our gender and race concepts-that is, to change how people in general use and understand terms such as 'man', 'woman' and 'race'. However, so interpreted, Haslanger's project appears unsuccessful: she has not succeeded in changing how people use and understand 'man', 'woman' or 'race'. On the second interpretation, she is seeking to reveal the semantic meanings of those terms. However, Haslanger's project then appears not to involve conceptual engineering at all: it appears rather to be a project of descriptive semantics. This is something of an embarrassment for supporters of conceptual engineering. It would be a bad sign for conceptual engineers if Haslanger's project-a paradigm case-was not a successful conceptual engineering project. It might not follow that conceptual engineering should be abandoned as a philosophical methodology, as it might be more effective in other cases. But, if the paradigm of conceptual engineering is a failure, or else not an instance of conceptual engineering at all, it raises serious doubts about the purported promise of conceptual engineering.
In this paper, I articulate and defend an alternative interpretation of Haslanger's project. On this alternative-I call it the terminological interpretation-Haslanger's project is a successful conceptual engineering project. In making the case, I develop what I call a particularist account of the success conditions for conceptual engineering. In contrast to recent univocal accounts, such as those defended by Simion & Kelp (2020) and Nado (2021b), I argue that each conceptual engineering process and project has its own success conditions. I proceed as follows. In § 2, I introduce the two standard interpretations of Haslanger's ameliorative project. In § 3, I develop the case that, on those interpretations, Haslanger's project is not a successful conceptual engineering project. In § 4, I develop and defend the terminological interpretation of Haslanger's project. In § 5, I argue that, so interpreted, Haslanger's project is a successful conceptual engineering project. In § 6, I draw out and defend the particularist account of success conditions that underlies the preceding discussion. I close, in § 7, with a brief discussion of the so-called 'inclusion problem' facing Haslanger's project.

Haslanger's ameliorative project
Conceptual engineering is the general process of improving our linguistic or conceptual repertoires. This may involve introducing new terms or concepts, revising or replacing existing terms or concepts, or eliminating defunct terms or concepts. One may engage in conceptual engineering for a variety of reasons. At a local level, one may introduce technical terminology (e.g. 'tonk') to play a particular role within an argument in the subsequent sentences. At a more global level, one may introduce a new term into English (e.g. 'Brexit') in pursuit of broad, political goals (such as exiting international treaties).
Haslanger's ameliorative project (2012) is taken to be a paradigm case of conceptual engineering. The project involves developing surprising definitions for 'man', 'woman', 'race', etc., in pursuit of social justice. For example, Haslanger defines 'woman' and 'man' in terms of hierarchies of privilege and subordination: S is a woman iff df S is systematically subordinated along some dimension (economic, political, legal, social, etc.), and S is "marked" as a target for this treatment by observed or imagined bodily features presumed to be evidence of a female's biological role in reproduction. S is a man iff df S is systematically privileged along some dimension (economic, political, legal, social, etc.), and S is "marked" as a target for this treatment by observed or imagined bodily features presumed to be evidence of a male's biological role in reproduction. (2012/2000: 230) What exactly are Haslanger's definitions supposed to tell us? Are they supposed to describe the semantic meaning of 'woman' and 'man', or to capture how we do or should use those terms, or something else? Our answer to this question will depend on how exactly we interpret Haslanger's project.

The revisionary interpretation
The revisionary interpretation is largely inspired by Haslanger's seminal Gender and Race: (What) are they? (What) do we want them to be? (Haslanger /2000, henceforth G&R, in which Haslanger first develops the above definitions. 1 On this interpretation, ameliorative projects begin with a reflection on what we want our concepts to do: we begin by considering more fully the pragmatics of our talk employing the terms in question. What is the point of having these concepts? What cognitive or practical task do they (or should they) enable us to accomplish? (pp.

223-224)
Then, when we have identified what we want the concepts for, [t]he responsibility is ours to define them for our purposes. In doing so we will want to be responsive to some aspects of ordinary usage […]. However, neither ordinary usage nor empirical investigation is overriding, for there is a stipulative element to the project: this is the phenomenon we need to be thinking about. (p. 224) Haslanger's principal concern is taken to be social justice: At the most general level, the task is to develop accounts of gender and race that will be effective tools in the fight against injustice. (p. 226) How might redefining 'woman' and 'man' help fight injustice? Richard understands the idea as follows: The, or an important, purpose of the concept woman is to subordinate people on the basis of their (perceived) female properties. We shouldn't be subordinating people on this basis; indeed, we should be fighting against such subordination. One way to do this is to reformulate the concept so that, so to speak, its noxious purpose is part of its definition. This will put the purposes for which the concept is actually being used front and center, allowing us to fight gender subordination. (Richard 2020: 359) Whatever the details, it is typically held that Haslanger needs to change how a large number of people use and understand relevant terms and concepts. Richard is unusual in restricting this to theorists: Her project […] would be successfully carried off only if a large number of those who theorize about gender and race were to come to use 'woman', 'Latino', and so on with the conscious intention that they should be understood as using the words with the relevant meanings, and indeed are so understood. (p. 360) Others understand Haslanger's project as requiring a change to how people in general use and understand the terms and concepts. Ritchie writes: As Haslanger stated, the semantic component involves "asking us to use an old term in a new way" [Haslanger 2020]. If 'us' only refers to theorists, it is dubitable that much social progress will be made. (Ritchie 2021: 480) Jenkins writes: 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. (Jenkins 2016: 395, my emphasis)

And Gibbons writes:
consider Sally Haslanger's accounts of race and gender, crafted with the intent to be "effective tools in the fight against injustice" [Haslanger /2000. These sort of ameliorative projects […] will not be successful without the coordination of potentially exceptionally large numbers of agents. (Gibbons 2022: 6) On the standard revisionary interpretation, then, Haslanger is seeking to revise how people use and understand 'woman', 'man', etc., in pursuit of social justice. Haslanger's definitions are the end goals: they capture the intended post-amelioration meaning of those terms.

The revelatory interpretation
The second interpretation of Haslanger's project, the revelatory interpretation, is largely inspired by Haslanger's What Good Are Our intuitions? Philosophical Analysis andSocial Kinds (Haslanger 2012/2006), henceforth Intuitions. In that article, Haslanger distinguishes three concepts that, for each speaker, are associated with any given term: the manifest concept is 'the concept I thought I was guided by and saw myself as attempting to apply'; the operative concept is 'the concept that best captures the distinction that I in practice draw'; and the target concept is 'the concept that, all things considered (my purposes, the facts, etc.), I should be employing' (p. 388).
For example, consider Ela's use of the word 'spicy': she takes herself to apply the word to any dish cooked with lots of spices; she in fact applies the word to any dish cooked with lots of spices or herbs; and perhaps (given her communicative aims and how others around her use the term) she ought all-things-considered to apply the word to any dish cooked with lots of chilli. Then, the manifest concept is cooked with lots of spices, the operative concept is cooked with lots of spices or herbs, and the target concept is cooked with lots of chilli. Now, in Intuitions, Haslanger appears to argue that her proposed definitions are operative concepts-that the above definitions of 'man' and 'woman' track the distinctions we in practice draw with those terms. In justifying this view, Haslanger sketches a form of metasemantic externalism: Objective Type Externalism: Terms/concepts pick out an objective type, whether or not we can state conditions for membership in the type, by virtue of the fact that their meaning is determined by ostension of paradigms (or other means of reference-fixing) together with an implicit extension to things of the same type as the paradigms. (p. 398) Appealing to Objective Type Externalism, Haslanger claims that the semantic meanings of 'woman' and 'man' are fixed by the social type to whose instances we typically apply those terms. So, if (as Haslanger suggests) we typically apply 'woman' to individuals who satisfy Haslanger's proposed definiens for 'woman', and mutatis mutandis for 'man', then her proposed definitions effectively pick out the current extensions of those terms. According to Haslanger: 'This is not to propose a new meaning, but to reveal an existing one' (ibid.  (Cappelen 2018: 80) He adds that Haslanger is endorsing 'the view of conceptual engineering as revelation' (p. 81). This interpretation of Haslanger's project is also advocated by Deutsch. He writes that 'Haslanger's analysis of womanhood is just that. It is an analysis of womanhood' (2020b: 11).
Haslanger [2012/2006] suggests that her analysis of 'woman', for example, might well be the correct descriptive analysis of our actual concept 'woman' […].
[H]er proposal concerns the semantic meaning and reference of the term 'woman', one to the effect that the semantic extension of this term includes all and only people subordinated on the basis of perceived biological features indicating a female role in reproduction. (Deutsch 2020a: 3949) On the revelatory interpretation, then, Haslanger's ameliorative project is to reveal (and analyse) the semantic meanings of terms such as 'woman', 'man', etc. On this interpretation, Haslanger's definitions are the analyses: they are intended to capture the meanings of, or operative concepts associated with, 'woman', 'man', etc.

Not a successful conceptual engineering project
On neither interpretation is Haslanger's project a successful conceptual engineering project.

The revisionary interpretation
On this interpretation, Haslanger is seeking to change how people use and understand 'woman', 'man', etc. We are to assume that people's use and understanding of those words did not, at the time of G&R, encode hierarchies of privilege and subordination; and we judge the success of Haslanger's project by whether people's use and understanding of those words come to encode hierarchies of privilege and subordination. I will understand the idea in terms of semantic meaning, so that Haslanger's project (on the revisionary interpretation) involves revising the semantic meanings of 'woman', 'man', etc. 2 So construed, on the revisionary interpretation, Haslanger's project assumes that the semantic meanings of 'woman' and 'man' did not encode hierarchies of privilege and subordination at the time of G&R, and that the success of Haslanger's project depends on her changing the semantic meaning of those terms so that they do encode such hierarchies.
On this interpretation, Haslanger's project has not been successful: she has not successfully changed the meanings of 'woman', 'man', etc., in the intended way. This is not to deny that the meanings of 'woman' and 'man' may have changed since G&R, nor to deny that Haslanger has in some way influenced the meaning of those terms. The meanings of words naturally change over time ('semantic drift') and, assuming that patterns of use are a causal factor in this process, every speaker contributes at least minutely to this process. Moreover, given Haslanger's status, it is likely that she has had a larger effect on any such changes than most individuals. Nonetheless, the particular changes associated with her project have not come about.
Consider 'woman'. Let us suppose (for reductio) that the meaning of 'woman' has recently changed in such a way that Haslanger's definition for 'woman' now captures the meaning of 'woman'. As semantic facts do not float free, there must be an explanation for this change. 3 To count as adequate, a proposed explanation should draw on the kinds of factors that can cause semantic changes-such as folk and expert dispositions and the underlying nature of referents. But, for the case at hand, none of these causal factors can do the relevant explanatory work. Neither folk nor experts (whoever the relevant experts are) have acquired the disposition to use 'woman' as though, for example, (1) is false by definition: (1) One day woman won't be oppressed on the basis of their perceived sex.
After all, for the most part, folk and experts alike can still conceive of scenarios in which women are not oppressed on the basis of their perceived sex. Moreover, for a change in the underlying nature of women to do the relevant explanatory work, it would need to be the case that women were not by nature oppressed at the time of G&R, but are by nature oppressed now-but I see no plausible arguments for this being the case. So, absent any plausible explanation of the semantic change, we should reject the initial supposition: the meaning of 'woman' has not changed in the supposed way. Mutatis mutandis for 'man', etc.
Secondly, on the revisionary interpretation, the so-called implementation problem gives us reason to doubt that Haslanger's project will be successful. 4 To illustrate, consider the question: What precisely must Haslanger do in practice to successfully bring it about that (say) her definiens for 'woman' becomes the meaning of 'woman'?
The implementation problem is that there appears to be no plausible answer to this question.
One immediate difficulty, as Richard points out, concerns how Haslanger's project is framed: Insofar as this particular version of the project involves extension shifting, it strikes me that it was never likely to be successful. Haslanger tells us that she wants to answer such questions as What is it to be a man? […]. An extension shifting answer strikes me as one very difficult to make stick-as very difficult to get people to accept-if it is not grounded in something about prior usage that can be adduced to make plausible that the answer "simply reveals what we were talking about all along", or that the answer is an apt response to an ambiguity, confusion, or inconsistency in prior use. (Richard 2020: 375) Nonetheless, Sterken (2020) has proposed a mechanism for changing the meaning of words: transformative communicative disruption. According to Sterken, the ameliorator is to adopt her new definition to disrupt ordinary communication, causing audiences to reflect upon and change their own usage.
Return to the example involving Amanda's utterance of 'We need to get rid of women'. Consider someone not initiated in or not knowing about Amanda's project […]. On the supposed conventional reading, this is horrific. That reaction of horror, which is a result of the miscommunication […], is constructive. It makes the audience stop and think, and that […] will lead her, one might hope, to start reflecting on the meaning of her words. (Sterken 2020: 430) But, while this mechanism will undoubtedly have some effect, it is not a plausible story for how Haslanger could bring the meaning of 'woman' into line with her definition. Haslanger's definitions are long, carefully constructed and somewhat surprising: even the most reflective audience would not reconstruct Haslanger's definitions from an utterance such as 'we need to get rid of women'. Sterken would perhaps agree: Disruptiveness is not sufficient for meaning transformation. In meaning transformation, coming to understand the proposed amelioration is transformative. The deviant communicative event, together with a number of other events, triggers a full-on meaning change that is transformative. (ibid.) So the proposed strategy, I take it, is this: Haslanger and followers disrupt communication, causing their audiences to reflect on the meaning of 'woman', during which reflection Haslanger and followers explain the definitions at hand, which (together with other events) triggers the intended meaning change for the audiences. To achieve genuinely large-scale meaning change, presumably a chain reaction would be required: audiences would have to join the effort to undertake further transformative communicative disruptions.
However, it is extremely unlikely that this process would be successful. Too few audiences would reflect in the envisaged way; of those who did, too few would come to genuinely understand Haslanger's definitions; of those who did, too few would accept or take seriously the definitions; and, of those who did, too few would take on the task of undertaking further communicative disruptions. The practical difficulties in overcoming these challenges would be immense. Put simply, none of us has sufficient influence (or control) over how other speakers behave to implement these kinds of changes. As it stands, this is not a plausible and practical story for bringing the meaning of 'woman', 'man', etc., into line with Haslanger's definitions. 5 Finally, it is worth noting that, even if Haslanger convinced people to change how they use and understand the terms in question, it remains an open question whether this would effect the desired semantic changes. Deutsch argues: on any view according to which more is required for the determination of semantic meaning and reference than speakers' intentions to use the relevant term in a stipulated sense, there will remain the question: What more must take place or be done in order for a semantic shift to actually be implemented? Not only do conceptual engineers rarely offer any answer to this question, a case can made that there is no plausible answer to it: we are, all of us, simply ignorant of the precise mechanisms of semantic change. (Deutsch 2020a: 3953-3954) To be clear, none of the above is an objection to Haslanger's project per se: I do not believe that the revisionary interpretation captures Haslanger's intentions (past or present). The point is simply that, on that interpretation, Haslanger's project is, and will inevitably remain, an unsuccessful conceptual engineering project.

The revelatory interpretation
On this interpretation, Haslanger is engaged in 'conceptual engineering as revelation': she is seeking to reveal the semantic meaning of, or to reveal the operative concept associated with, 'woman', 'man', etc. However, so interpreted, Haslanger's project appears not to be a project of conceptual engineering at all.
The concern is well-known. Cappelen clearly has it in mind when he describes Haslanger as engaged in 'conceptual engineering as revelation'. He writes: in the case I just imagined [the extension of 'marriage' excluding same-sex couples], the revelation of what our term meant revealed something we object to-it reveals that the word doesn't denote what it ought to denote. That, as I see it, is the point where conceptual engineering kicks in. The first stage, the revelation of what we refer to, is a descriptive project, and is not distinctively ameliorative. (Cappelen 2018: 81) Deutsch makes the point more explicitly: we should […] understand Haslanger's method as not any variety of conceptual engineering at all. Rather, Haslanger's analysis of womanhood is just that. It is an analysis of womanhood; it specifies the purported conditions on being a woman. (Deutsch 2020bb: 11) One reason Deutsch gives for this view is that: In [Haslanger /2006], Haslanger herself suggests that she should be understood as offering a descriptive analysis of womanhood, as opposed to trying to re-engineer the meaning of 'woman'. (2020b: 11n14) The argument is clear. If Haslanger is trying to articulate the operative concepts associated with 'woman' and 'man', she is engaged in a project of descriptive semantics: she is trying to analyse the extension of 'woman' and 'man'. But then, she is not engaged in conceptual engineering: she is not seeking to improve our linguistic or conceptual repertoires.
The argument is decisive with respect to the revelatory interpretation, but there are two alternative interpretations in the vicinity. Firstly, note that, in Intuitions, Haslanger suggests that the manifest and operative concepts associated with (for example) 'woman' come apart. Haslanger describes this kind of situation as an 'awkward position', deserving of resolution. Haslanger suggests three strategies for resolution (2012/2006: 388): (i) change our understanding of how we are using the term, thereby bringing the manifest concept into line with the operative concept; (ii) change our patterns of use, thereby bringing the operative concept into line with the manifest concept; or. (iii) change both our understanding and patterns of use, bringing them both into line with what we ought (all things considered) to have been doing all along, and thereby bringing the manifest and operative concept into line with the target concept.
Now, each of (i)-(iii) plausibly counts as conceptual engineering-of the manifest concept, of the operative concept, and of both, respectively. Accordingly, one might construe Intuitions as the first stage of a two-stage conceptual engineering project: Haslanger is revealing the precise nature of the conceptual engineering problem (i.e. by revealing how the manifest and operative concepts come apart) so that she can select the right conceptual engineering strategy for resolution (i.e. (i), (ii) or (iii)). So construed, revelation is merely the first step in Haslanger's conceptual engineering project. Call this the extended revelatory interpretation of Haslanger's project. Now, on the extended revelatory interpretation, Haslanger's project is a conceptual engineering project-but Haslanger has thus far only undertaken the pre-engineering, descriptive stage. And, problematically, the conceptual engineering stage will face difficulties comparable to those raised in connection to the revisionary interpretation: Haslanger does not have the kind of influence (or control) to bring about a scenario in which people in general associate her precise definitions with 'woman' and 'man', nor in which patterns of use are in line with those definitions. While, on the extended revelatory interpretation, Haslanger's project is a conceptual engineering project, there are no obvious grounds for optimism about its prospects for success.
The second interpretation in the vicinity is Ball's (2020) retrospective revelatory interpretation. According to Ball, when a revisionary analysis is offered for a term: the analysis is successful if everyone (within the relevant discourse) comes to accept the analysis; and, in such a case, the revisionary analysis retrospectively determines what the term meant (within the discourse) all along-even before the revisionary analysis was proposed. On this view, Haslanger is indeed seeking to reveal the meaning of 'woman' and 'man', but she is aiming to do so by retrospectively determining the meaning of those terms.
On the retrospective revelatory interpretation, Haslanger's project is again a project of conceptual engineering. Ball demurs from this conclusion, as he ties conceptual engineering closely to 'changing the subject' (pp. 36ff), and on his view Haslanger is trying to fix (rather than change) the subject. However, pace Ball, there is no essential connection between conceptual engineering and changing the subject: it is coherent to understand conceptual engineering as sometimes not changing the subject, as evidenced by the plethora of explanations of why this is so. 6 On the retrospective revelatory interpretation, Haslanger engineers new and surprising definitions, and tries to convince everyone within the discourse to use them-and that is one good sense in which she is engaged in conceptual engineering. If Ball's metasemantic views are right, then it simply turns out that successful conceptual engineering sometimes reveals the meaning of a term-not through descriptive semantics, but by retrospective determination. 7 However, on the retrospective revelatory interpretation, Haslanger's conceptual engineering project has again not been successful: Haslanger has not convinced everyone to accept her revisionary analyses. And, there are no obvious grounds for optimism about its future success: the tides are not turning in favour of her analyses.
Again, none of the above is an objection to Haslanger's project per se. The point is simply that, on any of the revelatory interpretations, Haslanger's project is not a successful conceptual engineering project.

The terminological interpretation
To my eyes, none of the above provides a compelling interpretation of Haslanger's project. In this section, I defend an alternative: the terminological interpretation.
In Going on, not in the same way (Haslanger 2020), Haslanger sets out the historical and political context of her earlier work. It is useful to begin with Haslanger's concepts of cultural technē and ideology.
[A] cultural technē is a set of social meanings-including concepts, scripts, background assumptions ("analytic truths"), inferential patterns, salient metaphors, metonyms, conceptual oppositions, and (broadly speaking) grammar-that provides tools for interpreting and responding to each other and the world around us, and does so in ways that facilitate (better or worse) forms of coordination.
[…] The cultural technē provides the frames of our (fragmented, dynamic) practical orientation […] that enables us to engage in social life. An ideology is a cultural technē gone wrong: it may fail to provide us the tools to appreciate relevant parts of the world, or what's valuable and how things are valuable; it may organize us in unjust ways. When we are in the grip of an ideology, however, our practical orientation positions us to enact […] practices and structures that sustain injustice. (Haslanger 2020: 232) For Haslanger, we are all in the grip of gender and race ideology: the cultural technē orients us towards matters of gender and race in such a way that we enact practices and structures that sustain gender and racial injustices.
If one is to pursue emancipation, an important first step is to understand the situation, 'to see how the local cultural technē is ideological' (2020: 233).
Questions of aptness or justice don't arise for what is taken for granted. However, once articulated, ideology can (in principle) be debated and reformed. So one goal of ideology critique is to elucidate the conceptual and narrative frameworks that undergird our social interaction, thus making them available for critical examination. (2012: 19) Such elucidation requires theoretical work: one cannot simply observe the conceptual and narrative frameworks that undergird our social interaction, nor the ways in which the local cultural technē is ideological. Thus, for Haslanger, the first step of emancipation relies on critical theory-theory that exposes and critiques ideology. It is unsurprising, then, that she writes in the introduction to Resisting Reality that: [i]t would not be unreasonable to claim that the accounts of race and gender that I offer are theoretical and should be evaluated by theoretical standards. In pursuing emancipation, one starts with theory.
In G&R, Haslanger's aim was to develop theories of gender and race that were emancipatory in this sense; she sought to pursue emancipation by exposing, through theory, one respect in which our cultural technē is ideological. This explains both her comment that 'the task is to develop accounts of gender and race that will be effective tools in the fight against injustice' (2012/2000: 226, my emphasis), and the less-quoted critical-theoretical goals that immediately follow: (i) […] to identify and explain persistent inequalities between females and males, and between people of different "colors"; […] to identify how social forces, often under the guise of biological forces, work to perpetuate such inequalities. (ii) […] to be sensitive to both the similarities and differences among males and females, and the similarities and differences among individuals in groups demarcated by "color"; […] to identify the effects of interlocking oppressions, for example, the intersectionality of race, class, and gender [….]. (iii) […] to track how gender and race are implicated in a broad range of social phenomena extending beyond those that obviously concern sexual or racial difference […].
(iv) […] to take seriously the agency of women and people of color of both genders, and [to] develop an understanding of agency that will aid feminist and anti-racist efforts to empower critical social agents.

(pp. 226-227)
From this perspective, Haslanger's definitions in G&R serve to introduce theoretical terminology-theoretical terminology that makes certain gender and racial injustices within the local cultural technē explicit. 9 Importantly, though, Haslanger is not seeking to change how people (folk or theorists) in general use and understand 'woman', 'man', etc. The theory she develops in G&R is better thought of as a tool (2012/2000: 226). The function of the tool is to expose people to a different way of thinking about gender and race categories, to cause (or jolt) them to see those categories in a new light-a light that reveals the injustices within the cultural technē. So, in the following passage, Haslanger is asking "us"-herself and her readers-to use her terms as part of this ideology-exposing effort: In one sense this […] is just "semantics": I'm asking us to use an old term in a new way. But it is also politics: I'm asking us to understand ourselves and those around us as deeply molded by injustice and to draw the appropriate prescriptive inference.
[…] The point is not to legislate what terms to use in all contexts, but to offer resources that should be used judiciously. (2012/2000: 242) As the final sentence indicates, there is no suggestion here that anyone ought to be using Haslanger's definitions in general. There is also no suggestion that anyone needs to convince ordinary folk to use any terms in a new way, nor (except when exposing them to the theory) to convince them to understand any terms in a new way. In G&R, the aim is simply to engineer a set of newly-designed definitions that, when engaged with, can cause people to see gender and race categories in a new, ideologyexposing way.
Does this really enhance social justice? Only indirectly.
Disruption of this sort does not assume that people who come to question their identity can or should immediately act on it. They may not have the power, security, or resources to do so. Social emancipation must be a collective effort and change more than minds. But under conditions of ideological oppression, ideology critique matters. It invites and sometimes produces a shift in one's practical orientation. (2020: 236) The ultimate vision, then, is that enough people come to recognise the ideology to collectively challenge it and, perhaps, achieve social change. This goes far beyond the project of G&R, however. The project in G&R, then, involves giving theories of gender and race. So interpreted, Haslanger's revisionary definitions invite the objection that they 'change the subject'-that the resulting theories are not really about gender and race at all. However, as Haslanger principally conceives of the theories as tools for exposing ideology, we would expect the 'change of subject' objection to seem of secondary importance to her. Indeed, this is precisely what we find in G&R. Haslanger raises and addresses the 'change of subject' objection in the following passage: in an explicitly revisionary project, it is not at all clear when we are warranted in appropriating existing terminology. […] [I]t isn't entirely clear when a project crosses over from being explicative to revisionary, or when it is no longer even revisionary but simply changes the subject.
[…] The issue of terminological appropriation is especially important, and especially sensitive, when the terms in question designate categories of social identity such as 'race' and 'gender'.
Are there principles that determine when it is legitimate to appropriate the terms of ordinary discourse for theoretical purposes? An answer, it seems to me, should include both a semantic and political condition […]. The semantic condition is not surprising: the proposed shift in meaning of the term would seem semantically warranted if central functions of the term remain the same, for example, if it helps organize or explain a core set of phenomena that the ordinary terms are used to identify or describe. Framing a political condition in general terms is much more difficult, however, for 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. We will return to some of these issues later […]. (2012/2000: 225) Notice that Haslanger raises the 'change of subject' objection in the context of a more general (and 'especially important') issue-namely, that she might not be warranted in appropriating 'woman', 'man', etc. Haslanger suggests that semantic and political conditions should be satisfied to warrant terminological appropriation, but she is not particularly concerned about the semantic condition. Haslanger devotes just one sentence (found in the above quotation) to it, whereas she spends an entire section (pp. 240-242) to defending her terminological appropriation from a political perspective. On the terminological interpretation, all of this is to be expected.
As an aside, given that Haslanger does not develop her informal comment about 'central functions', she presumably takes her definitions to straightforwardly satisfy the 'semantic condition'. Perhaps she takes it to be enough that her definitions of 'woman' and 'man' delineate two social categories that align closely (albeit imperfectly) with who we usually call 'woman' and 'man', and that those definitions better organise the gender hierarchies that theorists had previously used 'woman' and 'man' to identify. Certainly, the preservation of these 'central functions' makes plausible that Haslanger's definitions, for the purpose of introducing theoretical terminology, do not deviate from ordinary usage to an unreasonable degree. 10 So far, I have focused on G&R. According to the terminological interpretation, Haslanger there introduces theoretical terminology principally for a specific (and quite limited) emancipatory end. She was explicitly not engaged in descriptive semantic analysis: Although the analyses I offer will point to existing social kinds (and this is no accident), I am not prepared to defend the claim that these social kinds are what our race and gender talk is "really" about. (2012/2000: 224) In Intuitions-building on intermediate views in e.g. her 2012/2003 and 2012/2005that descriptive semantic claim is precisely what Haslanger does defend. She argues that her definitions capture the operative concepts associated with the terms. As quoted earlier: 'This is not to propose a new meaning, but to reveal an existing one' (2012/2006: 398). Her willingness to defend this claim derived from an increased commitment to metasemantic externalism. 11 This is a development of, rather than a change to, her project in G&R. In G&R, Haslanger demarcates various social categories via her definitions of 'woman', 'man', etc., in order to provide tools for emancipation. In Intuitions, she argues that those definitions in fact point to the actual semantic extensions of those terms. This is intended to strengthen her original theoretical accounts, enhancing their capacity to fulfil their emancipatory function. To see this, consider some of the theoretical benefits to the development that Haslanger highlights: • It enables a more direct response to the charge that her initial definitions were 'changing the subject' (Haslanger /2006, along with the more general concern that her terminological appropriation might not be warranted. • It captures the sense in which the use of 'gender and race terms actually track hierarchical social formations' (2020: 237). • It exposes another 'potential site of ideology', namely the 'false semantic beliefs [that] can mask or distort how and whether our terms track kinds, sometimes contribute to the construction of them, and enable us to avoid taking responsibility for their effects' (ibid.) More broadly, the development yields a clearer route to exposing ideology. When G&R-era Haslanger presents her theories, she appropriates gender and race terms, using them to denote unjust social kinds in an attempt to cause (or jolt) people to see gender and race categories in a new light; but when Intutions-era Haslanger presents 10 Cf. Carnap on the desiderata for explicata: "close similarity is not required, and considerable differences are permitted" (1962/1950: 7). 11 This is largely due to the influence of Robert Stalnaker, with whom Haslanger became colleagues in 1998 (personal communication).
her theories, she reveals the injustices underlying the audience's own gender and race categories. It is unsurprising that, as Haslanger became a more committed externalist, she saw and utilised its promise for her project. 12 So, several years after engineering new theoretical definitions for 'woman', 'man', etc., Haslanger argues-in a continuation of the original project-that those definitions had perhaps captured the semantic meaning of those terms all along.

A successful conceptual engineering project
I now argue that, given the terminological interpretation, Haslanger's project has been a successful conceptual engineering project.
To do so, it is helpful to distinguish between conceptual engineering processes and conceptual engineering projects. Firstly, a conceptual engineering process is any process or method aimed at improving our linguistic or conceptual repertoires in some specified respect. Such processes may be large scale (e.g. using transformative communicative disruption to produce widespread meaning change), and they may be developed and set out explicitly (e.g. Carnap's method of explication), 13 but they need not be: when a theorist spends an afternoon using trial and error to develop a technical definition for a term, she is informally engaged in a local conceptual engineering process.
There are two important points to note about conceptual engineering processes. Firstly, different conceptual engineering processes have different success conditions. For example, a process designed to produce widespread meaning change is successfully undertaken if the relevant widespread meaning change is produced, whereas a process designed to yield a useful theoretical definition is successfully undertaken if it yields a useful theoretical definition. Secondly, conceptual engineering processes need not be designed to change the semantic meaning of a term. For example, the process of designing and using a new and surprising definition for a term counts as a conceptual engineering process, regardless of what, according to some given metasemantic theory, the original term happens to denote. 14 Now, conceptual engineering processes, when undertaken, are undertaken in the service of broader goals: widespread meaning changes are typically sought in pursuit of broader political goals, and useful theoretical definitions are typically sought in pursuit of broader theoretical goals. So let a conceptual engineering project be a broader, goal-driven package of work in which a conceptual engineering process has a substantial role, but where the success of the project is judged against the broader, driving goals.
How does this distinction bear on Haslanger's ameliorative project? Firstly, in G&R, Haslanger aims to introduce theoretical terminology by way of a simple conceptual engineering process: determine what cognitive or practical task the new terminology is to accomplish, and then design, stipulate and use definitions to that 12 This development continued, as is clearly evident in e.g. Haslanger /2010Haslanger and 2020 See Sterken 2020and Carnap 1962/1950 14 Cf. the discussion of Ball 2020 in § 3.2. end. 15 The task Haslanger highlights is to make explicit relevant gender and racial injustices within the local cultural technē. Plausibly, then, Haslanger undertook the process successfully just in case she introduced theoretical terminology that accomplishes that task.
Secondly, Haslanger's project has broader-albeit still quite limited-political goals. Although at the most general level she is working towards enhanced social justice, her project has a particular place within that: to develop a tool for exposing gender and race ideology, as a contribution to the first step to emancipation. Given that the conceptual engineering process described above had the principal role in initially developing the tool, the project is a conceptual engineering project-regardless of the fact that later developments of the tool (i.e. in Intuitions) did not involve additional conceptual engineering processes. And this conceptual engineering project has been successful just in case it has yielded a tool that can be used-and perhaps has been used-to expose gender and race ideology. Now, in both cases, it is reasonable to conclude that Haslanger's conceptual engineering endeavours have been successful. Firstly, in G&R, Haslanger did introduce theoretical terminology that makes explicit relevant gender and racial injustices within the local cultural technē. And this is so even if the descriptive semantic claim she makes in Intuitions is right. In G&R, Haslanger introduced terminology by stipulating that, within a particular theoretical context, 'woman', 'man', etc., were to be understood as having particular, theoretically-motivated definitions. This introduction was successful (qua introduction), as there were no extant terms with those precise definitions. Now, if Haslanger's claim in Intuitions is right, then it follows that the social kinds those definitions point to happened to be the semantic extensions of 'woman', 'man', etc., all along. But it does not follow from that claim that, at the time of G&R, 'woman', 'man', etc., had particular definitions at all, let alone the precise definitions formulated by Haslanger. Speakers did not associate those particular definitions with the terms, nor any other definitions that point to the corresponding social kinds. In contrast to Haslanger's theoretical terminology, everyday terms such as 'woman', 'man', etc., tend not to be explicitly defined. Haslanger's claim in Intuitions-that her definitions picked out those terms' extensions-does not undermine the point that, in G&R, she used those definitions to introduce theoretical terminology.
Secondly, Haslanger's project has been successful as it resulted in a theory that is a tool that can be used-and perhaps has been used-to expose gender and race ideology. To see this, note that Haslanger's theory encourages those engaging with it to recognise the possibility of defining gender and race categories-categories that (at least) overlap significantly with familiar gender and race categories-as classes within a hierarchy of privilege and subordination. The point is not that anyone who engages with Haslanger's theory experiences a change in how they conceive of gender and race: no one needs to reconceive of gender and race as essentially hierarchi-15 It is tempting, but not essential, to see this as a speaker-meaning approach to conceptual engineering (see e.g. Pinder 2020and Jorem 2021: using a stipulated definition need involve no more than speaker-meaning the definiens by the term in question. Nothing in this process, or the project more generally, depends on substantive linguistic mechanisms. cal. Rather, the point is that, when one engages with Haslanger's theory, one obtains theoretical resources for thinking about gender and race that puts hierarchies of privilege and subordination at centre stage; and recognising that we can define broadly familiar gender and race categories in these terms can be enough to come to see that we are all, to some extent, moulded by gender and racial injustices. And, although it is ultimately an empirical question, it is reasonable to think that many of the significant number of people who have engaged with Haslanger's work-e.g. a significant minority of philosophy graduates, interested lay people, etc.-thereby came to have these theoretical resources to hand. If this is right, then, given any reasonable standard for success in a practically-oriented philosophical project, Haslanger's project has, on the terminological interpretation, been successful.

Univocal vs. particularist success conditions
In the recent literature, a variety of alternative accounts of the success conditions for conceptual engineering have been proposed. On some of these views, given the terminological interpretation, Haslanger's project has been successful; on other views, it has been unsuccessful.
For an example of the latter, consider Cappelen's (2018) Austerity Framework. According to Cappelen, the aim of conceptual engineering is to change the semantic meaning of some or other term-and to engage in conceptual engineering successfully, one must change the semantic meaning of the target terms in the intended way. Such a view implies that, on the terminological interpretation, Haslanger's conceptual engineering endeavours have been unsuccessful-she has not managed to change the semantic meaning of target terms in the intended way.
For another example, consider the view developed by Simion & Kelp (2020). They argue that conceptual engineering is successful just when: (a) the engineered concept continues to be used, and (b) what explains this continued use is that the engineered concept does that which the engineer intended it to do. On this view, given the terminological interpretation, Haslanger's conceptual engineering endeavours have been successful just in case: (a) her definitions continue to be used, and (b) what explains this continued use is that the definitions make explicit relevant gender and racial injustices within the local cultural technē. On this account, again, I doubt that Haslanger has been successful, as there is perhaps a better explanation for why Haslanger's definitions continue to be used to whatever extent they are in fact still used: that they constitute a carefully defended, novel and provocative approach to an important topic from a well-known philosopher.
Conversely, consider Nado's view that '[s]uccess is measured in efficacy' (2021b: S1523): conceptual engineering is successful just in case the designed concept or definition fulfils its intended function. This is a much weaker criterion than that advocated by Simion and Kelp: Nado does not require the concept or definition to be used because it fulfils its intended function. Accordingly, on this account, I suspect that Haslanger does count as having undertaken conceptual engineering successfully: her definitions do fulfil their intended functions-they do make explicit relevant gender and racial injustices within the local cultural technē-regardless of whether this fact explains their continued use. Now, the above views are univocal: they offer the same success conditions for all instances of conceptual engineering. However, in light of the preceding discussion, univocal views may seem problematic, as they do not take into account the variety of possible conceptual engineering processes and projects. I have urged that, if we want to ascertain whether Haslanger's conceptual engineering endeavours have been successful, we need to look at her project, and the processes she used. The reason for this is that each project and process comes with its own success conditions: the success conditions of a conceptual engineering project are fixed by the engineer's broader goals; and the success conditions of a conceptual engineering process are fixed by whatever it was designed or intended to do. To stipulate, based on general considerations, that Haslanger's conceptual engineering endeavours were successful only if she changed the semantic meaning of some terms, or only if her definitions continue to be used because they fulfil their intended functions, is to miss the point of what Haslanger is doing.
Herein, then, I am recommending a particularist view of the success conditions for conceptual engineering. To ascertain the success conditions of a conceptual engineering process, we must look at that particular process. And to ascertain the success conditions a conceptual engineering project, we must look at that particular project. There need be no general account of the conditions under which conceptual engineering is successful and, in proposing such an account, one would risk missing the unique aims and purposes that different conceptual engineers have.

The inclusion problem
I have not, herein, provided a detailed introduction to Haslanger's work in social philosophy. Rather, I have focused on high-level interpretations of Haslanger's goals and methodology. So, in response to the arguments above, one might insist that Haslanger's project fails for low-level reasons I have not considered. For example, some have objected-and Haslanger has conceded-that her definitions 16 problematically excluded some women from being counted as women and some men from being counted as men.
[…] [S]ome women are prevented from presenting as women, and some men are prevented from presenting as men, and so do not meet the conditions I proposed. (Haslanger 2020: 236) Now, this is not the place to examine the details of Haslanger's project. No philosophical view is without objections. Nonetheless, I close by responding briefly to this so-called inclusion problem.
The problem, as I understand it, targets Haslanger's project qua revisionary interpretation. Folk in general should not adopt Haslanger's definitions because the exclusive nature of the definitions would serve (in particular) to deepen trans oppres-sion-whether directly (in the case of trans people who do not present as their gender) or indirectly (in the case of those who do). But the inclusion problem does not automatically undermine Haslanger's project qua terminological interpretation. Theoretical terminology is often designed to focus on particular aspects of the phenomenon under investigation and need not be appropriate more generally. Haslanger's definitions of 'woman' and 'man' reveal a certain subspecies of gender ideology, say patriarchal ideology: the ideology whose revelation is the first step to the emancipation of oppressed cisgender people. But her definitions do not reveal cisnormative ideology: the ideology whose revelation is the first step to the emancipation of trans people. But this is not so much an objection to Haslanger's project as an acknowledgement of its limitations.
Haslanger draws the following lesson from her reflections on the inclusion problem: linguistic choices that might be emancipatory at one moment, or for some individuals, […] may be inadequate in a broader context and even deepen other forms of oppression. (2020: 237) In using any tools for emancipation, we should work hard to be sensitive to, and to mitigate against, intersecting forms of oppression. Of course, Haslanger was always aware of this. Nonetheless, at the time of G&R, we were all firmly in the grip of cisnormative ideology. At the present moment, trans liberation movements are fighting to expose cisnormative ideology: the grip has loosened, although much work remains to be done. For this reason, at the present moment, engaging in emancipatory projects that reveal patriarchal ideology while perpetuating cisnormative ideology may be particularly harmful to trans people. While Haslanger's tools have played a role in revealing patriarchal ideology, and may have such a role to play again, it is perhaps inappropriate to use those tools in the present moment.
The appropriate conclusion is that Haslanger's project-like any other emancipatory project-is to be understood within a particular sociopolitical context. That is, the inclusion problem shows only that her tools are not universally applicable. The problem does not undermine the claim that her project, on the terminological interpretation, has been successful. 17

Conflict of interest None.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission